Approach To Arabic Linguistics Diglosie 403 PDF

Approaches to Arabic Linguistics
Studies in Semitic Languages

and Linguistics
Editorial board
T. Muraoka
VOLUME 49
Kees Versteegh
Approaches to Arabic
Linguistics
Presented to Kees Versteegh on the Occasion
of his Sixtieth Birthday
Edited by
Everhard Ditters and Harald Motzki
LEIDEN BOSTON
2007
This book is printed on acid-free paper.
Library of Congress Cataloging-in-Publication Data
A C.I.P. record for this book is available from the Library of Congress.
ISBN 0081-8461
ISBN 978 90 04 16015 6
Copyright 2007 by Koninklijke Brill NV, Leiden, The Netherlands.

Koninklijke Brill NV incorporates the imprints Brill, Hotei Publishing,
IDC Publishers, Martinus Nijhoff Publishers and VSP.
All rights reserved. No part of this publication may be reproduced, translated,

stored in a retrieval system, or transmitted in any form or by any means, electronic,
mechanical, photocopying, recording or otherwise, without prior written permission
from the publisher.
Authorization to photocopy items for internal or personal use is granted by

Koninklijke Brill NV provided that the appropriate fees are paid directly to
The Copyright Clearance Center, 222 Rosewood Drive, Suite 910,
Danvers, MA 01923, USA.
Fees are subject to change.
printed in the netherlands

CONTENTS
Preface .................................................................................................. xi
Bibliography Kees Versteegh ............................................................ xv
HISTORY
Inside the Speakers Mind: Speakers Awareness as Arbiter of

Usage in Arabic Grammatical Theory ......................................... 3
Ramzi Baalbaki
Pragmatics and Contractual Language in Early Arabic

Grammar and Legal Theory ......................................................... 25
Michael Carter
Idmr in the Man of al-Farr: A Grammatical Approach

between Description and Explanation ......................................... 45
Kinga Dvnyi
Arabic allad as a Conjunction: An old Problem and a New

Approach ......................................................................................... 67
Werner Diem
Les origines de la grammaire arabe, selon la tradition:

description, interprtation, discussion .......................................... 113
Pierre Larcher
Sbawayhis View of the zarf as an mil ......................................... 135

Aryeh Levin
Problems in the Medieval Arabic Theory of Sentence Types ......... 149

Yishai Peled
viii contents
Arabic avant la lettre. Divine, Prophetic, and Heroic Arabic ....... 189
Stefan Wild
Inflection and Government in Arabic According to Spanish

Missionary Grammarians from Damascus (XVIIIth Century):
Grammars at the Crossroads of Two Systems ............................. 209
Otto Zwartjes
LINGUISTICS
The Linguistic Analysis and Rules of Pause in Arabic ................... 247

Salman H. Al-Ani
The Explanation of Homonymy in the Lexicon of Arabic ............. 255

Georges Bohas and Abderrahim Saguer
The Periphrastic Bilingual Verb Construction as a Marker of

Intense Language Contact. Evidence from Greek, Portuguese
and Maghribian Arabic ................................................................. 291
Louis Boumans
Faula, faila, faala: dispersion et rgularits smantiques dans

les trois schmes simples du verbe arabe ..................................... 313
Joseph Dichy
Featuring as a Disambiguation Tool in Arabic Natural

Language Processing ....................................................................... 367
Everhard Ditters
Arabic on the Media: Hybridity and Styles ..................................... 403

Mushira Eid
The Use of Morphological Patterns in Arabic Grammars of

Turkic ............................................................................................... 435
Robert Ermers
Lexical Gaps in Arabic: Evidence from Dictionaries ...................... 455

Jan Hoogland
contents ix
Masdar Formation .............................................................................. 475

Joost Kremers
Mthodologie linguistique: organisation de la langue arabe.

Organisation gnrale des langues ............................................... 501
Andr Roman
DIALECTS
How to be KOOL in Arabic Writing: Linguistic Observations

from the Side Line .......................................................................... 527
Gert Borg
Hello, I say, and welcome! Where from, these riding men?

Arabic Popular Poetry and Political Satire: a Study in
Intertextuality from Jordan ........................................................... 543
Clive Holes
Notes on the Dialects of the Lgt and H amda h of Southern

Sinai ................................................................................................. 565
Rudolf de Jong
Classical and Colloquial Arabic Archaisms ..................................... 595

Alan S. Kaye
Do They Speak the Same Language? Language Use in Juba

Local Courts .................................................................................... 607
Catherine Miller
Paradigmatic Stability and Final Laryngeals in Nigerian Arabic:

Why History Repeats itself, without Actually Doing so ............. 639
Jonathan Owens
Some Aspects of Diglossia as Reflected in the Vocabulary of

Literary and Colloquial Arabic ..................................................... 653
Judith Rosenhouse
x contents
Everything you Always Wanted to Know about l, yil to say

in Egyptian Arabic ......................................................................... 675
Manfred Woidich
Index .................................................................................................... 701

PREFACE
From when we began to compile a Festschrift for Kees Versteegh on the

occasion of his sixtieth birthday, the first problem we had to tackle was:
Who should we invite to contribute? The broad range of his scholarly
interests, his expertise in different fields, his academic contribution
to them, as well as his network of global contacts made it clear that
the circle of his colleagues and friends is too large to invite all to con-
tribute. So we decided to confine this volume to Kees Versteeghs core
contribution: Arabic linguistics. This decision will surely be regretted
by his colleagues working in related fields of scholarship in which Kees
Versteegh is engaged as well, but we had to make a choice. Moreover,
our time schedule forced us to only select the first set of early contribu-
tions brought in. In the end, what does it matter? Someone will have
to prepare another for his sixty-fifth!
Even with this limitation, the present Festschrift has become volumi-
nous, since the eagerness of scholars in the field of Arabic linguistics
to contribute has been overwhelming. It clearly demonstrates: Kees
Versteegh has been widely considered to be one of the most eminent
scholars in this discipline. He also is appreciated as a tirelessly working
editor and co-editor of renowned book series, collective volumes and
encyclopaedias. Moreover, he has won many friends by his kindness
and dependability.
With the list of Kees Versteeghs publications at hand, one realises the
number, diversity, and depth of his scholarly interests, from Hellenistic
elements in Arabic linguistics and other fields of Islamic culture, to
the history of Arabic grammar as a scholarly discipline from classical
to modern times, grammatical and linguistic phenomena described
by scholars of Arabic grammar, changes in written and spoken Arabic
through the ages, as well as early Qurnic exegesis as source for the
beginnings of Arabic grammar. In addition, he ranslated and commented
on Arabic texts, grammatical treatises and other genres including a novel.
His books Greek Elements in Arabic Linguistic Thinking; Pidginization
and Creolization: the Case of Arabic; Arabic Grammar and Qurnic
Exegesis in Early Islam; and The Arabic Language are handbooks. Every
single scholar and student of Arabic linguistics can not bypass them.
xii preface
Several of his publications have been translated into Arabic, and others
will follow.
At Nijmegen University, where Kees Versteegh is working since
1972, first as lecturer and later as full professor, he organized several
workshops about the History of Arabic Grammar (1984 and 1987)
and the Model of Arabic Grammar in other Languages (1997). With
intense pleasure the participants of these conferences remember these
fruitful academic gatherings occurring in a very informal and convivial
atmosphere. These gatherings resulted in two collective volumes he
edited in cooperation with one of his many colleagues. Thanks to his
co-editorship, international publication projects such as the History of
Language Sciences and the Encyclopaedia of Arabic Language and Lin-
guistics, Arabic linguistics acquired a firm place in the field of linguistics
in general. Generations of Dutch students, translators and interpreters
of Arabic will know his name from the two volumes of the dictionary
Woordenboek Arabisch-Nederlands/Nederlands-Arabisch of which he is
one of the editors.
This Festschrift will be a monument for Kees Versteegh in Arabic
studies, as a homage to his scholarly oeuvre. Several contributions
camd from former students who wrote their Ph.D. theses under his
supervision. As a matter of fact, all contributors studied with him. We
divided the collected articles into three chapters reflecting the foci of
his scholarly oeuvre: history of Arabic grammar, Arabic linguistics, and
Arabic dialectology. We have put the history of Arabic grammar first,
since it is Kees Versteeghs true domain. He wrote his Ph.D. in this field,
dedicated most of his publications to it, and returns to it. He translated
and commented upon al-Zajjjs Kitb al-dh , a theoretical treatise on
Arabic grammar and one of the classical works on this issue. Since the
1980s, his research and publications also have been directed towards
other historical and thematical topics in linguistics, and he certainly
will recognize many of his own ideas in the contributions in all three
sections. His studies on pidginization and creolization in Arabic and his
interest in Arabic dialects express his feeling for the historical develop-
ment of the Arabic language.
Kees Versteegh is a polyglot. When invited to lecture outside the
Netherlands he enjoys lecturing in the language of is audience, in Eng-
lish, French, German and Arabic, as well as Spanish or Czech. From
the start of his academic careerhe first studied Greek and Latinhe
cherished a fondness for these languages. In his research this became
preface xiii
apparent in his search for Hellenistic elements in Arabic linguistics.

Together with a colleague of Classical Studies, he also teaches regular
classes on linguistics. His interest inwell, let us sayexotic languages
is exceptional. His eyes sparkle when he tells how he learned Inuktitut,
the language of the Inuit, from an indigenous scholar in Alaska or how
he mastered Hottentot clicks in South Africa.
Many colleagues enjoyed Kees Versteeghs hospitality in his sextons
house in picturesque Batenburg. They will never forget the tasty meals
he prepares himself as the evenings conversation quickly passes from
scholarly issues to personal and private ones, as for example his experi-
ences as director of the Netherlands-Flemish Institute in Cairo or the
birds he suddenly noticed by a Zamalek window. One may be engaged
with Kees in a really basic discussion of Arabic linguistics, only to be
interrupted by his quick sprint outdoors to observe an ornithological
event, afterwards resuming the discussion exactly at the comma or full
stop of his last intervention. Bird watching remains his hobby and on
a walk along the Maas (or any other river), he never forgets his bin-
oculars. Besides linguistics, a guest learns a lot about the birds of the
region, their names in Dutch and other languages, including Arabic.
Guests get free access to his rich private library with its rare books
on Arabic linguistics and even the privilege of borrowing without any
check for solvability.
The study of Arabic has been Arabic-speaking scholars domain for
many centuries. Kees Versteegh has unearthed their understanding of
the language and translated it for Western scholarship. Western scholars
linguistic study of Arabic, according to current standards, started only
at the beginning of the 19th century. So while the western linguistic
approach is much younger than that of those native scholars, Western
Arabic linguistics are on par with even native intellectual efforts, and
Kees Versteegh has contributed to this state of affairs. Had classic schol-
ars of the Arabic language like Sbawayhi or al-Zajjj the change to
witness the linguistic achievements of their modern Western colleagues
like Kees Versteegh, they would have been impressed. They might have
cited the verses of Ibn H amds (Sicily, 447527/10561133):
al adawti tuyrin fish in al anna afsah ah ajamu
lahunna ardu inda l-halli muhammalatu l-wazni l tulamu
turajjiu fh durba l-luh ni fa-tutribun wa-hya l tufhamu
On the tunes of birds that speak true Arabic,

but the most eloquent of them are foreigners
xiv preface
who use metres that are neglected by al-Khall and unknown

in which they quaver notes of various melodies,
delighting us although they are unintelligible.
This Festschrift would not have seen the light without several helping
hands. Greetje Heemskerk compiled the bibliography of his publications.
Marjolein van der Heul and Ine Smeets produced the raw version of
the index. Elizabeth Bishop copy-edited the English contributions. Joed
Elich of Brill has been willing to publish the book. Ingrid Heijckers
lead us safely through the whole of the production process and beyond.
Renee Otto supervised the publication process. We are very grateful for
their commitment.
Everhard Ditters and Harald Motzki

BIBLIOGRAPHY KEES VERSTEEGH
1977
Greek Elements in Arabic Linguistic Thinking. (= Studies in Semitic
Languages and Linguistics, 7). Leiden: E. J. Brill. Also published as
PhD thesis University of Nijmegen. [Arabic translation by Mahmud
Kankr, 2000: Ansir ynniyya f l-fikr al-luaw al-arab. Amman:
Jamiyyat umml al-mat bi at-tawuniyya].
1978
The Arabic Terminology of Syntactic Position. Arabica 25, 261281.
1979
Die Mission des Kyrillos im Lichte der arabo-byzantinischen Bezie-
hungen. Zeitschrift der deutschen Morgenlndischen Gesellschaft 129,
233262.
Reviews
Ephrem Hunayn Festival Baghdd 47/2/1974. 1974. Baghdad: Mat bt
Majma al-Lua al-Suryniyya, Bibliotheca Orientalis 36, 9697.
Rundgren, Frithiof. 1976. ber den griechischen Einflu auf die arabische
Nationalgrammatik. Uppsala: Acta Universitatis Upsaliensis. N.S. 2(5),
119144. Bibliotheca Orientalis 36, 235236.
1980
Hellenistic Education and the Origin of Arabic Grammar. Progress in
Linguistic Historiography: Papers from the International Conference
on the History of the Language Sciences, Ottawa, 2831 August 1978.
Edited by Konrad Koerner. Amsterdam: J. Benjamins. 333344.
Logique et grammaire au dixime sicle. Histoire, pistmologie, Lan-
gage 2. 3952.
Notice bibliographique. Histoire, pistmologie, Langage 2. 6775.
The International Project Onomasticon Arabicum. Bibliotheca Orientalis
37, 291294.
The Origin of the Term qiys in Arabic Grammar. Zeitschrift fr ara-
bische Linguistik 4, 730.
The Stoic Verbal System. Hermes 108, 338357.
xvi bibliography kees versteegh
Reviews
Ambros, Edith. 1979. Sieben Kapitel des arh Kitb Sbawaih von ar-
Rummn in Edition und bersetzung. Wien: Verlag des Verbandes der
wissenschaftlichen Gesellschaften sterreichs. Bibliotheca Orientalis
37, 361362.
Malti Douglas, Fedwa and Genevive Fourcade. 1976. The Treatment
by Computer of Medieval Arabic Biographical Data: An introduction
and guide to the Onomasticum Arabicum. Paris: Centre national de
la recherche scientifique. Bibliotheca Orientalis 37, 362.
Traini, Renato. 1977. Sources biographiques des Zadites (annes 1221200
h.): Letters alif-ha. Paris: Centre national de la recherche scientifique.
Bibliotheca Orientalis 37, 362363.
1981
A Dissenting Grammarian: Qutrub on declension. Historiographia Lin-
guistica 8, 403429. [See also The History of Linguistics in the Near
East. Edited by Kees Versteegh, Konrad Koerner and Hans-Josef
Niederehe. 1983. Amsterdam: J. Benjamins. 167193].
De taalsituatie in de Arabische wereld. De taal van de Islam: Opstellen
over Arabische, Turkse en Afghaanse cultuur. Nijmegen: Nederlandse
Vereniging voor de Studie van het Midden-Oosten en de Islam.
1938.
La conception des temps du verbe chez les grammairiens arabes. Ana-
lyses, thorie 3, 4768.
Reviews
Klein-Franke, Felix. 1980. Die klassische Antike in der Tradition des
Islam. Darmstadt: Wissenschaftliche Buchgesellschaft. Bibliotheca
Orientalis 38, 734737.
1982
Progress and Change in the History of Arabic Grammar. Linguistics
in the Netherlands 1982. Edited by Saskia Daalder and Marinel Ger-
ritsen. Amsterdam: North-Holland. 3950.
Structural Change and Pidginization in the History of the Arabic
Language. Papers from the 5th International Conference on Histori-
cal Linguistics, Galway, April 610 1981. Edited by Anders Ahlqvist.
Amsterdam: J. Benjamins. 362373.
Vertalingen voor 1ste jaarsstudenten Arabisch (with Gert Borg). Nijmegen:
Katholieke Universiteit Nijmegen.
bibliography kees versteegh xvii
Reviews
Dagorn, Ren. 1981. La geste dIsmal daprs lonomastique et la tradi-
tion arabes, Genve: Droz; Paris: Champion. Bibliotheca Orientalis
39, 720726.
Rowson, Everett and Seeger Bonebakker. 1980. A Computerized Listing
of Biographical Data from the Yatmat al-Dahr by al-Thalib. Malibu:
Undena Publications. Bibliotheca Orientalis 39, 727729.
1983
A Dissenting Grammarian: Qutrub on declension. The History of Lin-
guistics in the Near East. Konrad Koerner, Hans-Josef Niederehe, and
Kees Versteegh (Eds.). Amsterdam: J. Benjamins. 167193. [See also
Historiographia Linguistica 1981. 8, 403429].
Arabic Grammar and the Corruption of Speech. Arab Language and
Culture. Edited by Ramzi Baalbaki. Beirut: American University of
Beirut. [= al-Abhth 31]. 139160.
Current Bibliography on the History of Arabic Grammar. Zeitschrift
fr arabische Linguistik 10, 8689.
History of Eastern Linguistics in the Soviet Union. Historiographia
Linguistica 10, 289307.
The History of Linguistics in the Near East. (= Studies in the History of
Linguistics, 28). Editor (with Konrad Koerner and Hans-Josef Nie-
derehe). Amsterdam: J. Benjamins.
1984
Arab Grammatical Studies before Sbawayh. Matriaux pour une histoire
des thories linguistiques = Essays toward a history of linguistic theories
= Materialien zu einer Geschichte der sprachwissenschaftlichen Theo-
rien. Edited by Sylvain Auroux, Michel Glatigny, Andr Jolly, Anne
Nicolas and Irne Rosier. Lille: Universit de Lille III. 227238.
Piginigado, kreoligado kaj Esperanto. Hungara Vivo 4, 127129.
Pidginization and Creolization: The case of Arabic. (= Current Issues in
Linguistic Theory, 33). Amsterdam: J. Benjamins.
xviii bibliography kees versteegh
Reviews
Bakalla, Muhammad. 1983. Arabic linguistics. London: Mansell. Biblio-
theca Orientalis 41, 751754.
Carter, Michael. 1982. Arab Linguistics: An introductory classical text
with translation and notes. Amsterdam: J. Benjamins. Bibliotheca
Nebes, Norbert. 1982. Funktionsanalyse von kna yafalu. Hildesheim:
Olms. Bibliotheca Orientalis 41, 754757.
1985
La Grande tymologie dIbn Jinn. La linguistique fantastique. Edited
by Sylvain Auroux, Jean-Claude Chevalier, Nicole Jacques-Chaquin
and Christiane Marchello-Nizia. Paris: Denol. 4450.
Studies in the History of Arabic Grammar. Papers Presented during a
Workshop held at the University of Nijmegen, April 1619, 1984. Editor
(with Hartmut Bobzin). Wiesbaden: O. Harrassowitz. (= Zeitschrift
fr arabische Linguistik 15).
Survey of Journals. Arab Journal of Language Studies / al-Majalla
al-arabiyya li-d-dirst al-luawiyya 2, 189197.
The Development of Argumentation in Arabic Grammar: The declen-
sion of the dual and the plural. Studies in the History of Arabic Gram-
mar. Edited by Hartmut Bobzin and Kees Versteegh. Wiesbaden:
Harrassowitz. 1985. 152173. [= Zeitschrift fr arabische Linguistik
15].
Translation
Salih, Tayyib. Seizoen van de trek naar het noorden. Uit het Arabisch
vertaald en van een nawoord voorzien. Amsterdam: Meulenhoff.
[Dutch translation of Mawsim al-hijra il -aml].
Reviews
Sezgin, Fuat. 1982, 1984. Geschichte des arabischen Schrifttums. VIII.
Lexikographie. IX. Grammatik. Leiden: E. J. Brill. Historiographia
Linguistica 12, 452461.
1986
History of Arabic Grammar. Nijmegen: Instituut voor Talen en Culturen
van het Midden Oosten, Katholieke Universiteit.
bibliography kees versteegh xix
Latinitas, Hellnisms, Arabiyya. The History of Linguistics in the Clas-

sical Period. Edited by Daniel J. Taylor. (= Historiographia Linguistica
13, 425448). Amsterdam: J. Benjamins, 1987. 251274.].
The Origin of the Romance Languages and the Arabic Dialects. Islo e
arabismo na pennsula ibrica: Actas do XI. congresso da Unio Euro-
peia de Arabistas e Islamlogos, Evora, Faro, Silves, 29 set.6 out. 1982.
Edited by Adel Y. Sidarus. vora: Universidade de vora. 337352.
Word Order in Uzbekistan Arabic and Universal Grammar. On the
Dignity of Man: Oriental and classical studies in honour of Frithiof
Rundgren. Edited by Tryggve Kronholm and Eva Riad. (= Orientalia
Suecana 3334). Uppsala: Almqvist and Wiksell. 443453.
1987
al-Arab f jibl al-Alp. al-Azmina 7, 2630.
Die arabische Sprachwissenschaft. Grundri der arabischen Philologie,
II. Edited by Helmut Gtje, Wiesbaden: L. Reichert. 148176.
Het Arabisch: Norm en realiteit (with Arie Schippers). Muiderberg: D.
Coutinho.
Marginality in the Arab Grammatical Tradition. Papers in the History
of Linguistics: Proceedings of the Third International Conference on
the History of the Language Sciences (ICHoLS III), Princeton, 1923
August 1984. Edited by Hans Aarsleff, Louis G. Kelly and Hans-Josef
Niederehe. Amsterdam: J. Benjamins. 8796.
Nahwiyyna wa-luawiyyna wa-mawqif Dozy iz at-turt an-nahw
al-arab. F l-mujamiyya al-arabiyya al-musira. Edited by Ahmed
El-Ayed and Ibrahim Ben Mrad. Tunis: Dr al-arb al-Islm.
401413.
Reviews
Sublet, Jacqueline. 1985. Cahiers donomastique Arabe, 19821984. Paris:
Centre national de la recherche scientifique. Bibliotheca Orientalis
44, 130133.
1988
De ontwikkeling van de technische woordenschat in Modern Standaard
Arabisch: Naar aanleiding van een recente publicatie. Sharqiyyt 1,
8085.
xx bibliography kees versteegh
Geschiedschrijving in de klassieke Arabisch-Islamitische samenleving.

Tussen traditie en wetenschap: Geschiedbeoefening in niet-westerse
culturen. Edited by R.B. van de Weijer, P.G.B. Thissen and R. Schn-
berger. Nijmegen: Katholieke Universiteit Nijmegen. 133145.
Pourquoi tudier la tradition grammaticale? Le Maroc et la Hollande:
tudes sur lhistoire, la migration, la linguistique et la smiologie de la
culture. Edited by Abdelmajid Kaddouri, Jilali Sab and Abdelmajid
Zeggaf. Rabat: Universit Mohammed V. 207217.
1989
A Sociological View of the Arab Grammatical Tradition: Grammarians
and their professions. Studia linguistica et orientalia memoriae Haim
Blanc dedicate. Edited by Paul Wexler, Alexander Borg and Sasson
Somekh. Wiesbaden: Harrassowitz. 289302.
Early and late Grammarians in the Arab Tradition: The morpho-
nology of the hollow verbs. Zeitschrift fr arabische Linguistik 20,
922.
La tradition arabe. I : Le langage, la religion et la raison. Histoire des
ides linguistiques, I : La naissance des mtalangages en Orient et en
Occident. Edited by Sylvain Auroux. Lige: P. Mardaga. 243259.
The Definition of Philosophy in a Tenth-century Grammarian. Jerusa-
lem Studies in Arabic and Islam 12, 6692.
Vulgair Latijn en Koine-Grieks: De verhouding tussen standaardtaal
en volkstaal. Lampas 22, 7491.
Reviews
Bergter, Annette. 1989. Das Kapitel inna wa-ahawtuh aus dem Man-
haj as-slik des Grammatikers Ab H ayyn al-arnt (12561344).
Hildesheim: G. Olms. Historiographia Linguistica 16, 180184.
1990
Are Linguists Ridiculous? Notes on a heavenly discussion between
grammarians in the 11th century. History and Historiography of
Linguistics: Proceedings of the Fourth International Conference on the
History of the Language Sciences (ICHoLS IV), Trier, 2428 August
1987, I: Antiquitity17th Century. Edited by Konrad Koerner and
Hans-Josef Niederehe. Amsterdam: J. Benjamins. 147155.
Borrowing and Influence: Greek grammar as a model. Le langage dans
lantiquit. Edited by Pierre Swiggers and Alphons Wouters. Leuven:
Peeters. 197212.
bibliography kees versteegh xxi
Freedom of the Speaker? The term ittis and related notions in Arabic
grammar. Studies in the History of Arabic Grammar, II. Edited by
Michael G. Carter and Kees Versteegh. Amsterdam: J. Benjamins,
1990. 281293 [Arabic translation by Bouchaib Barramou: H urriyyat
al-mutakallim? Mussa lah al-ittis wa-l-mafhm al-murtabiha bihi f
n-nahw al-arab. Fikr wa-naqd 3:24 (1999). 99110].
Grammar and Exegesis: The origins of Kufan grammar and the Tafsr
Muqtil. Der Islam 67, 206242. [Reprint in The Early Islamic Gram-
matical Tradition. Edited by Ramzi Baalbaki. Aldershot [etc.]: Ash-
gate/Variorum. 2007. 3773].
Over taal en verandering. Nmegen: Katholieke Universiteit. Inaugural
lecture, University of Nmegen.
Studies in the History of Arabic Grammar, II. Proceedings of the 2nd Sym-
posium on the History of Arabic Grammar, Nijmegen, 27 April1 May,
1987. Editor (with Michael G. Carter). Amsterdam: J. Benjamins.
The Arab Presence in France and Switzerland in the 10th Century.
Arabica 37, 359388.
The Earliest Commentary on the Qurn: Muqtils Tafsr. Makalah-
makalah yang disampaikan dalam rangka kunjungan menteri Agama
R.I.H. Munawir Sjadzali, M.A. ke Negeri Belanda (31 Oktober
7 November 1988). Edited by Wim Stokhof and Nico Kaptein. Jakarta:
INIS. 213219. [with Indonesian translation: Tafsir Quran paling
awal: Tafsir Muqatil].
Reviews
Owens, Jonathan. 1988. The Foundations of Grammar: An introduction
to medieval Arabic grammatical theory. Amsterdam: J. Benjamins. Die
Welt des Islams 30, 248250.
1991
Arabic Language Teaching and the Status of Standard Arabic. Proceed-
ings of the Symposium on Differentiation in LSP, Learning and Teach-
ing, Leuven, 710 November 1990. Edited by Serge Verlinde. Leuven:
Instituut voor levende talen. 95102.
Greek Translations of the Qurn in Christian Polemics (9th century
A.D.). Zeitschrift der Deutschen Morgenlndischen Gesellschaft 141,
5268.
Man. 1. In grammar. The Encyclopaedia of Islam: New edition. Edited
by C. E. Bosworth, E. van Donzel, B. Lewis and Ch. Pellat, assisted
by F. Th. Dijkema and S. Nurit. Leiden: E. J. Brill. VI, 346.
xxii bibliography kees versteegh
New approaches towards the history of Arabic. Taqaddum al-lisniyyt

fi l-aqtr al-arabiyya [Progrs de la linguistique dans les tats arabes].
Edited by Abdelkader Fassi Fehri. Beirut: Dr al-arb al-Islm.
199216.
The substratum debate in creole linguistics. Diachronica 8, 5980.
Two Conceptions of Irreality in Arabic Grammar: Ibn Him and Ibn
al-jib on the particle law. De la grammaire de larabe aux gram-
maires des arabes. Edited by Pierre Larcher. Damas: Institut Franais
de Damas. 7792 [= Bulletin dtudes Orientales, 43].
Reviews
Holes, Clive. 1990. Gulf Arabic. London: Routledge. Bibliotheca Orien-
talis 48, 293296.
Wouters, Alfons. 1988. The Chester Beatty Codex Ac. 1499: A Graeco-
Latin lexicon on the Pauline Epistles and a Greek grammar. Leuven:
Peeters. Bibliotheca Orientalis 48, 523527.
1992
De val van de Barmakiden volgens Tabar. Elf wijzen van interpreteren:
Essays over het lezen van teksten uit het islamitisch cultuurgebied.
Edited by Ed de Moor. Nijmegen: Instituut voor Talen en Culturen
van het Midden-Oosten. 117126.
Grammar and Rhetoric: Jurjn on the verbs of admiration. Studies in
Semitic Linguistics in Honor of Joshua Blau. Edited by M. Bar-Asher
[et al.]. Jerusalem: Hebrew University of Jerusalem. 113133. [=
Jerusalem Studies in Arabic and Islam 15].
The Debate about Latin and Early Romance. Diachronica 9, 259285.
The History of Linguistics in the Low Countries. (= Studies in the His-
tory of the Language Sciences 64). Editor (with Jan Noordegraaf and
Konrad Koerner). Amsterdam: J. Benjamins. [Published in 1988 as
Historiographia Linguistica 15, 1+2].
Reviews
Malina, Renate. 1987. Zum schriftlichen Gebrauch des kairenischen
Dialekts anhand ausgewhlter Texte von Sadaddn Wahba. Berlin:
Schwarz. Bibliotheca Orientalis 49, 526528.
Mitchell, T. F. 1990. Pronouncing Arabic, I. Oxford: Clarendon. Biblio-
theca Orientalis 49, 523526.
Sluiter, Ineke. 1990. Ancient Grammar in Context, Amsterdam: VU
University Press. Historiographia Linguistica 19, 401408.
bibliography kees versteegh xxiii
1993
Arabic Grammar and Qurnic Exegesis in Early Islam. Leiden: E. J.
Brill.
Esperanto as a First Language: Language acquisition with a restricted
input. Linguistics 31, 539555.
Leveling in the Sudan: From Arabic creole to Arabic dialects. Interna-
tional Journal of the Sociology of Language 99, 6579.
Three is a Crowd: lawyers and linguists on Qurn 4/11. Wolfdietrich
Fischer zum 65. Geburtstag. Edited by Hartmut Bobzin and Otto Jas-
trow. Wiesbaden: Harrassowitz. 302315 [= Zeitschrift fr arabische
Linguistik 25].
Reviews
Jastrow, Otto. 1990. Der arabische Dialekt der Juden von Aqra und Arbl.
Wiesbaden: O. Harrassowitz. Bibliotheca Orientalis 50, 858862.
Johnstone, Barbara. 1991. Repetition in Arabic Discourse: Paradigms,
syntagms, and the ecology of Language. Amsterdam: J. Benjamins.
1994
Arabic Outside the Arab World. Editor. New Delhi: Bahri Publications.
[= Indian Journal of Applied Linguistics, 20].
The Arabic Linguistic Tradition in a Comparative Perspective. Arab
Journal of the Humanities 12 (no. 46), 282297.
The Notion of Underlying Levels in the Arabic Grammatical Tradi-
tion. Historiographia Linguistica 21, 271296. [Arabic translation by
Bouchab Barramou: Mafhm al-mustawayt at-tahtiyya f t-turt
an-nahw al-arab. Recherches Linguistiques 6:2 (2001), 81126].
Reviews
kesson, Joyce. 1990. Ahmad b. Al b. Masd on Arabic morphology =
Marh al-arwh, edited with translation, commentary and introduc-
tion. Part 1: The strong verb as-sahh. Leiden: E. J. Brill. Bibliotheca
Broselow, Ellen, Mushira Eid and John McCarthy (Eds.). 1992. Perspec-
tives on Arabic Linguistics, IV. Amsterdam: J. Benjamins. Bibliotheca
Comrie, Bernard and Mushira Eid (Eds.). 1991. Perspectives on Arabic
Linguistics, III. Amsterdam: J. Benjamins. Journal of the American
Oriental Society 114, 107109.
xxiv bibliography kees versteegh
Robins, Robert H. 1993. The Byzantine Grammarians: Their place

in history. Berlin: Mouton de Gruyter. Bibliotheca Orientalis 51,
719723.
1995
The Explanation of Linguistic Causes: Az-Zajjjs Theory of grammar,
introduction, translation, and commentary. Amsterdam: J. Benjamins.
(Translation of Ab l-Qsim Abd ar-Rahmn b. Ishq az-Zajjj, Kitb
al-dh f ilal an-nahw).
Arabic Outside the Arab World. (= Indian Journal of Applied Linguistics,
20). (Co-editor). New Delhi: Bahri Publications.
1996
Arab Grammarians: Ab l-Aswad ad-Dual, 3 i-ii; Ab H ayyn, 4
i-5 ii; Ab Ubayd, 5 ii; Ab Ubayda, 6 i; al-Axfa al-Asar, 11 ii;
al-Axfa al-Awsat, 11 ii-12, i; al-Asma, 49 ii-50 i; al-Astarbd, 50
i-51 i; al-Azhar, 59 i-ii; al-Batalys, 70 ii71 i; al-Frb, 288 ii-290
i; al-Fris, 290 i-ii; al-Farr, 291 i-292 i; al-Frzbd, 300 i-ii;
al-Jarm, 326 ii-327 i; al-Jawhar, 330 iii; al-Jurjn, 380 i-381 ii;
Xalaf al-Ahmar, 384 i-ii; Xall, 385 i386 i; Ibn Ab Ishq, 438 i; Ibn
jurrm, 438 i-ii; Ibn al-Anbr (Ab l-Bakr), 438 ii; Ibn al-Anbr
(Ab l-Barakt), 438 ii-439 ii; Ibn Aql, 439 i-440 i; Ibn Durayd, 441
i-ii; Ibn Fris, 442 ii-443 i; Ibn Jinn, 445 ii-447 ii; Ibn al-H jib, 447
i-ii; Ibn Xlawayh, 447 ii-448 i; Ibn Him, 448 i-ii; Ibn Kaysn, 448
ii-449 i; Ibn Mad, 449 ii-450 ii; Ibn Mlik, 450 ii-451 i; Ibn Manzr,
451 i-ii; Ibn Qutayba, 452 i-ii; Ibn as-Sarrj, 452 ii-453 i; Ibn Sdah,
453 ii-454 i; Ibn as-Sikkt, 454 i-ii; Ibn uqayr, 454 ii-455 i; Ibn Usfr,
455 ii-456 i; Ibn Walld, 456 i-ii; Ibn Ya, 456 i-ii; Ixwn as-Saf,
458 i-ii; s ibn Umar at-Taqaf, 461 ii-462 i; al-Kis, al-Mzin, 617
ii- 618 i; al-Mubarrad, 651 ii-657 ii; an-Nahhs, 666 ii- 667 i; al-Ql,
766 i-ii; Qutrub, 767 ii- 768 i; ar-Rummn, 808 i-ii; Sbawayhi, 856
i-859 i; as-Srf, 863 i-864 i; as-Suyt, 895 ii-896 i; at-Talib, 904 i-ii;
Talab, 905 i-ii; Ukbar, 946 i-ii; Yahy ibn Ad, 1028 i-ii; az-Zabd,
1037 i-ii; az-Zajjj, 1038 i-ii; az-Zajjj, 1038 ii-1039 ii; az-Zamaxar,
1039 ii-1040 i; az-Zubayd, 1045 i-ii. Lexicon Grammaticorum. Edited
by Harro Stammerjohann. Tbingen: Niemeyer.
Jahilia. Het woordenboek van De duivelsverzen. Edited by Maarten
Asscher [et al.]. Amsterdam: Contact. 5253.
Linguistic Attitudes and the Origin of Speech in the Arab World.
Understanding Arabic: Essays in contemporary Arabic linguistics in
bibliography kees versteegh xxv
honor of El-Said Badawi. Edited by Alaa Elgibali. Cairo: American

University in Cairo Press. 1531.
The Linguistic Introduction to Rzs Tafsr. Studies in Near Eastern
Languages and Literatures. Memorial volume of Karel Petrek. Edited
by Petr Zemnek. Prague: Academy of Sciences of the Czech Repub-
lic. 589603.
Western Studies on the History of Arabic Grammar, 19691994. Pro-
ceedings of the Colloquium on Arabic Linguistics, Bucharest August 29
September 2. 1994. Edited by Nadia Anghelescu and Andrei Avram.
Bucharest: University of Bucharest. I, 927.
Reviews
Fernndez, Mauro. 1993. Diglossia: A comprehensive bibliography
19601990, Amsterdam: J. Benjamins. Word 47, 288291.
Peled, Yishai. 1992. Conditional Structures in Classical Arabic. Wies-
baden: O. Harrassowitz. Bibliotheca Orientalis 53, 557560.
Wagner, Daniel A. 1993. Literacy, Culture, and Development: Becoming
literate in Morocco. Cambridge: Cambridge University Press. Interna-
tional Journal of the Sociology of Language 119, 129134.
1997
Bedouin and Townspeople: Ibn Khaldns view on the development of
Arabic. The Diversity of Development: Essays in honour of Jan Klein-
penning. Edited by Ton van Naerssen, Marcel Rutten and Annelies
Zoomers. Assen: Van Gorcum. 332340.
Landmarks in Linguistic Thought, III: The Arabic linguistic tradition.
London: Routledge.
Sarf. The Encyclopaedia of Islam: New edition. Edited by C. E. Bosworth,
E. van Donzel, W. P. Heinrichs and G. Lecomte, assisted by P. J. Bear-
man and S. Nurit. Leiden: E. J. Brill. IX, 5354.
al-Shaybn, Ab Amr Ishk b. Mirr. The Encyclopaedia of Islam: New
edition. Edited by C. E. Bosworth, E. van Donzel, W. P. Heinrichs
and G. Lecomte, assisted by P. J. Bearman and S. Nurit. Leiden:
E. J. Brill. IX, 394395.
The Arabic Language. Edinburgh: Edinburgh University Press. [2nd
revised edition 2001] [Arabic translation by Muhammad a-arqw:
Al-lua al-arabiyya. Cairo: Al-Majlis al-al li-t -ta qfa, 2003].
The Arabic Tradition. The Emergence of Semantics in Four Linguistic
Traditions: Hebrew, Sanskrit, Greek, Arabic. Edited by Wout van
xxvi bibliography kees versteegh
Bekkum, Jan Houben, Ineke Sluiter and Kees Versteegh. Amsterdam:

J. Benjamins. 225284.
The Emergence of Semantics in Four Linguistic Traditions: Hebrew, San-
skrit, Greek, Arabic. (with Wout van Bekkum, Jan Houben and Ineke
Sluiter). Amsterdam: J. Benjamins.
Wetenschap in de Islamitische samenleving. In het huis van de Islam:
Geografie, geschiedenis, geloofsleer, cultuur, economie, politiek. Edited
by Henk Driessen. Nijmegen: SUN. 278295.
Reviews
Bohas, Georges (Ed.). 1993. Dveloppements rcents en linguistique
arabe et smitique. Damas: Institut Franais de Damas. Bibliotheca
Dialectologia Arabica: A collection of articles in honour of the sixtieth
birthday of Professor Heikki Palva. Helsinki: Finnish Oriental Society.
1995. Bibliotheca Orientalis 54, 800802.
Durand, Olivier. 1995. Introduzione ai dialetti arabi. Milan: Centro Studi
Camito-Semitici. Bibliotheca Orientalis 54, 799800.
Gille, Christiane. 1995. Das Kapitel al-Mausl (Das Relativum) aus dem
Manhaj as-slik des Grammatikers Ab H ayn al-arnt (12561344).
Hildesheim: G. Olms. Bibliotheca Orientalis 54, 793795.
Khan, Mohammad-Nauman. 1994. Die exegetischen Teile des Kitb al-
Ayn: Zur ltesten philologischen Koranexegese. Berlin: K. Schwarz.
Watson, Janet C. E. 1993. A Syntax of Sann Arabic. Wiesbaden:
O. Harrassowitz. Bibliotheca Orientalis 54, 795799.
1998
Reviews
Endress, Gerhard and Dimitri Gutas. 1994, 1995. A Greek and Arabic
Lexicon: Materials for a dictionary of the mediaeval translations from
Greek into Arabic, fasc. 2 and 3. Leiden: E. J. Brill. Journal of the
American Oriental Society, 118, 108109.
Ingham, Bruce. 1994. Najdi Arabic: Central Arabian. Amsterdam:
J. Benjamins. Bibliotheca Orientalis 55, 294297.
Marais, Jean (Ed.). 1996. Quebecs Aboriginal Languages: History, plan-
ning, development. Clevedon: Multilingual Matters. Applied Linguistics
19, 131135.
bibliography kees versteegh xxvii
1999
Loanwords from Arabic and the Merger of d /d. Compilation and
Creation in adab and lua: Studies in memory of Naphtali Kinberg
(19481997). Edited by Albert Arazi, Joseph Sadan and David
J. Wasserstein. Winona Lake, IN: Eisenbrauns. 273286. [= Israel
Oriental Studies 19].
The Dutch/ArabicArabic/Dutch Dictionary Project: Aims and
methods (with Jan Hoogland). Revue de la lexicologie 1415,
1326.
Zayd ibn Als Commentary on the Qurn. Arabic Grammar and Lin-
guistics. Edited by Yasir Suleiman. London: Curzon Press. 929.
Reviews
Rizk, Attia. 1992. Ibn a-ajar. M ttafaqa lafzuh wa-xtalafa manhu.
Stuttgart: F. Steiner. Der Islam 76.184186.
2000
Ed de Moor (19362000). Sharqiyyt 12, iii-vi.
Editors Foreword. (With Sylvain Auroux, Konrad Koerner and Hans-
Josef Niederehe). History of the Language Sciences / Geschichte der
Sprachwissenschaften / Histoire des sciences du langage. I. Berlin:
W. de Gruyter. Volume I, HSK 18.1. XXVXXXV.
History of the Language Sciences: An international handbook on the
evolution of the study of language from the beginning to the present =
Geschichte der Sprachwissenschaften: Ein internationales Handbuch zur
Entwicklung der Sprachforschung von den Anfngen bis zur Gegenwart
= Histoire des sciences du langage: Manuel international sur lvolution
de ltude du langage des origines nos jours, I. Editor (with Sylvain
Auroux, Konrad Koerner and Hans-Josef Niederehe). Berlin: W. de
Gruyter. Volume I, HSK 18:1.
Language and logic in the Arabic grammatical tradition. History of the
Language Sciences / Geschichte der Sprachwissenschaften / Histoire des
sciences du langage, I. Edited by Sylvain Auroux, Konrad Koerner,
Hans-Josef Niederehe and Kees Versteegh. Berlin: W. de Gruyter.
Volume I, HSK 18.1. 300306.
Tasrf. The Encyclopaedia of Islam: New edition. Edited by P. J. Bearman,
Th. Banquis, C. E. Bosworth, E. van Donzel and W. P. Heinrichs.
Leiden: E. J. Brill. X. 360361.
xxviii bibliography kees versteegh
Reviews
Enk, Gerrit J. van, and Lourens de Vries. The Korowai of Irian Jaya:
Their language in its cultural context. Oxford: Oxford University Press.
Word 51, 134137.
Lallot, Jean. Apollonius Dyscole, De la construction (syntaxe). I. Intro-
duction, texte et traduction. II. Notes et index. Paris: J. Vrin. Word
51, 304308.
2001
Arabic in Europe: From language of science to language of minority.
Lingua e stile 36, 335346.
Arabic in Madagascar. Bulletin of the School of Oriental and African
Studies 64, 177187.
History of the Language Sciences / Geschichte der Sprachwissenschaften /
Histoire des sciences du langage, II. Editor (with Sylvain Auroux,
Konrad Koerner and Hans-Josef Niederehe). Berlin W. de Gruyter.
Volume II, HSK 18:2.
Linguistic Contacts between Arabic and other Languages. Arabica 48,
470508.
Naphtali Kinberg. Studies in the Linguistic Structure of Classical Arabic.
Editor (with Leah Kinberg). Leiden: E. J. Brill.
Reviews
Kropp Dakubu, M. E. Korle meets the sea: A sociolinguistic history of
Accra. Oxford: Oxford University Press. Word 52, 163166.
Owens, Jonathan. 1998. Neighborhood and Ancestry: Variation in the
spoken Arabic of Maiduguri, Nigeria, Amsterdam: J. Benjamins. Journal
of African Languages and Linguistics 22, 101105.
2002
Dead or Alive? The Status of the Standard Language. Bilingualism in
Ancient Society: Language contact and the written word. Edited by
J.N. Adams, Mark Janse and Simon Swain. Oxford: Oxford University
Press. 5274.
al-Zadjdjdj, Ab Ishk Ibrhm b. al-Sar. The Encyclopaedia of Islam:
New edition. Edited by P. J. Bearman, Th. Banquis, C. E. Bosworth,
E. van Donzel and W. P. Heinrichs. Leiden: E. J. Brill. XI, 377378.
al-Zadjdjdj, Abu l-K sim Abd al-Rah mn b. Ish k. The Encyclo-
paedia of Islam: New edition. Edited by P. J. Bearman, Th. Banquis,
bibliography kees versteegh xxix
C. E. Bosworth, E. van Donzel and W. P. Heinrichs. Leiden: E. J.

Brill. XI, 378379.
al-Zamakhshar, Abu l-K sim Mahmd b. Umar. The Encyclopaedia
of Islam: New edition. Edited by P. J. Bearman, Th. Banquis, C. E.
Bosworth, E. van Donzel and W. P. Heinrichs. Leiden: E. J. Brill. XI,
432434.
Translations
Al Jahiz: Traktaat over de lof van de zwarten boven de blanken. De
taal der engelen: 1250 jaar klassiek Arabisch proza. Edited by Arnoud
Vrolijk. Amsterdam: Contact. 135139.
Hunayn ibn Ishaq: De behandeling van staar. De taal der engelen: 1250
jaar klassiek Arabisch proza. Edited by Arnoud Vrolijk. Amsterdam:
Contact. 279282.
Reviews
Abu-Manga, Al-Amin. 1999. Hausa in the Sudan: Process of adaptation
to Arabic. Cologne: R. Kppe. Word 53, 292296.
2003
Arab tradition. International Encyclopedia of Linguistics (2nd ed.). Edited
by William J. Frawley. Oxford: Oxford University Press. 202204.
De Arabische taalkunde in Nederland. Sharqiyyt 15, 2942.
The Arabic component of the Indonesian lexicon. Rintisan kajian
leksikologi dan leksikografi. Edited by Lilie Suratminto and Muna-
war Holil. Jakarta: Fakultas Ilmu Pengetahuan Budaya Universitas
Indonesia. 216229.
Woordenboek Arabisch-Nederlands. Amsterdam: Bulaaq. Editor (with
Jan Hoogland and Manfred Woidich).
Woordenboek Nederlands-Arabisch. Amsterdam: Bulaaq. Editor (with
Jan Hoogland and Manfred Woidich).
2004
Approaches to Arabic Dialects: A collection of articles presented to Manfred
Woidich on the occasion of his sixtieth birthday. Editor (with Martine
Haak and Rudolf de Jong). Leiden: E. J. Brill.
From Classical Arabic to the Modern Vernaculars. Morphologie: Ein
internationales Handbuch zur Flexion und Wortbildung = Morphology:
an international handbook on inflection and word-formation. Edited
xxx bibliography kees versteegh
by Geert Boo, Christian Lehmann, Joachim Mugdan and Stavros

Skopeteas in collaboration with Wolfgang Kesselheim. Berlin: W. de
Gruyter. II, 17401754.
Grammar and Lexicography.. Encyclopaedia of Islam and the Muslim
World. Editor in chief: Richard C. Martin. USA: MacMillan Refer-
ence USA. I, 279281.
History of the Arabic Language. Encyclopaedia of Islam and the Muslim
World. Editor in chief: Richard C. Martin. USA: Macmillan Refer-
ence USA. 5862.
Meanings of Speech: The category of sentential mood in Arabic gram-
mar. Le voyage et la langue: Mlanges en lhonneur dAnouar Louca et
dAndr Roman. Edited by Joseph Dichy and Hassan Hamz. Damas-
cus: Institut Franais du Proche-Orient. 269287.
One Letter does not make a Word: The doubly weak imperatives in
Arabic. Words, Texts and Concepts Cruising the Mediterranean Sea:
Studies on the sources, contents and influences of Islamic philosophy
and Arabic philosophy and science dedicated to Gerhard Endress on his
sixty-fifth birthday. Edited by R. Arnzen and J. Thielmann. Leuven:
Peeters. 131141.
Phonological Constraints in Arabic Grammatical Theory: The Iltiq
as-skinayn. Romano-Arabica 3, 221235.
Pidginization and Creolization Revisited: The case of Arabic. Approaches
to Arabic Dialects: A collection of articles presented to Manfred Woidich
on the occasion of his sixtieth birthday. Edited by Martine Haak,
Rudolph de Jong and Kees Versteegh. Leiden: E. J. Brill. 343357.
Taalkunde en letterkunde in de kleine letteren. Gij letterdames en letter-
heren: Nieuwe mogelijkheden voor taalkundig en letterkundig onderzoek
in Nederland. Edited by Joost R. Swanborn. Amsterdam: Koninklke
Nederlandse Akademie van Wetenschappen. 189203.
The Future of Creolistics. Linguistics Today: Facing a greater chal-
lenge. Edited by Piet van Sterkenburg. Amsterdam: J. Benjamins.
305319.
Whats Where and Hows What? Interrogatives in Arabic dialects.
Estudios de dialectologa norteafricana y andalus 8, 239251.
2005
Breaking the Rules Without Wanting to: Hypercorrection in Middle
Arabic texts. Investigating Arabic: Current parameters in analysis and
learning. Edited by Alaa Elgibali. Leiden: E. J. Brill. 318.
bibliography kees versteegh xxxi
Reviews
Talmon, Rafael. 2003. Eighth-century Iraqi Grammar: A critical explora-
tion of pre-H allian Arabic Linguistics. Winona Lake, IN: Eisenbrauns.
Jerusalem Studies in Arabic and Islam 30, 528535.
2006
A new treatise about the ilal an-nah w : Ibn al-Warrq on inna
wa-axawtuh. Grammar as a Window onto Arabic Humanism: A
collection of articles in honour of Michael G. Carter. Edited by Lutz
Edzard and Janet Watson. Wiesbaden: O. Harrassowitz. 5165.
Arabic Linguistic Tradition. Encyclopedia of Language and Linguistics:
Second edition. Editor in chief: Keith Brown. Oxford: Elsevier. I,
434440.
D d. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J.
Brill. I, A-Ed. 544545.
Encyclopedia of Arabic Language and Linguistics. Leiden: E. J. Brill.
General editor (with Mushira Eid, Alaa Elgibali, Manfred Woidich
and Andrzej Zaborski). I, A-Ed.
Introduction. (With Mushira Eid, Alaa Elgibal, Manfred Woidich and
Andrzej Zaborski. Eds.). Encyclopedia of Arabic Language and Lin-
guistics. Leiden: E. J. Brill. I, A-Ed. v-x.
History of Arabic language teaching. Handbook for Arabic Language
Professionals in the 21st Century. Edited by Kassem M. Wahba, Zeinab
A. Taha and Liz England. Mahwah, NJ: Lawrence Erlbaum. 312.
History of the Language Sciences / Geschichte der Sprachwissenschaften /
Histoire des sciences du langage. Editor (with Sylvain Auroux, Konrad
Koerner and Hans-Josef Niederehe). Berlin: W. de Gruyter. Volume
III, HSK 18.3.
Leerboek modern geschreven Arabisch. Grammatica, woorden, idioom
en oefeningen. (with Everhard Ditters and Lieke de Jong). Nijmegen:
Radboud Universiteit. 1, 2, 3.
The Study of the non-Western Linguistic Traditions. History of the
Language Sciences / Geschichte der Sprachwissenschaften / Histoire
des sciences du langage. Edited by Sylvain Auroux, E. F. K. Koerner,
Hans-Josef Niederehe and Kees Versteegh. Berlin: W. de Gruyter.
Volume III, 18.3. 27912802.
xxxii bibliography kees versteegh
Reviews
Crowley, Terry. 2004. Bislama Reference Grammar. Honolulu. University
of Hawaii Press. Bdragen tot de taal-, land- en volkenkunde 162,
572575.
2007
Encyclopedia of Arabic Language and Linguistics. General editor (with
Mushira Eid, Alaa Elgibali, Manfred Woidich and Andrzej Zaborski).
Leiden: E. J. Brill. II: Eg-Lan.
Encyclopedia of Arabic Language and Linguistics. General editor (with
Mushira Eid, Alaa Elgibali, Manfred Woidich and Andrzej Zaborski).
Leiden: E. J. Brill. III: Lan-Ser.
H araka. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J.
Brill. II: Eg-Lan. 232236.
Idfa. (With Karin C. Ryding). Encyclopedia of Arabic Language and
Linguistics. Leiden: E. J. Brill. II: Eg-Lan. 294298.
Idmr. (With Michael C. Carter). Encyclopedia of Arabic Language and
Linguistics. Leiden: E. J. Brill. II: Eg-Lan. 300302.
Il. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J.
Illa. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J. Brill.
II: Eg-Lan. 308311.
Inna wa-axawtuh. Encyclopedia of Arabic Language and Linguistics.
Leiden: E. J. Brill. II: Eg-Lan. 354358.
Interrogative Pronoun. Encyclopedia of Arabic Language and Linguistics.
Leiden: E. J. Brill. II: Eg-Lan. 387389.
Isnd. Encyclopedia of Arabic Language and Linguistics. Leiden: E. J.
HISTORY
INSIDE THE SPEAKERS MIND:
SPEAKERS AWARENESS AS ARBITER OF USAGE
IN ARAB GRAMMATICAL THEORY
Ramzi Baalbaki
American University, Beirut
1. Introduction
The Arab grammatical theory as represented by the later grammar-

ians is generally characterized by its focus on the formal aspects of the
utterance within well-defined, albeit complex and often controversial
rules of usage. Although it is very difficult to represent linearly the shift
which took place during the evolution of the theory towards formal
aspects at the expense of meaning since there have indeed been some
attempts to restore to meaning its primary position in linguistic analy-
sis, it may be safely argued that this shift started immediately after the
first major grammatical work, namely Sbawayhis (d. 180/796) Kitb.
It is, of course, true that at the levels of morpho-phonology and mor-
pho-syntax Sbawayhis formal considerations were adopted almost in
their entirety by subsequent grammarians. But Sbawayhis method of
probing the relationship, at the syntactico-semantic level, between form
and meaning was continuously eroded by the grammarians attempt
to codify rules, systematize usage, and analyze structure largely on the
basis of formal considerations which govern its constituent elements.
Generally speaking, the continuous shift from Sbawayhis method of
syntactico-semantic analysis culminates in works from the seventh
century A.H. onward, such as Alfiyya commentaries and the extensive
sources (mutawwalt), where pedantic formulae and rigid rules almost
fully replace the vivid and dynamic nature of Sbawayhis analysis which
takes into account both the formal and the semantic aspects of citations
and utterances.1
1
In his Muqaddima (10811084), Ibn Xaldn (d. 808/1406) praises Sbawayhis
Kitb on the grounds that its author did not confine it to the formal rules related to
irb and that it is replete with proverbs and citations from poetry and speech. Those
who study the Kitb are therefore likely to enhance their malaka (natural linguis-
4 ramzi baalbaki
One of the most significant aspects of Sbawayhis method of gram-

matical analysis is the pragmatic role he assigns to the speaker (mutakal-
lim), and by extension to the listener (muxtab), as part of the social
interaction which language represents for him. This aspect which lies
at the core of Sbawayhis understanding of language as a form of social
behavior and which embodies his originality, and perhaps the essence
of his value in the history of linguistic ideas, has unfortunately been
severely diminished, if not totally annulled, by later authors. Although
this development is obvious in the works which immediately follow the
Kitbsuch as Mubarrads (d.285/898) Muqtadab and Ibn as-Sarrjs
(d. 316/929) Uslthe shift of focus in linguistic analysis from social
interaction and context of situation to formal considerations becomes
more dramatic in later stages of the history of grammar. Equally unfor-
tunate is that modern scholarship has only recently paid any meaningful
attention to the role which Sbawayhi assigns to the speaker, the lis-
tener, and the context in which speech takes place. Carter (2004, 5657)
observes that one of the most striking features of Sbawayhis analysis
is that it concerns itself almost exclusively with language as behavior:
speech is a set of actions, each named according to its intention, e.g.
istifhm asking a question, tatniya making something dual, tanbh
drawing attention to something. He further observes (p. 57) that every
utterance takes place in a context of a speaker . . . and listener and that
this approach places great emphasis on the pragmatic roles of speaker,
listener and context and invites the analyst to propose psychological
explanations of linguistic phenomena.2 Bohas et al. (1990, 38) convin-
cingly argues that, from a typological perspective, grammatical and lin-
guistic systems can be divided into two rough classes: on the one hand,
those which analyze utterances in terms of formal relationships between
their components; on the other hand, those which analyze them in terms
tic ability), although some of them end up mastering grammar as a sina (craft), but
not as a malaka. Contrarily, the books of the later authors (kutub al-mutaaxxirn) are
void of poetry and the speech of the Arabs, and contain nothing but grammatical rules
(al-qawnn an-nahwiyya). Readers of such works, according to Ibn Xaldn, can hardly
be expected to enhance their malaka and can only master the craft. See also Zakariyy
(1986, 23 ff.).
2
See also Carter (2004, 9598) for further discussion of the speakers role and its
significance to Sbawayhis reasoning. It would be particularly interesting to examine
in more detail the effect of the speakers choice (96) and the speakers intention (97) on
utterances in Sbawayhis analysis of speech.
speakers awareness as arbiter of usage 5
of operations performed by the speaker in order to achieve a specific

effect on the allocutee. Based on this distinction, they maintain that
Sbawayhis approach basically belongs to the latter category, while the
classical grammarians, typically belongs to the former.
It must be stressed that what we shall call the linguistic awareness of
the speaker is, according to Sbawayhi, an essential component of the
competence which is demonstrated in successful interaction with the
listener. By linguistic awareness we refer to what he and a few other
authors perceive as the speakers alertness to the various tools which
language places at his disposal and his ability to use them to decide what
form best expresses the meaning which he intends to convey to the lis-
tener. Within this context, the purpose of the present paper is twofold.
First, we shall try to show that although Sbawayhis method, which
assigns a central role to the meaning and to the speakers awareness, was
generally not followed by subsequent authors, some of them have indeed
attempted to restore that central role, albeit from different perspectives.
In a cursory look at the basic elements of the most important of these
attempts, it may be possible to establish that what they have in common
with Sbawayhi is a genuine concern for meaning and for the role of the
speaker. The latter part of the paper looks into how Sbawayhis assump-
tion of the speakers awareness is practically applied in his analysis of
particles whose etymology is essential in determining usage. For this,
we shall examine his analysis of those compound particles which may
be used as single words or split into their constituent elements.
2. Meaning and speakers awareness
The relationship between meaning and the speakers awareness in

Sbawayhis analysis cannot be overemphasized. As a general rule,
whenever Sbawayhi highlights the role of the speaker in the utter-
ance or his competence in producing correct speech, the reader should
expect meaning to be at the center of the authors argument. The Kitb
abounds with examples which support this conclusion, and it is cer-
tainly beyond the scope of this paper to investigate the larger number
of formal, semantic and contextual elements which contribute to the
link between meaning and the speakers awareness in Sbawayhis analy-
sis. This notwithstanding, it is important to establish the link between
meaning and one of the most basic concepts in the Kitb, namely taqdr
6 ramzi baalbaki
(suppletive insertion of elided elements), mainly because, as a gram-

matical tool, taqdr embraces the formal, semantic and contextual ele-
ments that are at the core of correct speech according to Sbawayhi.
Another compelling reason for establishing this link is that it can help
us understand the sharp contrast between Sbawayhis approach and
that of the later grammarians, for whereas they have largely adopted the
formal aspects of his taqdr, they have, more often than not, ignored the
semantic and contextual dimensions of the process of supplying missing
elements to structure. We shall not, however, make any detailed com-
parison between Sbawayhi and the later grammarians in this respect as
this would require an independent study. Rather, we shall point out two
of the most illuminating principles of the relationship between meaning
and taqdr in Sbawayhis system of grammatical analysis.3
The first principle is that in the process of taqdr, the proposed con-
struction should not contradict the meaning of the original construc-
tion, i.e. before the suppletion of the elements which are judged to be
elided. In an earlier study of the harmony which Sbawayhi tries to estab-
lish, through taqdr, in several types of constructions (Baalbaki 1979,
714), I discussed his method of breaking up one sentence into two,
both of which share a common feature and thus demonstrate an under-
lying harmony. For example, the two sentences Zaydan darabtuhu and
a-Abdullhi daraba axhu Zaydan are interpreted, at the level of deep
structure, as *darabtu Zaydan darabtuhu and *a-daraba Abdullhi
daraba axhu Zaydan respectively (Kitb I:81, 102). If a nominal sen-
tence is conjoined to a verbal sentence, as in raaytu Zaydan wa-Amran
kallamtuhu, he intervenes to restore the harmony by supplying a verb to
the nominal sentence, hence the proposed construction *raaytu Zay-
dan wa-kallamtu Amran kallamtuhu (Kitb I:88). In defending the res-
toration of a verb to produce a verbal sentence parallel to the first one,
he argues that the introduction of the verb causes no contradiction in
meaning (l yanqud man; Kitb I:8889). Closely related to this argu-
ment is his assertion that in utterances which express amr (command)
or nahy (prohibition), the verb may be uniformly elided, as in (idrib)
Zaydan, (l taqrabi) l-asada, and (xalli) t-tarqa (Kitb I:253254).
Sbawayhis discussion of such constructions reveals that the element of
3
Carter (1991, 127128) notes that although Sbawayhi uses the term taqdr exceed-
ingly sparingly, only 24 times in fact (see Troupeau 1976, 167), he does give plenty of
advice on reconstruction without calling it taqdr.
meaning which justifies the elision of the verb is present in the context
of situation since the listener would assume the virtual existence of a
verb of which he is the agent. Consequently, the grammarians interven-
tion to restore this verb would only elucidate, rather than contradict
meaning, based on his understanding of what the speaker has in mind.
The second principle which governs the relationship between taqdr
and meaning represents a further stage in Sbawayhis analysis of con-
structions. While he remains faithful to the rule that taqdr should not
contradict meaning, he explores the niyya4 (intention) of the speaker to
explain why he may well utter a certain part of the construction although
he has another usage in mind. An example of this is his belief that each
of the two constructions lam tika and l tka has the status (manzila)
of a noun at the level of the speakers niyya, and can thus be interpreted
as lam yakun itynun (Kitb III:2829). The grammatical implication of
this niyya is only fully revealed when the uttered part is virtually replaced
by what the speaker intended to say. Thus, in Farazdaqs line: mamu
lays muslih na aratan * wa-l nibin ill bi-baynin urbuh,
nibin is in the genitive although at the level of the actual utterance it
is conjoined to muslih na, which, being the predicate of laysa, is in the
accusative. According to Sbawayhi, the recurrent use of the preposi-
tion bi- with the predicate of laysa (e.g., lays bi-muslih na) reveals the
true intention of the speaker, and hence nibin is in reality conjoined
to a genitive noun which does not feature in speech, but is as valid as
an uttered noun. In other words, the preposition, which the speaker
has in mind, is syntactically valid and operational, and it determines
usage as if it were actually uttered (h att kaannahum qad takallam
bih f l-awwal; Kitb III:29). As far as meaning is concerned, Sbawayhi
asserts that the assumption of the preposition bi- in bi-muslih na does
not alter the meaning5 (l yuayyir al-man) because the preposition is
indeed frequently used with the predicate of laysa. Similarly, Sbawayhi
intervenes in constructions of the type marartu bihi fa-id lahu sawtun
sawta h imrin/surxun surxa t-takl to supply a verb (i.e. yusawwitu,
4
According to Troupeau (1976, 208), the terms naw and niyya occur 13 and 27
times respectively in the Kitb. The concept of intention, however, is often expressed
by much more frequent termsincluding an (136 times), man (891 times), arda
(1361 times), etc. (Troupeau 1976, 150; 102)or by expressions such as kaannahum
ql, tawahham, ixtra, etc.
5
The significance of preserving the meaning of the construction in this line as well
as the meaning of other constructions within the context of Sbawayhis analysis of the
taqdr of an after f is discussed by Baalbaki (2001, 186209, esp. 188).
8 ramzi baalbaki
yasruxu) before the accusative noun (Kitb I:355f.). The intention of

the speaker is expressed here by the term tawahhama,6 and the actually
uttered words lahu sawtun are said to have the same status as yusawwitu.
Syntactically, the outcome of taqdr is the use of the accusative in sawta
and surxa, whereas at the semantic level, it is implicit that the introduc-
tion of yusawwitu and yasruxu does not contradict the meaning because
these verbs share the same root with the nouns they govern.
The study of the relationship between meaning and taqdr in the Kitb
strongly indicates that in his analysis of constructions, Sbawayhi tran-
scends the levels of grammatical correctness and the effect of the oper-
ants on case-endings to examine the speakers thinking and the mental
processes involved in the choices he makes. This linguistic awareness on
the part of the speaker becomes a real arbiter of usage and allows him,
for example, to use nibin where niban is expected, or to use sawta
h imrin in spite of the absence of a verb from the utterance. It is this
feature in Sbawayhis analysis that one so much misses in the work of
the later grammarians. In spite of that, some authors were indeed inter-
ested in the role of the speaker and the effect of his awareness on the
speech he produces. From this perspective we shall briefly examine the
contribution of two leading figures, Ibn Jinn (d. 392/1002) and Jurjn
(d. 471/1098), and show how their priorities are largely consistent with
those of Sbawayhis. In the light of the speakers awareness and his inter-
nal thinking, we shall then examine the analysis proposed by Sbawayhi
for constructions in which the etymology and word-class of certain par-
ticles are crucial for the speaker to achieve correct speech and for the
listener to comprehend what is meant.
3. Post-Sbawayhi authors
The first post-Sbawayhi author whose work reflects serious concern for
the speakers awareness is certainly Ibn Jinn. Groomed in the gram-
matical tradition and himself author of several works which are in full
conformity with the general grammatical theorymost notably Sirr
sinat al-irb, al-Luma f l-Arabiyya, and at-Tasrf al-mulkIbn
6
For the various senses in which the term tawahhum is used in the Kitb, see Baal-
baki (1982, 234237). In this particular case of Farazdaqs line, tawahhum refers to the
speakers mental restoration of elided parts in the utterance, resulting in their govern-
ment of parts actually uttered.
Jinn stands out as a unique scholar whose innovative approach in his

most impressive work, al-Xasis, represents a major step in the direc-
tion of determining the theoretical bases of language usage. The tradi-
tional phonetic, morphologic and syntactic data becomes in al-Xasis
the subject of study from a methodological and epistemological per-
spective. Within this framework, the speaker takes a central role in
Ibn Jinns interpretation of linguistic activity. His point of departure
for this is a firm belief in the intuitiveness7 of native speakers of Arabic
as well as in their mental abilities, which are manifest in their produc-
tion of speech and subtle analysis of a host of linguistic phenomena.
A particularly illuminating example is that in which he tries to exam-
ine the linguistic awareness of a Tamm informer of his. When asked
why he would say darabtu axka, and not *darabtu axka, but would
still use axka in daraban axka, the Tamm expresses his astonish-
ment and wittily comments that each of the two expressions has a dif-
ferent perspective (ixtalafat jihat l-kalm; Xasis I:76; cp. 1:250). Ibn
Jinn confidently concludes that the Arab native speakers scrutinize the
syntactic positions of speech elements (adall ay al taammulihim
mawqi al-kalm) and that they knowingly and consciously (an mza
wa-al basra) assign to each element the position and case-ending it
merits. Similarly, he describes vowel mutation which results from the
speakers sense of lightness (istixff ) and heaviness (istitql ) as proof
of keen insight and fineness of perception and observation (li-quwwat
nazarihim wa-lutf istiffihim wa-tasaffuh ihim; I:78).
Ibn Jinns interest in the intuitiveness of Arab native speakers and
in their mental awareness of the various phonetic, morphologic, and
syntactic processes involved in speech is best understood as part of his
attempt to reveal the h ikma which underlies Arabic and which consti-
tutes its intellectual basis. Whether the discussion relates to the reasons
(ilal ), purposes (ard) and intention (qasd ) associated with speech
(I:237, 245), or to the change which forms, etc. undergo due to recur-
rent usage (II:31), or to the temporal precedence of one part of speech
7
Ibn Jinn often expresses the notion of intuitiveness by derivatives of the root t-b-
(e.g., sun al-br subh nahu f an tabaa n-ns al hd; tahjum bihim tibuhum al
m yantiqna bihi; a-turhu l yuh sin bi-tabihi . . . hd l-qadr; Xasis II:117; III:273,
275 respectively). Cf. also the terms salqiyya and najr (I:76) for intuitiveness, and the
expressions min lutf al-h iss wa-safihi wa-nasat jawhar al-fikr wa-naqihi; quwwat
nafsihi wa-lutf h issihi; I:239; III:75). See also the comments of Suleiman (1999, 6465)
on the intuition of native speakers and the rationality of Arabic within the more general
framework of Ibn Jinn s study of tall (causation).
10 ramzi baalbaki
over others (II:33), or to the onomatopoeic nature of phonemes within

words (II:162, 164), or to the occurrence of two dialects side by side in
speech (III:317), the term h ikma is used as an inalienable trait of the
original wdi (creator) of Arabic and of his logic in deciding on what is
to be used or not and how it should be used. It can thus be argued that
the abstract notion of the wdi finds its practical dimension in the role
of the speaker (or, more generally, the speech community) since Ibn
Jinn ascribes to him an awareness of the linguistic processes that are
the result of the original h ikma. In this respect, Ibn Jinns ideas are very
much in line with Sbawayhis not only because both of them place the
speaker at the center of their linguistic analysis, but also due to their
mutual interest in the communication of meaning as the ultimate aim
of successful speech.
One of the most striking results of Ibn Jinns focus on the speakers
role and of his firm belief that, at all levels of analysis, the linguistic phe-
nomena of Arabic are essentially rational is his discussion of grammati-
cal awmil or operants and their relatedness to meaning. Although he
adopts the traditional division of these operants into two types, lafz (for-
mal; expressed) and manaw (abstract), his interpretation of the awmil
in the light of the speakers role redresses the imbalance between the
two types in the tradition. For example, Jurjn in his al-awmil al-mia
n-nahwiyya (8586; 312), labels ninety-eight of the awmil as lafz and
only two as manaw.8 This imbalance, of course, predates Jurjn, and
it is interesting to note that Ibn Jinns own master, Ab Al al-Fris
(d. 377/987), is reported to have authored a book entitled al-awmil
al-mia (Sezgin 1984, 107). By insisting that all types of amal (rection)
i.e., raf , nasb, jarr and jazm, which cover the three nominal types
(nominative, accusative and genitive) and the three verbal types (indica-
tive, subjunctive and jussive)are in reality ( f l-h aqqa) produced by
the speaker (Xasis I:109110), Ibn Jinn effectively reduces the tradi-
tional divide between lafz and manaw operants merely into a didactic
technique that tries to distinguish between amal which is accompa-
nied by an uttered operant and amal which lacks such accompaniment
(I:109). In other words, whether an operant is actually uttered or not,
rection is in all cases the result of the speakers internal thinking and
8
Note also the assertion of some grammarians that no mil may be classified as
manaw unless it cannot possibly be explained as lafz (l yudal il jal al-mil
manawiyyan ill inda taadd
u
r al-laf as-slih ; Suyt, Ham I:159).
a reflection of the meaning he intends. Based on this and on the simi-

larly untraditional view that qiys lafz (formal analogical extension)
is not devoid of meaning, Ibn Jinn confidently formulates the general
principle that one cannot but bestow a manaw dimension on what is
lafz, whereas what is manaw may well do without a lafz dimension
(I:111).
It may be useful at this point, at the risk of disrupting the historical
sequence, to bring into the discussion the sixth-century Z hirite scholar,
Ibn Mad (d. 592/1196). Ironically, Ibn Jinns attribution of amal to the
speaker is enthusiastically received by Ibn Mad. He quotes Ibn Jinns
statement fa-l-amal . . . innam huwa li-l-mutakallim nafsihi l li-ay
ayrihi and highlights his use of nafsihi as a corroborative to emphasize
al-mutakallim, followed by the assertion that rection is attributable to
nothing other than the speaker (l li-ay ayrihi; Radd, 77). Ibn Mad
quickly recognizes his odd position as a Z hirite embracing a Mutazilite
view, and thus hastens to resolve the situation by explaining that con-
trary to the Mutazilites, the doctrine of ahl-al-h aqq (i.e., the Z hirites)
stipulates that case-endings (here, aswt) are in reality produced by
God (innam hiya min fil Allh tal) but metaphorically attributed
to man. Irrespective, however, of this modification and of Ibn Mads
argument that neither the uttered forms of the operants nor their mean-
ings cause rection, his merciless criticism of the grammarians focus on
amal and taqdr rests in part on their disregard for both the speaker and
the intended meaning. In fact, he often refers to the speakers intention
(cf. yanw, 89; yurd, 93) and assesses the relationship between meaning
and taqdr (80, 109). These two aspects of his theory obviously form his
best defense against traditional grammar and firmly place him, albeit
from the different perspective of his Z hirite doctrine, with the few
authors who challenged its shortcomings.
After Ibn Jinn, the most important attempt to reinstate a primary
role to meaning and the speaker who intends it is undeniably that of
Jurjn. Like his predecessor, Jurjn followed in the footsteps of the tra-
ditional grammarians in some of his works. In addition to al-Awmil
al-mia n-nahwiyya mentioned earlier, such works include al-Jumal,
most of which is a didactic summary of awmil (chapters 2 to 4), and
a commentary on Friss dh entitled al-muqtasid f arh al-dh .
However, Jurjns two major works in the field of stylistics, primarily
Dalil al-ijz but also Asrr al-bala, represent a major shift from
the traditional syntactical analysis of the grammarians. In his Dalil, he
12 ramzi baalbaki
makes a zealous appeal for placing meaning at the center stage of gram-
matical analysis and repeatedly argues that nazmwhose narrow sense
corresponds to word order, but which more generally refers to the com-
plex relations among the constituents of a structureis nothing other
than the proper adherence to the discipline of grammar (laysa n-nazm
ill an tada kalmaka l-wad allad yaqtadhi ilm an-nahw, 64). As such,
nazm should aim at what he calls the meanings of grammar (an-nazm
huwa tawaxx man n-nahw, 276, 282, 310, 403404; cf. Asrr, 65).
This means that syntactical rules, which govern the relationships among
the various parts of any utterance, can express the exact meaning
intended by the speaker since speech formation begins in ones mind ( f
n-nafs) and only then are words arranged to formulate the outcome of
this mental process according to a set of syntactical relationships (43f.).
Based on the conviction that the arrangement of meanings (at-tartb f
l-man) is prior to the arrangement of words, and that form is subsid-
iary to meaning (4445), it follows that any change in syntax is necessar-
ily accompanied by a change in meaning (86f.) The speakers awareness
of the intricacies of syntactic relations hence acts as a virtual arbiter in
his choice of the nazm, which best expresses the intended meaning.
Jurjns theory of nazm is to a large extent a reaction against tradi-
tional grammar in which formal aspects acquired greater prominence at
the expense of meaning. Even Sbawayhi, it has been suggested (Baalbaki
1983, 12f.), may have been the target of some of Jurjns critical com-
ments in which he accuses the grammarians of giving too little attention
to meaning. This notwithstanding, Jurjn is surely much closer in spirit
to Sbawayhi than to the later grammarians. Both authors strive to inves-
tigate the internal thinking of the speaker and examine its influence on
actual utterances. On a wider scale, Sbawayhi, Ibn Jinn and Jurjn, the
three most original authors in the related fields of nahw (grammar), phi-
lology (ilm al-lua), and stylistics (bala) respectively, share the view
that meaning should be the main focus of linguistic analysis. A variety
of conceptssuch as the speakers intuitiveness, competence, intention
and awareness of the tools at his disposalfeature in the works of the
three authors as part of their study of meaning and the mental processes
to which it is related. Unfortunately, however, the three have one more
thing in common, for although their focus on meaning and the speakers
awareness represents the most significant and original aspect of their
contribution, that focus gave way in later writings to an ever-growing
shift towards formal considerations and pedantic formulae which rel-
egate meaning and those concepts that are related to it to a much lower
position in their scale of linguistic analysis.
4. The importance of the particles
The importance which Sbawayhi attaches to the speakers competence

and his alertness to the means of successful communication of mean-
ing feature distinctly in his discussion of some particles.9 More specifi-
cally, Sbawayhi is interested in proving that the speaker can express the
intended meaning only if he is aware of certain peculiarities of these
particles. The rest of this paper shall deal with etymology of particles
as one such peculiarity, the awareness of which by the speaker, accord-
ing to Sbawayhi, is crucial for correct speech. It should be noted here
that the later grammarians largely adopt Sbawayhis analysis of particles
and their usage, but this does not at all mean that they have preserved
his method of syntactical analysis or embraced the psychological and
contextual explanations which he emphasizes so much. Rather, they
seem to dwell on the formal side of his analysis, or simply repeat his
views without incorporating them into their own methods of analysis
or developing them in any meaningful way.
Initially, it is essential to introduce the concept of h ikya, which
Sbawayhi applies to those particles which he discusses from the etymo-
logical perspective. The original sense of the root h -k-y, to report; to
imitate, is preserved in Sbawayhis use of h ikya for direct speech. For
example, in the construction qla Zaydun Amran xayru n-nsi (III:142),
Zayds words are reported verbatim and therefore qla does not govern
Amrun, which remains in the nominative as in the original utterance.
Derived from this sense is the use of the term to refer to elements that
are intentionally not integrated into the syntactical build of the construc-
tion. Commenting on the sentence marartu bi-Zaydin, for example, one
may respond by saying man Zaydin, in the genitive (II:413). This strat-
egy in dialogue can result in the creation of forms that are only used in
9
The term particle is used here in a general sense which includes not only what the
traditional grammarians classify as particles proper (h urf ), but also what they consider
to be verbs or nouns. We shall therefore refer, for example, to amm, interrogative m,
and the halumma as particles although they are traditionally classified as particle, noun,
and verb respectively.
14 ramzi baalbaki
h ikya, as in the use of ayyayni/ayyna or manayni/manna in response

to raaytu rajulayni/rijlan (II:407408). Further removed from the
original sense of the term h ikya is its use to describe proper nouns that
are syntactically whole sentences or phrases and are uttered verbatim
regardless of their grammatical position. These include attested proper
nouns such as taabbata arran (II:269; III:326), baraqa nah ruhu (II:269;
III:326), ba qarnh (II:85; III:207, 326), ibn jal (III:207), and forms
such as xayran minka, driban rajulan and min Zaydin (III:328329),
which are hypothetically proposed as proper nouns and serve as a test-
ing device for the full potential of h ikya. Sbawayhis use of the term
h ikya in connection with proper nouns that are made of more than one
element is most probably what facilitated his generalization of the term
to apply to compound particles, that is, particles which he believes are
etymologically made up of more than one element.
As a preliminary example, we can consider the particle h aytum
which Sbawayhi describes as h ikya. This sense of the term is clearly
distinguished in the Kitb from neighboring concepts, most notably
ism wh id and law. The latter term, law, refers to elements of speech
which are otiose or redundant. The negative particle l, for example,
is considered to be law when it does not signify negation, as in liall
yalama ahlu l-kitbi (Q 57:29), which is interpreted as li-an yalama in
the affirmative (IV:222).10 As for the term ism wh id, Sbawayhi recur-
rently uses it in connection with the place name H adramawta, which he
cites as an example of two distinct nouns that coalesced into one (e.g.,
the four chapter titles in II:267; III:296, 374, 475). Although at times he
refers to particles which exemplify h ikya as ism wh id (II:417418),
h arf wh id (II:418), or kalima wh ida (III:115), Sbawayhi clearly
demarcates h ikya, as in h aytum, from ism wh id in the sense which
H adramawta embodies. He achieves this by looking into the function of
-m in h aytum and similar particles (see below) and comparing it with
the corresponding element -mawta in H adramawta. The difference lies
in the relationship which each of -m and -mawta has with the preceding
element (III:331). The introduction of -m to h aytu-, he observes, does
not result in the retention of the final damma in h aytu- since h aytam,
with a fath a, is also attested. Moreover, and more importantly, the intro-
duction of -m causes a semantic change since to the adverbial sense of
10
Another example of law is m in mahm (interpreted as m- + -m), and after
conditional in, adverbial id, and pronominal ayy (III:5960).
h aytu- is added to the conditional sense which h aytum expresses. Sim-

ilarly, -m is introduced to an- in amm (amm <an- + -m) to cause
a grammatical change, namely the annulment of the rection of an-.
Because of these semantic and grammatical changes, which its intro-
duction brings about, -m in h aytum and amm is called muayyira.
The difference between -m and -mawta becomes clear semantically and
grammatically. Although Sbawayhi does not explicitly describe the dif-
ference, it is implied that h adra- can only have a fath a and not any other
vowel, and that one cannot speak of semantic or grammatical change
caused by -mawta, at least not in the sense which applies to particles.
Furthermore, the term muayyira draws a sharp distinction between
the nature of -m in h ikya and the concept of law which is essentially
antithetical to change.
Based on the above discussion, particles that are classified as h ikya
can be described as typically made up of two constituent elements (e.g.,
h aytum> h aytu- + -m; innam> inna- + -m),11 and the element
which is identified as being introducedeither as a prefix, such as la-
in laalla, or suffix, such as -m in h aytumis not law. But there is
another dimension to this issue since there are particles, such as excep-
tive ill, which may seem to be examples of h ikya according to the
above criteria, but which Sbawayhi interprets as non-compound par-
ticles. Since Sbawayhi does not use a special term for such particles,
we shall refer to them as non-h ikya or non-compound particles. The
differentiation between the two types is crucial for our understanding
of the role which Sbawayhi assigns for the speaker in using certain par-
ticles. Following is an illustrative list of h ikya particles, with the most
essential characteristics which Sbawayhi ascribes to each. These will be
analyzed and then compared to non-h ikya particles.
The h ikya particles12 comprise the following:
11
Only rarely does h ikya involve more than two constituent elements, as in dlika
which Sbawayhi mentions side by side with other h ikya particles that are made up of
two elements (III:332), and laallam in which -m is added to laalla- (IV:221), itself a
compound particle (la- + -alla; III:332).
12
Sbawayhi also uses the term h ikya in connection with personal and demonstra-
tive pronouns (e.g., anta, hd, huli, dka, dlika, etc.; III:332; IV:218), but these are
not relevant to our discussion. Also beyond our scope are particles like lan which may
well be the result of merging two elements (III:5) but which speakers do not normally
recognize as compound particles.
16 ramzi baalbaki
1. innam: h ikya (III:331); h arf wh id (II:418; III:57); laysat -m bi-

law (III:57); -m changes inna- by imparting a new meaning to it
(IV:222; cf. II:138); -m changes (i.e., annuls) the rection of inna-
(IV:222).
2. annam: h ikya (III:331).
3. h aytum: h ikya (III:331); -m does not prevent the final vowel of
h aytu- from being either -u- or -a- (III:331); -m imparts to h aytu-
the meaning of conditional (III:331; cf. III:59, 518; IV:222); -m is
muayyira (III:331; cf. III:59), is unlike -mawta in H adramawta
(III:331), and is not law (III:331); h arf wh id (II:418; cf. III:57);
has the status of in, imm (III:59) and ayna (IV:221).
4. imm: h ikya (III:331); -m is attached (madmma) to in- and
may be elided (III:331332; cf. I:266; III:141; IV:222); has the status
of amm in amm anta muntaliqan intalaqtu maaka (III:332);
see also h aytum.
5. amm (in amm anta, as in 4 above); introduction of -m
prevents an- from governing the subjunctive and -m is hence
muayyira (III:331, 332).
6. halumma: h ikya in both H ijz and Tamm dialects (III:332; cf.
I:252; III:529, 534).
7. idm: -m imparts to id- the meaning of conditional (III:56); has
the status of innam and kaannam (III:57); laysat -m bi-law
(III:57); h arf wh id (III:57).
8. lawm and lawl: h ikya (III:333); -m and -l impart a new mean-
ing to law- (III:115; IV:222); -m is muayyira (IV:222223); h arf
wh id (III:115); kalima wh ida (II:180); as excitative (tah dd) par-
ticles, they precede only verbs (III:115).
9. lamm (which governs the jussive): -m is muayyira (i.e., it changes
the syntactic properties of lam-; IV:223).
10. kam: h arf wh id (III:116).
11. kad: h ikya (III:332; cf. III:151); ay wh id (III:171).
12. kaayyin: h ikya (III:332; cf. II:171; III:151).
13. kaanna: h ikya (III:332; cf. III:151, 164); ay wh id (II:171).
14. kaannam: h ikya (III:331); h arf wh id (II:418; III:57); laysat -m
bi-law (III:57); -m changes the rection of kaanna- (IV:221; cf.
II:138).
15. laalla: h ikya (III:332).
16. laallam: -m changes the rection of laalla- (IV:221; cf. II:138).
17. rubbam: kalima wh ida, like qallam and other similar particles
(abhuhum; III:115); precedes verbs, unlike rubba- (III:116).
18. h abbad: kalima wh ida (II:180).
19. al and am (as interrogative particles): h ikya (III:332).
20. all and hall: h arf wh id (III:5, 115); -l imparts a new meaning to
hal- (IV:222).
21. ill (as a conditional particle): h ikya (III:332).
22. imm (as a conditional particle): h ikya (III:332); see also
h aytum.
23. md and badam: see below.
Sbawayhi spares no effort to demonstrate the coherency of the above

group through cross-references in various and often distant parts of the
Kitb and by using a largely unified terminology to describe the prop-
erties of the particles within the group. Furthermore, this coherency is
supported by the characteristics, which its members share particularly
at the level of meaning (nos. 1, 3, 7, 8, 20), syntax (8, 9, 17) and rec-
tion (1, 5, 14, 16). This notwithstanding, the particles within this group
considerably vary in their clarity, from the perspective of the speaker,
as to whether they are compound in nature or not. For example, one
can safely assume that idm, kad and laallam are much more easily
recognizable by the speaker as compound particles than, say, halumma
or amm which are far removed from their supposed origin and which
can even be the object of disagreement among grammarians concern-
ing what that origin really is (cf. the case of halumma in Suyt , Ham
II:106107). The most revealing examples, as far as speakers aware-
ness is concerned, are those in which the particle may be either bro-
ken down into two elements or used as a single entity. To use the late
grammatical term lamh al-asl13 (lit. recognition of origin), one can say
that the speaker who splits such particles into two elements is aware
of their etymological origin as compound words, unlike the speaker
13
The term lamh al-asl is used by the later grammarians mainly to refer to the
speakers recognition of word class as reflected in usage. One example is that proper
nouns such as H rit and H asan, contrary to the norm, may be prefixed with the definite
article al- since they are originally adjectives. The article is hence called al-allat li-lamh
al-asl (Suyt , Ham I:174175), and the speakers recognition of the adjectival origin
(lamh as-sifa) of such proper nouns justifies its prefixation to them (Ibn Aql, arh
91; Umn, arh I:8586). Due to the obvious similarity between the recognition of an
original word class and the recognition of a particles etymology, the term lamh al-asl is
perfectly applicable to m d, man d, etc. when they are split into two units.
18 ramzi baalbaki
who treats them as single-word particles. Sbawayhi specifically deals

with this issue in a chapter which he designates for md and mand.
Depending on whether d enjoys the independent status of the rela-
tive pronoun allad or is integrated with m into a single unit, which
Sbawayhi typically describes as ism wh id (II:417), the speaker respec-
tively uses either the nominative or the accusative in the noun which
follows md. In Labds line al tasalni l-mara m d yuh wilu *
a-nah bun fa-yuqd am dallun wa-btilu, d is given an independent
status equivalent to allad and m by itself acts as interrogative. The cor-
roborative noun nah bun thus acquires the nominative because it modi-
fies m, which has the grammatical function of the subject of a nominal
sentence (i.e., mubtada). Conversely, md could have been treated as
a single interrogative particle which grammatically serves as the direct
object of yuh wilu. In this case, corroborative nah ban would be, like
md which it modifies, in the accusative.
It is clear in Sbawayhis discussion that the speaker has the choice of
using md as a single-word particle or splitting it into its constituent
elements. The mental process which leads up to the speakers decision is
not restricted to the deconstruction of the particle into its elements or
its retention as one unit. It also has to do with the syntactical ramifica-
tions of the speakers choice since the case-endings of the noun which
follows the particle have to be consistent with that choice.14 The speakers
competence is thus demonstrable at two levels, namely his awareness
of the nature of md as a compound particle which admits two pos-
sibilities of usage, and his confirmation of this awareness by observing
the syntactical implications of each possibility. Furthermore, Sbawayhi
takes into consideration the listeners interpretation of the speakers use
of the particle in order to prove that successful communication also
depends on the listeners awareness of how the speakers choice between
two possibilities of usage affects his own response. The correct response
to m d raayta, for example, would be matun h asanun in the nomi-
native since m itself is mubtada and hence nominative. On the other
hand, md raayta is equivalent to m raayta since the verb governs
14
Note that Sbawayhi does not refer to the fact that, in actual speech, stress may be
an essential factor in differentiating md from m d. As a single word, md would
normally receive stress on its first syllable. Conversely, the separation of the two ele-
ments would be indicated by a stronger stress on d than m, perhaps to underline the
likely demonstrative function of d. In short, the difference in meaning between the two
options can be best demonstrated by translating md faalta and m d faalta as What
have you done? and What is this that you have done? respectively.
the single-word interrogative particle which precedes it in both cases,

and the answer should thus be in the accusative as in md anzala rab-
bukum ql xayran (Q 16:30).15
As part of his discussion of md/m d and mand/man d,
Sbawayhi mentions two other particles which the speaker can use as
a single word or as two separate elements. These are kaannam and
h aytum (II:418; cf. also badam II:139). Sbawayhi often examines the
function of the suffixed -m in these and similar particles (see detailed
list above), including annulment of rection (cf. innam) and shift in
the part of speech which the particle precedes. For instance, rubba and
qalla can only precede nounsas one cannot say *rubba yaqlu and
*qalla yaqlubut the suffixation of -m causes the new particle to
precede verbs only (axlashum li-l-fil; III:115). In such cases, correct
speech is contingent on the speakers reanalysis16 of such compound
particles with respect to their constituents as this would ensure proper
distinction between utterances like qallam yaf alu, innam yaf alu,
kaannam yaf alu, etc. and their counterparts with relative m; i.e.,
qalla m yaf alu, inna m yaf alu, kaanna m yaf alu, etc.17 For his
part, the listener is obviously expected to differentiate between the two
types for successful communication to take place.
As we pointed out earlier, there is a group of particles which may
seem to be examples of h ikya but which Sbawayhi interprets as simple
or non-compound (i.e., non-h ikya) particles. In order to highlight the
contrast between these particles and the h ikya particles, Sbawayhi dis-
cusses both types side by side in one of his chapters (III:331332). It
is important, however, to determine at which level this contrast exists.
By classifying, for example, the exceptive particle ill with non-h ikya
particles, Sbawayhi does not want to specifically deny that it is histori-
cally the result of a merger between in- and -l, in contrast with the
15
Sbawayhi cites the reverse usage, i.e., the accusative after m d and the nomina-
tive after md. But although this is grammatically explicable, he asserts that to use the
nominative after m d and the accusative after md is the most appropriate manner
of response (wajh; aqrab il an taxud bihi; II:418419).
16
Cf. the role of reanalysis in the use of tlam as discussed by Anghelescu (2004,
115116).
17
To illustrate this distinction in the case of innam, for example, we can replace it
by inna m in some Qurnic verses where this is syntactically possible. The resulting
constructions are grammatically sound, but they obviously differ in meaning from the
original constructions. Cf. innam/inna m ttuhu al ilmin ind (Q, 28:78); wa-
man yabxal fa-innam/fa-inna m yabxalu an nafsihi (Q, 47, 38); innam/inna m
tuxzawna m kuntum tamalna (Q, 66:7).
20 ramzi baalbaki
conditional particle ill (see no. 21 in the list of h ikya particles). But
irrespective of whether or not he actually differentiates between the two
types of ill at the etymological level, it is obvious that he intends to
establish the contrast between h ikya and non-h ikya particles at the
level of the speakers awareness of their compound versus non-com-
pound nature. In other words, the fact that linguistic analysis may well
prove the compound nature of exceptive ill and similar particles is of
no interest here to Sbawayhi since it has little to do with the speakers
awareness of their etymology. Based on the fact that proper usage and
comprehension of pairs of utterances which have two apparently identi-
cal particles is dependent on the distinction between the simple versus
the compound nature of each particle in a pair (e.g., exceptive versus
conditional ill), we can interpret Sbawayhis interest in such pairs as
part of his overriding interest in the competence of the speaker in cor-
rectly communicating the intended meaning to his listener. As for why
the particles which he describes as non-h ikya are considerably fewer
in number than the h ikya particles, two reasons may be suggested. The
first is Sbawayhis faith in the speakers ability to break down compound
particles into their componentsalthough some of them are more eas-
ily recognizable as compounds than othersand thus arrive at what
the later authors call lamh al-asl. The second reason may have to do
with the grammarians own failure, due to their largely synchronic and
non-comparative approach, to ascertain the compound nature of a large
number of particles (e.g., lam, layta, lta, laysa, kayfa, ayna, etc.).
Within the above confines, the particles which Sbawayhi explicitly
describes as non-h ikya (III:332) are the following:
1. ill: As an exceptive particle, it has the status of difl (oleander)i.e.,

it is a simple or non-h ikya particleand stands in contrast to the
conditional particle ill (see h ikya particles, no. 21).
2. h att: Like ill, it has the status of difl. It should be noted that h att
is the only member of Sbawayhis group of non-h ikya particles
which is apparently not made up of two elements and which has no
counterpart in the h ikya group. We list it here, nevertheless, for the
sake of completeness.
3. amm: As an inceptive particle in constructions like amm Zay-
dun fa-muntaliqun, it is a non-h ikya particle which has the status
of arw (the like of a thing) and stands in contrast to amm as in
amm anta muntaliqan intalaqtu maaka (see h iyka particles,
no. 5).
4. al: When used inceptively, it is a non-h ikya particle which has the
status of words like qaf (back) and rah (quern), in contrast to inter-
rogative al which is a h ikya particle (no. 19 above).
5. am: Like al, it is a non-h ikya particle when used inceptively, in
contrast to interrogative am (no. 19 above).
The distinction between non-h ikya particles and their h ikya counter-
parts in context is, of course, dependent on the speakers competence
in using the appropriate syntax, stress intonation, etc. But proper com-
municationas is implied in Sbawayhis textis also to a large extent a
function of the listeners ability to identify each of two identical particles
one of which is simple and the other is compound. This etymological
distinction is to be assumed in the listeners comprehension of sentences
in which these particles appear. Cf., for example:
l tatn fa-tuh additan ill izdadn fka rabatan (III:32) and wa-ill
tafir l wa-tarh amn akun min al-xsirna (Q 11:74; note that the
choice of these two examples is ours, since Sbawayhi does not provide
contrasting examples for ill);
amm Zaydun fa-muntaliqun (III:332), and amm anta muntaliqan
intalaqtu maaka (III:332); and
al innahu dhibun (IV:235) and al rajula imm Zaydun wa-imm
Amrun (I:289).
The fact that the distinction between formally identical particles in

the above sentences on the basis of their simple or compound nature
is accompanied by certain syntactic peculiarities is not without paral-
lels in the grammatical corpus. The mandatory use of f after simple
or non-h ikya amme.g., amm anta fa-muntaliqun versus amm
anta muntaliqan intalaqtu maakais strikingly similar to the man-
datory use of lm in constructions like in kna la-slih an (cf. III:104).
Although Sbawayhi does not use a special term for this lm, it acquired
later the name of al-lm al-friqa because its presence indicates that in
is not a negative particle but the lightened form of inna, known as in
al-muxaffafa. In other words, lm contributes to the distinction of two
formally identical particles both of which, unlike the case of amm,
are not compound. In Sbawayhis own words, this lm is mandatory
(alzamah l-lm) in order that the lightened form of inna not be con-
fused with in which has the status of m (li-all taltabis bi-in allat hiya
bi-manzilat m allat tanf bih; II:139).
22 ramzi baalbaki
5. Conclusion
Sbawayhis interest in the speakers awareness of the etymology of par-

ticles and its effect on his usage and on the listeners response is yet
another proof of his method of grammatical analysis which aims at
examining the mental operations which the speaker performs and at
determining the formal and semantic effects of these operations. As the
study of md and mand versus m d and man d in the Kitb shows,
successful communication between the speaker and the listener (who in
turn assumes the role of speaker) is measured by their ability to associ-
ate each of the two possibilities of usage (i.e., simple versus compound
particle) with the formal and semantic aspects which pertain to it. In
this particular case, what the grammarians refer to as etymology and
is intuitively discerned both by the speaker and the listener represents
the crucial factor upon which successful communication hinges. The
close association between form, meaning, and speakers awarenessas
exemplified in Sbawayhis analysis of h ikya and non-h ikya parti-
cleshas certainly been degraded, and at times even totally obliter-
ated by subsequent grammarians. On a wider scale, Sbawayhis model
of linguistic analysis which, like that of Ibn Jinns and Jurjns, largely
rests on exploring the dialectics or interplay, so to speak, between
form and meaning, gave way to an alternative model which is heavily
tipped in favor of formal considerations. Proponents of this model con-
sequently failed to delve, as did Sbawayhi, into the mind of the speaker
in order to pursue the complex processes which result in his choice of
the form that most appropriately expresses the intended meaning and is
expected to have the desired effect and elicit the correct response from
the listener.
6. References
6.1 Primary sources

Ibn Aql, arh = Bah ad-Dn Abdallh Ibn Aql, arh Ibn Aql al Alfiyyat Ibn Mlik.
Ed. by Ramz Munr Baalbak. Beirut: Dr al-Ilm lil-Malyn, 1992.
Ibn Jinn, Xasis = Ab l-Fath Utm
n Ibn Jinn, al-Xasis. Ed. by Muhammad Al an-
Najjr. Cairo: Dr al-Kutub al-Misriyya, 195256.
Ibn Mad, Radd = Ab l-Abbs Ahmad b. Abd ar-Rahmn Ibn Mad al-Laxm, ar-
Radd al n-nuh t. Ed. by awq D ayf. 3rd ed. Cairo: Dr al-Marif, 1988.
Ibn as-Sarrj, Usl = Ab Bakr Muhammad b. Sahl Ibn as-Sarrj, al-Usl f n-nahw.
Ed. by Abd al-H usayn al-Fatl. Beirut: Muassasat ar-Risla, 1985.
Ibn Xaldn, Muqaddima = Waliyy ad-Dn Ab Zayd Abd ar-Rahmn b. Muhammad
Ibn Xaldn, al-Muqaddima. Beirut: Dr al-Kitb al-Lubnn, 1956.
Jurjn, Asrr = Ab Bakr Abd al-Qhir b. Abd ar-Rahmn al-Jurjn, Asrr al-bala.
Ed. By Helmut Ritter. Istanbul: Government Press, 1954.
Jurjn, Awmil = Ab Bakr Abd al-Qhir b. Abd ar-Rahmn al-Jurjn, al-Awmil
al-mia n-nahwiyya f usl ilm al-Arabiyya, bi-arh Xlid al-Azhar. Ed. by al-Badrw
Zahrn. 2nd ed. Cairo: Dr al-Marif, 1988.
Jurjn, Dalil = Ab Bakr Abd al-Qhir b. Abd ar-Rahmn al-Jurjn, Dalil al-ijz.
Ed. by Muhammad Rad Rid. Repr. from the Cairo edition, Beirut: Dr al-Marifa,
1981.
Jurjn, Jumal = Ab Bakr Abd al-Qhir b. Abd ar-Rahmn al-Jurjn, al-Jumal. Ed. by
Al H aydar. Damascus: Dr al-H ikma, 1972.
Mubarrad, Muqtadab = Ab l-Abbs Muhammad b. Yazd al-Mubarrad, al-Muqtadab.
Ed. by Muhammad Abd al-Xliq Udayma. Cairo: Dr at-Tahrr, 196568.
Sbawayhi, Kitb = Ab Bir Amr b. Utm n Sbawayhi: al-Kitb. Ed. by Abd as-Salm
Muhammad Hrn. Cairo: al-Haya l-Misriyya l-mma, 1977.
Suyt, Ham = Jall ad-Dn Ab l-Fadl Abd ar-Rahmn b. Ab Bakr as-Suyt , Ham
al-hawmi arh jam al-jawmi f ilm al-Arabiyya. Cairo: Matb aat as-Sada, 1327
A.H.
Umn, arh = Ab l-H asan Al b. Muhammad al-Umn, arh al-Umn al
Alfiyyat Ibn Mlik al-musamm Manhaj as-slik il Alfiyyat Ibn Mlik. Ed. by
Muhammad Muhy d-Dn Abd al-H amd. Cairo: Dr al-Kitb al-Arab, 1955.
6.2 Secondary sources

Anghelescu, Nadia. 2004. La langue arabe dans une perspective typologique. Bucharest:
University of Bucharest.
Baalbaki, Ramzi. 1979. Some Aspects of Harmony and Hierarchy in Sbawayhis Gram-
matical Analysis. Zeitschrift fr arabische Linguistik 2, 722.
. 1982. Tawahhum: An Ambiguous Concept in Early Arabic Grammar. Bulletin of
the School of Oriental and African Studies 45, pt. 2. 22344.
. 1983. The Relation between nahw and bala: A Comparative Study of the Meth-
ods of Sbawayhi and Jurjn. Zeitschrift fr arabische Linguistik 11. 723.
. 2001. Bb al-f [ f + Subjunctive] in Arabic Grammatical Sources. Arabica 48.
186209.
Bohas, Georges, J. P. Guillaume and D. E. Kouloughli. 1990. The Arabic Linguistic Tradi-
tion. London: Routledge.
Carter, Michael, G. 1991. Elision. Proceedings of the Colloquium on Arabic Grammar,
Budapest 17 September 1991, Kinga Dvnyi and Tams Ivnyi: eds. Budapest: Et-
vs Lornd University and Csoma de Krs Society. 12133.
. 2004. Sibawayhi. London and New York: I.B. Tauris and Oxford University Press.
Sezgin, Fuat. 1984. Geschichte des arabischen Schrifttums. IX. Grammatik. Leiden: E.J.
Brill.
Suleiman, Yasir. 1999. The Arabic Grammatical Tradition: A Study in tall. Edinburgh:
Edinburgh University Press.
Troupeau, Grard. 1976. Lexique-index du Kitb de Sbawayhi. Paris: Klincksieck.
Zakariyy, Michel. 1986. al-Malaka l-lisniyya f Muqaddimat Ibn Xaldn. Beirut: al-
Muassasa l-jmiiyya li-d-dirst wa-n-nar wa-t-tawz.
PRAGMATICS AND CONTRACTUAL LANGUAGE IN
EARLY ARABIC GRAMMAR AND LEGAL THEORY
Michael Carter
University of Sydney
It is well known that Arabic grammar and Islamic law enjoy a peculiarly
close relationship. The two sciences are united by a common purpose,
to control linguistic and general behaviour respectively, and they share
a common methodology, namely the inductive derivation of rules from
a linguistic corpus and the deductive application of these universal rules
to particular acts of the Muslim. Where they differ is in their sources.
Grammar relies on the natural, worldly speech of a select range of
human speakers (Bedouin), law on the inspired texts of the Qurn and
the H adt, which are supernatural in origin and holy in status. All legal
systems are linguistic codes of one sort or another, spoken or written,
but the total dependence of Islamic law on a finite body of revealed and
prophetic language is unique. Its modern secular analogue is the type of
law which is derived from a written constitution, and here too, the law
has to be discovered by an essentially linguistic process, whereby there
is often disagreement over the presumed intentions of those (invariably
dead) who framed the document.
The development of a method for interpreting the language of the
Qurn and H adt took several centuries, and at risk of oversimplifica-
tion it can be said that the two sciences of grammar and law, aided by
imported Aristotelian logic, leap-frogged each other in an evolutionary
series, where the advances of one made further progress possible in the
other. This paper will review the general similarities in the approach
to language in early grammar and law, especially Sbawayhis intuitive
pragmatism (here in the non-technical sense) and his awareness of the
legal implications of grammatical form. There follows a brief account of
some grammatical/legal problems discussed in an intermediate phase
in the 3rd4th/9th10th centuries, and the paper concludes by listing a
number of features of legal methodology which can be linked with ideas
first noted in Sbawayhi, but which only acquired their fully developed
form after the maturing of us l al-fiqh as a discipline.
26 michael carter
Islamic legal hermeneutics proceeds from the axiom that, regard-

less of its supernatural origin, the language of the holy texts conforms
entirely to the principles of human discourse. Two short quotations, one
from Sbawayhi (d. ca 180/796) and one from a-fi (d. 204/820, and
probably born about the same time as Sbawayhi) will testify that this
notion was well defined at the very birth of the sciences of grammar and
law :
Gods servants were spoken to in their own speech and the Qurn came
down in their language and according to what they mean
ibd[u] llhi kullim bi-kalmihim wa-ja l-Qurnu al luatihim wa-
al m yanna (Kitb Der. I:139/Bl. I:167).
In almost identical wording from a-fi (the original Arabic could not
be checked: this is from Khadduri 1987, 94, and looks very like a quota-
tion from or paraphrase of the Kitb):
God has addressed his book to the Arabs in their tongue in accordance
with the meanings known to them.
The context in a-fis case was the dispute over the possibility of for-
eign words in the Qurn, while for Sbawayhi it was a syntactical issue
of indefinite expressions such as salmun alayka; although a definite
as-salmu would be expected, the indefinite is an old-established Arab
usage that must be accepted, especially when it appears in the Qurn.
A-fi is regarded as the first legal theorist to give proper weight
to the linguistic aspects of the law, since which time both grammar-
ians and lawyers have shown themselves to be remarkably strict and
uncompromising in subordinating the language of God to the linguis-
tic conventions of ordinary Arabic. The latter, for Sbawayhi, was a dia-
logue between speaker and listener, both being required to conform to
what are basically ethical criteria to speak well (h asan) and rightly
(mustaqm). For the lawyers there could be no dialogue with God, only
the contemplation of the written record of what he and his Prophet said,
but over time they evolved a system of interpretation in which they
played the role of silent listeners to a speaker of their own tongue, under
the same conditions as natural speech.
For this they constructed an elaborate hermeneutical mechanism
(usl al-fiqh) which, as documented by Ali (2000), exhibits an impres-
sive congruence in many details with the modern branch of linguistics
pragmatics and contractual language in early arabic 27
known as Pragmatics (henceforth with capital P as a school of thought).

The thesis of the present article is that a form of Pragmatics can be dis-
cerned in Sbawayhis analysis of speech (kalm) long before the the
usl al-fiqh were codified, and, further, that this approach to language
passed through at least two stages before it grew into the explicit type of
Pragmatics seen in Alis sources.
For the purposes of this article usl al-fiqh will be taken in the broad
sense of the principles of legal argumentation in deriving law from the
texts. As for Pragmatics, here is a recent definition which so closely
reflects the usls position that it might have been written by one of
them, or even by Sbawayhi:
The study of language from the point of view of the users, especially of the
choices they make, the constraints they encounter in using language in a
social interaction, and the effects their use of language has on the other
participants in the conversation [. . .] including aspects of deixis, impli-
catures, presuppositions, speech acts and discourse structure. [. . .] It has
been characterized as the study of the principles and practice of conversa-
tional performancethis including all aspects of language usage, under-
standing and appropriateness (Crystal 2000, s.v. Pragmatics).
Readers of the Kitb will find all these notions very familiar, and some
will be illustrated below. But first it is necessary to introduce the essen-
tial elements of Pragmatics as laid down by Grice.
Grice (1989) treats speech (writing, significantly, does not fit easily
into his model, nor into Sbawayhis, see below) as a cooperative activity
that is both purposeful and rational, in which the participants under-
stand each other by a logically structured process of conversational
implicature. Speech is a quasi-contractual matter, governed by four
maxims (Grice 1989, 26f):
The maxim of Quantity :

1. Make your contribution as informative as is required (for the cur-
rent purposes of the exchange).
2. Do not make your contribution more informative than is required.
The maxim of Quality :
1. Do not say what you believe to be false.
2. Do not say that for which you lack adequate evidence.
The maxim of Relation:
Be relevant.
28 michael carter
The maxim of Manner:

1. Avoid obscurity of expression.
2. Avoid ambiguity.
3. Be brief (avoid unnecessary prolixity).
4. Be orderly.
All these maxims can be deliberately ignored or flouted, sometimes for

personal reasons, e.g. a desire to mislead or deceive, sometimes for rhe-
torical purposes, e.g. exaggeration, irony, implying what one is reluctant
to say explicitly, intentional ambiguity etc. There can be and often is a
difference between the words actually spoken (sentence meaning) and
what is really meant (utterance meaning).
In Larcher (1998) an earlier form of Pragmatics is applied to the
category of in, i.e. performative utterances of the type qabiltu hd
n-nikh a I hereby accept this marriage [proposal]. This is the Pragmat-
ics of J. Searle and J. L. Austin of How to do things with words fame (bib-
liographical details in Larcher), out of which Grices maxims were later
elaborated. However, Grice has been invoked at least twice over the past
few years in articles on Arabic linguistics. Moutaouakil, who quotes the
above maxims in full (1990, 233), deals only with the Pragmatics of the
relatively late scholastic author as-Sakkk (d. 626/1229) and will not
detain us further, except to remark, following Simon (1993, 15) (where
further sources), that it was not until grammar and usl al-fiqh had
themselves achieved systematic perfection that an independent science
of rhetoric could emerge, effectively the last of the Islamic sciences to
appear.
The other is a direct comparison between Grice and Sbawayhi made
by Buburuzan. She interprets the constructions nima l-rajulu abdullhi
What a fine man Abdullh is! and abdullhi nima l-rajulu Abdullh,
what a fine man [he is]! (1993, 424, Kitb Der. I:259f/Bl. I:300f) as
compound expressions where the second part presupposes a question
from the listener. The first part, she says, has violated Grices maxim of
quantity and requires completion in answer to the question who? or
what about him?1
1
It is relevant here to recall a similar analysis of the syntax of the zaydun ja abhu
structure, Zayd, his father came, which Bravmann (1953, 136) explained as deriv-
ing from a self-answered question, What about Zayd? His father came; Sbawayhi loc.
cit. actually compares the nima construction to this same type, viz. abdullhi dahaba
axhu Abdullh, his brother has gone. Bravmanns Isolated Natural Subject is not
Sbawayhi makes more or less the same statement about the syntax
of marartu bi-rajulayni muslimin wa-kfirin/muslimun wa-kfirun I
passed by two men, one Muslim, one pagan, where the speaker will
choose between the oblique case ( jarr) in adjectival agreement or the
independent (raf ) case as if muslimun wa-kfirun were predicates of
elided subjects, because, as Sbawayhi observes (Der. I:182/Bl. I:214,
and see Der. I:215/Bl. I:252 for another, similar case), the speaker tries
to anticipate the question the listener might pose, either, what sort of
men? or, who were these two men?
Like Grice, Sbawayhi is very concerned with the listeners role in
conversation, and there are many linguistic events in which the listener
influences the speakers choices. In what would be a neither [. . .] nor
[. . .] construction in English, it is qabh . i.e. structurally incorrect, to
say marartu bi-rajulin l frisin I passed by a man who was neither
a knight without completing it with wa-l ujin nor a valiant per-
son or the like, because it is an answer to someone who asked youor
whom you have put in the status of having askedwhether you passed
by a knight or a valiant man (Der. I:313/Bl. I:358, translators italics).
By the same token a listener who answers no to the disjunctive ques-
tion Is it Zayd who is with you or Bir? when one of them is known
to be there, has broken the communicative contract so gravely that his
answer is classified as muh l [morally] wrong, [semantically] absurd,
i.e. an utterance which is self-contradictory and therefore meaningless
(Der. I:432/Bl. I:483). As if to reinforce the importance of the listener,
Sbawayhi comments that in talking to oneself, e.g. hall af alu why
dont I do this, you are like the listener (Der. I:114/Bl. I:136, kunta
fhi ka-l-muxtabi).
There is even a discussion of what looks like body-language when
Sbawayhi describes how, on seeing the figure of an unknown person,
some sign (ya, the same word as for the verses of the Qurn!) appears
by which you identify him, so you exclaim, Abdullh! Good Lord!
Not only that, the same elliptical exclamation (that is, a predicate with-
out a subject) can be uttered when the sign by which you identify a
person is his voice or perfume, or simply what you hear said about him
(Der. I:240/Bl. I:279).
intrinsically Gricean, but it accords well with the eminently Pragmatic principle stated
by Sbawayhi (Kitb Der. I:346/Bl. I:394, a notion he acquired from his teacher al-
Xall b. Ahmad), that every subject must have a predicate because the listener is expect-
ing it.
30 michael carter
It will be apparent that as well as the speaker and listener, the physical
context is also linguistically relevant. Both Grice and Sbawayhi like to
present their data in the form of utterances set in a described situation.
In Grice it is undoubtedly fictitious and often involves broken-down
cars or sherry parties, in Sbawayhi it probably reflects actual observa-
tion, e.g.:
an example of the suppression of the verb which could be expressed in
normal usage is when you see a man who has just returned from a journey
and you say, the best of returns xayra maqdamin, [. . .] where the depen-
dent (nasb) form is as if [the speaker] had [syntactically] constructed it on
the basis of having said may you return, viz. qadimta xayra maqdamin,
and even though he was not heard to say this expression, the arrival of the
other person and the sight of him have the same [linguistic] status as the
speakers saying qadimta (Der. I:114f./Bl. I:136f.).
The similarity with Gricean Pragmatics is unmistakable here: not only
does the speaker engage in a cooperative activity with the listener in a
real context, but that context itself can become an active constituent in
the grammatical form of the utterance (cf. Carter 2002, 7, where the
above item is discussed).
In the same way the vocative particle can be dispensed with when
the listener is standing right in front of the speaker (Der. I:104, 274/
Bl. I:125, 316); the object of a blessing does not have to be indicated
if the intended recipient is obvious from the context (Der. I:131/Bl.
I:157)the speaker however remains free to name the recipient for pur-
poses of emphasis). In a rather violent scenario (Der. I:107/Bl. I:128)
the speaker can dispense with the verb and merely shout the persons
name if he sees someone about to be killed or being abused. This is com-
mon in warnings, e.g. al-jidr(a) [mind] the wall!, al-asad(a) [dont
go near] the lion!, and the cry of at-tarq(a), at-trq(a) [get out of] the
way! [Get out of] the way!2
A striking feature of the Kitb is the sheer quantity of commercial and
contractual talk, much of it admittedly trivial, though it does tell us how
preoccupied the Bedouin in the Mirbad were with the price of sheep,
camels and wheat.3 But many items are strictly legal in form and con-
2
These are always printed with dependent (nasb) case endings, however, in the cir-
cumstances they are bound to be in pausal form.
3
The Mirbad was not only a market but also a place where H adt scholars came to
check their vocabulary (EI 2, art. Mirbad by C. Pellat), and this may be one of the
reasons why Sbawayhi came to Basra to study tr or H adt. As it happens mirbad
tent, involving contracts, debts, sale of goods and property, testimony,

deposits, theft and the activity of fiqh itself. The following list appeared
in Carter 1972, 90ff:
inna f alfi dirhamin la-madrabun (Der. I:99/Bl. I:119); baya l-malat
l ahda wa-l aqda (Der. I:115/Bl. I:137); tabu d-dra h addun minh
kad wa-h addun minh kad (Der. I:137/Bl. I:165: note that the sale of
undefined property was not legal, at least among the H anafs, see Hamilton
1870, 257), lahu ilmun ilmu l-fuqahi (Der. I:151/Bl. I:181); lahu alayya
alfu dirhamin urfan (Der. I:160/Bl. I:190); bitu -a tan wa-dirha-
man, qmartuhu dirhaman f dirhamin, bituhu dr diran bi-dirhamin,
etc. (Der. I:165/Bl. I:196); hd dirhamun waznan and others (Der. I:235/
Bl. I:275); inna alfan f darhimika bdun (Der. I:245/Bl. I:285); kam
minkum hidun al fulnin (Der. I:256/Bl. I:297); alayhi aru kalbayni
daynan (Der. I:257/B I:298); al-wadatu ayyuh l-biu (Der. I:284/Bl.
I:326); qadiyyatun wa-l Ab H asanayni = the caliph Al, (Der. I:310/Bl.
I:355); ti l-amra l yaqtau l-lissa (Der. I:402/Bl. I:435); amm juhda
ray (Der. I:418/Bl. I:470). The pre-emptive in a llhu to avoid being
bound by an oath may also be mentioned here: it is called istitn both by
Sbawayhi and in later legal terminology (Der. I:399/Bl. I:448, see also
Carter 1972, 90 n. 2).
Additional legal or commercial material which has come to light since
includes (and is not exhaustive: marginal items have been ignored such
as have you barley, wheat or dates?, though this may well reflect a mer-
cantile context, Der. I:434f/Bl. I:485f):
Placement of merchandise (Der. I:64/Bl. I:76: even in those days it was
better to buy from the top of the pile rather than the bottom!); expres-
sions of time (important for contracts, Der. I:9093, 176/Bl. I:110114,
208); litigation (Der. I:94/Bl. I:114); price variations (Der. I:122, 170/Bl.
I:147, 200); profit, sadaqa and zakh, rendering accounts (Der. I:165f/Bl.
I:196f); weights and measures (Der. I:141f, 183/Bl. I:216, 292f); sub-
stitution of goods (Der. I:245/Bl. I:285); giving testimony (Der. I:421/
Bl. I:473); default masc. for mixed genders, i.e. slaves (Der. II:180/Bl.
II:174).
One specifically contractual type of utterance will be discussed here in
some detail, involving the way prices are stated. In a long analysis (over
four chapters, 9295, Der. I:165168/Bl. I:1958) Sbawayhi explores
how deals are struck, beginning significantly with the grammar of
kallamtuhu fhu il fiyya I spoke to him face to face and byatuhu
is mentioned once in the Kitb (Der. II:265/Bl. II:248) but only as an example of the
mif al pattern.
32 michael carter
yadan bi-yadin I traded with him hand in hand (i.e. for cash), where
the dependent (nasb) forms are obligatory, since they are not literal, but
simply mean immediately, on the spot, no matter how physically close
the listener might be. In other words the legal (utterance) meaning is
different from the overt (sentence) meaning: by saying these words in
this form a legal obligation is created regardless of their literal meaning.
This leads to a whole string of commercial expressions, bitu -a tan
wa-dirhaman, qmartuhu dirhaman f dirhamin, bituhu dr diran
bi-dirhamin, bitu l-burra qafzayni bi-dirhamin, axadtu zakta mlihi
dirhaman li-kulli arbana dirhaman, bayyantu lahu h isbahu bban
bban, in all of which the phrase indicating the unit and price (scil. tan
wa-dirhaman one sheep and one dirham) must be stated in its entirety,
otherwise the meaning will not be valid l yasih h u l-man, i.e. legally.
By this Sbawayhi means only in the case of a contractual intent, since
the shortened expressions omitting the price are still meaningful but not
in any contractual sense, e.g. bitu tan tan I hereby sell my [col-
lective] sheep, sheep by sheep, bituhu dr diran I hereby sell him
my house, one cubit, but this would lead the listener to believe that you
were selling your sheep one at a time or that your house was only one
cubit in size, and so on.
Nevertheless, as he observes, the price or the unit are frequently
omitted in ordinary speech, and people will say kna l-burru qafzayni
the wheat was [for sale] at two bushels, omitting the price, or al-burru
bi-sittna the wheat is for sixty [dirhams], omitting the unit of quantity.
They do this, Sbawayhi says, because in the first instance they know
in their hearts ( f sudrihim) that bi-dirhamin is meant and that the
dirham is the standard price unit (allad yusaaru alayhi), so it is as if
they were answering the question, How much you get for a dirham?,
while in the second they and the listener both know what they mean,
as if someone had asked What is the price of a load? and received the
answer The load is [for sale] at sixty [dirhams] (Der. I:166/Bl. I:196).
Sbawayhi advises us to follow the practice of the Arabs in this, though
al-Xall complicates the picture somewhat by pointing out alternative
formulations.
Sbawayhis Pragmatic approach is self-evident here: he puts the
conversation in a real-life setting, which assumes all the Gricean max-
ims: he distinguishes between utterance and sentence meaning, and he
accounts for the grammatical features of the expressions in terms of the
extralinguistic situation and the intentions of the participants. If a state-
ment such as yajzu an taqla bitu d-dra dirun bi-dirhamin (Der.
I:166/Bl. I:197) were taken out of context there would be no way of

knowing whether it was from a legal work, scil. it is [legally] permit-
ted4 to utter [the binding contractual formula] I hereby sell etc., or a
grammatical work, scil. it is allowed [by the rules of grammar] to say, I
hereby sell etc. 5 Only when Sbawayhi goes on to discuss the possibility
of the independent (raf ) case in dirun here (as advocated by al-Xall)
do we learn that the topic is syntax. The goal, however, is to determine
the legal consequences of the syntactic options. Why else would he raise
the matter at all?
In the 3rd/9th century a new weapon enters the methodological
armory, the concept, borrowed from Aristotle, that science has to be
logically structured and its enquiries carried out along systematic, that
is, logical lines. For about a century and a half there was a great deal of
experimentation before the sciences reached the point where they could
be defined and classified, notably in such works as the Mafth al-ulm
of al-Xwrazm, written between 366/976 and 387/997. Two debates
from that period will be mentioned here as evidence of the transitional
stage in the development of grammar and law.
Ibn Walld (d. 332/943) reports a dispute which originated with
Mubarrad (d. 285/898) over the semantic status of commands and pro-
hibitions, whether they were logically complementary, i.e. whether a
command is equivalent to a prohibition from doing the opposite and
a prohibition is equivalent to a positive command to do the opposite.
Ibn Wallds wording (Intisr 42) is: kullu amrin amarta bihi fa-anta f
l-man nhin an xilfihi [. . .] fa-id nahayta amarta bi-xilfihi hence
the positive exclamation h adaraka beware! has the negative meaning,
do not approach. The probable Stoic origins of this controversy (cf. Ver-
steegh 1977, 181) are not the main point of interest here:6 what is impor-
tant is to observe how the intellectual environment stimulated such
arguments while the usl al-fiqh were still in gestation. Al-Mubarrad
is among the earliest of the grammarians to reveal the direct influence
4
The old definition of jiz by Bergstrer (1935, 32) is still the most informative:
zulig im moralisch-religisen und zugleich rechtlichen Sinne, und daher rechtsgltig,
rechtswirksam. Replace legal by linguistic to see what the term meant to Sbawayhi:
permissible in a religious-moral and at the same time linguistic sense, hence linguisti-
cally valid, linguistically effective.
5
Sbawayhis examples in this section are not unambigously performatives, and they
can be, and have been translated elsewhere as literal statements I have sold etc.
6
The Stoic term pragmata for the things done which are represented by words is the
basis of our linguistic Pragmatics in the sense of doing things with words.
34 michael carter
of Greek ideas before there is any reliable documentation, suggesting

that the ideas circulated informally, perhaps from Christian to Muslim
scholars. Ibn Walld himself is very sceptical, and he flatly asserts that
there is nothing intrinsically positive or negative about commands: only
by inference (istidll) can we determine whether an imperative verb is
ordering us to do something or prohibiting us from doing the opposite.
The language and style of this argument are exactly what we encoun-
ter when the usls take up the theme later in the century. In al-Bqilln
(d. 403/1013, and a strong supporter of the notion of complementarity),
the debate is given a thorough airing (Taqrb II:198207), and by this
time both the grammatical and the logical techniques are well advanced,
and the level of argumentation shows that the Aristotelian dialectic had
been fully absorbed. There is a good deal of sophistry from al-Bqillns
opponents, who claim, amongst other things, that imperative and pro-
hibitive verbs have different forms, therefore cannot be complementary
to each other. He matches their sophistry with a reductio ad absurdum,
putting it to them that by their own criteria a positive act would have
to be defined as not refraining from doing the opposite ayru tarki
diddihi and he concludes with the practical argument that the mean-
ings of commands or prohibitions are understood naturally and grasped
immediately from the words themselves, without the need for any kind
of logical inference (while this may not look like Pragmatics, it is really a
question of differing opinions on conversational implicature: the meth-
ods of inference were a very contentious issue among usls, see Ali
2000, especially ch. 5). There is thus a clear line of progression from
Sbawayhis casual recognition of amr and nahy as grammatical catego-
ries (e.g. Der. I:105/Bl. I:1267), through Ibn Wallds early dialectical
treatment to the fully structured arguments of al-Bqilln and his fel-
low usls.
In other words, there is now a theory where before there was only data.
Under these circumstances there is a universal tendency for theory to tri-
umph over data, and a very illuminating example is the treatment of the
word kad such and such [an amount]. For Sbawayhi, it was no more
than a dummy numeral (kinya, as he calls it, an allusion to any number,
just as fuln so-and-so is an allusion to any person), and it behaves like
interrogative kam, with a dependent (nasb) complement, kad wa-kad
dirhaman so and so many dirhams (Der. I:256/Bl. I:297).
By the end of the 4th/10th century a more complicated system makes
its appearance, when ar-Rummn (d. 384/994, arh fol. 161r) proposes
that kad by itself has the value of 1120 (because these numbers are
regarded as single words) while kad wa-kd stands for compound

numbers from 21 upwards.7 Ar-Rummn was famous for the exagger-
ated logical rigor of his experiments with language, but he did not go as
far as he could have done. Others did go further, and by the 7th/13th
century we find the fully elaborated system in Ibn Mut (d. 628/1231,
Fusl 23, further references in 2445), exactly mimicking the syntax of
the numerals:
kad wa-kad darhima = from 3 to 10 dirhams.
kad wa-kad dirhaman = from 11 to 99 dirhams.
kad wa-kad dirhamin = from 100 dirhams upwards.8
This exquisitely artificial and almost certainly unattested scheme is con-
nected with the H anafs and (which is not much different in this con-
text) with the Kfans, and it survived at least until the 19th century in
the work of Nsf al-Yzij (d. 1871, in Nr al-Qir, see Fleischer I:568
for references). What it shows is that as the linguistic and legal sciences
evolved they became increasingly abstract, passing beyond the limits of
actual usage. It does not seem likely that Ab H anfa himself explicitly
correlated the syntax of kad with that of the numerals in this manner,
but it fits the reputation of the H anafs for artificial and over-systematic
reasoning, and can be seen as a fine specimen of what can happen when
a legal judgement (scil. how much is meant by kad in such and such a
case?) has to depend more on the methodology than on actual speech.
Classical Arabic was no longer a living language at this time, and the
usls could only consult the rules of the dead language.
In doing so they nevertheless assumed that their one-way commu-
nication with the texts was a natural use of language. Here follow some
illustrations of the inherent similarities between Sbawayhis concept of
speech and the lawyers approach to consultations with God.
It has already been said above that Gods speech had to follow the
formal rules of Arabic grammar. Furthermore, as the usls were well
aware, in order to make himself understood, God had to obey the con-
ventions of human communication. For this reason the usls, exactly
like Sbawayhi, strove to account for meaning in terms of the motives
of the speaker (man or God) and the real-life context of the utterance.
A good specimen from Sbawayhi of the dependence of meaning on
7
Through a scribal error the single kad (scil. kad dirhaman) and paired kad (scil.
kad wa-kad dirhaman) are not distinguished in the manuscript.
8
Further references in Carter 2003a, endnote 27 referring to p. 180.
36 michael carter
motive and context is the statement m atn rajulun: it can mean no

man [at all] came to me, not one man came to me[but several], no
[real] man came to me [but a weakling], or no man came to me [but a
woman did] (Der. I:20/Bl. I:27). There are no formal linguistic clues
whatsoever as to which meaning the speaker has in mind: the clues lie
in the extra-linguistic context, in this case, what question might the
speaker be answering, and this is why Sbawayhi spends so much time
offering psychological explanations for the speakers choices. Some cen-
turies later the lawyers had to do the same for God, trying to penetrate
his words to divine his purpose, which they could only do by assuming
that he spoke to his servants in the same way as a rational human being
would under normal circumstances.
Sbawayhi took this for granted. The muxtabna in this next quota-
tion are the same people who are the muxtabna of everyday conversa-
tion, but this time it is God speaking to them:
When [God] said your mothers are prohibited to you and so on to the
end of his speech (kalm), those being addressed (al-muxtabna) knew
that this was proscribed for them, and affirmed, but God went on to say
kitba llhi by written decree to add emphasis, tawkdan (Der. I:160/Bl.
I:191, on Q 4/2324).
Striking here is the attempt to explain away the mention of the written
decree: writing is ill-suited to Pragmatic analysis for the simple reason
that the recipient of writing (or in this case the sender!) is usually absent.
There can be no conversation, still less any conversational implicature,
without the presence of both participants. Writing is mentioned only
occasionally in the Kitb, and is completely marginal and secondary.
For the usls, too, it was not part of normal communication: it was a
special case, a kind of act of faith in the future, like farming, lending or
borrowing, and other actions which require a presumption of continu-
ity, istish b al-h l, i.e. that the recipient would still be alive when the
letter was delivered (see Ali 2000, 80). We have to admire the usls
for their commitment to the belief that the spoken words of God and
Muhammad can be directly experienced through their written record:
the paradox was to a large extent resolved by the device of learning the
texts by heart so that, once implanted in the memory, they ceased to be
a document (mush af, kitb) and became a virtual oral event (Qurn,
h adt), neurologically the same as a remembered discourse.
God as a speaker also has the same privilege as humans of presuming
knowledge in his listeners: in Q 3/180 (Der. I:347/Bl. I:395) he says let
not those who are miserly with the bounty that has come to them reckon
[being miserly] is better for them . . ., and Sbawayhi explains that God
here omits the word al-buxla being miserly which would be required
as the first direct object of reckon, because the listener, al-muxtab,
will know from the verb yabxalna they are miserly that miserliness
is meant. Anyone who has consulted a tafsr will be familiar with this
method of filling gaps in the Qurn, and the usl sources are no excep-
tion in the use of this procedure.
A second point of similarity between Sbawayhi and the usls is the
presumption of sanity. Sbawayhi assumes, without spelling it out, that a
speaker will be mentally capable of formulating an idea and conveying
it successfullythis is implicit in his criterion of mustaqm right, used
for utterances which are fully understood in their intended meaning,
and in the term murd what is intended for the meaning of speech
acts. The lawyers turned this into an overt legal principle. They had to
decide who was allowed to speak, and to eliminate those who were not
legal persons, and therefore had no voice, such as infants and the insane,
and they produced the following conditions of legally valid speech (here
paraphrased from Ali 2000, 42 based on al-mid, d. 631/1233):
1. It must be uttered intentionally.
2. It must be intended for a particular listener.
3. The listener must be rational and understand it.
The first condition presupposes sanity, because only a sane person can
form an intention at all (legal or otherwise), or indeed be a Muslim for
that matter (sanity later formed part of the definition of a Muslim); the
second criterion excludes soliloquy, and the third anchors speech in a
sane society, as well as giving us a hint as to how the usls saw them-
selves in the dialogue. To recall the notion of leap-frogging introduced
above, it should be noted that these new legal definitions of speech found
their way back into grammar, where the speech of the insane, or of those
talking in their sleep, or even of birds imitating humans, were excluded
for the sole reason that their speech could not be intentional (e.g. a-
irbn d. 977/1570, Nr 10, though it appeared before his time).
God himself comes under the same constraints: in order to commu-
nicate with humanity he must speak rationally. His attributes allow for
this: he has an intellect, a will and the power of speech, and his language
is that of the people he is addressing (there is some literature, which can-
not be looked at here, on the requirement that God address his prophets
in the language of their own people otherwise the revelation will be in
38 michael carter
vain: you cannot punish sinners for disobeying laws they are unable to
understand). For the usls Gods mode of communication (bayn) is
inevitably a major theme in their treatises: they either begin their works
with a detailed review of the nature of Arabic and of legal semantics,
e.g. the introductory chapters of Ab l-H usayn al-Basr (d. 436/1044),
Mutamad, or else the subject is raised after the epistemological topics
have been covered, e.g. al-Bqilln (d. 403/1013), Taqrb, from I:316.
For Sbawayhi, and later the usls, lexical meaning is arbitrary. Defi-
nition by synonyms only leads to infinite regression (Der. II:339/Bl.
II:312), and meaning is nothing more than intention, hence the verb
arda to want and its derivatives are among the commonest terms in
the Kitb (1,362 times, plus 20 in the passive, according to Troupeau
(1976), s.v., and there are also synonyms). It is clear, too, that Sbawayhi
fully aware of the distinction between utterance meaning and sentence
meaning: he refers more than once to man l-kalm and man l-h adt
the [integral] meaning of the utterance i.e. not simply the sum of its
lexical parts, and man itself is almost exclusively used to denote the
meaning of speech acts, not of words, such as the acts of expressing sur-
prise, asking a question, giving an order etc., e.g. the meaning of swear-
ing an oath man l-qasam, the meaning of calling man n-nid, and
even of grammatical categories, the meaning of the dependent form
man n-nasb, the meaning of tanwn, etc.
A significant similarity between Sbawayhi and the lawyers is that they
both define the meanings of the particles (h urf ) in terms of their dis-
course functions: thus wa- and is used to to bring one thing together
with another and join them without indication of order, and fa- and
[then] is the same except that you leave some scope for one to be after
the other (Der. II:330/Bl. II:304), cf. Ab l-H usayn, Mutamad I:20,
very concisely, wa- is li-l-jam for joining while fais li-l-taqb for
arranging consecutively. Sbawayhis definition of naam yes is inter-
esting: instead of the expected agreement or consent he gives us a
rather legalistic definition: naam indicates promise and belief , ida
wa-tasdq (Der. II:339/Bl. II:312), the former implying some kind of
contractual commitment (yes, I promise do it), the latter indicating
assent to a proposition (yes, I believe what you say), which in our con-
text could mean believing the sellers description of the goods or the
terms of a contract.
Sbawayhi never even asks where meaning originates, but the usls
were obliged to agree on an answer before they could proceed to the
derivation of law from the texts. Here is a case where the debates of
the intervening century and a half carried the lawyers a long way from
Sbawayhis agnostic position. They had to reconcile the potentially infi-
nite backward extension of meaning with the historical fact that Arabic
is not the oldest human language. Part of the solution, which will not be
discussed here, was to attribute to Adam the bridging role connecting
the supernatural Arabic he spoke in Heaven with the temporal world he
inhabited after the Fall. He himself did not speak Arabic on earth which
appeared only later, evolving naturally until it reached perfection in the
time of Adams prophetic heir and descendant, Muhammad.9
After Sbawayhi there was considerable discussion of the origins of
language, which has been investigated for the grammarians by Lou-
cel (196364), with the general conclusion that the choice lay between
divine ordination (tawqf) or human convention (tawdu). The lacuna
which Loucel pointed out, that there was no comparable study of the
origins of language in the legal sciences, remains unfilled and will not be
dealt with here. For our purposes it must suffice to note that there was
overall preference for the view that language is in some way convention-
ally imposed (by wad) but the identity of the imposer, wdi, is left
obscure, perhaps deliberately: it may be God, it may be the first users of
Arabic, it may be all users of Arabic who agree amongst themselves on
the meaning of a word.
The real dispute concerned whether meaning could be imposed inde-
pendently of a words being used. At one end of the spectrum (Mutazil),
a word does have a meaning before it is used, and at the other (Ibn
Taymiyya), a word cannot have a meaning until or indeed unless it is
used. These issues are well described by Ali (2000): what is important
for this paper is that none of these ideas, including the term wad in
this sense, are found in the Kitb. The concept must have emerged later,
probably under the influence of the Platonic debate over whether words
had meaning by their nature ( physis, cf. Arabic taba, replaced in Islm
by the creating God) or by imposition (thesis, the same as the wad of the
grammarians and lawyers).
From the lawyers point of view it was important to detach meaning
from prehistory: in spite of disagreement about the origins of language,
lexical meaning was taken as given, either a priori or as recorded by lexi-
cographical experts (ahl al-lua), or synchronically by mere usage. It
9
See Carter (2003b).
40 michael carter
could then be treated as purely conventional, and in this way the lawyers,
after a long debate in which Sbawayhi took no part, eventually returned
to his Pragmatist position. Meaning, regardless of where it comes from,
is what you do with the language, or, put another way, language is only
meaningful for usls when it has legal effects outside the actual utter-
ance. This restriction of meaning to habit and community usage enables
the law to control that communitys behaviour. The result is the same for
both grammarians and lawyers: speakers are obliged to stay within the
habitual codes, whether linguistic or social.
We might say that Sbawayhi took a lawyers view of language and
lawyers a Sbawayhian view. Consider his Pragmatist interpretation of
such verbs as ra etc. to see, consider, regard, be of the view that, of
which he says, even a blind man can use ra to see and say, I regarded
Zayd as the good man raaytu zaydan-i s-slih a (Der. I:13/Bl. I:18).10
The equivalent lawyers position is stated by a-aybn (b. 132/749, d.
189/804, and therefore a contemporary of Sbawayhi): a blind man who
has to feel the goods for sale when making a purchase is in the same
place (mawdi, i.e. legal situation) as a sighted man ( Jmi 81).
The primacy of usage and habit is asserted many times by Sbawayhi,
and there is no better illustration (because one senses a tongue in cheek
here) than his discussion of expressions of praise and blame. They are
not unconstrained, he says, you must follow the speech habits of the
Arabs, so you cannot, for example, praise someone for being a tailor
or a seed-merchant, still less praise a person in terms normally used of
God, e.g. al-h amdu li-zaydin praise be to Zayd!. Sbawayhi, or pos-
sibly a commentator, allows himself a pun here by saying that would
be a grave sin, am, playing on tam magnification, the name for
this laudatory construction (Der. I:214f/Bl. I:251). And although it is
correct to use such attested idioms as he is as close to me as where my
waist-cloth is tied huwa minn maqida l-izri, you cannot say he is as
close to me as where the horse is tethered *huwa minn marbita l-farasi
(Der. I:174/Bl. I:206). Ab l-H usayn discusses the interdependence of
meaning and use in a similar way (e.g. Mutamad I:17f, 2228), and it is,
of course the central problem of usl semantics, as Ali (2000) demon-
strates in great detail.
10
The example is perhaps deliberately perverse, as the natural reading would be I
saw the good man Zayd, but this is in a chapter on verbs of the heart, so ra must have
the complete sentence Zayd [is] the good man as its direct object.
Finally Sbawayhis recognition of the performative and illocutionary

uses of speech can be compared with the far more elaborate and self-
conscious expressions of the same ideas in the legal works, written after
the lawyers had had time to absorb Aristotles categories of sentences (a
post-Sbawayhian innovation), which include vocatives, requests, com-
mands and entreaties.
Thus when Sbawayhi defines the vocative noun in y abdallhi as
made dependent (nasb) through the suppression of the verb which [in
this context] is not expressed (Der. I:262/Bl. I:303, nasbun al idmri
l-fili l-matrki ihruhu), he does leave it open for others to supply a
verb such as und I call out to, though his choice of the verb here
is significant, and in fact there are situations where a verb cannot be
restored, e.g. subh na llhi glory be to God, for which there is no verb
available to be suppressed. This is enough (especially when taken with
Sbawayhis other observations on elliptical utterances, see above with
xayra maqdamin) to permit the assumption than he was aware of the
performative aspects of y and other speech elements. In the usl treat-
ment, those who did regard y as representing an elided verb such as I
call are roundly rebuked by Ab l-H usayn (Mutamad, I:20). He firmly
rejects this on the grounds that merely to say the word y and its noun
conveys in itself the information that an act of calling has occurred. Here
he makes explicit what is left implicit in Sbawayhi, as so often happens.
By Ab l-H usayns time it had become a polemical issue, and he accuses
those who would restore a verb in this situation of either oversimplify-
ing for pedagogical purposes (how condescending!) or misinterpreting
what Sbawayhi meant by h arf nid a particle of calling, a performative
element by nature, which needs no verb, as Ab l-H usayn observes.
What this article has tried to show is that Sbawayhi, perhaps by his
own genius, perhaps encouraged and inspired by the community of
pioneering Islamic scholars around him, chose to treat language in a
manner which shows a remarkable affinity with modern Pragmatics.
The nine points which Ali (2000, 3f) regarded as essential pragmatic
insights are taken from usl sources, but equivalents to most of them
can easily be found in the Kitb. Indeed it might have made a tidier ver-
sion of this paper simply to match them seriatim with material from the
Kitb.
These ideas were, so to speak, embryonic in Sbawayhi, but the mar-
riage of Aristotelian logic and Arab linguistic studies in the 4th/10th
century led to the birth of a much more self-conscious and program-matic
version of the same attitude in legal science, reaching its peak of
42 michael carter
development in the 7th/13th century works which provide the bulk

of the material for Ali (2000). Symbolically we may point to the term
qnn law as an indication of the change of direction, though the term
has never carried much weight in medieval Islm. It would not have
meant much to Sbawayhi, who had half a dozen terms for the correct
use of Arabic, one of them sunna, nor is it prominent in medieval legal
contexts, but perhaps because it was circulating in the 4th/10th century,
e.g. in Frb (d. ca 339/950) Ih s 10, it may have fostered the notion
that a systematic code of behaviour along the lines of Greek ethics was
desirable and possible. As with grammar from that point on, works on
usl al-fiqh display an impressive mastery not only of the large body of
historical Islamic data but also the methods of logical enquiry and dia-
lectical disputation. The two disciplines henceforth continue in paral-
lel, grammar becoming increasingly legalistic to stop the language from
changing, and jurisprudence becoming more and more a grammatical
analysis of the unchangeable texts.
One result is that it was possible to formulate a set of five principles of
communication which, as set out by Ali (2000, 64), present an instruc-
tive analogue to Grices four maxims, here briefly paraphrased:
1. The speakers disposition to make his intention manifest.
2. The speakers truthfulness.
3. That what is said should have its due effect.
4. That what is said should be grasped immediately.
5. That the existing conventions should be maintained.
With great diligence and subtlety the classical usls applied these and
other principles to the interpretation of Gods speech in search of
Gods law as it has been put. This had not been Sbawayhis goal, but
his exhaustive description of how the Arabic language works between
speaker and listener prefigured the Pragmatics of the usls and pro-
vided a basis (filtered through the subsequent grammatical tradition)
for their scholarly exertions.
In a sense there was always an implicit Pragmatics in Arabic grammar
and Islamic law. On the macro-level Islm itself is a covenant with God,
and Islamic society is a kind of social contract in which everything said
between Muslims has a contractual dimension, exactly as in Grices per-
ception of speech. That is one reason why Muslims are enjoined not to lie
to each other, particularly when transmitting the religious knowledge on
which the survival of their faith depends. Sbawayhi could afford to take
truthfulness for granted, since it does not affect linguistic form, but the
usls had to stipulate truthfulness as a condition of a valid legal utter-

ance precisely because there was no way to tell from the words them-
selves whether the speaker was lying, i.e. breaking Grices unenforceable
maxim of Quality, where lying is likewise formally undetectable.
The last point to make is that all this legal matter found in the Kitb
goes to confirm two other aspects of Sbawayhis life and work which tie
him closely to the lawyers. Firstly, from what little we know of his biog-
raphy we can deduce that he associated not only with revered authorities
on Arabic but also with early legal and religious scholars, and secondly
it is evident from the Kitb that these scholars were the inspiration for
much of his technical vocabulary and methodology. While this does not
solve the problem of the origins of grammar completely, it does make
it more likely that the Kitb is the creation of a single mind, an unpre-
cedented description of Arabic in all its domains, religious, poetic, pub-
lic and private, in a theoretical framework which drew deeply upon the
principles of the nascent legal system and owes almost nothing to exter-
nal traditions. This view may not meet with the approval of the dedica-
tee of the present volume, but it is a tribute to his belief in the right of
dissent that such heresies can be published without fear of legal action.
References
2.1 Primary sources

Ab H usayn, Mutamad = Ab l-H usayn al-Basr, Kitb al-Mutamad f usl al-fiqh.
M. H amdullh with M. Bekir and H . H anaf, eds. Damascus, 1964.
al-Bqilln, Taqrb = Ab Bakr Muhammad b. at -Tayyib al-Bqilln, at-Taqrb wa-l-
ird as-sar. Abd al-H amd b. Al, Ab Zayd, ed. Beirut, 1998.
al-Frb, Ih s = Ab Nasr Muhammad b. Muhammad, Ih s al-ulm. A. Gonzles
Palencia, ed. Madrid, 1932.
Ibn Mut, Fusl = Ab l-H usayn Yahy b. Abd an-Nr, Ibn Mut , al-Fusl al-xamsn.
Mahmd Muhammad at -Tanh, ed. Cairo, 1976.
Ibn Walld, Intisr = Ahmad b. Muhammad b. al-Wald, Ibn al-Walld, Kitb al-Intisr.
Monique Bernards, ed. Changing Traditions. Al-Mubarrads Refutation of Sbawayh
and the Subsequent Reception of the Kitb, Arabic pp. 1212. Leiden, New York, Kln,
1997.
ar-Rummn, arh = Ab l-H asan Al b. s ar-Rummn, arh Kitb Sbawayhi. MS
Feyzulla 1984.
a-aybn, Jmi = Muhammad b. al-H asan a-aybn, al-Jmi as-sar f l-fiqh. Mar-
gin of Ab Ysuf, Kitb al-xarj. Blq, 1884.
Sbawayhi, Kitb = Ab Bir Amr b. Utm n Sbawayhi, al-Kitb. (1) Hartwig Deren-
bourg, ed. Le livre de Sibawaihi. Paris, 18811889. Repr. Hildesheim, 1970. (2) Kitb
Sbawayhi. Blq Press, 18981900. Repr. Baghdad, 1965. (3) Kitb Sbawayhi. Abd
as-Salm Muhammad Hrn, ed. Cairo, 196877 (cross-paginated with the Blq
edition). References to the Kitb are usually in the form Der/Bl.
44 michael carter
as-Srf, arh = Ab Sad as-Srf, arh Kitb Sbawayhi. MS Atif Efendi 2548.
a-irbn, Nr = Muhammad a-irbn al-Xatb, Arab Linguistics, an introductory clas-
sical text with translation and notes [Nr al-sajiyya f h all alfz al-jurrmiyya].
Michael G. Carter, ed. Amsterdam. 1981.

Ali Mohamed M. Yunis. 2000. Medieval Islamic Pragmatics. Sunni legal theorists models
of textual communication. Richmond.
Bergstrer, G. 1935. G. Bergstrers Grundzge des islamischen Rechts: bearbeitet und
herausgegeben von Joseph Schacht. Berlin, Leipzig.
Bravmann, Meir. 1953. Studies in Arabic and General Syntax. Cairo.
Buburuzan, Rodica. 1993. Exclamation et actes de langage chez Sbawayhi. Revue Rou-
maine de Linguistique 38, 421437.
Carter, Michael G. 2002. Patterns of reasoning: Sibawayhis analysis of the h l. Pro-
ceedings of the 20th Congress of the Union of European Arabists and Islamicists, Part
One, Linguistics, Literature, History [= The Arabist, vol. 2425]. K. Dvnyi, ed. Buda-
pest. 315.
. 2003a. Legal Schools and Grammatical Theory. Arabistikai islamoznanie. Tom 2.
Studi po sluchai 60godishnata na dots. d.f.n. Penka Samsareva, Simeon Evstatiev, ed.
Sofia. 177183.
. 2003b. Talking with and about God, Adam and the Arabic language. Majz,
culture e contatti nellarea del Mediterraneo. It ruoli dell Islam (21st Congress of the
Union of European Arabists and Islamicists, Palermo 2002) [= La Memoria vol. 15].
Antonino Pellitteri, ed. Palermo. 197208.
Crystal, David. 2000. A Dictionary of Linguistics and Phonetics. 4th ed. Oxford.
Fleischer, Heinrich Leberecht. 188588. Kleinere Schriften, gesammelt durchgesehen und
vermehrt von A. Huber, prof. Thorbecke, und F. Bhlau. Leipzig. Repr. Osnabrck.
1968.
Grice, H. Paul. 1989. Logic and Conversation. Studies in the Way of Words, 2240.
Cambridge, Mass. and London.
Khadduri, Majid. 1987. Al-Imm Muh ammad ibn Idrs al-Shfis al-Risla f usl al-
fiqh. Treatise on the foundations of Islamic jurisprudence. Translated with an introduc-
tion, notes and appendices. 2nd ed. Cambridge.
Larcher, P. 1990. lments pragmatiques dans la thorie grammaticale arabe post-clas-
sique. Studies in the History of Arabic Grammar II. Kees Versteegh, Michael G. Carter,
eds. Amsterdam. 193214.
. 1998. Une pragmatique avant la pragmatique: medivale, arabe et islamique.
Histoire, Epistmologie, Langage 20, 101116.
Loucel, Henri. 1963, 1964. Les origines du langage daprs les grammairiens arabes.
Arabica 10, 188208, 253291; 11, 5772, 151187.
al-Marnn, Burhn ad-Dn. 1870. The Hedaya or Guide: a commentary on Musulman
laws, trans. by Charles Hamilton, 2nd edition. Standish G. Grady, ed. London (refer-
ence is to the reprint 1963).
Moutaouakil, Ahmad. 1990. La notion dactes de langage dans la pense linguistique
arabe ancienne. Studies in the History of Arabic Grammar II. Kees Versteegh, Michael
G. Carter, eds. Amsterdam. 229238.
Simon, Udo. 1993. Mittlelalterliche arabische Sprachbetrachtung zwischen Grammatik
und Rhetorik. Heidelberg.
Troupeau, Grard. 1976. Lexique-index du Kitb de Sbawayhi. Paris.
Versteegh, Kees [C.H.M.]. 1977. Greek Elements in Arabic Linguistic Thinking. Leiden:
E.J. Brill.
ID MR IN THE MAN OF AL-FARR:
A GRAMMATICAL APPROACH BETWEEN DESCRIPTION
AND EXPLANATION
Kinga Dvnyi
Corvinus University, Budapest
1. Introduction
Kees Versteegh stimulated the discussion on the history and develop-

ment of Arab grammatical thinking in a number of his publications.
In one of his books (Versteegh 1993, 150), he reflects upon my earlier
analysis of al-Farrs linguistic methods in his Man (Dvnyi 1990).
He pointed out the insufficient analysis of idmr and its related terms
in this authors work. In another chapter of the same work (Versteegh
1993, chapter five), he assembled data in an attempt to present the inter-
relationship between grammarians, readers and commentators who
worked in the 2nd/8th century.
The present contribution would like to pick up these two threads and
examine, on the one hand, in some detail the role of idmr in the Man
l-Qurn of al-Farr and, on the other hand, analyze the role and place
of this grammatical commentary of the Qurn from the point of view of
other grammars (mainly Sbawayhis Kitb) and other exegetical works,
like for example those of al-Axfa and at- Tabar.
It is a well-known fact that al-Farr (d. 207/822) held in great esteem
Sbawayhis (d. 180/796) Kitb, which more than twenty years predated
his composition. Their starting points and approaches were, however,
widely different. Versteegh (1993, 180) has already pointed out that the
interests of the two authors lay elsewhere. To this, we can add that al-
Farr and Sbawayhi, though working within the framework of prac-
tically one grammaror one grammatical idealhad widely different
aims. While al-Farr, in his Man, used his grammatical knowledge
for the analysis of an existing corpus which he described from the point
of view of the listener to this text, Sbawayhi aimed at creating, from
the point of view of the speaker, a comprehensive grammar in which he
used poetical and Qurnic excerpts only by way of illustration.
46 kinga dvnyi
2. An overview of al-Farrs methods
In order to have a brief overview of the methods used by al-Farr, a

few examples from the beginning of the Man will be presented first.
That the corpus al-Farr is working on is the text of the revelation has
special importance. On the one hand, al-Farrlike grammarians and
later rhetoriciansconsiders that irb is a necessary prerequisite for
the understanding of any text and so the text of the Qurn. On the other
hand, since the Qurnic text is usually understood without relying on
the irb endings, there is a strong tendency to analyze different end-
ings at a given place without entailing a change in the meaning. 1 This
method aims at eliminating the problems posed by the different qirt.
Although in most of the cases the grammatical analysis only underlines
and systematizes the interpretation given by the first exegetes of the
Qurnic text, in several cases, however, it is the grammatical analysis
which helps to disclose the meaning of the ya.
2.1 Man I:3 regarding Q 1:2 al-h amdu li-llhi

The task here is to determine the vowel ending of the word al-h amd.
Step (1): The examination of the readings (qirt):

According to al-Farr the readers are in total agreement concerning the
raf ending.2 It is also interesting to note, that al-Farr only mentions
this ending but does not present a grammatical explanation for it.3
Step (2): The elicitation of extra-textual linguistic source:

The Bedouins (ahl al-badw) say three things:
(a) al-h amda li-llh,
(b) al-h amdi li-llh,
(c) al-h amdu lu-llh.
1
A notable exception, where different readings reflect a difference in meaning is e.g.
Q 5:6. For the analysis of this ya, see Dvnyi 198788 and Burton 1988.
2
Makram and Umar (1985, I:5) also list al-h amda and al-h amdi among the readings
of this verse.
3
This is in contrast with al-Axfa (Man I:9 ff.) whose analysis at this place is rather
similar to that of al-Farr, but who also provides a detailed grammatical analysis of the
raf ending.
R in the MAN of al-farr
IDM 47
Step (3): The explanation of the variants:

For (a), i.e. the nasb ending, al-Farr
(i) gives a grammatical rule according to which h amd is a masdar
in place of which a verb could also have been used. This is an
example that fixes the rule;
(ii) supports it with other similar instances in the Qurn, like:
Q 47:4: fa-id laqtumu lladna kafar fa-darba r-riqbiinstead
of which one could say in kalm: fa-drib r-riqba
Q 12:79: mada llhi an naxuda . . .which is the same as:
nadu bi-llhi;
(iii) and props it with Bedouin usage (qawl al-arab) where saqyan
laka may be used instead of saqka llh.
The (b) variant, i.e. al-h amdi li-llh is treated as one word. As such, it is
compared to ibil, where two i vowels follow each other. This is a descrip-
tive explanation making reference to usage. Other parallels are pre-
sented as well, among them the (c) variant: al-h amdu lu-llh. Al-Farr
approached this corpus from the point of view of the listener to the text.
And it seems from the second step employed by al-Farr that the listener
could hear some variants. And because al-Farr was interested not only
in the text of the Qurn, but obviously placed it in the context of the
Arabic language as a whole, he analyzed these versions. What is even
more, he proceeded to explain these variants in detail thoughaccord-
ing to his knowledgethese variations were not Qurnic readings.
2.2 Man I:78 regarding Q 1:7 [. . . anamta alayhim] ayri

l-madbi [alayhim wa-l d-dllna]
The task here is to determine the vowel ending of the word ayr. There
are neither variant readings nor extra-textual variants.
Step (1): Grammatical analysis:

Al-Farr explains the i in ayri as a nat to alladna, mentioning that it is
definite (marifa) because of the following word (al-madbi).4 He also
4
It is important to note in this respect, that while the analysis of al-Axfa goes along
the same lines (Man I:18) (with the usual difference in terminology), but with one
basic difference, i.e. that he does not define negatively the ending of the word, in other
words he does not say what it is not, rather contents himself with saying what it is. That
is to say, he does not deal with the refutation of grammatically incorrect endings or with
48 kinga dvnyi
categorically refuses to connect ayri to the preceding word, alayhim.

Doing this he seems to argue against the view of those among his con-
temporaries who carry the surface descriptive analysis or the analysis
based upon proximity (itb) to the extremes.
Step (2): Semantic analysis:

Turning to the semantic side of the explanation, he states that ayri is
connected to wa-l d-dllna which follows it and its meaning is l. He
also adds that if it meant siw, it could not have been followed by l. So
it can be established that al-Farr connects the two types of analysis, the
semantic and the grammatical.
2.3 Man I:1112 regarding Q 2:2 [dlika l-kitbu l rayba fhi]

hudan li-l-muttaqna
The question is whether hudan is in raf or nasb. Though in the case
of this particular word both endings are realized in the same surface
form, the question should be decided both from the point of view of the
semantic interpretation of the structure and for the sake of other similar
structures in the Qurn where even the surface realization is different.
In the case of similar phrases in the Qurn a number of qirt have
been preserved with readings in both raf and nasb, as e.g. Q 31:13 (
lm tilka yt al-kitb al-h akm hudan wa-rah matun/an li-l-muh sinn)
and Q 11:72 (a alidu wa-hd bal ayxun/an).
Step (1): Grammatical statement:

Both endings can be explained in different ways (wajh). If the interpreter
of grammatical structures does not deviate from the intended mean-
ing, he may freely choose between raf and nasb or the different ways.
This freedom is expressed by al-Farr in the use of the 2nd person: id
aradta . . ., wa-in jaalta . . . rafata . . . (Man I:11, 12ff.).
Step (2): The detailed grammatical analysis:

(i) raf
(1) It can be the xabar of dlika l-kitb whichin this caseis ana-
lyzed as the mubtada. A paraphrase is given: dlika hudan.
the refutation of grammatically not permissible analyses. This difference will remain
characteristic throughout the two books.
IDM 49
(2) It can be a nat (modifier) of the xabar: l rayba fhi. A Qurnic

parallel is quoted: wa-hd kitbun anzalnhu mubrakun (Q
6:92, 6:155).
(3) It can also be considered the beginning of a new structure which
comes after a complete phrase (al stinf li-tamm m qablahu).5
A Qurnic parallel is quoted: wa-hd bal ayxun (Q 11:72).6
(ii) nasb al l-qat
(1) It is either cut off from al-kitbu, which in this case would be
analyzed as the xabar of dlika.
(2) Or it is cut off from the -hi in l rayba fhi. A paraphrase is given
as: l akka fhi hdiyan.
A ready-made grammatical rule is given as an explanation for both pos-

sibilities: li-anna hudan nakira ittasalat bi-marifa qad tamma xabaruh
fa-nasabtah li-anna n-nakira l taknu dallan al l-marifa.
2.4 Man I:1415 regarding Q 2:16 fa-m rabih at tijratuhum

The task is to define the reason why certain structures are permissible in
contrast to other seemingly similar structures that are not.
Step (1): Parallels from kalm al-arab:

hd laylun nimun
Step (2): Parallels from the Qurn:

Q 47:21 fa-id azama l-amru
Step (3): Semantic definition

The permissibility of specific structures greatly depends upon their
communicational value, i.e. they can only be permitted if they can be
understood unambiguously (ulima manhu I:14, 17). In these exam-
ples, the reference of the verbs and the adjective is unambiguously not to
5
In connection with this third analysis, attention should be called to the difficulties
of grammatical and semantic analysis arising from the lack of punctuation. No wonder
that writings on qat and istinf developed into a special branch of the Qurnic sci-
ences. On the use of istinf and related terms in early grammar and exegesis in general,
see Versteegh 1993, 132136 and the literature cited there.
6
The reading in nasb can be found at this place in todays printed editions, and this
was al-Farr s reading as well (Man I:12, 2). The raf ending was read by Ibn Masd
and Ubayy among others (see Makram and Umar 1985, III:125).
50 kinga dvnyi
the other word in the phrase which is a common noun (tijra, layl, amr)
but to the people who act behind these words. Similar structures are
not permitted in the case of possible ambiguity, e.g. *qad xasara abduka
is not permitted because of the ambiguity of the word abd (he can both
be trader and the object of trade), as al-Farr puts it, its meaning cannot
be known: fa-l yulamu manhu (Man I:15, 3).
2.5 Man I:16 regarding Q 2:1718 wa-tarakahum f ulumtin . . .

summun bukmun umyun
The task is to explain the endings of summ, bukm and umy.
(i) The raf ending is explained by istinf and the completeness of the
preceding clause in itself (li-anna l-kalm tamma). An additional
reason is that the whole phrase is split between two yas, but this is
not a prerequisite of istinf. It should, however, be mentioned that
istinf is frequently found at the beginning of yas.
(ii) The nasb reading (summan bukman umyan) is explained in two
ways:
(1) according to the meaning (al l-man), i.e. by referring back
to the verb in the preceding clause (tarakahum);
(2) by their being expressions of blame (damm).
A parallel Qurnic passage quoted by al-Farr is Q 9:111112:
wa-man awf bi-ahdihi mina llhi
. . . at-tibna l-bidna l-h midnaexplained as istinf
or the reading in nasb
. . . at-tibna l-bidna l-h midnaexplained as qat
The use of the explanatory terms istinf and qat shows clearly that for
al-Farr the most important issue was to understand the text as a whole.
And one of the first issues to be dealt with was the correct segmentation
of the text. The term qat expresses more clearly the syntactic structure,
whereas the term h l can only refer to the meaning.7
2.6 Man I:17 regarding Q 2:19 yajalna asbiahum . . . h adara

l-mawti
Meaning, however, can also play a decisive role in the explanation of an
irb ending and the determination of the syntactic structure, as it is the
7
For the difference in al-Farrs usage between hl and qat, see Kinberg 1996, 194.
IDM 51
case, e.g. in Q 2:19 (Man I:17, 1ff.): yajalna asbiahum . . . h adara

l-mawti. The word h adara is in nasb not because the verb yajalna
affects it (*yajalnah h adaran) but by way of specification (tafsr). It is
the same as e.g. in the phrase ataytuka xawfan where fear is not given
but something is given because of it.
3. The explanatory technique of idmr in the Man
The above examples were presented to show the steps followed by

al-Farr in his explanations.
One of the basic explanatory techniques used by al-Farr in the
Man is the term idmr (together with its related forms). In the course
of the explanation of the text of the Qurn, al-Farr is basically con-
fronted with three types of problems:
(i) The meaning of the ya needs clarification;

(ii) Though the meaning is clear, there is a problematical irb-ending;
(iii) The qirt of similar structures at different places in the Qurn
should be harmonized and explained.
It will be seen that the technique of idmr features as a prominent solu-

tion in all of these cases. It is outside the scope of the present paper
to examine all the occurrences of idmr and its related terms in the
Man. In the following, however, we shall try to present a typology of
the most significant types of occurrences of this term.
3.1 idmr is supported by another place in the Qurn where the

suppressed element occurs in the text
3.1.1 Man I:13 regarding Q 2:7 xatama llh al qulbihim wa-al
samihim wa-al absrihim iwatun
This ya has a variant reading, iwatan, which entails no differ-
ence in the meaning of the ya (wa-manhum wh id ).8 This read-
ing is explained by otherssays al-Farrby the idmr of jaala. This
8
It might be interesting to note that al-Axfa did not mention the existence of differ-
ent readings here (see al-Axfa, Man I:34).
52 kinga dvnyi
is supported by Q 45:23 where the phrase actually occurs as wa-jaala

al basarihi iwatan. Al-Farr accepts this explanation, i.e. the use of
another Qurnic place to support the explanation by idmr, though he
emphasizes that were it not for the Qurnic parallel, the idmr in this
sentence could not have been accepted. This is because according to
al-Farr, idmr is acceptable only in those utterances (kalm) which
are coherent ( yajtami), i.e. where the beginning refers ( yadullu al) to
the end. The idmr is acceptable (yah sunu) if it is well known (urifa; or
at other places malm).
A kalm example is brought in to illustrate what is meant by the
term yadullu al, and how it operates in structures. The example is as
follows: qad asba fulnun al-mla fa-ban d-dra wa-l-abda wa-l-
ima wa-l-libsa l-h asana. It can be seen from this example, proceeds
al-Farr, that the notion of building (bin) cannot be extended to
slaves and clothes. By quoting this example from kalm, al-Farr appeals
to the listeners linguistic insight and competence in the Qurnic text.
He explains that the first part of the utterance (asba) refers to the
slaves, etc., i.e. qad asba fulnun al-mla fa-ban d-dra wa-[asba]
l-abda wa-l-ima wa-l-libsa l-h asana. Without saying that it should
be repeated, instead he uses the term dalla, yadullu al. Sbawayhi would
rather say explicitly that asba should be repeated before al-abda.
Al-Farr only hints at the idmr of asba.
In the following (Man I:14), Q 56:22 is mentioned as a similar
example: yatfu alayhim wildnun muxalladna bi-akwbin . . . wa-
h rin nin.9 Al-Farr accepts the -in reading on the basis of proxim-
ity (itb) but mentions that those who put it in raf (wa-h run nun)
do this on the basis of the meaning, since the cups cannot be co-ordi-
nated with the beautiful companions, and by the idmr of a word like
indahum, fh, maa dlika. Again, as in the previous example from
kalm, al-Farr does not use the term idmr here, but only hints at it.
The limits of idmr are imposed by understandability and coherence,
or rather their lack. This is called by al-Farr qillat al-ijtim. So. e.g. the
idmr of qataltu is not permissible in the following utterance: darabtu
fulnan wa-fulnan [*qataltu]. The reason is that there is no reference
9
Mainly because he considered it the accepted reading. This was the reading of a
great number of readers, among them al-Kis (cf. Makram and Umar 1985, VII:65),
but in todays printed mush af the -un reading can be found.
IDM 53
(dall) to the meaning here, the speakers intention cannot be known by

the listener.
At- Tabar adds (Tafsr I:87) that xatama cannot refer to the eyes
because they are never described by xatama either in the Qurn or in
the kalm.
3.1.2 Man I:35 regarding Q 2:72: wa-id qataltum nafsan

fa-ddratum fh
The jawb of id is present neither in this ya nor several others simi-
lar to it in the Qurn. Al-Farr explains the structure by the idmr of
udkur before id, and supports this explanation by similar utterances
in the Qurn where this word is present, as e.g. in Q 8:26 (wa-dkur id
antum qall . . .). Suppression of udkur with id can be inferred from
places like Q 8:26 and Q 7:86. al-Farr adds that if udkur was not
mentioned in Q 7:86, you could still infer that this is what is intended
because it occurs previously.
A parallel case of idmr is also mentioned: Q 7:73 (wa-il Tamda
axhum Slih an) where the known meaning allows the explanation of
this structure by the idmr of arsaln, i.e. arsaln Slih an. This exam-
ple in itself can be regarded as a sub-case of the acceptance of idmr on
the basis of parallel Qurnic passages, since though the verb arsala is
used in the Qurn, it is mainly the content of the ya which determines
the selection of the suppressed element.10
3.1.3 Man I:166 regarding Q 2:246 wa-m lan all nuqtila

Al-Farr starts his analysis with comparing similar utterances from dif-
ferent parts of the Qurn, some of them with an and some other with-
out it, like e.g.:
Q 14:12: m lan all natawakkala al llh

Q 57:8: wa-m lakum l tuminna bi-llhi wa-r-raslu yadkum li-
tumin bi-rabbikum
Dropping (ilq) an here cannot be considered deficiency (illa) accord-

ing to al-Farr because of its frequent usage in the arabiyya. The usage
of an relies on the meaning of manaa. According to this explanation,
10
On the meaning of this term in the Man, see Kinberg 1996, 377.
54 kinga dvnyi
the utterance m laka l tusall f jama means: m yamnauka an

tusalliya. The an may be inserted into the m laka phrase because the
latters meaning is identical with manaa. And a last evidence is put
forward from the text of the Qurn itself: Q 7:12: m manaaka all
tasjuda id amartuka runs parallel to Q 15:32: m laka all takna maa
s-sjidna.
3.2 idmr is supported basically by kalm
3.2.1 Man I:32 regarding Q 2:48 wa-ttaq yawman l tajz nafsun

an nafsin ayan
Al-Farr explains this passage by the idmr of fhi after tajz, insisting
on the necessity of distinguishing by reference to the meaning adver-
bial structures allowing idmr from verbal structures with sifa (prepo-
sitional phrase) complement, where idmr is not possible. For example,
anta llad takallamtu fhi should be distinguished from anta llad takal-
lamtu. In doing this he argues with al-Kis and others who understand
too rigidly the formal description and identify the two, formally simi-
lar but semantically different types. On the basis of this identification
al-Kis refutes both structures with idmr while others permit both
of them. The same holds true, says al-Farr, for the fhihu alternation
within adverbial structures of time and place but not in verbal prepo-
sitional phrases. The phrase tka yawma l-xams is interchangeable
with tka f yawmi l-xams, since the sifa and h here agree in meaning
(muttafaq manhum). But when the meaning differs it is not allowed
to suppress f in place of h and vice versa: yuh ibbuh does not equal
with yuh ibbu fh.
at-Tabar, who quotes extensively (Tafsr I:203, 11ff.) al-Farrs inter-
pretation, adds that the idmr in this ya is possible because it is well-
known (malm), this being the generally accepted explanation for
idmr by the exegetes (ash b at-tawl).
3.2.2 Man I:113, 9, Q 2:185: wa-li-tukmil l-idda

The li- + verb cannot be art11 (cause) of a previous verb because the ww
blocks its impact. Al-Farr discusses in many places that some particles,
11
On the meaning of this term in the Man, see Kinberg 1996, 377.
IDM 55
especially the ww, do not allow us to suppose a grammatical link with

a preceding regent. This phenomenon is explained on the basis of kalm
examples: jituka li-tuh sina ilayya cannot be transformed to jituka
wa-li-tuh sina ilayya only if you mean (jituka) wa-li-tuh sina ilayya
jituka. This second occurrence of the same verbal phrase is, naturally,
suppressed (al-arab tudxiluh f kalmihim al idmr fil badah).
This rule stated, al-Farr cites many Qurnic places where the
same causal structurewa-li and the idmr of a verbal phrasecan
be observed signifying the result of this cause, e.g. wa-kadlika nur
Ibrhma malakta s-samawti wa-l-ardi wa-li-yakna min al-mqinna
(Q 6:75) where the suppressed (mudmar) phrase is a repetition of the
verbal phrase before ww: araynhu.
At-Tabar (Tafsr II:88, 3ff.) quotes two opinions. One is a verbatim
quotation of al-Farrs view without mentioning his name (qla bad
nahwiyy l-Kfa). The other is the view of bad ahl al-arabiyya which
was rejected by al-Farr. This supports our analysis that whenever
al-Farr rejects a seemingly only theoretical possibility of analysis, he
rejects the opinion of a certain group of people without attributing their
view to anybody.
3.2.3 Man I:271 regarding Q 4:46 mina lladna hd yuh arrifna

l-kalim
The meaning is: man yuh arrifna l-kalim. This is supported by kalm
examples where man is suppressed in the beginning of the utterance
(mubtada al-kalm): minn [man] yaqlu or minn [man] l yaqlu.
This is possible according to al-Farr, because min, i.e. the first part
(awwal) of the utterance refers to the meaning of what has been left out,
being its part (bad). Al-Farr considers it important to note that man
cannot be left out (idmr) as a rule, but only in those cases where the
prepositional phrase refers to it. This latter sometimes may be f too: fh
slih na wa-fh dna dlika.
3.3 idmr is supported by the immediately preceding context (the

first part of the utterance or the previous utterance): Man I:141142
regarding Q 2:220: wa-in tuxlithum fa-ixwnukum
With idmr of the rfi, i.e. the nominal subject: hum, or if it is nasb, with
idmr of the second occurrence of the verbal phrase: fa-ixwnakum
tuxlitna. In both cases the suppressed element is given in the first
56 kinga dvnyi
part of the utterance or, sometimes, in a previous utterance, as in

Q 2:239: fa-in xiftum fa-rijlan. This is called ijtim al-kalm. Here,
says al-Farr, it would not be correct to suppose huwa because it is not
an (everlasting) state (dim) but an action. The previous ya (h fi
al s-salawti . . . wa-qm li-llhi . . .) gives a clue for the explanation.
Hence the meaning can be restored as: in xiftum an tusall qiyman
fa-sall rijlan.
At-Tabar (Tafsr II:208) follows al-Farrs argument, and refers to the
same Qurnic passage, etc. but everything is more detailed. It is as if he
worked from a fuller version of the Man.
3.4 idmr is supported primarily by the readings of Abdallh and

Ubayy: Man I:156 regarding Q 2:240: wa-lladna yutawaffawna
minkum wa-yadarna azwjan wasiyyatan
Here al-Farr presents two readings, nasb (wasiyyatan) and raf
(wasiyyatun), and both of them are explained on the basis of idmr. The
first idmr relies on the rule of the absolute object, the suppressed verb
being a jussive of the same root (li-ys . . . wasiyyatan). He refuses, how-
ever, to accept the supposition that the previous verb in the yadarna
azwjan phrase could have put wasiyya into nasb. This may have been
held by the supporters of the proximity-theory who always preferred the
explanation on the basis of the neighboring word.
The second, raf, reading (wasiyyatun) may also be explained by
idmr. The source of this idmr is the reading of Ibn Masd or Ubayy:12
fa-matun li-azwjikum or kutiba alaykum al-wasiyyatu li-azwjikum,
both containing raf , the first also a lexical variation.
12
From among the 64 Qurn readers whom al-Farr quotes by name, Ibn Masd
is by far the most frequently quoted, with 411 references. Ubayy is the seventh most
frequently quoted reader, with 92 occurrences. Cf. Dvnyi 1991, 160161.
IDM 57
3.5 The use of major, meaning-triggered grammatical rules in

supposing idmr
3.5.1 h l
3.5.1.1 Man I:24 regarding Q 2:28: kayfa takfirna bi-llhi

wa-kuntum amwtan
The meaning is established as qad kuntum, since the second verb is a
h l to the first one and it refers back to a state prior to the past action
denoted by the first verb.
After the definition of the meaning, al-Farr formulates the following
rule: this type of h l necessitates qad, either overtly (ihr) or in a sup-
pressed way (idmr). If this condition is not fulfilled, i.e. the first verb
does not refer to the past (like e.g. in the case of kda, as) then the
second cannot contain qad in either way.
This rule is reinforced by another type of proof, a parallel place in
the Qurn where qad is being suppressed and where it is generally
understood to be necessary for the correct meaning (Q 12:27: in kna
qamsuhu qudda min duburin fa-kadibat).
At-Tabar (Tafsr I:146, 18ff.) follows al-Farrs arguments but he
adds that the explanation for the suppression of qad lies in the fact that
if the verbal form faala takes the place of h l it is self-evident (malm)
that it requires qad. It is to be emphasized that al-Farr also considers
that only those elements can be suppressed which are well-known to the
speakers.
3.5.1.2 Man I:372 regarding Q 7:4: fa-jah basun baytan aw

hum qilna
The meaning is established: [wa]-hum qilna is a h l, parallel to
baytan. Its grammatical rule is: wa-huwa filun, accordingly the idmr
of wa- is compulsory. The wa- should be present either overtly or in a
suppressed way. All this is supported by an example from kalm where
both ways (ihr and idmr) are correct: ataytan wliyan aw wa-an
mazl or awan mazl.13
13
Az-Zajjj (Man II:317) refutes this explanation without mentioning al-Farr
and says that it is not necessary to suppose wa- in h l in general.
58 kinga dvnyi
3.5.2 The delimitation and rules of idmr in quotation (hikya)
3.5.2.1 Man I:38 regarding Q 2:58 wa-ql h itta tun

Considering h itta tun as a h ikya entails the idmr of the rfi (i.e. the
nominal subject) which is supposed to be m umirtum bihi or simply
hiya. Though not called idmr in this place, it is analyzed as such, as will
be seen below.
It is a quotation (h ikya) which presupposes that it should be correct
(saluh a) if the rfi/nsib or xfid is suppressed. Al-Farr exemplifies
this rule as follows: qultu l ilha ill llh fa-yaql al-qil: qultu kali-
matan slih atan. But kalimatan slih atan cannot stand alone without a
nsib.
To decide whether an utterance is quotation or indirect speech is very
important for the Qurnic text where change(s) of speaker and hearer
can occur even within the same ya.
3.5.2.2 Man I:40 regarding Q 11:69: ql salman qla salmun

The word salm is used here in two meanings (al manyayni). In the
first step the two endings (nasb and raf ) are explained. The explanation
of the endings relies on the grammatical rule differentiating indirect
speech from quotation (h ikya). Having fixed this al-Farr adds that
in the case of h ikya the idmr of alaykum should be supposed. In a
third step he demonstrates that this idmr is possible because salm
frequently occurs in kalm alone, i.e. without alaykum.
3.5.2.3 Man I:93 regarding Q 2:154: wa-l taql li-man yuqtalu f

sabli llh amwtun
It is another instance of h ikya where the raf ending is explained by the
idmr of hum.14 The irb ending and the difference between quotation
and indirect speech is demonstrated by substitutional analysis (huwa bi-
manzilat . . .). He says that the nasb ending is not permitted here because
amwt is an ism and not a qawl, i.e. it is not a nominalized phrase.
He illustrates the possible and impossible usages as follows:
qultu laka xayran, i.e. kalman h asanan.
qultu laka xayrun, is similar to: qultu laka mlun.
14
Az-Zajjj (Man I:229) also accepts this analysis.
IDM 59
In the case of xayr both structures are correct whereas their substitutes
cannot be used in the other structure. Since there are no different read-
ings here it might be supposed that by illustrating these structures in
detail al-Farr may possibly refute analyzing techniques that take into
account only the surface structure.
3.5.2.4 Man I:296, 7 regarding Q 4:171: wa-l taql taltatun

This and similar examples show that it is a general grammatical rule that
necessitates the use of idmr, scil. if there is a marf after the qawl (as
a quotation after the verb qla), there must be a rfi (subject) as well, in
the above example: wa-l taql hum taltatun
General rule: if something is in raf after qla (yaqlu, qawl), there
should be a rfi either overtly or in a suppressed way (idmr). Here al-
Farr gives a formal explanation of an irb ending.
3.5.3 The id in the relative clause: Man I:157 regarding Q 2:246

ibat lan malikan nuqtil f sabli llhi
The verb may not be nuqtilu (in raf ) referring to the preceding noun
as a sila, because there is no reference back to malikan. However, if one
accepts the yuqtil reading, it can be explained as either sila ( yuqtilu)
or jaz (yuqtil) after an imperative (amr), as in the case of nuqtil.
Al-Farr shows that the form yuqtilu can also be used when there is
no antecedent, with idmr: ibat lan llad yuqtilu. Then he presents a
kalm example: allimn ilman antafiu bihi and allimn llad antafiu
bihi. But if bihi is dropped, then only the jazm is correct, since there is
no reference back (id).
3.5.4 jaz: Man I:178, 4, Q 2:265: fa-in lam yusibh wbilun

fa-tallun
Al-Farr here refers to the grammatical rule of jaz that requires md
forms, thus he supposes the suppression of kna before tallun. The
interesting point of his analysis is, however, when he says: udmirat kna
fa-saluh a l-kalm, because it sheds light on the meaning of idmr: It is
not simply suppression or deletion but the supposition of a suppressed
element which corrects the utterance and makes it fit the grammatical
rules.
60 kinga dvnyi
3.5.5 amm . . . fa-: Man I:228 regarding Q 3:106: fa-amm lladna

swaddat wujhuhum a-kafartum
This is a good example of a well-known grammatical rule (the obliga-
tory use of fa- after amm) which makes the grammarian suppose the
idmr of fa-. At the same time the meaning of the ya also requires the
supposition of a suppressed phrase yuql. After dropping yuql, how-
ever, the fais also dropped. In reconstructing the meaning one needs to
reconstruct the utterance as well: fa-amm lladna swaddat wujhuhum
fa-yuql a-kafartum.
In the above five examples idmr is necessitated by some basic gram-
matical rules considered by al-Farr to be generally accepted and evi-
dent to such an extent that he does not even try to explain them with the
exception of the last example.
3.6 The impossibility of idmr: Man I:195197 regarding Q 3:15: qul

a-unabbiukum bi-xayrin min dlikum li-lladna ttaq inda rabbihim
janntun
Jannt is in raf because of the influence of lm. This intervening lm
prevents referring back to the beginning of the utterance (awwal
al-kalm).
He then draws the conclusion: the xfid (here bi-) cannot be sup-
pressed (lam yudmar), therefore one cannot suppose that the bi- (in bi-
xayrin) is suppressed before jannt causing it to be in xafd. This also
means that the irb-ending xafd cannot be used independently. The
same principle is stated in connection with Q 6:96 (Man I:346) where
a h l stands in the way of the effect of a preceding verbal noun (jil):
wa-jilu l-layli sakanan wa--amsa wa-l-qamara h usbnan. Since
a-amsa and al-qamara cannot be in xafd, therefore in al-Farrs view
they take nasb ending according to their meaning.
4. Concluding remarks
4.1 A summary of idmr types

To sum up the method of idmr, it can be established that al-Farr
derives the explanations of specific Qurnic passages from the follow-
ing main types of sources:
IDM 61
(i) the end of an utterance is related to its beginning, the preceding

context (sometimes on the basis of the different segmentations of
the text), whereas al-Farr (Man I:13, 1415) puts itwa-
innam yah sunu l-idmru f l-kalmi llad yajtamiu wa-yadullu
awwaluhu al xirihi, i.e. the context is coherent. It is contrasted
with inqataa, where the context is incoherent.
(ii) another place in the Qurn;
(iii) an extra-textual linguistic source: kalm al-arab/badw or some-
times a poem, though it should be pointed out that poems do not
have their central significance they have in the Kitb;
(iv) a grammatical rule often quoted without example or proof. Its main
types are:
(a) only the rule is mentioned;
(b) a rule-like example is given (kam taql . . .);
(c) a grammatically incorrect example is used to support his rea-
soning (l yajz . . .);
(v) a qira;
(vi) the codex of Abdallh b. Masd or, less frequently, that of Ubayy
b. Kab or the codices of their followers which sometimes contain
non-canonical variations.
4.2 The meaning of idmr

The meaning of idmr is not simply h adf (deletion), but rather the
obligatory supposition of an element. Thus when al-Farr writes that l
budda min idmr kna li-anna l-kalm jaz (Man I:178, 4) it means
just the opposite of obligatory h adf (the obligatory deletion); it means
that it is obligatory to suppose the existence of an element either present
in the utterance or suppressed.
The use of the terms uskitat and ulqiyat also sheds light on the main
characteristic of idmr, i.e. its obligatory nature. For example at Man
I:163 regarding Q 2:246 there are two options to express the same mean-
ing: either with all or only l. The dropping of an, however, cannot be
considered idmr, since both variants are equally correct Arabic (al
wajh al-arabiyya), the structure with only l not being brought back to
the one with an.
62 kinga dvnyi
4.3 Two kinds of traditions in the Man

It can be established that al-Farr relied on two kinds of traditions in
his Man.
I The grammatical tradition
A) al-Farr and Sbawayhi

Disregarding the terminological differences it can be stated that on the
whole he followed the same grammatical tradition that became exem-
plified by Sbawayhis Kitb.
a) He takes sides with a Sbawayhi-like grammatical analysis on the basis

of the mil theory, without, however, using the same terminology.15
b) Frequent reference can be found to grammatical rules considered
self-evident and thus in no need of further explanations.
c) We can see the main difference between the Man of al-Farr and
Sbawayhis Kitb not in the method of grammatical analysis but in
their approach. Al-Farr, starting from a complete text, always has
in mind the text as a whole, as a series of utterances, and he proceeds
accordingly. Whereas Sbawayhi cites only specific examples to illus-
trate the grammatical rules he wishes to present.
B) al-Farr and al-Axfa

The difference between al-Farr and al-Axfa lies in the fact that the
latter basically writes about grammar, while the former deals essentially
with Qurnic exegesis on a grammatical basis. This may account for
al-Axfas lack of interest in eliciting all the known readings at a given
place whereas al-Farr seems keen on mentioning whatever can be
heard (see, e.g. al-h amdu/al-h amda/al-h amdi above), whereas does not
seem to attach great importance to the explanation of structures or end-
ings that he considers self-evident.
15
According to Talmon (2003, 309312) Sbawayhis main concern in syntax is irb
carried out by amal effect. Talmon also postulated that in the Kitb al-h udd, al-Farr
seemed to focus in his syntactic description on sentence-types and the determination
of syntactic relations. We can also experience in the Man that while al-Farr dealt
with irb-endings in a somewhat flexible way, he did not make allowances in the case
of syntactic structures.
IDM 63
C) al-Farrs opinion on the linear interpretation of the text

a) At least half of al-Farrs analysis of idmr deals with the interpreta-
tion of different irb endings which he considers fundamental in
clarifying the meaning of the Qurnic text. Therefore he is in con-
stant struggle with those views according to which the short vowel
endings are merely phonetic or morphological phenomena and not
irb endings. The most frequent of these solutions is the choice of
the final vowel on the basis of proximity, itb.16
b) He also introduced some formal descriptive interpretations in which
he might have followed Kfan exegetical methods. He accepts these
explanations if they fit into his concept of the linear descriptive text
analysis. The best examples to demonstrate this are those places
where he insists that certain particles, such as wa-or the preposi-
tions, may block the impact of the previous grammatical structure
(e.g. Q 2:185 as analyzed in 3.2.2 above).
c) al-Farr, however, refuses to accept superficial formal analyses which
run contrary to the meaningful interpretation (e.g. Q 2:48 when he
refuses identification of f in temporal adverbs with f in prepositional
verbs). In such cases he does not say whose opinion, whose analy-
sis is rejected by him. Sometimes he uses the passive (yuql), some
other times it can only be supposed that he is at variance with some
people who only looked upon the text as a linear string of utterances
and derived the explanations quasi mechanically from the preceding
string.
II The exegetical tradition

Al-Farrs grammatical activities in the field of Qurnic exegesis cannot
be overestimated. It becomes especially evident if we compare his expla-
nations to the relevant passages in at- Tabars Tafsr. It goes without say-
ing that the scope of at -Tabars commentary is much wider than that of
al-Farrs work. But if we limit our comparison to the linguistic exege-
sis, we find that al-Farrs explanations are most of the time taken over
verbatim without his name being mentioned. One can also suppose that
16
Al-Farr accepts these variant forms in case of different existing sam in the col-
lected corpus, as e.g. in the case of al-h amda/i/u (Q 1:2). If, however, the difference
appears not in the given vowel ending but in the explanation, then he always advocates
the explanation based on iml, as e.g. in Q 1:7 where he does not accept the expla-
nation according to which the ending of ayri would be determined by the preceding
alayhim.
64 kinga dvnyi
at-Tabar had access to a fuller version of the Man, it is, however, also
possible that al-Farr was only one prominent exponent of a common
thinking about grammatical issues in the Qurn and the analyses he
dictated from his memory belonged to a common stock of knowledge, a
long line of grammatical exegetical tradition, with which at-Tabar was
still familiar. It is interesting to note, however, that the two other Man
works, Ab Ubaydas Majz and al-Axfas Man do not seem to have
been incorporated into what became the definitive commentary of the
Qurn for centuries. So it might be concluded that while the grammati-
cal tradition as it was shaped in Basra outshone the Kfan, its trace in
the grammatical analysis of the Qurn is not significant.17
5. References
5.1 Primary sources

Ab Ubayda, Majz = Ab Ubayda Mamar b. al-Muta nn at-Taym, Majz al-Qurn.
Muhammad Fud Sazgn [Fuat Sezgin], ed. 2 vols. Cairo: al-Xnj, n.d.
al-Axfa, Man = Ab l-H asan Sad b. Masada al-Muji al-Axfa al-Awsat , Man
l-Qurn. Fiz Fris, ed. 2 vols. Amman: Dr al-Bar, 1981.
Farr, Man = Ab Zakariyy Yahy b. Ziyd al-Farr, Man l-Qurn. Ahmad
Ysuf Najt and Muhammad Al an-Najjr, eds. 3 vols. Cairo: al-Haya al-Misriyya
l-mma li-l-Kitb, 19802.
Sbawayhi, Kitb = Ab Bir Amr b. Utm n Sbawayhi, al-Kitb. Abd as-Salm
Muhammad Hrn, ed. 5 vols. Beirut: lam al-Kutub, 19661977. n.d.
at-Tabar, Tafsr = Ab Jafar Muhammad b. Jarr at-Tabar, Jmi al-bayn f tafsr
al-Qurn. 30 vols. Cairo: al-Mat baa al-Maymaniyya, n.d.
az-Zajjj, Man = Ab Ishq Ibrhm b. as-Sar az-Zajjj, Man l-Qurn wa-irbuhu.
Abdaljall Abduh alab, ed. 5 vols. Beirut: lam al-Kutub, 1988.
17
An interesting feature of al-Farr s explanatory method is his usageat certain
placesof the 2nd person singular in the case of irb endings and other grammatical
interpretations. The use of 2nd person singular is on the one hand the usual practice
in Sbawayhis Kitb, but Sbawayhi presents his linguistic analysis from the point of
view of the speaker, the producer of different utterances, whereas al-Farr deals with
a concrete text, and what is even more, the text of the revelation. The use of the 2nd
person singular in this case might indicate that al-Farr considered that the sacred text
was the one without the short vowel endings or that it had been revealed according to
what is termed as sabat ah ruf which leaves the reader of the text some freedom in the
vocalic realization. It should, however, be noted that al-Farrsimilarly to other com-
mentators, or in fact Sbawayhi when he deals with the Qurn (e.g. Kitb II:155, 10 ad
Q 5:69)does use the 3rd person singular or the passive when he deals with different
irb endings in the Qurn.
IDM 65

Burton, John. 1988. The Qurn and the Islamic Practice of wud. BSOAS 51:1:2158.
Dvnyi, Kinga. 198788. Mujwara: A Crack in the Building of irb. Quaderni di
Studi Arabi 56, 196207.
. 1990. On Farrs linguistic methods in his work Man l-Qurn. Studies in the
History of Arabic Grammar II. Proceedings of the 2nd Symposium on the History of Ara-
bic Grammar, Nijmegen, 27 April1 May 1987, Michael G. Carter and Kees Versteegh,
eds. Amsterdam: J. Benjamins.
. 1991. Al-Farr and al-Kis: References to Grammarians and Qurn Readers
in the Man l-Qurn of al-Farr. The Arabist. Budapest Studies in Arabic 34, 159
176.
Kinberg, Naphtali. 1996. A Lexicon of al-Farrs Terminology in his Qurn Commentary.
Leiden: E.J. Brill.
Makram, Abd al-l Slim and Ahmad Muxtr Umar. 1985. Mujam al-qirt al-
Qurniyya. 8 vols. Kuwait: Jmiat al-Kuwayt.
Talmon, Rafael. 2003. Eighth-century Iraqi Grammar. A Critical Exploration of Pre-
Xallian Arabic Linguistics. Winona Lake, Indiana: Eisenbrauns.
Versteegh, C.H.M. 1993. Arabic Grammar and Qurnic Exegesis in Early Islam. Leiden:
E.J. Brill.
ARABIC ALLAD AS A CONJUNCTION:
AN OLD PROBLEM AND A NEW APPROACH
Werner Diem
Cologne
Sunt aliquot quoque res, quarum unam dicere causam

non satis est, verum pluris, unde una tamen sit.
Lucretius: De rerum natura VI 703f.
1. Introduction
This paper is devoted to the Arabic sub-standard phenomenon of the

relative pronoun allad in the function of a conjunction predominantly
meaning that; because. The historical interpretation of this phenom-
enon, which had been noted by Arab purists for the formula al-h amdu
li-llhi llad Praise be to God that as early as the 9th century C.E.,
has been a topic for Arabists during the last fifty years. Considering the
intensity with which the historical dimension of allad that; because
has been discussed, it may seem superfluous to want to take it up yet
again. This, however, would be a rash conclusion as no communis opinio
concerning the origin of the conjunctional allad has ever been agreed
upon.
The aim of this paper is threefold. First, I shall discuss the theories
as to the diachronic aspects of allad as a conjunction in chronological
order. Then I shall present additional early evidence from documen-
tary sources dating from the 11th12th centuries C.E. and later. After
that, I shall present my own theoretical approach combining important
insights of my predecessors with hitherto neglected aspects. Finally, I
shall deal with the origin of the formula al-h amdu li-llhi llad.
A point that I shall not consider in this paper is the general function of
the H amdalah in formulaic expressions used by todays Arab Muslims.
The important role of this formula, simple and expanded, in everyday
Arabic speech is borne out by numerous examples in Piamenta (1979,
247f., Index) and (1983, 209, Index), among them examples of the
68 werner diem
al-h amdu li-llhi llad type with allad as a conjunction meaning that;
because (Piamenta 1979, 89, 172f., 176).
2. Theoretical approaches
2.1 Meir M. Bravman (1953)

In the chapter The development of the psychological (logical) subject-
predicate relation in his Studies in Arabic and General Syntax (1953),
Meir M. Bravmann, after stating that in constructions of the al-h amdu
li-llhi llad type and similar expressions allad can be interpreted as
that, maintains that the construction is already found in Classical Ara-
bic, for which he adduces a verse by Imra al-Qays. This verse need not
be discussed here as it allows of other interpretations. More important
is Bravmanns general theory about the origin of allad as a conjunction
(1953, 41):
[. . .] we have to assume [. . .] relative clauses which do not link up with
the immediately preceding expression (as al-h amdu li-llhi etc.), but are
parts of an independent new sentence or, more exactly, predicates whose
subjectknown from the preceding sentence or from the situationis
mentally supplied but not linguistically expressed [. . .]. [. . .] the sentence
al-h amdu li-llhi llad lam amut would mean: Thank God. [I am one]
who has not died etc.
Spitaler (1963) rejected Bravmanns theory (see below 2.3), and it was
not discussed by other scholars. I tend toward Spitalers rejection of this
theory as it does not explain satisfactorily the phenomenon in question,
and therefore I shall disregard it.
2.2 Joshua Blau (1961)

In his grammar of Judaeo-Arabic, Joshua Blau treats the conjunctional
allad in 346, which begins with the words allad introducing noun
clauses.1 The paragraph falls into five sections according to the syntactic
status of the allad clauses: (a) attributive clauses, (b) subject clauses,
(c) predicate clauses, (d) object clauses, (e) clauses in which allad has a
causative function. Most of the examples refer to past events, but there
1
This and the following translations from the Hebrew of Blau are mine.
arabic allad as a conjunction 69
are also some referring to future events with the heads of the clauses
expressing prevention and necessity.
As to the origin of the conjunctional allad, Blau begins section (a)
with the following words (1961, 226):
Relatively usual are attributive clauses (apposition), which represent the
transition from relative clauses,
a statement which is continued in a note with the remark (words in
brackets are my additions):
However, this does not mean that allad introducing noun clauses has
originated from attributive clauses only. It is, for example, possible that
from sentences such as al-h amdu li-llhi llad anan Praise be to God
who helped me allad = k, e- [that; because] originated, marking
object clauses or mipne e [because].
These are quite general statements, which hint to the direction where,
according to Blau, a historical interpretation should be looked for. In his
opinion, as can be grasped from both remarks cited above, the origin of
allad as a conjunction has to be seen (a) in attributive relative clauses
and (b) in the al-h amdu li-llhi llad type.
2.3 Anton Spitaler (1963)

Spitalers approach differs fundamentally from Bravmanns and Blaus
as it is monocausal. He starts from the al-h amdu li-llhi llad type and
assumes that this construction was first re-interpreted (Spitaler does not
use this word) as meaning gottlob, da, after which it was generalized.
The following quotations will illustrate Spitalers approach:
al-h amdu li-llhi llad ist der Anfang eines normalen, von dem Wort Allh
abhngigen Relativsatzes, aber aus seinem eigentlichen Zusammenhang
gelst und sekundr in eine neue Konstruktion bertragen. (1963, 101)
Es ist nun ganz eindeutig, dass das stereotyp wiederkehrende al-h amdu
li-llhi im Lauf der Zeit einer Funktionsschwchung unterlegen ist [. . .].
Dadurch verlor das nachfolgende allad zwangslufig den lebendigen
Zusammenhang mit seinem Beziehungsnomen und der ganze Ausdruck
wurde zu einem starren, wenn auch mit einem ganz bestimmten Affekt-
gehalt geladenen Syntagma, bei dem allad nur mehr als berleitendes
Element, als verbindende Partikel, letztenendes eben als Konjunktion
empfunden wurde. Und nunmehr war es natrlich gleichgltig, welche
syntaktische Form der anschlieende Satz hatte. (1963, 235)
70 werner diem
After presenting some examples semantically corresponding to al-h amdu

li-llhi llad praise be to God that which, however, contain finite forms
of h md I and kr I instead of the nominal al-h amdu li- (1963, 105), and
after rejecting Bravmanns above-mentioned theory (2.2), Spitaler con-
tinues with the words:
Mit der Herauslsung aus seinem ursprnglichen legitimen Relativsatz-
gefge und der bertragung in eine ganz neue Konstruktion hat nun aber
unser Typus seine Entwicklung nicht abgeschlossen; es zeigt sich nmlich,
dass die Formel al-h amdu li-llhi als solche zurcktritt und durch andere
Ausdrcke ersetzt wird, die ebenfalls der usserung des Dankes, der
Freude, frohen berraschung, Befriedigung usw. dienen. [. . .] Die dabei
gegebenen phraseologischen Mglichkeiten sind sehr mannigfaltig [. . .].
Sie werden aber noch dadurch bereichert, dass die Konstruktion nunmehr
[. . .] auch fr die damit kontrastierenden Gefhle des Bedauerns, der Reue
usw. verwendet wird. (1963, 109)
There follow in Spitalers article numerous examples of these latter con-
structions both from post-classical writings and modern dialects, all of
them referring to events in the past. Additionally, Spitaler presents some
examples of allad that after verbs from other semantic fields, among
them some where the allad-clause, as in some of Blaus examples, refers
to future events expressed by the Arabic imperfect.
While Spitalers approach to the problem is definitely a monocausal
one, he was well aware of the fact that a multicausal approach would
also in principle have been possible, all the more in view of Hebr. aer
(on its origin see now Rubin 2005, 49f.) and Aram. d-, which, besides
their original function as relative particles, developed the meaning
that; because (1963, 106f.). In his final remarks, Spitaler emphatically
defended his own theory on the grounds: (a) that the Arab grammarians
noted the conjunctional allad only for the H amdalah, which therefore
seems to be central, (b) that the al-h amdu li-llhi llad type possesses the
chronological priority within the evidence of the conjunctional allad,
and (c) that it would be implausible to assume two different starting
points of the conjunctional allad converging into one phenomenon
(1963, 111f.).
There is no reference to Blaus Diqduq (1961) in Spitalers article,
probably due to the fact that he was unaware of it when he wrote the
article. In fact, Blau in some respect anticipated Spitalers theory about
the central role of al-h amdu li-llhi llad.

In his Emergence and Linguistic Background of Judaeo-Arabic (1965),
which appeared four years after Diqduq, Blau is essentially of the same
opinion as in Diqduq (1961) about the origin of allad that, and even
the wording reminds one of that used in Diqduq although it is more
explicit:
Since a relative clause may often be mistaken for a substantive clause, the
transition of allad into a conjunction introducing substantive clauses was
easily accomplished. (1965, 109)
After this general statement, Blau gives two examples: wa-kam alimta
min h adti bni Dwrdy jama m faalahu llad abida s-sanama and li-
llhi l-h amdu llad aslamahu f aydkum, in Blaus translation and like-
wise, you know the story of B. D., everything that he had done, that he
had served the idols and praise be to God that he has delivered him
into your hands. According to Blau, it is in both cases only the word
order that indicates that allad introduces a substantive clause and not
a relative clause, i.e. that allad is a conjunction and not a relative pro-
noun. While the second example indeed illustrates this transition, as
allad can be interpreted both as a relative particle and a conjunction
(praise be to God, who delivered him into your hands and praise be to
God that He delivered him into your hands), this, in my opinion, is not
possible with the first example as it admits of the interpretation of allad
as a conjunction only. From a semantic point of view, the allad clause in
this sentence is an apposition to m faalahu, which means that it cannot
be interpreted at the same time as an attribute to Ibn Dwrdy. Thus, this
example does not represent the starting point of the re-interpretation of
allad but allad in its new conjunctional function.
Spitalers theory is not taken into consideration by Blau, but his article
is mentioned in a footnote (1965, 109).

In contrast with Emergence (1965), where Spitalers article (1963) is
mentioned in a footnote only, having no impact on Blaus theoretical
approach, it is fully considered in the third volume of Blaus monumen-
tal Grammar of Christian Arabic, which appeared in 1967. Blau writes
there:
72 werner diem
alladh opens substantive clauses. Beginnings of this construction already

appear in CA2 and it becomes rather frequent in MA. A. Spitaler [. . .]
has brought out in full relief the early and very interesting history of the
type al-h amdu li-llhi-lladh thank God that and its development. The
use of alladh for introducing substantive clauses in ASP, insofar this is
not due to literal translation of Aramaic d-, is mainly from this type and
its developments. This is all the more remarkable since in JA other sub-
stantive clauses (especially apposition clauses) introduced by alladh are
quite common. It stands to reason that, in accord with Spitalers findings,
this use of alladh first developed in the type al-h amdu li-llhi. Early ASP
exhibits mainly this stage, whereas in later JA other kinds of alladh intro-
ducing substantive clauses have arisen. (1967, 526f.)
Among Blaus examples there is none of the nominal al-h amdu li-llhi
llad type. As to his examples of the corresponding verbal type that con-
tains a reference to God, such as h amida llhi llad or akara llhi llad,
it must be stated that none of them unequivocally represent the new
type with allad as a conjunction because each can also be interpreted
as a relative clause referring to Allh. Of Blaus remaining examples,
many are likewise syntactically ambiguous, while some do represent the
new type. In several cases, Blau himself draws attention to the alterna-
tive interpretation as relative clauses. Apart from that, as Blau remarks
himself, some examples might be calques on the Aramaic d-, which, as
has been mentioned above, is both a relative particle and a conjunction
meaning that; because. Thus unequivocal evidence of the new type is
less for Ancient South Palestinian than might be inferred from the num-
ber of Blaus examples at first glance.
2.6 Manfred Woidich (1980)

Manfred Woidichs article illi als Konjunktion im Kairenischen (1980)
is an investigation into illi, which goes back to allad, as a conjunction in
modern Cairene Arabic. In his masterly analysis, which shows an inter-
est and a competence in syntax not common for Arabic dialectology,
Woidich gives a thorough picture of illi as a conjunction from a syn-
chronic perspective. Woidichs examples are mostly drawn from written
sources, but he also asked native speakers of what is called Educated
Cairene Arabic. It would be beyond the scope of this paper to comment
2
CA = Classical Arabic, MA = Middle Arabic, ASP = Ancient South Palestinian, JA
= Judaeo-Arabic.
extensively on what Woidich found out about illi as a conjunction in

Cairene Arabic. Therefore I shall concentrate on some points that are
also important from a historical perspective.
Woidich begins with the statement that the A (bergeordnete Aus-
drcke), i.e. the heads on which the illi clauses as uS (untergeordnete
Stze) depend, are:
uerungen des Dankes, der Freude, der Befriedigung, des Bedauerns usw.
[. . .] Der Sprecher nimmt jeweils emotional Stellung zu einem im unter-
geordneten Satz (uS) geschilderten Sachverhalt und wertet ihn in diesem
Sinne. [. . .] Wir teilen das gesammelte Material nach der Art der AA ein,
von denen sich zwei Typen unterscheiden lassen. Die einen (Gruppe A)
werten den Sachverhalt des uS direkt, indem sie [. . .] emotional Stellung
nehmen, die anderen (Gruppe B) werten auf eine indirekte Weise, indem
sie angeben, wie eine Person sich oder andere wertet aufgrund des Sach-
verhalts des uS, oder zu welcher affektisch bestimmten Handlung eine
Person durch diesen veranlat wird. (1980, 225f.)
Examples of type A are il-h amdu li-llh illi ultaha b-nafsak Gott sei
Dank, da du es selbst gesagt hast! and kuwayyis illi wiit fi gbak inta
Gut, da es in deine Tasche gefallen ist! Characteristic of this type is
according to Woidich the absence of resumptive pronouns:
Eine Verknpfung des A mit dem uS durch pronominale Rck- und Ver-
weise kann hier nicht stattfinden, da im A keine Pronomen auftreten.
(1980, 226).
Though this statement is corroborated by the above-cited examples, the
last two examples of Woidichs type A do have a pronominal concatena-
tion, among them ayizni lli dayyat il-bazburt-i bti Es rgert mich,
da ich meinen Pa verloren habe. Woidich considers examples of this
type einen bergang zur Gruppe B (1980, 227). In my opinion, such
examples do not belong to group A but rather to group B, as expressions
like ayiz + object pronoun are a kind of pseudo-verbs, the pronoun
being the logical subject, and there are similar examples in group B.
As for type B, examples are d-ana batnaddim illi gt Ich bereue, da
ich gekommen bin and inta karihni lli bal il-h a Du kannst mich
nicht leiden, weil ich die Wahrheit sage. Concerning the syntactic char-
acteristics of this type B, Woidich says:
Logisches Subjekt des A ist [. . .] eine Person, die im uS als Subjekt auf-
tritt. Dadurch kommt eine Verknpfung der beiden Teilstze durch pro-
nominalen Verweis zustande, der ein Charakteristikum fr diese Gruppe
darstellt. (1980, 227)
74 werner diem
Contrary to this statement, the person in the A (i.e., the head) to whom
the illi clause refers is not always the logical subject, as is, e.g., borne
out by the above-mentioned example inta karihni lli bal il-h a, where
the Person, die im uS als Subjekt auftritt is the (logical) object, not
the (logical) subject of the sentence. Furthermore, there are examples of
type B declared by Woidich to be marginal where there is no pronomi-
nal concatenation between the illi clause and an element of the head at
all, e.g. ana mabst illi ma-g I am glad he did not come (1980, 229).
Considering this evidence, it seems to me that the feature of coreferenti-
ality is questionable for Woidichs type B in Cairene Arabic, and it is evi-
dent from Blaus, Spitalers and my evidence that it cannot be sustained
in a general diachronic and diatopic perspective. A certain degree of
coreferentiality is per se bound to exist for the simple reason that when
in the heads of sentences emotions are spoken of the causes of these
emotions as expressed in substantival clauses depending on those heads
are in most cases connected with the persons mentioned in the heads of
the sentences as having these emotions.
Important insights in Woidichs article are that in type B the element
of the head to which the illi clause refers must be human (1980, 230),
that the predicates are faktiv, that is, refer to real facts, mostly past
events (1980, 231), and that this kind of illi can always be replaced by
inn- that, with the difference that illi is considered by the informants
as being strker than inn-, that is, more affective (1980, 234).
2.7 Manfred Woidich (1989)

While Woidichs article of 1980 is essentially descriptive, his contribu-
tion illi dass, illi weil und zayy illi als ob: zur Reinterpretation von
Relativsatzgefgen im Kairenischen, which was published in 1989,
is diachronically oriented. As announced in the title, Woidich distin-
guishes between three different kinds of illi: illi that, illi because and
zayy illi as if .
(a) illi that

The type illi that corresponds to type A of Woidich (1980), e.g., il-h amdu
li-llh illi sabitak gottlob, da sie dich verlassen hat and kuwayyis illi
gt gut, da du gekommen bist (1989, 110f.). As for the origin of this
type, Woidich follows Spitaler (1963) in saying that il-h amdu li-llh in
sentences like il-h amdu li-llh illi waaik fiyya Lob sei Gott, der dich
mit mir zusammengebracht hat:
unterlag einer Funktionsschwche [. . .]. Als starres Syntagma wurde
diese Formel nicht mehr als analysierbar aufgefat, was zur Folge hatte,
da insbesondere N = allh nicht mehr als Nomen gesehen wurde, dem
ein syndetischer Relativsatz angeschlossen werden konnte. (1989, 111)
While in the assumption of Funktionsschwche Woidich explicitly fol-
lows Spitaler, he differs from him in assuming that the al-h amdu li-llhi
llad formula was re-interpreted because there are similar sentences in
Cairene Arabic having the same marked structure of Rhema-Thema,
that is, comment-topic, such as h ilwa di prima ist die! or b ikkalm
da eine Schande sind solche Worte! He sums this up by saying
da die Reinterpretation von illi als Relativpronomen zu illi da ausgelst
wurde durch den Umstand, da die ursprngliche syntaktische Struktur
mit dem eingebetteten Relativsatz nicht der thematisch-kommunikativen
Funktion dieser Stze entsprach. Sie war markiert und wurde durch Rein-
terpretation der bei diesen Stzen blichen funktionalen Satzstruktur
Prd.Subjekt angeglichen. Dadurch konnte dem illi die Funktion einer
Konjunktion zugeordnet werden, die Subjektsstze einleitet. (1989, 115)
Here we have to ask what Woidich intends to demonstrate from a
general Arabic perspective. Does he want to demonstrate a historical
development valid for Cairene Arabic only, or a general development
in Neo-Arabic the results of which are palpable in Cairene Arabic also?
A reconstruction of the first kind would be flawed by the fact that the
Cairene conjunctional allad cannot be detached from the common
history of Neo-Arabic, and a reconstruction of the second kind by the
fact that it would be problematic to reconstruct a common Neo-Arabic
development on the basis of the specific evidence of one modern dialect,
while disregarding other evidence, older and newer. Since we are con-
cerned here with the second perspective only, which is tantamount to a
general reconstruction of the conjunction allad in Neo-Arabic, we have
to ask what Woidichs theory means for the history of the conjunctional
allad in early Neo-Arabic.
In Neo-Arabic the equivalents of Woidichs above-mentioned two
Cairene examples would probably be something like h ilwa hd and ayb
hd l-kalm. Comparing these sentences with a sentence like al-h amdu
li-llhi llad jita, it seems very improbable that the two types of sen-
tences should have been mentally connected by any speaker of Arabic,
76 werner diem
and thus Woidichs theory seems to me to be farfetched. Nevertheless,

his drawing attention to the Rhema-Thema (or comment-topic) struc-
ture, which syntactically corresponds to a predicate-subject structure, is
an important insight, to which I shall come back below when discussing
my own theory (4.3).
Goldenberg comments on Woidichs theory with the words This form
is structurally parallel to the classical Arabic constructions of cleft sen-
tences with m having the same order of constituents (1994, 16/261),
but I doubt the validity of this alleged parallel, as the cleft sentences
mentioned by Goldenberg, apart from their Rhema-Thema structure,
differ fundamentally from the sentences mentioned by Woidich.
(b) illi because

illi because corresponds to type B of Woidich (1980), and examples
are ana farh n illi uftak I am happy because (that) I saw you (1989,
116) or itnaddimit illi gat maya sie bereute es, da sie mit mir gekom-
men war (1989, 117). As for the meaning of illi in the second example,
it remains to be shown that in Cairene Arabic illi in such sentences can
be replaced by the causal conjunctions an and li-ann-. Interestingly,
no transformations of sentences containing expressions of this type are
among those adduced by Woidich (1980, 235) for the replacement of illi
by the causal conjunctions an and li-ann-.
In view of the semantic difference between the two kinds of illi in
Cairene Arabic as seen by him and also in view of the feature Refe-
renzidentitt (that is, coreferentiality) in the second type, which he
thinks is essential, Woidich assumes for the second type (type B in
Woidich 1980) an origin independent of the al-h amdu li-llhi type (type
A in Woidich 1980). This origin he finds in sentences such as ana h mr
illi dafat il-h isb ich bin ein Esel, der ich die Rechnung bezahlt habe!,
which contains a direct relative clause depending on ana. Since in sen-
tences of this kind there exists a causal connection between the relative
clause and the head, Woidich assumes that illi could be re-interpreted
as a causal conjunction, ich bin ein Esel, da ich die Rechnung bezahlt
habe!, adding that this re-interpretation is also true of alle anderen
Stze dieser Struktur (1989, 118). It goes without saying that if we are
to assume this kind of re-interpretation for early Neo-Arabic in general
(and not only for Cairene Arabic), we would have to assume that allad
had been generalized as a relative particle by then, an assumption which
poses no problem.
The fact that Woidich declares his example (13) itnaddimit illi gat
maya sie bereute es, da sie mit mir gekommen war not to belong
to the original type, confirms that, in his opinion, it is in nominal sen-
tences that this type originated (1989, 119). Against this it can be argued
(a) that verbal sentences (such as itnaddamit illi) seem to prevail over
nominal sentences (such as ana h mr illi), and (b) that it seems highly
improbable to me that speakers should have used sentences like ana
h imruni llad addaytu d-dannra I am an ass, (I) who have paid the
dinars, an farh nuni llad najawtu I am glad, (I) who have escaped
or an mutaassifuni llad taaxxartu I am sorry, (I) who was late at all.
For this theory to be accepted, it would be prerequisite to find unam-
biguous relative clauses of this kind in Classical Arabic. I have checked
more than a thousand items with allad in Tradition, many of them in
dialogues, without finding even one example of such constructions, and
I dare say that it is very improbable that they occurred at all.
To sum up, as long as the syntactic type ana h imr illi dafat il-h isb,
which Woidich presumes to be the starting point of this type, is not
shown to have been a normal construction in pre-Neo-Arabic, I con-
sider sentences of this kind in Neo-Arabic the result of a specific devel-
opment, which should be explained otherwise, rather than the origin
thereof.
(c) zayy illi as if

In interpreting this third type, Woidich starts with cases such as ir-rgil
firih bna zayy illi laa lya der Mann freute sich ber uns, wie einer, der
einen Fund gemacht hat, in which the syntagma zayy illi wie jemand,
der; like somebody who could be interpreted as meaning als ob; as if .
In cases representing unequivocally the result of this re-interpretation,
only the interpretation of zayy illi as as if is possible, such as ma-kan-i
f tagwub, zayy illi kunt-i bakallim h agar es gab keine Resonanz, es war,
als ob ich mit einem Stein sprche.
As for earlier evidence of this type, the ka-llad mentioned in gram-
mars, which structurally corresponds to zayy illi, has a different func-
tion, namely that of a conjunction of comparison with real facts, such as
xudtum ka-llad xd You have plunged as they have plunged Qurn
9:69, already adduced by Reckendorf (1921, 192, 4). For more evi-
dence of this ka-llad, see Blau (1967, 527f.), Hopkins (1984, 238) and
Goldenberg (1994, 27f./276).
78 werner diem
An example of ka-llad that functionally corresponds to zayy illi as a

particle of comparison with unreal or hypothetical facts, as described by
Woidich, was adduced by Spitaler (1962, 108) in a footnote: fa-l yazlu
kadlika ka-llad yuh ibbu an yasxara min sh ibih 3 and he continues to
behave in this manner like one (or as if) intending to mock his friend,
sc. the tomcat (as-sinnawr) who plays with the mouse (al-farah) after
having caught her (al-Jhiz, H ayawn V, 202). Spitalers second example
concerns the well-known other ka-llad type: fa-sallaw ka-llad kn
yaf alna and they prayed as they always had. Another example of
ka-llad as a particle of comparison with hypothetical facts is wa-man
axadahu bi-irfi nafsin lam yubrak lahu fhi ka-llad yakulu wa-l
yabau and who takes it (sc. certain money) with haughtiness of mind
will not be blessed regarding it, like one (or as if) eating without getting
replete (al-Buxr, Sah h , Kitb az-zakh, No 1379), and more evidence
of this kind can be found in Tradition. Since in examples of this kind
allad can still be interpreted as a relative pronoun, they represent ka-
llad in the meaning as if in statu nascendi. Obviously, Spitaler did not
realize that his two examples belong to two fundamentally different types.
The first ka-llad is a conjunction corresponding to kam (accord-
ingly it should perhaps be transcribed as kallad), while the second
ka-llad consists of the particle ka- and the relative pronoun of the 3rd
masc. sing. In Cairene Arabic, this second type was grammaticalized as
zayy illi, with zayy standing for ka-, which is unusual in Cairene Arabic.
Future research should pay attention to the existence of two different
types of ka-llad.
It is, then, evident that ka-llad has to be separated from allad that,
which is the topic of this paper. Therefore, these remarks must suffice,
and I shall not return to this special type.
2.8 Gideon Goldenberg (1994)

In his article allad al-masdariyyah in Arab grammatical tradition
(1994), which was reprinted in 1998, Gideon Goldenberg begins with
general observations, then mentions Spitalers article (1962) and Woid-
ichs second article (1989). Goldenberg describes the aim of his paper
as follows:
3
There follow four further an-clauses dependent on ka-llad yuh ibbu. Spitalers quo-
tation has the variant bi-sh ibih.
About the distribution of infinitival allad in Arabic writings in general

I have nothing to add; in the present paper I just wish to adduce some
further evidence for a fuller understanding of the relevant structures as
described and treated in Arabic grammatical literature. Such examination
is important, because syntactical constructions that are considered gram-
matical by the great masters of Arabic grammar cannot easily be discarded
as inadmissible or non-Classical. (1994, 16f., 262)
Some of the examples that the Arab grammarians commented on are
from the Qurn, while other examples were made up by the grammar-
ians themselves. A survey of the examples in the Concluding remarks
shows that the examples belong to disparate types. In some of them,
allad is combined with a preposition, such as ka-llad as, al llad for
that, bada llad after, in which allad stands for the more usual m. In
other cases, it is again either m, or an, which allad stands for. It might
even be doubted that some of the sentences on which the grammarians
dwelt in length ever occurred in normal speech, such as allad mararta
mamarrun h asanun It is a good passing that you passed, apart from
the fact that the syntactic interpretation of allad in this sentence is far
from being unequivocal.
As far as I can see, Goldenberg makes no attempt to draw historical
conclusions from his examples within the general discourse of conjunc-
tional allad, nor can I do this myself. So, in spite of Goldenbergs thor-
ough and learned approach, the historical dimension of his examples
has yet to be demonstrated.
3. Additional early documentary evidence
In the following, I shall give documentary evidence of allad as a con-

junction, which is the result of a wide and systematic reading of pre-
modern Arabic documents. The major part of the evidence comes
from Judaeo-Arabic documents found in the Genizah in al-Fustt (Old
Cairo). Considering that I likewise have read Arabic documents written
by Muslims as well as Christians, the preponderance of the Judaeo-Ara-
bic evidence is noteworthy, being due to the more sub-standard charac-
ter of Judaeo-Arabic writings as compared with Muslim and Christian
ones. The value of documentary examples lies in that they are chrono-
logically well-defined, which is not necessarily the case with literary
examples, especially when these occur in sub-standard texts which have
come down to us through late manuscripts.
80 werner diem
Meanwhile, many of the Judaeo-Arabic documents which I had read

in their editiones principes have been re-edited by Moshe Gil in his two
comprehensive collections of Genizah documents. Where this is the case
I shall cite both editions, partially in order to give credit to the original
editors and partially because there are occasionally differences between
Gils re-editions and the original versions. While some interesting or
difficult cases of dialectal divergences from the literary language will be
commented upon in footnotes, substandard elements as such are not
systematically marked. Some of my examples have already been cited by
Blau (1961), as became clear to me while preparing this article. I have
retained these examples for the reason that they belong to an exclusively
documentary corpus, whereas Blau draws his examples from both liter-
ary and documentary sources, with the majority being literary.
My examples will be presented according to a linguistic typology
which mirrors the historical development of allad as I see it. My divi-
sion owes much to Woidichs division of his Cairene material into his
types A and B (see above 2.6 and 2.7), even if his division and mine do
not fully overlap from a historical perspective as I assume more sub-
types to have existed than he does. Many examples contain vulgarisms,
which I shall not mark systematically.
A. The head of the sentence expresses praise of or gratitude to God

(a) Nominal type
(1) li-llhi l-h amdu llad knati l-qibatu li-xayrin4
Praise be to God that the result (of the affair) was good
Goitein, Arkiyon, No 39r, 21 = Gil, Texts, No 215r, 20 (letter
to al-Fustt, early 11th c. C.E.)
(2) fa-li-llhi l-h amdu wa-l-minnatu llat5 knati l-aqbatu
h amdatan
So to God be praise and gratitude that the result (of the
6
affair) was praiseworthy
4
Goitein reads [il l-]xayri, Gil li-xayrin. Gil does not mention Goiteins reading.
5
allat, which seemingly refers to al-minnah, is due to a kind of hypercorrection.
6
Another example with allat that, or, more exactly, li-llat in order that, is li-
anna qawm mina l-maribati qad sr yusall indan li-llat (!7%%) yutbahum in
Goitein, Kneset, No 4v, 4f. Indeed, Goitein rendered !7%% as the Hebrew final conjunc-
tion *.)%. However, Gils re-edition in Texts, No 328, has %% instead of !7%%. So what
we have here is the simple prayer Allhu yutbhum May God reward them!sc. the
Maghrebis for praying in the synagogue of the Jerusalemites in Old Cairo.
Diem, Geschftsbriefe Wien, No 45r, 10 (letter to a merchant

in Egypt, 12th13th c.s C.E.)
(b) Verbal type
(3) fa-h amidn llha al dlika llad nazara ilayn wa-lam
yumit bin
and we praised God for this (happy outcome),7 namely that
He cared for us and did not have us mocked at8
Goitein, Iggeret r, 16 = Gil, Texts, No 616r, 17 (letter of
al-Mahdyah to al-Fust t, 11th c. C.E.)
(4) wa-h amidtu llha]9 an wa-ax llad masayn10 lah iqn al
xayr wa-fiyah li-Msr11
and we praised God], I and my brother, that we (finally)
reached Egypt on foot well and in good health
Toledano, Teudot, No 2, 1 (letter from Egypt, 1540 C.E.)
B. The head of the sentence does not express praise of or gratitude to God
(a) The head of the sentence contains an abstract noun expressing a
non-personal emotional evaluation of the contents of the allad
clause
(5) wa-m wajadn azan li-qulbin ayra annahu l-waylu
lan nah nu llad in li-hdihi l-masibi wa-narab hdihi
l-akwsa l-murrata
We did not find consolation for our hearts but (all we can
say is) that woe is us that we have lived (long enough to go
through) these disasters und (that we) have to drink these
bitter cups!
Gottheil and Worrell, Fragments, No 27r, 79 = Gil, Texts, No
501r, 79 (letter from Jerusalem to al-Fustt, 1065 C.E.)
7
The details of the happy outcome of the affair in question are imparted immedi-
ately before the passage cited.
8
Gil in his translation considers dlika an antecedens of allad in the sense of the
German darber, da. However, the alladclause is an apposition to dlika, which in
its turn refers to details mentioned before. In other words, the sentence is an expansion
of the usual fa-h amidn llha al dlika.
9
The addition is mine. Other additions are possible, but praise of God is the most
probable one.
10
Dialectal form for maayn.The language of the letter is substandard and exhib-
its features of Moroccan Jewish Arabic.
11
The grapheme renders Masar < Masr < Misr.
82 werner diem
(b) The head of the sentence contains a verb or a participle express-

ing the emotion of a person caused by the contents of the allad
clause
Joy
(6) wa-an bi-h amdi llhi mutabitun bi-llad ttasaltu ilayhim
I am, thank God, rejoicing that I joined them12 (in
marriage)13
Ashtor, Documentos, No 2r, 11 = Gil, Documents, No 457r,
12 (letter from Jerusalem to Toledo, 1057 C.E.)14
(7) wa-qad sarran llad anfadta lahu rah lahu
It pleased me that you sent him his merchandise
Gottheil and Worrell, Fragments, No 9v, margin, 1113 (let-
ter from al-Fustt to Aden, probably 12th c. C.E.)
(8) fa-qad radtu laka bi-llad uxidtum fa-huli nsun
muhtaimn wa-llhi z-zabbln m rad bi-llad uxidtum
I was content concerning you that you (both)15 were pun-
ished <as well as certain other persons were content about
it> for these are decent people. Only the street-sweepers 16
were, by God, not content that you (two) were punished
Diem, Geschftsbriefe Wien, No 10, 8 (letter of a jealous wife
to her husband, 12th c. C.E.)
Wonder
(9) wa-tumma inn ajabu minka llad lam tusb man yaktubu
laka kitb ill daf atan
Furthermore, I am astonished at you that you (allegedly)
found only once somebody writing a letter down for you17
Goitein, Saloniqi, No 1r, 37f. (letter from Saloniki to al-
Fustt, 11th c. C.E.)
12
Sc. the male relatives of the bride.
13
Ashtor conceives of allad as standing for alladna, translating Estoy contento de
aquellos con que [he emparentado por matrimonio].
14
According to Ashtor, the date is 1053. The difference consists in the reading of the
last letter in the date according to the Jewish era on the margin (ttyd vs. ttyh )
15
Sc. the addressee and his mistress, the plural standing for the dual in vulgar
language.
16
That is, the mob.
17
Sc. a letter to be sent to the writer of the letter, who is the addressees father.
Anger
(10) wa-an adbnu alayka y-ax katr allad wasal[ta il]
Misra wa-lam tasil il Adana
I am very angry about you, O brother, that you came to
Egypt (or Cairo), whereas you did not come to Aden18
Braslavski, Mishar, r 12 (Letter from Aden to al-Mahdyah,
c. 1149 C.E.)
Reproach19
(11) m baqiya alayn ayun ill llad lam tuarrifn kayfa
knat wasyatu xlika naxuduhu (!) llad lam tuarrifn in
kna wasala laka ayun mina l-kutubi
Nothing remains for us (to say) except that (until now)
you have not informed us as to how your uncles last will
was. What we reproach him (!)20 with is that you did not
inform me as to whether you received any of our letters
Gottheil and Worrell, Fragments, No 9v, 40f. (letter from
al-Fust t to Aden, probably 12th c. C.E.)
Grief
(12) wa-azza alayn dlika katr allad lam yakn ah adun
minn indaka yuwinuka f-m jar alayka f taabika f
m yaxussu amra xlika
and we were very much grieved by that,21 (namely) that
none of us was with you to assist you in that which you
had to endure in your concerns regarding the affair of (the
illness and death of) your uncle
Gottheil and Worrell, Fragments, No 9r, 12f. (letter from
(13) wa-dqa sadrun katr allad lam yakn laka maahu
kitbun yutamminun
and we were very much distressed that he had no letter
of yours with him setting our minds at rest (with regard
to you)
18
The writer intends to say that continuing the travel from Egypt to Aden would have
been easy for the addressee, his brother.
19
Only the second allad in (11) is an example of allad following a verb expressing
reproach. For the first allad, see (17).
20
Scribal error for you.
21
Reference to the illness and death of the addressees uncle, the details of which are
recapitulated by the writer before.
84 werner diem
Gottheil and Worrell, Fragments, No 9r, 28f. (letter from

(14) nulimukum ann bi-xayrin f fiyatin wa-qad azza alayya
katr allad lam takn h dir h att tuxallisa laka rah laka
I inform you (herewith) that I am well (and) in good health,
but (that) I am much grieved that you are not present so that
you might free your luggage (from the authorities)
Amari, Diplomi, 53, 2f. (letter from Tunis to Pisa, 12th
13th c.s. C.E.)
(c) The head of the sentence contains a verb or noun not expressing
an emotion
(15) wa-qad akartu tafaddulahumh arasahum llhullad
qad dakarn f kitbihim bi-s-salmi wa-bi-fili l-jamli f
bb
I am also grateful for their22 kindnessmay God protect
them(consisting in) that in their letter they gave greet-
ings to me and (also consisting) in performing good deeds
to me
Assaf, Meqorot, 51, line 20f. = Gil, Documents, No 298r,
20f. (letter from Jerusalem to al-Qayrawn, 1039 C.E.)
(16) wa-jb lan kitbaka wa-nah nu laysa indan xabarun
h att jbahu r-rajulu l-warrqujazhu llhu ann xayr
allad nalamu
and he brought us your letter while we (still) had no news
of you until that man, the book-seller, brought itmay
God requite him in our stead with good that we (now)
know (how you are)
Gottheil and Worrell, Fragments, No 9v, 368 (letter from
(17) See (11), first allad.
I am the first to admit that 17 examples of allad that are a somewhat

meagre result of a decade-long reading of texts, but this result simply
shows how rare this phenomenon is in texts whose writers intended lit-
erary Arabic. Surprisingly scarce is especially evidence of the type A(a)
al-h amdu li-llhi llad.
22
Reference to two addressees, whose letter the writer answers.
Asking whether we can conclude from the scarcity of this type in doc-
uments, especially letters, that it was likewise marginal in the spoken
language, we must allow for some reservations. The H amdalah occurs
frequently in letters, mostly at the beginning but also in other parts.
Usually the H amdalah is mentioned in the context of news which are
deemed praiseworthy, as, for example, the writers or other persons
good health or a good outcome of a difficult situation. However, this
kind of the H amdalah is usually preceded by the report of the fact to
which it refers, as, e.g., katabtu ilayka . . . wa-an wa-man qibal bi-
xayrin wa-fiyatin wa-l-h amdu li-llhi (al dlika) I am writing you . . .,
while I and my family are in good healthPraise be to God (for this).
This conventional structure prevents the al-h amdu li-llhi llad type
from frequently occurring in letters, which in its turn must also lead
to lower frequency of the re-interpreted al-h amdu li-llhi llad in let-
ters than may have been the case in the spoken language. Documentary
texts where the al-h amdu li-llhi llad type usually does occur are, e.g.,
waqf documents, appointments of high officials and marriage contracts
between persons belonging to the upper class. In these kinds of texts, an
introductory al-h amdu li-llhi llad is expanded into long complicated
passages praising God for bounties related to the topic of the text, but
they are of so elaborated a style that they lack any deviation from the
literary language. To sum up, the rare occurrence of the re-interpreted
al-h amdu li-llhi llad in documents is of no relevance for the question
of its potential frequency in the spoken substandard language.
Scarce is also my evidence of the type B(a), where an abstract noun
expresses a non-personal evaluation of the contents of the allad clause,
especially when compared with the abundant evidence of this type in
Cairene Arabic as collected by Woidich (1980 and 1989). Again it is pos-
sible that the scarcity of early evidence of this type in documents is due
to specific circumstances. Expressions of this type are mostly exclama-
tory and thus have a clear Kundgabefunktion, which makes it more
likely for them to be used in the spoken language than in writing.
4. Historical typology
4.1 Introductory remark

Since the question of whether, and if so, how the different subtypes of
allad that are connected with each other is still open at this stage
86 werner diem
of this investigation (even if the answer to this question is admittedly

anticipated by the arrangement of my own material in the preceding
paragraph), I shall deal with them one by one. In doing so I shall con-
sider whether there is reason to assume that the subtype in question
had an origin of its own independent of the other subtypes or whatever
other origin there might be. The sequence of the subtypes corresponds
to decreasing syntactic and semantic markedness and specificity.
4.2 Types A(a) al-h amdu li-llhi llad Praise be to God that and type
A(b) h amidtu llha llad I praised God that
There seems to be nobody so far denying the validity of Spitalers theory
that in al-h amdu li-llhi llad, to which my type A(a) corresponds, the
relative particle came to be re-interpreted as a conjunction meaning
da; that. In Spitalers view, this development is due to a Funktions-
schwchung of allad, whereby das nachfolgende allad zwangslufig
den lebendigen Zusammenhang mit seinem Beziehungsnomen [ver-
lor] (for details see above 2.3), and this opinion is whole-heartedly
shared by Woidich (see above 2.7).
In my opinion, this approach is flawed by the fact that, besides the
nominal type al-h amdu li-llhi llad, attested in the Qurn and else-
where, there existed a corresponding verbal type, as e.g., in the saying
ascribed to Muhammad, y-maara l-arabi h mad llha llad rafaa
ankumu l-ura O company of the Arabs, praise God, who took the
tithes from you! Ibn H anbal, Musnad, Musnadu l-aarati l-mubairna
bi-l-jannati, No 1566. This verbal type is likewise attested in the form of
the re-interpreted verbal type A(b), along with the re-interpreted nomi-
nal type A(a). The existence of the re-interpreted verbal type A(b), whose
verb varies according to the syntactic context, clearly proves that when
the re-interpretation of al-h amdu li-llhi took place its constituents, in
spite of its formulaic character, must have been present in the minds of
the speakers and cannot have been a de-etymologized complex as, e.g.,
add how much? (< *qadr ayyi ayin) in Syrian Arabic or izzayy
how? (< * zayy) in Cairene Arabic. Both the nominal al-h amdu
li-llhi and the verbal h amidtu llha are equally attested as early as in
pre-Islamic poetry,23 which proves that they had existed side by side
from the first.
23
An example of al-h amdu li-llhi occurs in the Muallaqah of Imra al-Qays, verse
124, in Ahlwardts edition, and for examples of the verbal type see Brockelmann (1922,
Rather, the re-interpretation took place because the implicit logi-

cal structure of many items of the Qurnic al-h amdu li-llhi llad is
causal. A typical example is al-h amdu li-llhi llad najjn mina l-qawmi
z-zlimna Q 23:28, in Arthur J. Arberrys translation: Praise belongs to
God, who has delivered us from the people of the evildoers. This literal
translation of the relative clause is of course correct. However, when
considering the propositional structure of the Arabic sentence we can
easily discern that there is between the relative clause and the head of
the sentence an intrinsic connection consisting in the implication that
God is to be praised because He delivered the believers from the evil-
doers.24 It should be noted that this causal structure does not automati-
cally exist in all al-h amdu li-llhi llad sentences but, it seems, only or
mainly in those which refer to certain individual bounties of God. In this
respect, it may, e.g., be doubted that the relation in al-h amdu li-llhi llad
xalaqa s-samwti wa-l-arda Praise be to God, who created the heav-
ens and the earth Qurn 6:1 or al-h amdu li-llhi llad wasia samuhu
l-aswta Praise be to God, whose ear comprises all sounds an-Nas,
Sunan, Kitb at-ta lq, No 3406 and Ibn H anbal, Musnad, Bq musnad
al-Ansr, No 23064, is as causal as the first example or causal at all. But
the cases of al-h amdu li-llhi llad in the Qurn and still more those in
Tradition refer mostly to individual deeds of God and thus express also
an inherent causal relation.
A combination of a non-causal relation and a causal relation is, e.g.,
found in the following passage in Ibn Ab d-Dam a-fis (d. 1244
C.E.) Kitb adab al-qad: al-h amdu li-llhi llad ahidati l-uqlu
bi-qidamihi wa-wah dnyatih / wa-waqafat dna idrki jallihi wa-
azamatih || allad btadaa l-maxlqti bi-badi h ikmatih / wa-sawwara
116). Additional verbal examples occur in Labds Dwn No 6, 1 (h amidtu llha wa-
llhu l-h amdu etc.) and No 5, 1 (wa-llhu rabb mjidun mah mdun). For examples of
the related bi-h amdi llhi, see Brockelmann (1922), and an additional example occurs in
the Muallaqah of an-Nbiah, verse 12, likewise in Ahlwardts edition.
24
Such causal function of relative clauses is, it seems, not dealt with by Lehmann
1989 in his otherwise comprehensive morphological and semantic analysis of the rela-
tive clause in a great number of languages, including Arabic. Nor is his chapter Vom
Relativpronomen zur Konjunktion (1989, 389393) of much help for the problem of
allad as a conjunction, as he considers any relative pronouns that have lost their inflec-
tion to be Konjunktionen if there is no resumptive pronoun in the subordinate clause.
How Lehmann thinks the transition from uninflected relative pronouns to true con-
junctions meaning that or because to have been remains unclear, in spite of some
final remarks (1989, 391).
88 werner diem
l-kinti bi-lutfi sanatih etc.25 Praise be to God to whose pre-existence

and uniqueness all minds testify and whose magnificence and power
they are unable to grasp, who made creation with His wonderful and
singular wisdom and designed the existing world with His kind work,
etc. The first allad clause is non-causal as it describes attributes of God,
and thus it is not possible to paraphrase it with a causal clause such as
*Praise be to God because all minds testify to His priority and unique-
ness and are unable to grasp His magnificence and power. In contrast
to the first allad clause, the second one is inherently causal, which is
evident from the fact that it can easily be re-formulated as a causal
clause: Praise be to God because He made creation with His wonder-
ful and singular wisdom and designed the existing world with His kind
work, etc. The causal character of this allad clause is due to the fact
that men own their existence to God having created them and thus have
to thank Him for His benevolence. It is also interesting that the content
of the first allad clause is of a more inherent nature than the second in
that it describes inherent, inalienable attributes of God, while the sec-
ond refers to a deed that God performed of His own will. This hierarchy
of the two allad clauses conforms with a rule of Arabic syntax which
says that attributes follow the noun they refer to in order of decreasing
inherence.
Considering that there is an inherent causal relation in many items
of the Qurnic al-h amdu li-llhi llad between the head of the sentence
and the relative clause, it comes as no surprise that the allad that con-
nects the two parts of the sentence and thus holds the position which a
conjunction explicitly expressing the causal relation would have, should
have been re-interpreted as a causal conjunction. This re-interpretation
was in all probability triggered by parallel constructions in which the
same causal relation is expressed explicitly. These constructions are as
follows:
(a) al-h amdu li-llhi al + noun/pronoun praise be to God for . . . and
h amidtu llha al + noun/pronoun I praised God for . . . In both
constructions, the al phrase indicates the reason why God shall
be praised. They are so frequent, particularly the former, that it is
superfluous to give examples.
25
Adab al-qad I, p. 247.
(b) al-h amdu li-llhi (al) an(na) praise be to God (for) that . . . and
h amidtu llha (al) an(na) I praised God (for) that . . . The prepo-
sition al can be elided as can any preposition preceding an(na).
Sentences of this kind seem absolutely normal, but I can adduce
only few examples of the verbal type, one from pre-Islamic times
and the other examples later: h amidtu llha an ams Rubayun *
bi-dri l-hni malh yan muqm (Wfir) I praised God that
Rubay got * in the house of disgrace, placed there in a shameful
way (Maqil b. Xuwaylid) Dwn al-Hudalyn I, No 14, 3; istahalla
yah madu rabbahu an l yakna asbahu d h aqqin f l-fayi and
he began to praise his Lord that none who had a right to the booty
had injured him ad-Drim, Sunan, Kitb al-muqaddimah, No 91;
allha ah madu al an jaalan min ulami l-arabyati God I
praise that he made me belong to the scholars of the Arabic lan-
guage az-Zamaxar, Mufassa l, 2, 2. Another example of the verbal
type with God as the subject of an anna-clause is cited by Blau for
Christian Arabic for kr I: akara llha annahu lam yuxallhi yusu
il abdihi He thanked God that He had not let (or made) him
act unjustly to His servant (19661967, 526). With subordinate
clauses whose subject is coreferential with Allh (second and third
examples), this construction is optional instead of (a), whereas it
is the only possible construction for subordinate clauses whose
subject is not coreferential with Allh (first example). It cannot be
excluded that this construction is more usual in or even restricted to
the verbal type, which is less formulaic than the nominal type. More
examples would be desirable; their being so scarce so far is perhaps
due to my (and my predecessors) not noting them simply because
they are so normal and therefore do not seem to deserve attention.
(c) ah maduhu (al) an(na) I praise Him (for) that . . . Whenever
Allh is replaced by the personal pronoun, only this construction
is possible. This construction is especially frequent as a continu-
ation of the relative clause type in the religious arengas (xutbahs)
of Mamluk documents: al-h amdu li-llhi llad . . . nah maduhu al
annahu . . . This construction would also be the only one possible
for the nominal head lahu l-h amdu but I have no examples.
(d) h amidtu llha id/fa- I praised God because. In this type, the sub-
ordinate clause is not dependent on the verb of the head as in the
preceding types but forms a subordinate causal clause. Examples
of each of the two conjunctions are the pre-Islamic verse h amidtu
90 werner diem
ilh bada Urwata id naj * Xirun wa-badu -arri ahwanu min
26
badin (Tawl) I praised my God after (the death of my brother)
Urwah (in battle) because (my son) Xir was (at least) saved, *
and some evil is easier to bear than another (Ab Xir) Dwn
al-Hudalyn III, No 14, 1, and, in Tradition, y-iata h mad llha
fa-qad barraaki llhu O iah, praise God because God has excul-
pated you! al-Buxr, Sah h , Kitb a-ahdt, No 2467.
These functional and syntactic correspondences are set off in the fol-
lowing table, in which all sentences express the notion of praise being
due to God for the salvation of the speakers, which is directly or indi-
rectly ascribed to Him. Of the types marked with ?, no evidence has
been adduced so far, but they would be normal Arabic from a syntactic
point of view.
(a) Nominal type
Head Expansion Head Expansion
al-h amdu li-llhi llad najjn Praise be to God who saved us

al-h amdu li-llhi al najtin Praise be to God for our salvation
?al-h amdu li-llhi (al) an najjn Praise be to God (for) that He saved
us
?al-h amdu li-llhi (al) an najawn Praise be to God (for) that we were
saved
(b) Verbal type
Head Expansion Head Expansion
h amidtu llha llad najjn I praised God who saved us

h amidtu llha al najtin I praised God for our salvation
h amidtu llha (al) an najjn I praised God (for) that He saved
us
h amidtu llha (al) an najawn I praised God (for) that we were
saved
h amidtuhu (al) an najjn I Praised Him (for) that He saved
us
h amidtu llha id/fa-najjn I praised God because He saved us
h amidtu llha id/fa-najawn I praised God because we were
saved
26
bad in the rhyme.
It is from this specific constellation that the re-interpretation of allad

in al-h amdu li-llhi llad and h amidtu llha llad as a causal conjunction
meaning (for) that started. Since the relative clause type shared with
the other types both the overall construction and the causal meaning of
the subordinate clause, allad could by analogy be interpreted as a par-
ticle expressing this causal relation. It is in sentences such as al-h amdu
li-llhi llad najawn Praise be to God (for) that we were saved and
h amidtu llha llad najawn I praised God (for) that we were saved,
with the subject of the allad no longer being coreferential with Allh of
the head, that the re-interpretation surfaced.
It should be stressed that this re-interpretation of the relative clause
as a causal subordinate clause was only possible because of the three fol-
lowing characteristics of al-h amdu li-llhi llad (and the corresponding
h amidtu llha llad), of which the first is syntactic, the second semantic
and the third morphological.
(a) The relative clause of al-h amdu li-llhi llad is non-restrictive,
which means that it does not serve the purpose of identifying Allh, who
is the object of praise. It is the loose connection typical of non-restric-
tive relative clauses that permitted the syntactic shift of the subordinate
clause from its status as a relative clause depending on the antecedent
Allh to its new status as a causal subordinate clause which no longer
depends on Allh but on the head of the sentence as a whole.
(b) Verbs such as to praise or to thank have, due to their specific
semantics, an inherent causal complement indicating the grounds for
praise or gratitude. Even if in the case of an isolated H amdalah in a
religious context the reason is not explicitly expressed it is neverthe-
less there, consisting in Gods general bounties, the knowledge of which
exists in the religious subtext. Thus whenever something positive is
mentioned in the syntactic context of a H amdalah it is in principle sub-
ject to being interpreted as being the reason for it. Exactly this happened
in the case of the relative clause of the H amdalah, but it happened only
there because it is only the non-restrictive clause that, due to its struc-
ture, has the syntactic potential of such a subordinate clause, while this
is not the case with, e.g., nominal attributes referring to Allh.
(c) The re-interpretation of allad may have been helped by the fact
that at the time when it happened the relative pronouns had probably
already been reduced in the spoken language to allad, which etymo-
logically is the pronoun of the 3rd masc. sing. As a result of this develop-
ment, the congruence between the antecedent and the relative pronoun
in terms of gender, number and, in the case of the dual, case had
92 werner diem
disappeared, and thus the syntactic connection between the antecedent

and the relative clause had been weakened.
However, behind this development there is yet another factor, which
is of an ontological nature. Of the constructions mentioned above sub (a)
and (b), those where the subject of the subordinate clause is not coref-
erential with Allh are as a rule not or less possible with other syntactic
objects of to praise, unless the specific reason of praise is mentioned
before or otherwise known to the person addressed. Thus one could say
in English (and correspondingly in German and Arabic and many other
languages), I would like to thank you that you helped me (coreferen-
tial subject of the subordinate clause), but usually not I would like to
thank you that we were helped (non-coreferential subject of the sub-
ordinate clause), whereas the second, non-coreferential construction is
a normal construction with God as the syntactic object of to thank, to
praise: (I) thank God that we were helped. The difference lies here in
that whenever something (positive) happens it can be attributed by the
believer to God as the one who with His power has caused it. This is why
the H amdalah (nominal or verbal) could be followed from the begin-
ning by non-coreferential causal clauses (al an, an, id) as mentioned
above, and this is also why in al-h amdu li-llhi llad the coreferentiality
of the subject of the relative clause with its antecedent Allh was not an
inherently indispensable syntactic feature, a fact which made it possible
for coreferentiality to be given up in the course of the re-interpretation
of the relative clause.
As for the question of chronological priority of the re-interpreted
nominal and verbal types, Spitaler seems to have assumed the verbal
type to have been secondary (1962, 105). In support of this assumption it
may be argued, as Spitaler has, that there is evidence of the nominal type
from as early as the 9th century C.E. whereas the verbal type is attested
somewhat later, but this does not necessarily mean that the verbal type
had not existed before the earliest written evidence of it. More convinc-
ing is the argument that it is only the nominal type al-h amdu li-llhi
llad that is part of the Qurn and thus a formula the Muslims had been
familiar with from the beginning of Islam. Accordingly, we find in Tra-
dition numerous examples of (the correct) al-h amdu li-llhi llad + a
coreferential verb and many examples of al-h amdu li-llhi al + a noun,
but few examples of the verbal type. We could even say that al-h amdu
li-llhi unexpanded or expanded has been and still is part of everyday
Muslim speech, while this is not the case with the verbal type to the same
extent. On the other hand, it is the verbal type who seems to exhibit
greater variability, and thus it may have been this type, or perhaps rather
the existence of this type in its various subtypes, that may have triggered
the re-interpretation of the more formulaic nominal type.
Besides the examples of h md, there are examples of the same basic
structures containing synonymous expressions, which shows a first
semantic generalization of the re-interpreted allad. There is my example
(2) fa-li-llhi l-h amdu wa-l-minnatu llat (!) knati l-aqbatu h amdatan,
and among Spitalers examples there is ukuri r-Rah ma r-Rah mn /
allad lam tajid al-malik Qays f hd l-makn danke dem Allbarm-
herzigen, dass du den Knig Qais nicht an diesem Ort gefunden hast
from the Antar novel, where additionally Allh is replaced by its syn-
onym ar-Rah mn ar-Rah m, reversed there because of the rhyme. Blau
adduces further examples of synonymous verbs (sbh II, kr I, mjd II),
which, however, are syntactically ambiguous (1967).
We have reason to assume that the synonymous roots had already
been in use, besides h md, as alternatives of h md from the first, as we have
nominal examples such as wa-li-llhi l-h amdu wa--ukru To God praise
and thanks are due! (Dietrich, Briefe Hamburg, No 42a, 3 [c. 916/7 C.E.],
and likewise in Tradition fa-laka l-h amdu wa-laka -ukru Ab Dwd,
Sunan, Kitb al-adab, No 4411); wa--ukru li-llhi Thanks be to God!
(Anawati and Jomier, Papyrus chrtien, line 3 [9th c. C.E.]); wa-li-llhi
-ukru katran kam huwa ahluhu wa-mustah iqquhu To God repeated
thanks are due as he is entitled to and worthy of it (Rib, Lettres II,
No 17r, 5 [9th c. C.E.]). Furthermore, there are the Christian formulae
as-subh u li-llhi Praise be to God! and al-majdu li-llhi Glory be to
God!, for which the reader is referred to my Briefe Heidelberg, 24 and 25
respectively. An example of ukr and a relative clause is fa-aqimi -ukra
li-llhi llad snaka bi-dlika Therefore extend in the right way your
thanks to God, who preserved you thereby (Rib, Lettres I, No 5, 22
[9th c. C.E.]). When al-h amdu li-llhi llad and h amidtu llha llad were
re-interpreted, this development comprised the synonyms of h md also.
A final point to be made regards the translation of the re-interpreted
allad as that, da. Asking why allad can be translated with simple
that da, while, according to what I have tried to point out above, its
re-interpretation is based on an implicit causal relation, we should be
aware of a hitherto neglected fact: the syntactic opacity of that, da.
As with Arabic an and anna, prepositions preceding that, da in
English and German can or must be elided, which leads to the basic
94 werner diem
logical relations, as expressed by the prepositions, becoming blurred on

the surface. But the prepositions can also be retained on the surface,
probably more so in German than in English. Thus, instead of the syn-
tactically opaque Lob sei Gott, da wir gerettet worden sind the more
explicit Lob sei Gott dafr, da wir gerettet worden sind is also pos-
sible.
4.3 Type B(a) malh allad (it is) good that

Half a dozen examples of this type were adduced by Spitaler (1962, 109f.)
for modern dialects, intermingled however with examples of the farihtu
llad type (for this latter type, see below 4.4). It was Woidich who distin-
guished between the two groups, of which he characterized the first one
as satzquivalente Ausdrcke (1980, 226). Examples of his type A for
Cairene Arabic are, apart from il-h amdu li-llhi illi Gott sei Dank, da,
e.g., baraka illi ein Segen, da, ya-xsra lli schade, da, and kuwayyis
illi gut, da. Corresponding examples were given by Spitaler for other
modern dialects, whereas he had, according to himself, not come across
examples in literary texts (1962, 110). Earlier evidence of this kind is my
example (5) from the 11th century C.E. containing al-wayl lan nah nu
llad woe is us that.27 Two other examples from that time quoted by
Blau are y-baxtik allad what good fortune for you (fem. sing.) that
and y-h ayf al abk llad what a pity for your (fem. sing.) father
that (1961, 227 sub ). These three early examples contain references to
persons in their heads, but this is also the case with two modern Cairene
examples cited by Woidich for this type A, and therefore this should not
lead us to assume historically different types.
As Woidich (1980) begins his Cairene evidence of this type with two
examples of the re-interpreted il-h amdu li-llh illi it is clear that he con-
siders this type to be somehow the prototype of the whole group. This
conclusion is confirmed by Woidich (1989), where he derives this group
from the al-h amdu li-llhi llad expression, assuming as a first step
Funktionsschwche and as a second step a re-interpretation accord-
ing to the Satzstruktur Prd.Subjekt as in h ilwa di prima ist die!
or b ikkalm da eine Schande sind solche Worte! As I have remarked
above (2.7), sentences of this kind are not likely to have been connected
by native speakers with al-h amdu li-llhi llad.
27
allad might also be interpreted here as a relative pronoun standing for alladna.
In my opinion the solution to this problem should not and need not
be looked for in Cairene examples such as h ilwa di or their early Neo-
Arabic equivalents if any existed. But before going into details, I think it
appropriate to recapitulate some important points:
(a) That the type A(a) al-h amdu li-llhi llad is likely to be the starting
point of the type B(a) malh un allad may be assumed for the simple
fact that it is only al-h amdu li-llhi llad where the transition of the
relative particle allad to a conjunction can be explained, whereas
this is not feasible for malh un allad and the other expressions of
this type.
(b) The sentences of my type B(a) have a comment-topic structure.
(c) When asking for a link between the type A(a) al-h amdu li-llhi llad
and the type B(a) malh un allad, one can rightly assume, as Woid-
ich did, that this link lies in that al-h amdu li-llhi llad had (finally)
been reduced to being a (positive) evaluation of the fact mentioned
in the allad-clause. This reduced meaning of al-h amdu li-llhi llad
would have been equivalent to good that; nice that.
However, there remains the question of the structure of the type B(a)
malh un allad. In this respect, my theory differs from Woidichs. In my
opinion, we should not start from sentences such as Cairene h ilwa di
or b ik-kalm da nor their Neo-Arabic equivalents in order to explain
malh un allad and similar expressions. Asking rather whether there
are other examples of the structure comment + that + topic in Classi-
cal Arabic (and Neo-Arabic), we can answer in the affirmative: there is
an absolutely normal type consisting of a noun + an(na) clause, as for
example malmun anna it is known that, h asanun anna it is good
that, barakatun anna it is a blessing that and so on.28 This is why
we can transpose all examples of my type B(a) into normal Arabic by
replacing allad by an(na), with potential small changes due to anna
demanding an accusative. It should be noted that the transformation
is always possible from allad to an(na) but not always from an(na)
to allad. Thus the transformation of malmun an(naka) jita into
?
malm allad jita would be very questionable, and for Cairene Arabic
?
malm illi gt would probably not be acceptable to most speakers. This
28
In modern literary Arabic, the type mina l-malmi anna would be more likely to
correspond.
96 werner diem
is due to the restriction of this allad type to specific heads referring to

emotions.
Starting from sentences evaluating events in an emotional manner
of the type barakatun (h asanun) an(na) + perfect, which were struc-
turally and semantically equivalent to al-h amdu li-llhi llad + perfect,
the next step of the development, as I see it, was the replacement of
an(na) by allad in analogy with al-h amdu li-llhi llad. This develop-
ment was made possible by two factors: (a) A semantic-syntactic con-
nection between the type al-h amdu li-llhi llad + perfect and the type
barakatun (h asanun) an(na) + perfect was mentally established, due
to Funktionsschwche of the H amdalah. (b) The allad of al-h amdu
li-llhi llad was mentally connected with an(na), because it overlapped
with it in function after it had been re-interpreted as a conjunction.
The generalization of allad led to an opposition between an(na) and
allad, with allad as the marked element being subject to certain restric-
tions as compared with the unmarked an(na). These restrictions, which
the type B(a) had inherited from the primary type al-h amdu li-llhi llad
Praise be to God who via the secondary type A(a), are as follows:
(a) The structure of the type B(a) is always comment + allad + topic
and cannot be reversed. Thus malh allad jita It is nice that
you came is possible, whereas *?allad jita malh is not. Woidich
remarked this for his type B, with which my type B(b) partially
overlaps, but this restriction is valid of his type A and my type B(a)
also (1980, 235).
(b) The head of the sentence consists of an evaluating expression.
Thus malh allad jita It is nice that you came is possible, whereas
*?malm allad jita It is known that you came is usually not.
(c) The logical subject of the allad-clause is human. Woidich remarked
this for his type B for the syntactic subject of the allad clause, but
the same holds true for this type (1980, 230). Thus a sentence such
as malh llad wasala is possible in the meaning It is nice that he
(sc. a certain person) arrived, whereas the meaning It is nice that it
(sc. a letter or a parcel) arrived is excluded or questionable. What,
however, is attested in both the old and new evidence are allad
clauses with a non-human syntactic subject which however refer
to a logical human subject. Thus, a sentence such as malh allad
wasalan It is nice that it (sc. the letter) reached me or, more freely,
It is nice that I received it would probably be possible, though

perhaps not accepted by all speakers of individual dialects.
(d) The subject of the allad clause is almost exclusively pronominal
and as such mostly implied in a finite verb. Thus again a sentence
such as malh allad wasala It is nice that he arrived is possible,
whereas malh allad wasala Ah mad It is nice that Ahmad arrived
is with all probability either not possible at all or at least doubtful in
most dialects.
(e) The allad clause refers to actual, mostly past events. Woidich found
out this feature of Faktivitt for his type B, but this is also true of
his type A and my type B(a) (1980, 231). Thus malh allad jita It is
nice that you came is possible, whereas malh allad taj It would
be nice for you to come is not in most dialects.
These restrictions of allad in type B(a) as compared with an(na) cor-

roborate the assumption of its being the result of a generalization of
the original al-h amdu li-llhi llad Praise be to God who, whereby the
head was replaced with similar expressions while the other features of
the construction were retained:
(a) al-h amdu li-llhi llad najjn has the structure al-h amdu li-llhi +
allad clause, which is not reversible.
(b) The head is evaluating.
(c) The subject of the allad clause, which is Allh, is, in a anthropo-
morphistic perspective, human.
(d) The subject of the allad clause is pronominal, being implied in the
verb that refers to Allh.
(e) The relative clause refers to past events only due to the semantics of
to praise in the meaning of to thank.
4.4 Type B(b) farihtu allad I was glad that

Turning now to the type farih a allad he was happy that, that is, allad
after verbs expressing emotions, the function of allad as a conjunction
is clearly borne out by cases where it follows a preposition as in (6) and
(8) of my examples which have bi-llad by that.
While Spitaler holds that this type goes directly back to the al-h amdu
li-llhi llad type by way of generalization (1962; see above 2.3), Woidich
98 werner diem
declares it to have had a separate origin in relative clauses where the subject
of the head and the subject of the allad clause are coreferential, as in Cairene
ana h mr illi dafat il-h isb (1989). In my opinion, this interpretation
is flawed in several respects, for which the reader is referred to para-
graph 2.7.
Even if I do not find Woidichs historical explanation of this type con-
vincing, it is worthwhile to mention that the sentence which, according
to him, was the starting point shares with al-h amdu li-llhi llad, in spite
of their different structures, the three characteristics mentioned above:
(a) The relative clause is non-restrictive, (b) there is an underlying causal
relation between the head and the relative clause (I am an ass! Why?),
(c) illi goes back to the uninflected allad, to which the relative pronouns
had been reduced.
Before continuing, I think it appropriate and useful to draw atten-
tion to the fact that this type B(b) shares with type B(a) the restrictions
(b)(e) mentioned above, that is, all restrictions with the exception of
the one concerning specific heads. As in the case of type B(a), a histori-
cal theory must take account of these restrictions and explain why they
exist. The assumption that this type had developed from coreferential
relative clauses as claimed by Woidich would explain the restrictions
(a)(c) but not (d) nor (e), as some emotions can concern both past
and future events. Therefore, it is plausible to assume that this type B(b)
developed from the verbal type A(b) h amidtu llha llad in the same way
as the type B(a) developed from the nominal type A(a) al-h amdu li-llhi
llad. In the first case, features of an impersonal expression were general-
ized, and in the other case, features of a personal expression.
This development, as I see it, was as follows: The inherited conjunc-
tion in Arabic sentences of the type I was glad that he came was in
Classical Arabic, and still is in most dialects, an(na), e.g. farihtu an (or
annahu) ja in the classical language. In a certain sense, praising God
for an agreeable event that has happened is an expression of an emo-
tion. Saying, for example, in German Ich danke Gott, da ich davon
verschont geblieben bin, is equivalent to Ich bin froh, da etc. This
semantic overlap made it possible for the allad of h amidtu llha llad
I praised (thanked) God that to be generalized to verbs expressing
likewise the positive emotions of joy, contentment etc. Subsequently, as
Spitaler assumed, this generalization was extended, by Kontrastanalo-
gie, to verbs expressing negative emotions such as anger, reproach, grief
and so on (1962, 109).
As a result of this development, the replacement of the marked allad

in B(b) by the unmarked an(na) is always possible, whereas the replace-
ment of an(na) by allad is only possible where the an(na) clause fulfills
the above-mentioned restrictions.
4.5 Type B(c) alamtuhu llad I informed him that

In Spitalers, Blaus and my material examples remain where allad that
is preceded by verbs not expressing emotions. As I have mentioned
above (2.3), this was a major problem for Spitaler.
In my opinion, the problem is less serious than Spitaler thought it to
be. It is true that the heads of these sentences do not contain expressions
of emotion, but the contents of the allad-clauses are something calling
for positive or negative comments. Take, for example, the two following
examples cited by Spitaler: wa-axbarhu llad qatalahu Antar und sie
teilten ihm mit, dass Antar ihn gettet hatte (1962, 111) and, quoting
Michel Feghali (1928, 313) for Lebanese Arabic, min hayk illi twaffiqti
b-jztik cest cause de cela que tu as t heureuse dans ton mariage!,
with Feghali adding: avec ironie. Supporting evidence for this are also
my example (11) with its first allad and my examples (15)(16). In sen-
tences of this type, allad has the function of expressing a certain emo-
tion or the empathy of the speaker or narrator, the kind of which can be
inferred from the context only. Thus, allad in axbarhu llad qatalahu
Antar indicates a certain emotion on the part of the narrator, which,
depending on the specific context, can either be positive (joy) or nega-
tive (grief).
It should be stressed that this kind of affective allad is not a result of
a secondary development but something inherent in allad that from
the beginning. From the moment when in al-h amdu li-llhi llad and
h amidtu llha llad the relative pronoun was re-interpreted as a con-
junction it became the marked counterpart of an(na), which through
this development became unmarked. The difference between them lay
in that an(na) continued to express the simple that, while allad as its
marked counterpart expressed a certain emotional involvement of the
speaker.
Should one try to express the specific affective content of allad in
sentences of this type B(c), one could do this, for example, by means
of adverbs. Consequently, the first of the two sentences could be ren-
dered, depending on the context, alternatively with and they informed
100 werner diem
him that Antar had fortunately/regrettably killed him. Strictly speaking,

such explicit translation would be appropriate for any case of the con-
junction allad as long as it is restricted to this and the aforementioned
types, translating, e.g., Cairene ana mutaassif illi taxxart with I am so
sorry that I was late.
4.6 Remaining cases

There remain some examples in Spitalers corpus and elsewhere where
the clause following allad (illi) does not refer to past, but to future or
possible events. It is interesting that two of Spitalers three examples are
from Tunisian Arabic, among them ammin elli tsil l-martek w-uldek
sei sicher, dass du zu deiner Frau und deinen Kindern kommen wirst.
Here we have the final point of the development of allad. As Hans-
Rudolf Singer pointed out for the dialect of Tunis, illi is used there in
every context (1984, 669), apparently in the same sense and distribu-
tion as would be an(na) in Classical Arabic and inn- in many modern
dialects including Cairene Arabic, e.g. andna z-zhar illi lqnhum29 wir
hatten Glck, da wir sie trafen on the one hand (referring to a past
event) and th il-ahid illi ma yfh h atta . . . er gab ihm das Verspre-
chen, da er sie bis [zum Hochzeitstag] nicht anschauen wrde (future
30
event). In this dialect, illi that is unmarked, and its marked emo-
tional counterpart for expressing that is kf (< kayfa how), for which
this function is not an isolated phenomenon in modern dialects nor in
the pre-modern substandard language. This remark about kf typical of
the dialect of Tunis must suffice here as the hitherto unwritten story
of kayfa as a conjunction is too long to be told here in a few words.
The dialect of Tunis is not the only one to have generalized the usage
of allad as a conjunction. The same usage of illi/li that is found in
Maltese Arabic (Schabert 1976, 216 and Aquilina 1987, I 566), which is
no coincidence as La langue maltaise a pour origine un dialecte arabe,
vraisemblablement proche des vieux dialectes citadins de Tunisie (Van-
hove 1993, 1). The Jewish dialect of Tripolis in Libya is yet another dia-
lect using lli/li that (Yoda 2005, 278). In addition, similar extreme
examples of illi that are known from other dialects but we lack compre-
29
I have simplified Singers complicated transcription.
30
See also Woidich (1980, 224) for the same function of illi in the Judaeo-Arabic
dialect of Tunis.
hensive studies of their exact distribution. For Cairene Arabic, one can
state on the basis of Woidichs two studies that examples as cited above
for Tunisian Arabic would most certainly be declared to be ungram-
matical or unusual by native speakers.
As for the early history of Cairene Arabic, one should be cautious
about taking the language of the Jewish-Arabic documents of the Cairo
Genizah from the 11th century C.E. and later as early evidence of Cairene
Arabic, because many of the writers of those documents hailed from
the Maghreb, especially from what is today Tunisia, and there remained
strong bonds between the Jewish traders who had settled in al-Fust t
and elsewhere in the Islamic East and their relatives and partners in the
Maghreb. This means that deviations in those documents in the usage of
allad as a conjunction as compared with todays Cairene Arabic might
be ascribed to the Maghrebine background of their writers. Therefore
extreme examples of the conjunction allad in Judaeo-Arabic docu-
ments are to be connected to modern Tunisian Arabic rather than to
modern Cairene Arabic.
4.7 allad vs. an(na)

In describing the development of allad as a conjunction, I have not
dealt in detail with the various functions which it can assume. Some
additional remarks might therefore be helpful.
According to the development of allad as I see it, the relative pronoun
allad in al-h amdu li-llhi llad and h amidtu llha llad was first re-inter-
preted in the sense of that, which is normally expressed by an(na), and
then, by a gradual generalization, it replaced an(na) in specific syntactic
and semantic contexts. We can express this by saying that allad took
over part of the functions of an(na) step by step, a process that led to
an(na) and allad becoming the unmarked and marked members of an
opposition in many, if not all, dialects.
There is a well-known rule in Arabic that any preposition preceding
an(na) can be deleted. When allad replaced an(na) in specific con-
texts, which have been described above, allad consequently was felt to
be submitted to the deletion rule as its counterpart allad was. This meant
that where, according to the syntactic context, a preposition was to be
expected before allad in analogy with an(na), this preposition could
be considered to exist in the deep structure of the construction. On the
other hand, since a preposition could be retained before an(na) without
being deleted, allad could likewise be preceded by prepositions, again in
102 werner diem
analogy with an(na). To put it another way, the existence of cases such
as (6) wa-an . . . mutabitun bi-llad ttasaltu ilayhim I am . . . rejoicing
that I joined them, where allad is preceded by the preposition bi- typi-
cal of the verb bt VIII on which it depends, proves that the process
of the re-interpretation of allad as a conjunction meaning that had
taken place to its full extent. In the modern dialects, the combination
of illi with prepositions does not seem possible; at least Woidich does
not mention it for Cairene Arabic (1980 and 1989), and I do not have
examples thereof either.
Let us now have a look at some of my examples and see what syn-
tactic status allad has and which prepositions are possibly missing. I
shall insert the prepositions missing in the surface structure but existing
in the depth structure in brackets, and also give a translation of these
prepositions:
(1) li-llhi l-h amdu {al} llad knati l-qibatu li-xayrin

Praise be to God for the result having been good
Remark: The verb is h md I fulnan al to praise s.o. for s.th.
(15) wa-qad akartu tafaddulahum . . . {bi-}llad qad dakarn f
kitbihim bi-s-salmi wa-bi-fili l-jamli f bb
I am also grateful for their kindness . . . of giving greetings to me
and of performing good deeds to me
Remark: The verb is fdl V bi- to be so kind as to do s.th.
(13) wa-dqa sadrun katr {li-}allad lam yakn laka maahu kitbun
yutamminun
and we were very much distressed due to his not having a letter
of yours with him setting our minds at rest
Remark: (li-)llad indicates the cause as (li-)an(na) does.
On the other hand, there do exist many sentences where allad intro-
duces a subject or an object clause so that no preposition can be supple-
mented in the deep structure:
(7) wa-qad sarran llad anfadta lahu rah lahu

It pleased me that you sent him his merchandise
Remark: the allad-clause is the subject of the sentence.
There is also the adnominal usage of allad, forming, as is frequent with

an(na), a syntagma that seems to be an apposition to a noun. It would
be difficult to elucidate the true syntactic status of allad here. However,

this is not necessary; it suffices to say that the status of allad is here
analogous to the status of an(na), by which it can be replaced.
(9) inn ajabu minka llad lam tusb man yaktubu laka kitb ill
daf atan
I am astonished at you that you (allegedly) found only once some-
body writing a letter down for you
5. Summary
In the preceding paragraphs, I have endeavored to sketch a picture of

the development of allad as a conjunction valid of all varieties of Neo-
Arabic as far as they are known. The stages of the development are set
off in the following table:
(a) Original expressions containing relative clauses:

al-h amdu li-llhi llad h amidtu llha llad
Praise be to God who I praised God who
(b) Re-interpretation of allad as a causal conjunction on the pattern of
parallel constructions with semantically explicit syntactic means,
this process being due to (a) the relative clause being non-restric-
tive, (b) verbs such as to praise having an inherent complement
indicating the cause of praise, (c) the relative pronouns having been
reduced to allad, whereby the connection of the relative clause to
the head was weakened:
31
al-h amdu li-llhi llad h amidtu llha llad
Praise be to God that I praised God that
(c) Generalization of allad for heads expressing positive emotions, by
(optional) replacement of an(na) with allad, while the syntactic
and semantic restrictions typical of (a)(b) were retained:
malh allad (for malh anna) farihtu llad (for farihtu anna)
It is nice that I was glad that
(d) Generalization of allad for heads expressing negative emotions cet-
eris paribus as in (c):
31
And by analogy semantically related verbs such as kr I, sbh II and mjd II.
104 werner diem
wayl laka llad (for wayl laka taassaftu llad (for taassaftu
anna) anna)
Woe is you that I regretted that
(e) Generalization of allad for heads not expressing emotions, with
allad still having an affective value and retaining its original restric-
tions as in (a)(d):
alamtuhu llad
I informed him (of the pleasant/regrettable fact) that
(f) Generalization of allad as an unmarked conjunction that without
syntactic or semantic restrictions:
th il-ahid illi ma yfh
He made a pledge that he would not see her (Tunisian Arabic)
Stage (f), which is found in the dialect of Tunis and the dialect of the
Jews of Tripolis (Libya), is the final point of the development of allad as
a conjunction. The development of allad (illi) in Cairene Arabic, which
is the best-known of all Arabic dialects, reached stage (d) only.
6. Appendix: Notes on the origin of al-h amdu li-llhi llad
As demonstrated in paragraph 4.2, there were, besides the relative clause,

other, more explicit means of expanding the H amdalah. This raises the
question, which has not yet been given the attention it deserves, why
in the case of al-h amdu li-llhi a causal relation between a subordinate
clause and the head is expressed by a relative clause if the subject of
the relative clause is coreferential with Allh of the head, and not more
explicitly.
Indeed, in terms of normal Arabic one is in some respect entitled
to say that al-h amdu li-llhi llad expressing causal relations is a special
case, which is, however, so innate in the religious language of Islam that
Muslims are not likely to be aware of this fact. On the other hand, the
very fact that the relative pronoun was re-interpreted, at some time in the
history of Arabic, as a conjunction demonstrates that this relative clause
was something peculiar not wholly in line with the normal syntax of
Arabic. That Western Arabists do not seem to have felt al-h amdu li-llhi
llad to be something special either, must be ascribed to the existence
of corresponding relative clauses in the religious style of the Western
languages. For German, I can say by way of introspection that this kind
of causal relative clauses, apart from the religious language, e.g. Lob
sei Gott dem Herrn, der etc., is restricted to a high stylistic level and is
possible for verbs such as danken, preisen, loben only, e.g. Es ist mir
ein Anliegen, an dieser Stelle Herrn N.N. zu danken, der dieses Projekt
hochherzig gefrdert hat. A corresponding formulation in less formal
language is, e.g., Ich mchte Herrn N.N. dafr danken, da er dieses
Projekt untersttzt hat. So the fact that al-h amdu li-llhi llad seems
familiar to us must not prevent us from enquiring as to the reason for
this construction. To this question the following lines will be dedicated.
Considering that al-h amdu li-llhi was in use in pre-Islamic times
already (see above 5.2), we can assume that its expanded form al-h amdu
li-llhi llad is likewise pre-Islamic, although there is no explicit evidence
of it so far, at least none that I am aware of. The basic formula itself is
probably a calque on the Syriac ubh l-alh, as remarked by Theodor
Nldeke and Friedrich Schwally (1909, 112, footnote 1) for the Ftih ah
of the Qurn: entspricht genau syrischem ubh l-alh32 bezw.
teboht l-alh und neutestamentlichem . In this context,
Nldeke and Schwally also mention the so-called Berk in the Old Tes-
tament and the Christian liturgy as a parallel to the Syriac and Greek
formulae, without, however, connecting it to the Qurn. Anton Baum-
starks article Jdischer und christlicher Gebetstypus im Koran (1922)
is in a certain sense a comment on the short remarks of Nldeke and
Schwally, although he does not refer to them. In the following, I shall
first sum up some important points of Baumstarks article and then add
some deliberations of my own:
(a) In the Old Testament, there is the so-called Berk (blessing) of the
structure brk Yahw Blessed is (or be) Yahwe, to which a nomi-
nal attribute or a relative clause can be added, e.g. brk Yahw aer
hiss l etke m miy-yad Misrayim -miy-yad Par Blessed be Yahwe,
who delivered you from the hand of the Egyptians and the hand of
Pharaoh Exodus 18:10.33 A corresponding Qurnic expression is
tabraka in the subtypes tabraka llhu rabbu l-lamna Blessed
be God, the Lord of the whole world Q 7:54 and tabraka llad
bi-yadih l-mulku Blessed be the One in whose hand is the king-
dom Q 67:1, as well as the exceptional brika in brika man f n-nri
32
In the original text in Syriac script.
33
This passage has already been cited by Spitaler (1962, 107) but in another con-
text, where he asks whether this expression might be a parallel to the development of
al-h amdu li-llhi llad. Instead of Spitalers h iss l, read hiss l.
106 werner diem
wa-man h awlah Blessed be He Who is in the fire and around it

Q 27:8. Baumstark tends to assume that the Berk was transmit-
ted to Muhammad via Christian formulae in the New Testament
and elsewhere, in which the Jewish Berk lives on, of the type
, a last echo of this development being Benedictus
dominus (Baumstark 1922, 231ff.). A fact not mentioned by Baum-
stark is that the Qurnic formula is syntactically different both
from the Biblical Berk and the H amdalah, as it lacks a subtype
tabraka llhu llad Blessed be God who.
(b) The type which in the Christian liturgy, especially of the East, has
become the prevailing one is the so-called doxology, which has prob-
ably developed from the Hebrew Berk, e.g. o , Gloria tibi
domine, Laus tibi Christe, and so forth, the doxology being [eine]
possessive Form der Prdikation, durch welche Herrlichkeit, Ruhm,
Lob, Ehre oder wie immer man das schillernde griechische
wiedergeben will, als der Gottheit eigen oder gebhrend bezeich-
net wird (Baumstark 1922, 234). The dative of the Greek formula
can be expanded by eine partizipiale Apposition oder Anrede, and
the reason for Gods praise can also be given in the form of a
clause. This Christian doxology is likewise found in the Qurn, in
two forms: as subh na + a pronoun or a genitive (which does not
interest us in this context) and as al-h amdu li- (Baumstark 1922,
234ff.). However, as Baumstark remarks:
Der frhchristlichen partizipialen Ergnzung einer Doxologie entspricht
diejenige eines
fast immer vielmehr durch einen Relativsatz [. . .]
(1922, 237)
After dealing with details of subh na and al-h amdu li- in the Qurn and
comparing them to Christian liturgical formulae they are likely to be
derived from, Baumstark finally poses the question of how these formu-
lae may have passed to Muhammad. He supposes that they were brought
to the Arabs by Nestorian missionaries, who translated their Syriac reli-
gious texts into Arabic, which he says is typical of the Nestorians:
Die nestorianische Kirche hat aber immer wieder eine auffallende Geneigt-
heit bekundet, ihre angestammte syrische Kultsprache beim Betreten
neuer Missionsgebiete der Volkssprache derselben zu opfern (1922,
247f.)
So far my summary of Baumstarks article. As we have seen, Baumstark
is aware of the fact that the Qurnic al-h amdu li-llhi is expanded by a
relative clause while there are participles and clauses in the Greek
doxologies, but he does not pay further attention to this difference. In
my opinion, the solution to this problem is found in the Syriac context,
through which the doxology was probably transmitted to the Arabs.
Baumstark (1922, 235) mentions himself that the Syriac doxology is
ubh l-, without however further dwelling upon it, while Nldeke and
Schwally (1909), as mentioned above, were of the opinion that al-h amdu
li-llhi goes back to ubh l-alh. It is also interesting in this context
that for Barhebrus, as cited by S. Payne Smith (1879 II, 4026), the
ubh l- formula must have been closely connected to the H amdalah, as
he renders in his dictionary the Syriac ubh l-mrayym l-lmn Praise
be to the Elevated One to all eternity with the Arabic al-h amdu li-l-l
il d-dahri.
In the Syriac New Testament, the simple doxology is part of a sentence
in w-kulleh amm da-h z yab-w ubh l-alh Et omnis plebs ut vidit,
dedit laudem Deo Lk 18:43. Syntactically independent simple doxolo-
gies are ubh l-alh Praise be to God (Payne Smith 1903, 563a) and
ubh l-ab w-la-br wa-l-rh qadd Praise be to the Father, the
Son and the Holy Spirit (Payne Smith 1879 II, 4026). As for expanded
doxologies, we find evidence of them in the Breviarium Chaldaicum, e.g.
ubh l-tb da-b-yad h ubbeh gl teboht la-bnayn Praise be to the
Good One, who through His love revealed glory to men (Breviarium
I, kh = 28) or ubh l-mry da-b-yad tuqpeh hg
am la-trn wa-p
raq
l-abdaw Praise be to the Lord, who through His power overthrew
the tyrant and saved His servants (Breviarium III, 333). Examples of
expanded doxologies in colophons of books are ubh l-ab w-la-br
wa-l-rh qadd d-h ayyel wa-dar w-sayya Praise be to the Father,
the Son and the Holy Spirit, who strengthened, assisted and supported
(respectively) and ubh l-ab d-h ayyel w-la-br d-sayya wa-l-rh
d-qud d-aml Praise be to the Father, who strengthened, the Son,
who assisted, and the Spirit of Holiness, who accomplished (Payne
Smith 1879 II, 4026).
We can safely infer from these examples that the expanded Ara-
bic doxology al-h amdu li-llhi llad is nothing but a rendition of the
Syriac ubh l-alh d- in the same way as the simple Arabic doxology
al-h amdu li-llhi is a rendition of the Syriac ubh l-alh. This means
that the Arabic relative pronoun allad corresponds to the Syriac d-,
which in its chief function is also a relative pronoun. It is true that d- is
also a conjunction of wide and vague meaning so that the d-clauses of the
108 werner diem
above-mentioned Syriac doxologies might be translated as causal clauses

as well, e.g., ubh l-ab d-h ayyel w-la-br d-sayya wa-l-rh d-qud
d-aml as Praise be to the Father because He strengthened, the Son
because he assisted, and the Spirit of Holiness because he accomplished.
On the other side, it may be argued that the d-clauses of Syriac doxolo-
gies, at least to my knowledge, are all coreferential with the nouns of the
l-phrases and thus likely to have been conceived of as relative clauses
rather than as causal clauses.
Although the relative pronoun of the Qurnic al-h amdu li-llhi llad
is thus accounted for, there remains the question why the Syriac doxo-
logy is expanded by d-clauses while the older Greek formulae have par-
ticiples and clauses. The solution to this problem lies probably in the
difference of the verbal systems. The Greek verbal system, which per se
is more complex than the West Semitic one, is still more so with regard
to the active participle, as there are different Greek participles derived
from the corresponding tenses, whereas the Semitic languages have but
one active participle, whose temporal value depends on various factors.
Therefore, when Greek doxologies with active participial attributes
were translated into Syriac, the problems connected with this disparity
seem to have been avoided by translating the Greek participle, if the
specific context demanded this, with a relative clause containing a finite
verb in the (Syriac) perfect. As the Syriac relative pronoun d- is also a
causal conjunction, Greek expansions of the doxology by means of the
conjunction that; because could likewise be rendered with d-. In
this process of rendering Greek attributive participles and clauses
with Syriac d- clauses, the translators may also have been influenced
by the Berk of the Old Testament with its aer, which is also both a
relative particle and a causal conjunction, and which in the Syriac Old
Testament is rendered with d-. Thus the Hebrew brk Yahw aer hiss l
etke m miy-yad Misrayim -miy-yad Par Exodus 18:10 is rendered
as the Syriac brk- mry d-p
ass kn men d d-Misry w-men deh
d-P
ern, with d- exactly corresponding to aer.
Even if there exists, as I have tried to demonstrate, a historical con-
nection between the Hebrew brk Yahw aer, the Syriac ubh l-alh
d- and the Arabic al-h amdu li-llhi llad, this does not fully explain why
relative clauses were chosen from the beginning and not explicit causal
conjunctions as the Hebrew k, the Syriac mettu l d- or the Arabic al
an or li-an. The reason is probably that the implicit indication of the
cause is more in line with the religious attitude deemed appropriate
towards God. If an explicit causal conjunction were used the praise of

God would be limited to a specific deed of His as mentioned in the sub-
ordinate clause, that is, God would be praised for this specific deed and
for nothing else. But this would not be adequate as God, according to
Jewish, Christian and Islamic thinking, is entitled to praise in general,
as is clearly borne out by the unexpanded Berk and the unexpanded
doxology. Thus by not explicitly expressing an (implicitly existent)
causal relation the relative clause is much more in line with a general,
unconditioned praise of God than an explicit causal clause would be.
Accordingly, the Arabic H amdalahs, nominal and verbal ones, where
the subordinate clause is introduced by an explicit conjunction (includ-
ing the re-interpreted allad followed by non-coreferential clauses) do
not imply a general praise of God but are limited to praise of God for
specific facts. In this respect, sentences of this kind belong to the lan-
guage of everyday life and not the language of religious worship, a fact
which is demonstrated by the contexts in which such sentences occur.
A final remark concerns the Arabic verbal type h amidtu llha llad/
an. There is the Syriac verb abbah to praise, and this verb is likely to
have been used with God as the object. However, the dictionaries do
not contain evidence of a formula such as abbh et l-alh d- I praised
God who (or that) used for the expression of gratitude to God for indi-
vidual bounties as the Arabic h amidtu llha llad/an is. Furthermore,
the verbal forms in both languages are different. Consequently, the ver-
bal h amidtu llha llad/an is the result of a specific Arabic development
rather than a calque on a Syriac expression. This in turn would suggest
that the doxology had a history of its own in Arabic after it had been
taken over in its formulaic nominal form from Syriac. Now, as Hartmut
Bobzin (2004, 59) has stressed, it can be concluded from certain pas-
sages of the Qurn that even in pre-Islamic times Allh had been con-
sidered a Schpfer- und Rettergott, a designation of which the notion
of Rettergott (saviour God) is significant in this context. For if Allh
was considered a Rettergott he was especially qualified for the H am-
dalah as in every day life the H amdalah, nominal or verbal, is mostly
used for the happy outcome of an affair. It seems, therefore, that in the
Arabic H amdalah the Christian doxology might well have melted in a
syncretistic way with pre-existing pagan notions of the role of Allh.
110 werner diem
7. References
7.1 Primary sources

Ab Dwd, Sunan = Ab Dwd Sulaymn b. al-Aat as-Sijistn, Kitb as-Sunan.
See Mawsat al-hadt.
Ahlwardt, Wilhelm see Arazi and Masalha
Amari, Diplomi = Michele Amari, I diplomi arabi del R. Archivio fiorentino. Florence:
Tipografia di Felice le Monnier, 1863.
Anawati and Jomier, Papyrus chrtien = PP. Anawati and Jomier, Un papyrus chrtien
en arabe (gypte, IXe sicle Ap. J.-C.). Mlanges islamologiques 2, 1954, 91102.
Arazi and Masalha, Six Early Arab Poets = Albert Arazi and Salman Masalha, Six Early
Arab Poets. New Edition and Concordance Based on W. Ahlwardts The Divans of the
Six Ancient Arabic Poets. Jerusalem: The Hebrew University, 1999.
Ashtor, Documentos = Eli Ashtor, Documentos espaoles de la Genizah. Sefarad 24,
1964, 4180.
Assaf, Meqorot = Simha Assaf, Meqorot u-meh qarim be-toledot Yisrael. Jerusalem, 1946.
Braslavski, Mishar = Josef Braslavski, Al ham-mishar hay-yehudi ben hay-yam hat-
tikon wa-Hoddu bam-meah ha-12. Zion 7, 19411942, 135139.
Breviarium Chaldaicum = Breviarium iuxta ritum Syrorum orientalium id est Chaldaeo-
rum. Slawwt qnnyt d-khn. Rome, 2002. (Reprint of Eugne Tisserants edi-
tion, Rome 1938, itself a reprint of the original edition, Rome 1886.)
al-Buxr, Sah h = Muhammad b. Isml al-Buxr, Kitb al-Jmi as-sah h. See Mawsat
al-hadt.
ad-Drim, Sunan = Abd Allh b. Abd ar-Rahmn ad-Drim, Kitb as-sunan. See
Mawsat al-hadt.
Diem, Briefe Heidelberg = Werner Diem, Arabische Briefe auf Papyrus und Papier aus der
Heidelberger Papyrus-Sammlung. Wiesbaden: O. Harrassowitz, 1991.
Diem, Geschftsbriefe Wien = Werner Diem, Arabische Geschftsbriefe des 10. bis 14.
Jahrhunderts aus der sterreichischen Nationalbibliothek in Wien. Wiesbaden: Harras-
sowitz, 1995.
Dietrich, Briefe Hamburg = Albert Dietrich, Arabische Briefe aus der Papyrussammlung
der Hamburger Staats- und Universitts-Bibliothek. Hamburg: J. H. Augustin, 1955.
Dwn al-Hudalyn = Kitb arh ar al-Hudalyn sanat Ab Sad al-H asan ibn
al-H usayn as-Sukkar. Ed. by Abd as-Sattr Ahmad Farrj and Mahmd Muhammad
kir. Cairo: Dr al-Urbah, 1965.
Gil, Documents = Moshe Gil, Palestine during the First Muslim Period (6341099). Part I.
Studies. Part II. Cairo Geniza Documents. Part III. Cairo Geniza Documents. Tel Aviv:
Tel Aviv University etc., 1983.
Gil, Texts = Moshe Gil, In the Kingdom of Ishmael. Volume I. Studies in Jewish History
in Islamic Lands in the Early Middle Ages. Volume II. Texts from the Cairo Geniza.
The Jews of Iraq and Persia (nos. 1101). Letters of Jewish Merchants (nos. 102303).
Volume III. Texts from the Cairo Geniza. Letters of Jewish Merchants (nos. 304607).
Volume IV. Texts from the Cairo Geniza. Letters of Jewish Merchants (nos. 608846).
Indexes. Jerusalem: Tel Aviv University etc., 1997.
Goitein, Arkiyon = Shelomo Dov Goitein, Olelot me-arkiyono el Ysuf ibn Awkal.
Tarbiz 38, 1968, 1842.
Goitein, Iggeret = Shelomo Dov Goitein, Iggeret Labrat ben Moe ben Simr dayyan
ha-ir al-Mahdyah. Tarbiz 36, 1967, 5972.
Goitein, Kneset = Shelomo Dov Goitein, Bet hak-kneset we-siyyudo lepi kitbe hag-
Genizah. Erez Israel 7, 1964, 8197.
Goitein, Saloniqi = Shelomo Dov Goitein, Eduyot qedumot min hag-Genizah al
qehillat Saloniqi. Sefunot, 11, 1970, 1133.
Gottheil and Worrell, Fragments = Richard Gottheil and William H. Worrell, Fragments
from the Cairo Genizah in the Freer Collection. New York: Macmillan, 1927.
Ibn Ab d-Dam a-fi, Adab al-qad = ihb ad-Dn Ab Ishq Ibrhm b. Abd
Allh al-Hamdn al-H amaw al-marf bi-bn Ab d-Dam a-fi, Kitb adab
al-qad. Ed. by Muhy Hill as-Sarhn. 2 volumes. Bagdad: Matbaat al-ird, 1984.
Ibn H anbal, Musnad = Ahmad b. H anbal, Kitb al-Musnad. See Mawsat al-hadt.
al-Jhiz, H ayawn = Ab Utm n Amr b. Bahr al-Jhiz, Kitb al-h ayawn. Ed. by Abd
as-Salm Muhammad Hrn. Cairo: Matbaat Lajnat at-talf wa-t-tarjamah wa-n-
nar, 1969.
Labd, Dwn = arh Dwn Labd b. Rabah al-mir. Ed. by Ihsn Abbs. Kuwait:
Governmental Press, 1962.
Mawsat al-hadt = Mawsat al-hadt a-arf. al-Kutub at-tisah. CD-Rom. 1st edition.
arikat Saxr li-barmij al-hsib (19911996).
an-Nas, Sunan = Ab Abd ar-Rahmn an-Nas, Kitb as-Sunan. See Mawsat
al-hadt.
Rib, Lettres III = Ysuf Rib, Lettres arabes. Annales islamologiques 14, 1978,
1535; 16, 1980, 129.
Toledano, Teudot = Jacob Moses Toledano, Teudot mik-kitbe yad. Hebrew Union
College Annual 4, 1927, 449467.
az-Zamaxar, Mufassa l = Ab l-Qsim Mahmd b. Umar az-Zamaxar, Kitb
al-Mufassa l f n-nahw. Ed. by J. P. Broch. Christiania etc.: P. T. Mallingius, 1879.

Aquilina, Joseph. 1987. Maltese-English Dictionary. Malta.
Baumstark, Anton. 1927. Jdischer und christlicher Gebetstypus im Koran. Der Islam
16, 229248.
Blau, Joshua. 1961. Diqduq ha-arabit ha-yehudit el yeme hab-benayim. Jerusalem:
Magnes Press, The Hebrew University.
. 1965. The Emergence and Linguistic Background of Judaeo-Arabic. A Study of the
Origins of Middle Arabic. Oxford: Oxford University Press.
. 19661967. A Grammar of Christian Arabic Based Mainly on South-Palestinian
Texts from the First Millennium. 13. Louvain: Secrtariat du Corpus SCO.
Bobzin, Hartmut. 2004. Der Koran. Eine Einfhrung. Mnchen: C.H. Beck, 5th edition.
Bravmann, Meir M. 1953. Studies in Arabic and General Syntax. Cairo: Imprimerie de
lInstitut franais darchologie orientale.
Brockelmann, Carl. 1922. Allah und die Gtzen, der Ursprung des islamischen Mon-
theismus. Archiv fr Religionswissenschaft 21, 99121.
Feghali, Michel. 1928. Syntaxe des parlers arabes actuels du Liban. Paris: P. Geuthner.
Goldenberg, Gideon. 1994. allad al-masdariyyah in Arab grammatical tradition.
Zeitschrift fr Arabische Linguistik 28, 735 = Gideon Goldenberg, Studies in Semitic
Linguistics. Selected Writings, 250285. Jerusalem: Magnes Press, The Hebrew Uni-
versity, 1998.
Hopkins, Simon. 1984. Studies in the Grammar of Early Arabic Based upon Papyri Dat-
able to before 300 A.H./912 A.D. Oxford: Oxford University Press.
Lehmann, Christian.1984. Der Relativsatz. Typologie seiner StrukturenTheorie seiner
FunktionenKompendium seiner Grammatik. Tbingen: G. Narr.
Nldeke, Theodor. 1909. Geschichte des Qorns. Bearbeitet von Friedrich Schwally. Erster
Teil. ber den Ursprung des Qorns. Leipzig. Reprint Hildesheim: G. Olms, 1961.
Payne Smith, J. 1903. A Compendious Syriac Dictionary founded upon the Thesaurus
Syriacus of R. Payne Smith, DD. Ed. by J. Payne Smith (Mrs. Margoliouth). Oxford:
Oxford University Press.
112 werner diem
Payne Smith, R. 1879. Thesaurus Syriacus. Oxford: Clarendon.

Piamenta, Moshe. 1979. Islam in Everyday Arabic Speech. Leiden: E.J. Brill.
. 1983. The Muslim Conception of God and Human Welfare as Reflected in Everyday
Arabic Speech. Leiden: E.J. Brill.
Reckendorf, Hermann. 1921. Arabische Syntax. Heidelberg: C. Winter.
Rubin, Aaron D. 2005. Studies in Semitic Grammaticalization. Winona Lake, Indiana:
Eisenbrauns.
Schabert, Peter. 1976. Laut- und Formenlehre des Maltesischen anhand zweier Mundar-
ten. Erlangen.
Singer, Hans-Rudolf. 1984. Grammatik der arabischen Mundart der Medina von Tunis.
BerlinNew York: W. de Gruyter.
Spitaler, Anton. 1962. al-h amdu li-llhi llad und Verwandtes. Ein Beitrag zur mit-
tel- und neuarabischen Syntax. Oriens 15, 1962, 97114 = Anton Spitaler, Philo-
logica. Beitrge zur Arabistik und Semitistik, ed. by Hartmut Bobzin, 230247, 248
(Zustze). Wiesbaden: Harrassowitz, 1998.
Vanhove, Martine. 1993. La langue maltaise. Etudes syntaxiques dun dialecte arabe pri-
phrique. Wiesbaden: Harrassowitz.
Woidich, Manfred. 1980. illi als Konjunktion im Kairenischen. Studien aus Arabistik
und Semitistik. Anton Spitaler zum siebzigsten Geburtstag von seinen Schlern berre-
icht, ed. by Werner Diem and Stefan Wild, 224238. Wiesbaden: Harrassowitz.
. 1989. illi dass, illi weil und zayy illi als ob: zur Reinterpretation von Relativ-
satzgefgen im Kairenischen. Mediterranean Language Review 45, 1989, 109128.
Yoda, Sumikazu. 2005. The Arabic Dialect of the Jews of Tripoli (Libya). Grammar, Text
and Glossary. Wiesbaden: Harrassowitz.
LES ORIGINES DE LA GRAMMAIRE ARABE, SELON LA
TRADITION : DESCRIPTION, INTERPRTATION, DISCUSSION1
Pierre Larcher
Universit de Provence (Aix-en-Provence, France)
1. Introduction
Le texte, bien connu, que nous traduisons en annexe, est extrait du

Kitb al-dh f ilal al-nahw daz-Zajjj. Az-Zajjj le Zajjjien est
le surnom sous lequel est connu Ab l-Qsim Abd al-Rahmn b. Ishq.
Ce grammairien dorigine iranienne du IVe/Xe sicle doit son surnom
au fait quil tudia Bagdad auprs du grammairien Ibrhm b. as-Sar
az-Zajjj (m. 311/9234). Il sinstalla ensuite en Syrie, Alep, puis
Damas, avant de se rendre en Palestine, Tibriade, o il mourut en
337/949.
On connat un peu moins dune vingtaine douvrages daz-Zajjj. Une
dizaine environ a t publie. Parmi ceux-ci, deux se distinguent :
le Kitb al-Jumal, qui est un ouvrage de grammaire devant son nom

au fait, non quil traite de phrases, mais quil est constitu de notes
de synthse (jumal) sur les diffrents chapitres de la grammaire. Il a
t publi en 1926 Alger par Mohammed Ben Cheneb et republi
Paris en 1957.
le Kitb al-dh f ilal al-nahw, qui nest pas un ouvrage de gram-
maire, mais sur la grammaire. Il a t publi par Mzin Mubrak au
Caire en 1959, puis republi Beyrouth en 1973 et 1979.
Vers la fin des annes 60 et le dbut des annes 70, il y eut, avec lex-
plosion de la linguistique, celle dune sous-discipline : lhistoire de la
linguistique. Le mouvement atteignit mme les arabisants. De par sa
nature mme, le dh dal-Zajjj attira lattention, en particulier celle
1
Ce texte est la version crite de la leon ERASMUS faite au sminaire du Pr. Dr.
Andreas Kaplony, lOrientalisches Seminar de lUniversit de Zrich, le mardi 19 Avril
2005. Que les collgues et tudiants de lOrientalisches Seminar soient remercis pour
leurs remarques et questions, dont a bnfici la version finale.
114 pierre larcher
de Kees Versteegh, qui le traduisit dans le cadre de son MA (1971), et

lutilisa abondamment dans sa thse Greek Elements in Arabic Linguistic
Thinking (1977).
Nanmoins, cest seulement en 1995 que Versteegh publia cette tra-
duction, sous le titre de The Explanation of Linguistic Causes. Az-Zajjjs
Theory of Grammar.
Louvrage est typique du IVe/Xe sicle, en ce que sy rvle partout lin-
fluence de la falsafa (ou philosophie hellnisante), commencer sur la
forme mme de louvrage : celle, dialectique, hrite de lAntiquit grec-
que, par question et rponse.
Notre texte constitue le chapitre XIV (8990) de louvrage. Il rpond
la question de savoir pourquoi la grammaire a t nomme nahw en
arabe. Notons que cest le mme mot de illa, pluriel ilal, qui apparat
dans le titre du chapitre et dans le titre de louvrage. En revanche, dans
la formulation de la question, au dbut du chapitre, apparat celui de
sabab. Cela peut amener penser que les deux termes sont synonymes.
Et cest srement ce qui dtermine Versteegh traduire ilal par causes
dans le titre de louvrage et illa et sabab dans le texte du chapitre par le
mme mot de reason. Il existe cependant entre sabab et illa la mme
diffrence quen franais entre cause et justification. La cause est objec-
tive et relve de lordre logique : A parce que B. La justification en revan-
che est intersubjective et relve de lordre dialectique : A car/puisque B.
Noublions jamais que le rapport des autres disciplines la falsafa est
dialectique : par les questions quelle pose, la falsafa les oblige rpon-
dre, en se justifiant.
Ce texte est capital, non seulement pour lhistoire de la grammaire
arabe, mais encore celle de la langue arabe. Telle quelle a t comprise
par la tradition arabe, la premire tire en effet son origine de la corrup-
tion ( fasd) de la seconde.
2. Description
Le texte nous dit o, quand et comment ce processus a lieu et il nous dit

aussi en quoi il consiste.
O : Basra, cest--dire dans une des villes nouvelles cres la suite
de la conqute islamique, ce qui rpond en mme temps la question du
quand : Basra a t fonde en 16/637.
Comment : par un double processus de sdentarisation des Bdouins
et de mlange des populations arabes avec des populations non-arabes.
les origines de la grammaire arabe 115
Apparaissent dans le texte quelques termes fondamentaux. Dabord celui

de Arab, dans lexpression abn al-Arab (litt. Fils, enfants des Arabes),
oppos tout la fois al-h dira et abn al-Ajam (litt. Fils, enfants des
non-Arabes). Cela veut dire que Arab ne soppose pas seulement ici
Ajam comme Arabes non-Arabes (et, plus particulirement, dans le
contexte local, Persans), mais encore comme Bdouins sdentaires.
Au tmoignage mme du Lisn al-Arab (dsormais LA) de Ibn Manzr,
m. 711/1311 (art. hdr) : al-h adar, al-h adra et al-h dira sont le contraire
de al-bdiya et il sagit des villes, des villages et de la campagne (al-
h adar wa-l-h adra wa-l-h dira xilf al-bdiya wa-hiya al-mudun wa-
l-qur wa-r-rf ). Al-h dira dsigne donc bien le pays sdentaire par
opposition al-bdiya ou pays bdouin.
Voil pour la sdentarisation. Venons en maintenant au mlange
des populations. Celui-ci nest pas voqu explicitement au travers du
terme habituel de muxlata, mais implicitement au travers de celui de
muwalladna-na. On traduit ordinairement par mtis (< lat. mixti-
cius mlang ). On a voulu voir dans muwallad ltymon de multre
(Kazimirski 18467, art. wld), via lespagnol mulato, mais ce dernier
terme se rapporte peut-tre plus simplement au latin mulus/mula
( mule(t) , cf. en espagnol mme, outre mulo/mula mule(t) , muleto/
muleta jeune mule(t) ), tout le moins a subi une contamination de
cette famille lexicale. Dans la mme veine larticle muwallad de EI2 indi-
que quil sagit dun terme appartenant au vocabulaire des leveurs et
dsignant le produit dun croisement (tawld) entre deux races animales
diffrentes, donc un hybrid, un sang ml et que cest par analogie
que le terme a t tendu aux humains. Mais larticle ne donne aucune
rfrence pour ce sens, que nous ne trouvons pas, par exemple, dans LA
(art. wld). La question se pose donc au linguiste de larticulation de la
dsignation historique du terme et de sa signification. Morphologique-
ment, muwallad est le participe passif du verbe wallada. Wallada est le
factitif du verbe de base, mais il renvoie lactif ou au passif de celui-ci,
donc walada-yalidu ou wulida-yladu, selon quil est doublement ou
simplement transitif. Le verbe de base walada-yalidu, simplement tran-
sitif, signifie engendrer un enfant (et plus particulirement accou-
cher dun enfant , sil se dit dune femme, ou mettre bas un petit ,
sil se dit dune femelle). Le verbe doublement transitif wallada-h -hu
signifie faire en sorte quune femme accouche ou quune femelle mette
bas , cest--dire laider accoucher ou mettre bas. Muwallida est
un des noms de la sage-femme (qbila). Le verbe simplement transitif
wallada signifie faire natre quelquun ou gnrer quelque chose .
116 pierre larcher
Dans les deux cas, cependant, muwallad signifie, comme seul objet ou
second des deux objets de wallada, engendr . Cest srement par
lide de mis au monde que le terme muwallad a pris, tout la fois
par mtaphore et gnralisation, le sens de tout ce qui est nouveau,
moderne (al-muh dat min kulli ay LA, art. wld). Il peut se dire, soit
de quelquun, soit de quelque chose : il se dit en particulier des potes
modernes (al-muwalladn min a-uar) et des nologismes (sum-
miya al-muwallad min al-kalm muwalladan id istah dathu wa-lam
yakun min kalmihim fm mad ce quil y a de muwallad dans le par-
ler a t ainsi appel, quand on le produit, sans quil ait exist dans le
parler auparavant ). Comme tout ce qui est nouveau, le terme peut sen-
tendre en mauvaise part comme quelque chose de fabriqu, controuv,
apocryphe. Cest sans doute par une extension de ce dernier sens que
trs tt le terme a pris le sens de non purement arabe , pouvant se
dire, l encore, soit de quelquun soit de quelque chose, cf. LA, art. WLD
arabiyya muwallada wa-rajul muwallad id kna arabiyyan ayr mah d
de larabe ou un homme muwallad(a), sil nest pas purement arabe .
Historiquement, le terme sest dit des enfants ns, la suite des conqutes
islamiques, dunions mixtes, gnralement entre des pres arabes et des
mres non arabes. Si donc lon suit le mouvement smantique suggr
par LA, le sens de mtis nest pas mettre au dpart, mais au contraire
larrive dun processus dvolution smantique . . . Notons que ce der-
nier sens pourrait aussi satteindre par un simple et banal processus de
tadmn, consistant faire entrer dans un mot le sens de toute une
collocation, muwallad tant mis pour muwallad min muxlatat al-Arab
al-Ajam ( issu/produit du mlange des Arabes et des non-Arabes ).2
Maintenant, en quoi consiste prcisment ce processus de corrup-
tion de la langue ? Celui-ci est dcrit au travers dune anecdote mettant
en scne Ab l-Aswad ad-Dual et sa fille.
La tradition a vu dans ce personnage du Ier/VIIe sicle le pre de
la grammaire arabe. Pourquoi lui plutt quun autre ? Dabord, parce
quil est dorigine arabe : sa gnalogie complte, telle que donne par les
Tabaqt (21) de Zubayd, m. 379/98990, le rattache aux Kinna, tribu
2
Cest un processus fondamental, tant dans le lexique de larabe classique (e.g. jihd
guerre sainte mis pour jihd f sabl li-llh combat pour Allah , siysa politi-
que , mis pour siysa madaniyya ( gouvernement de la cit ) que dans celui de larabe
moderne (tlib tudiant mis pour tlib al-ilm celui qui cherche le savoir , amn
secrtaire mis pour amn as-sirr dpositaire du secret ).
de la rgion de la Mecque. Il serait n vers 606 de notre re et aurait

sjourn chez les Quayr dArabie centrale (nous verrons ultrieurement
limportance de ces notations). Ensuite, ralli Al, il est nomm par ce
dernier cadi puis gouverneur de Basra, en 36/656, mais la disparition de
Al en 41/661 ne lempche pas, comme le montre notre texte (et dautres
sources) dentretenir des relations avec le gouverneur Umayyade Ziyd
b. Abhi. Il serait mort vers 69/688. Enfin, il tait galement pote. Le
personnage est donc le prototype de larabe sdentaris, que son origine
et son parcours dsignent comme un matre et un gardien de la langue.
Il est de Basra, dont la tradition ultrieure fera le sige de lcole domi-
nante de la grammaire arabe, et voir dans un Basrien le pre de la gram-
maire arabe nest rien dautre quune manire de signaler lanciennet du
travail grammatical dans cette ville et surtout son antriorit par rapport
lcole rivale de Kfa. Il est consensuel socialement et politiquement :
nomades et sdentaires, partisans de Al comme des Umayyades peu-
vent sen rclamer.
Venons en maintenant lanecdote elle-mme. La fille dit son pre
m aaddu l-h arri. Celui-ci interprte le propos comme une question
sur la chaleur la plus intense ( Quelle est la chaleur la plus intense ? )
et rpond, par suite, ar-ramdau f l-hjira, cest--dire la canicule en
plein midi . Mais la fille rejette cette interprtation lam asalka an hd
je ne tai pas demand cela , ajoutant je me suis tonn de lintensit
de la chaleur (innam taajjabtu min iddati l-h arri). Alors dis, rtor-
que son pre, m aadda l-h arra ( Quelle chaleur intense !) .
Autrement dit le pre reproche sa fille davoir confondu les structu-
res interrogative (istifhm) et exclamative (taajjub) et davoir employ
lune pour lautre. Cest cela un lah n, autre terme fondamental apparais-
sant dans le texte : non pas une faute de langage en gnral, mais une
faute contre la flexion dsinentielle, casuelle et modale, en particulier.
Cest cette flexion quon appelle en arabe mme irb. Une telle faute
serait dautant plus grave quelle crerait un quiproquo.
De cette anecdote, il existe plusieurs versions (comme le reconnat
le texte lui-mme) : Versteegh (1997b, 59) cite la version donne par
Srf (m. 368/979), dans son Axbr (d. Krenkow, 19) avec les exem-
ples de m ah sanu as-sami What is the most beautiful thing in the
sky? / m ah sana s-sama How beautiful is the sky! .
A travers cette anecdote, il sagit en fait de mettre au centre de la
arabiyya ou langue des Arabes le irb (mot tymologiquement li
Arab) et de suggrer que cet irb est pertinent, en ce quil distingue des
118 pierre larcher
significations (muwaddih , mubayyin, mufarriq, munabbi . . . li-man

al-lua), pour citer ici quelques-uns des mots que lon relve dans les
sources.
Et cest laggravation de la situation qui amne la constitution de la
grammaire, son nom arabe de nahw tant justifi par le fait que le gram-
mairien se fait lindicateur de la voie suivre (unh hd n-nahw).
Nahw est en effet le masdar du verbe nah -yanh , quon emploie tou-
jours comme circonstanciatif, fig laccusatif (nahwa), de sens vers .
Peu importe si cette tymologie est fantaisiste ou non. Ce qui nous
intresse ici, ce sont les diffrentes interprtations quun linguiste est
susceptible de faire de ce texte pour lhistoire de la langue.
3. Interprtation
3.1 De lhistoire . . .
Au XIXe sicle, la linguistique, ne au tournant du XVIIIe et du XIXe,
sous forme de la grammaire compare (des langues indo-europennes),
devient historique. On ne cherche plus seulement reconstruire en
amont des protolangues (Ursprache). Plus modestement, on cherche
retracer en aval lvolution des langues existantes.
La linguistique historique est une spcialit essentiellement allemande.
On ne stonnera donc pas que ce soient les arabisants allemands qui, les
premiers, se sont intresss lhistoire de larabe. Ils rinterprteront le
fasd al-lua de la tradition arabe, caractrise par les lah n ou fautes de
irb, comme le signe dune volution dun type ancien arabe (en alle-
mand Altarabisch et en anglais Old Arabic) vers un type no-arabe (en
allemand Neuarabisch et en anglais New ou Neo-Arabic). Le type ancien
arabe est videmment caractris par lexistence dune flexion dsinen-
tielle, casuelle et modale, le type no-arabe par la disparition de cette
flexion. Le type ancien arabe est donc plus synthtique et le type no-
arabe plus analytique. Corollairement, dans le type ancien arabe, lordre
des mots est plus libre, mais, dans le type no-arabe, moins libre. Quand
le type ancien arabe commence se dgrader en type no-arabe, nous
entrons dans le moyen arabe (en allemand Mittelarabisch et en anglais
Middle Arabic). Cest Heinrich Leberecht Fleischer (18011888) qui
projettera, implicitement dans un article de 1847, explicitement dans
un article de 1854, sur larabe cette tripartition clbre en linguistique
historique. Voici ce quil crit en 1854 (23) :
Als heilige Sprache des Islam, Organ der Gelehrsamkeit und hhern Wis-
senschaftlichkeit, Mittelpunkt oder vielmehr ausschliesslicher Gegen-
stand aller Schulphilologie, steht das Altarabische seinem Abkmmling,
dem Neuarabischen, in der Anschauung des Morgenlandes selbst schroff
gegenber. Nur jenes heisst bei den Gelehrten al-luah, die Sprache, al-
arabiyyah, das Arabische schlechthin, dieses al-lisn al-mm oder al-
mm, die gemeine Mundart, la lingua volgare.
Il est clair, daprs la description mme quen donne Fleischer, que lan-
cien arabe est larabe classique et le no-arabe larabe dialectal et non
moins clair que si les deux varits coexistent en synchronie, larabe dia-
lectal est explicitement compris comme tant historiquement le des-
cendant (Abkmmling) de larabe classique.
Un peu plus loin (4), il mentionne le moyen arabe. Il lavait dj exac-
tement dcrit (du point de vue de la linguistique historique) en 1847,
propos de la langue dun codex grco-arabe (155), quil compare celle
des Mille et une nuits, auxquelles il avait consacr sa dissertatio en 1836 :
Wie in der Tausend und Einen Nachten sind auch hier einzelne jener
ltern Formen mit der neuern gleichsam noch im Kampfe begriffen ;
willkrlich tritt bald die eine, bald die andere ein . Ltat moyen dune
langue se caractrise en effet par lalternance, en synchronie, dlments
interprtables, en diachronie, comme relevant encore de ltat ancien
(lter) ou dj de ltat moderne (neuer).
Un de ses lves, Ignaz Goldziher (18501921), dans un crit de jeu-
nesse rdig en hongrois et aujourdhui traduit en anglais, compare expli-
citement la relation entre ancien arabe et no-arabe celle du latin et des
langues romanes, appeles jadis no-latines (Goldziher 1994, 20) :
As French abandoned the case inflection of Latin and developed the
Roman synthesis into analysis, making de lhomme from hominis, so did
the living Arabic of today dissolve the old rajulin into met r-rajul ; as latin
scrip-si developed into French jai crit (. . .), so was Old Arab[ic] aktubu
turned into biddi aktub ou bi-aktub.
Cette conception allemande de lhistoire de larabe se retrouve au
XXe sicle, en particulier chez Johann Fck (18941974) dans son grand
ouvrage Arabya (Fck, 1955[1950]), et, aujourdhui encore, chez Jos-
hua Blau, le grand matre du moyen arabe (e.g. Blau 2002, 16).
Objectivement, cette conception rejoint celle que sen faisait au VIIIe/
XIVe sicle Ibn Xaldn (m. 808/1406) et quil expose deux reprises
dans la Muqaddima : une premire fois dans la section 22 (6757), inti-
tule f lut ahl al-amsr ( Des parlers des habitants des villes ) du
chapitre IV, consacr aux villes et pays , et une seconde aux chapitres
120 pierre larcher
47 et 48 (10731080), consacrs respectivement aux parlers noma-

des et aux parlers sdentaires de son temps, du chapitre VI, lui-mme
consacr aux sciences . Pour Ibn Xaldn, ce quil appelle langue de
Mudar (luat Mudar, al-lisn al-mudar) est la langue premire (al-
lisn al-awwal) et originelle (al-lisn al-asl) de lArabie et la langue
du Coran et du h adt. Cette langue est parle de manire dautant plus
chtie ( fash a) que ceux qui la parlent sont plus loigns des autres
nations, la manire de parler des Quray tant pour cette raison la plus
chtie et la plus pure des manires de parler arabes (wa-li-hd knat
luat Quray afsah al-lut 3 al-arabiyya wa-asrah uh): si lon retraduit
en termes gographiques la gnalogie, lappellation langue de Mudar
revient dsigner le centre et louest de lArabie comme le domaine de
larabe fash . Ibn Xaldn fait ici la synthse de deux thses, sur lesquel-
les nous reviendrons ci-dessous : dune part la thse philologique (qui
voit ce domaine comme constitu de deux sous-domaines dits Hedjaz
et Tamm),4 et dautre part la thse thologique (qui identifie, sur la base
de Cor. 14:4, la langue du Coran la luat Quray et, sur une base dog-
matique, la luat Quray la lua al-fush ), tout en les croisant avec une
thse philosophique, issue du Kitb al-h urf de Frb (m. 339/950),5
et qui est le corrollaire de la thse liant corruption et mlange .
A linverse, cette langue est dj corrompue ds lpoque prislamique,
l o les Arabes sont en contact avec dautres nations, et se corrompt
encore davantage aprs la conqute islamique et les nouvelles fondations
urbaines, et donc en milieu sdentaire plus encore que nomade, jus-
qu donner naissance de nouvelles langues. Loriginalit dIbn Xaldn
est en effet de ne pas considrer les dialectes comme de simples formes
dgrades de la langue de Mudar , mais comme des varits autono-
mes par rapport celle-ci et distinctes delle, en ce quelles ont substitu
la syntaxe base sur la flexion dsinentielle une syntaxe de position
(at-taqdm wa-t-taxr).6
3
Cest cette expression qui donne, par rcriture, celle de al-lua al-fush .
4
Sur cette subdivision, cf. Rabin (1951). On comprend pourquoi Ab l-Aswad
d-Dual, natif du Hedjaz, est dit avoir fait un dtour par lArabie centrale . . .
5
Du moins la version de ce texte connue par le Muzhir (I:211212) ou, mieux, le
Iqtirh (20) de Suyt (m. 911/1505), non celle publie par Mahdi en 1969. Sur les deux
versions de ce texte, cf. Langhade (1994, 248258) et Larcher (2006a).
6
Sur Ibn Xaldn et lhistoire de larabe, cf. Versteegh (1997a, 153165) et Larcher
(2006b).
3.2 la sociolinguistique
En citant lexpression italienne de lingua volgare, certes attire par celle
de al-lisn al-mm(),7 Fleischer montre quil na pas seulement en tte la
linguistique historique, mais aussi le modle italien. Or, dans ce modle,
la langue vulgaire nest videmment pas dsigne en termes diachro-
niques, cest--dire historiques, comme descendant du latin ; elle lest
au contraire, en termes synchroniques et sociolinguistiques, comme
langue du vulgum pecus, par opposition au latin langue des clercs .8
Mais, l encore, cette conception, qui, pour larabe, sera baptise ult-
rieurement par les arabisants diglossie,9 rejoint objectivement celle que
se font les auteurs de langue arabe, commencer dailleurs par az-Zajjj
lui-mme. Au chapitre XVII, intitul bb dikr al-fida f taallum al-
nahw, il demande (95) :
quoi sert dapprendre la grammaire, la plupart des gens parlant naturel-
lement10 sans flexion dsinentielle, quils ne connaissent pas, tout en com-
prenant les autres et en [se] faisant comprendre deux ( fa-m al-fida f
taallum an-nahw wa-aktar an-ns yatakallamn al sajiyyatihim bi-ayr
irb wa-l marifa minhum bihi fa-yafhamn wa-yufhimn ayrahum
mitl dlika).
7
Fleischer ne donne aucune rfrence pour ces deux expressions. Cest dommage,
car si, lpoque o il crit (milieu du XIXe sicle), lexpression al-arabiyya est, comme
il le note, couramment utilise, par une mtonymie significative, pour dsigner larabe
classique, cest lexpression de al-lua ad-drija qui est utilise pour dsigner larabe dia-
lectal. Lexpression dal-lua al-mmiyya (vs al-lua al-fush) napparatra que vers la
fin du XIXe sicle, du moins comme nom de cette varit, mais ds le Moyen Age, on la
rencontre pour dsigner un vulgarisme au sein de la langue.
8
La comparaison avec la situation italienne ne peut dailleurs tre pousse trop
loin sans aporie. Le domaine arabe na pas connu la rvolution qua connue le domaine
roman (et, mutatis mutandis, lEurope entire), savoir la promotion des langues vul-
gaires au rang de langues littraires, ce qui fera du latin (et seulement pour un temps,
plus ou moins long selon les pays) le vhicule de la seule culture savante. Ainsi, aprs
Il cantico delle creature (1226) de Saint Franois dAssise (11821226), Dante Alighieri
(12651321), logiquement, crit La Divine Comdie en langue vulgaire, mais traite de
celle-ci en latin (De vulgari eloquentia).
9
Ce terme, venu de la linguistique no-hellnique (1885), a t explicitement intro-
duit en linguistique arabe par William Marais (18741956), dans un article de 1935,
avant que le concept ne soit thoris, partir de larabe et dautres langues, par Charles
A. Ferguson (19211998), dans un article de 1959. Pour le dtail, cf. Larcher (2003).
10
Blanc (1979, 165, n. 20) traduit par spontaneously et Versteegh (1995) par
intuitively .
122 pierre larcher
Il revient sur ce point la fin du chapitre (96) :

Quant aux gens du commun qui parlent larabe sans flexion dsinentielle,
on les comprend. Mais cela est seulement possible pour ce qui est bien
connu et dusage courant, ce dont on a une connaissance familire et est
usit. Mais si, daventure, lun deux se risquait claircir une ambigut,
sans le faire comprendre au moyen de la flexion dsinentielle, il ne le pour-
rait pas ( fa-amm man takallama min al-mma bi-l-arabiyya bi-ayr
irb fa-yufham anhu fa-innam dlika f al-mutaraf al-mahr wa-l-
mustamal al-malf bi-d-dirya wa-law iltajaa ah aduhum il al-dh an
man multabis min ayr fahmihi bi-l-irb lam yumkinhu dlika).
Alors que le chapitre XIV concerne le Ier/VIIIe sicle, le chapitre XVII
concerne lpoque dal-Zajjj lui-mme, cest--dire le IVe/Xe sicle. La
situation dcrite dans ce chapitre semble pouvoir tre interprte comme
laboutissement du processus dcrit au chapitre XIV. Au Ier/VIIIe sicle
le type ancien arabe commence se dgrader en type no-arabe. Trois
sicles plus tard, cette dgradation a abouti, non une substitution dun
type lautre, mais, le type ancien arabe subsistant, une coexistence
des deux, chaque type tant caractris non seulement linguistiquement
par la prsence/absence du irb, mais encore socialement et culturel-
lement : dire que le type non flchi est lexpression naturelle de la
plupart des gens revient dire que le type flchi est celle non seulement
dune minorit, mais encore lexpression artificielle de cette minorit
(artificielle, puisque scolairement apprise et non naturellement acquise
et pratique !). Qualifier explicitement la majorit de mma revient non
seulement qualifier implicitement la minorit de xssa , mais encore
semble prfigurer lappellation de al-lua al-mmiyya, aujourdhui uti-
lise pour dsigner larabe dialectal. Enfin reconnatre par deux fois que
labsence de flexion dsinentielle ne nuit en rien la communication
explique que, pour sauver cette flexion (son gagne pain !), le grammai-
rien distingue entre communication quotidienne et communication
savante et cherche, voire fabrique, des structures ambigus, que seule
la flexion, censment pertinente, est mme de dsambiguser (on voit
tout de suite que lexemple propos au chapitre XIV relve exactement
de ce cas de figure et, comme tout arabisant en a fait lexprience, ce petit
jeu perdure, jusqu aujourdhui, dans certains milieux puristes !).
La vrit oblige cependant dire quil subsiste un doute sur cette
interprtation. On doit se rappeler ce que dit al-Jhiz (m. 255/869) dans
le Bayn (I:137). Il noppose nullement mma xssa comme la masse
illettre llite lettre, mais seulement, parmi les gens lettrs, les gens
ordinaires aux happy few. Cest exactement le sens que les spcialistes
donnent mma dans la littrature spcialise des ouvrages de lah n al-

mma.11 Il ny a rien l que de logique : lah n prsuppose irb ; par suite,
sil ny a pas irb, il ny a pas lah n ! Si lon projette le sens jhizien de
mma sur le texte daz-Zajjj, il ne sagirait plus alors de deux vari-
ts darabe, mais seulement de deux registres dune mme varit : lun
soutenu (avec ralisation du irb) et lautre relch (sans ralisa-
tion de ce irb). Jusqu aujourdhui lcole enseigne le irb qui terro-
rise (irhb) les apprenants, parce que sa ralisation est source de lah n,
do le sage conseil ijzim taslam Supprime la voyelle brve finale, tu
seras prserv de lerreur ! . Un seul lment du texte daz-Zajjj peut
faire pencher la balance en faveur de la premire interprtation, plu-
tt que de la seconde : cest la notation que les locuteurs de larabe sans
flexion dsinentielle nont pas connaissance de celle-ci . La mma, au
sens dal-Jhiz, ne lignore pas ; elle en a seulement une connaissance
imparfaite12 . . .
4. Discussion
On le voit : tout tourne autour du irb, tant dans le texte daz-Zajjj que
dans linterprtation, tout la fois historique et sociolinguistique, qui en
est faite. Pourtant, ds avant le XIXe sicle et lessor de la linguistique his-
torique, une tout autre tendance tait apparue chez certains arabisants.
Ds le XVIIIe sicle, des arabisants, galement allemands dailleurs,
comme Johann Davis Michaelis (17171791) et Johann Gottfried Hasse
(17591806), avaient dvelopp ce que Gruntfest (1991), qui la tu-
die, appelle A Early theory of Redundancy of Arabic Case Endings .
Notons quils rinterprtaient des ides dj exprimes, au XVIIe
sicle, par lItalien Antonius ab Aquila (Antonio dellAquila), Franciscain
envoy en mission auprs des Chrtiens dAlep et auteur de la premire
11
On lira avec profit larticle Lah n al-mma, de EI2, d Charles Pellat (1914
1992).
12
On notera que dans lunique manuscrit, dat de 617H, qui sert de base ldition du
dh , il y a dans lavant-dernier paragraphe de ce chapitre XVII, une magnifique faute de
irb : lan yumkin ah ad [corrig par lditeur en ah adan] min al-muwalladn iqmat-hu
ill bi-marifat an-nahw ( Nul, parmi les muwalladn, ne pourrait ltablir (la posie),
sauf par la connaissance de la grammaire ). Labsence du alif rvle au minimum quil
nest pas prononc, voire suggre que ah ad a t trait comme le sujet de yumkin et par
suite iqma comme lobjet, ce que peut galement suggrer le masculin yumkin.
124 pierre larcher
grammaire de la langue (arabe) vulgaire (1650).13 Notons galement

que tous les arabisants du temps savaient le latin et le grec : quand on a
souffert dans sa jeunesse sur une version grecque ou latine, on voit tout
de suite la diffrence entre grec et latin classiques et arabe classique. En
grec et en latin, cest la dclinaison qui permet de construire la phrase ;
en arabe, cest la construction de la phrase qui permet de restituer la
dclinaison . . . Les grammairiens arabes le reconnaissent eux-mmes
implicitement, en couplant, dans leur thorie, le irb au amal : le irb
est leffet de l action dun lment dit mil sur un autre dit maml,
ainsi lan, dit nsib, qui dtermine le nasb (subjonctif ) du mudri (inac-
compli) dit mansb. Par ailleurs, le fait mme que les grammairiens ara-
bes cherchent des cas o cette flexion serait pertinente suffit prouver
quelle ne lest pas ! On voit tout de suite le caractre artificiel de lexemple
propos par az-Zajjj. Dans beaucoup de langues, structures interroga-
tives et exclamatives sont confondues, tant distingues par lintonation,
ainsi en franais ou en allemand Quel artiste/Welcher Knstler ? vs Quel
artiste/Welch ein Knstler !
Les exemples donns de l aggravation de la situation , qui font
dfaut chez az-Zajjj lui-mme, mais quon trouve chez dautres auteurs,
ne sont gure plus convaincants. Ainsi le fameux tuwuffiya abn wa-
taraka banna ( notre pre est mort et a laiss des fils ) est si peu pro-
bant pour un fasd al-lua interprt en termes historiques que Versteegh
(1997b, 51), qui le cite daprs le Nuzha dIbn al-Anbr (m. 577/1181),
linterprte en termes sociolinguistiques (alors mme que lanecdote
concerne le temps de Ziyd, donc le Ier/VIIIe sicle). En termes histori-
ques, on se serait en effet attendu lemploi des formes no-arabes ab
et bann (respectivement nominatif de la flexion triptote ab// et cas
rgime de la flexion diptote ban/(na) du type ancien arabe), de sorte
quavec la phrase donne en exemple, il ny aurait pas eu de faute !
En termes sociolinguistiques, en revanche, la coexistence de deux types
dont lun a plus de prestige que lautre peut amener le locuteur, dans une
circonstance formelle, substituer lunique forme du type no-arabe,
quil emploie ordinairement, mais pense fautive, lautre forme du type
13
Cf. Fck 1955, 78. Fck considre que la Fabrica overo Dittioniario della lingua vol-
gare arabica et italiana (1636) de Dominicus Germanus de Silesia (15881670) nest pas,
malgr son titre, un dictionnaire, mais une introduction, presque sans valeur, larabe
vulgaire. En tout cas, on voit que ce sont des clercs, Italiens ou lis lItalie, qui, sous
lappellation, valant signature, de langue vulgaire , sont les inventeurs de larabe
dialectal. Cette double qualit, jointe au fait quils taient des hommes de terrain et non
de cabinet, les y prdisposait.
ancien arabe, quil nemploie pas et pense correcte : pour viter une faute,
il en commet une autre !
Certes, ces arabisants voyaient dans les dsinences casuelles une
invention des grammairiens arabes, position encore dfendue au XIXe
sicle par Alos Sprenger (18131893). Si personne aujourdhui ne croit
plus cela, en revanche la redondance de la flexion en arabe classique
amne beaucoup darabisants douter que lhistoire de larabe consiste
en une volution du type ancien arabe vers le type no-arabe, caractri-
ss respectivement par la prsence et labsence de cette flexion.
Deux positions se sont fait jour. Lune est trs rpandue chez les ara-
bisants. Lakkadien possdant une flexion casuelle triptote et dautres
langues smitiques exhibant des restes de flexion casuelle, les arabisants
admettent que la flexion casuelle de larabe classique est un trait de haute
antiquit, qui sest maintenu dans le seul registre potique de la langue.
Sous linfluence de la reprsentation diglossique de larabe, rtoprojete
sur lhistoire de la langue, ce registre potique est souvent vu comme
une langue commune (koin), vhiculaire, oppos aux vernaculaires que
sont les anciens dialectes arabes. Cette koin serait galement la langue
du Coran, quelques hedjazismes prs. Malgr les apparences, cette
hypothse arabisante nest pas si diffrente de la thse thologique ! Cel-
le-ci, on la vu, se rsume en une double identification : lune, opre sur
une base scripturaire, de la langue du Coran avec la langue de Quray
et lautre, opre sur une base purement dogmatique, de la langue de
Quray avec al-lua al-fush . Il y a un sicle, Vollers (1906) acceptait la
premire identification, mais refusait la seconde. Il supposait en effet que
le Coran avait dabord t nonc et crit dans le vernaculaire de la Mec-
que, parler ouest-arabique dpourvu de irb, avant dtre rcrit dans la
langue vhiculaire de la posie, parler est-arabique pourvu de ce irb.
Kahle (1959[1947]) reprit cette hypothse en lattnuant : si le ductus
coranique (rasm) reflte le vernaculaire de la Mecque, les qirt, elles,
refltent la langue vhiculaire de la posie. La plupart des arabisants (cf.
Blachre 1952, 6682) refusent la premire identification et, par suite,
la seconde, mais acceptent la troisime, dcoulant de la suppression
du moyen terme, i.e. langue du Coran = al-lua al-fush . Or, si lon relit
le clbre texte du Sh ib (5253) dIbn Fris (m. 395/1004), dont la
source nest autre que celui attribu al-Farr (m. 207/822) et jadis
exhum par Kahle (1959[1947]), on saperoit que pour concilier vrit
thologique et vrit philologique (i.e. le fait que la langue du Coran,
identifie la luat Quray, exhibe des traits, les fameux hedjazismes ,
qui ne sont pas ceux de la lua al-fush ) ces auteurs imaginaient un
126 pierre larcher
scnario expliquant comment la luat Quray tait devenue la lua

al-fush : comme produit dun processus de koinsation justifi par le
fait que, la Mecque tant un centre de plerinage intertribal, les Quray
avaient ainsi pu slectionner (taxayyur) le meilleur de chaque parler
arabe !14
Mais une autre position, originale et marginale, est en train de gagner
du terrain. La posie archaque pratique la qfiya mutlaqa ( rime abso-
lue ), cest--dire ralise les voyelles brves u, a et i, avec ou sans tanwn,
uniformment comme des voyelles longues , , . Cest dire si le tanwn
ne sert rien !15 La rcitation psalmodie du Coran (tajwd) va plus loin :
elle pratique gnralement la qfiya muqayyada ( rime lie ), cest--
dire supprime les voyelles brves u, a et i en finale. Elle supprime gale-
ment -un et -in, et ralise -an comme -. Cette prononciation pausale de
-an en - est la seule trace de flexion casuelle qui apparaisse dans le mai-
gre matriel pigraphique conserv16 et cest aussi sa seule manifestation
dans les dialectes modernes (marh aban ou ahlan wa-sahlan prononcs
marh ab et ahl w-sahl). Le cas du suffixe relateur -(V)n quon trouve
dans maint dialecte ancien et moderne est disjoindre, dans la mesure
o sa fonction (marquer la relation mawsf/sifa, celle-ci pouvant tre un
N, un SP ou une phrase) ne le relie en rien au tanwn de larabe classi-
que et amne penser que cest la conservation dun trait archaque (cf.
Ferrando 2000).
Quon doive supprimer les voyelles brves en finale (y compris donc
les voyelles flexionnelles) est videmment un argument en faveur de la
non pertinence de cette flexion sur le plan syntaxique.17 En revanche,
quen dehors de la pause, elle soit ralise, suggre que cette flexion nest
pas syntaxique, mais prosodique. Lide tait dj mise au XIXe sicle
par Johann Gottfried Wetzstein (18151905), dans un article paru en
1868. Les historiens de la grammaire arabe savent aujourdhui, grce
Versteegh (1981[1983]), quil sest trouv au moins un grammairien
arabe, Qutr ub (m. 206/821), pour anticiper cette ide. Nous connais-
sons sa thorie grce az-Zajjj (ch. VII, 6971, bb al-qawl f al-irb
14
Sur ces deux textes, cf. Larcher 2004 et 2005b.
15
Comme le suggre en outre la ralisation du long rsultant de la qfiya mutlaqa
comme . . . -an (tanwn at-tarannum) attribue aux Tamm, dans la rcitation potique
(ind). Dans les deux cas, on peut donc avoir larticle et voir et/ou entendre /an.
16
Un exemple dans linscription dUmm al-Jiml (Ve ou VIe sicle ap. JC ?), avec un
mot lu successivement par Littmann en 1929 et 1949 comme iyran et afran. Nous
parlons bien sr ici de la seule flexion marque par des voyelles brves.
17
Cf. lanalyse qui est propose de Cor. 85 :2122 dans Larcher (2005a).
lim daxala f-kalm pourquoi la flexion sest-elle introduite dans le

discours ? ), qui explique que pour Qutr ub les voyelles de flexion sont
lies lenchanement (wasl) des mots dans la phrase. Des ides, sinon
identiques, du moins comparables sont dfendues aujourdhui par
Owens (1998).
Un bon argument en faveur de la conception prosodique et non syn-
taxique de la flexion peut tre trouv dans le traitement contradictoire de
Tamd dans le Coran. Contradictoire, dans la mesure o il est trait par-
tout comme un diptote sans tanwn Tamd-u/a, sauf en 11 : 68 ; 25 :38 ;
29 :38 ; 53 :51, o apparat un alif. Alors que H afs an Asim (Coran du
Caire) neutralise ce alif, War an Nfi (Coran du Maghreb) le traite
bien, en ces endroits, comme un triptote avec tanwn et lit Tamdan.
Pourquoi, donc, deux dclinaisons pour un mme mot, alors quaucune
des deux ne sert rien sur le plan syntaxique (en 17 :59, nous avons un
Tamd complment dobjet sans alif) ? On a tt fait dobserver que dans
trois des quatre cas o apparat ce alif, Tamd est coordonn d, ga-
lement muni de ce alif (et partout trait lui comme un triptote d-u/
a/i-n). Lapparition du alif peut sexpliquer par un simple et banal ajus-
tement forme/sens : cest la coordination syntaxique de deux mots qui,
smantiquement, vont ensemble, comme noms de peuples reprouvs ,
qui explique quon donne au second le traitement syntaxique du pre-
mier. Lexplication ne vaut cependant pas pour le quatrime cas (11 :68),
o Tamd apparat seul : (ka-al-lam yanaw fh) a-l inna Tamda
(H afs)/an (War) kafar rabbahum a-l budal-li-Tamd[a]. Sauf lire
ce alif comme il est crit, cest--dire non comme -an, mais bien comme
-, auquel cas il forme aussitt assonance avec les nombreux - de son
environnement syntaxique immdiat, que nous avons mis en gras . . . On
peut aller plus loin. Le dbut du verset, que nous mettons entre paren-
thses, va avec ce qui prcde. Le reste sorganise en un paralllisme
marqu par lanaphore de a-l. On voit que si on lit, non budan, do,
par assimilation (idm) du nn au lm budal-li-, mais bud, on na
plus seulement alors une assonance en , mais encore une rime interne
en -d, soit : a-l inna Tamd kafar rabbahum a-l bud li-Tamd
Hol ! Oui, les Thamoud, ils ont t infidles leur seigneur ! Hol !
Arrire aux Thamoud ! .18
18
Tamd est loin dtre un cas unique. On peut galement citer salsil de 76 :4 et
qawrr de 76 :1516. Par ailleurs on a compris que, pour ma part, je ne considre
pas - comme la prononciation pausale de -an, mais -an comme une rinterprtation
ultrieure dun - fondamental . Linscription du Jabal Usays (5289 ap. JC) montre
128 pierre larcher
On ne perdra pas de vue que lhistoire de larabe est toujours dpen-

dante des ides du moment. Ainsi Versteegh (1984), surfant sur la vague
des tudes croles, a propos de rinterprter le fasd al-lua dans les
termes dun processus de pidginisation-crolisation. La conqute isla-
mique a pour effet de mettre en contact des populations arabophones
et des populations non arabophones. On peut donc imaginer, pour les
besoins de la communication, lmergence de langues de contact ht-
rognes, ce quon appelait jadis des sabirs et quon appelle aujourdhui
des pidgins (Fck 1955 [1950], 8, en fait dj lhypothse). Si les enfants
qui naissent dunions mixtes (les muwalladn) la prennent pour langue
maternelle, cette langue de contact devient un crole. Et, enfin, si au
contact de larabe, le crole se rarabise (ou se dcrolise), ce pidgin-
crole dcrolis peut tre vu comme le point de dpart des dialectes
arabes, sur le modle du Juba-Arabic au Sud Soudan.19 Lhypothse de
Versteegh suscita un vaste dbat et fut gnralement rejete (cf. Holes
1995, 1924). Il me semble nanmoins quon ne devrait pas jeter le bb
avec leau du bain !20 On ne peut exclure a priori que des processus de ce
genre aient jou, localement, un rle. Pour ma part, je me contenterai
de signaler et souligner ici une phrase de la Muqaddima dIbn Xaldn
qui ma fait parler, pour tenter une comparaison loquente, de case de
lOncle Tom sur les bords du Tigre et de lEuphrate .21 Traitant de la
gense des parlers sdentaires dans les diffrentes rgions du monde
musulman (Maghreb, Machrek, Andalousie), il crit propos du second
(10771078) :
De mme, quand les Arabes eurent vaincu les nations de lOrient, Per-
sans et Turcs, et se furent mlangs eux, les langues de ceux-ci se dif-
fusrent parmi eux, par lintermdiaire des laboureurs, des paysans et
des captifs quils prenaient comme intendants, sages-femmes, pres nour-
riciers et nourrices : leur langue se corrompit du fait de la corruption de
quavant linvention du t marbta les deux ralisations dun mme morphme se tra-
duisent par deux graphies : -h la pause, mais -t en liaison (il reste des traces de cet tat
de choses dans le ductus coranique, avec des exemples de rah mat ou nimat). On voit
dautant moins pourquoi les deux ralisations phoniques du tanwnan de larabe classi-
que auraient donn lieu dans le matriel prclassique une graphie unique (correspon-
dant -) que, dans les textes du moyen arabe, le suffixe relateur -(V)n se traduit dans
la graphie par un n.
19
Ma collgue Catherine Miller, consulte, mindique qu lheure actuelle le Juba-
Arabic est encore loin dtre un dialecte arabe. En outre les choses vont un peu dans tous
les sens, manifestant des tendances contradictoires.
20
Cf. dailleurs, Versteegh lui-mme (2004).
21
Cf. Larcher (2006b).
lhabitus, au point de muter en une autre langue (wa-kad al-Mariq

lamm al-Arab alaba al umamihi min Fris wa-t-Turk fa-xlatahum
wa-tadwalat baynahum lutuhum f l-akara wa-l-fallh n wa-s-subiyy
alladna ittaxadhum xawalan wa-dytin wa-adh ran wa-mardi fa-fa-
sudat luatuhum bi-fasd al-malaka h att inqalabat lua uxr).
Il serait videmment intressant de remonter ici aux sources mmes
dIbn Xaldn. Bien sr, la notation va dans le sens de largumentation
dIbn Xaldn (interruption du processus naturel dapprentissage de la
langue de Mudar, par transmission dune gnration lautre dArabes,
qui seule cre l habitus ). Mais mme le plus intgriste des crolis-
tes ne peut nier que la phrase souligne dcrit ce que ce genre de cro-
listes considre tre la base objective, sur le plan social, dun processus
de crolisation : une population servile, parlant une ou dautres langues
que celle de ses matres. En outre deux, voire trois, des quatre catgo-
ries mentionnes renvoient ltymologie mme de crole (< lat. criare
nourrir, lever ), terme qui peut dsigner aussi bien les enfants blancs
que les esclaves noirs et leurs langues (gnralement le parler des Noirs,
mais au moins en un cas, celui de Saint-Barthlemy dans les Antilles,
celui des Blancs).
La linguistique arabe est aujourdhui sous linfluence de la sociolin-
guistique amricaine variationniste , issue des travaux de William
Labov (n en 1927). Cela a amen au dpassement de la reprsentation
diglossique de larabe. Mais cela a aussi amen, en matire dhistoire de
la langue, renouer avec la vision des plus anciens grammairiens arabes,
notamment Sbawayhi (m. 177/793 ?) et al-Farr. Ceux-ci concevaient
larabe tout la fois comme une langue une et plurielle, lua faite de
lut, cest--dire non pas de varits autonomes, mais de variantes, tri-
bales ou rgionales, bonnes ou mauvaises (cette hirarchie suffit bien sr
pour ne pas en faire des variationnistes avant lheure !). Aujourdhui,
larabe dit classique est gnralement compris, non comme un tat de la
langue, mais comme une langue standardise, cest--dire retenant cer-
taines lut et en liminant dautres. Elle est un aboutissement, non un
point de dpart, et, par suite, les dialectes arabes modernes ne sauraient
en tre des corruptions .
Dans la mesure o beaucoup des traits dcrits par les grammai-
riens arabes (notamment les plus anciens dentre eux : Sbawayhi et
al-Farr) se retrouvent, identiques ou analogues, dans les parlers ara-
bes daujourdhui, il ny a pas de raison de penser quils ne prolongent
pas les anciens parlers arabes, selon des modalits complexes en fonc-
tion des lieux et des temps (cest le travail de la dialectologie historique
130 pierre larcher
que de dbrouiller lcheveau). Simplement, mme si les parlers arabes

daujourdhui relvent uniformment de ce qui est pour la linguistique
historique le type no-arabe, on se gardera bien de croire quils des-
cendent uniformment dun type ancien arabe. Lopposition des deux
types, compte tenu de la non-fonctionnalit de la flexion dsinentielle
en arabe classique mme, apparat trs largement outre et, par suite,
ils ont pu trs bien coexister date ancienne et, non pas en opposition,
mais en continuit . . . Un argument en ce sens peut dailleurs tre trouv
dans la faute releve dans lun des deux plus vieux papyrus dats (22
de lHgire), o lon a Ab Qr pour Ab Qr (cf. Diem 1984, 271). Ab
Qr nest pas un nom arabe, mais larabisation dun nom grec, qui est
Apa Kyros. Pour parvenir Ab Qr, il faut donc bien passer par Ab
Qr, rinterprt comme laccusatif dune flexion triptote Ab//. Cela
nempche pourtant pas le scripteur dutiliser le nominatif Ab Qr
une place rclamant le gnitif ! Le papyrus tant contemporain de la
conqute de lEgypte (celle-ci, commence en 18 de lHgire, sachve en
25 avec loccupation dfinitive dAlexandrie) et, plus prcisment de la
fondation de Fust t (22H), il parat difficile, pour ne pas dire impossi-
ble, de voir dans ce fait le signe dune volution dun type lautre, lie
au mtissage des populations dans les centres urbains nouvellement
crs la suite de la conqute ! Et si on a donc bien ici, en mme temps,
rfrence implicite au type ancien arabe et utilisation explicite du type
no-arabe, il vaut mieux y voir la continuation dune situation de langue
plurielle , non encore standardise. Un autre argument dans le mme
sens peut tre trouv dans le Coran mme, avec les incertitudes de la
flexion visible, casuelle ou modale (pour des exemples et une analyse,
cf. Larcher 2005a).
5. Conclusion
Le rcit traditionnel des origines de la grammaire arabe, ne de la cor-

ruption de la langue , a lapparence dun rcit historique. Les orienta-
listes du XIXe sicle, sicle dhistoire et desprit critique, sy sont laiss
prendre, mme si, en 1906, Karl Vollers jeta un fameux pav dans la
mare ! Cent ans aprs, les linguistes arabisants voient dans larabe dit
classique, comme dans toute autre langue classique, une construction.
Ils verront alors ce rcit comme une reconstruction, ayant pour seul but
la lgitimation de cette construction.
6. Rfrences
6.1 Sources primaires

al-Frb, Ab Nasr. 1969. Kitb al-h urf (Alfarabis Book of Letters), d. Muhsin Mahdi,
coll. Recherches publies sous la direction de lInstitut de Lettres Orientales de Bey-
routh n46, Premire srie : Pense arabe et musulmane. Beyrouth : Dar el-Machreq.
al-Jhiz, Bayn = Ab Utmn Amr b. Bahr al-Jhiz, Kitb al-bayn wa-t-tabyn, d.
Abd as-Salm Muhammad Hrn, 4 parties en 2 vols, Le Caire, 1367/1948.
Ibn Fris, Sh ib = Ab l-H usayn Ahmad Ibn Fris as-Sh ib f fiqh al-lua wa-sunan
al-arab f kalmih, d. Moustafa El-Choumi. Coll. Bibliotheca philologica arabica,
publie sous la direction de R. Blachre et J. Abdel-Nour, vol. 1. Beyrouth : A. Badran
& Co. 1383/1964.
Ibn Manzr, Lisn al-Arab = Muhammad b. Mukarram b. Al b. Ahmad al-Ansr
al-Ifrq al-Misr Jaml ad-Dn Ab l-Fadl Ibn Manzr. Lisn al-Arab al-muh t. Ed.
par Ysuf Xayyt , 4 vols. Beyrouth: Dr Lisn al-Arab. s.d.
Ibn Xaldn, Muqaddima = Wal d-Dn Abd ar-Rahmn b. Muhammad Ibn Xaldn.
al-Muqaddima, t. I du Kitb al-ibar. Beyrouth: Maktabat al-madrasa et Dr al-kitb
al-lubnn, 1967.
Suyt, Iqtirh = Jall ad-Dn Abd ar-Rahmn Ab Bakr as-Suyt , Kitb al-Iqtirh f
ilm usl an-nahw, H aydarbd, 1359 H. [reimp. Alep : Dr al-Marif, s.d.].
Suyt, Muzhir = Abd ar-Rahmn Jall ad-Dn as-Suyt al-Muzhir f ulm al-lua
wa-anwih, d. Muhammad Ahmad Jr al-Mawl, Al Muhammad al-Bajw et
Muhammad Ab l-Fadl Ibrhm, 2 vols. Le Caire : s l-Bb l-H alab. s.d.
az-Zajjj, Ab l-Qsim. 1973. al-dh f ilal an-nahw, d. Mzin Mubrak, 2me edi-
tion, Beyrouth, Dr an-Nafis.
. 1957. Al-Jumal. Prcis de grammaire arabe publi avec une introduction et un
index par Mohamed ben Cheneb, coll. Etudes arabes et islamiques, Premire srie :
Manuels. Paris : Klincksieck.
Zubayd, Tabaqt = Ab Bakr Muhammad b. al-H asan az-Zubayd al-Andals, Tabaqt
an-nahwiyyn wa-l-luawiyyn, d. M. Ab l-Fadl Ibrhm, coll. D axir al-Arab 50.
Le Caire, Dr al-Marif, 1984.
6.2 Sources secondaires

Blachre, Rgis. 195219641966. Histoire de la littrature arabe des origines la fin du
XV e sicle de J.C., I, II et III. Paris : Adrien-Maisonneuve.
Blanc, Ham. 1979. Diachronic and Synchronic Ordering in Medieval Arab Gramma-
tical Theory. in Studia Orientalia Memoriae D.H. Baneth Dedicata. Jerusalem: The
Magnes Press, The Hebrew University. 155180.
Blau, Joshua. 2002. A Handbook of Early Middle Arabic, The Max Schloessinger Memo-
rial Studies Monographs 6, Institute of Asian and African Studies, Faculty of Huma-
nities, The Max Schloessinger Memorial Foundation and The Hebrew University,
Jerusalem.
Diem, Werner. 1984. Philologisches zu den arabischen Aphrodito-Papyri. in: Der
Islam 61, 251275.
EI2 = Encyclopdie de lIslam, nouvelle dition, Leiden, Brill, 1960.
Ferrando, Ignacio. 2000. Le morphme de liaison /an/ en arabe andalou : Notes de
dialectologie compare , in : Oriente moderno, 80:1, 2546.
Ferguson, Charles A. 1959. Diglossia. Word, 15:2, 325340.
Fleischer, H. 1847. Ueber einen griechisch-arabischen Codex rescriptus der Leipziger
Universitts-Bibliothek. ZDMG, I, 148160 [repris dans Kleinere Schriften, 1885
1888 t. III, ch. XXII, 378388].
132 pierre larcher
Fleischer, Heinrich. 188588. Ueber arabische Lexicographie und Talibs Fikh al-
luah. Berichte ber die Verhandlungen der Kniglich Schs. Gesellschaft der Wissen-
schaften. Philol.-histor. Cl. 114 [repris dans Kleinere Schriften, 18851888, t. III, ch.
IX, 152166].
Fck, Johann. 1955 [1950]. Arabya. Recherches sur lhistoire de la langue et du style
arabe, Paris, Didier [tr. fr. de Arabya. Untersuchungen zur arabischen Sprach- und Stil-
geschichte, Abhandlungen der schsischen Akademie der Wissenschaften zu Leipzig,
Philologisch-historische Klasse. Band 45, Heft 1, Berlin, Akademie Verlag, 1950].
. 1955. Die arabischen Studien in Europa bis in den Anfang des 20. Jahrhunderts.
Leipzig: Otto Harrassowitz.
Goldziher, Ignaz. 1994. On the History of Grammar among the Arabs. An Essay in Lite-
rary History, translated and edited by K. Dvnyi et T. Ivnyi, Studies in the History of
the Language Sciences, 73, Amsterdam: Benjamins, 1994 [traduction anglaise de A
nyelvtudomny trtnete az araboknl, Nyelvtudomnyi Kzlemnyek 14, 307375,
1878].
Gruntfest, Y. 1991. From the History of Semitic Linguistics in Europe: an Early Theory
of Redundancy of Arabic Case-endings. in: K. Dvnyi et T. Ivnyi, eds. Proceedings
of the Colloquium on Arabic Grammar, The Arabist. Budapest Studies in Arabic 34.
Budapest. 195200.
Holes, Clive. 1995 [2004]. Modern Arabic. Structures, Functions, and Varieties. London:
Longman [Revised Edition, Washington D.C.: Georgetown University Press].
Kahle, Paul. 1947 [1959]. The Cairo Geniza, First Edition: 1947, Second Edition: 1959.
Oxford: Basil Blackwell.
Kazimirski, A. de Biberstein. 18467. Dictionnaire arabe franais, 2 vols. Paris : Tho-
phile Barrois.
Langhade, Jacques. 1994. Du Coran la philosophie. La langue arabe et la formation du
vocabulaire philosophique de Farabi. Prface de Jean Jolivet. Damas : IFEAD.
Larcher, Pierre. 2003. Diglossie arabisante et fush vs mmiyya arabes : essai dhistoire
parallle . Auroux, Sylvain et al. eds. History of Linguistics 1999. Selected Papers from
the Eighth International Conference on the History of the Language Sciences (ICHoLS
VIII), Fontenay-St.Cloud, France, coll. SIHoLS 99. Amsterdam: Benjamins. 4761.
. 2004. Thologie et philologie dans lislam mdival : relecture dun texte clbre
de Ibn Fris (Xe sicle) , dans Le discours sur la langue sous les rgimes autoritaires,
Cahiers de lILSL, n 17. Universit de Lausanne. 101114.
. 2005a. Arabe prislamique, arabe coranique, arabe classique : un continuum ? ,
dans Karl-Heinz Ohlig et Gerd-Rdiger Puin (Hrsg) : Die dunklen Anfnge. Neue
Forschungen zur Entstehung und frhen Geschichte des Islam. Berlin : Verlag Hans
Schiler. 248265.
. 2005b. DIbn Fris al-Farr ou un retour aux sources sur la lua al-fush , Asia-
tische Studien/Etudes asiatiques, LIX, 3, 797814.
. 2006a. Un texte dal-Frb sur la langue arabe rcrit ? Lutz Edzard and Janet
Watson (eds.). Grammar as a Window onto Arabic Humanism. A Collection of Articles
in honour of Michael G. Carter. Wiesbaden: Harrassowitz. 108129.
. 2006b. Sociolinguistique et histoire de larabe selon la Muqaddima dIbn Xaldn
(VIIIe/XIVe sicle) . Pier Giorgio Borbone, Alessandro Mengozzi e Mauro Tosco
(eds.), Loquentes Linguis Studi linguistici e orientali in onore di F.A. Pennacchietti /
Linguistic and Oriental Studies to Honour F.A. Pennacchietti / Lingvistika kaj orientaj
studoj honore al Fabrizio A. Pennacchietti. Wiesbaden : Harrassowitz. 425435.
Littmann, Enno. 1929. Die vorislamisch-arabische Inschrift aus Umm ij-Jiml , ZS,
VII, 197204.
. 1949. Arabic Inscriptions (Syria : Publications of the Princeton University Archaeo-
logical Expedition to Syria in 19045 and 1909. Division IV, Section D). Leiden: E. J.
Brill.
Marais, William. 1930. La diglossie arabe , in LEnseignement publicRevue pdago-
gique, tome 104 n12 (1930), 401409.
Owens, Jonathan. 1998. Case and proto-Arabic. Bulletin of the School of Oriental and
African Studies, I:61/1, 5173 et II:61/2, 215227.
Rabin, Cham. 1951. Ancient West-Arabian. London: Taylors Foreign Press.
Versteegh, C. H. M. 1977. Greek Elements in Arabic Linguistic Thinking. Leiden: E. J.
Brill.
Versteegh, Kees. 1981 [1983]. A dissenting grammarian: Qutr ub on declension. Histo-
riographia Linguistica 8:23, 403429 [repris dans Cornelis H.M. Versteegh, Konrad
Koerner and Hans-J. Niederehe, eds. The History of Linguistics in the Near East, Stu-
dies in the History of Linguistics, 28. Amsterdam: Benjamins. 1983.167193].
. 1984. Pidginization and Creolization: The Case of Arabic. Amsterdam Studies in
the Theory and History of Linguistic Science, Series IV-Current Issues in Linguistic
Theory. Amsterdam: Benjamins.
. 1995. The Explanation of Linguistic Causes, Az-Zajjjs Theory of Grammar. Intro-
duction, Translation, Commentary. Studies in the History of the Language Sciences
75. Amsterdam: John Benjamins.
. 1997a. Landmarks in Linguistic Thought III. The Arabic Linguistic Tradition. Coll.
History of Linguistic Thought. London: Routledge.
. 1997b [2001]. The Arabic Language. Edinburgh: University Press [2me d. 2001].
. 2004. Pidginization and Creolization revisited: The Case of Arabic. dans Mar-
tine Haak, Rudolf de Jong and Kees Versteegh, eds. Approaches to Arabic Dialects.
A Collection of Articles presented to Manfred Woidich on the Occasion of his Sixtieth
Birthday. Leiden: E. J. Brill. 343357.
Wetzstein, Johann Gottfried. 1868. Sprachliches aus den Zeltlagern der syrischen
Wste. Zeitschrift der Deutschen Morgenlndischen Gesellschaft 22. 69194.
7. Traduction du texte dal-Zajjj

Pourquoi cette sorte de science a t appele nahw
Si lon demande pour quelle raison cette sorte de science a t appele

nahw et sest vue attribuer ce nom, on rpondra : la raison en est lhis-
toire rapporte au sujet de Ab l-Aswad ad-Dual. Lorsquil entendit
parler les mtis dArabes Basra, il rprouva les fautes de langage quils
commettaient, du fait de leur contact avec la vie sdentaire et les enfants
des non-Arabes. Une fille lui dit un jour : Papa, m aaddu l-h arri ?
[quelle est la chaleur la plus intense ?] La canicule en plein midi,
ma petite fille, lui rpondit-il, ou quelque chose de ce genre, car il y a
divergence dans le rcit. Je ne tai pas demand cela, lui dit-elle, je me
suis seulement tonne de la chaleur intense . Alors dis, reprit-il, m
aadda l-h arra ! [quelle chaleur intense !] . Et dajouter : Nous apparte-
nons Dieu; la langue de nos enfants sest corrompue . Il songea faire
un ouvrage, o il rassemblerait les fondements de larabe, mais Ziyd len
empcha. Nous ne croyons pas, dit-il, que les gens se fient ce livre, ni
quils abandonnent la [bonne] langue et cessent de tirer la puret [lin-
guistique] de la bouche des Arabes . Et, ce, jusqu ce que les fautes de
langage se rpandent, deviennent nombreuses et affreuses. Alors, il lui
ordonna de faire ce quil lui avait interdit. Et Ab l-Aswad fit un livre
134 pierre larcher
contenant la syntaxe de larabe et dit aux gens : suivez cette voie, cest-
-dire allez dans ce sens ; nahw signifie direction et cest pourquoi la
grammaire a t appele nahw.
On dit quil fut le premier crire dans un ouvrage que le discours
est nom, verbe et particule, dote dune valeur smantique. Interrog
ce sujet, il dclara : je lai emprunt au commandeur des croyants Al b.
Ab Tlib (Dieu tende ses bndictions et son salut sur lui !) .
Il arrive quun nom, un qualificatif ou un surnom lemporte pour une
chose. Celle-ci est alors connue sous ce nom spcifiquement, lexclu-
sion de tout autre objet entrant dans la comprhension de ce nom. Le
fiqh, on le sait, est lintelligence des choses. On dit faqihtu le rcit aussi
bien que je [l] ai compris et un homme faqh ou faqih, cest--dire qui
comprend. Puis le fiqh est devenue la science religieuse, spcialement.
Et quand on dit un homme faqh, on vise seulement lhomme savant en
matire de Loi, mme si toute personne qui comprend une science et y
excelle est un faqh en cette science. Et, de mme, tibb est lhabilet. Cest
de l que lon dit un homme de tibb et tabb, sil est habile. Puis tabb est
devenu insprable de ceux qui sintressent la science des philosophes
ayant pour effet la conservation de la sant et, plus spcialement, per-
mettant de la recouvrer. Les exemples de ce genre de choses abondent.
SBAWAYHIS VIEW OF THE ZARF AS AN MIL
Aryeh Levin
The Hebrew University, Jerusalem
1. Introduction
1.1 The meaning of the term zarf

In Sbawayhis terminology the term zarf (plural zurf ), designates an
expression denoting place or time. The zarf is an accusative as in al-lay-
lata to-night,1 or a combination of an accusative + genitive as in maa-
hu with him,2 xalfa-ka behind you3 and yawma l-jumati Friday,4 or a
combination of a h arf jarr + genitive, as in f-h in it.5
1.2 The syntactic status of the zarf

In given syntactic constructions the zarf is an indispensable part of the
sentence, while in others it is a dispensable part.
When the zarf is an indispensable part it occurs as a predicate in
some types of the nominal sentence, as in the examples abdu llhi fh
(Sb. I:222, 18) and fh abdu llhi (Sb. I:222, 17) Abdallah is in it6
and abdu llahi fh qiman Abdallah is standing in it (Sb. I:222, 15).
A zarf also occurs as the indispensable predicate in sentences begin-
ning with inna and the kind of kna the later grammarians called kna
an-nqisa, as in the examples inna fh zaydan and inna zaydan fh
verily Zayd is in it (Sb. I:222, 20); inna zaydan lafh qiman verily
Zayd is standing in it. (Sb. I:242, 12); m kna fh ah adun xayrun
minka nobody better than you was in it (Sb. I:21, 7) and m kna
1
See Sb. I:176, 1720.
2
The form maa is conceived of by Sbawayhi and the other grammarians as a noun
taking the accusative (see Sb. I:177, 1415). When following the particle min ma takes
the genitive, as in the example min maihifrom him (ibid.).
3
See Sb. I:170, 1720.
4
See Sb. I:176, 1720.
5
See Sb. I:207, 2021.
6
See Sb. I:222, 1420. Cf. Sb. I:207, 2021.
136 aryeh levin
ah adun mitluka fh nobody like you was in it (ibid.). When the zarf
is the indispensable predicate of the sentence it is called mustaqarr,7 lit.,
a place where someone is. As a grammatical term mustaqarr desig-
nates an indispensable predicate of the nominal sentence, denoting the
place where the subject is.8 The term mustaqarr is sometimes restricted
by an expression denoting its grammatical quality as an indispensable
part of the sentence, as in mustaqarran taktaf bihi a predicate denoting
the place where the subject is, with which you content yourself [when
intending to express a complete sentence] (Sb I:21, 11), and mustaqar-
ran lizaydin yastan bihi s-suktu a predicate denoting the place where
Zayd is [occurring in the sentences inna fh zaydun and inna zaydan
fh], with which a complete sentence can be satisfied as its comple-
ment.9 (Sb. I:222, 2021).
When the zarf is not an indispensable part of the sentence it is called
ayr mustaqarr not a mustaqarr.10 It is said that the zarf in this case is
mulan or law a dispensable zarf which does not operate as an mil
(see below 4).
The form mustaqarr sometimes occurs in combinations referring to a
dispensable part of the sentence. These combinations include restrictive
expressions indicating that the whole combination refers to a dispens-
able part: in referring to the example fh abdu llhi qimun Abdal-
lah is standing in it (Sb. I:223, 2), where fh is a dispensable part of
the sentence, Sbawayhi says that fh here is a mustaqarr lil-qiym an
expression denoting the place where the act of standing [expressed in
the predicate qimun] takes place.11 Similarly, in referring to the exam-
ples fha abdu llahi qiman and abdu llhi fha qiman Abdallah is
standing in it (Sb. I:222,15), Sbawayhi says that qiman is a h lun
mustaqarrun fh [an expression denoting] a h l (= a state) where [the
subject abdu llhi] is. Note that the above combination refers to a part
of the sentence that is a h l and not a zarf.
7
See Sb, I:21, 717; 222, 1422.
8
This definition is inferred from Sb. I:222, 1422; See also Sb. I:21, 411; as-Srf
III, 11, 914; as-Srf according to Jahn, 1895, I/2, 73, note 16.
9
Lit. with which silence can be satisfied [after expressing a complete sentence].
10
See Sb. I:21, 1415.
11
See Sb. I:223, 19.
sbawayhis view of the zarf as an mil 137
2. The amal in sentences containing a zarf
2.1 A zarf cannot operate as the mil producing the nominative in the
subject
Sbawayhi believes that in nominal sentences a zarf cannot operate
as the mil producing the nominative in the subject (= al-mubtada),
irrespective of whether the zarf occurs as an indispensable predicate
(= mustaqarr) or as a dispensable part of the sentence. Hence, in a sen-
tence like fh abdu llhi qiman (Sb. I:222,15), the mil of the subject
abdu llhi is not the predicate fh, but the abstract mil called al-ibtid.12
In Sbawayhis view, the sense of al-ibtid is the act of putting the noun
in a position where it is unaffected by any word operating as an mil.13
The view that the zarf cannot be the mil of the subject derives from
the notion that in a nominal sentence, a word operating as an mil pro-
ducing the nominative must be logically identical with the noun affected
by it. For example: in the sentence abdu llhi axka Abdallah is your
brother (Sb. I:6, 11), the subject abdu llhi is logically identical with
the predicate axka, since Abdallah is your brother and your brother
is Abdallah.14 Hence the subject abdu llhi is the mil producing the
nominative in the predicate axka.15 In contrast, in fh abdu llahi
qiman, the predicate fh is not identical with the subject abdu llhi,
since it is an expression denoting the place where the subject is, and
hence fh cannot be the mil producing the nominative in the subject
abdu llhi.16 As a result, this sentence does not include any word that
can operate as an mil producing the nominative in the subject abdu
llhi. Hence, abdu llhi takes the nominative because of the effect of
al-ibtid.17
Sbawayhi contends that examples beginning with inna, such as inna
fha zaydan (Sb. I:222, 20), confirm that fh does not produce the
12
See Sb. I:222, 14223, 18.
13
See Levin, (forthcoming), mubtada, 4.1.
14
This notion is discussed in detail in Levin, 1979, 199202; Levin, 2002, 359,
15360, 11; Levin, 2006, 110111, 5; Levin, forthcoming, Cahiers linguistiques, 3.2.
15
For this notion see Sb. I:239, 59.
16
See Sb. I:222, 14223, 18.
17
See Sb. I:222, 1419.
138 aryeh levin
nominative in the subject, since in the above example inna is the mil
of the subject zaydan, as is shown by its accusative ending.18
Although Sbawayhi believes that the zarf cannot be the mil of
the mubtada, it is inferred that he holds that the zarf can produce the
accusative in nouns occurring as a h l, as in nominal sentences of the
type abdu llhi fh qiman and fh abdu llhi qiman Abdallah is
standing in it (Sb. I:222, 15).19 In his view, in these examples, the mil
producing the accusative in the h l qiman is the zarf fh.20 Sbawayhi
also explicitly says that the zarf is the mil producing the accusative in
words denoting measures of distance occurring as a tamyz 21 (see below
3).
2.2 Arguments confirming that in Sbawayhis view the zarf is the mil
of the hl
Sbawayhi does not explicitly say that the zarf is the mil producing the
accusative in the h l in certain constructions of the nominal sentence.
However, his view in this respect is inferred from some places in the
Kitb text.
2.2.1 In his discussion of sentences beginning with m kna, Sbawayhi

refers to two types:
(i) sentences where the zarf fh occurs as a mustaqarr, i.e., as the

indispensable predicate of the sentence, as in m kna fh ah adun
xayrun minka nobody better than you was in it (Sb. I:21, 7).
(ii) sentences where the zarf fh occurs as a dispensable part of the
sentence (mulan or law), as in m kna ah adun xayran minka
fh (Sb. I:21, 10).
In this discussion Sbawayhi explicitly says that a zarf which is a mus-

taqarr can operate as an mil.22 He also adds here that since a mustaqarr
18
Sb. I:222, 19223, 1.
19
Sbawayhi refers to the accusative in these examples both as a h l (Sb. I:223, 1) and
a xabar (Sb. I:222, 140). For this special use of xabar see Levin, 1979, 193196, 2.4.
20
See Sb. I:222, 49. See also Sb. I:167, 1116; Sb. I:218, 616, especially lines
1213. See below 2.22.2.3.2.
21
The later grammarians term tamyz does not occur in the Kitb.
22
See Sb. I:21, 719.
can operate as an mil, it is preferable to put it in a position where it

precedes the noun affected by it. The more the speaker makes it precede
this noun the better, he says. In contrast, he says that when the zarf is
not a mustaqarr, it is preferable to put it at the end of the sentence, or at
least in a position close to the end.23
2.2.2 In his above discussion, dealing with the possibility that a zarf
can operate as an mil, Sbawayhi does not give any example illustrating
this point. However, it is clear that his statements in this respect refer to
examples such as fh abdu llhi qiman and abdu llhi fh qiman,
discussed elsewhere in the Kitb.24 Since Sbawayhi explicitly says that
in these examples the mil of the mubtada abdu llhi is not the zarf,
but al-ibtid (see above 2.1), it is inferred that the mil producing
the accusative in qiman, which is a h l, is the zarf fh. It should be
stressed that the zarf fh is the only part of the sentence which can be
the mil producing the accusative in qiman, since the mubtada abdu
llahi is logically identical with qiman, and a noun which is logically
identical with another noun can produce in it only the nominative, as in
fh abdu llhi qimun (see above 2.1).
2.2.3 The above inference is supported by the following considerations:
2.2.3.1 In his discussion of the example laka -u tan bidirhamin

tan bidirhamin [The sale of] the sheep became binding on you25
[in a situation where their price is] one dirham a sheep, one dirham a
sheep (Sb. I:167, 13), Sbawayhi says that the accusative tan occur-
ring twice in this example is a h l. He explains how the accusative is
produced in tan as follows: wasra laka -u id nasabta bimanzilati
wajaba -u kam kna fh zaydun qiman bimanzilati staqarra zay-
dun qiman when you put [tan] in the accusative [the utterance]
laka -u is equivalent to wajaba -u, just as [the utterance] fh
zaydun qiman is equivalent to istaqarra zaydun qiman (Sb. I:167,
13).26 Since Sbawayhi believes that in all verbal sentences the verb is the
23
See Sb. I:21, 919.
24
See Sb. I:222, 14224,2.
25
The translation here is based on Lanes rendering of the expression wajaba l-bayu
(see Lane VIII, 2922A,1517). According to Sbawayhi laka -u is equivalent to
wajaba -u.
26
See Sb. I:167, 1116. Cf. Sb. I:222, 14223, 1.
140 aryeh levin
mil of the h l,27 it is clear that in his view, the mil of tan in wajaba
-u atan bidirhamin tan bidirhamin is the verb wajaba, while the
mil of qiman in istaqarra zaydun qiman is the verb istaqarra. Since
Sbawayhi holds that laka and fh are equivalent to wajaba and istaqarra
which operate as awmil, it is inferred that in his view laka and fh
also operate as awmil in the corresponding examples laka -u atan
bidirhamin tan bidirhamin and fh zaydun qiman respectively. For
similar examples where a combination of li + genitive or li + a relative
clause operates as an mil see Sb. I:223, 18224, 2.
2.2.3.2 In referring to the example fh zaydun qiman fh In it is

Zayd, in it (Sb. I:238, 15) Sbawayhi says: fainnam ntasaba qimun
bistigni zaydin bifh [The h l qiman] takes the accusative because
of the fact that when [the first] fha is added to zaydun, it makes the
sentence complete.28 The significance of this wording is that since the
first fh in the above utterance is the indispensable predicate making
the sentence complete, it is the mil producing the accusative in the h l
qiman in fh zaydun qiman fh.
3. The zarf as an mil of a tamyz denoting a measure of distance
Sbawayhi explicitly says that in the example dr xalfa drika farsaxan

the wandering territory of my tribe is situated in a distance of one
parasang from that of yours (Sb. I:176, 67), the mil producing the
accusative in farsax is the zarf xalfa drika, which is the indispensable
predicate of the sentence.29 He adds that this zarf produces the accusa-
tive in farsaxan and in other measures of distance, since it is not logi-
cally identical with farsaxan, and since farsaxan is not the sifa of xalfa
drika. He compares the effect of xalfa drika on farsaxan with that of
irna on dirhaman, in the example irna dirhaman.30
27
See Sb. I:15, 1822.
28
See Sb. I:238, 1418.
29
See Sb. I:176, 612. For xalfa as a zarf see Sb. I:177, 14.
30
See Sb. I:176, 611. For the amal in irna dirhaman see Carter, 1972. See also
Levin, (forthcoming), Cahiers linguistiques, 3.2.
4. The il of the zarf
According to Sbawayhi, one of the syntactic qualities of the zarf is al-

il, lit. the abolishment. It is inferred from the text of the Kitb that
this term, when referring to a zarf occurring in certain syntactic con-
structions, denotes the abolishment of the status of the zarf as an indis-
pensable predicate of the sentence. In this case al-il is opposed to
al-istiqrr the occurrence of the zarf as an indispensable predicate.31
In other syntactic constructions al-il denotes the abolishment of the
status of the zarf both as an indispensable predicate and as an mil. In
these constructions al-il is opposed both to al-istiqrr and al-iml,
i.e., the appliance of the amal of the zarf to the case ending of a certain
noun.32
The following examples illustrate the contrast between al-il on the
one hand, and al-istiqrr and al-iml on the other hand in certain syn-
tactic constructions:
(1) In the example m kna fh ah adun xayrun minka nobody better

than you was in it (Sb. I:21, 7), the zarf fh is the indispensable
predicate of the sentence33 (= al-mustaqarr). In contrast, in m kna
ah adun xayran minka fh (Sb. I:21, 10), the il of the zarf fh
takes place, since the indispensable predicate is xayran minka, and
hence the status of fh as the indispensable predicate is abolished.34
(2) In the examples abdu llahi fha qiman and fh abdu llhi qiman
(Sb. I:222, 15) the zrf fh is the indispensable predicate,35 and it
is also the mil of the h l qiman . In contrast, in fh abdu llhi
qimun (Sb. I:223,2) the il of fh takes place, since qimun is
the predicate, and fh loses its status as an indispensable predicate
and as an mil.36
31
See Sb. I:21, 1415.
32
The contrast between il and iml is inferred from Sb. I:21, 914, where
Sbawayhi uses the forms milan and yamalni in contrast to al-il and alayta. For
the occurrence of the term iml in the Kitb, also when referring to other awmil, see
Troupeau, 1976, 149, voc. iml.
33
The expression indispensable predicate here is based on Sbawayhis view of xabar
kna (see Levin, 1979, 203205, 2.6).
34
See Sb. I:21, 714.
35
See Sb. I:222, 14223,1.
36
See Sb. I:223, 12.
142 aryeh levin
Sbawayhi compares the il of the amal of the zarf with that of the
verbs later called afl al-qulb.37 He says that when the il of the zarf
takes place, it is preferable to pronounce it at the end or close to the end
of the sentence. In contrast, when the zarf is a mustaqarr occurring as an
mil, it is preferable to put it at the beginning of the sentence, like verbs
such as zunnu and ah sibu, which are pronounced at the beginning of
the sentence when they are awmil.38
Sbawayhi illustrates two types of taqdr construction of sentences
containing a zarf mulan:
(i) In referring to the sentence inna bika zaydan maxdun Zayd is
enchanted by you (Sb. I:242, 2), where the il of the zarf bika takes
place, Sbawayhi says that when the speaker expresses this sentence, it
is as if he were saying inna zaydan maxdun. Similarly, when saying
inna fka zaydan la ribun Zayd covets you (Sib. I:242, 5) it is as if the
speaker were saying inna zaydan ribun.39 These taqdr constructions
illustrate the notion that when the above sentences are pronounced it
is as if the zarf bika and fka are not spoken, and hence they cannot
operate as the mil producing the accusative in maxdun and ribun
respectively.
It appears that Sbawayhi holds this view in order to solve a theo-
retical difficulty arising from one of the main principles of the theory
of amal: in his view, the effect of an mil producing the nominative
or the accusative in the noun is always applied, irrespective of whether
this mil is an indispensable part of the sentence or not.40 This principle
seems to be violated if one assumes that when a zarf such as fh is an
indispensable predicate, as in fh zaydun qiman, it is the mil pro-
ducing the accusative in the h l qiman, but when fh is a dispensable
part, as in fh zaydun qimun, its amal is abolished. In order to solve
this difficulty Sbawayhi says that when the amal of the zarf is abol-
ished, the zarf does not occur in the taqdr construction. Since accord-
ing to the grammarians the relevant construction, as far as grammatical
analysis is concerned, is that of the taqdr, Sbawayhi assumes that fh
37
The il of the amal of this category of verbs is discussed in chapter 31 of the
Kitb (= Sb. I:49, 452, 15).
38
See Sb. I:21, 1013.
39
See Sb. I:242, 28. Sbawayhis words expressing this notion are very clear:
. . . kaannaka aradta inna zaydan ribun wainna zaydan maxdun walam tadkur fka
wal bika fauliyat hhun kam uliyat f l-ibtidi (Sb. I:242, 78).
40
See Sb. I:223, 613.
does not occur in the taqdr construction,41 and hence it is clear that its
amal cannot be applied.
(ii) Sbawayhi says that when the speaker expresses the sentence
fh abdu llhi qimun he intends it is as if he were saying abdu llhi
qimun fha. This taqdr construction illustrates the view that when
the zarf is not an indispensable predicate and hence is not an mil, the
speaker imagines that it is as if he were pronouncing the zarf at the end
of the sentence, since as regards grammatical theory it is preferable to
pronounce a zarf which is not an mil at the end of the sentence, or at
least in a position close to the end (see above 2.2.1). Sbawayhi says
that when the il of the zarf occurs in an example like fh abdu llhi
qimun the speaker imagines that it is as if he were saying abdu llhi
qimun fh, since in this taqdr construction the zarf which is not an
mil occurs at the end of the sentence.42
5. An interpretation of a difficult passage from the Kitb

(Sb. I:207, 1721)
Sbawayhis discussion in Sb. I:207, 1721 is one of the most difficult

passages in his text. It contains some points relevant to the topic of this
paper.
In referring to the example marartu birajulin maahu saqrun sidan
bihi adan I passed by a man with a hawk who was intending to hunt
with it tomorrow (Sb., Hrn II:52, 67),43 Sbawayhi says:
walam annaka id nasabta f hd l-babi faqulta marartu birajulin maahu
saqrun sidan bihi adan fal-nasbu al h lihi lianna hd laysa bibtidin
wal yubihu fh abdu llhi qimun adan lianna z-zurfa tul h att
yakna l-mutakallimu kaannahu lam yadkurh f hd l-mawdii, faid
sra l-ismu majrran aw milan fhi filun aw mubtadaun lam tulihi
liannahu laysa yarfauhu l-ibtidu, waf z-zurfi id qulta fha axawka
qimni yarfauhu l-ibtidu
41
For taqdr constructions which are shorter than their corresponding literal con-
structions see Levin, 1997, 146148, 3.3.
42
See Sb. I:222, 22223,6.
43
Derenbourgs edition has idan instead of sidan (see Sb. I:207, 1718). How-
ever, Hrns version sidan is supported by Sb. I:206, 8 in Derenbourg. The version
sidan also occurs in all the later grammarians treatises (see, for example, al-Fris, I:
250, 710; as-Srf, VI: 131, 7 (in a quotation from Sbawayhis text).
144 aryeh levin
know that if you put a noun in the accusative in [syntactic constructions

of] this type,44 and you say marartu birajulin maahu saqrun sidan bihi
adan, the accusative ending [occurring in sidan] remains unchanged,
because this [utterance]45 is not [an independent sentence] occurring at
the beginning of the utterance,46 and it does not resemble [the nominal
sentence] fh abdu llhi qimun adan (= Abdallah is standing in it
tomorrow) [where the form qim can take either the nominative or the
accusative],47 because [when expressing sentences such as fh abdu llhi
qimun adan ] the [amal] of the expressions denoting place (= zurf )
is abolished, [and the feeling of] the speaker is that [when expressing his
literal utterance he intends] it is as if he were not pronouncing the zarf
in this place at all.48 When the noun [rajulin] takes the genitive [in the
example marartu birajulin maahu saqrun sidan bihi adan], or when
it is affected by a verb [as in the example raaytu rajulan maahu saqrun
sidan bihi adan]49 or [when it is affected by the mubtada [in an example
such as hd rajulun maahu saqrun sidan bihi adan]50 you do not abol-
ish the amal [of the zarf maa which produces the accusative in sidan],
because the ibtid does not produce the nominative in the noun [rajulin
in the example marartu birajulin maahu saqrun sidan bihi adan], [and
hence this noun is not a mubtada which can be the mil producing the
nominative in the word sid]. [On the other hand, in sentences beginning
with] a zarf,51 when you say fh axawka qimni (= your two brothers
are in it), the ibtid produces the nominative in [the mubtada, which is
axawka]52 [so this mubtada is the mil producing the nominative in the
xabar, which is qimni (Sb. I:207, 1721).53
44
I.e., in syntactic constructions including a relative clause beginning with a zarf,
such as marartu birajulin maahu ksun maxtmun alayhi I passed by a man having
with him a sealed sack (Sb. I:207, 1516). Many examples of this type are discussed in
Chapter 112 of the Kitb (=Sb. I:206,5210,2).
45
I.e., the clause maahu saqrun sidan bihi adan.
46
The term al-ibtid here denotes a position occurring at the beginning of the utter-
ance. For al-ibtid in this sense, see, for example, Sb. II:295, 16; 296, 1115; 297, 36;
362, 1623.
47
See Sb. I:222, 14223, 2. See above 4.
48
I.e., when the speaker says fh abdu llhi qimun he intends it is as if he were say-
ing abdu llhi qimun (for this notion see as-Srf, VI:135, 18136, 5; see above 4).
49
This example does not occur in the Kitb. It has been introduced here according
to al-Friss interpretation (See al-Faris, I:251, 35) in order to explain Sbawayhis
intention.
50
This example does not occur in the Kitb. It is introduced according to al-Friss
interpretation (see al-Faris, I:251, 37), in order to explain Sbawayhis intention.
51
In Sbawayhis manner of expression a combination such as f az-zurf denotes the
sense of in sentences beginning with a zarf . Similarly, the expression f l-fil denotes
in sentences beginning with a verb (see Sb. I:17, 17); f darabain a sentence begin-
ning with daraba (see Sb. I:16, 1820); f knain a sentence beginning with kna
(Sb. I:17, 12).
52
For this notion see Sb. I:222, 14223, 18.
53
Ibid.
The following remarks and conclusions are inferred from the above pas-
sage. These conclusions are supported by other texts in the Kitb, dis-
cussed in this paper.
(1) A zarf can operate as an mil only when it is a mustaqarr, i.e.,
only when it is the indispensable predicate of a certain sentence or of a
certain clause.
(2) The zarf cannot be an mil producing the nominative in the sub-
ject or in the predicate. In a nominal sentence such as fh abdu llhi
qimun the mubtada abdu llhi takes the nominative because of the
amal of al-ibtid, and abdu llahi is the mil producing the nominative
in the predicate qimun.
(3) In a sentence such as marartu birajulin maahu saqrun sidan
bihi adan, the noun rajulin cannot be the mil producing the nomi-
native in sid, since rajulin is not affected by the ibtid, and hence it
is not a mubtada. In Sbawayhis view, in a nominal sentence, only a
mubtada can produce the nominative in a noun occurring as a predi-
cate. The noun affected by the mubtada must be logically identical with
it. Since the sentence marartu birajulin and the clause maahu saqrun
sidan bihi adan do not contain a mubtada logically identical with
sid, sid cannot take the nominative. The only word, which can be the
mil of sid in the above utterance is the zarf maa, which can produce
the accusative in sid. Hence it is impossible to abolish the amal of
maa, since if its amal were to be abolished the word sid would remain
without an mil.
(4) But in fh axawka qimni , the mubtada axawka takes the
nominative because of the amal of al-ibtid, and hence it can be the
mil of the predicate qimni. It is also possible to say fh axawka
qimayni in it are your two brothers standing. In this structure
axawka takes the nominative because of the amal of al-ibtid, and
fh, which is a mustaqarr, is the mil producing the accusative in the
h l qimayni.
(5) The form ibtid contained in the expression lianna hd
laysa bibtidin denotes an expression occurring at the beginning of
the utterance. It does not denote here any of the terms it designates
in Sbawayhis terminology of the nominal sentence. The words lianna
hda laysa bibtid express the notion that the clause maahu saqrun
sidan bihi is not an independent sentence occurring at the beginning
of the utterance.
(6) There is another argument which, according to as-Srf, pre-
vents the il of maahu in the clause maahu saqrun sidan bihi adan:
146 aryeh levin
maahu, as-Srf says, includes the antecedent -hu , referring to rajulin

in marartu birajulin maahu saqrun sidan bihi adan. If the il of
maahu were applied, it would have been dropped from the taqdr con-
struction, which in this case would be marartu birajulin saqrun sidun
bihi adan. This taqdr construction cannot exist, since the clause con-
tained in it does not include an antecedent referring to rajulin. Since
in the grammarians view the taqdr construction is the relevant one
as far as grammatical analysis is concerned, it is impossible to apply
here the il of maahu, since this would create an unacceptable taqdr
construction.
6. Conclusions
1 Syntactically, Sbawayhi distinguishes between two kinds of a zarf:

1.1 A zarf which is an indispensable predicate of a nominal sen-
tence, as fh in the example fh abdu llhi qiman In it is
Abdallah standing. This kind of zarf is called mustaqarr [A
zarf] denoting the place where the subject is.
1.2 A zarf which is a dispensable part of the sentence, as fh in the
example abdu llhi qimun fh Abdallah is standing in it.
This kind of zarf is sometimes called mulan or law or ayr
mustaqarr.
2 A zarf which is an indispensable predicate (= mustaqarr) is liable to
operate as the amil producing the accusative in a part of a sentence
occurring as a h l or a tamyz denoting a measure of distance. For
example:
(i) In fh abdu llhi qiman, fh is the mil of the h l
qiman.
(ii) In dr xalfa drika farsaxan the wandering territory of
my tribe is behind that of yours, at a distance of one paras-
ang, the zarf xalfa drika is the mil of the tamyz far-
saxan.
3 In contrast, a zarf which is not an indispensable part of the sentence
cannot operate as an mil. In Sbawayhis view, this zarf undergoes
the process of al-il, i.e., the process of the abolishment of its sta-
tus as an indispensable predicate and as an mil. Hence, this zarf
is called mulan or law [a zarf whose] status as an indispensable
predicate and as an mil is abolished.
4 Sbawayhi illustrates two types of taqdr construction of sentences

containing a zarf mulan:
4.1 In referring to examples like inna bika zaydan maxdun Zayd
is enchanted by you (Sb. I:242, 2), where the il of the zarf
bika occurs, Sbawayhi says that when the speaker expresses
this sentence, he intends it is as if he were saying abdu llhi
maxdun. This taqdr construction illustrates the notion that
when pronouncing the above sentence the speaker intends that
it is as if the zarf bika is not pronounced, and hence it cannot
operate as the mil producing the accusative in maxdun.
4.2 When the speaker expresses the sentence fh abdu llhi
qimun he intends it is as if he were saying abdu llhi qimun
fha. This taqdr construction illustrates the view that when the
zarf is not an indispensable predicate and hence is not an mil,
the speaker imagines that it is as if he were pronouncing the
zarf at the end of the sentence, since as regards grammatical
theory it is preferable to pronounce a zarf which is not an mil
at the end of the sentence, or at least in a position close to the
end.
7. References
7.1 Primary sources

al-Fris, Ab Al al-H asan b. Ahmad b. Abd al-affr. (d. 377/987). at-Talqa al
Kitb Sbawayhi. Iwad b. H amad al-Qz, ed. 1410 A.H. = 1990. Cairo.
Sbawayhi. (d. 177/793). Le livre de Sbawaihi. Trait de grammaire arabe. Hartwig
Derenbourg, ed. 18811889. Paris. 2 vols.
Ab Bir Amr b. Utm n b. Qanbar. (d. 177/793). al-Kitb. Kitb Sbawayhi. Abd
as-Salm Hrn, ed. Cairo, 1977. 5 vols.
as-Srf, Ab Sad. (d. 368/979). arh Kitb Sbawayhi. Ramadn Abd at-Tawwb and
others, eds. Cairo, 19882004. 6 vols.

Carter, Michael G. 1972. Twenty Dirhams in the Kitb of Sbawayhi, Bulletin of the
School of Oriental and African Studies. London 35, 485496.
Jahn, G. 1895. Sbawaihis Buch ber die Grammatik, bersetzt und erklrt von G. Jahn.
Vol. I, second pagination. Berlin.
Lane, E.W. 18631893. Arabic-English Lexicon. London (8 volumes).
Levin, Aryeh. 1979. Sbawayhis view of the Syntactical Structure of kna wa-axawtuh,
Jerusalem Studies in Arabic and Islam 1, 185211.
. 1997. The Theory of al-Taqdr and its Terminology. Jerusalem Studies in Arabic
and Islam 21, 142166.
148 aryeh levin
. 2002. An Interpretation of a Difficult Passage from the Kitb. Jerusalem Studies in

Arabic and Islam 27, 356362.
. 2006. An Interpretation of Two Difficult Passages from al-Kitb Referring to the
mil in Elliptical Sentences. Jerusalem Studies in Arabic and Islam 31, 107117.
. (forthcoming). Sbawayhis View of al-mubtada and al-ibtid. Moshe Bar-Ashers
Festschrift.
. (forthcoming). The mil of the xabar in Old Arabic Grammar. In: Cahiers lin-
guistiques de Inalco.
Troupeau, Grard. 1976. Lexique-Index du Kitb de Sbawayhi. Paris: Klincksieck.
PROBLEMS IN THE MEDIEVAL ARABIC THEORY OF
SENTENCE TYPES
Yishai Peled
Tel Aviv University
1. Introduction
In medieval Arab grammatical tradition, the two basic sentence types

jumla filiyya and jumla ismiyya, are normally defined by the first
occurring predicative constituent. A verb followed by its subject sig-
nals a jumla filiyya (verbal sentence), e.g. daraba abdu-llhi zaydan
(Abdullh hit Zayd). By contrast, a sentence introduced by a nomina-
tival noun is a jumla ismiyya (nominal sentence), e.g. zaydun rajulun
(Zayd is a man). This binary division corresponds with the grammar-
ians theory of amal (regimen), or, to be more specific, with two basic
types of amal. A jumla filiyya correlates with a verbal mil (opera-
tor), whereas a jumla ismiyya corresponds with ibtid, which is con-
sidered an abstract mil.1 The ibtid is normally said to consist of a
zero phonological mil and the predicatival relationship between the
mubtada and the xabar, respectively the subject and predicate in this
type of sentence. The basic principle of amal, stipulating that the mil
should precede the maml, applies in both sentence types. In a jumla
filiyya, the verbal mil affects the complements following it; in a jumla
ismiyya, the abstract mil, ibtid, occupies in principle a pre-mubtada
position, from where it assigns the raf case to the mubtada; the latter,
in turn, assigns raf to the xabar (according to Sbawayhis [Kitb I:239]
version). As for cases such as daraba zaydan abdu-llhi and rajulun zay-
dun, these were presented as cases of taqdm wa-taxr (preposing and
postposing), i.e. as the inverted versions of daraba abdu-llhi zaydan
and zaydun rajulun; in other words, as inverted jumla filiyya and jumla
ismiyya respectively.
1
For a detailed discussion of this correspondence, see Levin 1985.
150 yishai peled
However, the linkage between the concept of two sentence types on

the one hand, and the theory of amal on the other, turned out to consti-
tute a major problem with regard to the binary division into jumla filiyya
and jumla ismiyya. The grammarians realized that such types as daraba
abdu-llhi zaydan and zaydun rajulun, with their inverted versions, leave
various constructions that do not easily fit into any of the two categories.
The two apparently most problematic cases may be represented by the
two model sentences qimun zaydun (Standing is Zayd), where a par-
ticiple is followed by a definite noun phrase, and zaydun f d-dr, or f
d-dri (or fh) zaydun (Zayd is in the house/in it), where the predicate
position is occupied either by a definite prepositional phrase or by an
adverbial phrase such as hun, hunka etc. In the latter type, a definite
subject noun may either precede or follow the adverbial/prepositional
predicate; an indefinite subject must obligatorily follow its predicate.
As it were, sentences such as qimun zaydun and f d-dri (or fh)
zaydun may be considered as cases of an inverted jumla ismiyya, prag-
matically motivated. And modern linguists, uncommitted to the the-
ory of amal, would probably regard them as such. Yet for many of the
medieval grammarians, they represent, rather, a sentence type in its own
right. To be more precise, f d-dri zaydun is explicitly presented as such
by some of the grammarians; qimun zaydun, by contrast, is often dealt
with in a way that leads one to believe that it was considered by certain
grammarians as representing a third sentence type.2 I would argue that
the controversies that arose over these (and other) constructions point
to what may be viewed as gaps in the medieval theory of sentence types.
In other words, an attempt will be made to show that the theory, based
on a binary conception of jumla filiyya and jumla ismiyya, representing
two types of amal, was highly vulnerable and far from stable. This paper
concentrates on the theoretical problems presented by the above two
constructions. We start with qimun zaydun.
2
To be sure, there were grammarians who analyzed both constructions as an inverted
jumla ismiyya with a fronted xabar; others accepted more than one type of analysis. For
a discussion, cf. Ibn Ab r-Rab, Bast I:583ff.
the medieval arabic theory of sentence types 151
2. qimun zaydun
In his bb al-ibtid, Sbawayhi (Kitb I:239) discusses the option of

mubtada-xabar inversion. His starting point is that the standard pre-
ferred (al-h add) structure is for the mubtada to precede the xabar rather
than the reverse, much as the standard word order in the verbal sentence
is for the fil to precede the maf l. When dealing with inversion, he sin-
gles out qimun zaydun as a markedly complex case deserving special
attention. He quotes his teacher al-Xall as saying that qimun zaydun is
an ill-formed (qabh ) sentence unless analyzed as the inverted version of
zaydun qimun. As is well known, some of the later grammarians made
the point that such an inversion is quite problematic, since it places the
maml before the mil, and the Kfans saw a further problem in that it
makes the pronoun implicit in the participial form qimun precede its
antecedent (al-idmr qabla d-dikrsee, e.g. Ibn al-Anbr, Insf I:65).
But these problems were easily dismissed by the claim that qimun zay-
dun represents a secondary ( far) or surface (lafz) structure, whereas in
the basic structure (man, niyya, taqdr) zaydun, the zhir and mil
precedes qimun, the maml, with the implicit pronoun (the mudmar)
referring backas requiredto zaydun (for a detailed discussion, see,
e.g. Ibn al-Anbr, Insf I:6566, 68).
However, the real problem with qimun zaydun was associated with a
differentverbalanalysis of the participle, known to have been advo-
cated by some of Sbawayhis contemporaries. Citing al-Xall, Sbawayhi
points out that:
fa-id lam yurd hd l-man wa-ard an yajalhu filan ka-qawlihi
yaqmu zaydun wa-qma zaydun qabuh a li-annahu smun
If, however, they do not accept this analysis [= inversion], and want to
treat [qimun] as a verb, in analogy to such sentences as yaqmu zaydun
and qma zaydun, this should be rejected, because [qimun] is a noun
(Sbawayhi, Kitb I:239).
Yet Sbawayhi immediately makes it clear that under certain conditions
an active participle, while categorized as a noun, may implement a verbal
function (yajr majr l-fil). This could be accepted (h asuna indahum),
he maintains, if the participle functions as part of an asyndetic relative
clause (sifa) linked to some antecedent (mawsf), or, otherwise, governed
by a preceding operator such as a mubtada. In other words, qimun
zaydun is disallowed with a verbal analysis, much as dribun zaydan
(hitting Zayd) is unacceptable as a complete sentence. However,
152 yishai peled
qimun zaydun, as part of a larger sentence, is acceptable with a verbal

analysis, in analogy to such cases as an dribun zaydan (I am hit-
ting Zayd).3 As we shall see shortly, Sbawayhis approach to the verbal
analysis of qimun zaydun was later established as a firm principle in
medieval Arab grammatical thinking.
Ibn as-Sarrj (Usl I:5960), whose position regarding mubtada-
xabar inversion seems to be similar to that of Sbawayhis, readily accepts
muntaliqun zaydun as an inverted version of zaydun muntaliqun. As
for analyzing qimun zaydun as analogous to yaqmu zaydun where
qimun is not preceded by any supporting element (see below), in
other words, construing zaydun as fil to muntaliqunin principle, Ibn
as-Sarrj, much like Sbawayhi, regards such an analysis as misguided
(qabuh a), yet he admits it as jiz (acceptable). What both grammar-
ians seem to accept without reservation is that the noun following the
participle may be analyzed as a kind of fil provided that it is anchored
(yatamidu al, in Ibn as-Sarrjs words) to some preceding constitu-
ent. As an illustration of how this condition can be met, Ibn as-Sarrj
adduces such sentences as: marartu bi-rajulin qimin abhu (I passed
by a man whose father was standing), zaydun qimun abhu (Zayd
his father is standing), a-qimun abka (is your father standing?).
This rule would be developed by later grammarians into a general
principle of itimd, designed to specify the conditions under which a
non-verbal predicate may be analyzed as analogous to a verb preced-
ing its subject.4 This principle stipulates that a non-verbal predicatival
constituent, such as an active participle, or an adverbial/prepositional
phrase (see section 3 below) may exercise amal upon the constituent
following it (the subject) only if supported by (yatamidu al) some ele-
ment such as an interrogative particle, a relative pronoun, or, otherwise,
when the clause as a whole functions as an asyndetic relative clause or as
xabar to a preceding mubtada. In such cases, the first predicatival con-
stituent is perceived as behaving analogously to a verb. And as a verb-
like constituent it acts as a mil, assigning the raf case to the following
3
Sbawayhi (Kitb I:239) asserts that while the active participle and the verb may be
similar in some respects, one must appreciate the difference between them. Other gram-
marians (e.g. Ibn al-Anbr, Asrr, 70) pointed out that the active participle is weaker
than the verb, and cannot, therefore, exercise verbal amal, unless supported by some
preceding element (see below).
4
For the concept itimd as it is used in al-Xalls Kitb al-Ayn with reference to other
grammatical structures, see Talmon 1997, 210.
nominal constituent.5 For discussion, see, e.g. Ibn al-Anbr, Asrr: 70;
cf. Goldenberg 2002, 199201.
Ibn as-Sarrj is aware, however, of the implications of a verbal analy-
sis of qimun zaydun for the theory of amal. He argues (Usl I:60) that
in qimun abka (Your father is standing), qimun is assigned the
raf case by the ibtid, and abka is assigned the same case by the verb
preceding it. He indicates further that abka fills a xabar position.
In any event, both Sbawayhi and Ibn as-Sarrj reject the use of dribun
bakran amrun ( Amr hits Bakr) as an independent sentence, on the
ground that the active participle, while being analogous to the verb, is
by definition a nominal, and as such cannot be made to function identi-
cally to a verb in terms of case assignment. The above examples, where
the participle is linked to a preceding antecedent (mawsf), a mubtada,
or an interrogative particle, are viewed as analogous to the construc-
tion dribun bakran when anchored, under the principle of itimd, to
some external constituent (mah ml al ayrihi), such as a mubtada,
thus presenting a well-formed independent sentence (e.g. hd dribun
bakranthis [person] is hitting Bakr) (Ibn as-Sarrj, Usl I:60; and cf.
Sbawayhis position above; Levin 1985, 125126).
Like Sbawayhi and Ibn as-Sarrj, as-Zajjj (Jumal, 3738) was aware
of the theoretical problems raised by sentences consisting of an active
participle followed by a noun phrase. In particular he demonstrated the
implications for the grammatical agreement between the two constitu-
ents. To the extent that qimun in qimun zaydun is conceived of as
xabar muqaddam (a fronted xabar), it must be replaced by qimni or
qimna, once zaydun is substituted by a dual or a plural form respec-
tively. But under the alternative analysis cited by as-Zajjj, in which
qimun is assigned a verbal function, the active participle preceding
its subject should invariably take the singular form. In other words, the
proponents of this analysis would have qimun az-zaydni/az-zaydna
rather than qimni z-zaydni and qimna z-zaydna.6
5
Ibn Ab r-Rab (Bast I:585) remarks that some grammarians rejected the idea of
an adverbial/prepositional phrase assigning case. They argued that such phrases were
different in status (manzila) from the adjective. The latter, they argued, is capable of
inflection, and as such is more powerful than the adverbial/prepositional phrase. There-
fore, they concluded, the adverbial/prepositional phrase may not be analyzed as a case
assigner even where the principle of itimd is met. I return to this issue later.
6
See, e.g. Ibn Ab r-Rab (Bast I:584), who also indicates that the proponents of
akaln l-bart must, by extension, say qimni z-zaydni and qimna z-zaydna,
154 yishai peled
This rule, as stated by as-Zajjj, highlights the verbal function of the

participle under this particular analysis, relating qimun az-zaydni/az-
zaydna to qma z-zaydni/z-zaydna. It should be noted that the medi-
eval grammarians treated participles, as well as other types of adjective,
as complex forms incorporating a personal pronoun (for discussion, see
Goldenberg 2002, 195). Ibn Ya (arh I:8788) states clearly that parti-
ciples and other adjectives are derived from the verb, and that in virtue
of having a verbal meaning (man fil) they must have a fil.7 How-
ever, construing muntaliqun zaydun as modeled on yantaliqu zaydun
implies neutralizing the pronominal element in muntaliqun exactly as
it is done in yantaliqu.8 The argument is that the participle, much like a
regular fil, cannot assign the raf case twice (l yarfau filayni).9
Zajjj (Jumal, 38) indicates that in such cases the active participle
introducing the sentence is assigned raf by the ibtid, whereas the con-
stituent following it is assigned the same case by its verb (bi-filihi
apparently referring to the active participle; cf. Ibn as-Sarrjs analysis
above). Zaydun in qimun zaydun, it is argued, replaces the xabar (ya-
suddu masadd al-xabar); and the participle preceding it, he points out,
is invariably singular li-annahu qad jar majr l-fili l-muqaddami (for
it behaves analogously to a verb preceding [its subject]).10
Observe that, unlike the vast majority of grammarians, Zajjj did not
make the point of linking the verbal analysis of qimun zaydun to an
obligatory application of the principle of itimd. It should, indeed, be
noted that some of the later grammarians held a narrower version of
this principle, restricting the use of a participle in sentence-initial posi-
tion to cases where the participle is attached to a negative or interroga-
since in this version of the language the verb preceding the subject agrees with it in
number and gender (for a detailed discussion of akaln l-bart, see Levin 1989).
7
The grammarians, however, recognized that the personal pronoun incorporated
in an active participle cannot qualify as fil in the way an implicit personal pronoun
in a verb can. Thus, while allad daraba zaydun (The one who hit is Zayd) is a per-
fectly grammatical sentence, allad dribun zaydun is not, since, unlike allad daraba,
allad dribun cannot implement the function of a subject clause (see Jurjn, Muqtasid
I:463464).
8
This is, perhaps, why the Kfans, who rejected the analysis of qimun zaydun as an
inverted nominal sentence, could accept it as modeled on a verbal sentence: under the
verbal analysis the pronoun in qimun is disabled so there is no problem of cataphora
(cf. above).
9
For Ibn Ya (arh I:8788), then, a sentence such as zaydun qimun abhu
(Zayd, his father is standing) consists of a mubtada (zayd) and a xabar, the latter ana-
lyzed as a complex construction consisting of a fil (qimun) and a fil (abhu).
10
This type of analysis is attributed to Axfa; see, e.g. Ibn Usfr, arh I:341.
tive particle: m qimun az-zaydni (the two Zayds are not standing),
a-qimun az-zaydni (Are the two Zayds standing?). According to Ibn
Aql (d. 1367) (arh I:189), qimun, in each of the last two sentences,
functions as a mubtada, whereas az-zaydni is a fil sadda masadd al-
xabar (a fil substituting for the xabar). Indeed, this is the common
formula employed by those later grammarians who adopted the verbal
analysis of the construction in question (cf. Carter 1981, 189).
At this point one might ask how frequent in classical Arabic are such
sentences as qimun az-zaydni? I looked for this construction in the
Qurnic text, but no example of it was attested. In all the recorded cases,
a singular participle is followed by a singular noun phrase, or, otherwise,
a singular feminine participle by a plural (non-human) noun phrase.
It is interesting to note, however, that all cases display some kind of a
supporting element. In the vast majority of examples, the construc-
tion in question functions as predicate to a preceding subject realized
as either a referential nominal (typically, but not necessarily, in a sen-
tence introduced by inna or one of its sisters), or, otherwise, as a non-
referential damr a-an: wa-zann annahum mniatuhum h usnuhum
(They believed that their fortresses would protect themQ. 59:2),
wa-huwa muh arramun alaykum ixrjuhum (You are not allowed to
expel themQ. 2:85). Huwa in the latter example functions as damr
a-an. One example was noted where the supporting element is the
interrogative particle a-: a-ribun anta an lihat (do you loathe my
gods?Q. 19:46).
What is then the effect of the supporting element that makes [zay-
dun/rajulun] qimun abhu or a-/m qimun abhu an acceptable
verbal construction, as opposed to qimun abhu? Al-Xall (see above)
does not provide an elaborate answer. He argues, however, that a partic-
iple cannot easily replace a verb in pre-subject position, because it is an
ism. A verb and a noun, he maintains, may in certain positions imple-
ment similar functions, but they must still be differentiated. Nor did later
grammarians elaborate on the function of the supporting element. But
their discussion of the relevant cases might give us a clue. To phrase the
question differently, how does the supporting element impart further
verbal force to the adjectival predicate that enables it to act analogously
to a verb in such cases? If we compare the two constructions qma zay-
dun and qimun zaydun, we can see that the difference between the
two is that the finite verb, while devoid of a pronominal element, is still
inflected for person, whereas the participle is not. Lacking either a pro-
nominal element or inflection for person, the participle is excluded as a
156 yishai peled
pre-subject verbal predicate in an independent sentence. The function

of the supporting element zaydun in zaydun qimun abhu, is to make
up for the lack of person inflection in qimun, and thus empower the
latter to implement a verbal function.11
Further, as has been indicated, the supportive function may be
implemented not only by a noun but also by an interrogative or a nega-
tive particle. These are, in other words, further sources from which the
adjective could derive a verbal force. Already Sbawayhi attributed a
verbal effect to certain interrogative particles (Sbawayhi, Kitb I:39f.,
41f.the latter dealing specifically with a-). Ibn Ab r-Rab (Bast
II:712) points out that the interrogative particle a- requires a verb (al-
hamza tlibatun bi-l-fil), and that underlying (taqdr) a-zaydun daraba
amran is a-daraba zaydun amran (Did Zayd hit Amr?) (and cf. his
similar attitude to halII:679). Regarding the construction at issue he
maintains that underlying a-qimun zaydun is a-yaqmu qimun
zaydun, and that the former is derived from the latter by suppressing the
redundant yaqmu, for which qimun serves as an exponent (tafsr).
This explains why the interrogative a- qualified in the grammarians
view as a supporting element in sentences consisting of an adjectival
predicate followed by a subject.
However, the verbal analysis of such constructions as (a-)qimun
zaydun raises a difficult problem for the medieval theory of sentence
types. For if a sentence consists of a mubtada followed by a fil, how
should it be categorized in terms of sentence types? The fact that the
fil is presented as replacing the xabar does not make the issue any
simpler. For under the suggested analysis, a verb-acting constituent is
considered, in terms of irb, as a mubtada assigned the raf case by
the ibtid. The conception of the participle in such cases as mubtada is
quite understandable, given that it is a nominatival constituent in sen-
tence-initial position. If one declines the inversion analysis of sentences
such as qimun zaydun, how else can one account for the raf case of
qimun? The main problem with this analysis lies in its stipulating that
qimun as a mubtada assigns raf to a fil occupying a xabar position.
11
That, I believe, is what is intended by al-Xall (Sbawayhi, Kitb I:239) when he
refers to the participle in such cases as [kna] sifatan jar al mawsfin aw jar al
smin qad amila fhi [the participle] is an adjective agreeing with a head or [other-
wise] with a noun acting upon it, and further when he says: l yaknu maf lan f
dribin h att yaknu mah mlan al ayrihi fa-taqlu hd dribun zaydan . . . drib
cannot take an object unless it is linked to some other constituent, as for example in
hd dribun zaydanthis [person] is hitting Zayd.
Once we correlate qimun zaydun with qma zaydun, rather than with
zaydun qimun, with the consequence that qimun is to remain in
the singular irrespective of the number of the following noun (qimun az-
zaydni/az-zaydna), one can hardly see how sentences such as qimun
zaydun under a verbal analysis, let alone qimun az-zaydni, may be
viewed other than as cases of jumla filiyya. Indeed, when presenting the
fil, some of the grammarians, like Ibn Ya (arh I:74), indicate that
the position preceding the fil is available for a verb ( fil) or a nominal that
is analogous to a verb (abahuhu, m huwa f man l-fil min al-asm).
In this latter category they normally include the active and passive par-
ticiples, as well as such adjectives as h asan (sifa muabbaha bi-smi l-
filan adjective analogous to the active participle, e.g. Ibn Ya, arh
I:87). It is argued that in a sentence such as zaydun dribun ulmuhu12
(Zayd, his slave is hitting), dribun, much like yadribu, assigns raf to
ulmuhu. One may infer, then, that Ibn Ya would regard a sentence
such as qimun zaydun as a verbal sentence. Yet I have not recorded
any explicit reference to this type of sentence as a jumla filiyya. In any
case, the prevalent analysis of the construction under discussion was, as
already indicated, mubtada+fil sadda masadd al-xabar. It is not sur-
prising, however, that the grammarians adhering to this analysis did not
commit themselves to explicitly categorizing such sentences as either
jumla filiyya or jumla ismiyya.
A remarkable exception is Ibn Him al-Ansr (d. 1360) who, in his
famous book Mun l-Labb, provides an elaborate discussion of Arabic
sentence types. Ibn Hims classification will be discussed in detail in
section 5 below. As we shall see, he defined three sentence types (rather
than two!) by the kind of constituent introducing the sentence. Thus,
a sentence introduced by a nominal element is a jumla ismiyya. And
among his examples of jumla ismiyya we find the sentence qimun az-
zaydni. Ibn Him was, indeed, aware of the controversy surrounding
this sentence, indicating that it was accepted as a well-formed sentence
by Axfa and the Kfans. As we saw above, qimun az-zaydni was
adduced as an acceptable sentence in Arabic also by Zajjj, but the latter
did not classify it as jumla ismiyya.
12
Note, however, that in Zamaxars and Ibn Yas examples the construction at
issue is itself a xabar following a mubtada. As we have seen, this is consistent with the
principle of itimd.
158 yishai peled
In sum, then, the analysis of qimun zaydun into a mubtada+fil

replacing a xabar appears to reflect a twofold attempt:
(1) to make this kind of sentence conform to the principle of ibtid;

this is motivated by the fact that the first constituent is a nominal
exhibiting a raf case ending, and;
(2) to apply the principle of verbal tadiya, given the verbal properties
of the active participle.
But, to the extent that this analysis holds, does not it follow (at
least from the grammarians viewpoint) that qimun zaydun rep-
resents a sentence type in its own right? To my knowledge, no such
proposition has ever been advanced in medieval Arab grammatical
literature.13
3. fh / f d-dri zaydun
Only a small minority of the grammarians suggested that sentences

such as fh zaydun (In it there is Zayd) should be regarded as repre-
senting a sentence type in its own right. They designated this type jumla
zarfiyya, but differed on whether this term should or should not cover
also sentences such as zaydun fh/f d-dr (Zayd is in it / in the house).
I return to this later. At this point, let us examine the grammarians con-
ception of this construction, starting with Sbawayhi.
3.1 Sbawayhi
In his bb al-ibtid (chapter 132), Sbawayhi does not develop any dis-
cussion of this sentence sub-type. But elsewhere in the Kitb (I:170171;
cf. Levin 1987, 362, and Owens 1989, 224) he argues that in cases such as
huwa xalfaka (He is behind you) it is the subject huwa that assigns the
nasb case to xalfaka. Indeed, this is consistent with his argument (Kitb
13
Badawi (2000, 8f.) claims that the grammarians recognized three types of sentences
namely: filiyya, ismiyya and wasfiyya, introduced, respectively, by a verb, a noun and an
adjective (a participle or otherwise). He emphasizes the use of different terms for the
subject and predicate in each sentence type, indicating that in the jumlat wasf these are
referred to as mubtada and fil sadda masadd al-xabar. However, the term jumlat wasf
has not been attested in the grammarians writings studied for the present work.
I:239) that in zaydun muntaliqun, it is the subject that assigns raf to the
predicate.14 However, what about sentences such as fh zaydun, con-
sisting of an adverbial/prepositional phrase followed by a nominatival
noun phrase? Sbawayhi deals with these cases within the framework of
his discussion of sentences such as fh abdu-llhi qiman and abdu-
llhi fh qiman (Abdullh is in it, standingcf. Talmon 1993, 281).
He starts his discussion analyzing qiman as an accusatival xabar to
abdu-llhi. Then he goes on to indicate that abdu-llhi in these cases:
irtafaa bi-l-ibtidi li-anna llad dukira qablahu wa-badahu laysa bihi
wa-innam huwa mawdiun lahu wa-lkinnahu yajr majr l-ismi l-mab-
niyyi al m qablahu
is assigned the raf case by the ibtid since the constituent occurring
either before or after it [= the adverbial] is not it [= is not identical in refer-
ence with it], but rather signals its location. Yet [this adverbial] functions
analogously to a noun built upon the [subject] preceding it(Sbawayhi,
Kitb I:222).
Sbawayhi, as can be seen, points out that fh is non-coreferential with
abdu-llhi, but rather refers to Abdullhs location (mawdi). But it is
precisely this observation that underlies his endeavor to establish, first
of all, the acceptability of fh abdu-llhi/abdu-llhi fh as a complete
independent sentence. To this end he draws an analogy between fh
abdu-llhi and hd abdu-llhi (This is Abdullh), claiming that
in terms of completeness, the former, much like the latter, is a kalm
mustaqm (a correct sentence) that h asuna [badahu] s-suktu (that
may appropriately be followed by silenceSbawayhi, Kitb I:222 and
239240). Similarly, abdu-llhi fh is presented by him as analogous
to abdu-llhi axka (Abdullh is your brother): in both cases the
second constituent is built (mabn) upon the first. As for abdu-llhi,
Sbawayhi states clearly that, whether preceding or following the prepo-
sitional phrase, it is assigned the raf case by the ibtid.
But once a predicatival relationship is established between fh and
abdu-llhi, Sbawayhi reanalyzes the sentence assigning fh the func-
tion of xabar and abdu-llhi the function of mubtada. The position of
qiman is then demoted to that of h l (see Figure 1 below). He points
out, however, that since fh represents the persons location, fh abdu-
llhi is paraphrasable by istaqarra abdu-llhi. In other words, fh
14
Note, however, that the amal in huwa xalfaka is presented by Sbawayhi as analo-
gous to that in his model construction irna dirhaman (Twenty dirhams).
160 yishai peled
mubtada xabar
fh abdu-llhi qiman
xabar mubtada hl
Figure 1
behaves analogously to the verb istaqarra. As we shall see in 3.2, the verb
istaqarra, or otherwise the participle mustaqirrun, have since become
the grammarians most common device for explaining the grammatical
structure of sentences such as zaydun fh / fh zaydun.
A further indication of Sbawayhis consideration of fh as a verb-like
element is his statement (Kitb I:223) that qiman in the above sen-
tence may, alternatively, be replaced by qimun in the nominative. This,
he explains, is the result of abrogating (alayta) fh. In the medieval
grammarians writings, ilg is normally used as a technical term denot-
ing the annulling of amal. It is typically used with reference to potential
awmil, that is, elements that normally exercise amal upon other ele-
ments in the sentence (for discussion see, e.g. Peled 1992a, 150152).
One may infer, then, that Sbawayhi considered fh, in virtue of its act-
ing analogously to istaqarra, as an mil assigning nasb to qiman in
fh abdu-llhi qiman. As we shall see later, such adverbials as fh
were considered by some early grammarians as an mil assigning raf
to the following subject in such cases as fh zaydun. This view is typi-
cally attributed to the Kfans. Yet, Sbawayhi (Kitb I:223224) then
enters into an extensive discussion designed to exclude the possibility
that fh in sentences such as fh abdu-llhi qimun is the mil assign-
ing raf to abdu-llhi. He draws an analogy between this sentence and
bika abdu-llhi maxdun (Abdullh is fascinated by you). He argues
that an operator assigning case to an optional constituent (qimun in
the former sentence) has the same status (manzila) as an operator act-
ing upon an obligatory constituent (maxdun in the latter).15 Sbawayhi
emphasizes that in both cases (as well as in similar ones adduced by
him) the adjective is built upon the noun, thus establishing a predicati-
val relationship between the two. The prepositional phrase, by contrast,
is a law, i.e. a constituent that neither assigns nor receives amal. In fh
15
Indeed, in later grammatical writings, the model sentence bika zaydun maxdun
features regularly in the Basran arguments against the Kfan claim that in fh zaydun
it is fh that functions as the mil assigning raf to zaydun (cf. Ibn al-Anbr, Insf
I:5253).
abdu-llhi qimun, fh is only designed to specify the location where

the standing is taking place.
Let us now return to the original construction fh abdu-llhi to
which Sbawayhi devotes a separate bb, following his bb al-ibtid. In
chapter 133 of the Kitb, he deals with such cases as fh abdu-llhi,
tamma zaydun and ayna zaydun. Compared to his discussion above,
Sbawayhi here seems to be less specific about the rfi of zaydun:
wa-llad amila fm badahu h att rafaahu huwa llad amila fhi h na16
kna qablahu
The operator assigning raf to the following constituent is the same oper-
ator that assigned it the raf case when that constituent was before it [i.e.
before fh](Sbawayhi, Kitb I:239; cf. editors notes as well as the edi-
tions of Blq and Hrn).17
Sbawayhi thus asserts that the operator assigning raf to zaydun in
fh zaydun is the same one that assigns the raf case to zayd in zaydun
fh, where/when zayd occurs before it (i.e. before fh). Apparently,
he leaves it to the reader to conclude that it is the ibtid that functions
as mil in both cases. This could lead one to believe that for Sbawayhi,
fh zaydun represents an inverted version of zaydun fh. Yet Sbawayhi
does not make any explicit claim for taqdm wa-taxr in this particular
case. He presents fh zaydun as a case in which fh:
yaqau mawqia l-ismi l-mubtadai wa-yasuddu masaddahu
occupies the position of the mubtada and replaces it(Sbawayhi, Kitb
I:239; quoted also in Kouloughli 2002, 9).
As we saw in section 2, there were among later grammarians those who
analyzed qimun in qimun zaydun as a mubtada followed by a fil
(replacing a xabar). It seems more than likely that they were influenced
by the above Sbawayhian passage.
Regarding ayna zaydun, Sbawayhi points out that ayna is paraphras-
able by f ayyi makn (in which place?), and emphasizes that ayna,
as an interrogative, must obligatorily occur sentence-initially. In other
words, fh zaydun is distinct from ayna zaydun only in that in the latter
case the xabar occupies sentence-initial position obligatorily.
16
The words h aytu and h na alternate in this position in two different versions of
the text.
17
Talmon (1993, 283284) confronts the long version cited here with a shorter one
that to me looks rather obscure.
162 yishai peled
It is interesting to note that Sbawayhi does not include in his discus-

sion such cases as fh rajulun (In it there is a man), where the subject,
being indefinite, obligatorily follows the predicate (much like in ayna
zaydun). Indeed, as we will see later, this construction received little
attention from the grammarians, compared to fh zaydun. As we shall
see, while the grammarians never failed to point out that the predicate-
subject order in fh rajulun is obligatory, only a small minority of them
regarded this construction as representing a sentence type in its own
right. For the vast majority, fh rajulun, much like fh zaydun, repre-
sented an inverted jumla ismiyya. A detailed discussion of constructions
with an obligatorily fronted xabar is provided in section 4 below.
3.2 The istaqarra/mustaqirrun hypothesis

Following Sbawayhi, the medieval grammarians continued to address
themselves to the twofold problem posed by such sentences as zaydun
fh and fh zaydun. This problem, as we have seen, consisted in estab-
lishing a predicatival relationship between the nominal and the adverb-
ial/prepositional phrase, and accounting for the mil assigning case to
each. The suggested solution of positing an underlying linking element
such as istaqarra/mustaqirrun gave rise to extensive discussions that
developed into what may be referred to as the istaqarra/mustaqirrun
hypothesis. As we saw in 3.1, the origins of this theory can be easily
traced back to Sbawayhis Kitb.18
Throughout the centuries, the istaqarra/mustaqirrun hypothesis
has become the common strategy used by the medieval grammarians
in order to account for such sentences as zaydun f d-dr (or, for that
matter, zaydun xalfaka, where xalfaka alternates with min xalfikafor
discussion of the status of the zarf, see Levin 1987, 351357).19 As for
the grammatical status of the adverbial following the mubtada, here the
grammarians differed. For Sbawayhi (Kitb I:222), muntaliqun and fh
were equally admissible as xabar for zaydun. In Ibn as-Sarrjs (Usl
I:6263) view, the xabar in zaydun f d-dr is the underlying mustaqir-
18
As we have seen, however, Sbawayhi used this device in dealing with sentences
such as fh abdu-llhi qiman, to account for the nasb of qiman; he did not employ
it in cases such as fh zaydun or zaydun fh, pointing rather to the ibtid, in both, as
the mil assigning raf to zaydun.
19
For the Kfan theory of xilf (or muxlafa), see Astarbd, arh I:214; Ibn al-
Anbr, Insf I:245247; Muji, arh , 87, n. 216; and cf. Carter 1973).
run which, when coreferring with the mubtada, is optionally omitted

(h adf). The adverbial f d-dr fills the xabars slot, acting as a comple-
ment to the omitted genuine xabar (for a similar view, see Ibn Usfr,
arh I:347). Zaydun xalfaka/f d-dr is thus paraphrasable by zaydun
mustaqirrun xalfaka/f d-dr. Ibn as-Sarrj points out that the omitted
element normally conveys some meaning of existence, and is redundant
because it is retrievable from the adverbial. 20 He emphasizes that postu-
lating an underlying constituent in such cases is obligatory, because, in
itself, f d-dr (or xalfaka) does not qualify as a predicate; that is, it does
not predicate any quality of zayd (laysa bi-h adt),21 but only specifies the
location (mawdi) of Zayd.
Jurjn (Muqtasid I:274275) argues that an adverbial, with either
an explicit or implicit preposition, in principle presupposes a verb with
which it is linked to form a syntactical unit (cf. Ibn al-Anbr, Insf
I:246). Consequently, underlying f d-dr, in zaydun f d-dr, is the verbal
clause istaqarra f d-dr. This indeed represents the view of the major-
ity of grammarians for whom positing an underlying (muqaddar) finite
verb (istaqarra) is consistent with the clausal status of f d-dr in sen-
tences such as zaydun f d-dr. Obviously, once established as the (head
of the) xabar, istaqarra/yastaqirru is referred to as the mil, assigning
case to the following adverbial/prepositional phrase.
However, the assumption of an underlying finite verb in cases such
as zaydun f d-dr was not universally accepted. The controversy here is
linked to the fact that not all the grammarians regarded f d-dr in zaydun
f d-dr as having a clausal status. Some grammarians posited a parti-
ciple rather than a verb as the underlying element linking the adverbial/
prepositional phrase to the preceding mubtada. We have just seen that
Ibn as-Sarrj was one of the proponents of this hypothesis; indeed, he
20
Ibn Usfr (arh I:347348) emphasizes that using an adverbial/prepositional
phrase as a xabar substitute is only admissible when the deleted element is recoverable
from the surface constructionotherwise, the xabar should appear in full. Thus, for
example, zaydun f d-dr is only allowed if it is intended to convey the meaning mus-
taqirrun f d-dr, because f, signalling a receptacle (wi), is compatible in meaning
with istiqrr (staying). If, however, zaydun f d-dr is intended to convey the mean-
ing of dh ikun f d-dr ([Zayd] is laughing in the house), then the word dh ik must
occur; for, unlike the meaning of staying, that of laughing cannot be recovered from
the preposition f. Cf. Astarbd, arh I:215, for linking elements like h sil and kin
(be); Levin 1987, 360.
21
Ibn as-Sarrjs use of the term h adt in this case is significant, for it signalizes predi-
cate realized whether as fil or as xabar (cf. Goldenberg 1988, 4649).
164 yishai peled
viewed sentences such as zaydun xalfaka as displaying a single-phrased-

rather than a clausal xabar (Ibn as-Sarrj, Usl I:63). Astarbd (arh
I:215) cites Ibn as-Sarrj and Ibn Jinn as two grammarians who advo-
cated the participle rather than the verb hypothesis, on the ground that
the participle as a single phrase (mufrad) is compatible with the basic
structure of the xabar. Another proponent of the participle hypothesis
is Muji (arh , 87) who derives such sentences as zaydun ammaka
(Zayd is in front of you) and amrun min al-kirm ( Amr is one of the
honorable) from the underlying (taqdr) structures zaydun mustaqir-
run ammaka and amrun kinun min al-kirm respectively.22 Muji
makes it clear that for him an adverbial/prepositional phrase in xabar
position has the status of, and is therefore a substitute for, a participle
(not a clause). Postulating a personal pronoun implicit in the participle,
Muji argues further that this pronoun moves to, and resides in, the
adverbial/prepositional xabar occupying the position of the deleted par-
ticiple (in Mujis words: wa-afd d-damru llad kna f smi l-fil il
n-nib anhu fa-statara fhi).23 For further discussion of the istaqarra/
mustaqirrun hypothesis, see Jurjn, Muqtasid I:275ff; Ibn al-Anbr,
Insf I:245247; Ibn Usfr, arh I:344, 349351.
We can see that the istaqarra/mustaqirrun hypothesis is used by the
grammarians to fit the construction zaydun fh/fh zaydun into their
theory of amal, and, by implication, to their binary system of sentence
types. Once zaydun in both zaydun fh and fh zaydun was recog-
nized as a mubtada, both constructions could be said to represent a
jumla ismiyya, the latter being an inverted version of the former. The
xabar, when following the mubtada, is presented as either clausal or
phrasal, depending on whether one assumes yastaqirru or mustaqirrun
to be the underlying linking element. In both cases, this element is made
accountable for the case of f d-dr (see Kouloughli 2002, 1316, for
further discussion). However, the istaqarra/mustaqirrun hypothesis was
not universally accepted. And as we shall see, the alternative hypotheses
had substantial implications for the theory of sentence types.
22
Astarbd (arh I:215) claims that the underlying element is obligatorily deleted,
rejecting such sentences as zaydun kinun f d-dr. He indicates that Ibn Jinn did allow
such constructions, but points out that there is no evidence to support this position.
23
This is evidently Mujiis way of claiming a xabar status for the adverbial/prepo-
sitional phrase. Astarbd (arh I:216217) points to Fris and his followers as advo-
cating the same hypothesis. But Srf is cited by him as claiming that the pronoun is
deleted as part of the linking constituent.
3.3 Ab Al l-Fris
Ab Al l-Fris (d. 987) is considered to be one of the first grammar-
ians who advanced in an explicit way and developed the idea of sentence
types in Arabic (cf. Owens 1988, 3637). He defined each type, and
spelled out the problematic nature of the dichotomy verb+noun ( jumla
filiyya) versus noun+noun ( jumla ismiyya).24 Indeed, he was the first to
present a detailed argument with the conclusion that zaydun f d-dr is
neither a jumla filiyya nor a jumla ismiyya.
Having defined the two basic sentence types in Arabic, Fris turns
to concentrate upon the construction represented by zaydun f d-dr.
Indeed it looks as though Friss definition of the two basic sentence
types is meant as an introduction to his discussion of this particular
construction (Fris, Askariyya, 105109). He starts by indicating that,
although such sentences are composed of a nominal element (i.e. the
two nouns) and a particle (the preposition), they do not have the same
status as inna sentences, where the particle enters into a sentence made
up of two nouns. This is because f d-dr is non-coreferential with zay-
dun. And since zayd and f d-dr are not identical in reference, they
cannot be analyzed simply as subject and predicate. However, given that
zaydun f d-dr is definitely a well-formed sentence in Arabic, one must
assume some underlying (muqaddar) linking element to account for the
predicatival relationship between its two constituents. As we saw in 3.2,
this linking element must inevitably be either a noun or a verb (a par-
ticle does not bear any reference). To the extent that either of these can
be posited, a sentence such as zaydun f d-dr must eventually belong
either to the verb+noun or to the noun+noun type.25
24
Anxious to provide accurate and valid definitions, Fris (Askariyya, 104105)
points further to the option of a particle (h arf) entering into either of the two defined
jumlas, to form a kalm. What the reader is invited to infer is that the resulting con-
struction is an independent grammatical sentence whose basic type (i.e. filiyya or
ismiyya) is unaffected. He exemplifies this by sentences introduced by hal, inna, m,
qad and lam. (As a matter of fact, the same principle had already been stated by Ibn
as-Sarrj, Usl I:43.)
25
Fris (Askariyya, 109) draws a comparison between the case in question and
address (nid) expressions. He argues that y zaydu (O, Zayd!), much like f d-dri
zaydun, consists of nominal elements and a particle, and constitutes an independent
sentence. The difference between the two, he maintains, is that in the case of y zaydu a
verbal element should be assumed, which renders the address expression a sub-type of
a jumla filiyya, whereas in the case of zaydun f d-dr/f d-dri zaydun no such element
can be posited (see below).
166 yishai peled
However, from this point onward, Friss argument continues in a

direction designed to prove that this is not the case. In other words, a
sentence such as zaydun f d-dr, while conforming to the general prin-
ciple governing the production of well-formed sentences, constitutes an
exception in that it does not fall into any of the two basic categories,
jumla filiyya and jumla ismiyya. In Friss words:
a-l tar anna l-kalma wa-in kna l yaxl mimm dakarn f l-asli fa-
qad sra lahu l-na h ukmun yaxruju bihi an dlika l-asli
Notice that although a sentence must obey the basic principles indicated
by us, in this case there is an [overriding] rule leading the sentence away
from the basic principles (Fris, Askariyya, 105).
What Fris is now trying to prove is that neither a verb nor a noun can
be posited as a linking element establishing a predicatival relationship
between zaydun and f d-dr, a relationship modeled on that obtain-
ing between the subject and predicate of a regular verbal or nominal
sentence. And if it can be proved that neither a verb nor a noun can be
posited as an underlying linking element between zaydun and f d-dr,
the conclusion must be that sentences such as zaydun f d-dr represent
a sentence type in its own right.
He starts (Fris, Askariyya, 105) by adducing the sentence inna f
d-dri zaydan, were the particle inna enters into the sentence f d-dri
zaydun (his choice of this construction rather than zaydun f d-dr is
significant, as will be seen below.) Then he makes the following two
points:
(1) An underlying verbal link cannot be assumed, because a verb would

exclude the use of inna. In other words, while f d-dri zaydun may
be preceded by yastaqirru, inna and yastaqirru are mutually exclu-
sive: inna f d-dri zaydan is a perfectly grammatical sentence in
Arabic, but *inna yastaqirru f d-dri zaydan is disallowed.
(2) A linking noun cannot be posited either, because that would
amount to assumingfalselythat inna exercises its effect (amal)
upon zaydan across the underlying linking noun (Fris, Askariyya,
108).
Having disqualified both noun and verb as possible underlying linking

elements in cases such as zaydun f d-dr and f d-dri zaydun, Fris
(Askariyya, 108) argues further that in such cases the adverbial con-
stituent as such cannot be claimed to implement a verbal function. This,
he maintains, is borne out by the fact that the adverbial may not be pre-
ceded by a circumstantial phrase; a sentence such as *qiman f d-dri
zaydun is inadmissible, but it would be allowed if f d-dr had a verbal
value (a sentence such as qiman dah ika zaydun is considered as per-
fectly acceptable by the grammarians.)
All the above boils down to a rejection of the istaqarra/mustaqirrun
hypothesis, and that, in turn, leads Fris to the conclusion that sen-
tences such as zaydun f d-dr/f d-dri zaydun should be considered
neither as jumla filiyya nor as jumla ismiyya; they must be thought of,
rather, as representing a sentence type in its own right. Note, however,
that Fris did not assign the type of sentence under discussion any spe-
cial designation. The term jumla zarfiyya to which we will be introduced
below was coined in a later period.
But if one is supposed to assume no underlying element linking the
two predicatival constituents in zaydun f d-dr/f d-dri zaydun, what
is then the assigner of raf to zaydun in such cases? As we saw in 3.1,
Sbawayhi, who was not committed to any theory of sentence types, had
no problem presenting the ibtid as raf assigner to zayd in both zaydun
f d-dr and fh zaydun. But for Fris, making a similar claim would
imply classifying fh zaydun as a jumla ismiyya. Regarding the raf
assigner in zaydun f d-dr Fris does not develop any elaborate discus-
sion, apparently because in such cases one would automatically refer to
the ibtid as the raf assigner. However, when it comes to fh zaydun,
the construction on which he focuses his discussion, Fris presents a
clear position as to the rfi of zaydun. Having shown that neither a verb
nor a noun can be posited as a linking element, and having proved, fur-
ther, that the adverbial itself cannot be claimed to function as a verb,
Fris (Askariyya, 108109) refers the reader to Ab l-H asan [al-Axfa]
(d. 733), explaining that these are the reasons why Ab l-H asan regarded
the adverbial per se as the rfi when preceding a noun functioning as
muh addat anhu (of whom the message is predicated, subject). Notice
that it is not the term fil that is used with reference to that noun, but
rather muh addat anhu, a term that cuts across all sentence types. As
we shall see, however, later grammarians did not refrain from using the
term fil in this particular context.
Obviously, attributing the assigning of raf to an adverbial/preposi-
tional phrase constitutes a serious problem for the theory of amal. Since
the formulation of this theory, the grammarians always insisted that the
function of case assignment is implemented by either a verb or a par-
ticle. Various elements, notably active participles and other adjectives
168 yishai peled
were claimed to have a verbal force. But in our case, as we have seen,
Fris argued that f d-dr was not verbal enough to allow a circum-
stantial phrase to precede it. So one might ask what it is that qualifies
fh as raf assigner. To my knowledge, this point has never been clari-
fied by the grammarians. And it is no wonder that the concept of jumla
zarfiyya, where a predicatival prepositional phrase assigns raf to the
following subject, remained marginal and never became part of main-
stream medieval Arab grammatical thinking. Anyhow, for Fris, Axfas
position regarding the raf assigner in fh zaydun constituted further
support for his thesis that this construction represents a sentence type
in its own right.
4. Obligatory fronting of the xabar
4.1 Formal aspects

A remarkable feature of the grammarians (including Friss) discussion
of the adverbial/prepositional xabar is that they base their argument on
such sentences as zaydun f d-dr/f d-dri zaydun, where the definite
subject may either precede or follow the predicate (cf. Kouloughli 2002,
10, n. 7). But for the third sentence type advocated by Fris, sentences
such as ind mlun (I have money) would surely be a better example.
For in this case the subject follows the predicate obligatorily; reversing
the order is disallowed.
Most of the grammarians adduce such sentences by way of illustrating
an obligatorily fronted (taqdm) xabar. However, within the framework
of their grammatical discourse, obligatory fronting of the xabar presents
a major conceptual problem. A xabar, by definition, must follow, not
precede, the mubtada. The very concept of an obligatorily fronted xabar
appears to conflict with two fundamental principles in medieval Arab
grammatical theory : 1. The formal principle stipulating that the mil
precede the mamlsee, e.g. Ibn Ab r-Rab, Bast I:587); 2. The func-
tional principle placing the constituent representing the given informa-
tion before the one representing what is new for the addressee. To the
extent that the mubtada is identified with the given, and the xabar with
the new, such sentences pose a serious problem.
Obviously, in cases such as f d-dri rajulun, the grammarians could
not present rajulun f d-dr as the asl of f d-dri rajulun, since the for-
mer is disallowed as an independent sentence in Arabic. One may argue,
therefore, that in f d-dri rajulun, as well as in other cases of obliga-

tory fronting to be discussed below, the very concept of taqdm is not
normally intended in the sense that the above structure is the output of
reversing the order of some basic structure in which the indefinite raju-
lun precedes f d-dr. Rather, one must assume that taqdm in such cases
is used in the sense of placing in initial position, with no transformation
involved. Ibn Ya (arh I:86) argues that rajulun f d-dr is excluded
(1) because it could be wrongly interpreted as a noun phrase (with f d-
dr functioning as the attributesifaof rajulun) rather than as a com-
plete sentence (see also Ibn Usfr, arh I:343), and (2) in order to avoid
introducing a declarative (wjib) sentence by an indefinite noun.26
The idea of transformation in cases such as f d-dri rajulun was not,
however, universally excluded. Not surprisingly, it was suggested, albeit
in a rather idiosyncratic manner, by Ibn Jinn, a grammarian noted
for his originality and for frequently advancing dissenting arguments
incompatible with mainstream medieval Arab grammatical thinking.
In his Sirr sinat al-irb, Ibn Jinn (Sirr I:276) uses the concept asl
marfd (a rejected basic construction) in dealing with cases which
he regards as transformed constructions, but whose underlying struc-
ture (asl) is inadmissible. The principle of asl marfd is not explicitly
applied by him to rajulun f d-dr. But in his Xasis (I:300) he maintains
that, while mubtada-xabar is the basic word order of a jumla ismiyya,
a certain intervening factor (rid) might impose the reversal of that
order. The occurrence of an indefinite mubtada at the beginning of an
affirmative sentence constitutes in Ibn Jinns view such an rid, a kind
of contingency imposing the movement of the mubtada into the second
position in the sentence (cf. Peled 1992b, 105106).27 This is regarded
26
Astarbd (arh I:232) maintains that the problem of ambiguity between xabar
and sifa is acute, owing to the common occurrence of an adverbial in xabar position
in Arabic. He cites, however, one case where an adverbial xabar follows an indefinite
mubtada, pointing out that it is perfectly acceptable when the sentence is used as an
exclamation (du). Astarbd also remarks that fronting a non-adverbial xabar to an
indefinite mubtada does not eliminate the ambiguity. Thus, if you transform rajulun
qimun into qimun rajulun, rajulun could be analyzed as xabar of qimun or as an
apposition (badal) to it, whereas a fronted adverbial in similar cases is bound to be inter-
preted as xabar, due to its nasb case, whether explicitly marked (lafzan), or understood
by position (mah allan).
27
Note that an indefinite mubtada introducing a negative or interrogative sentence
is readily accepted by Ibn Jinn. Thus he admits (Xasis I:300) sentences such as hal
ulmun indaka (Is there a boy with you?) and m bistun tahtaka (There is no
carpet under you), claiming that they are communicatively useful, as opposed to sen-
tences such as rajulun indaka (A man is with you). The argument is that one can
170 yishai peled
by him as a corrective procedure (islh al-lafz):28 the second position,

he reminds us, is in principle the xabars position; and since the xabar is
essentially indefinite, the indefinite mubtada now fills the appropriate
slot as far as (in)definiteness is concerned. He emphasizes, however, that
in the underlying theoretical (muqaddar) level the mubtada precedes
the xabar (Ibn Jinn, Xasis I:318).29
There are other cases adduced by the grammarians as examples of
obligatory fronting of the xabar. The first example presented by Ibn
Ab r-Rab (Bast I:587) is the interrogative construction ayna zaydun
(Where is Zayd?). This grammarian strived to demonstrate that such
cases of obligatory xabar fronting do not violate the basic principle that
the mil must precede the maml. Regarding ayna zaydun he indi-
cated that the basic underlying (asl) structure in this case is a-zaydun
f d-dri am f s-sqi am f l-h nti (Is Zayd at home, in the market, or
in the shop?). The word ayna is an economy device designed to replace
both the adverbials and the interrogative particle a-, as well as the par-
ticle am. It implements a xabar function in virtue of its being a replace-
ment for the adverbials. At the same time, it is obligatorily fronted as a
substitute for the interrogative particle a-. Another semantic compo-
nent in ayna is that of specification (tayn), formally represented in the
basic underlying structure by the particle am. The transformation from
the basic to the final surface structure thus proceeds in the following
stages: First, the xabar ( f d-dri am f s-sqi etc.) is moved to a frontal
position immediately following a- (a-f d-dri am f s-sqi . . . zaydun
the interrogative a- is always positioned sentence-initially). Then ayna
is introduced to replace all the constituents preceding zaydun. It is thus
the adverbial semantic component in ayna that warrants its occurrence
in pre-mubtada xabar position, whereas the interrogative component
accounts for the obligatoriness of the movement. Ibn Ab r-Rab indi-
cates that the same applies to other interrogatives such as mat, kayfa,
negate the existence of, or pose a question with regard to, an unknown entity (mankr
l yuraf ), while there is no communicative value in predicating of an unknown entity
affirmatively.
28
Similarly, Astarbd (arh I:232) regards the obligatory fronting of the xabar in
cases such as f d-dri rajulun as a corrective (musah h ih ) procedure designed to handle
the indefiniteness of the mubtada.
29
Ibn Jinn (Xasis I:319320) then refers to sentences displaying an indefinite
mubtada in sentence-initial position. He argues, however, that these are not predicatival
sentences, in the sense that they are meant to express a wish or imprecation rather than
convey information. Another case is explained by him as paraphrasable by a negative
sentence (cf. n. 27 above).
man and m: kayfa axka (How is your brother?), man axka (Who
is your brother?), and the like.
Ibn Ab r-Rabs second case of obligatory xabar fronting is the con-
struction f d-dri rajulun. This, however, is dealt by him in pragmatic
rather than in purely formal terms, and will, therefore, be reviewed in the
following sub-section. The third case is exemplified by Ibn Ab r-Rab
(Bast I:588) by the sentence al t-tamrati mitluh zubdan (on the date
there is butter of an equal amount). He indicates that the reverse order
(mitluh zubdan al t-tamrati) is disallowed since -h in mitluh is an
anticipatory pronoun in both lafz (surface) and martaba (underlying)
structures, thus violating the rule of the anticipatory pronoun (al-idmr
qabla d-dikrcf. section 2 above). For further discussion of the relation-
ship between anaphora and the position of the xabar, see Astarbd,
arh I:232233.
Introducing his fourth and final case of obligatory xabar fronting, Ibn
Ab r-Rab (Bast I:588) cites the exceptive sentences m frisun ill
zaydun (No one is a horseman but Zayd) and m f d-dri ill amrun
(No one is in the house but Amr). In these two sentences the subject
nominal occurs sentence-finally and is dominated by the exceptive par-
ticle ill. Reversing the order in such cases, Ibn Ab r-Rab points out,
would not violate the rules of Arabic grammar, but result in a sentence
different in meaning from the original one. The sentence m frisun ill
zaydun assigns to Zayd, and only to him, the attribute of horsemanship.
This sentence, however, is neutral as to whether or not Zayd possesses
other qualities as well. But if the order of constituents is reversed so
as to make the subject zaydun precede the predicate, the resulting sen-
tence m zaydun ill frisun unmistakably excludes the possibility that
Zayd possesses any quality beside horsemanship. Similarly, the sentence
innam frisun zaydun is equivalent in meaning to m frisun ill zay-
dun, whereas innam zaydun frisun is synonymous with m zaydun
ill frisun, which explains why a mubtada-xabar order is inadmissible
in this related case as well.
Ibn Usfr (arh I:353) adds two more cases where the xabar is oblig-
atorily placed sentence-initially: 1. When the mubtada is a nominalized
clause introduced by anna: f ilm annaka qimun (It is known to me
that you are standing); 2. When the xabar is a kam al-xabariyya phrase:
kam dirhamin mluka (How many dirhams you have!).
The first of these two cases is dealt with also by Astarbd (arh
I:233234) who cites Fris as claiming that the adverbial/prepositional
phrase in such cases exercises amal (raf ) upon the following anna
172 yishai peled
clause with no supporting element (cf. 3.3 above; for itimd, see sec-
tion 2 above). Astarbd (arh I:233) explains that the reason for the
obligatory fronting of the xabar (whether adverbial or not) in such cases
is that if the anna clause were placed sentence-initially, the word nna
could be misread as inna rather than anna. For, between the two par-
ticles, it is the former rather than the latter that is associated with the
initial position in the sentence. Astarbd points out further that if the
xabar precedes the anna clause it is bound to be correctly analyzed as
xabar to the following clause as a whole rather than as a fronted constit-
uent governed by anna, because a constituent within the scope of inna/
anna cannot be preposed to either of these particles. Furthermore, once
the adverbial/prepositional phrase is established as the xabar of the fol-
lowing clause, then the particle heading that clause will be easily read as
anna, because a mubtada clause, being a noun clause, cannot be intro-
duced by inna. For further discussion of this issue, see Ibn Ya, arh
VIII:5960.
As we have just seen, f d-dri rajulun was only one item, and not nec-
essarily the first, on the list of constructions presented by the grammari-
ans as examples of obligatory fronting of the xabar. But it was apparently
the most difficult to deal with in purely syntactic terms. For one thing,
like the related f d-dri zaydun, it presented a challenge to the theory
of amal. For another, the indefiniteness of the mubtada could not, in
itself, constitute sufficient grounds for ruling out its occurrence in sen-
tence-initial position (cf. Astarbd, arh I:202207 for a detailed dis-
cussion of cases of an indefinite mubtada in sentence-initial position).
The grammarians main formal explanation for the obligatoriness of
predicate-subject order in this case was that an adverbial/prepositional
phrase following an indefinite nominal could be wrongly interpreted
as an attribute rather than a predicate. But as we have just indicated,
sentences with an indefinite mubtada do occur in Arabic. Indeed, as
we will see shortly, the strongest argument against *rajulun f d-dr was
pragmatic rather than syntactic.
4.2 f d-dri rajulunpragmatic aspects

When examining the construction f d-dri rajulun in terms of informa-
tion structure, most of the grammarians appreciated that sentences of
this kind represent a special case. Indeed they recognized that in these
cases it is the definite adverbial/prepositional phrase, occupying sen-
tence-initial position that represents the given information, whereas the
following nominatival indefinite phrase signals the new information,

and not the other way around (see, e.g. Ibn Ya, arh I:8687).30 Ibn
Ya (arh I:86) argues that, judging by the meaning of sentences such
as laka mlun (You have money), it is the definite complement of the
preposition (in this case the pronoun -ka) that represents the muh addat
anhu (what the sentence is about), even though, formally ( f l-lafz), it
is the nominatival noun that implements that function.31 In support of
his claim he indicates (1) that laka mlun is paraphrasable by anta d
mlin, and (2) that an indefinite nominal is inadmissible in the position
of the complement of the preposition: *li-rajulin mlun is disallowed
(lam yakun kalmanis not an [acceptable] sentence).
Another attempt to relate the construction in question to a mubtada-
xabar order was made by Ibn Ab r-Rab. This grammarian presents
a number of cases where the xabar is obligatorily fronted (cf. 4.1
above), taking great pains to demonstrate that fronting the xabar in
these cases is wholly justified, if not on purely formal, then on func-
tional/semantic, grounds. Regarding f d-dri rajulun, Ibn Ab r-Rab
states the following:
fa-hd yulzimu t-taqdma wa-l yajzu taxruhu fa-taqla rajulun f
d-dri li-annahu l yubtadau bi-n-nakirati wa-innam jza l-i-btidu
hun bi-n-nakirati li-anna l-maqsda l-ixbru an-i d-dri bi-annah
masknatun wa-laysati n-nakiratu l-maqsdata bi-l-ixbri wa-kna l-aslu
an taqla ad-dru mamratun bi-rajulin tumma ard l-i-xtisra fa-ql
f d-dri rajulun wa-alzam d-dra t-taqdma li-annah l-muxbaru anh
bi-l-h aqqati
in such cases [the xabar] is preposed obligatorily. It may not be post-
posed to yield rajulun f d-dr, because an indefinite noun may not fill a
mubtada position. In our case, the mubtada32 may be indefinite, because
the intention is to predicate of the house that it is inhabited, rather than
to predicate of the indefinite noun. Underlying [our sentence] is the sen-
tence ad-dru mamratun bi-rajulin. But for the sake of brevity, they say
30
According to Talmon (1993, 285287), the idea is already attested in ninth century
writings where the locative is typically referred to as sifa and the nominatival noun fol-
lowing it as xabar as-sifa.
31
This obviously rests on the assumption that in a sentence containing only one
nominatival noun, it is this noun that should be construed as the muh addat anhu. This
term, while referring literally to a pragmatic function, signals in the grammarians usage,
the subject, irrespective of sentence type; its counterpart h adt signals the predicate (see
Goldenberg 1988, 4649, for discussion).
32
Notice that yubtadau and ibtid are both construed in this case as used as
mubtada or implementing a mubtada function.
174 yishai peled
f d-dri rajulun. They obligatorily prepose the [phrase f] d-dr, because

it is really the house that is predicated of [i.e. the topic] (Ibn Ab r-Rab,
Bast I:587588).
In other words, the sentence f d-dri rajulun is paraphrasable by ad-dru
mamratun bi-rajulin (The house is inhabited by a man), where
ad-dru obviously functions as mubtada. Ibn Ab r-Rab explains that
the meaning intended by f d-dri rajulun is that the house is inhabited;
it is ad-dr of which something is predicated, not the indefinite rajulun.
The version f d-dri rajulun, Ibn Ab r-Rab maintains, is preferred to
ad-dru mamratun bi-rajulin for reasons of economy. The constituent
ad-dr is obligatorily preposed, for it is the one that implements the
function of muxbar anhu, which obviously entails that rajulun func-
tions as xabar. (For the terms muxbar/muxbar anhu, see Goldenberg
1988, 4651).
A different, indeed exceptional, view on this issue is held by Jurjn
(Muqtasid I:308309). Sentences such as ind mlun (I have money)
are dealt with by him within the framework of his discussion of sen-
tences with an indefinite mubtada. For Jurjn, the mubtada in this
particular case, albeit indefinite, implements the same pragmatic func-
tion as a definite sentence-initial mubtada. Mlun in the above sentence
implements the function of mubtada
li-ajli h usli l-ixtissi f l-xabari id kullu wh idin l yalamu anna indaka
mlan
because the xabar signals some specification [regarding the mubtada],
for it is not common knowledge that you have money (Jurjn Muqtasid
I:308).
For Jurjn, then, ind in its sentence-initial position makes the same
contribution to the communicative value (ifda) of the sentence as
would a xabar occupying a post-mubtada position. Much like other
grammarians, however, Jurjn argues that in such cases the xabar is
obligatorily fronted because mlun ind would be wrongly interpreted
as a noun phrase, with ind analyzed as a complement (sifa) to mlun
(see 4.1 above).
Most of the grammarians, however, did not follow this line of thought.
Rather, they discerned a discrepancy between the syntactic analysis of
sentences such as f d-dri rajulun into xabar and mubtada, and the
pragmatic functions of muxbar anhu and xabar implemented by f d-
dr and rajulun respectively. They did, however, emphasize that in terms
of definiteness sentences like laka mlun conform to the principle that

the definite constituent, representing the given information, should pre-
cede the indefinite noun signalling the new information. In any case,
the grammarians could not accept an analysis of f d-dri rajulun into
a mubtada followed by a xabar, on the ground that if a sentence of this
kind is introduced by inna, the noun phrase rajul automatically takes
the nasb case, which marks it unmistakably as subject. For discussion,
see Kouloughli (2002, 1617), and his references.
5. Ibn Him al-Ansrs tripartite division
5.1 Background
Returning now to the question of the raf assignment in cases such as
fh zaydun, most of the grammarians from Sbawayhi onwards regarded
ibtid as the operator assigning raf to zaydun. This, however, was by no
means universal, as we have seen. Indeed, Ibn al-Anbr (Insf I:5155)
attributes this position to the Basrans while presenting the other (much
less common) position as Kfan. We learn that the Kfans, as well as
the Basran grammarians Mubarrad and Axfa (cf. 3.3 above), regarded
the adverbial/prepositional phrase in the above case as the assigner of
raf to the following noun zaydun.33 Ibn al-Anbr maintains that both
the Basrans and the Kfans resort to Sbawayhi for support for their
respective claims. The Basrans obviously point to chapter 133 in the
Kitb, where, as we have seen, Sbawayhi refers to the ibtid (though
without using the actual term) as the mil assigning raf to zaydun. The
Kfans, for their part, draw upon a number of cases where, according to
Sbawayhi, an adverbial assigns raf to a following noun (Ibn al-Anbr,
Insf I:52). It is upon such cases, Ibn al-Anbr argues, that the Kfans
base their claim that in fh zaydun it is fh that should be regarded as
the rfi of zaydun.
Astarbd (arh I:218) indicates that the analysis of zaydun in f
d-dri zaydun as fil of the adverbial/prepositional phrase was advanced
by the Kfans, as well as by Axfa in one out of two statements he made
33
This position is clearly evidenced in Farrs Man l-Qurn (e.g. I:195196;
III:133). See Talmon (1993, 279) for further details and references.
176 yishai peled
on the subject.34 The argument in both cases is that the adverbial/prepo-

sitional phrase has a verbal force (man l-fil), analogously to qim in
qimun zaydun.
In Astarbds view, the Kfans position emanates from their cat-
egorical objection to xabar fronting, irrespective of whether the xabar
is a phrase or a clause. This objection, in its turn, is designed to fore-
stall the occurrence of an anticipatory pronoun. To this, however,
Astarbd offers an outright rejection, claiming, as could be expected,
that the anticipatory pronoun occurs only in the surface structure; in
the basic structure the mubtada precedes the xabar with no anticipa-
tory pronoun involved (cf. section 2 above).
As for Axfa, according to Astarbd he did not object to xabar
fronting, and (in his other statement) indeed regarded the ibtid as the
assigner of raf to zaydun in f d-dri zaydun. Axfas position here is
divergent from the one cited above. It rests upon two assumptions: 1.
The verbal force of the adverbial/prepositional phrase is weaker than
that of the adjective. 2. The acceptability of the construction f drihi
zaydun (in his house Zayd [is located]). In this case, the option of ana-
lyzing zaydun as fil is excluded on the ground that it would lead to an
unacceptable anticipatory pronoun (since the pronoun, under this anal-
ysis, would be cataphoric both in the lafz and man configurationscf.
Astarbd, arh I:202).
The debate, as could only be expected, concludes with the Basrans
having the upper hand. There is evidence to suggest, however, that, as in
many other cases, the position dismissed as Kfan represented a view
that was much more widely accepted than the mainstream medieval
grammarians would have us believe. As we shall see in the next sub-sec-
tion, it remained viable centuries later, in the writings of one of the most
prominent medieval grammarians, Ibn Him al-Ansr. The relevance
of this issue to our discussion is clear: it is closely related to the question
of whether or not sentences such as fh zaydun represent a sentence
type in its own right.
34
Ibn al-Anbr (Insf I:51) adds Mubarrad to the proponents of this kind of analy-
sis (and cf. Ibn Usfr, arh I:158159).
5.2 Ibn Hims categorization and definitions

Coming now to the 14th century grammarian Ibn Him al-Ansr
(d. 1360), we find both a critical discussion of the concept jumla as com-
pared with kalm, and an unambiguous division into three sentence
types. Recall that Friss starting point was that there were two basic
sentence types in Arabic. This was followed by an elaborate argument
designed to prove that zaydun f d-dr represented a sentence type in its
own right. For Ibn Him the tripartite division is an established linguis-
tic fact, and he makes it the starting point of his discussion. He does not,
however, ignore the problems raised by this division, as we shall see.
Ibn Him (Mun, 492) starts by defining the concepts kalm and
jumla. The latter is defined by him as a construction made up of either
fil + fil or, otherwise, mubtada + xabar. Then, however, he distin-
guishes three types of jumla, ismiyya, filiyya and zarfiyya, introduced
respectively by a noun, a verb and an adverbial (zarf) or a prepositional
phrase ([ jrr wa-] majrr).35 Being aware of the problem arising from
his defining each type by the initially occurring constituent, he remarks
(Mun, 492) that the definitions refer only to predicative constituents
(musnad and musnad ilayhi). Thus, the sentence a-zaydun axka (Is
Zayd your brother?) and the conditional clause in qma zaydun (If
Zayd stands up) are, respectively, ismiyya and filiyya, even though the
first noun in the former and the verb in the latter are each preceded by
a particle.
Once the problem of the particles is settled, Ibn Him appears to
be remarkably uncompromising in applying his principle, that Arabic
sentence types must be defined by the initial predicative constituent.
And this, indeed, leads to some conspicuous peculiarities. The jumla
ismiyya, for instance, is exemplified by him by the following three sen-
tences: zaydun qimun (Zayd is standing), hayhti l-aqq (How far
is the ravine!) and qimun az-zaydni (Standing are the two Zayds).
While the first of the three sentences is straightforward, the other two
are not. The word hayht in the second example is regarded by the
medieval grammarians as ism fil representing, as the term suggests, a
special word category whose members are considered as neither nouns
nor verbs (see, e.g. Astarbd, arh III:165ff. for details). Such asm
35
A fourth type, jumla artiyya, which, as he indicates, was proposed by Zamaxar,
is rejected by Ibn Him on the ground that the conditional clause should be categorized
as jumla filiyya.
178 yishai peled
al-af l, in constructions like hayhti l-aqq, are normally described as

occupying a verb position, with the implication that the following noun
implements the function of fil. However, since hayht as an ism fil is
viewed as a special kind of noun (rather than as a kind of verb), a sen-
tence introduced by it must, according to Ibn Hims rigid principle of
classification, be regarded as a jumla ismiyya rather than as jumla filiyya.
The special problems relating to the third case (qimun az-zaydni)
have already been discussed in section 2 above, and will not be repeated
here. It is, however, noteworthy that for Ibn Him the fact that qimun
az-zaydni is introduced by a participle (viewed by the grammarians as
a nominal element) was sufficient for classifying this sentence as jumla
ismiyya; the fact that qimun is followed by a noun in the dual form did
not require, in his eyes, any further argument or elaboration.
The jumla filiyya is illustrated by Ibn Him by six sentences intro-
duced by a verb. Out of these, five are straightforward: he uses qma,
yaqmu, qum, duriba and zanna to demonstrate that a jumla filiyya
may be introduced by any finite verb form. Specifically, by adducing
the sentence zanantuhu qiman (I believed him to be standing), Ibn
Him makes the point that a cognitive verb may, like any other verb,
introduce a jumla filiyya. As is well known, in medieval Arab gram-
matical theory, from Sbawayhi onwards, cognitive verbs such as zanna
(zanna wa-axawtuhzanna and sisters) are presented as analogous
to inna and kna (and their respective sisters), in that they enter into
(yadxulna al) sentences composed of a mubtada and xabar, nullifying
in the process the abstract operator ibtid, and assuming in its stead the
function of a formal mil assigning case to both nominal constituents in
the sentence (see, e.g. Sbawayhi, Kitb I:6; Ibn Ya, arh VII:7778).
In any case, the status of zanna as a verb was never disputed. By
contrast, the verbal status of kna was a matter of controversy among
the grammarians. The vast majority of grammarians considered kna
as a semantically deficient (nqis) verb,36 in the sense that it lacks the
semantic component of action.37 As such, its only function is to signal
the time of the nominal sentence into which it enters. Significantly,
36
Unless signalling existence, in which case it is labelled kna at-tmma complete
kna, and treated as an ordinary verb.
37
This had been recognized already by Sibawayhi, although in his account (Si-
bawayhi, Kitb I:16) he depicted kna as analogous to daraba in terms of transitivity
(tadiya): like daraba it takes two nominal complements, one in the nominative, the
other in the accusative.
while acknowledging the predicative relationship between zaydan and

qiman in zanantu zaydan qiman, the grammarians analyze the two
nominal constituents as direct objects (maf l) of zanna. By contrast,
in kna zaydun qiman, zaydun and qiman are normally analyzed in
terms of a jumla ismiyya, namely as mubtada and xabar, or otherwise,
as ism kna and xabar knarespectively.
True to his rigid definitions of the three sentence types, Ibn Him
includes in his examples of jumla filiyya the sentence kna zaydun
qiman (Zayd was standing). And this already seems to run counter
to the mainstream conception of jumla filiyya. Indeed, the grammar-
ians do not normally bring kna into their discussions of sentence types.
But given their special treatment of this verb, it is highly unlikely that
they would have classified kna zaydun qiman as jumla filiyya just by
the verbal morphology of kna.
But the most interesting for our present discussion is Ibn Hims
jumla zarfiyya. This he illustrates by the two sentences a-indaka zaydun
(Is Zayd with you?) and a-f d-dri zaydun (Is Zayd in the house?),
where the first predicatival constituent is an adverbial and a preposi-
tional phrase, respectively. Ibn Him points out that sentences such as
these can qualify as jumla zarfiyya only
id qaddarta zaydan filan bi-z-zarfi wa-l-jrri wa-l-majrri l bi-l-
fi wa-l mubtadaan muxbaran anhu bihim
istiqrri l-mah d
if you assume zayd to be a fil [acted upon] by the adverbial/preposi-
tional phrase, not by a deleted [verb/participle conveying the meaning of]
istiqrr, and [only if] you do not analyze zayd as a mubtada for which
the adverbial/prepositional phrase serves as xabar (Ibn Him al-Ansr,
Mun, 492).
The significance of this passage lies in that it seems to suggest that Ibn
Hims sentence-type definitions were not as rigid as they appeared
to be when we looked at his definitions and illustrations of the jumla
ismiyya and the jumla filiyya. It now turns out that for him, the predica-
tive constituent that comes first in the sentence is not in itself the only
criterion for determining the type of sentence. Rather, for a predicati-
val constituent to qualify as sentence-type identifier it must act as mil
upon the second predicatival constituent.
Another important point to note is that in both of Ibn Hims exam-
ples the adverbial/prepositional phrase is preceded by the interrogative
particle a-. This may be taken to suggest that by the time of Ibn Him
the principle of itimd, which, as we have seen (section 2 above), can be
180 yishai peled
traced back to Sbawayhi, had already been firmly established in medi-

eval Arab grammatical thought. When Ibn Him states that a sentence
can only qualify as a jumla zarfiyya if the second constituent functions
as fil to the first, illustrating this with examples displaying the inter-
rogative particle a- preceding the first predicative constituent, one is
bound to conclude that for him a jumla zarfiyya is a sentence whose first
predicatival constituent is an adverbial/prepositional phrase, where that
phrase acts as a verb, thus assigning raf to the following constituent on
the strength of the principle of itimd.
Ibn Him then goes on to make a critical remark directed at
Zamaxar. He indicates that Zamaxar exemplified jumla zarfiyya by
the phrase f d-dr in zaydun f d-dr (cf. Ibn Ya, arh I:88). This posi-
tion, he argues, is based:
al anna l-istiqrra l-muqaddara filun l ismun wa-al annahu h udifa
wah dahu wa-ntaqala d-damru il z-zarfi bada an amila fhi
on [the assumption] that [the underlying word conveying] istiqrr is a
verb, not a noun, and that that verb was deleted alone while the pronoun
implicit in it moved to the adverbial phrase, after [the verb] had exercised
amal upon the adverbial (Ibn Him al-Ansr, Mun, 492).
Recall that zaydun f d-dr was described by Fris as representing a
sentence type in its own right on the ground that, in his view, neither a
verb nor a noun could be posited as a linking element between zayd and
f d-dr. Here Zamaxar is quoted as elevating f d-dr to the status of
a clausal xabar (jumla zarfiyya). This is done, so the argument goes, by
positing an underlying verb that is deleted while the pronoun implicit
in it is transferred to the adverbial, the verb having exercised amal upon
that adverbial. In other words, the clausal status of f d-dr stems from
the pronoun it receives from the deleted verb istaqarra. What we see
here is, indeed, another attempt to account for the predicatival relation-
ship between zaydun and f d-dr, two non-coreferential elements, as
well as for the irb of the xabar. However, this attempt is based on the
istaqarra hypothesis (3.2 above), and that is precisely the reason why it
is rejected by Ibn Him. Zamaxars analysis is incompatible with Ibn
Hims conception of jumla zarfiyya. For Ibn Him, once an underly-
ing verb is assumed, the clause should be regarded as a jumla filiyya;
exactly as, when one posits a xabar-mubtada relationship between f
d-dr and zayd, the sentence must be considered as jumla ismiyya. This
will be further clarified below.
In 3.3 we pointed to Friss reference to Axfa, who had attributed
to the adverbial/prepositional phrase the function of operator assigning
raf to the nominal following it. He did not, however, refer to that nomi-
nal explicitly as fil. The term muh addat anhu which he used, signals
in medieval Arabic grammatical literature the subject of the sentence,
whether a fil or a mubtada. Indeed, the specific grammatical status of
the nominatival constituent, as determined by the mil assigning it the
raf case, never ceased to be a topic of debate among the grammarians.
Yet one thing emerges quite clearly. The analysis of zaydun as fil in
both f d-dri zaydun and qimun zaydun, normally attributed to the
Kfans and Axfa (see section 2 above), always comes up when these
two constructions are discussed. It was never abandoned. However, of
these two constructions, it is only fh/f d-dri zaydun that was con-
sidered, albeit by a small number of grammarians, as a sentence type in
its own right. The reason for this should by now be clear. The opening
predicative constituent in each of the three sentence types was regarded
as an operator (mil) assigning case to the following constituent(s): sen-
tence types were unmistakably correlated with amal types. And since
the participle (qimun) could not be viewed as other than a verbal or
a nominal element, it could not be regarded as introducing a sentence
type in its own right. By the same token, a sentence such as zaydun f d-
dr, introduced as it is by a nominatival noun, could only be defined as
jumla ismiyya. The concept jumla zarfiyya was by and large associated
with cases where a zarf could be claimed to be a mil assigning raf to
the nominal following it. As we will see in the next sub-section, it was
Ibn Him, an eminent proponent of the tripartite division, who also
appreciated and spelled out the problems arising from the actual notion
of sentence types in Arabic, whether two or three.
5.3 Problems
As we have seen throughout, the problems the grammarians encoun-
tered in categorizing Arabic sentences stemmed from the fact that their
conception of sentence types was deeply embedded in the theory of
amal. This is manifested also in the way these problems are illustrated
by Ibn Him (Mun, 493497). He offers an illuminating discussion
of ten cases where a sentence can be construed as either a jumla filiyya
or a jumla ismiyya, or, otherwise, raise a controversy among grammar-
ians as to the right categorization. Significantly, no case is cited as an
unambiguous jumla zarfiyya. Since the basic arguments recur through-
out his discussion, I will review only four of his examples that, I believe,
well illustrate the problematic aspects of the traditional categorization
of sentence types.
182 yishai peled
Let us start with Ibn Hims fourh example md sanata (What have
you done?). He points out that this sentence may be paraphrased as
either m llad sanatahu, or as ayya ayin sanata. Since llad sanatahu
is a nominalized constituent, the sentence, according to the first para-
phrase, must be categorized as a jumla ismiyya. Ibn Him indicates that
the first constituent m is analyzed as a fronted xabar by Axfa, and
as mubtada by Sbawayhi. By contrast, the proponents of the second
paraphrase, ayya ayin sanata, would categorize the same sentence as a
jumla filiyya, analyzing ayya ayin as a fronted direct object. (And see,
further, Ibn Hims discussion of the sentence md sanatahu.)
Ibn Hims sixth example reads qm axawka (Your two brothers
stood up). This sentence is presented by him as acceptable, subject to
specific types of analysis. (To what extent this construction was in actual
use in medieval Arabic is immaterial for the present discussion). First,
the sentence could be categorized as jumla filiyya if (1) the ending - in
qm is interpreted as a dual-marking particle (h arf tatniya), much as
the -t in qmat Hindun is analyzed as a feminine marker (and not as a
pronoun); or alternatively if (2) the ending - is interpreted nominally
and the following axawka is analyzed as apposition (badal) to it. Sec-
ond, qm axawka may be categorized as jumla ismiyya with a fronted
xabar (with the ending - interpreted nominally and axawka analyzed
as a postposed mubtada). Note that Ibn Him does not mention the
possibility of analyzing qm axawka as a jumla filiyya with axawka
functioning as fil to qm (luat akaln l-bart see above, section
2, n. 6).
The seventh example presented by Ibn Him is nima r-rajulu zaydun
(What a nice man is Zayd). This sentence, he explains, may be viewed
as an inverted jumla ismiyya, with nima r-rajulu functioning as a pre-
posed xabar to zayd. Under an alternative analysis, however, zaydun
could function as xabar to a deleted mubtada. Ibn Him argues that
under the latter analysis, nima r-rajulu zaydun consists of two asyndeti-
cally coordinated clauses, the first one (nima r-rajulu) verbal, and the
second nominal.
But perhaps the most interesting is Ibn Hims second example,
where he makes the following statement regarding a-f d-dri zaydun
and a-indaka amrun:
fa-inn in qaddarn l-marfa mubtadaan aw marfan bi-mubtadain
fin taqdruhu kinun aw mustaqirrun fa-l-jumlatu ismiyyatun dtu
mah d
xabarin f l-l wa-dtu filin munin an-i l-xabari f t-tniyati wa-in
qaddarnhu filan bi-staqarra fa-filiyyatun aw bi-z-zarfi fa-zarfiyyatun
if we analyze the nominatival constituent as mubtada, or otherwise, as

a nominal assigned the raf case by a deleted mubtada such as kinun or
mustaqirrun, then the sentence should be considered as nominal, with a
xabar under the first analysis, or with a fil replacing the xabar under the
second. If, however, we analyze it [i.e. zaydun or amrun] as fil of [an
underlying] istaqarra, then the sentence is verbal; if [the operator assign-
ing raf to the fil] is the adverbial, then the sentence should be consid-
ered as a jumla zarfiyya (Ibn Him al-Ansr, Mun, 494).
Here Ibn Him offers four ways for analyzing sentences such as a-f
d-dri zaydun, correlating each analysis with a different sentence type.
These are the four options as presented in the above passage, in Ibn
Hims order:
1. zaydun could be analyzed as mubtada. This would imply that the

sentence is a jumla ismiyya, with the adverbial/prepositional phrase
implementing the function of (a preposed) xabar.
2. We could posit an underlying mubtada, such as kinun or musta-
qirrun, assigning the raf case to zaydun. In this case zaydun would
implement the function of fil replacing (munin an) the xabar. The
sentence under such an analysis would be regarded, according to Ibn
Him, as jumla ismiyya.
3. zaydun could be analyzed as fil assigned the raf case by the under-
lying verb istaqarra. In this case the sentence would be considered as
jumla filiyya.
4. If, however, we analyze zaydun as a fil receiving its raf case from the
preceding adverbial/prepositional phrase, then the sentence should
be regarded as jumla zarfiyya.
In Ibn Hims view, then, a fil is not necessarily preceded by a finite

verb. But it is only when the operator assigning raf to the fil is a finite
verb (whether explicit or underlying) that the sentence may be con-
ceived of as jumla filiyya. When the raf assigner is a participle (whether
explicit or underlying) or an adverbial/prepositional phrase, the sen-
tence must be conceived of as a jumla ismiyya in the first case, and as a
jumla zarfiyya in the second, even though the actual use of the term fil
suggests that the participle and the adverbial / prepositional phrase in
such cases behave analogously to a verb.
As can be seen, Ibn Him presents the four options without any
attempt to defend his categorization. His analyses are consistent with
his definitions of the three sentence types (see 5.2 above), and mani-
festly reflect the controversies relating to the constructions in question.
184 yishai peled
The proponents of the first option would presumably regard a-f d-dri
zaydun as the inverted version of a-zaydun f d-dr. The occurrence of
the interrogative a- in this case is irrelevant, as is the case also under
the third analysis, where the sentence is presented as an unmistakable
jumla filiyya. Indeed, positing an underlying verb like istaqarra in order
to account for the raf case of zaydun in sentences of this kind was com-
mon practice among the grammarians, as we saw in 3.2. What is really
remarkable in Ibn Hims third analysis is that it leads to the important
conclusion that under a certain analysis a sentence such as a-f d-dri
zaydun could be conceived of as jumla filiyya.
Under the second analysis, zaydun implements the function of fil
following a deleted mubtada, thus occupying a xabar position. The
adverbial/prepositional phrase, under this as well as under the third
analysis (see above), would be analyzed as an adjunct. Obviously, the
second analysis is reminiscent of the analysis of (a-)qimun zaydun
into a mubtada followed by a fil replacing the xabar, as we saw in sec-
tion 2. Note that, unlike Sbawayhi (Kitb I:239; and cf. 3.1 above), Ibn
Him does not view the prepositional phrase as occupying a mubtada
position. Rather, the mubtada in this case is an underlying participle.
Here, at any rate, a sentence whose subject is labeled fil is categorized
as jumla ismiyya.
As we have already indicated, the fourth analysis is consistent with
Ibn Hims theory of three sentence types, each defined by the pre-
dicative constituent introducing the sentence (and acting as mil upon
the second constituent). But we have already seen (3.3 above) that it
remains unclear how an adverbial/prepositional phrase can function
as a verb assigning raf to a following nominal constituent. Following
Sbawayhi, it was often argued that such a phrase may act analogously to
a verb when preceded by a supporting element such as the interroga-
tive particle a- (itimdcf. sections 2, 4.1 above). But does that in itself
warrant categorizing the construction a-f d-dri zaydun as represent-
ing a sentence type in its own right? The concept of jumla zarfiyya does
not seem to have been seriously discussed in the writings of the medi-
eval Arab grammarians. Evidently, the vast majority found it difficult to
fit the concept of jumla zarfiyya into their theory of amal. Indeed, this is
manifested even in Ibn Hims position, which does not present a-f d-
dri zaydun as a straightforward jumla zarfiyya. Rather, it makes it clear that
the actual identification of a sentence as jumla zarfiyya is essen-
tially dependent upon conceiving the adverbial/prepositional phrase as
a mil assigning raf to the following nominal. The other two types,
in contrast, could be determined straightforwardly, since both the verb

and the ibtid were established awmil in the medieval theory of amal.
Hence the grammarians adherence to the binary system of jumla filiyya
and jumla ismiyya.38
6. Summary
Since the grammarians theory of sentence types grew out of, and has
always been closely related to, their theory of amal, it is not surprising
that in elaborate discussions of Arabic sentence types, particularly those
of Fris and Ibn Him, problems relating to the categorization of cer-
tain constructions were couched in terms of case assignment (amal).
The basic types of jumla filiyya and jumla ismiyya are shown through-
out to represent two types of amal: verbal tadiya and ibtid. Sentences
such as qimun zaydun and f d-dri zaydun/rajulun are shown to be
problematical in terms of amal. With regard to qimun zaydun, we have
seen that many grammarians advocated the rather awkward analysis of
mubtada+fil sadda masadd al-xabar. This was designed to deal with
the essentially nominal nature of the participle occurring sentence-ini-
tially, as well as with its verb-like behavior in this particular case. Apart
from Ibn Him who regarded this construction as an example of jumla
ismiyya, the proponents of the above analysis did not commit themselves
to any clear-cut categorization of this particular structure. Regarding f
d-dri zaydun/rajulun, the very fact that this construction displays an
adverbial/prepositional predicative constituent in sentence-initial posi-
tion, gave rise to the hypothesis that it represents a sentence type in its
own right, a jumla zarfiyya. And it comes as no surprise that this was
associated with the hypothesis that in such cases it is the adverbial/prep-
ositional phrase that assigns raf to the nominal constituent following it.
Obviously, this hypothesis and the long established istaqarra/mustaqir-
run hypothesis were mutually exclusive. In 3.3 we saw Friss attempt to
refute the istaqarra/mustaqirrun hypothesis, arguing from the theory of
amal. This was his line of defending a tripartite sentence-type system.
38
For a modern study advocating a three-type division, see Kouloughli (2002, 2124)
who argues that sentences such as f d-dri rajulun/zaydun (referred to by him as loca-
tive sentences) should be viewd as representing a sentence type in its on right, since
they exhibit a number of syntactic and semantic properties not shared by regular topic-
comment sentences.
186 yishai peled
For him, indeed, claiming that fh zaydun represented a sentence type

in its own right was tantamount to presenting fh as the assigner of raf
to zaydun. But such an argument could never be accepted by the major-
ity of grammarians, because it was considered a major violation of a cen-
tral principle of the theory of amal. The vast majority of grammarians
did find the istaqarra/mustaqirrun hypothesis a convenient tool for fit-
ting such constructions as zaydun fh and fh zaydun into their theory
of amal. The question of sentence type was apparently secondary. Once
they established the status of mubtada to zaydun in both cases, they
could argue that both represent a jumla ismiyya, irrespective of whether
the underlying element linking the adverbial/prepositional phrase to
the mubtada is a verb or a participle. Fh zaydun was thus conceived
of as the inverted version of zaydun fh. To Ibn Him, as we have seen,
this was unacceptable. For, if underlying fh zaydun is the structure
yastaqirru fh zaydun, it follows that zaydun is assigned the raf case
by the underlying verb occupying sentence-initial position. And Ibn
Him, indeed, drew the conclusion following from that assumption,
namely that under the above analysis fh zaydun must be categorized
as jumla filiyya. But this position of Ibn Hims is clearly exceptional
in the medieval grammatical literature. The majority of grammarians,
conforming to the theory of amal, did posit istaqarra as the underlying
case assigner to zaydun, but they never went as far as categorizing fh
zaydun as jumla filiyya. Rather, fh zaydun has always been regarded
an ibtid construction, clearly associated with jumla ismiyya.
7. References
7.1 Primary sources

Astarbd, arh = Rad ad-Dn Muhammad b. al-H asan al-Astarbd, arh Kfiyat
Ibn al-H jib. Emil Bad Yaqb, ed. Beirut: Dr al-Kutub al-Ilmiyya, 1998.
Fris, Askariyya = Ab Al l-Fris, al-Masil al-askariyyah. Muhammad a-ti r
Ahmad Muhammad Ahmad, ed. Cairo: Matb aat al-Madan, 1982.
Farr, Man = Ab Zakariyy Yahy b. Ziyd al-Farr, Man l-Qurn. Ahmad
Ysuf Najt and Muhammad Al an-Najjr, eds. Cairo: ad-Dr al-Misriyya li-t-Talf
wa-t-Tarjama, 19551972.
Ibn Ab r-Rab, Bast = Ubaydallh b. Ahmad b. Ubaydallh Ibn Ab r-Rab, al-Bast f
arh jumal az-Zajjj. Ayyd b. d at-Tabt, ed. Beirut: Dr al-arb al-Islm, 1986.
Ibn al-Anbr, Asrr = Ab l-Barakt Abd ar-Rahmn b. Muhammd b. Ab Sad al-
Anbr, Kitb Asrr al-Arabiyyah. Muhammad Bahjat al-Bayt r, ed. Damascus:
Matbt al-Majma al-Ilm al-Arab bi-Dimaq, 1957.
Ibn al-Anbr, Insf = Ab l-Barakt Abd ar-Rahmn b. Muhammd b. Ab Sad al-
Anbr, Kitb al-Insf f masil al-xilf bayn an-nahwiyyn al-basriyyn wa-l-kfiyyn.
Muhammad Muh ad-Dn Abd al-H amd, ed. Beirut: Dr al-Fikr, n.d.
Ibn Aql, arh = Bah ad-Dn Abdallh Ibn Aql, arh Ibn Aql al alfiyyat Ibn Mlik.
Muhammad Muhy ad-Dn Abd al-H amd, ed. n.p: Dr Sab, n.d.
Ibn Him, Mun = Jaml ad-Dn Ibn Him al-Ansr, Mun l-labb an kutub al-
arb. Mzin al-Mubrak and Muhammad Al H amdallh, eds. Beirut: Dr al-Fikr,
1985.
Ibn Jinn, Sirr = Ab l-Fath Utmn Ibn Jinn, Sirr sinat al-irb. H asan Hindw, ed.
Damascus: Dr al-Qalam, 1985.
Ibn Jinn, Xasis = Ab l-Fath Utmn Ibn Jinn, al-Xasis. Muhammad Al an-Najjr,
ed. Cairo: al-Haya l-Misriyya l-mma li-l-Kitb, 19861988.
Ibn as-Sarrj, Usl = Ab Bakr Muhammad b. Sahl Ibn as-Sarrj, Al-Usl f n-nahw.
Abd al-H usayn al-Fatl, ed. Beirut: Muassasat ar-Risla, 1987.
Ibn Usfr, arh = Al b. Mumin b. Muhammad b. Al Ibn Usfr, arh jumal az-
Zajjj. Shib Ab Janh, ed. Mosul: Ihy at-Turt al-Islm, 19801982.
Ibn Ya, arh = Muwaffaq ad-Dn Ya b. Al Ibn Ya, arh al-Mufassa l. Cairo:
Maktabat al-Mutanabb, n.d.
Jurjn, Muqtasid = Abd al-Qhir al-Jurjn, Kitb al-muqtasid f arh al-dh . Kzim
Bahr al-Marjn, ed. Baghdad: Dr ar-Rad li-n-Nar, 1982.
Muji, arh = Ab l-H asan Al b. Faddl al-Muji, arh uyn al-irb. Abd al-
Fatth Salm, ed. Cairo: Dr al-Marif, 1988.
Sbawayhi, Kitb = Ab Bir Amr b. Utm n Sbawayhi, Al-Kitb. Hartwig Derenbourg,
ed. Hildesheim and New York: G. Olms, 1970.
Zajjj, Jumal = Ab l-Qsim Abd ar-Rahmn b. Ishq az-Zajjj, Kitb al-jumal f n-
nahw. Al Tawfq al-H amad, ed. Beirut: Muassasat ar-Risla, Dr al-Amal, 1988.

Badawi, El-Said M. 2000. Ray f man l-irb f fush t-turt: H lat al-jumla l-
ismiyya. Diversity in Language: Contrastive Studies in Arabic and English Theoretical
and Applied Linguistics, Z. M. Ibrahim, S. T. Aydelott and N. Kassabgy, eds., 120.
Cairo: The American University in Cairo Press.
Carter, Michael G. 1973. Sarf et H ilf: Contribution lhistoire de la grammaire arabe.
Arabica 203, 292304.
. (ed.). 1981. Arab Linguistics: An introductory classical text with translation and
notes (irbns Nr al-sajiyyah). Amsterdam: John Benjamins.
Goldenberg, Gideon. 1988. Subject and Predicate in Arab Grammatical Tradition.
Zeitschrift der Deutschen Morgenlndischen Gesellschaft 1381, 3973.
. 2002. Two Types of Phrase Adjectivization. Sprich doch mit deinen Knechten
aramisch, wir verstehen es!: 60 Beitrge zur Semitistik, Festschrift fr Otto Jastrow
zum 60. Geburtstag, W. Arnold and H. Bobzin, eds. 193208. Wiesbaden: Harras-
sowitz.
Kouloughli, D. E. 2002. On locative sentences in Arabic. Zeitschrift fr arabische Lin-
guistik 41, 726.
Levin, Aryeh. 1985. The distinction between nominal and verbal sentences according
to the Arab grammarians. Zeitschrift fr arabische Linguistik 15, 118127.
. 1987. The views of the Arab grammarians on the classification and syntactic func-
tion of prepositions. Jerusalem Studies in Arabic and Islam 10, 342367.
. 1989. What is meant by akaln l-bartu? Jerusalem Studies in Arabic and
Islam 12, 4065.
Owens, Jonathan. 1988. The Foundations of Grammar: An Introduction to Medieval Ara-
bic Grammatical Theory. Amsterdam: John Benjamins.
. 1989. The Syntactic Basis of Arabic Word Classification. Arabica 36, 211234.
Peled, Yishai. 1992a. Amal and Ibtid in Medieval Arabic Grammatical Tradition.
Abr-Nahrain 30, 146171.
. 1992b. Cataphora and Taqdr in medieval Arabic grammatical theory. Jerusalem
Studies in Arabic and Islam 15, 94112.
188 yishai peled
Talmon, Rafael. 1993. Two Early non-Sbawaihian Views of amal in Kernel-Sen-

tences. Zeitschrift fr arabische Linguistik 25, 278288.
. 1997. Arabic Grammar in its Formative Age: Kitb al-Ayn and its attribution to
Xall b. Ah mad. Leiden: E. J. Brill.
ARABIC AVANT LA LETTRE. DIVINE, PROPHETIC, AND
HEROIC ARABIC
Stefan Wild
Bonn
1. Introduction: the Qurn
The history of the Arabic language is indelibly marked by the fact that
the Qurn has made Arabic a prophetic language with its own holy
book and a worldwide appeal. Arabic had been a mantic language in
pre-Islamic times. When the poet recited verses or when the soothsayer
uttered his sayings, reciter and listener were sure that behind his voice
there was another voice. This voice really speaking was that of a higher
power.
With the Qurn, the mantic voice behind the voice of the Prophet
Muhammad became in the believers ear the voice of the one and only
God. Muslim dogma and the consensus of the unlearned considered
the Qurn to be the direct, undiluted Arabic word of God. The status of
Classical Arabic, the standardization of Arabic including the develop-
ment of Modern Standard Arabic, the diglossia Standard Arabic versus
Arabic dialects, the nature of the Arabic vocabulary, Arabic orthography,
Arabic style and vocabularyall are unthinkable without the Qurnic
fact. The history of the Arabic language down to our times cannot be
written without constant reference to the Qurn.
Conversely, the Qurn is deeply marked by its Arabness. The Qurn
is the first literary document in Arabic. It is also the first Arabic docu-
ment to mention the Arabic language. In contradistinction to the atti-
tude of the Jewish Bible and the Christian New Testament toward their
own linguistic forms, the language of the Qurn is an important topic
of Qurnic self-reflection. While neither the Bible nor the New Testa-
ment refer to their Hebrew (or Aramaic) and Greek literary forms, the
Arabic character of the Qurn is in its self-view a cornerstone of its
divine quality.
The adjective arab in the Qurn always refers to the language, never
to a tribe or a social class. The term as an ethnic designation is also
190 stefan wild
very rare in Jhiliyya poetry.1 In the Qurn, arab refers always to the
language of the holy text. It occurs eleven times in the Qurn, and only
in Suras traditionally dated to the Middle and late Meccan period. In
six passages, the adjective arab is a qualification of the word Qurn,
a word meaning primarily recitation, reading aloud . . . (Sura 12:12;
20:113; 39: 2728; 41:14; 42:7; 43:14). The conjunction lisn arab
Arabic tongue occurs three times in the Qurn (Sura 16:103; 26:195,
46:12), and is used to describe the language of the Qurn. The conjunc-
tion h ukm arab (Sura 13:37) an Arabic judgment also refers to the
holy text, and in 41:44, in which the possibility of a Qurn ajam a
non-Arabic Qurn is dismissed, the adjective arab again refers to the
holy text. The Arabic quality of Qurnic revelation could scarcely be
more solidly established.
On the other hand, the Qurn seems indifferent to the linguistic
shape of preceding revelations. In the Qurn, the only language used is
Arabic. Arabic is also the only language mentioned by name. The Qurn
does not specify in which languages Nh, Ibrhm, Isml, Ms, s or
other prophets and messengers spoke to their peoples or in which lan-
guages their holy books might have been.
God speaks Arabic to Adam and his wife, Satan whispers in Arabic
(Sura 20:120), the angels and the jinn speak Arabic (Sura 72:115),
Moses addresses the Pharaoh, Joseph addresses the Egyptian ministers
wife in Arabic, Jesus speaks Arabic from his cradle, D l-Qarnayn and
the People of the Cavethey all use Arabic. Every single soul is made to
speak Arabic at the Day of Judgment, animals like the ant (Sura 27:18) or
the hoopoe (Sura 27:22), even inanimate entities like Hell (Sura 50:30)
speak Arabic. Everybody and everything that speaks in the Qurn must
necessarily speak Arabic, because Arabic is the only language used
throughout the Qurn.
But the intention of the text is in no way to convey that all mankind
throughout history shared and will share the same language. I do not
know of any exegete who concluded from the Qurnic accounts that
the language used between Ms and the Egyptian Pharaoh was Arabic
or that the language used between the Egyptian notables wife and her
lady-friends must have been Arabic, or that the Messiah spoke in Ara-
bicjust because the Qurn reproduces their words in Arabic. It is a
different matter for Adam and Isml (see below).
1
Cf. Agha and Khalidi: Poetry and Identity 70.
divine, prophetic, and heroic arabic 191
On the other hand, the Qurn does mention the existence of lan-
guages other than Arabic and emphasizes that the plurality and variety
of human languages is a sign of divine grace. The divine creation of dif-
ferent languages is as important as the creation of heaven and earth:
wa-min ytih xalqu s-samawti wa-l-ardi wa-xtilfu alsinatikum
wa-alwnikum nna f dlika la-ytin li-l-limn
And of His signs is the creation of heavens and earth and the variety of
your tongues and hues. Surely in this are signs for people who know (Sura
30:22).
Even as God created man and woman, heaven and earth, he created
different colors (alwn, or: kinds of human skin) and different human
languages (cf. the enumeration in Sura 30:2025). The existence of dif-
ferent languages is one of Gods signs for those who know. Unless the
word ya is here taken to mean a portent of warning, such a view seems
difficult to reconcile with the myth of the Babylonian tower (Gen 11:
19), according to which the origin of a multitude of human languages
is divine punishment. Sura 30:22 may even be an inter-textual stab at the
narrative of the Babylonian Tower. There were, however, extra-Qurnic
traditions that preserved the motif that the difference between human
languages was due to an act of divine wrath (p. 196).
A second important element in the Qurnic linguistic outlook flows
forth from this esteem of other languages. In the course of history, God
sent each messenger (rasl) to a specific people (qawm), and this mes-
senger brought the divine message to that people in its language.
And We have sent no messenger save with the tongue of his people that he
might make all clear to them
wa-m arsaln min raslin ill bi-lisni qawmih li-yubayyina lahum
(Sura 14:4).
The primary raison dtre of the Qurn is that the Prophet Muhammads
message was in Arabic. Whereas the other prophets and messengers had
been sent earlier with a message in the languages of their peoples, who
did not speak Arabic, Muhammad is sent to the Arabs. The Arabic lan-
guage vouchsafes the understanding of the Arabic-speaking audience.
For a great part of the Qurn, the fact that this revelation was in Arabic
was the most important difference between Muslim revelation and all
previous revelation.
These Qurnic elements influenced a theological-linguistic scenario
that gave rise to a particular image of linguistic history. The claim that
192 stefan wild
Arabic as the language of the Qurn had become a divine language,

led to the construction of a religious pre-history of Arabic. There were
three competing strands of knowledge that were interwoven to form a
comprehensive imaginaire of the history of Arabic and by implication of
human languages in general:
1. the philosophical question of the origin of language in general;

2. exegesis of the Qurn, including the Qisas al-anbiy literature, i.e.,
reports on earlier prophets, which depended heavily on Biblical and
Aggadic material; and
3. historical traditions about pre-Islamic history, including the genealo-
gies of the Arab tribes.
One of the main issues was the origin of Arabic, an issue often framed as
the question Who was the first human being to speak Arabic? Muslim-
Arab scholarship tried sometimes to disentangle, sometimes to com-
bine and often to harmonize the heterogeneous strands of philosophical
speculation, exegetical H adt, and genealogical traditions. In many cases,
however, contradictory reports were just left standing side by side. An
extensive and useful overview of much of the Arabic material was given
by as-Suyt in his Muzhir.2 The following remarks will outline the main
lines of this colorful and often contradictory linguistic yarn.
2. Adams Arabic and Nhs Syriac
The basic divide between Arab philosophers and theologians concerned

with the origin of language was whether it rested on human convention
(isti lh , tawdu) or on a divine act of revelation (ilhm, wahy, tawqf ).3
The Platonic controversy whether human language was what it was phy-
sei or thesei, by nature or by imposition, here took a new form. A further
problem was whether, theologically speaking, Arabic had a special lin-
guistic status that set it apart from all other languages.
2
Andrzej Czapkiewicz, The Views of the Medieval Arab Philologists on Language and
its Origin in the Light of Al-Suyutis Al-Muzhir, Krakow 1989. Czapkieviczs translations
are sometimes hard to understand.
3
For the difference between tawqf and ilhm, cf. J. van Ess, Theologie und Gesell-
schaft IV 325.
An important part of tradition seems to be in general agreement that

Adam, who is in Islam a prophet, was the first human being to speak Ara-
bic. The revelationists based their view mainly on the early exegesis of
Qurn 2:31 wa-allama dama l-asma kullah and He (God) taught
Adam all names. Weiss explains: According to at-Tabar, the majority
of early exegetes, including the noted Companion and tafsr-authority
Ibn Abbs, interpreted this verse as meaning that God taught Adam the
Arabic names of all existing things. This meant that Adams language
was revealed by divine teaching. It also clearly implied that God taught
Adam language in its entirety.4 According to Ibn Jinn (d. 392/1002),
most Mutazilites, on the other hand, taught that all languages including
Arabic rested on convention, not on revelation.5
Even those who accepted the revelationist theory could argue that
Sura 2:31 meant that Adam was taught the names of all things created in
all languages, in Arabic, of course, but also in Persian, Syriac, Hebrew,
Greek, and all other languages as Ibn Jinn explains. Adam and his off-
spring spoke all these languages. But in the course of time, Adams off-
spring spread over the world and each of them stuck to one of these
languages until this language became his main language and other
languages eluded him because of his lack of contact with them (qad
fussira hd bi-an qla inna llha subh nah allama dama asma
jami l-maxlqti bi-jami l-lugti l-arabiyyati wa-l-frisiyyati wa-l-
suryniyyati wa-l-ibrniyyati wa-r-rmiyyati wa-gayri dlika min siri
l-lugti fa-kna damu wa-waladuh yatakallamna bih tumma inna
waladah tafarraq f d-duny wa-aliqa kullun minhum bi-lugatin min
tilka l-lugti fa-galabat alayhi wa-dmahalla anhu m siwh li-budi
ahdihim bih, Ibn Jinn, Xasis I 41). This model safeguarded the pri-
ority of Arabic; Arabic as the language of the divine word must be prior
to all other languages, whether their origin was revealed or due to con-
vention. But according to some, all other languages were also revealed
to Adam by God. Others with a different timetable thought that differ-
ent linguistic communities emerged only after the Flood. The idea that
Adam had known all human languages found a parallel in the belief of
some Shiites in the 9th century that the Prophet Muhammad had known
4
Weiss, Muslim Discussions 37.
5
aktaru ahli n-nazari al anna asla l-lugti innama huwa tawdu wa-sti lh l
wahy wa-tawqf (Al-Xasis I 4041).
194 stefan wild
all languages.6 The plurality of human languages is sometimes seen as

due to a benevolent divine intervention: He (God) shaped every nation
conforming to a language in which He made them speak and which
He made easy for them.7 God had made His Arabic revelation easy
for the Arabs (e.g. Sura 54:17); in a similar way, earlier revelations had
been made in other languages in order to make them easy for their
listeners.
The most widely accepted historical model that explained the exis-
tence of languages other than Arabic was built on the supposition that
Arabic had been at a certain time the universal language. The idea of
a universal common language can also be found in Gen 11:1, where,
however, this language does not have a name. For most Arabic scholars,
Arabic was the earliest existing language.8 But this universality had at
some point in history come to an end, and decadence, corruption, and
confusion ( fasd, tah rf, tabalbul) had set in. This statement was hard
to reconcile with the Qurnic assertion that God himself had created
the variety of human languages (Sura 30:22). The reason for the devel-
opment of other languages was nevertheless frequently seen either in
a divine act of punishing Adam or mankind, or in a general confusion
(balbala), which mixed up what had been one common human lan-
guage in a gradual process of corruption (tah rf ). Tah rf and tabalbul
were sometimes expressly linked to a punitive act of God, sometimes
they seem to be seen more as a general tendency of history. According
to Ibn Askirs chronicle, Arabic was Adams language in Paradise until
he disobeyed God. Then God deprived him of Arabic and he started
speaking Syriac, evidently considered a lesser language or a corrupted
form of Arabic. When Adam repented, however, God gave the Arabic
language back to him.9
A second less revelationist tradition was the following:
The first language with which Adam came down from Paradise was Ara-
bic. When the contact with Arabic became far and distant, Arabic was
corrupted (h urrifa) and became Syriac (suryniyya) which is a Nisba
to ard sr or Suryna. This is the land of al-Jazra, where Nh and his
6
Van Ess, Theologie und Gesellschaft IV 324.
7
az-Zubayd, Tabaqt an-nahwiyyn (ed. Muhammad Ab l-Fadl Ibrhm), Cairo
1954/1373, 1 jabala kulla ummatin mina l-umami al lugatin antaqahum bih wa-
yassara lahum.
8
lugatu l-arabi asbaqu l-lugti wujdan, as-Suyt , Muzhir I 28, 4.
9
as-Suyt, Muzhir I 30, 10.
people lived before the flood. Syriac resembled Arabic, but it was cor-
rupted (muh arraf ), and Syriac was the language of all people in Nhs ark
except for one man, whose name was Jurhum, and his language was still
the original Arabic. When they left the ark, ram b. Sm married one of
his (Jurhums) daughters. From them the Arabic language came down on
his offspring Aws Ab d, Ubayl, Jir Ab Tamd, and Jdis. d was
given the name of Jurhum, because he was their maternal forefather. And
Syriac stayed with the offspring of Arfaxad b. Sm until it reached Yajub
b. Qaht n. He was in Yemen. There the Ban Isml settled and the Ban
Qaht n learned from them the Arabic language.10
It is interesting to see that the term tah rf, which is the usual word
denoting the falsification of the revealed scriptures by Jews and Chris-
tians, is here used to explain a fact of imagined linguistic history : the
corruption and subsequent loss of Arabic. According to this account,
Nhs language was not Arabic, but Syriac. Arabic had survived only
with Jurhum and his tribe.
Abdalmalik b. H abb (d. 238/852) developed a stemma of prophetic lan-

guages. According to him, these prophetic languages are Arabic, Syriac,
and Hebrew. All the sons of Israel spoke Hebrew, the first to speak it
was Ishq. Syriac was the language of five prophets: Idrs, Nh, Ibrhm,
Lt and Ynus. And twelve of the Prophets spoke Arabic: Adam, Sht,
Hd, Slih, Isml, uayb, al-Xidr, the three in surat Ysn i.e. the three
nameless messengers who were sent as the Companions of the City
(ash b al-qarya, sura 36:13ff.), Xlid b. Sinn al-Abs, the legendary
forerunner of the Prophet Muhammad, and the Prophet Muhammad
himself.11
Others added more and more details. According to al-Azraq, Nh
had eighty men with their families on his ark. When the ark came to a
halt on Mount Jd, Nh founded a village called Tamnn (eighty).
The next morning, they found that their tongues/languages had been
confused and that there were now eighty languages, one of which was
Arabic. They did not understand each other any more. 12 This account
offers no explanation for the emergence of these eighty languages, but
10
as-Suyt , Muzhir I 30,-6 according to Abdalmalik b. H abb.
11
Claude Gilliot and Pierre Larcher Language and Style of the Qurn. EQ 3, 109
135, especially The mythical narratives on the superiority of Arabic 118ff.
12
fa-btan qaryatan wa-sammh Tamnin fa-asbah dta yawmin wa-qad tabal-
balat alsinatuhum al tamnn lughatan ih dh l-arabiyyatu (al-Azraq, Axbr Makka,
/ed. F. Wstenfeld) vol. 1, 20.
196 stefan wild
is based on the idea that Nh and his followers originally did not speak
Arabic.
In other accounts, Ibrhm is said to have spoken two languages,
Syriac and Hebrew. When he fled from the persecution by Namrd in
Kta in the land of Babylon, his language was Syriac (lisnuh suryn).
Namrd wanted to catch him and gave order that anybody speaking
Syriac should be arrested and brought before him. But when Ibrhm
crossed the Euphrates from H arrn, God changed his language and
Ibrhm miraculously started speaking Hebrew (ibrn). He escaped
because his persecutors did not know this language.13
A further version closer to the narrative of the Babylonian Tower is
reported by at- Tabar: Namrd has a high building (sarh ) made, until it
reaches the sky. God destroys the building and on that day the languages
of people become confused from fright. They then speak in seventy-
three languages. This is why the place was called Bbil. The language of
mankind before that was Syriac.14 A different kind of etiological expla-
nation is linked to Yarub (see below section 4). The etymological con-
nection between balbala and the name Bbil is often invoked and finds
its way into Arabic lexicography.15 This connection between the Arabic
word for confusion (balbala) and the name of Bbil is prefigured in the
Bible, in which the name of the city of Babel is linked to the Hebrew
verb balal, also meaning to confuse (Gen 11:7). The number of exist-
ing languages is given in different ways. According to al-Masd, there
were seventy-two languages divided under Nhs sons: the descendants
of Sm spoke nineteen languages, the descendants of H m seventeen,
and those of Yfit thirty six.16
As far as I can see, nobody in pre-modern times ever claimed that the
prophet Ibrhm spoke Arabicalthough his unequalled importance
for the link between earlier monotheistic religions and Islam, especially
through his role in building the Kaba, would have made such a claim
attractive. Abraham is called a Muslim in the Qurn (h anf muslim
13
Ibn Sad, Tabaqt ed. Eugen Mittwoch, Leiden 1905, I:1 21, 14; a similar tradition
in Ibn Mutarrif at -Taraf, Qisas al-anbiy no. 124.
14
fa-tabalbalat alsunu n-nsi min yawmaidin mina l-fazai fa-takallam bi-taltati
wa-sabna lisnan fa-li-dlika summiyat Bbil wa-innam kna lisnu n-nsi qabla
dlika s-suryniyyata., At-Tabar, Trx I 322.
15
Lisn al-Arab s.v. bll: summiyat ardu Bbil li-anna llha h na arda an yuxlifa
bayna alsinati ban dama baatha r-rh an fa-h aarahum min kulli ufuqin il Bbil fa-
balbala llhu bih alsinatahum tumma farraqahum tilka r-rh u fi l-bild.
16
al-Masd, Murj ad-dahab, ed. Barbier de Meynard I 78.
sura 3:67)but never an Arab. This claim was put forward only by zeal-
ous Arab nationalists. When in 1999 Pope John Paul II announced his
plan to visit Ur in Iraq, members of the ruling Bat -party criticized that
the Pope in his announcement had not spoken of Ibrhm as an Arab.
They argued: Ibrhm was born in the land of the Arabs and he lived
in Iraq, in the Sumerian city of Ur. When he was chased out of his place
of birth because of his monotheist creed, Ibrhm started his combative
journey through the lands of the Arabs, H arrn, Palestine, Egypt, and
Mecca . . . Combative was a favorite attribute of praise in the Bat-party.
The Bats do not expressly identify Arabic as Ibrhms languageeven
though the reader is forced to conclude that the language that the com-
bative Arab Ibrhm spoke must have been Arabic.17 Such a linguistic
myth could be elaborated. There is frequently a tendency in modern
popular Arab discourse to call Arabic what is elsewhere called Semitic,
in order to extol the importance of Arabic. The Canaanite tribes preced-
ing the Israelite settlers are then called Arab Canaanite tribes, and the
variant of Akkadian to be found in the Ebla texts is called Arabic.18 For
the claim that the Aramaic of the Nabatean inscriptions is really Ara-
bic, see below.
3. Ismls Arabic
Next to Adam, the most important figure with whom the introduc-
tion of Arabic is firmly connected is Ibrhms son Isml. In one of the
most important foundational Muslim narratives, God orders Ibrhm
to migrate to Mecca with young Isml and the latters mother Hagar.
Ibrhm builds the Kaba together with Isml. There they meet mem-
bers of the (Arabic-speaking) Jurhum tribe. Isml grows up with their
children, learns how to shoot (the bow) and to speak in their language,
and he takes a Jurhum-wife.19 Another report says, without reference to
where and how, Ismail learned Arabic: Sarah gave Hajar to Ibrhm, he
slept with her and she bore him Isml, who was Ibrhms eldest son.
His name used to be Iml which was later arabicized (wa-kna smuh
17
Amatzia Baram, Der moderne Irak, die Baath-Partei und der Antisemitismus in
Jahrbuch fr Antisemitismusforschung 12 (2003) 99119 p. 114.
18
Welt des Islams 21 /1982, 240f.
19
Ibn Qutayba, al-Marif, ed. Tarwat Uka, Cairo 1960, 34.
198 stefan wild
mwyl fa-uriba).20 This Arabic tradition closely follows the Biblical

account (Gen 21:20)even the bow is already thereand preserves
in Iml a dialectical Hebrew form of the classical Hebrew Yiml.
Abraham had given his son a Hebrew name, because he spoke Hebrew.
The link between Isml and the Arabic language became part of the
awil-literature. The Prophet had said: all Arabs descend from the off-
spring of Isml b. Ibrhm. The question who was the first to speak
Arabic could, therefore, be answered and the answer is contained in a
short didactic dialogue in which an Ab Jafar Quray b. Uqba b. Bar
asks and a Muhammad b. Al answers:
Q: Who was the first to speak Arabic?
A: The first to speak Arabic was Isml b. Ibrhm, when he was thirteen
years old.
Q: And what did people speak before that?
A: Hebrew.
Q: And what was Gods language that was sent down to his messengers
and servants?
A: Hebrew.21
Here, Adam is not even mentioned as the first recipient of Arabic. And
Gods language of pre-Arabic revelation is simply Hebrew. In awil-
collections, the particular that Ishmael was the first to write in Arabic is
not too frequent.22 See below p. 200.
The report that Isml picked up Arabic from Arabic-speaking tribes
was widely disseminated, but there was a competing tradition saying
that Isml received the Arabic language by revelation on the day he was
born. The three other sons of Ibrhm stuck to their fathers (Hebrew)
language.23 According to a H adt, the Prophet Muhammad recited (the
verse) an Arabic Qurn for people who know (Sura 41:3); then he said:
Isml received this Arabic language by a great revelation (ulhima
Ismlu hd l-lisna l-arabiyya ilhman). This tradition contradicted
the view that Isml had learnt Arabic from the Jurhum tribe. In the
latter case, the tribe must have logically preceded Isml in speaking
Arabic. Another H adt mentions Ismls age: when he learned Arabic,
he was 14 years old.24
20
Ibn Sad, Tabaqt I/1 23, 9.
21
Ibn Sad, Tabaqt I/1 24, 16. Tottoli p. 82. Cf. Sibt Ibn al-Jawz, Mirt 310; Ibn
Askir, Trkh II 331.
22
See the comment of ibl, Mah sin al-wasil f marifat al-awil 143.
23
24
as-Suyt, Muzhir I 22, 13.
Ibn Sad transmits a report that contradicts the idea that Isml was
born with Arabic and traces Arabic only to Ismls offspring. Accord-
ing to this tradition, Isml never spoke Arabic, because in his filial
piety he did not deem it permissible to act differently from his father.
According to this report, the first of his offspring to speak Arabic were
the Ban Rala bint Yajub b. Yarub b. Ldan b. Jurhum b. mir b. Saba
b. Yaqtn b. bir b. likh b. Arafxaad b. Sm b. Nh.25 Nevertheless,
the fact that Isml forgot his fathers Hebrew is expressly stated in a
tradition traced to Muhammad b. Salm:26 The first one to speak Arabic
and to forget his fathers language was Isml. Muslim Arab tradition,
therefore, agrees that Ibrhms language was not Arabic, whereas there
is disagreement on whether it was Isml who was the first to speak
Arabic and, if so, when and how Isml learned it. For the tribal aspects
of such traditions, cf. below under 4. Yarubs Arabic.
The most frequently quoted account, successful as a canonical H adt,
was, however, that Isml learned Arabic from the Jurhum tribe in his
youth (wa-shabba l-ghulmu wa-taallama l-arabiyyata minhum, i.e.,
Jurhum, s. Buxr, Sah h , Anbiy 21). He marries twice, in both cases
a wife from the Jurhum tribe. This again meant, of course, that Isml
was, strictly speaking, not the first one to use Arabic, because he had
to learn it from somebody else. In some reports, the gift of the Arabic
language is mentioned next to other privileges of Isml: the tradition-
ists report that Isml was the first one to speak Arabic, the first one to
build the h aram after his father Ibrhm, and the first one to install the
rites of pilgrimage. He was also the first to ride full-grown horses, which
before were wild and could not be ridden. Some say: Isml was the first
whose tongue God opened to speak Arabic. And when he grew up, God
gave him the Arabic bow. This report implies that God revealed Arabic
to Isml and that Isml did not have to learn it from the Jurhum.27
This report became part of Adab literature. At- Talib mentions it in his
Latif al-Marif. The first person to speak Arabic was Isml, peace be
upon him; all the Arabs came subsequently from his progeny, except for
three tribes, those of Auz, H adramawt and Taqf. He was the first to
25
26
as-Suyt , Muzhir I 32, -2.
27
al-Yaqb, Trx I 22.
200 stefan wild
ride horses, these mounts being originally wild horses which had never
been ridden before.28
There was also some concern about the fact that the long time that had
elapsed between the period in which Isml had started speaking Arabic
and the time of the Prophet Muhammad made it difficult to accept that
the Arabic language should not have changed during this interval. The
Prophet gave the answer: Ismls language had been obfuscated (daru-
sat). But Jibrl came and made me retain the language and I retained it,29
i.e. the Qurnic message saved Arabic.
In other versions, it is not only the Arabic language whose origin is
attributed to Adam or Isml, but also the Arabic script, which is in
this case imagined as written on clay tablets: The first who installed
the Arabic script, the Syriac script, and all other scripts was Adam, 300
years before his death. He wrote them in clay and baked them. When the
earth was hit by the flood, each people received its script and used it to
write. Isml b. Ibrhm received the Arabic writing.30
Some of these conflicting mythological reports are woven together
and transformed into scholarship by modern Wahhabi scholars. Taking
such accounts as factually historical, Muhammad Musta f al-Azam is
led in a recently published book31 to a re-writing of linguistic history
on a grand scale. He asks himself the question What language did the
Nabateans speak? And the answer is:
Growing up in Makkah from his earliest childhood Isml, eldest son of
Ibrhm, was raised among the Jurhum tribe and married within them
twice. This tribe spoke Arabic, and so, undoubtedly, must have Isml.
The Jurhum Arabic probably lacked the sophistication and polish of the
Quraishi Arabic, preceding it as it did almost by two thousand years.
Ibn Ushta records a statement from Ibn Abbs, that the first person to
initiate set rules for the Arabic grammar and alphabet was none other
than Isml. Eventually, Allah commissioned Isml as a messenger and
prophet, to call his people for the worship of the one true God Allah, to
28
In: C.E. Bosworth (ed.), The Book of Curious and Entertaining Information. The
Latif al-marif of al-Thalib, Edinburgh 1968, p. 40, cf. Lammens, La cit arabe de
Taif la veille de lHgire, Beirut 1922, 5768.
29
as-Suyt , Muzhir I 35, 3.
30
as-Suyuti, Itqn IV 143 quoting Ab Bakr M. b. Abdallh b. Muhammad b. Uta
al-Isbahn (d. 360) and his Kitb al-mash if after Kab al-Ahbr: cf. as-Suyt, Bugya
59; as-Safad, Wf III 347; and Nldeke GdQ II 53 and GdQ III 1 fn. 2.
31
Muhammad Musta f al-Azam, The History of the Quranic Text. From Revelation
to Compilation. A Comparative Study with the Old and New Testaments, UK Islamic
Academy, Leicester 2003.
establish prayers and pay alms to the poor. Since Allah sends every mes-
senger in the language of his own people, Isml must have preached in
Arabic. Genesis credits Isml with twelve sons, among them Nebajoth/
Nabat; born and nurtured in these Arabian surroundings they must have
adopted Arabic as their mother tongue. These sons may have preserved
their fathers message by using the prevailing Arabic script; certainly, they
would not have resorted to whatever script was then current in Palestine
(Ibrhms homeland), since two generations had already lived in Arabia.
When Nabat subsequently migrated northwards he must have taken the
Arabic language and alphabet with him. It was his descendants who estab-
lished the Nabatean Kingdom (600 B.C.E.105 C.E.).
Al-Azam dates Isml and his early Arabic at around 1400 B.C.E. and
concludes: The Nabatean language and script were . . . a form of Arabic
(121).
Such a sentence disregards that the Nabateans spoken language was

indeed Arabic, while the texts they wrote were in Aramaic. The thesis
that Nabatean is a form of Arabic comes close to a linguistic panorama,
which sees in all languages usually called Semitic a form of Arabic.
The Arabic language and script, in their primitive forms, gave birth to
the Nabatean and most probably predated the Syriac (121) is but one
example of such a view.
4. Yarubs Arabic
A further source in tracing the first Arabic speaker was tribal history.
The Arabs, in their own self-view, were not only a linguistic community,
but also marked by common ancestors. Descent was a primary symbol
of a tribal community, and the self-esteem and prestige of individuals
and communities were linked to the purity of descent. The efforts of the
Arab genealogists to establish a link between living Arab tribal groups
and past forefathers were, of course, highly tenuous. The skeptical obser-
vation of Ibn H azm (d. 456/1064) that on the face of the earth there is
no one whose descent from them is verifiable did not prevent the emer-
gence of the most speculative lineages.32 By and large, the Arab tribes
claimed descent from one of two ancestors, either a North Arabic origin
32
Ibn H azm, Jamharat ansb al-arab, ed. Lvy-Provencal, Cairo 1948, 8, quoted by
EI 2nd ed. I 546.
202 stefan wild
connected to the North Arab Adnn (also to Maadd and Nizr, later
Qays) or a South Arabic origin connected to Qaht n (also called Yama-
nis, later also Kalb). Some tribes such as d, Tamd, ram, Jurhum,
Tasm and Jads were believed to have disappeared before Islam. Hostil-
ity between the offspring of Qaht n and those of Adnn emerged after
the advent of Islam in the form of the cleavage between the Ansr in
Medina and the Quray. The fact that the Prophet Muhammad belonged
to the Quray brought enormous prestige to the Adnn, i.e., the North
Arabs. The narrative that linked Isml to Arabic was counterbalanced
by the argument that Isml had learned Arabic from the Jurhum, a
South Arabic tribe. While al-Masd accepted that Isml had been
given Arabic by God, he did not deny that Yarub b. Qaht n, the ances-
tor of the Yamanis, was the first to speak Arabic. Which of the three
Arab group, the extinct Arabs, the Nizris, or the Yamanis was the first
to speak Arabic? The Nizri and Yamani groups vigorously upheld their
own claims to this honour. The Nizris held that Isml was given the
language by God, while the Yamanis contended that Isml had learned
the language from a Yamani tribe living in Mecca.33 This conflict of
claims masked a deeper social and political conflict between the two
groups. Al-Masd, while granting that Yarub Ibn Qaht n, the ancestor
of the Yamanis, was the first to speak Arabic (Murj sec. 71), believes
that Ishmael too was granted this honour by God, independently of his
association with the Yamanis.34 The tribal family trees set up by Arab
genealogists always have an agenda. When names are inserted or omit-
ted, insertion and omission usually serve a purpose. When in the list
of Qaht ns ancestors there are two new names, those of Yajub and
Yarub, inserted between Saba and Qaht n (Yoqtan) in a family tree
that is otherwise based on Gen 10, 132, this happens because these
two names fulfill two important functions. Yarub symbolizes and per-
sonalizes the change from Syriac to pure Arabic, while Yajub/Yaman
gives his name to the land Yaman. The appearance and etiological func-
tion of persons with such telling names is a common feature of Arab
genealogy.35
A tradition quoted by Yqt and traced to ad-Dnawars Mujlasa
links Bbil with Yarub and Arabic as the heavenly language: When God
33
Tarif Khalidi, Islamic Historiography. The histories of Masudi, Albany 1975, 116;
conflicting claims are set forth in Tanbh 7983 and Murj 99699.
34
Tarif Khalidi, Islamic Historiography ibid.
35
Manfred Kropp, Geschichte der reinen Araber II 379f.
assembled mankind in Babylon, He sent them an Eastern, a Western, a

Southern, and a Northern wind. And He assembled them in Babylon
and they assembled to see why they had been called. Suddenly, a voice
called out: He who has the West on his right, the East to his left, and
faces Meccahis is the language of the people of heaven. Then Yarub b.
Qaht n stood up, and the voice said: Yarub b. Qaht n b. Hd, you are
the one. And he was the first to speak Arabic.36
According to Ibn Durayd, the pure Arabs were seven tribes d,
Tamd, Imlq, Tasm, Jads, Umayyim und Jsim. Most of them have
vanished except for some remnants dispersed in the sub-tribes. And
Yarub, whose name was Muhazzim b. Qaht n, received his byname
because he was the first whose tongue switched from Syriac to Arabic.
This is the meaning of al-Jawhars statement in as-Sih h : the first to
speak Arabic was Yarub b. Qaht n. 37 Here, the eponymon Yarub is
linked to the root rb, another example for the favorite method of the
genealogists of connecting proper names with imagined history, ety-
mology with linguistic history.
In this mixture of traditions, Yarub b. Qaht n is finally also located in
Babylon and celebrates his virtues in the style of a poetic self-praise.
The majority of historians among them, the author of Tawrx al-

umam (al-Isfahn), and the author of al-Marif (Ibn Qutayba) tell us
that Yarub b. Qaht n was the first to speak in clear Arabic, the first to
bear the crown in Yaman, the first who was greeted by his sons with the
formula May you avoid the curse and your day be happy!
Ibn Hishm, the author of the Book of Crowns, which he wrote on
the kings of H imyar reports: It was Yarub who went with the Arabs
to Yaman and settled there. Therefore Yaman was called after him,
because Yarubs name was Yaman. Al-Bayhaq says: Yarub was the
first to speak in clear Arabic and left the confused way of talking used
by the Arabs. He and his offspring were successful and God bestowed
on them the land of Yaman as heritage. To them belong the kings of
Tubba, who conquered the regions of the earth. On the strength of this
privilege they are ranked before Adnn, while the Adnn can boast the
prophethood of Muhammad. . . . At the time, in which the languages
had already been confused, the Ban H m had come to the highlands of
36
Yqt, Mujam al-buldn ed. F. Wstenfeld, I 447, 19.
37
as-Suyt , Muzhir I 31, -3.
204 stefan wild
Yaman. . . . And Yarub was the most high-minded youth in Babylon and
spoke clear Arabic, while the others spoke in confused language. 38 And
Yarub recited the following verse:
I am the youth favored by the richest gift, the happy one, well known for
his virtue. I am the son of Qaht n, the influential and rich, I spoke in
Arabic, while the people were in (linguistic) confusion. (I spoke) in the
clearest unambiguous language and in the perfect language of the king-
doms after me.
an l-gulmu d n-nasbi l-ajzali / al-aymanu l-marfu bi-t-tajammuli /
an bnu Qaht na l-hammmi l-aqyali / arabtu wa-l-ummatu f tabalbuli /
bi-l-mantiqi l-abyani gayri mukili / wa-mantiqi l-amlki badi l-kamali.
This is pure poetic self-praise in the rajaz meter as we know it from the
earliest Arabic poetry. According to this report, Yarub b. Qaht n was
the first to speak clear Arabic and turned away from the confused ver-
sion of Arabic (namatu l-arabiyyati l-mubalbalati). He let his sons greet
him with the greeting abayta l-lana (may you avoid the curse!) and
with good morning! This Arabic is not so much seen as a prophetic
language but as a heroic language spoken by a tribal noble forefather.
The hero of this poem, Yarub, praises himself for speaking pure Arabic,
a royal language of noble kings, while lesser Arab mortals used mixed
and confused languages.
In a further verse, Yarub predicts the coming of the Prophet
Muhammad:
Muh ammadu l-hd n-nabiyyu l-mursalu / li-llhi darru l-mjidi l-mus-
taqbili
Muhammad, the guide, the God-sent Prophet/ how praiseworthy is the
blessed one who is coming.
Such a tradition combined tribal history with revelation. And to round
things out, the report claims that the same Yarub was the first to recite
Arabic poetry and to put it into meters. He invented the poetic genres
and composed praise-poetry, self-praise, and love poetry.39 Earlier tra-
ditions see the Ban Hshim and the Ban Yarub in fierce competi-
tion. When the rival of the Prophet Muhammad, Musaylima the Liar,
who traced his descent to the Ban Yarub, heard of Muhammads death,
he hoped to outstrip the Prophet and is said to have recited the verse:
38
Kropp, I Arab. text 9ff.
39
Kropp 11, transl. 149.
Passed away the prophet of the Ban Hshim / and rose up the prophet
of the Ban Yarub!40
5. The Prophet Muh ammads Arabic
As we have seen, after the advent of Islam, the unique character of Ara-
bic was extrapolated by theologians, philosophers, and genealogists back
to times immemorial. In their view, after the revelation of the Arabic
Qurn, Arabic became even more of a special language. Arabic itself,
not only its Qurnic form, was something sacred, superhuman. Yet, in a
famous description, a-fi (d. 204/820) explained the special charac-
ter of Arabic mainly by its most important hermeneutical challenge, i.e.,
its vast and ramified vocabulary. According to him, this was the primary
reason for the privileged status of Arabic:
We do not know that any man except for a prophet can claim to know the
Arabic language completely. However, nothing Arabic escapes the Arabs
collectively, so that there would be nobody among the Arabs who knows
it. The knowledge of Arabic among the Arabs resembles the knowledge
of the traditions (sunan) among the scholars of religious Law. We do not
know anybody who could claim to know all traditions, so that not a single
one would escape him. But when the knowledge of traditions of all schol-
ars is collected, the whole of tradition is found. When the knowledge of
each of these scholars is divided, something of the traditions will escape
each of them. But what escapes him will be found with someone else.41
The Prophet, who as an individual can claim to know Arabic completely,
is, of course, Muhammad. His knowledge of Arabic is superhuman. But
the Arabs collectively also know the whole language.
The Prophets own mastering of Arabic was related to the Qurnic
revelation, as well. When a man admires the Prophets rhetorical tal-
ent and says, What a good speaker you are: We have not seen anybody
speaking better Arabic than you! the Prophet answers: This is my right.
For the Qurn was revealed to me in clear Arabic.42 And the Prophet
could boast: I am the best Arabic speaker (an afsah u l-arab43 or an
40
Ibn Katr, Bidya vi, 341, quoted according to M. J. Kister, Musaylima in EQ 3
(2003) 462.
41
A-fi, Risla, ed. Ahmad Muhammad kir, Cairo 1940, 42, 8ff.
42
as-Suyt , Muzhir I 35, 6.
43
Muzhir I 209, 2.
206 stefan wild
arabukum).44 It is not always clear whether such statements refer to

Qurnic revelation or to the Prophets non-Qurnic words. In the same
vein, the Prophet was immune against linguistic mistakes (lah n) in his
speech. He was reported to have said: I am from the Quray and I grew
up among the Ban Sadhow should I commit a linguistic error? (an
min Quray wa-naatu f ban Sad wa-ann fiyya l-lah n).45 The Ban
Sad b. Bakr b. Hawzin were the clan of Muhammads wet-nurse. In this
case, the Prophets claim to linguistic excellence was definitely rooted in
his tribal background and not in divine grace. Among the further signs
of the Prophets unprecedented eloquence was that he was said to have
coined expressions that allegedly had never been used in Arabic before
him and that later became part of an elevated Arabic style, such as mta
h atfa anfih, he died a natural death.46
In some traditions, the Prophet became a second Adam. In a H adt,
he said: My community appeared to me in water and clay. And I was
taught all names even as Adam was taught all names.47
Arab grammarians connect the linguistic quality of the Arabic of the
Quray with pre-Islamic times and with the role the Quray played at
the Kaba in Mecca. A privileged position vis--vis the Arabic language
was also conferred on the companions of the Prophetthey also spoke
nothing but the purest Arabic. The proof was that they never coined
a new word (m alimnhumu sta lah al xtiri lugatin aw ih dthi
lafzatin lam tataqaddamhum).48 This was most probably connected to
the fact that the companions were the guardians of the Prophets word
and deeds. If their Arabic had been doubtful, it could have influenced
the correctness of tradition. In a comparable manner, the ethical, philo-
sophical, and religious letters and pronouncements, ascribed under the
title Nahj al-balga (The Method of Eloquence) to Al b. Ab Tlib,
were considered by many Shiites in essence and style as second only to
the Qurn.49
44
Ibn Sad, Tabaqt I 113.
45
Ab t- Tayyib al-Lugaw, Martib an-nahwiyyn, ed. Muhammad Ab l-Fadl
Ibrhm, Cairo n.d. <1955>, Maktabat Nahdat Misr 6,1.
46
Muzhir I 209, 5.
47
Muzhir I 35, 4.
48
Muzhir I 10, 9.
49
Nahj al-Balgawa-huwa majm m xtrahu -arf Ab l-H asan Muh ammad
ar-Rd b. al-H asan al-Msaw min kalm Amr al-muminn Ab l-H asan Al b. Ab
T lib, ed. Subh Slih, Beirut 1982; Moojan Momen, An Introduction to Shii Islam, The
History and Doctrine of Twelver Shiism, New Haven/London 1983, 25.
6. The fate of Arabic after the Islamic conquest
The Arabic language celebrated its greatest triumph with the vast expan-
sion of the Islamic conquest and the concomitant expansion of the Ara-
bic language. But most Arab grammarians and lexicographers did not
see it that way. In an almost incredible volte-face, they saw the fate of the
Arabic language sealed and its purity doomed to corruption precisely
at the moment when the Muslim faith and the Qurn won over the
world:
wa-lam tazal il-arabu tantiqu al sajiyyatih f sadri islmih wa-md
jhiliyyatih h att azhara llhu l-islma al siri l-adyni fa-daxala
n-nsu fhi afwjan wa-aqbal ilayhi arslan wa-jtamaat fhi l-alsinatu
l-mutafarriqatu wa-l-lugtu l-muxtalifatu fa-fa l-fasdu f l-lugati l-ara-
biyyati.50
The Arabs did not cease to speak Arabic according to its disposition in
the beginning of the era of Islam and in the pre-Islamic past until God let
Islam prevail over all other religions. Then people entered Islam in masses
and turned to it in flocks. In Islam all separate languages and different
tongues came together and therefore corruption ( fasd) spread in the
Arabic language.
This rather anticlimactic statement marks the beginning of Arabic as
the language of an Islamic world civilization. For many an Arab gram-
marian and lexicographer, this period was at the same time the starting
point of its decadence and corruption.
7. References
Agha, Saleh Said and Tarif Khalidi. 2002/03. Poetry and Identity in the Umayyad Age.
Al-Abhth 5051, 55119.
al-Azam, Muhammad Musta f. 2003. The History of the Quranic Text. From Revelation
to Compilation. A Comparative Study with the Old and New Testaments. Leicester: UK
Islamic Academy.
al-Azraq. 1858. Die Chroniken der Stadt Mekka (F. Wstenfeld, ed.), vol. 1 Axbr Makka.
Leipzig.
Czapkiewicz, Andrzej. 1988. The Views of the Medieval Arab Philologists on Language
and its Origin in the Light of Al-Suyutis Al-Muzhir. Krakow: Universitas Jagellonica
Acta Scientiarum Litterarumque CMIX.
50
az-Zubayd, Tabaqt an-nahwiyyn wa-l-lugawiyyn (ed. Muhammad Ab l-Fadl
Ibrhm), Cairo 1954, 1.
208 stefan wild
EQ 20012006 = Jane McAuliffe, ed. Encyclopaedia of the Qurn, vol. 15. Leiden:
E. J. Brill
Fischer, A. and A.K. Irvine. 1978. Kaht n. EI 2nd ed., 4: 447449.
Gilliot, Claude and Pierre Larcher. 2003. Language and Style of the Qurn. EQ 3,
109135, especially The mythical narratives on the superiority of Arabic, 118ff.
Goldziher, Ignaz. 1873. Beitrge zur Geschichte der Sprachgelehrsamkeit bei den
Arabern. Sitzungsber. d. Kaiserl. Ak. d.W. Wien, Phil-Hist.Kl. 73, 51152.
Ibn H azm. 1948. Jamharat ansb al-arab. Ed. . Levy-Provencal, Cairo.
Ibn Mutarrif at -Taraf, Ab Abdallh M. b. A. al-Kinn. 2003. Qisas al-anbiy (The
Stories of the Prophets). Roberto Tottoli, ed. Berlin: Islamkundliche Untersuchungen
253.
al-Jumah Muhammad b. Sallm 1916. Tabaqt a-uar. Joseph Hell, ed. Leiden.
Khalidi, Tarif. 1975. Islamic Historiography. The histories of Masudi. Albano.
Kropp, Manfred. 1975. Die Geschichte der reinen Araber vom Stamme Qaht n aus dem
Kitb Nashwat at-Tarab f trkh jhiliyyat al-arab des Ibn Sad al-Maghrib, hrsg. u.
bersetzt von Manfred Kropp, Bd. I: Einleitung und Text, Bd. II: bersetzung und
Anmerkungen, Diss. Heidelberg.
Loucel, Henri. 1963. Lorigine du langage daprs les grammairiens arabes. Arabica 10 I
188208; II 253281; 11 (1964) III 5772; IV 151187.
Rubin, Uri. 1990. H anfiyya and Kaba. An inquiry into the Arabian pre-Islamic back-
ground of dn Ibrhm. JSAI 13, 85112.
as-Suyt , Abdarrahmn Jalladdn. n.d. al-Muzhir f ulm al-luga. Muhammad Ahmad
Jd al-Mawl Bek, Muhammad Ab l-Fadl Ibrhm, Al Muhammad al-Bjw, eds.
Cairo n.d., 2nd ed., Dr Ihy al-Kutub al-arabiyya, vol. 12.
Tottoli, Roberto, ed. 2003. The Stories of the Prophets by Ibn Mutarrif al-Tarafi, edited
with an introduction. Berlin, 253.
. 2002. Biblical Prophets in the Qurn and Muslim Literature. Richmond.
Weiss, Bernard G. 1974. Medieval Muslim Discussions on the Origin of Language.
ZDMG 124, 3341.
INFLECTION AND GOVERNMENT IN ARABIC ACCORDING
TO SPANISH MISSIONARY GRAMMARIANS FROM
DAMASCUS (XVIIITH CENTURY): GRAMMARS AT THE
CROSSROADS OF TWO SYSTEMS?1
Otto Zwartjes
University of Amsterdam, NIAS
1. Introduction
Grammars of Hebrew written in Europe in the Renaissance by Chris-

tians could benefit from the Hebrew grammatical tradition. Johannes
Reuchlin (14551522) quotes in his De rudimentis hebraicis (1506)
Priscian and Rabbi David (i.e. Qimhi) as well (Law 2002, 247248). In
grammars of Sanskrit written in Europe we see also that the framework
of grammatical description has been derived from the Indian
grammatical tradition. The German Jesuit Heinrich Roth (16201668),
as Hauschild (1988, 1314) observes, uses with perfect familiarity the
technical terms of Indian grammar [. . .] Roth stands entirely within
the Indian grammatical tradition, and probably he used the practical
grammar of Anubhti Svarpcrya, called the Srasvata Vykarana,
which was in general circulation in Hindustn, Bihr and Benares. [. . .].
Another candidate, though a less likely one, would be the grammar
called Mugdhabodha, which was composed in the second half of the
13th century by Vopadeva, but the usage of which was more common
1
This article is an elaborated version of paper delivered at the IIIrd International
Conference on Missionary Linguistics, Hong Kong and Macau, 12th15th March 2005.
The organization of the conference and participation in Hong Kong has been made pos-
sible by financial support of the Norwegian Research Council (Norges Forskingsrd)
and the Language Centre of the Hong Kong University of Science and Technology. I
have to acknowledge Kees Versteegh who inspired me to do research in the field of the
History of Linguistics and particularly Michael Carter for his valuable corrections and
suggestions. Thanks to my colleague Maria Cndida Barros, I came across the reference
to the grammar of Lucas Caballero. Research has been made possible by the Radboud
University (TCMO) where my 2 research on Pedro de Alcal started. I continued this
research topic at the University of Oslo, supported by the NFR-project OsProMil (Oslo
Project on Missionary Linguistics). I am grateful to the Rogge Library (Strngns) for
the reproduction of the MS. I gladly acknowledge Pierre Winkler for his translations
from Latin.
210 otto zwartjes
in Bengal . . . The work of Roth was a masterpiece, which does not differ
considerably from current grammars, which similarly depend on the
linguistic achievements of Indias own grammatical tradition (Hauschild
1988, 1314). In the grammars of Japanese written by the Portuguese
Jesuit Joo Rodrigues (15611634), particularly in his description of
particles and verbal endings, we can also find information concerning
the study of tenifa or tenivofa (the study of particles and verbal
endings) from contemporary Japanese scholars (Maruyama 2004, 155).
As has been demonstrated by Gregory James (2007), some missionaries
describing the Tamil language, such as Bartholomus Ziegenbalg (1682
1719), were familiar with some ancient Tamil works on grammar.
These missionaries worked under favourable conditions compared to
their colleagues who described languages, particularly Amerindian and
Austronesian, without any written tradition or an adequate indigenous
grammatical framework they could rely on.
The use of Arabic grammatical terms in the first grammar written in
Europe of vernacular Arabic, the Arte para ligeramente saber la lengua
arabiga (1505) of Pedro de Alcal (Order of St. Jerome) has been the
subject of an article written by William Cowan (1981). In de Alcals
grammar, some technical terms were incorporated in the descriptive
framework, including terms such as damir, temiz, masdar, amr and
xucla. In this article, Spanish grammars of Arabicvernacular and clas-
sicalwritten by Franciscans in Damascus and completed, copied, or
printed in Spain in the 18th century occupy our attention, particularly
the grammars of Francisco Caballero and Juan de la Encarnacin (18th
century) and Francisco Caes (17301795). Of the first we have an
unpublished manuscript, which has escaped the attention of research-
ers until today,2 and of the second a printed work has been conserved,
2
Bibliographical information concerning Francisco Caes can be found in Schnur-
rer (1811, 79, no. 113) BICRES III, and in Monroe (1970) we find some historical back-
ground. However Caes is not mentioned by Fck (1955), Dannenfeldt (1955), and
Killean (1984) and particularly the grammar of Caballero has been neglected by all.
After a century-long period of silence and total neglect, an important monograph on
Bernardino Gonzlez appeared recently, together with a facsimile edition of the diction-
ary (Intrprete arbico) and his grammar (Eptome) (Lourido Daz 2005), not long after
this paper had been delivered in Hong Kong. When the proofs were almost ready for
publication, I received a copy of this monograph, courtesy of Emilio Ridruejo. Lourido
Daz (2005, I, 2122) had traced seven manuscripts of the dictionary, and six copies
of Eptome de la gramtica rabe made by Bernardino Gonzlez pupils, probably for
their own use. One copy was completed by Blas Francisco de Salamanca in 1704, the
second by Lucas Caballero and Juan de la Encarnacin between 1709 and 1710, the
so-called Tingstadius copy. Two copies were compiled in 1719, one from the El Escorial
inflection and government in arabic 211
studied sporadically by dialectologists describing the urban colloquial

Arabic of Damascus (Lentin 1997), but which also escaped the atten-
tion of researchers working in the field of the history of linguistics. In
the Damascene Franciscan tradition, the authors used a great number
of Arabic grammatical terms, adapted in a hispanicised form, which
substituted, accompanied or sometimes superseded Latin terminology :
examples are: arabicacin (irb), moziones (h arakt) or the partcu-
las nasbantes, chazmantes and charrantes. Obviously, these Franciscan
authors were informed by the Maronite Christians, but in the prologue
of the grammar of Francisco Caes, we read that Spanish Franciscans
not only were familiar with the work of Pedro de Alcal, but also that
they were inspired by north-European grammars and dictionaries of
Arabic, such as the famous grammar written in Latin by Thomas van
Erpen (= Erpenius 15851624) and lexicographical work of Jacob Golius
and the other from the University of Valencia and two further anonymous and undated
manuscripts from the Real Academia de la Historia. Lourido Daz states that all these
works were calques of that of Bernardino Gonzlez (Lourido Daz 2005, I, 13). All the
Latin grammars analysed in this article are also listed in two footnotes by Lourido Daz
(2005, I, 130 and 135) but very little importance is given to the influence of these on the
grammatical tradition of the Franciscans linked to Damascus. After having consulted
all these Latin grammars, we have come to the conclusion that the Spanish grammars
of Arabic completed by Franciscans in Damascus or in Spain were heavily inspired by
the Latin grammars, and in some cases they are Spanish calques, or translations of the
Latin examples. This is particularly evident in the grammatical examples and the use of
literally almost the same orientalising terminology, inherited from the Arabic tradition.
Thanks to the evidence of Lourido Daz study, a direct link between the Franciscans in
Damascus and Spain and the Holy Congregation of the Propaganda Fide and the San
Pietro di Montorio can now be confirmed. Juan de la Encarnacin learned Arabic from
his teacher Lucas Caballero, a pupil of Bernardino Gonzlez. The latter, in his turn, was
a pupil of the Italian Fray Bonaventura da Molazzana, who taught at the San Pietro di
Montorio and who arrived in Seville in 1693. It is known that the grammars and dic-
tionaries used at the San Pietro di Montorio were those of Dominicus Germanus and
Philip Guadagnoli, among others, and it is thus probable that Bernardino Gonzlez had
direct access to the Italian grammatical tradition. It is also important for the purposes
of this article to know that the work of Bernardino Gonzlez was also obligatory in
the curriculum for Spanish and Portuguese missionaries (Lourido Daz 2005, I, 34).
Morevoer, the Portuguese Arabist de Sousa was born in Damascus, so all these gram-
mars are thus linked and use common sources. Germanus of Silesia was educated in the
Holy Land, and was later an Arabic instructor in the El Escorial Monastery in Spain.
Although Lourido Daz monograph is without any doubt extremely important for all
those interested in the bio-bibliographical data related to Bernardino Gonzlez and his
successors, little importance is given in it to the influence of Latin sources and almost
nothing is said about possible Arabic sources, the significance of these works from the
perspective of the history of linguistics or the history of Arabic. Are these grammars to
be considered as key creative productions on the part of Spanish missionaries, or are
they nothing more than a chain in a long tradition? In the future we hope to give an
answer to this question.
212 otto zwartjes
(15961667). Other sources mentioned by Caes are the grammars of

Felipe Guadaoli,3 Antonio de Aguila,4 Agapito de Valle Flammarum,5
and Francisco Gonzalez,6 and the Franciscan grammarian also informs
us that he used a considerable number of manuscripts, without speci-
fying which sources these were. In Renaissance grammars, we find usu-
ally two subdisciplines in syntax. In the Spanish tradition, syntax, often
called construccin, can be subdivided in construccin de rgimen and
construccin de concordancia and Francisco Caes does not form an
exception. In this article we concentrate on the first and we particularly
attempt to analyze how the concept of mil (often translated as gov-
ernor) has been incorporated into this model. The subject of concord
(agreement) has been analysed in Zwartjes (2007). The use of Arabic
grammatical terms in these grammars will be analyzed, and we will con-
centrate on the following questions related to morphosyntax:
Which Arabic grammatical terms are used in these Spanish grammars

particularly in the sections dedicated to inflection and agreement?
What did they mean and why did they use them?
Which sources could they have used?
2. The grammar and dictionary of Arabic of Pedro de Alcal
As we all know, Arabic speaking communities are diglossic. Pedro de

Alcals dictionary and grammar of Granadan Arabic is obviously a
description of colloquial Arabic. His purpose was to teach the ordinary
people (los populares) and not the language of the wise (alfaques).
The aims to compose his dictionary are also slightly different if we
compare them to other missionary dictionaries in the New World,
where missionary composed dictionaries for their own use and for
the novices from the Old World. As we can read in the prologue to his
dictionary, Pedro de Alcal wrote his dictionary not only for the Old
Christians who wanted to learn Arabic, but also for the New Christians.7
3
Schnurrer (1811, 47, no. 72).
4
Antonius Ab Aquila (Schnurrer 1811, 50, no. 78).
5
Schnurrer (1811, 59, no. 85), or Agapito Valle Flemmarum (da Val di Fiemme).
6
I have not identified this author yet, but this could possibly be Bernardino Gonzlez,
as Lourido Daz suggests (2005).
7
Ca assi como los aljamiados (o cristianos viejos) pueden por esta obra saber el arauia,
viniendo del romance al arauia: assi los arauigos (o nueuos cristianos), sabiendo leer la
As has been demonstrated in recent research, mainly from Federico

Corriente (1988), it is true that the language described is predominantly
colloquial,8 but at the same time it is also obvious that there is some
interference between the colloquial and the classical registers. The main
objective of all missionary grammars is practical: the teaching of a
certain language. Nevertheless, missionary works can be predominantly
didactic, showing many paradigms with few explanations and linguistic
theory, while in other works the didactic-pedagogical approach is
much more theoretical. Pedro de Alcal also states that his grammar
is more a practical introduction, not a learned theoretical work. His
approach reveals itself to be universalist, since he observes that there are
definitions and explanations in one language in respect to the expression
of its concepts in its own terms, these are the same in all other languages
regarding the expression of their own concepts. Thus for the same reason
that this name Pedro is a proper noun in Latin, it is also such in Arabic
(Alcal 1505[1883], 2). This observation is characteristic of Renaissance
grammars in general and can be found in many other sources from this
period. As Vivian Law observed:
Questions as what is a proper noun?, what is a verb? how many word
classes are there? what are properties of the conjunction? are as close
to universal as any you are likely to find in a medieval grammar. Such
concerns apply equally well to any European language; indeed, they had
already been transferred from Greek to Latin. There is no inherent reason
why they should not also be asked about Old Irish or Old Icelandic: one
can find proper nouns (for instance) just as easily there as in Latin. (Law
2002, 191)
Obviously, Alcal did not find it necessary to give his pupils definitions
of the parts of speech, since they are the same for all languages. So one
would wonder why the author decided to include Arabic grammatical
terminology, which seems to be in contradiction with his own univer-
salistic approach. A possible explanation is that he did this only for
scientific reasons. He might have introduced them with the purpose to
have a more adequate or sophisticated framework to fit in phenomena
letra castellana: tomando primero el arauia, ligeramente pueden venir en conocimiento

del aljamia). (Pedro de Alcal 1505, prologue, ii v.) See also Cowan (1981, 358). [As
the aljamiados, or old Christians can learn Arabic through this work, coming from
Romance to Arabic, so the Arabs (or new Christians), having mastery of the Castilian
alphabeth, taking first the Arabic, easily can have knowledge of the aljamia].
8
His purpose was hazer vocabulista de la habla comun y usada de la gente deste.
(ibid.) to compose a dictionary of the common speech and used by the people.
214 otto zwartjes
he was not so familiar with. As has been demonstrated by Cowen, this

is only in a few cases a satisfactory explanation. Sometimes Arabic
terms are only used as equivalents or translations of Latin terms. Clear
examples are the names for the cases in the nominal paradigm. Pedro de
Alcal recognizes when dealing with declensions that the Arabic noun
has only one declension, since all Arabic nouns are invariable.9 However,
admitting that there is only one declension, he states that Arabic nouns
have six cases. We reproduce here the table according to Cowan (1981,
359):
TABLE 1 THE SIX CASES ACCORDING TO PEDRO DE ALCAL
Alcal Translation Classical term

mubted nominatiuo mubtada
subject
mudf genitiuo mudf
genitive
maxrr datiuo majrr
dative
maful10 acusatiuo maf l
object
munde vocatiuo mund
vocative
zarf ablatiuo zarf
adverb
As William Cowan observed:10

they are not true cases in either the Arabic or the Latin sense of being an
inflection added to a noun, a fact that Alcal was quite aware of, but are
regarded by him in the same way that modern theoreticians of case gram-
mar regard the syntactic positions in an utterance. Such relations are not
expressed by inflections, since colloquial Arabic has none, but through
abitudines or conocimientos. The markers of the equivalents of these Latin
cases are in fact combinations of prepositions and particles and the defi-
nite article (lil, maal, lal, etc.) (Cowan 1981, 359).
9
La declinacion de todos los nombres arauigos es vna solamente. Porque todos los
nombres arauigos son inuariables (Alcal 1505, capitulo nono).
10
In the original text, a small hamza in superscript is placed on the first vowel u.
The information given by William Cowan, however, is not complete,

since translations from his dictionary are not included here. Although
we do not find all the terms in his dictionary, we do find some of them:
mubted is not only translated by Pedro de Alcal as nominative, but
also as principio de oracion (Corriente 1988, 12), which is not without
importance (see below). zm mudf is rendered as posesivo nonbre.11
The term abitud seems to be inherited from other grammars than
the Latin and Castilian grammars of Antonio de Nebrija (c. 14441522).
The term habitudo is used by Ferdinandus Nepos in his Materies (com-
pleted between 1469 and 1485) and Juan de Pastrana Compendium
grammatica (1462).12
Item per hanc regulam primo iuvenes component per unum casum tantum
sic dicendo: la tierra terra; de la tierra terre, etc.; uel per duos, sic: la tierra
del rey; uel per tres et quatuor et amplius, sic: o leyente la lection a los sco-
lares en el general de las escuelas componitur o legens lectionem scolaribus
in generali scolarum, dando cuilibet casui propriam habitudinem, inter-
rogando cuius casus, numeri et cuius declinationis hoc principio. (Codoer
2000, 90).
However, it is not so clear what Pedro de Alcals exactly means with
the term abitud (pl. abitudines), since he uses it as a synonym of the
definite article (Nota que porque ay algunas abitudines en cada vno delos
casos que en alguna manera parescen preposiciones, porque se preponen
alos casos, avn que en verdad no lo sean, mas son articulos) (Alcal
1505[1971], 26). In another section, Alcal uses the term as a synonym
of preposicion, since according to his observations, the maxrr case
(see below) has the four abitudines la, lal, li, lil, whereas the term
abitud is not used for ba, bal, bi, bil, fa, etc. which are described in
the paragraph on the zarf case (Alcal 1505[1971], 27). Here they are
just called prepositions and not abitudines and there is no explanation
11
Other terms not mentioned by Cowen, are jezme, translated as consonante (letra
mazjun [sic] (Corriente (1988, 34), and mdde, translated as acento (Corriente 1988,
189).
12
With this rule, the novices first build constituents with only one case [= head of
the NP], saying la tierra terra, de la tierra, terre, etc., or with two cases [= head
+ complement of the NP], as la tierra del rey, or with four and so on, as: reading the
lesson in general for the scholars of the schools, assigning the appropriate habitudo
[= grammatical form] to whatever case [= grammatical function]. The term habitudo
casualis is also used by Nepos in relation with government: Haec enim regula maxima
est in construendo [. . .] quia talem casum regit dictio qualis fuerit habitudo casualis.
(ibid.).
216 otto zwartjes
why he used these different terms. In other sections, Alcal uses abitud
as synonym of conocimientos (see also Zwartjes 1993, and 1994).
In Arabic grammatical theory a nominal sentence can be divided into
topic for which mubtada (lit. what is begun with) is normally used,
and comment (xabar), or predicate (Owens 1988, 32), or according to
Sbawayhis terminology also called mabny al-l-mubtada (what is built
upon what is begun with) (Owens 1990, 45). Mudf is the word Pedro
de Alcal uses for the genitive, which in Arabic tradition means literally
what is added, i.e. possessed (Owens 1988, 34; 1990, 104). Majrr from
the same root as jarr (see below), means pulling, or governing the -i
inflected form. Pedro de Alcal follows in his sections about the prepo-
sitions the Latin system and tries to apply Arabic terms to them, without
realizing that in the dialect he describes, case-endings are not used, and
without realizing that in classical Arabic nominal declension, there are
three inflectional vowels, the -u, the -a, whereas for the verbal inflection
the three vowels -u, -a and (zero ending) can be distinguished. Pedro
de Alcal did not take the Arabic inflectional endings as starting point,
but the Latin prepositions in alphabetic order: prepositions + accusa-
tive, prepositions + ablative, etc. and at the same time he translated the
names of these Latin terminology into Arabic:
Capitulo XXXII. De las preposiciones.
Hallamos en el Arauia todas las preposiciones que en la gramatica [latina],
y ayuntadas a essos mesmos casos, que son maful y darf (que son acusatiuo
y ablatiuo), y son las del acusativo las siguientes:
A ad apud circa circa ante longe ante [. . .]
A al aynd carib qued acbal bad cudim [. . .]
Las preposiciones del darf (que es ablativo) son las siguientes, conuiene
saber:
Con con con con en en en en
Ba bal bi bil fa fal fi fil . . .
(Alcal 1505[1883], 26).
Chapter XXXII. About the prepositions.
We find in the Arabic language all the same prepositions as in Latin gram-
mar, and they are combined with the same cases, which are mafl and
zarf (which are accusative and ablative), and those which can be combined
with the accusative are the following:
A. ad apud circa circa ante longe ante [. . .]
A al aynd carib qued acbal bad cudim [. . .]
The prepositions of the darf (which is ablative) are the following:

con con con con en en en en
ba bal bi bil fa fal fi fil . . .
As we can see, Pedro de Alcal did not separate the prepositions from
the definite article (al).
Mafl is the term used for object. Sbawayhi distinguishes five sub-
types: mafl bihi direct object, mafl fhi locative object, mafl maahu
accompaniment object, mafl lahu reason object, and mafl minhu
object from it. (Owens 1990, 160). For the locative object, instead of
mafl fhi the term zarf is also used (Owens 1990, 51,141151), which
is the term Alcal uses here for the ablative. Mund is the direct trans-
lation of vocative. According to Cowan (1981, 360), Alcal was appar-
ently trying to make unfamiliar material intelligible to his audience, but
at the same time to avoid a direct equation with the Latin categories. In
fact, the first might be true, but we must be aware that in his paradigm of
the case-system, we do find an equation with Latin cases, and we never
find any traces of the four traditional Arabic inflectional endings, -zero,
-u, -a, and i and never the original Arabic names for these inflectional
endings are introduced here.
Other technical grammatical terms in Alcals grammar are: amir
(conocimiento;13 cl. Ar. damr conjunct pronoun), temiz (cl. Ar. tamyz
accusative of specification),14 amr (imperatiuo; cl. Ar. amr impera-
tive), xucla seal; cl. Ar. akl(a) orthographic sign:15
Es otrosi de notar, que los Arauigos non tienen letras vocales como los Lati-
nos, mas tienen en lugar dellas ciertas seales, que ellos dizen xclas, con
las quales y con todos los caracteres suso dichos leen y escriuen lo necessario
(Alcal 1505[1883], 4).
It has to be observed that the Arabs do not have the letters for the vowels
as the Latins, but instead of them, they have certain signs, which they call
xuclas, and with all the above-mentioned characters they read and write
anything which is necessary.
13
Cf. Zwartjes (1994).
14
Tem is also translated as conocimiento in his dictionary (Corriente 1988, 197).
15
In his dictionary Alcal translates the word alma as (signo por) seal (Corriente
1988, 140). The seal is also used in his grammar for the article (seal de demostracin),
so this term could be the technical grammatical term, but as happens often in these
grammars, it is not always possible to distinguish between language and metalanguage.
218 otto zwartjes
The origins of the mnemonic vowel terms with and without tanwn
(the nunated forms) which expresses indefiniteness in Arabic are
unknown to me and this could be a local teaching method, not recorded
elsewhere:
TABLE 2 THE VOWELS AND THE NUNATED FORMS

ACCORDING TO PEDRO DE ALCAL (1505[1883], 4)16
minib fath a a
minib kasra i
minib damma u
minenbn fath atn an
mininbn kasratn in
minunbn dammatn un
minb sukn16
The remaining Arabic terms analyzed by Cowan are alif cequin (alif skin
silent alif ), and in his dictionary we find iarab (oracin de gramtica;
clas. Ar. irb (the inflectional endings , a, i, u (see below)17 and harf
(letra; Cl. Ar. h arf), which are not analyzed at all in the grammatical
treatise.
Summarizing the preceding paragraphs, we can conclude that in most
cases Pedro de Alcal could easily use Latin terminology, such as the
names for the cases or the imperative. The use of an exo-grammatical
term damr for the conjunct pronoun does make sense since traditional
grammar did not have precise equivalents from contemporary sources
yet. Probably, Pedro de Alcal understood very well that the conjunct
pronoun in Arabic can be used differently from the Spanish pronouns;
they can also be affixed to prepositions and nouns, for instance and
that explains probably the reason why he used the Arabic term. In the
remaining cases, Arabic terms are used for mnemonic or pedagogical-
didactical reasons. It is questionable if these terms made his teaching
16
The terms fath a, kasra and damma are not found in his grammar which gives us
the impression that Pedro de Alcal did not know them.
17
Again, we find more information in the dictionary, neglected by Cowan: iarb is
translated as declinacion de palabras (Corriente 1988, 134), which is not unimportant
because the author avoids the term noun here, since irb is used for nouns and verbs
as well.
method more attractive or more easily accessible to his pupils. 18 Seen

through our modern eyes, the grammar of Pedro de Alcal was a real
novelty, a pioneer work. His transcription system of the Arabic alphabet
is the first in Europe, although his knowledge of classical Arabic was evi-
dently limited; there are quite a few inconsistencies while using his own
transliteration system. It is also obvious that Pedro de Alcal did not
have any knowledge of Arabic grammatical theory and his Greco-Latin
approach is particularly visible in the lacking of insights in the analysis
of derivations. As had been observed by Fck [Pedro de Alcal] erkennt
nicht die Bedeutung des Wurzelbegriffs, so dass dem Leser der arabische
Formenbau in der Nominal- wie in der Verbalbildung undurchsichtig
bleibt (1955, 33). Nevertheless, as has been stressed by several scholars
at the conferences on missionary linguistics,19 it is easy to point at the
shortcomings of these grammars and dictionaries measured by the stan-
dards of our own time, but when studied and analyzed in its own his-
torical and cultural context, there is no doubt that there is still a wealth
of material to be studied. As observed by Smith Stark certain precon-
ceptions about pre-modern descriptive work have resulted in its neglect
among those in the Western tradition (2005, 4). Monographs and stud-
ies on the description of Arabic by Spanish missionaries are still non-ex-
istent. Positive evaluations are scarce when the grammars are concerned,
whereas the majority agrees that the lexicographical work of Pedro de
Alcal and others have been of great value. Dannenfeldts observation,
according to whom Both of these [Alcals] works are based on solid
18
Missionaries usually emphasised that the language they were learning was easy
to learn, although others label the language under description as difficult. They tried to
use the most transparent and less obscure paradigms and explanations. The reason to
re-write existing grammars was almost always because predecessors were too obscure.
Probably, the grammar of Pedro de Alcal could be perfectly understandable without
the use of Arabic grammatical terminology. The same could be said of the use of Hebrew
posodical-grammatical terms by Oyanguren de Santa Ins in his grammar of Tagalog
(1742, 208209), such as milehal (stress on the penultimate syllable, instead of the usual
ultimate syllable, in connection with stress assignment), athnach (semicolon or pause)
and metheg (one type of the several secondary accents, avoiding the loss of vowels in
pronunciation, or a sign, pointing a vowel, which usually would be reduced to schwa
but which is to be fully pronounced in this particular place). The terms atnach and
meteg are both so-called cantillation marks in the Hebrew Bible from Masoretic times.
Did the pupils of Oyanguren de Santa Ins know Hebrew, or is this pure pedantry or
snobbism?
19
The International Conference on Missionary Linguistics took place in Oslo (2003),
So Paulo (2004), Hong Kong and Macau (2005), Valladolid (2006), Mrida (Yucatn,
2007) and the sixth will be organized in vora, Portugal.
220 otto zwartjes
philological methods and are evidences of humanist learning in Spain

(1955, 33),when compared to Fcks (1955, 33)is without any doubt
an exception. Although Pedro de Alcals grammar served as the model
for the Gramtica arbigo-castellana (still unpublished) composed by
the Hiernoymite Patricio Jos de la Torre (17601819),20 and although
we have some evidence the Maronite Joseph Simon Assemani of Syr-
ian origin (16871768)21 quoted from his grammar, Alcals dictionary
was particularly widely known and used throughout Europe. Johannes
Gabriel Sparvenfeldius possessed probably Alcals catechism, which has
been appended to his grammar and his dictionary already in 1706 but
in that period it had become already a rare book (Schnurrer 1811, 16).22
If the original work was not longer available, scholars used the original
editions or the re-edition from 1776 from Patricio Jos de la Torre with
transcriptions into Arabic characters, which was the base for the Sup-
plment of the Dutch Orientalist Reinhart Dozy (18201883) (Monroe
1970, 38). One of the main purposes was to publish the work of Pedro
de Alcal in Arabic script, something that was completely unnecessary
according to Pedro de Alcal. Although we find an entire page with the
Arabic alphabet in his grammar, an observation below it tells us all
characters can be substituted by Latin or Castilian letters:
Estos son los caracteres y nonbres de las letras arauigas, las quales todas se
pueden suplir con nuestras letras Latinas o castellanas, de manera, que para
la comun algarauia no ay necessidad de las saber ni conocer todas, mas
solamente quarto conuiene saber kha, dil , te, ay, cuyos sones no tenemos en
nuestro ABC latino. (Alcal 1505[1883], 34)
These are the characters and names of the Arabic letters, which all can
be substituted by our Latin or Castilian letters, so that there is no need
to learn or to know them all for the common speech, but only four [are
necessary], namely the kha, dil , t e, ay, whose sounds we do not have in
our Latin ABC.
We have seen that for didactic reasons, Pedro de Alcal used these
mnemonic words, which are not derived from canonical grammatical
works of the Arabic tradition. Other grammarians found a different
20
Schnurrer (1811, 88, no. 128). BICRES III, no. 84.
21
Schnurrer (1811, 16) observes that Assemani quotes from his grammar ex gram-
matical recitat, but also adds that he actually used material from Alcals dictionary
(non Grammaticae sunt, sed Vocabularii, ibid.).
22
Ioan. Gabr. Sparvenfeld. 1706. Catalogus Centuriae Librorum Rarissimorum Manu-
script.& partim Impressorum, Arabicorum, Persicorum, Turcicorum, Graecorum, Latino-
rum, &c. Upsala: John Henry Werner.
solution. One of the most remarkable solutions can be found in the

Fabrica Arabica (1640) of Dominicus Germanus of Silesia (15881670).
In this case, we have translations of the Latin names of the cases into
Arabic. In the following table we see these literal:
TABLE 3 THE CASES ACCORDING TO DOMINICUS

GERMANUS FABRICA ARABIC (1640)23
al-mutasamm Nominativus

al-mutawallid Genitivus
al-mustat Dativus
al-mutak Accusativus
al-mund Vocativus
al-mustaqti Ablativus23
3. The grammars of Lucas Caballero, Juan de la Encarnacin, and

Francisco Caes: Grammars at the crossroads of two systems?
3.1 Introduction
After the foundation of a great number of missions in the East, the
importance of the Arabic language for preaching the Christian faith
continued to increase. Paul V in a papal bull dated 1610 had commanded
the various religious orders to teach Oriental languages in their colleges.
In the early 16th century, Arabic was taught in Seville at the Colegio
trilinge.24 The Franciscans decided to found colleges in Salamanca, Alcal,
Paris and Toulouse for the teaching of Arabic, Greek and Hebrew (Monroe
1970, 26). According to Monroe, Bernardino Gonzlez (c. 16651735)
composed an Arabic dictionary in Seville, which was completed by
Franciscans in Jerusalem in 1709 (Monroe ibid.), an unpublished work.
Jos de Len began to compile a new dictionary of Arabic and Bernardino
Gonzlez was sent to Damascus in order to complete his work.25 As Monroe
23
From the verbs samm to denominate, walada to give birth, at to give, ak to
complain, nad to call and qatta to cut off, to disjoin, the tenth derived form means to
deduct, which seems to be an approximate translation of auferre.
24
In this short account, there will be no space to summarize the study of languages
during the Middle Ages in Muslim Spain.
25
Franciscans had already arrived in 1233 in Damascus. Propagating the Christian
faith was not permitted by Sultan Malik al-Araf, but they took care of Europeans who
settled there. After several cases of martyrdom and imprisonment, the Cadi of the city
222 otto zwartjes
observed, the work was the most complete ever to have been composed
in Spanish after that of Pedro de Alcal. [. . .] The work of Gonzlez and
his collaborators was lost until Asn Palacios (1901) came across a copy
containing additions up to the year 1727 made by the friars of Damascus
(Monroe 1970, 27). One of the successors of Bernardino Gonzlez was
Francisco Caes who settled in Damascus at the Spanish Franciscan
College in 1757 (Monroe 1970, 28). Caes grammar of Arabic has been
printed in two different editions, a first in 1775 (Madrid, Don Antonio
Perez de Soto, and a second in 1776 (Schnurrer 1811, no. 113, BICRES
959 and 971), entitled Gramatica arabigo-espaola, vulgar y literal. Con
un diccionario arbigo-espaol, en que se ponen las voces mas usuales
para una conversacin familiar, con el texto de la Doctrina Cristiana en
el idioma arabigo.
I came across another copy from the same Franciscan tradition, com-
posed by Lucas Caballero and Juan de la Encarnacin as we can read in
the colophon, which escaped the attention of scholars who have worked
in this field. The manuscript has been identified by Magnus and Aare
Mrner in his Spanien i svenska arkiv. The title of this manuscript is Com-
pendio de los rudimentos y gramtica rabe en que se da notizia de la len-
gua verncula y Vulgar y algunas reglas de la literal Iustamente, 1709, and
in the colophon 1710 (another author, Juan de la Encarnacin, finished
the text San Diego, Seville). The work is based on Bernardino Gonzlez
as we can read in the title, and Lucas Caballero, lector actual Arabo en
el Colegio de Damasco composed (recopilado) this manuscript, which
has been donated by Johan Adam Tingstadius (17481827),26 bishop of
Strngns, Sweden, from 1803, to the Rogge library, which has belonged
administratively to the Royal Library of Stockholm since 1968.
As the titles of the grammars of Caes and Caballero demonstrate, the
language under description is not only classical Arabic, but the urban
dialect of Damascus. Apart from Alcals grammar of colloquial Ara-
bic of Granada, European scholars usually did not pay much attention
to lower registers, so the linguistic works of these Franciscans work-
granted the privilege of being able to open a public chapel and in 1668 the Francis-
cans established themselves in a Maronite church, which they left in 1719 when they
acquired a new church in the Christian Quarter of Bb Tma. The foundation of the
college where Arabic was taught dates from this period.
26
Tingstadius was a professor in Oriental languages at Uppsala. He published, for
instance in 1770, a treatise entitled Dissertatio philologica de natura et indole linguarum
orientalium communi (Uppsala: Johan Edman) and in 1794 his Dictiones arabicae ex
carmine Tograi, hebraismum biblicum illustrantes. Uppsala: Johan Edman.
ing in Damascus are of great importance. However, they were not the
only grammarians who described non-Classical registers. Antonio ab
Aquilas grammar published in 1650 is not only a grammar of classical
Arabic (ad grammaticae doctrinalis intelligentiam) but also colloquial
Arabic (ad vulgaris dumtaxat idiomatic), probably the reason why he
called the grammar Arabicae linguae novae et methodicae institutiones.
Dominicus Germanus (Germanus of Silesia; 15881670)27 composed a
dictionary in 1636 with the title Fabrica overo Dittionario della lingua
volgare arabica et italiana, copioso di voci e locutioni, con osservare la
frase delluna e dellaltra lingua (Roma. Nella stampa della Sac. Congr. De
Propag. Fide) followed three years later by his Fabrica linguae Arabicae
cum interpretatione latina et italica, accommodata ad usum linguae uul-
garis et scripturalis (Roma. Typis Sac. Congreg. De Prop. Fid.).
In 1800 a work has been completed by Franciscus de Dombay (1758
1810) with the title Grammatica linguae Mauro-Arabicae juxta vernaculi
idiomatis usum, accessit vocabularium Latino-Arabicum (Vindobonae:
apud Camesina)28 but, according to Schnurrer, this title is misleading;
although this grammar describes the common speech of the people in
the Maghreb (Arabicus sermo in Mauritania quo vulgus uti solet), he
observes that all words are good Arabic (non sunt vulgari idiomati pro-
pria, sed omnia bene Arabica) (Schnurrer 1811, 95).29
27
Germanicus was a teacher of Arabic at the mission school St. Peter in Montorio,
Rome. He assisted with the preparation of the Arab Bible, he published dictionaries,
and commentaries on the Qurn. He was teacher and translator at the court of Philip
IV of Spain.
28
I have not been able to consult this grammar yet.
29
Although this is not the aim of this paper, I wish to show just a few elements from
these sources that are important records of colloquial Damascene Arabic from the
beginning of 1700. Particularly the word lists are full of colloquialisms, but also the
grammar of Lucas Caballero has many colloquial elements, to mention a few: omis-
sion of vowels: muqatla instead of muqtala, the use of the -u- vowel as a prefix for the
imperfect tense (64), b- future suffix (p. 24), which is colloquial (in Egypt it is the present
tense). However, Caballero is not always consistent, we find both faaltu as faaltum (2
person pl.m.), the use of - instead of -na for the second person feminine singular in
taf al (= Haztu fem.) (72). It is remarkable that sometimes we find even hybrid forms,
such as antum faaltu. However, we find also classical elements, such as the use of the
feminine plural in the verbal paradigms, which is not used in colloquial urban speech
today. It is also significant that the order of the persons singular in the verbal paradigms
is not the traditional one 73v72r. Caes has 3 (masc. Sing, 3 fem. Sing, 3. plur. 2 masc.
Sing. 2 fem. Sing. 2 plur. 1 sing 1 plur. Instead of 3,3,2,2,1 (sing.), 3, 3, 2, 2, 1. I am grate-
ful to Manfred Woidich for his comments on this footnote.
224 otto zwartjes
Grammars of Classical Greek were available throughout Europe, but

Spanish priests were engaged in the writing of vernacular Greek too, as
we can read in the prologue of Francisco Caes (1775, s.n.).30
In the following paragraphs we will treat the use of Arabic terminol-
ogy in both grammars separately.
3.2 Semantic and inflectional vowels and the concept of amal

The questions to be answered now are: which Arabic grammatical
terms are used, particularly in the sections dedicated to inflection and
government, what did they mean and why did Caballero and Caes use
them?
Lets start with the vowel system, according to the paradigm of Cabal-
lero are:
TABLE 4A THE VOWEL SYSTEM ACCORDING

TO CABALLERO31
bu Damma31 O que inclina a V

ba Fatha A que inclina a E
bi Kesra E que inclina a I
Unlike Pedro de Alcal, Caballero gives also in an appendix of this

Chapter the Arabic technical terms for these vowels when they are used
as inflectional endings, accompanied with translations into Spanish:
Los Gramaticos a estas mociones dan otros nonbres conbiene a saber que
significan, eleuacion, ereccion, y contraccion: al-raf, al-nasb, wa-l-jarr y
estas mociones duplicadas llaman tanun (tanwn) esto es nunacion que es
30
It must be emphasized that missionary sources, often written in Spanish, are in
many cases the only existent sources which can give us information of vernaculars once
spoken in early ages of languages of which we only have more detailed information con-
cerning the literary or classical register. Priests understood very well that in China the
teaching of Mandarin was not so useful in regions were other dialects were spoken.
Grammars of modern Greek circulated since 1638, the Grammatica lingu grc
vulgaris was printed by Simon Portius. The first Spanish grammar of modern Greek
was composed by Pedro Fuentes, as we can read in the following quotation from the
same prologue: Por lo tocante la lengua griega ha impreso su gramatica vulgar Fr.
Pedro Fuentes observante, que residi en el Seminario de Nicosa en Chipre, y ahora est
imprimiendo la gramatica literal.
31
In their tables of the vowel system, both authors also give the names of the vowels
in Arabic script, not reproduced here.
lo mismo que decir addicion de un sonido de nun o n y biene a sonar on, an,
en vg. Racholon, Racholan, Racholen, Homo, Hominem, Hominis [. . .] pero
esto ueras mas claro en el tratado del nonbre (77 r.).
The grammarians give other names to these motions [h arakt], namely
the names which signify elevation, raising, and contraction, ar-raf ,
an-nasb, wa-l-jarr, and these motions, when duplicated, are called tanwn,
which is nunation which is the same as adding the sound of a nn, or -n,
which approximately sounds as on, an, en vg. Racholon, Racholan, Racho-
len, Man (nom.), man, (acc.), man (gen.), but this you will see more clearly
in the chapter about the noun.
Caes table resembles that of Caballero, although there are some
differences. The order of the vowels is different, Spanish translations of
the Arabic vowel terms are given, and instead of de verb inclinar we
find declinar:
TABLE 4B THE INFLECTIONAL VOWELS ACCORDING TO CAES

(1775, f. 7)
ba Fatha, Apercion A que declina e, y a clara. Ba, Be.

bi Kesra, Fraccion E que declina en e, i claro.
bu Domma, Colleccion O que declina en o, y u. como Bo, Bu.
Also here, we find almost the same observation as above, which

demonstrates that the works are closely related to each other and that
they probably derived from a common source (or sources):
Los gramaticos las tres mociones dichas les dan otros nombres, es saber:
al Fatha le llaman Nasbo, esto es ereccion, al Kesra Charro, contraccion,
y al Domma Rafo, elevacion. Suelen tambien duplicar las dichas mociones

de esta suerte ( ) ( ), y entonces les dan el nombre tanuin, esto es, nunacion,
nun vocal, que viene sonar an, en, on. (Caes 1775, ff. 89).
The grammarians give other names to these three above-mentioned
motions, namely: they call the Fatha Nasbo which means erection, they
call Kesra Charro which means contraction, and Domma Rafo elevation.
They are used to dublicate the above-mentioned motions in this way: ( )

( ), and they give them the name tanwn which is nunation, or vocal nn
which sounds rougly like: an, en, on.
In Chapter 4, both authors deal with the cinco signos:
(1) Secun (quietud), which has according to both the synonym chezm
(caballero), or chiasmo (Caes), (jazm) which Caballero translates
as anputacion and Caes as corte;
226 otto zwartjes
(2) Tex did (duplicacion, corroboracion);

(3) Maddo (extension produccion);
(4) Vasalo (union, conjuncion); and
(5) Hamza (punzamiento).
A chapter dealing with how to read Arabic without vowel signs was
appended by Caballero, not present in the grammar of Caes. Although
there are differences, both grammars are from the same tradition and
both authors (re-)formulated probably the lost version of Bernardino
Gonzlez, or quoted directly from other sources, such as Agapito Valle
Flemmarum, who almost has the same definitions as Caess, although
the vocalizations of the Arabic terms and the order is slightly different:
TABLE 4C THE INFLECTIONAL VOWELS ACCORDING TO

AGAPITO VALLE (1687, f. 79V.) 32
bu Dzhamma, collectio o declinans ad u, & u claru

ba Fathha, aperitio a declinans ad e, & a claru
bi Kesra, fractio e declinans ad i, & i claru
The Tratado III deals with the noun. Here we find sometimes some
parallels with the grammar of Pedro de Alcal, particularly since
the term seal has been used in both sources, or notificacin for the
article33 and the abitudines of Pedro de Alcal resemble much the
seales del nombre. Since colloquial Arabic nouns are not inflected
according to cases, we find in the grammars of Alcal, Caballero and
Caes equivalents of the Greco-Latin cases for didactic reasons:
Los arabes aunque en la lengua vulgar reconocen tres numeros en el nombre,
es saber: singular, dual, y plural no conocen distinction de casos. Y asi el
nombre en qualquier caso termina con una misma voz. (Caes 1775, 59).
The Arabs, although they recognize in the colloquial speech three num-
bers in the noun, being, singular, dual and plural, they do not know the
distinction in cases. And thats why the noun ends with the same sound
in whichever case.
32
The vocalization of Martelottus is slightly different: Dzhammon, Phathhon and
Kafron.
33
Pedro de Alcal uses seal de demostracin. See Zwartjes (1992).
Elsewhere in his grammar, i.e. after the Numeros cardinales and

Numeros ordinales we find a comparable description:
Los Arabos en la lengua Vernacular o Vulgar no tienen terminacion de casos
en el nombre assi como los Castellanos (16v.).
In the vernacular or vulgar speech, the Arabs do not have the ending of
cases in the noun as the Castilians have.
The case system according to Caballero is as follows:
TABLE 5 THE INFLECTIONAL VOWELS ACCORDING TO

CABALLERO (1709, 14r.)
Nominativo Vocatiuo raf un Rajulun Vir, o vir
(this column in Arabic script)
Acusativo Vocatiuo nasbun Rajulan Virum o Virum
Genitiuo Viri
Datiuo jarr un Rajulin Viro
Ablatiuo A viro
What did these terms mean in the Arabic grammatical tradition? Before
Sbawayhi, no distinctions have been made between the vowels which
are used in classical Arabic for the declensional endings, and the other
vowels, for instance: there was no disctinction between both vowels i
in the genitive al-kitb-i. An important novelty of Sbawayhi is that
he distinguished the first i that is non-declensional, from the final i
which is declensional (Versteegh 1997, 19).
When we analyze Arabic terminology in our 18th century grammars,
the sections about the particles are even more interesting. Caes and
Caballero use both a metalanguage inherited from the Arabic tradition
that had already been developed by Sbawayhi:
These endings follow eight courses: accusative (nasb), genitive ( jarr),
nominative (raf ), apocopate ( jazm), a-vowel ( fath ), i-vowel (kasr),
u-vowel (damm), zero-vowel (waqf ). (Translation by Versteegh 1997, 36).
The main distinction is whether a certain vowel is declensional or not.
The Arabic terms used by Caballero have all to be related to the Arabic
term amal that generally is translated as governance, or dependence/
dependency, which resembles the 20th century theory of government
and binding. As explained by Owens:
228 otto zwartjes
The governor is said to govern the governed in some case or mode form
(irb):
For the nouns these forms are: u nominative (raf ); a accusative (nasb); i
genitive ( jarr);
For the verb only the imperfect verb shows mode inflection: u indicative
(raf ); a subjunctive (nasb); jussive ( jazm).(Owens 1988, 39).
Carter demonstrated in several publications (1991, 1993, 1994)dealing
with different grammatical theoriesthat the translation of technical
terms are in many cases inexact, problematic, anachronistic and can lead
to misconceptions.34 Carter argued that the basic meaning of the concept
of amal is different, stressing that the interpretation of Wei: amal
und regere: Es ist nicht recht verstndlich, wie man hier schwanken
kann. amala f kann ja doch gar nicht heien Gouverneur sein ber.
[. . .]. amala f ist ttig sein, arbeiten. The translation governor is
according to Wei an example of wishful thinking (Wunsch der Vater
des Gedankes), because the concept of governance was already widely
used in the Greco-Latin framework. A translation that brings us closer
to the original meaning an etwas arbeiten, auf etwas einwirken (Wei
1910, 384).
Particles (h urf )35 can also be defined as governors, since they can
govern cases. The subclasses of some particles can be defined in terms
of dependency, i.e. which nominal or verbal ending they govern. If a
particle (h arf ) governs a genitive, such particles are called h urf jarr
(Owens 1988, 10), for instance bi (by means of ), or min (from) etc.
In both the Compendio delos Rudimentos y Gramatica Araba of Lucas
Cauallero, as the Gramatica arabigoespaola, vulgar, y literal of Fran-
cisco Caes we find hispanicised forms of these subclasses of particles.36
At the beginning of the chapter dealing with the parts of speech, both
Caballero and Caes give us the tripartite division of the parts of speech,
which is an Arabic division, according to Caes (Pero los arabes las
34
The same happened when Romans translated Greek grammar. An illustrative
example is the term casus accusativus which is a wrong translation of the Greek term
ptosis aitiatike. It is not the anklagender Fall, but das von der Handlung Betroffene,
dasjenige, dem etwas gechieht. (L. Lersch: Die Sprachphilosophie der Alten, Bonn 1838
1841. vol. 2, 186, quoted in Carter 1993, 131).
35
H arf does not only mean particle, since it has in fact much more meanings, such
as edge, letter, sound, word. See for a detailed overview of the most important meanings
the first Appendix of Owens (1990, 245248).
36
We have not been able to consult a dictionary, which is particularly devoted to the
particles, the Diccionario de partculas rabes [s.a.], composed by Mariano Rizzi y Fran-
ceschi (18th century; BICRES III, no. 71).
[partes de la oracion] reducen a tres, que son: verbo, nombre, y particula

(h arf ). As a matter of fact, the authors follow this tripartite division, but
in the second chapter of the 4th treatise, which deals with the particles,
the author prefers to follow the Latin model:
Capitulo II. De las particulas separadas.
Muchos son los generos, que hay de estas particulas, y para evitar confusion,
las dividiremos al modo de los latinos, en adverbios, conjunciones, pre-
posiciones, y interjeciones. (71).
Chapter II. About the separate [= not suffixed] articles.
The subcategories of these particles are many, and in order to avoid confu-
sion, we divide them, according to the Latin manner, into adverbs, con-
junctions, prepositions and interjections.
In the pages dealing with classical Arabic, a separate chapter following
the final chapter in syntax, Francisco Caes again prefers the Arabic
model. In the following quotation he deals with them after the chapter
on syntax describing classical Arabic:
Aqui es preciso advertir, que asi como el nombre se declina por la variacion
de las terminaciones, que tiene en los casos, como adelante se ver; igual-
mente el verbo en el modsareo se conjuga de tres maneras: por variarse
su terminacion en las personas. Esta variacion proviene, de que los arabes
anteponen al verbo en el modsareo dos generos de particulas, que se llaman
nasbantes y chazmantes. Se llaman nasbantes del verbo nasaba puso,
plant, fix, porque en fuerza de ellas el domma de la tercera radical del
verbo se pierde, y convertido en la mocion fatha, la fixan, y plantan sobre la
dicha radical, sin que padezca mutacion. Las chazmantes se llaman asi del
verbo jazama sec,37 cort, porque estas particulas le cortan la mocion la
tercera radical; y poniendole el signo secun, la dexan quiescente, liquida.
(Caes 1775, 109110).
Here, it is necessary to observe, that as the noun declines through the varia-
tion of its endings which they have in the cases, as we shall see below, so
on the same manner the verb in the mudri38 conjugates in three ways: by
varying the endings according to persons. This variation results from the
fact that the Arabs put before the verb in the mudri two classes of parti-
37
Probably seccion, since secar means to dry.
38
Usually translated as imperfect, literally the resembling verb (Baalbaki 2004,
XIII, 23), because they resemble the nouns, since both share the same declensional
vowels (irb) -u (subject and indicative mood; and -a which is direct object and
subjunctive mood). The nouns do not have zero endings (apocopate or jussive mood
in the verbal system), whereas the nouns have the -i ending (genitive), which is not
present in the verbal paradigm.
230 otto zwartjes
cles, which they call nasbantes and chazmantes [= which co-occur with
the nasb (a ending) and those which co-occur with the jazm ( ending)].
They are called nasbantes from the verb nasaba, to put up, to plant, to fix,
because through their force, the domma [damma] of the third radical of
the verb is cut off and converted into the fath a [a ending/motion]; they
fix this radical and settles down on it, while it does not suffer any change.
The chasmantes [those which co-occur with the zero-ending] are called so
from the verb jazama, to cut off, to truncate, because these particles cut off
the vowel [motion] from the third radical, and put on this the sign sukn
[motionless, vowelless] and leave them silent or liquid.
In Chapter IV, dealing with the noun, Caes gives us three classes of
particles, charrantes, chazmantes y nasbantes:
Particulas charrantes son unas preposiciones, que antepuestas al nombre le
colocan en el caso charro, genitivo (139) [. . .] Aqui se debe advertir, como
en arabe lo mismo es de decir particula nasbante, que en latin preposicion
de acusativo, y asi antecediendo al nombre le colocan en dicho caso, [. . .]
Estas particulas manera de los verbos, admiten afixs, y rigen los nombres,
colocando el sujeto en nasbo, acusativo, y el predicado en rafeo, nomi-
nativo, de suerte, que se viene hacer una permutacion del nominativo con
el acusativo . . . (143)
the particles which are charrantes are some prepositions, which placed
before the noun, put it in the jarr [i] ending, or genitive. Here it must be
observed, that as in Arabic it is the same to call them particular nasbante,
as in Latin, prepositions which combine with the accusative, and as such
they are placed before the noun, they put it in this so-called case. [. . .]
These particles, when combined with verbs, allow affixes, and govern the
nouns, placing the subject in nasb, or accusative and the predicate in raf ,
or nominative, so that there is a permutation from the nominative to the
accusative.
Caballero uses the same classification and terminology, although his
hispanicised forms are slightly different: particulas charrantes, nas-
suantes, and chesmantes. Although in Caes description, only classical
Arabic is commented on, Caballero also dedicates a paragraph on
the particulas Chesmantes, ojo, Vulgares (particles which correspond
with the optative Utinam and other Latin forms, such as quare, quia,
quoniam, etc.).
If we compare the use of Arabic terminology in the works of Caba-
llero and Caes with those used by Erpenius, we can conclude that the
latter did not use Arabic terms so extensively. Worthy of mention are for
instance: Fatha, damma, kesra in the first book De elementis . . .. We find
a Latinized form of the term jazm: post gjezman constanter manent
(1620, 20), also used as the verb Gjezmare, gjezmant, or in the passive
form gjezmatur (47), verbum hamzatum (70). The term motiones is not
used as a translation of h arakt but it for the change a noun undergoes
if the feminine ending is added to the masculine form.39 Nunnatio is
used as well (141), but the Arabic terminology for inflectional endings
as used by the Spanish missionaries is not recorded.
3.3 Possible sources

The use of non-Western metalanguage in itself has its own tradition. Not
only Pedro de Alcal used non-Western terminology, but in Northern
Europe we see also that Hebrew grammars used Hebrew terminology
in an adapted form.40 Which sources could the Franciscans have used?
In the prologue of the grammar, Caes informs us that he completed
an eclectic grammar in agreement with the taste of everyone (que
sea del gusto de todos), using the most useful aspects (he procurado
aprovecharme de lo bueno que en ellas he visto ibid.) of earlier sources
and adding material from his own 16 years long experience:
He procurado con el mayor cuidado y desvelo leer, y releer para el ajuste de
esta, las gramaticas de Fr. Pedro de Alcal, de Fr. Felipe Guadaoli, de Toms
Erpenio, de Fr. Antonio de Aguila, de Fr. Agapito de Valle flammarum, de Fr.
Francisco Gonzalez, & c. Asimismo me he valido de un considerable numero
de manu-scritos, que me han franqueado gustosos algunos aficionados la
lengua arabe. Finalmente he aprovechado lo que me ense el estudio, y la
experiencia por espacio de diez y seis aos, que estube predicando, y confe-
sando en arabe en las misiones del Asia (ibid.).
I have taken the greatest care to do my best to read and read over again the
grammars of Fr. Pedro de Alcal, Fr. Philip Guadagnoli, Thomas Erpenius,
Fr. Antonio de Aguila, Fr. Agapito de Valle Flammarum, and Fr. Fran-
cisco Gonzlez, and others. Likewise I have used a considerable amount
of manuscripts, which some aficionados of the Arabic language have
passed to me. Finally I have benefited from what study and experience
39
Motio est nominis Masculini in Foemininum converse; sitque additione termina-
tionis foemininae (Chapter V).
40
Johannes Reuchlin, who published his Hebrew grammar one year after the publi-
cation of Alcals, introduces the verb dagessare: quando he uel aleph repellunt nun pas-
siue significationis, dagessatur prima ut . . . (1974 [1506] Liber III, 590). See also Geiger
(1871, 129). The presence of Oriental elements in Western grammars is ofcourse not
only present when loans are used. Translations and mistranslations produced some-
times terms which are not longer recognized as from oriental origin. An example is the
history and development of the concept of radix (root).
232 otto zwartjes
have taught me over the period of sixteen years that I spent predicating
and confessing in Arabic in the missions of Asia.
To start with the first grammar mentioned, Pedro de Alcals Arte, we
can conclude immediately that this grammar has not been the source
for the terms used for the inflectional endings. Thomas Erpenius has
been apparently used, but we observed that Erpenius reduced the exotic
grammatical terms in his grammars,41 although he maintained them in
his Latin translation of Arabic treatises written by Ibn jurrm and
al-Jurjn, which could have been also the direct sources of Caes. Ibn
jurrm was born in Morocco in 12731274 and died in Fez in 1323.
He is the author of a grammatical compendium entitled Muqaddima
al-jurrmya Mabdi ilm al-Arab where he exposes the inflectional
system of Arabic, called irb. This treatise on syntax has been widely
used until the present day and it is one of the later works downstream the
long tradition starting with Sbawayhi. The Muqaddima was known in
Europe since the 16th century.42 This work has been printed for the first
time in Europe in 1592 (Medici, Rome). A translation by Peter Kirsten
(15771640), into Latin appeared in 1610 (Breslae, 1610), followed by
a translation of Erpenius (Leidae, 1617). In 1631, another translation
appeared by Thomas Obicini (15851632).43 The Kitb al-awmil al-mia
n-nahwiyya (Book of the Hundred Regentia) of Abd al-Qhir al-Jurjn
(died in ca. 1080) has been translated by Erpenius in 1617 and published
together with the Muqaddima: Grammatica Arabica dicta gjarvmia &
libellus centum regentium cum versione Latina & comentarijs (Leidae,
1617). As we shall demonstrate below, scholars in Rome, such as Philip
Guadagnoli, knew this work. Erpenius used in his translations of these
works loans from Arabic in his specified grammatical terminology. To
mention an example:
41
In this article, we quote from the Rudimenta (1620). For a more complete analysis,
the Grammatica Arabica, quinque libris methodice explicata a Thoma Erpenio (Leidae,
1613) and the Grammatica Arabica (Leidae, 1636) have to be taken into account.
42
See his article Ibn jurrm (Encylopedia of Islam. New Edition, 3, 697), and Ben
Cheneb (1927, 381382).
43
Thomas Obicini (15851632) was abbot of the Franciscan convent at Aleppo
between 1613 and 1619 and in 1621 he returned to Rome where he founded the college
at the St Peter Convent of Montorio where arabic was taught for the missionaries who
were being prepared to spread the faith in the East. He was responsible for the supervi-
sion of the type designs of Oriental types at the Propaganda Press.
partes autem eijus sunt Rafa, Nasba, Chafda and Gjezma, e quibus convenit
Nominibus Rafa, nasba and Chafda, non autem Gjezma: verbis vero Rafa,
nasba, and Gjezma: non autem Chafda (f. 11)
However, as has been demonstrated by Fck (1955, 68), Erpenius
replaced in his own grammars, when possible, Arabic terminology by
Latin equivalents, and his Rudimenta could not have been either the
source of inspiration of Caballero and Caes either. Erpenius maintained
the Arabic terminology of the vowels:
TABLE 6 THE VOWEL SYSTEM ACCORDING TO THOMAS

ERPENIUS (1636, f. 16).4445
Nomen Figura Potestas

Phatha44 Ba Nunc a purum & clarum ut in amabam, nunc cum e mix-
tum, id est Graecum, ut multi id nunc pronuntiant45
Dhamma Bu Nunc u purum & clarum, nunc cum o mixtum, id est o ob-
scurum
Kesre Bi i simplex
Erpenius does not maintain the Arabic terms of the subclasses of the
particles, as he rendered them in a Latinized form in his translation of Ibn
jurrm, but obviously he attempted to fit them into the Latin model.46
If we compare the translation into Latin of Erpenius with Obicinis, we
can conclude that the latter also used the Arabic terms for inflectional
endings.47 Obicini firstly gives the Arabic term, written in the Arabic
script, then a translation is given followed by a description or paraphrase
with the purpose to explain the Arabic terms: ar-raf as elevatio, the
definition of an-nasb is accusativus, quasi patiens positum sub agente
(without translation), al-xaf is rendered as depressio, & amplectitur
44
In the left column the terms are also written in Arabic script, and in the second
column the Arabic letter b is given, together with its appropriate vocalisation.
45
Sometimes a pure and clear a as in amaba, other times mixed with an e i.e. as in
Greek as many pronounce it now, etc.
46
For instance, we find definitions such as: de syntaxi Particularum: Praepositiones
omnes tum separatae regunt genitivum f baitin . . ., instead of the terms nasbantes, etc.
Nevertheless, in his Rudimenta we still can find verbs as gjezmare (1620, f. 47).
47
Rafaa, & Nasba, & Chafda, & Gezma. At nominibus ex ijs (conueniunt). Rafaa, &
Nasba, & Chafda, non autem Gezma, Verbis ver ex eisdem, Rafaa, & Nasba, & Gezma,
non autem autem Chafdha (Obicini 1631, f. 3). Agapito Valle Flemmarum has almost
the same definition (1687, 194).
234 otto zwartjes
genitivum, dativum, & ablativum, quasi obliquus, and finally, al-jazm

as abscisio, idest casus privatio, quasi dictione cadentium scindat. This
means that Obicini decided to maintain Arabic terminology, and when
explained appropriately, the learner could take advantage of these Arabic
terms which were developed for describing the Arabic language.
And now lets come back to the concept of mil f. In Troupeau (1962),
the most important translations in this period is listed: In Kirstens and
Obicinis translation we have agens, whereas the first also uses efficiens
and the latter uses also regens. Erpenius uses both regens as operans.
As Michael Carter observed, we see until today that the basic meaning
has been neglected by many scholars:
Anderseits kann man fast an eine Verschwrung glauben, die Grundbe-
deutung des Terminus amal auch heute noch zu unterdrcken, um den
unbegrndeten Mythos zu verewigen, der arabische Begriff vom struktu-
rellen Verhltnis zwischen Satzteilen sei identisch mit dem lateinischen,
d.h. eine Art Rektion. Da die Grundmetapher der lateinischen Rektion
eine durchaus hierarchische, senkrechte Einordnung der Satzteile voraus-
setzt, die arabische amal Operation, aber im Gegenteil eine waagerechte
Beziehung schildert, scheint die moderne Sprachwissenschaft nicht aner-
kennen zu wollen. (Carter 1993, 133).
Carter observed that Wei was an exception and demonstrated that
the translation by Kirsten unfortunately did not have any impact on
later translations or interpretations: Wie eine Stimme in der Wste
verbleibt noch der Aufsatz von J. Weiss. Die wrtliche und systemtreue
bersetzung operans von Kirsten hat sich leider nicht durchgesetzt
(Carter 1993, 134).
In the first place we have to add that Franciscus Martelottus and Philip
Guadagnoli (15961656), both neglected by Troupeau, also deserve our
attention in this context, since we find in their grammars of Arabic a
very precise analysis and translation of the concept. To start with the
latter: the concept of al-irb is translated as arabicatio and amal as
efficientia or operatio, and al-awmil as regentia:
Ratione autem, quo vnuquodque in suo vel Casu vel Modo & quacumque
dispositione locatur, quam Constructionem latini dicunt, Arabes interdum
al-irb, Arabicationem, proprius autem amalun efficientiam, seu opera-
tionem appellant. Nomen autem illud, seu Verbum, seu Particula, eiusmodi
operationem exercens, scilicet cuius vi aliquid in tali vel tali dispositione
locatur, dicitur, milun operans, regens. Locatum ver ex eius dicitur
mamlun operatum, rectum. Exinde dicuntur, al-awmilu regentia. Qui-
bus omnibus notis, nihil superest Grammatico. Collegit autem quidam,
cognomine Giargianius, in libello Regentia omnia, quem propterea nomina-

vit De Centum Regentibus. Regentia enim, vel sunt lafziyyatun explicita, vel
manawiyyatun implicita.48 (Guadagnoli 1642, f. 248).
However, the system according to which a constituent has to be placed
in the appropriate case or mode and on which convenient position has
to be placed, this system which the speakers of Latin call Constructio
[= syntax], is system is called al-irb by the Arabs which means arabi-
cization, but they call this more appropriately amalun efficiens [an act
which produces a certain effect], or operation [an act caused by force of
an operator]. When a Noun, a Verb or a Particle produces such effects in
this manner, and when by force of this effect something has to be placed in
a certain disposition, we call this milun [producing a certain effect upon
something, to govern]. The element which has been effected is called
mamlun [the governed]. Consequently, al-awmilu are called regentia
[governers]. Although all this is well-known, nothing has been transmit-
ted by The Grammarian,49 but someone with the name Giargianius has
collected in a booklet all the regentia, which can be subdivided in two
subclasses, the lafziyya (expressed) and the manwiyya (abstract).
As we see, Guadagnolis description is not only accurate and complete,
since the original Arabic is given and the literarly meaning had been
maintained, but he also mentions his source, which is the hundred
regentia written by al-Jurjn (Giargianius).50
Do we find traces of these translations of amal other grammars writ-
ten in Latin of this period? The answer is positive. In chronological
order we shall summarize what other grammarians from Rome taught
us, starting with one of the earlier grammars that appeared in Rome
after the Medici translation of the work of Ibn jurrm, the Institutiones
of Franciscus Martelottus.
Martelottus not only mentions Erpenius in his prologue (1620, 38),51
but he also deals with methodology. Should the Arabic model with
48
Operators can be expressed (lafziyy), and abstract (manaw). The first class are
the particles or verbs or nouns that are either actually uttered or elided but understood,
while the latter are abstract causes that do not involve uttered or restored linguistic
elelements. (Baalbaki 2004, XV, 2358). This means that elided elements can produce
effects.
49
We could not identify this Grammarian.
50
Also in other paragraphs we see direct translations in the work of Guadagnoli,
which remain close to the Arabic original, such as ignoratum, for majhl, usually
erroneaously translated as passive, opposed to cognitum for the active (marf )
(Guadagnoli, 1642, f. 255).
51
Scripsit autem eleganter admodum de dictionibus hisce, quemadmodum & de lit-
teris eruditissimus Orientalium linguarum in Leidensi Academia professor Thomas Erpe-
nius uniusquam leuasset quoq. nobis huiis secundi libri labores partem (1620, 38).
236 otto zwartjes
its technical terms be maintained, or should they be abandoned and

replaced by Latin terminology? Martelottus decided to follow the tra-
ditional method when dealing with the single word classes (Priora
vero hic loci propri de singulis dictionibus singillatim ordine debito,
ac recta methodo explicanda sunt (1620, 37), but we can infer from his
preface, he obviously follows the Arabic model, and he does this explic-
itly.52 However, Martelottus does not leave the Arabic terminology out
of his volume:
Ubi omisso Arabicorum grammaticorum ordine Latinis admodum dissono,
in rebus pluribus nobiscum, quoad methodum conuenisse comperimus.
Caeterum quamvis praedicta Arabum methodo deflectentes, nostro nos
ordine Latinis magis consono procedamus, omnino grammaticalibus voci-
bus ubique utemur, eorumque ordines, ac procedendi rationes, divissiones,
ac series enucleabimus, ut faciliro cunctis ad eosdem perlegendos authores
pateat aditus. (Martelottus 1620, 38).
Whenever the difference between Arabic grammars and the Latin way of
construction was left out of consideration, we discovered that, with res-
pect to the system of the language, Arab had many similarities with ours.
After all, to what extent we ourselves may differ from this language system
of the Arabs and are differing more and more as well from the Latin way
of construction, we should still use in our research the Arab terminology
and developing our knowledge we will explain the systems, the methods
of construction and concatenation, in order to make it easier for us all to
read the same authors.
Martelottuss methodology is in our eyes extremely modern. It tries to
bridge the gap between exo- and endo-grammaticalization. An eclectic
approach, combining the best elements of both traditions and their
corresponding technical terms is the best way to understand the Arabic
language. Martelottus also dedicates an entire chapter to inflection (De
irb, seu Inflexione), translated as Arabificatio, or Arabicatio (1620,
98). The concept of mil is translated again as operans and the author
quotes directly from the Arabic grammatical tradition:
irb apud Grmaticos tayru awkhir al-kalm il al-ikhtilf al-
awmil: variatio ultimorum, seu extremitatum dictionum, ob diversitatem
operantium (1620, 98).
52
In his grammar we find Arabic terminology extensively, together with the Latinized
form, for instance the traditional Arabic classification of the consonants: Chalchiiton,
lahuiiaton, sciagiariiaton, asliiaton, natiiaton, dhalchiiaton, sciaphahiiaton, liniiaton
(1620, 35).
Variation of the final parts or extremities of the words, caused by the diver-
sity of the operators (by the different effects produced by the operators).
At the end of the 17th century, Agapito Valle Flemmarum treats the
particles in detail in his chapter entitled De syntaxi Particularum,
and particularly describes the effect they have as operators on the
inflectional system, translating al-h urf al-mila as:
1. particlae operantes in nomen simplex Giarrantes,

2. et in nomen simplex nasbantes,
3. et in summam,
4. et in verbum nasbantes,
5. et operantes in verbum Giezmantes (1687, 268).
Another Franciscan, Antonio ab Aquila, also used Arabic terminology

for the particles, although we do not find the same definitions. In his
classification of the subclasses, we find a Latin terminology, such as
particulae coniunctae and particulae separatae, with subclasses, such
as particula interrogationis, iurandi, status in loco, ad determinationem,
optativi, conformandi, negationis, conditionalis (1650, Caput XXV,
388 ff.). However, he also uses Arabic terms, such as particula Gizemeh
(f. 403).
And finally, the question rises wether we find traces of the concept
of mil in the two Spanish grammars. The answer is again positive, as
we shall demonstrate. Lucas Caballero gives rules for classical Arabic
(gramtica literal) where particles have a certain effects:
De las particulas charrantes
Despues de las particulas que hemos puesto en el Arabo corresponden a
nuestras quatro ultimas partes de la oracion, de las quales particulas, la
mayor parte son literales; siguiendo el intento de dar algunas reglas de la
grammatica literal, quiero explicar algunos efectos que tienen las particu-
las puestas o otras que a ellas siguen . . . (Cabellero 1709, f. 7r.; emphasis is
mine)
After the particles we have put in Arabic, four final parts of speech cor-
respond with ours, whose major part are literary utterances: following our
intention to give some rules of literary grammar, I want to explain cer-
tain effects these particles have when placed, or other which follow after
them.
In the grammar of Caes, we find exactly the same term, hispanicised
as operacion:
238 otto zwartjes
. . . quando se les une la particula m quedan absolutamente privadas de su

operacion, y regimen, de suerte que pierden la fuerza que tenian de colocar
el sujeto en el caso nasbo y se queda en el rafeo, nominativo, como cier-
tamente que escrivir Pedro, innam sayaktubu Butrus (Caes 1775, 144;
emphasis is mine).
. . . when combined with the particle m they remain absolutely deprived of
their operation and government so that they loose their force which they
had before to put the nasb (a inflection, or direct object) and it remains
in the raf (u ending, or subject) or nominative, as in certainly to write to
Peter innam sayaktubu Batrus.
To sum up, we have found the following authors as possible sources of
the two Franciscan grammarians of Damascene Arabic: Ibn jurrm
and al-Jurjn (through translations of the Medici edition, Erpenius,
Obicini and maybe also Kirsten) as the main Eastern sources. Western
sources mentioned by name are Erpenius, Golius (mentioned by Caes),
Agapito Valle and Guadagnoli. It has to be observed that missionary
grammarians in Rome were familiar with some of the most important
Arabic authors. The name of Al-Xall b. Ahmad al Farhid (died in 780)
who codified and established a system of 15 meters has been mentioned
by Guadagnoli. Since the grammars of Caballero and Caes do not have
a final chapter on prosody, they did not deel the need to use this source.
Another prominent pioneer, who established the foundations of Arabic
grammar is of course Sbawayhi, whose work became known in the
West through translations of Jahn. It is surprising that we find already in
1620 his name in the grammar of Martelottus, which is evidence for the
fact that missionary-grammarians in Rome knew who he was. Dealing
with the verba ternaria, Martelottus explains that the forma masdari is
irregular, and in that context he mentions Sbawayhi:
In verbis ternarijs, ut iam diximus, forma masdari irregularis est, omnes
autem quidem Author Arabicus, nomine Sibauai ad 32. reuocauit, videlicet
(1620, 213).
In the ternary verbs, as we said before, the masdar-form is irregular;
indeed, the Arabic author with the name Sbawayhi reduced all these to
32.
The same author, Martelottus, also mentions another Arabic source in
his chapter dealing with the Constructio Particularum where he divides
the particulas operantes in five subclasses, in agreement with an Arabic
work called Lucerna, or Al-Misbh , the particulae operantes in nomen
simplex giarrtes, particulae operantes in nomen simplex nesbantes,
in verbum nasbantes, in verbum gezmantes, and the original text in
Arabic script is given in the same table as well. Although Martelottus

does not give the name of the author, we think this work is probably
the treatise written by al-Muta rriz, entitled Al-Misbh f ilm an-nahw.
Al-Mutarriz (11441213) compiled this treatise which became a text-
book in the madrasas of the East. The Misbh itself wa sbased on three
small grammatical monographs of al-Jurjn (Lichtenstdter 1936, 847
and Sellheim (EI [New edition] 7, 773).
4. Conclusion
Summarizing, we can conclude that Pedro de Alcals use of Arabic

terminology seems to be unsystematic and the reasons why he used
them remain unclear, particularly when he uses the Arabic names for
the cases. They do not reflect the Arabic inflectional endings and his
model was obviously Latin grammar. The use of the term damr is an
exception, since the suffixed pronouns can be used in a different way-
attached to verbs, nouns and particlescompared to Latin. Pedro de
Alcals mnemonic terms of the vowels stand alone, and we do not find
any use of them in other works and in his dictionary we do not find the
terms fath a, kasra, and damma. Although we find in Alcals dictionary
the translation obrar for the Arabic root amala there are no traces of
Arabic theory concerning operators or government, related to this
term.
The earliest translations of al-Jurjn and Ibn jurrm are without any
doubt an important milestone for the development of the study of non-
Western grammatical theory in the West and probably for some of them
a real new discovery, which can serve as an enrichment of the West-
ern system, as Martelottus postulates. Direct influence of these works
can be found in the grammars of Martelottus, Ab Aquila, Agapito and
Guadagnoli, and the Franciscans in Damascus continued this tradition.
Although Caballero and Caes were not the pioneers themselves, they
were probably the first grammarians who introduced extensively Arabic
terminology in the Spanish metalanguage, as we have demonstrated.
In Northern Europe, however, we see a different approach. Erpe-
nius did not aim at orientalising Western grammatical terminology,
except for the terms for the vowels and some other terms. The concept
of mil has followed its own course in grammatical theory and in the
20th century it was absorbed by anachronistic terms such as govern-
ment and dependency, as Carter demonstrated. Nevertheless, we have
240 otto zwartjes
demonstrated that the Franciscans were totally aware of the right con-
notations of mil although we have to admit that sometimes they use
gobierno or regimen as a synonym for efficiens or operans. Gua-
dagnoli and Martelottus gave us without any doubt the most detailed
analysis, and probably Caballero and Caes have been inspired by their
works. It was surprising that not only works of Ibn jurrm, al-Jurjn
and al-Muta rriz are mentioned by name by some of the grammarians
working in Rome, but even Sbawayhi is mentioned by name in this
relatively early period (1620).
Missionary linguistics in Rome, particularly the achievements of
scholars and teachers who published grammars in the seventeenth
century at the Polyglot Press of the Sacra Congregatio de Propaganda
Fide deserve to be studied more in detail in future. Particularly those
authors who tried to combine exo- and endo-grammatical terminol-
ogy and approaches have been innovative. How the learners of Arabic
appreciated this bridging approach is another question. Many scholars
preferred in their teaching curriculum the more Latin-based grammar
of Erpenius and his work was without any doubt a great success dur-
ing many centuries. However, I agree with Martelottus that there is no
reason to postulate that the Arabs differ from our system. We can also
say, we ourselves may differ from this Arabic language system. While
using their own terminology, which has been developed for their own
linguistic phenomena, we will make progress in the understanding of
not only the language but of the linguistic model as well.
5. References
5.1 Primary sources

Alcal, Pedro de. [Before 5 Feb. 1505a]. Arte para ligeramente saber la lengua arauiga.
Salamanca: Juan Varela.
. [Before 5 Feb. 1505b]. Vocabulista arauigo en letra castellana. Salamanca: Juan
Varela.
. [ca. 1506a]. Arte para ligera mente saber la lengua arauiga emendada y aadida y
Segunda mente imprimida. Salamanca: Juan Varela.
. [ca. 1506b]. Vocabulista arauigo en letra castellana. Salamanca: Juan Varela. [Bib-
lioteca Nacional, R 31638.] Re-edition: Paul de Lagarde. 1883. Petri Hispani de lin-
gua arabica libri duo. Gottingae: In aedibus Dieterichianis Arnoldi Hoyer. Reprint:
Osnabrck: Otto Zeller, 1971.
Aquila, Antonio Ab. 1650. Arabicae linguae novae, et methodicae institutiones. Non ad
vulgaris dumtaxat Idiomatis; sed etiam ad Grammaticae doctrinalis intelligentiam, per
Annotationes in Capitum Appendicibus suffixas, accomodatae. Authore F. Antonio ab
Aquila Or. Min. Sancti Francisci strict. Obser. Theologo, atque in Collegio Sancti Petri
Martis Aurei Sacra Congregatione de propaganda Fide Arabicae linguae deputato

lectore. Opus tum omnibus Arabicae linguae studiosis, tum potissumum Apostolicis
Viris, per Asiam & Africam Fidem propagaturis, utile & necessarium. Romae: Typis
Sac. Cong. De Prop. Fide.
Caballero, Lucas. [Cauallero] 1709. Compendio de los Rudimentos y Gramatica Araba en
que se da suficiente notizia de la lengua Vernacula o Vulgar y algunas Reglas de la literal
Iustamente P.M.R.F. Bernardino Gonzalez hijo de la Pro de la Concepzion en Espaa
Lector Jubilado en Arabo y Misionero Apostolico del Oriente y recoplada por el Re.do P.
Fr. Mo Apostolico hijo de la Proa de los Angeles Lector actual Arabo en el colegio de
Damasco. Ms. Rogge Library, Handskriftssamlingen, Sweden.
Caes, Francisco. 1775. Gramatica arabigoespaola, vulgar, y literal. Con un diccionario
arabigo-espaol, en que se ponen las voces mas usuales para una conversacion familiar,
con el Texto de la Doctrina Cristiana en el idioma arabigo. Madrid: En la Imprenta de
Don Antonio Perez de Soto.
Dombay, Franciscus de. 1800. Grammatica linguae Mauro-Arabicae juxta vernaculi idi-
omatic usum, accessit vocabularium Latino-Arabicum. Vindobonae: apud Camesina.
Erpenius, Thomas. 1617. Grammatica Arabica dicta Gjarumia, & Libellus centum regent-
ium cum versione latina & commentariis Thomae Erpenii. Leidae: Ex typographia
Erpeniana Linguarum Orientalium.
. 1620. Rudimenta lingv arabicae. Accedunt ejusdem Praxis Grammatica; & Con-
silium de studio Arabico feliciter instituendo. Leidae: Ex Typographia Auctoris.
Germanus [of Silesia], Dominicus. 1639. Fabrica linguae Arabicae cum interpretatione
Latina & italica, accomodata ad usum linguae vulgaris, & scripturalis. Romae: Typis
Sac. Congreg. De Prop. Fide.
. 1640. Fabrica Arabica copiosioribus impensis atque vberiori structura. Romae.
Gonzlez, Bernardino 2005 [c. 1705]. Intrprete Arbico. Edicin y estudio preliminar
por Ramn Lourido Daz. 2 vols. Madrid: Real Academia de la Historia/Ministerio de
Asuntos Exteriores y de Cooperacin.
. 2005 [c. 1705]. Eptome de la gramtica arbiga. Edicin y estudio preliminar por
Ramn Lourido Daz. 2 vols. Madrid: Real Academia de la Historia / Ministerio de
Asuntos Exteriores y de Cooperacin.
Guadagnoli, Philippi. 1642. Breves Arabicae linguae Institutiones. Romae: Ex Typographia
Sac. Congregationis de Propaganda Fide.
Martelottus (= Martelotti), Franciscus. 1620. Institutiones Linguae Arabicae Tribus Libris
distributae. In quibus uberrime quaecumque ad litteras, Dictiones & Orationem atti-
nent, explicantur. Authore P. Francisco Martelotto Martinensi, Sacerdote, Theologo,
Clericorum Regularium Minorum. Romae: Excudebat Stepahsnus Paulinus.
Nepos, Ferdinandus. 2000.[c. 14681485]. Materies. (Ms. See Codoer).
Obicini, Franciscus. 1631. Grammatica Arabica Agrumia appellate. Cum versione Latina
ac dilucida expositione. Adm. R.P.F. Thomae Obicini Noniensis, Diae. Nouariau, ord.
Minorum Theologi, ac linguarum Orientalium in Collegio S. Petri Montis Aurei, de
mandato Sacrae Congregationis Fidei propaganda Magistri. Romae: Typis Sac. Con-
gregationis de Propaganda Fide.
Oyanguren de Santa Ins, Melchor. 1738. Arte de la lengua japona. Dividido en quatro
libros segun el Arte de Nebrixa, con algunas voces proprias de la escritura, y otras de
los lenguages de Ximo, y del Cami, y con algunas perifrasis, y figuras. Mxico: Joseph
Bernardo de Hogal.
. 1742. Tagalysmo elucidado, y reducido (en lo possible) la Latinidad de Nebrija, con
su Syntaxis, Tropos, Prosodia, Passiones, &c. y con la allusion, que en su uso, y composi-
cion tiene con el Dialecto Chinico Mandarin, con las Lenguas Hebrea, y Griega. Mxico:
D. Francisco Xavier Snchez, en la Calle de San Francisco.
Reuchlin, Johannes. 1974[1506]. De rudimentis hebraicis libri III. Pforzheim. Reprint:
Hildesheim New York: Georg Olms Verlag.
242 otto zwartjes
Rodrigues, Ioo. (Rodriguez) 16041608. Arte da Lingoa de Iapam composta pello Padre
Portugues da Copanhia de Iesv diuidida em tres Livros. Com Licena do Ordi-
nario, e Svperiores. Nagasaqui: no Collegio de Iapo da Companhia de Iesv Anno.
1604 (-1608 colofon.)
Rodrigves, Ioam. 1620. Arte Breve da Lingoa Iapoa tirada da Arte Grande da mesma
lingoa, pera os que comecam a aprender os primeiros principios della. Pello Padre Ioam
Rodrigvez da Companhia de Iesv Portugues do Bispado de Lamego. Diuidida em tres
Livros. Com Licenca do Ordinario, & Superiores. Amacao: no Collegio da Madre de
Deos da Companhia de Iesv.
Roth, Heinrich. 1988[ca. 1662]. Grammatica Linguae Sanscretanae Brachmanum Indiae
Orientalis. Facsimile edition: Arnulf Camps & Jean-Claude Muller: The Sanskrit
Grammar and Manuscripts of Father Heinrich Roth S.J. (16201668). Facsimile edition
of Biblioteca Nazionale, Rome, Mss.Or. 171 and 172. Leiden: E. J. Brill.
Valle Flemmarum, Agapiti . 1687. Flores Grammaticales arabici idiomatis collecti ex
optimis quibusque Grammaticis nec non pluribus Arabum monumentis, ad quam max-
imam fieri potuit brevitatem atque ordinem revocati: Studio, & Labore Fr. Agapiti
Valle Flemmarum Ordinis Minorum S. Francisci reformatorum Provinciae Tridentinae,
in Semionario Patavino lectoris. Opus omnibus Arabicae Linguae Studiosis perutile, &
necessarium. Patavii: Ex Typographia Seminarii Opera Augustini Candiani.
Ziegenbalg, Bartholomus. 1985 [1716]. Grammatica Damulica. Qu Per Varia Para-
digmata, Regulas & Necessarium Vocabulorum Apparatum, Viam Brevissimam Mon-
strat, Qua Lingua Damulica Seu Malabarica, Qu Inter Indos Orientales in Usu Est, &
Hucusque In Europa Incognita Fuit, Facile Disci Possit: In Usum Eorum Qui Hoc Tem-
pore Gentes Illas Ab Idololatria Ad Cultum Veri Dei, Salutemque ternam Evangelio
Christi Perducere Cupiunt . . . Hal Saxonum: Litteris & Impensis Orphanotrophei.
Ed. Burchard Brentjes & Karl Gallus, Halle: Martin-Luther-Universitt Halle-Witten-
berg, 1985.

Baalbaki, Ramzi. 2004. Grammarians and Grammatical Theory in the Medieval Arabic
Tradition. Aldershot: Ashgate Variorum.
Ben Cheneb, Mohammed 1927. Ibn djurrm. Enzyklopdie des Islm. 2.381382.
Leiden Leipzig: Brill Harrassowitz.
BICRES (see Niederehe).
Carter, Michael G. 1991. Review Owens 1988. Journal of the American Oriental Society,
111.2: 395397.
. 1993. Probleme bei der bersetzung von Fachsprache am Beispiel des Ara-
bischen. In Frank, Armin Paul, Kurt-Jrgen Maa, Fritz Paul, und Horst Turk, eds.
bersetzen, Verstehen, Brcken bauen. Geisteswissenschaftliches und literarisches
bersetzen im internationalen Kulturaustausch. Berlin: Erich Schmidt., 1.130141.
. 1994. Review Owens 1990. Journal of the American Oriental Society, 114.3: 472
475.
Codoer, Carmen. 2000. Gramticas Latinas de transicin. Juan de Pastrana, Fernando
Nepote. Introduccin y edicin crtica. Salamanca: Ediciones Universidad de Sala-
manca.
Corriente, Federico. 1988. El lxico rabe andalus segn P. de Alcal. (Ordenado por
races, corregido, anotado y fonmicamente interpretado. Madrid: Universidad Com-
plutense de Madrid.
Cowan, William. 1981. Arabic grammatical terminology in Pedro de Alcal. Histo-
riographia Linguistica, 8.2/3:357363.
Dannenfeldt, Karl H. 1955. The Renaissance Humanists and the Knowledge of Arabic.
Studies in the Renaissance, 2, 96117.
Escavy, R., M. Hernndez Terrs, A. Roldn, eds. 1994. Actas del Congreso Internacio-
nal de Historiografa Lingstica. Nebrija V-Centenario 14921992. Murcia, 14 abril
1992. Murcia: El Taller, vol. III: Nebrija y otros temas de historiografa lingstica.
Frank, Armin Paul, Kurt-Jrgen Maa, Fritz Paul, und Horst Turk, eds. 1993. bersetzen,
Verstehen, Brcken bauen. Geisteswissenschaftliches und literarisches bersetzen im
internationalen Kulturaustausch. Berlin: Erich Schmidt.
Fck, Johann. 1955. Die arabischen Studien in Europa bis in den Anfang des 20. Jahrhun-
derts. Leipzig: Otto Harrassowitz.
Geiger, Ludwig. 1871. Johann Reuchlin. Sein Leben und seine Werke. Leipzig: Duncker
and Humblot.
Hauschild, Richard. 1988. Notes on the Content of the Three Manuscripts of Heinrich
Roth. Arnulf Camps and Jean-Claude Muller: The Sanskrit Grammar and Manu-
scripts of Father Heinrich Roth S.J. (16201668). Facsimile edition of Biblioteca Nazio-
nale, Rome, Mss.Or. 171 and 172. Leiden: E.J. Brill.
James, Gregory. 2007. The terminology of declension in Early Missionary Grammars
of Tamil. In Zwartjes, Otto, Gregory James and Emilio Ridruejo, eds., Missionary
Linguistics III/Lingstica misionera III. Morphology and Syntax. Selected papers from
the IIIrd and IVth International Conferences on Missionary Linguistics, Hong Kong/
Macau/Valladolid, 167190.
Juynboll, W.M.C. 1931. Zeventiende-eeuwsche beoefenaars van het Arabisch in Neder-
land. Utrecht: Kemink en Zoon.
Kerkhof, Maxim, Hugo de Schepper, Otto Zwartjes, eds. 1993 Espaa: Ruptura 1492?.
Amsterdam: Rodopi.
Killean, Carolyn G. 1984. The development of Western Grammars of Arabic. Journal
of Near Eastern Studies, 43.3: 223230.
Law, Vivian. 2002. The History of Linguistics in Europe. From Plato to 1600. Cambridge:
Cambridge University Press.
Lentin, Jerome. 1997. Recherches sur lhistoire de la langue arabe au Proche-Orient
lpoque moderne. Thse de Doctorat dtat, Universit de Paris III.
Lichtenstdter, Ilse. Al-Mutarriz. Enzyklopdie des Islm, 3:847. Also New edition:
R. Sellheim, 7:773.
Lourido Daz, Ramn. 2005. Estudio preliminar. See: Gonzlez (primary sources).
Maruyama, Toru. 2004. Linguistic Studies by Portuguese Jesuits in Sixteenth and Se-
venteenth Century Japan. In Zwartjes, Otto and Even Hovdhaugen, eds. Missionary
Linguistics [I]/Lingstica misionera [I]. Selected papers from the First International
Conference on Missionary Linguistics, Oslo, 1316 March 2003. Amsterdam: John Ben-
jamins. 141160.
Monroe, James. T. 1970. Islam and the Arabs in Spanish Scholarship (Sixteenth century to
the present). Leiden: Brill.
Mrner, A.R. and Magnus Mrner. 2001. Spanien i svenska arkiv. Skrifter utgivna av
Riksarkivet, 16. Stockholm: Riksarkiv, 6061.
Niederehe, Hans-Josef. 2005. Bibliografa cronolgica de la lingstica, la gramtica y la
lexicografa del espaol (BICRES III). Desde el ao 1701 hasta el ao 1800. Amster-
dam: John Benjamins.
Owens, Jonathan. 1988. The Foundations of Grammar. An introduction to Medieval Ara-
bic Grammatical Theory. Amsterdam: John Benjamins.
. 1990. Early Arabic Grammatical Theory. Heterogeneity and Standardization.
Amsterdam: John Benjamins.
Schnurrer, Christianus Fridericus de. 1811. Bibliotheca Arabica. Halae ad Salam: I.C.
Hendelii.
Smith-Stark, Thomas. 2005. Phonological description in New Spain. In Zwartjes, Otto
and Cristina Altman, eds.. Missionary Linguistics II/Lingstica misionera II. Orthog-
raphy and Phonology. Selected Papers from the Second International Conference on
Missionary Linguistics, So Paulo, 1013 March 2004. Amsterdam: John Benjamins,
164.
244 otto zwartjes
Troupeau, G. 1962. Trois traductions latines de la Muqaddima dIbn urrm. tudes

dOrientalisme ddies la mmoire de Lvi-Provenal. Paris: Maisonneuve et Larose,
1.359365.
Versteegh, Kees. 1997. Landmarks in Linguistic Thought III. The Arabic Linguistic Tradi-
tion. London: Routledge.
Wei, Josef. 1910. Die Arabische Nationalgrammatik und die Lateiner. Zeitschrift der
deutschen Morgenlndischen Gesellschaft, 64:349390.
Zwartjes, Otto. 1993. El artculo en las gramticas pioneras de Nebrija y Alcal y las
gramticas grecolatinas. In Kerkhof, Maxim, Hugo de Schepper, Otto Zwartjes, eds.
Espaa: Ruptura 1492?. Amsterdam: Rodopi, 261286.
. 1994. Tradicin e innovacin en las gramticas pioneras de Antonio de Nebrija
y Pedro de Alcal: la categora gramatical del pronombre. In Escavy, R., M. Hernn-
dez Terrs, A. Roldn, eds. 1994. Actas del Congreso Internacional de Historiografa
Lingstica. Nebrija V-Centenario 14921992. Murcia, 14 abril 1992. Murcia: El
Taller, 3:651665.
. 2007. Agreement asymmetry in Arabic according to Spanish missionary gram-
marians from Damascus (XVIIIth century). In Zwartjes, James, Ridruejo, eds., forth-
coming. Missionary Linguistics III/Lingstica misionera III. Morphology and Syntax.
Selected papers from the IIIrd and IVth International Conferences on Missionary Lin-
guistics, Hong Kong/ Macau/Valladolid, 273303.
, and Even Hovdhaugen, eds. 2004. Missionary Linguistics [I]/Lingstica misionera
[I]. Selected papers from the First International Conference on Missionary Linguistics,
Oslo, 1316 March 2003. Amsterdam: John Benjamins.
, and Cristina Altman, eds. 2005. Missionary Linguistics II/Lingstica misionera II.
Orthography and Phonology. Selected Papers from the Second International Conference
on Missionary Linguistics, So Paulo, 1013 March 2004. Amsterdam: John Benja-
mins.
, Gregory James and Emilio Ridruejo, eds. 2007. Missionary Linguistics III/Lings-
tica misionera III. Morphology and Syntax. Selected papers from the IIIrd and IVth
International Conferences on Missionary Linguistics, Hong Kong/ Macau 1215 March
2005. Valladolid, 811 March 2006. Amsterdam: J. Benjamins.
LINGUISTICS
THE LINGUISTIC ANALYSIS AND RULES OF
PAUSE IN ARABIC
Salman H. Al-Ani
Indiana University, Bloomington
1. Introduction
The Arabic term waqf pause is a grammatical concept. It is a verbal

noun derived from the verb waqafa that could mean to stop, to come
to a standstill or to pause. This verb is used both in transitive and
intransitive form. The term waqf is from the transitive verb waqafa.
Another verbal noun wuqf standing still is derived from the intransitive
verb waqafa (Farrj 2001, 1314). The term waqf is more commonly used
and has a significant point of reference in both linguistic and religious
connotations that mean to hold something at a pause or stop. This term
waqf and its plural form awqf mean an endowment in Islamic law
which signifies the dedication of property or land that cannot be sold.
The chain of speech utterances may be divided into spoken group
events. This is usually regulated by the phonological, syntactic and
semantic rules of the language. Normally words in the chain of the
utterances of speech in prepausal forms either end with vowels or con-
sonants. In Arabic vowels at the end of a speech event in prejunctional
state are a sign of continuity and consonants signal waqf (H assn 1973,
270271). When speaking or reading aloud with the prepausal mode
almost all short vowels at the end of words, phrases and sentences will
be dropped. This linguistic phenomenon, in Arabic, is referred to as
pause forms.
Oral reading and recitation have been highly emphasized in the
Arab-Islamic tradition. Poets in Arabia in the pre-Islamic period com-
posed their odes to be recited and the members of the poets tribe in
turn memorized and recited these poems whenever the opportunity
availed itself for them to do so. When the Qurn was first revealed it
emphasized recitation. The very first verse states recite or read aloud,
iqra. The tradition of oral reading, in general, is engrained in the hearts
and minds of most Muslims and Arabs.
The Arabic sources that dealt with this topic were focusing on the
rules of waqf in Classical Arabic with special emphasis on the rules of
248 salman h. alani
recitation in the Qurn. This aspect of waqf will not be covered in this
paper. The emphasis is on the phenomenon of waqf with the explana-
tion of the basic linguistic features of waqf of Modern Standard Arabic.
Previous research on waqf, in western languages, is rather limited. There
are four studies that made a basic contribution to waqf or pause. The fol-
lowing is brief account of these studies.
1.1 F.T. Mitchell (1990)

Mitchell wrote that pause . . . relate(s) first and foremost to words that
occur in apocopated form before pause, that is predominantly, final in
the phrase and sentence. He added that: In discourse, it should not
be expected that pause will always neatly correlate with grammatical
phrasal divisions nor that the most appropriate divisions are always
observed by speakers or readers, but the principle should always be
followed that, wherever a pause is made, the preceding word should be
pronounced in its pausal form.
He postulated the following rules: (i) a final short vowel, for example
those of case, tense and mood is omitted; (ii) the sign of the indefinite,
i.e., in nunation (-n) is omitted together with the preceding vowel, in
the nominative (-un) and genitive (-in) cases of nouns and adjectives;
(iii) accusative (-an) may be replaced by a long fath a (); (iv) the femi-
nine singular and unit ending t marbtah is replaced by /-h/ in all
three cases (Mitchell 1990, 99100).
This brief account of waqf by Mitchell covers the basic rules of pause
of al-arabiyyah. It is intended to aid the learner of Arabic to pronounce
and read aloud correctly. He used a short transliterated Arabic passage
read aloud by what he called . . . a speaker trained in the high Classi-
cal tradition to illustrate the pause form rules. The comments and the
pause rules almost mirror the traditional statements of the Arab gram-
marians on waqf (Mitchell 1990, 100101).
1.2 Clive Holes (2004)

Holes defined waqf as: pause is defined (rather vaguely) as an audible
break in delivery (Holes 2004, 63). The focus of his analysis is on the
spoken Modern Standard Arabic. The illustrative examples he used were
taken from Syrian radio broadcast and television news and speeches of
a former Egyptian leader, Gamal Abdul Nasser. He (Holes) said that
linguistic analysis and rules of pause in arabic 249
pauses depend on what is said in the spoken Modern Standard Arabic,

the audience and the speakers intentions.
The conditions where prepause occur are: (i) final short vowels that
appear in prejunctural position are omitted; (ii) the feminine gender
morpheme t marbtah in certain nouns and adjectives in prepause,
together with the /-u, -i, and -a/ are deleted. He added that /-ah/ is pro-
nounced as /-a/; (iii) The inflectional suffixes /-un/ nominative, /-an/
accusative and /-in/ genitive were discussed individually. The /-un/ is
omitted in prepause. The nature of the accusative /-an/ depends on its
grammatical status. When it is used as an adverbial marker as in kullyan
completely the majority of speakers retain it, however few realize it as
/-/. When /-an/ is used as marker of an object of nouns or an adjective
that agrees with one, it is sometimes retained and sometimes omitted.
Holes analysis of the inflectional suffix /-in/ is very interesting. He made
a distinction of /-in/ when it is used as a marker of a noun in the genitive
case and /-in/ as a marker of an attributive adjective in agreement with
a genitive head noun. The suffix /-in/ usually is omitted, however, Holes
gave some examples where it is retained (Holes 2004, 6368).
1.3 Muh ammad Farrj (2004)

The focus of Farrjs long Arabic article is on the traditional grammatical
concept of waqf. The emphasis of this article was on the rules of waqf in
Classical Arabic Grammar and its application for the recitation of the
Qurn. He stated that there are two basic features of waqf. These are
wajh manner and mah all place. The purpose of mah alli l-waqf place of
pause is to regulate the speech event and leads to improving its division
where juncture or pause may occur. The wajh, plural awjuh, provides
the speaker or the reader with the guide for waqf so the speech will
appear harmonized in its utterances. He added that waqf has a wazfa
task which serves for clarity of pronunciation and meaning that help in
understanding of what is being read or spoken (Farrj 2004, 1223).
1.4 Tamm H assn (1973)

H assn stated that waqf through its various means by its nature is a
mifsal, separator of speech where it is possible that the chain of speech
may broken into spoken groups. Every one of these, when its meaning
is complete, is considered a speech event. However, if the speech is not
250 salman h. alani
complete then the speech event may consist of more than one event
(H assn 1973, 270).
2. The Arabic linguistic analysis of and rules for waqf
2.1 The waqf status of the t marbtah

One of the most common modes of waqf is applied to words that end with
a t marbtah. This t marbtah is considered to be a morpheme that
primarily marks the feminine endings of nouns and adjectives. However
there are some feminine nouns and adjectives that do not always end
in t marbtah. In Arabic script some feminine nouns and adjectives
may be written with t mabstah. This is the regular t that appears at
the end of words. This is especially true in the script of the Qurn. In
fact we find sometimes the same word written with t marbtah and in
other contexts it is written with a t mabstah. The word rah ma mercy
for example is written in either t marbtah or t mabstah.
There are nouns and adjectives that end in a t marbtah that have a
masculine meaning. Proper names such as H amzah, Talh ah, and several
others are written with t marbtah. The rules of waqf apply to them as
they apply to the feminine nouns and adjectives. Therefore it is not the
gender of the word but rather the form of the word that determines the
application of the waqf rules.
The t marbtah in nouns and adjectives that appears in prepause
form is deleted and replaced by /-h/. This deletion, of course, includes
the vowels: /-i/, -/u/ and /-a/ in the definite and /-in/, /-un/ and /-an/
in the indefinite. The deletion takes place in all three cases: nominative,
accusative and genitive. With prejunctural, on the other hand, the t
marbtah and the vowels both in the definite and indefinite are retained.
This is illustrated with the word madrasah school as in the prejunctural
forms below:
Nominative Genitive Accusative

Indefinite madrasa-tun madrasa-tin madrasa-tan
Definite al-madrasa-tu al-madrasa-ti al-madrasa-ta
The case endings in the indefinite /-tun, -tin, -tan/ and the definite /-tu,
-ti, -ta/ all will be deleted in the prepausal forms. The word madrasah
school and the word al-madrasah the school both in definite and
indefinite forms will have the same endings in the prepausal forms. The
reason is that the /-h/, placed between slashes, is sometimes weakened to
the point that one really cannot even hear it. This is especially the case of
Modern Standard Arabic, read aloud or spoken by radio and television
broadcasters. I have examined and analyzed the speech segments of
several announcers and observed both the dropping off of the /-h/ and
the retaining of it. In careful delivered speeches especially of religious
nature the /-h/ is almost always retained.
2.2 The deletion of final short vowels

The final short vowels, in waqf, are deleted. This deals with al-h arakti
l-irbiyyah, the vowel marks that signal case endings, tense and mood.
The following rules apply to nouns and adjectives derived from strong
verb roots:
(a) The case endings in the definite nouns are indicated by /-u/ in the
nominative case, /-i/ in the genitive case, and /-a/ in the accusative
case. All of these short vowels that mark the case endings are omit-
ted in waqf. The following sentences illustrate the prejunctural and
prepausal forms:
Prejunctural forms Prepausal forms Meaning

ja-l-walad+u ja-l-walad the boy came
mina-l-walad+i mina-l-walad from the boy
raaytu-l-walad+a raaytu-l-walad I saw the boy
(b) When the indefinite nouns and adjectives that end in nunation,
the sound /-n/ pronounced but not written. The case endings in the
indefinite nouns and adjectives are indicated by /-un/ in the nom-
inative case, /-in/ in the genitive case and /-an/ in the accusative
case. Both nominative and genitive endings are omitted in prep-
ausal forms. However the accusative case maker is changed into an
alif /-/ that requires a special treatment. The word walad boy in
the following sentences illustrates both prejunctural and prepausal
forms.
252 salman h. alani

ja walad+un ja walad a boy came
min walad+in min walad from a boy
raaytu walad+an raaytu walad+aa I saw a boy
(c) The indefinite accusative case ending is /-an/. Also certain types of
adverbs end in /-an/. This ending is replaced by /-/ in prepausal
forms as illustrated in the table above. In the case of adverbs some
speakers retain the nunation while others replaced it with /-/.
However, in prepausal and words in isolation may be retained but
oftentimes it changes to /-/ following the normal rules of waqf. The
word id an meaning if is written orthographically either with nn
/n/ or with the alif tanwn. In prepausal form pronounced as /-/.
(d) Tense and mood in verb-vowel endings are omitted.

katab+a katab he wrote
yaktub+u yaktub he writes
lan yaktub+a lan yktub he will not write
2.3 Words that end in long vowels

There are two main grammatical categories of words that end in long
vowels. These are called in traditional Arabic grammar as the al-manqs
defective as in the word al-wd the valley and al-maqsr shortened
as in the word al-fat the youth.
1. The rules of waqf on words that end in the long vowel /-/ such as
al-muh m the lawyer which is derived from a finally weak verb h am
and its imperfect yah m to defend ends in /y/. Nouns that are derived
from finally weak verbs like h am of the pattern of al-muh m always end
/-/. This category of words should not be confused with words ending in
y-n-nisbah as in words like lubnniyyun Lebanese or even with words
like zabyun deer both of these words are written in Arabic script with
/y/, however the source of this /y/ is not a final radical as the nouns
derived from finally weak verbs like h am and its imperfect yah m to
defend which ends in /y/ as a radical (Farrj, 2001, 7172).
The vowel endings of al-manqs defective are determined by the case
endings and whether they are definite or indefinite. When words of the
manqs defective are in the accusative case and indefinite they end in
/-/ in the waqf as qbaltu muh my I met a lawyer after the omitting
of the nunation. When they are definite they end in /-/ as in qbaltu
l-muh m I met the lawyer. When the manqs words are in genitive or
nominative case they end in the omitting of the final vowel /-/ as in fil-
wdin the valley and hd wd this is a valley. However when they are
definite the words of the manqs end in long vowel /-/ as in fil-wd in
the valley and hd l- wd this is the valley.
2. The rules of waqf of words that end in the long vowel /-/ such as
al-fat the youth and which are primarily derived from finally weak
verbs are called in traditional Arabic grammar al-maqsr the shortened.
The waqf on these words is always end in an alif /-/ in all three case
endings and regardless how orthographically they are written with alif
maqsrah or regular alif. What matters here is the pronunciation and
not the script.
3. Concluding Remark
It is worthwhile to mention that the rules of waqf are not always adhered
to by readers and speakers of Modern Standard Arabic. The Arabic
language for sometime has been going through processes of change.
Some readers and speakers are not using the al-h arakta-l-irbiyyah
case endings. The often heard statement that states sakkin taslam use
sukn and you will be safe reflects the state of affairs of the on going
change of the Arabic language. Therefore the rules of waqf outlined
above when considered should be used as guidelines not to be applied
in a rigorous and strict manner.
254 salman h. alani
4. References
4.1 Primary source

Ibn Ya. n.d. arh al-mufassa l. Cairo: Maktabat al-Mutanabb.

Farrj, Muhammad. 2001. Al-waqf wa-wazifuhu inda n-nahwiyyn wa-l-qurr. Annals
of Arts and Social sciences, Monograph 159 Volume 21. Kuwait: Kuwait University.
H assn, Tammm. 1973. Al-lua al-arabiyya mabnh wa-manh. Cairo: al-Haya
al-mma lil-Kitb.
Holes, Clive. 2004. Modern Arabic Structures, Functions and Varieties. Revised edition.
Georgetown, Washington, D.C.: Georgetown University press.
Mitchell, T.F. 1990. Pronouncing Arabic. Oxford: Clarendon Press.
THE EXPLANATION OF HOMONYMY1 IN THE LEXICON
OF ARABIC
Georges Bohas and Abderrahim Saguer

ENS Lettres et Sciences Humaines, Lyon
1. Introduction: the theoretical framework and earlier analyses
This article is a follow-up to our previous paper entitled Sur un point de

vue heuristique concernant lhomonymie dans le lexique de larabe.2 By
adopting a heuristic viewpoint, we take into account the fact that we have
not yet finished exploring all the matrices of Arabic, and thus proceed
by means of successive evaluations and provisional hypotheses.3 Certain
points remain to be clarified; nevertheless, the level of explanation we
can achieve has become clearer, as well as the explanatory methods that
we are able to offer within the framework of the Theory of Matrices and
Etymons (TME). Although there are already numerous publications on
this subject, it is worth underlining that, within the TME framework,
the lexicon is organized on three levels:4
1 / The matrix: a non linearly-ordered combination of a pair of pho-

netic feature vectors linked to a notional invariant; for example,
{[labial], [coronal]} to strike a blow.5
1
It is worth defining homonymy in opposition to polysemy. Polysemy is a word
which brings together several meanings between which users can recognize a link
(Nyckees, 1998: 194); the meanings are different but related. Homonymy is distinct
from polysemy in that, in the case of homonymy, it seems impossible to re-establish a
plausible semantic relationship (Nyckees, 1998: 194) between the different meanings,
for example: flies certain insects and flies the opening at the front of a pair of trousers
or to sound to make a noise and to sound to measure the depth of waterdifferent non
related meanings.
2
Bohas and Saguer (2006).
3
That is: in a not-rigorously demonstrated manner but justified by reasons of inter-
nal coherence (see the website www.memo.fr Einstein, Albert); and accepting that you
cannot explain everything.
4
See Bohas (1997, 2000), Dat (2002).
5
This is a property of the language that was proved both formally and semantically
by Bohas and Darfouf (1993), developed in Bohas (1997), which consists in the fact that
256 georges bohas and abderrahim saguer
2 / The etymon: a non linearly-ordered bi-consonantal base made up

of two phonemes taken from a given matrix exhibiting both the fea-
tures of this matrix and its notional invariant; for example, {b,t} to
strike a blow with a sharp object.
3 / The radical: an etymon that has developed by diffusion of the final
consonant or by incrementation, or that results from the blending of
two etymons; the radical includes at least one vowel and vectors the
notional invariant; for example: /bvtar/, to cut, to cut the tail (Bohas
2000, 9). The radical is the domain in which diverse morphological
and Ablaut processes take place (Guerssel and Lowenstamm 1993,
Segeral 1995).
So far, ten matrices have been accounted for; most of these have already
been subjected to in-depth studies (see footnote 14).
Matrix 1 {[labial],6 [coronal]}

Notional invariant: to strike a blow
Matrix 2 {[labial], [-voiced] }7
[+continuant]8
Notional invariant: movement of air
Matrix 3 {[labial], [pharyngeal]}9
Notional invariant: (a) tightening
Matrix 4 {[coronal], [pharyngeal]}
[-dorsal]10
[-voiced]
a binary combination {a, b} is realized in the order a+b and in the order b+a while keep-
ing the same notional invariant.
6
[labial] characterizes sounds produced with a constriction of the lips. For matrices
1, 2, 3, 6 we integrate on-going research which shows that the feature [labial] should not
be restricted by [-sonorant] (see Mansouri, 2006).
7
[voiced] Sounds produced with vibration of the vocal cords are said to be voiced
([+voiced]), whereas other sounds are said to be non-voiced ([-voiced]), see Dell (1973:
56).
8
[continuant] Sounds with the feature [+continuant] are produced without inter-
rupting the flow of air through the oral cavity, those with the feature [-continuant] are
produced with total interruption of the flow of air at the oral cavity, see Halle (1991:
208).
9
[pharyngeal] characterizes segments that the Arabic tradition calls gutturals, that
is: , h, , h , x, and q. For the problems posed by the characterization of this class, see
Kenstowicz (1994: 456ff).
10
[dorsal] characterizes sounds produced with a constriction created with the back
of the tongue between the soft palate and the uvula (velar and uvular consonants; rear
vowels).
explanation of homonymy in the lexicon of arabic 257
Notional invariant: low voice, muffled, hoarse noise

Matrix 5 {[coronal], [dorsal]}
Notional invariant: to strike a blow11
Matrix 6 {[labial], [dorsal]}
Notional invariant: curvature
Matrix 7 {[dorsal], [pharyngeal]}
Notional invariant: the cries of animals
Matrix 8 {[+sonorant12], [+continuant]}
[+lateral13]
Notional invariant: the tongue
Matrix 9 {[+nasal], [+continuant]}
Notional invariant: the nose
Matrix 10 {[+nasal], [coronal]}
Notional invariant: traction14
The data on which we have based our study are to be found in the
Kazimirski, and have been checked in the Qms and/or the Lisn.
When they are based on another source, this is mentioned.
2. Explanatory methods
In the paper quoted above, we demonstrated that the homonymy of a

radical may be attributed to three causes:
A. the fact that it is the result of blending: it manifests the meanings of

both the etymons that are its source.
11
See Diab (2005) who brings a modification to the formulation of the notional
invariant of this matrix.
12
[sonorant] Sounds with the feature [+sonorant] are produced with a constriction
which does not influence the capacity of the vocal cords to vibrate spontaneously. Those
with the feature [-sonorant] have a constriction which reduces the global flow of air and
makes voicing more difficult. Thus the natural state for sonorants is [+voiced] and for
non sonorants (termed obstruents) is [-voiced], see Kenstowicz (1994: 36).
13
[lateral] A [+lateral] sound is produced by making a constriction with the central
part of the tongue while lowering one or both lateral edges so that air escapes around the
side(s) of the mouth, see Kenstowicz (1994: 35).
14
For matrices 1 to 6, see Bohas (2000), Dat (2002), for an in-depth study of matrix 6,
see Serhane (2003), Bohas and Serhane (2003), for matrix 7, see Bohas and Dat (2005),
for matrices 8 and 9, see Bohas (to be published) and for matrix 10, see Saguer (2003).
B. the fact that its etymon is the realization of several matrices: it mani-
fests the meanings of these matrices.
C. the fact that two etymonial analyses are possible, such as [nX]Y and
n[XY].
Below, we illustrate each case with an example from Bohas and Saguer
(2006).
A. Homonymy resulting from blending

Let us consider the verb araza, which attests two meanings (hereafter
senses):
S1 to prick something with a needle, to drive in, to plunge (a sharp
instrument), to plunge a tail into the ground to lay eggs (of
locusts);
S2 to give but very little milk (of pregnant camel).
The same semantic load is found in rizun;
S1 that drives in, plunges a sharp instrument, a goad into something;
that plunges a tail into the ground to lay eggs (of locusts);
S2 that gives but little milk (camel).
Since it is not possible to establish a plausible semantic relationship

between the two senses, this is an obvious case of homonymy. And yet
we observe the existence of the following words:
arra15 F. III : to be found in small quantities (of milk of a female);
irrun : a small quantity, generally, such as a small quantity of
milk in a female
murrun : that has little milk in the udders (camel)
Phonetically, the etymonial analysis can only be { ,r}, since their radical
has no other consonant,16 and it is obvious that they attest sense S2.
Moreover, the precise meaning of the verb razza is: to plunge a tail into
the ground to lay eggs (of locusts); to stick, drive in and fix firmly one
object into another or into the ground. Thus it clearly attests sense S1
and is analysed as the etymon {r,z}.17 Therefore, the explanation is that
araza includes senses S1 and S2 because it results from blending of the
15
We use boldface for the segments that make up the etymon.
16
We call these non ambiguous radicals.
17
Non ambiguous radical.
two etymons {,r} and {r,z}. The way this blending occurs is represented
in model A:18
A
Cj Ci Ci Ck
Si Sj
Cj Ci Ck19
Si + SJ
More explicitly :
r rz
lack of milk x20 to drive a sharp object into
Si Sj
araz 1920
Henceforth, in similar cases, we shall talk of explanation through

blending.
B. Homonymy through the realization of several matrices

1) An easy case
Let us take the verb mata"a, which means:
S1 to strike somebody with a stick
S2 to tighten, stretch out a rope
The m is [labial], the t is [coronal]: the etymon {m,t} can thus be a real-
ization of matrix 1:
{[labial], [coronal]}
And for this reason assumes the sense S1 to strike somebody with a
stick.
But the m is also [nasal] and the etymon {m,t} can also be a realization
of matrix 10:
18
See Bohas (1997: 175f and 2000: 49).
19
The obligatory contour principle (OCP) explains the fusion of the two Ci into a
single segment. See McCarthy (1986) for the definition and various applied examples
referring to Semitic languages of this principle which forbids adjacent identical elements
at the same level. Since, OCP has given rise to a multitude of studies the list of which
would be superfluous here.
20
We use x to indicate blending.
{[+nasal], [coronal]}
Notional invariant: traction
And for this reason assumes the sense S2 to tighten, stretch out a
rope.
2) A more complex case
The verb natara attests the following meanings:

S1 to disperse
F. I : to scatter, disperse, disseminate
F. II : to scatter a lot, in large quantities: intensive of F. I
F. V : to be scattered, dispersed, to disperse
F. VI : to be scattered, dispersed, to disseminate, to spread
S2 actions concerning the nose
F. I natura : to blow ones nose
F. II : to blow ones nose
F. II : to draw up water through the nostrils
F. VIII : to blow ones nose
F. VIII : to draw up water etc. through the nostrils and to
expel it through the nostrils
S3 to pull, tear off/out
F. I : to remove, to take the clothes off the body of some-
body, to strip
S4 to strike a blow with a sharp object21
F. IV : to pierce somebody with a sharp instrument and to
make the blood flow
The first hypothesis we can formulate is that natara develops the ety-
mon: {n,t} and that for this reason, it is a realization of matrix 9:22
{[+nasal] [+continuant]}
Notional invariant: the nose
The phonetic substance of this matrix comprises, on the one hand, the
two nasals, m and n, and, on the other, the various fricatives.
21
We will propose no explanation for this sense; as we said in the introduction, we
have not yet explored all the matrices of Arabic, the notional invariant sharp is without
doubt important but as yet we know nothing of it.
22
See Bohas (to be published) for a detailed study of this matrix.
The ramifications of the notional invariant are as follows:

1. the nose
1.1. the organ itself and what affects it
1.2. the specification of parts (the top, the sides)
1.3. to be sharp, > protruding, > to precede
2.1. specifications of the organ (big, small . . .)
2.2. animal or human presenting these specifications
3. to lift the nose: movement of pride or contempt
4. the nose and air: to breathe in; to breathe out; to perceive odours,
to smell
5. the influence of the nose on the voice: nasal sound; similar animal
cries (buzzing-grunting)
6. various liquids (mucus, phlegm) which pass through the nose
We may note that, in this organization, senses S2 to blow ones nose,
to draw up water through the nostrils, fit into both 6. and 4.
But, since n is [+nasal] and t [coronal], {n,t} can also be a realization of

matrix 10:
{[+nasal], [+coronal]}
Notional invariant: traction
Moreover, other realizations of the {n,t} etymon can be found in:
natala [nt]l : to remove the garment or breastplate
nanaa [n]n : to remove (ones clothes)
Sense S3 to pull, tear off/out of F. I: to remove, take the clothes off the
body of somebody, to strip is thus incorporated into this matrix.
We can therefore understand why the {n,t} etymon is homonymic: if we
take into account the t the feature [coronal], then we see that it realizes
matrix 10 traction, while if we take into account the feature [+continu-
ant], then we see it realizes matrix 9 the nose.
In this case we shall speak of ambiguity originating from the fact that
an etymon is the realization of several matrices.
C. Homonymy due to the possibility of two etymonial analyses

Let us pursue the analysis of natara. We have seen that it also attests the
sense to disperse, to scatter. It is true that we can establish both a pho-
netic and a semantic relation with:
tarra [tr]r : to disperse, disseminate
the etymon of which can only be {t,r},23 and with:
23
No ambiguity.
tra t[w]r : to be lifted and scatter in the air

tartaratun [tr]tr : dispersion, dissemination.
farata f[rt] : to be dispersed, disseminated (of a tribe)
This leads us to analyse natara by positing {t,r} as etymon and the n as
an initial crement. Therefore, the homonymy stems from the fact that in
A and B the form is analysed as [nt]r, which may have two matrix links,
and in C as n[tr]. We shall say in the latter case that homonymy arises
because several etymonial analyses are possible.
3. The two levels of explanation
In Bohas and Saguer (2006) we studied cases concerning radicals

containing an n. In this paper, we shall extend the study to include the
analysis of radicals containing an l. According to previous studies,24 l may
have the status of initial crement/prefix25 or of matrix segment. Earlier,
Hurwitz (1913, 5560) had already recognized this prefix status:
The preformatives are thus seen to possess a fairly definite, though remote,
relationship to each other. The sibilants and gutturals are to be traced to
causative stems; the dental t and liquid n to reflexive stems; the liquids m,
l, and i are to be connected etymologically with the reflexive n, and the
preformative y may be considered to be a denominative stem.
When it has a matrix function, l may form an etymon with the second
radical: [lx]y, or the third radical: l[x]y.26 In other words, a radical [lvxy]
may present all its possibilities, as can be seen in the table below:
Etymon: {x,y} {l,x} {l,y}
Analysis: l[xy] [lx]y l[x]y
Radical: <lvxy>
24
The most recent are those of Saguer (2000, 2002, 2002b).
25
We use prefix when l has a semantic-grammatical value and initial crement when
it does not.
26
This expresses the fact that in this form the etymon is [ly] and x is an inset
crement.
3.1 The explanation through attribution to identified matrices

The verb labaxa demonstrates eight senses:
S1 to be fleshy (of the body)
S2 to beat, strike somebody
S3 to kill somebody
S4 to take a thing from somebody, from the hand of somebody
S5 to get something out of somebody using trickery
S6 F. III: to slap somebody in the face
S7 F. V: to perfume oneself with musk
S8 to say something foolish, insulting to somebody
It should be noted that several senses are polysemically related; in other

words, a plausible semantic relation can be established between them,
as follows:
S2, S6 and S3 are manifestations of the notional invariant to

strike a blow; to kill is related to this concept through the
cause>consequence relationship and S6 specifies the mode of
action.
S4 and S5 come under the notional invariant: to pull, bring some-
thing to oneself.
Thus five homonymic senses remain:
A S1 to be fleshy (of the body)

B S2+S6+S3 to strike a blow
S2 to beat, strike somebody
S6 F. III: to slap somebody in the face
S3 to kill somebody
C S4+S5 to pull, bring something to oneself
S4 to take a thing from somebody, from the hand of some-
body
S5 to get something out of somebody using trickery
D S7 F.V: to perfume oneself with musk
E S8 to say something foolish, insulting to somebody
We are going to attribute each of these homonymic senses to a source

matrix.
Take sense A S1labaxa: to be fleshy (of the body).
It is easily noted that this form is semantically linked to:

rabxun r[bx] : fat, thick and with a soft and loose body
habayyaxun h[bx] : young chubby
anbaxun [bx] : fat, thick, hard
muxabxabat [xb] xb : handsome and plump (camels)
This leads us to deduce that labaxa should be analysed as a form
incorporating the etymon {b,x} with initial incrementation of an l
which has the role of a prefix denoting the sense we have called static,
following Joon (1923, 95). This is defined as the appropriation by the
subject of the quality or the state X. The Hebrew examples cited by Joon
are: kbed he is heavy, qton: he is small (1923, 95).
This etymon {b,x} is a realization of the matrix 6 {[labial], [dorsal]}
which has curvature as its notional invariant. One of the rst mani-
festations of curvature in the convex form is precisely the concept of
fatness:27 fat, plump, robust, that which realizes a swollen form: , as
in:
bajja -F. VII : to be plump, to have rounded flanks (of animals fat-
tened by pasture)
bjilun : fat, replete
fajia : to have a fat belly
hijaffun : who has a fat belly, paunch
For sense A, labaxa to be fleshy (of the body) is analysed as l[bx] and is
thus a realization of the matrix {[labial], [dorsal]} which has the notional
invariant curvature.
Let us consider sense B S2+S6+S3 to strike a blow; S2 to beat,

strike somebody; S6 F. III: to slap somebody in the face; S3 to
kill somebody. It is easy to establish a link between the form and other
manifestations of the etymon {l,b} such as:
wabala w[bl] : to strike somebody with a stick
labana [lb]n : to strike somebody violently, to batter somebody to
death with a stick
lah aba l[h ]b : to strike somebody with a sabre
and for the implied sense:
habila h[bl] : to lose ones son through death
27
See Bohas (2000: 108).
This enables us to establish that labaxa includes the etymon {l,b}, itself a
realization of the matrix {[labial], [coronal]} which has to strike a blow
as its notional invariant.
However, note the existence of the following set:
laxxa [lx]x : to slap somebody in the face
laxaba [lx]b : to slap somebody in the face
laxama [lx]m : to strike somebody in the face
This enables us to extract the etymon {l,x} connected, once again, to
the sense to strike a blow. The segment l is analysed as [coronal} and x
as [dorsal], so that this etymon is a realization of the matrix {[coronal],
[dorsal]}, which leads us to the conclusion that labaxa in the B sense
results from blending between two etymons {l,b} and {l,x}, a B type
blending (Bohas 2000, 50).
For sense C. = S4.+S5. to pull, bring something to oneself : S4. to
take a thing from somebody, from the hand of somebody; S5. to get
something out of somebody using trickery we shall justify an etymonial
analysis l[b]x, that is, an etymon {l,x} with an inset crement b. Indeed, a
phono-semantic relation can be established with:
In the order xl:
xalaja : to pull, attract to oneself
xal (hly) : to pull out
xalxala : to completely strip a bone of its flesh
saxala : to take, remove, ravish through trickery
xalasa : to take away, remove ravish in an instant, unexpect-
edly, to pull out
F. III : to pull something from somebody; to seize; to grab
something
F. V : to remove, take away
F. VI : to reciprocally pull out, and to pull something each
from his own side
F.VIII : to pull something towards oneself in a hurry (the same
meaning as the first item of the list with the addition
of celerity).
In the order lx :
malaxa : to forcibly pull something towards oneself seizing it
with ones teeth or hands
F. V. : to pull out, to burst e.g. the eye of a prey (birds of
prey)
F.VIII : to pull, extricate, pull out (a tooth, an eye); to pull out
of the scabbard (a sabre, etc.)
Therefore the etymon {l,x} carries two homonymic senses: to strike a

blow, as a realization of the matrix {[coronal], [dorsal]} and to bring
something to oneself as a realization of another matrix, which remains
to be studied and, we suspect, includes the features:
{[+approximant28], [+continuant]}
[coronal]
with the notional invariant: to bring something to oneself
Besides the words above, this matrix manifests itself in:
Etymons with l
salla : to pull, extract gently one object from another
salaba : to snatch something with force from somebody
salata : to extract, to pull
salaxa : to skin, to remove the skin from a sheep; to remove
ones clothes and shirt
salaba : to pull, to extract the marrow from bones
sahala : to peel, to strip bark, skin
halaba : to pull out hair, bristles, horsehair
halata : to peel something by removing the skin, bark
halada : to pull, extract, pull out something from its place
lah : to remove the inner bark of a tree, wood
lahaba : to remove the bark from a piece of wood, to strip it of
its bark
lahata : to remove the bark from a piece of wood, to peel it
laxasa F. II : to remove, to extract the purest part
halaa : to remove the flesh from the skin of a dead or bled
white animal
halaba : to take the milk, to milk
halata : to pull out, to remove in flakes
halama : to remove tinea from the skin
alaha F. II : to strip, to remove the clothes from somebody
28
This composition, [+approximant], seems too complex to characterize the class r, l.
[coronal]
This is due to the fact that, according to Yeou and Maeda (1994), the pharyngeals and
uvulars of Arabic are also characterized by the feature [+approximant]. Indeed, the con-
tinuation of our studies will enable a discussion of this point. If the gutturals do figure
in this matrix, and assuming that it may then be formulated simply as {[+approximant],
[+continuant]}, this would constitute proof that the gutturals and r, l are indeed mem-
bers of the same class: [+approximant].
Etymons with r
haraba : to strip somebody, to pillage (a caravan, a tribe)
harasa : to steal something; in fields, pastures
harada : to take, by milking, all the milk from a camel
saraqa : to steal
sar : to remove, separate, distance something from some-
body
xaraa : to attract a camel to oneself with a hooked stick
xarata : to strip of bark and make even
xarafa : to pick a piece of fruit from a tree
'araza : to pull with force
'ar'ara : to move the cork of a bottle to uncork it; to remove,
take out the cork; to pull, to pull out, to burst an eye
'araqa : to strip a bone of the flesh that was on it by eating it
'arama : to eat the flesh that sticks to the bone
'ariy : to be naked, stripped of ones clothes (cause> conse-
quence relation)
qa'ara : to pull out with the root, from top to bottom and to
cause to fall
This abundance of data, the notional domain of which has still to be
organized, is nevertheless sufficient to justify the existence of this
matrix:
[+approximant], [+continuant]
[coronal]
to bring something to oneself
This data also justifies the analysis that posits that in labaxa sense C =
S4+S5 to pull, bring something to oneself : S4 to take a thing from
somebody, from the hand of somebody; S5 to get something out of
somebody using trickery is a manifestation of this matrix.
As to labaxa, D S7 F.V to perfume oneself with musk it is analysed as

l[bx] in which the etymon {b,x} is found, as in:
baxara [bx]r : to perfume somebody or something with incense
baxira : to smell bad
abxaru : who has unpleasant, fetid breath
baxxa [bx]x : to snore during sleep
This etymon is a realization of matrix 2.

{[labial], [-voiced]}
[+continuant]
which combines the labials b and f with voiceless fricatives.

The ramifications of the notional invariant of the conceptual field
are:
movement of air: wind, breath

expulsion of air in man or animal
29
consequences (various odours)
Below we provide some realizations of the matrix with f:
{ ft}
nafata : to blow (on something)
{ fh}
fahh a : to hiss (snake); to hiss during sleep
fahfah a : to be hoarse
fah -F. II : to season food
fha/fawah a/ : to spread ones perfume; to smell good or bad
lafaha : to blow (of warm wind)
nafaha : to spread ones odour; to blow (of cold wind)
{ fx}
faxxa : to snore and hiss talking of somebody who sleeps;
to spread (of an aroma)
fxa/fawaxa/ : to spread (of an odour); to hiss (wind); to release
wind (of a man)
nafaxa : to blow with the mouth; to break wind
Finally, sense: E S8 to say something foolish, insulting to somebody.

Note that labaxa commences with an l [+lateral] and nishes with a x
which is a [+continuant] segment. The analysis is thus l[b]x, of the
etymon {l,x}, which is also present in:
laxiya : to be very talkative and say a lot of rubbish
xatila : to talk a lot and to say only trivialities
In this case, this etymon is a realization of matrix 8, exactly like:
{l, }
la : to speak, in general to say futile things, to utter vain,
flippant or careless words, to make mistakes in talk-
ing, to make an error
29
This sign indicates that a semantic relationship exists, here: cause>consequence.
laiya : to make an error talking, to be mistaken

F. X : to gather locutions, idioms, or to pay attention to
words and locutions (lut), particularly of nomadic
Arabs, i.e. to draw from them knowledge of Arabic
words
Matrix 8 {[+sonorant], [+continuant]}
[+lateral]
Notional invariant: tongue
As we shall often be returning to the organization of the notional field of

this matrix, it is worthwhile examining it here:30
The ramifications of the notional invariant include:
0. The tongue and its characteristics
1. The tongue and the physical actions that are its peculiarities
1.1. to make an action of the tongue
1.2. to seize, to pull with the tongue
1.3. to lick
1.3.1. consequence (1): to moisten and stick
1.3.2. consequence (2): to be smooth, polished
1.4. to savour, to taste
2. The tongue as an instrument of language: to speak, to speak in various
ways, to be talkative, to malign, to wrestle in words with somebody, to
hurt with malevolent words; to speak with authority > to order.
3. The tip of the tongue: pointed, to be sharp, to become a point and
thus to prick. Here we have a relationship of the type part/whole:
only the tip of the tongue is taken into consideration. Note a possible
interference with 2. he who has a pointed tongue is likely to utter
hurtful remarks. We link to 3. the cases in which the pointed charac-
teristic is explicitly given as cause.
4. Tongue of fire: to blaze, to burn, to scorch.
Consequently, all the senses expressed by this radical are made explicit
in the complex diagram below which constitutes its lexicogenetic tree,
and traces its phono-semantic composition, thus accounting for the
homonymy :
30
This matrix is studied in detail in Bohas (2006).
Matrix level
M6 M2 M1 M5 M8
{[labial], {[labial], {[labial], {[coronal], {[+lateral],

[dorsal]} [+continuant]} [coronal]} [dorsal]} [+continuant]}
curvature air> odour to strike a blow to strike a blow the tongue
Etymoniall evel
{b,x}1 {b,x}2 {b,l} {l,x}1 {l,x}2
l[bx] [lb]x l[b]x

Radicall evel
labax
Once this tree is exhaustively constructed, the analysis is finished. In the

present case, we have managed to link each sense to a matrix. This is not
always possible.
3.2 The explanation through attachment to etymons

In other cases, the analysis only enables the etymons to be identified
without reaching the matrix level. Consider the verb laata which has
the senses:
S1 to hurt, to strike, to injure, to harm somebody (either by shoot-
ing an arrow at him, or by looking at him with the bad eye
S2 to print on an animal, on the neck, a mark with hot iron, to
mark it
S3 to hurry, to hasten
S4 to delay paying a debt, to put off its payment
S5 to go to pasture (livestock)
Each sense is homonymic in relation to the others, but, in each case,

it is possible to establish a phono-semantic relation with other words,
enabling us to identify etymons, as we shall demonstrate.
For S1 to hurt, to strike, to injure, to harm somebody (either by shoot-

ing an arrow at him or by looking at him with the bad eye. Taking into
account:
laata l[]t : to touch, attain somebody, e.g. with an arrow

lahat l[h]t : to strike, to attain somebody with an arrow
lta l[w]t : to strike, to attain somebody with an arrow or by look-
ing at him with the bad eye
we may establish that the three verbs all demonstrate the phonetic con-
stant lt and very similar meanings (indeed for laata and lta, the mean-
ing is identical), enabling the identification of an etymon {l,t}.
For S2, compare laata to print on an animal, on the neck, a mark with
hot iron, to mark it with the elements of the following paradigm:
'alata : to mark (a camel) on the neck with a transversal
mark
'alama : to mark, to distinguish by a mark, by some sign
'alaba : to mark something, either by making incisions or by
pressure
ra'ala F. II : to incise the ear of a beast to mark it
This enables us to establish the presence of a phonetic constant l and
a semantic constant to mark, thus identifying the etymon {l,' }. Let us
add the form:
laxafa : to print a large cautery, to make a large burn on some-
body
in which lx constitutes either a variant31 of l or a clue to identifying a
matrix.
For S3, it is worth comparing la'atato hurry, to hasten with:

'at F. II [' t]w : to hurry somebody, to tell him to hurry
'abata ' [b]t : to throw a horse into the race so as to make it
sweat
hata'a h[t' ] : to go, to advance rapidly and by lengthening the
neck (of camel)
'atawwad ['t]d : rapid, hurried and tiring (voyage)
This enables us to identify the phonetic constant ,t and the semantic
constant to go rapidly, which leads us to deduce that laata should be
analysed as an etymon {',t} with the incrementation prefix l bearing the
reflexive meaning.32 Here again, note the variant:
xta : to pass rapidly
31
We explain this notion the following pages.
32
Which corroborates Hurwitz earlier quoted in subsection 3.
We may relate S4 laata: to delay paying a debt, to put off its payment
to:
tla t[w]l F. II : to allow ones debtor an extension
tal [tl]w : to wait, to be waiting, to defer
talla [tl]l : to allow an extension, to give respite to ones
debtor
latat [lt]t : refusal to pay or to recognize what one owes to
somebody
The phono-semantic constant presented by the above data enables us to
identify the etymon {l,t}.
Note that S3 and S4 are two contradictory senses: to go rapidly and

to delay. This enantiosemy33 can be explained by the fact that the word
is analysed as a blending of two etymons with opposing senses: lt to
delay and x t to go rapidly (D type blending: (Khatef, 2003 and 2004)).
This enantiosemy is homonymic, since the two senses have nothing in
common. The enantiosemy, which is usually presented as a quirk of
Arabic is, in fact, trivially predicted by TME (for more examples, see
Bahri 2003).
Finally, for laata S5 to go to pasture (livestock) a semantic link may

be established with:
ra' : to graze in this or that place; to go to graze freely;
to put to graze, to take to graze
rata'a F. IV : to leave to graze freely
'ra : to go away, to move away (of, amongst others, a
horse that goes off grazing here and there)
raba'a : to graze freely
These forms are obviously related to the etymon {r,' } linked to the sense
to go to graze (freely). It seems plausible that the etymon l from laata
S5 to go to pasture (livestock) is an allophone of this etymon, r and
l being from the same class [+approximant] [coronal]. The definition
of the allophones of etymons is set out in Bohas and Dat (2007):
33
We use enantiosemy (i.e. reverse semantics) for words that mean something and its
opposite, such as big and small, to rent which means both to take temporary possession
for the payment of a fee and to give temporary possession for the receipt of a fee.
Phonetic evolution may provoke a modification of the signifier and bring

about the appearance of what we will call etymon allophones, which are
phonetic variants of matrix etymons (from the set of a given matrix).
Most frequently, the etymon allophones bring into play acoustic fac-
tors, which explains the confusion of segments in the communicational
process. We posit that if:
[b] / {a , _} and [c] / {a , _}
where {a,b} is a matrix etymon and [b] and [c] are phonemes with
one or several shared phonetic features other than the vector of features
demanded by the matrix combination, and that they correspond to two
lexical items that are conceptually related, linkable (not necessarily identi-
cal), then [b] and [c] are the free variants of the phoneme (belonging to
the paradigm defined by the vector of features) that enters into the com-
position of the matrix etymon.
The allophonic forms, which enlarge the logical number of etymons belong-
ing to a binary matrix of features, characterize the etymons, enlarged or
not, the articulation of which is weakened34 or loosened in verbal com-
munication; these allophones are, over time, recovered and incorporated
into the lexicon of the language. These are free phonetic variantshistor-
ical and/or dialectalof successful (widespread) innovations that co-exist
with the source-forms, such as the elements in the following list (from the
lexicon of Hebrew).
s s / /
ntas : to demolish, to knock over, to knock down, to

pull out
ntas : to break, to destroy
sdh Niph. (hapax) : to be desolated, ravaged
dad Pi. : to break up lumps of soil, to harrow, to level a
piece of land
dad : exercise violence, to desolate, to wreck, to
destroy, to devastate
In all these examples, the existence of the minimal pair engenders no
major lexical opposition, the semic difference often resulting from the
translation.
This argumentation enables us to consider laata S5 to go to pasture
(livestock) as a realization of the etymon {l,' } allophone of {r,' } given
the (obvious) semantic relation and the phonetic relation we have
established, but, in our analysis of this word, we have yet to trace it to a
clearly identified matrix.
34
The natural tendency of the speaker is to limit effort of his speech and to avoid
sharp shifts in the use of speech organs. (Lipinski, 1997, 186).
laata, which realizes the etymons: {',t}, {l,t} and {l,' }, is therefore
typical of the second level of explanation, namely identification of the
etymons. When we go on to study specific cases in the fourth part, we
will move from one of the two levels of explanation to the other. In some
instances we can link the etymon to a matrix; in others we may only link
the radical of a word to an etymon, while in others, we can provide no
analysis for the simple reason that, as things stand, identification of the
matrices and etymons is not yet complete. The objective we pursue is to
fully organize the lexicon into matrices. Is this a chimera, as some do
not fear to say? In our eyes, the best answer is provided by Darwin:
[. . .] it is those who know little, and not those who know much, who so
positively assert that this or that problem will never be solved by science.35
4. Case study
Within the framework defined above, we shall now analyse some

words.
4.1 lah ana

S1 to become fond of somebody, to fall in love with somebody
S2 to speak Arabic badly
S3 to speak a particular slang with somebody so as not to be under-
stood by others
S4 to understand, to hear a word, an expression (link speak/under-
stand)
Let us start by studying sense S1 to become fond of somebody. A

phonetic and semantic relation can be easily established with:
h anna [h n]n : to groan, to make a groan of tenderness (of
certain animals e.g. a female camel showing
tenderness to her young)
nh a n[w]h : to coo, to groan (of pigeons)
h anh ana [h n]h n : to have and to show tenderness, compassion,
with emotion and worry
35
Charles Darwin: The Descent of Man and Selection in Relation to Sex, 4th para-
graph, quoted in Quiniou (2006); for the full text visit www.gutenberg.org.
h an [h n]w : to have great tenderness for somebody (also

used of a mother who through love of her
children, does not want to re-marry)
h aniba [h n]b and F. II : to feel a sentiment of pity, of compassion for
somebody
sah ana s[h n] F. III : to treat somebody with goodness
The above leads us to deduce that, for this sense, lah ana should be
analysed as a form incorporating the etymon [h ,n] through initial
incrementation with an l which acts as a prefix marking the middle
voice.
In the semantic organization of matrices, the starting point is a con-
crete sense. In this we agree with Hurwitz (1913, 72):
It must also be borne in mind that primitive ideas are generally concrete,
and that an abstract idea is secondary, in that it is often based on some
objective aspect involved in the expression of the abstract idea, as when
anger is denoted by a reddening of the face, displeasure, by a falling of the
countenance etc.
The example under study is particularly illuminating. The word h anna
which has the abstract sense to be moved, to have compassion for
somebody, to feel great tenderness for somebody has the matrix
sense to groan, to make a groan of tenderness (of a female camel).
Exactly like anna to groan which comes from the same matrix:
Matrix 4 {[coronal] , [pharyngeal]}
[-dorsal]
[-voiced]
Notional invariant: low voice, muffled, hoarse noise
From the physical groan, as a noise, we pass to why we groan, to what
the groan expresses. In h anna, the two senses are maintained, whereas,
in lah ana, only the abstract sense appears.
We have thus identified the matrix of which lah ana to become fond
of somebody is a realization. Let us now examine senses S2 and S3
which are clearly linked. S2 to speak Arabic badly, S3 to speak a
particular slang with somebody so as not to be understood by others.
For these senses, the analysis [lh ]n is used, in other words, the etymon is
{l,h}, which is itself a realization of Matrix 8 {[+lateral], [+continuant]}
linked to the notional invariant tongue; furthermore, we have already
seen that to speak is one of the developments of the tongue notional
invariant:
2. The tongue as an instrument of language (cf. 3.1 above): to speak,

to speak in various ways, to be talkative, to malign, to wrestle in words
with somebody, to hurt with malevolent words; to speak with author-
ity > to order.36
The same sense is found in:
laxxa [lx]x : to be unintelligible, to speak (especially Ara-
bic) in an unintelligible way
laxlaxniyy [lx]lx : who has difficulty speaking or only speaks
lecherous language
laxiya [lx]y : to be very talkative and say a lot of rubbish
la [l]w : to say futile things, to utter vain, flippant or
careless words
S4 to understand, to hear a word, an expression that can be found

in lah ina with to be intelligent, seems to accept the same analysis. The
causal relation we assume, i.e. speak > understand > be intelligent is,
in fact, explicit in Syriac; in the latter, mll means both who speaks
and who is intelligent or in Greek in which logiks covers two series of
meanings: I. which concerns speech and II. which concerns reason.37
Therefore the lexicogenetic tree of the word may be constructed:
M8 M4
{[+lateral], {[consonantal],
[+continuant} [pharyngeal]}
tongue muffled voice . . .
{l,h} {h,n}
[lh ]n l[h n]
lah an
36
See for the organization of the conceptual field above at the end of subsection 3.1.
37
See Bailly (1950).
4.2 lasaa
Reorganizing the Kazimirski, the following senses can be said to be
included:
S1 to head into the interior of the land
S2 to prick somebody, to bite (of a scorpion or snake)
S3 to hurt somebody with malevolent words, with features of
satire
S4 to malign somebody
S5 F. IV to sow enmity between men
These five senses can be reduced to two:

A to make ones way into the land
B in relation with tongue, as we shall show.
As far as the first (A.) sense (to head into the interior of the land) is
concerned, this form is semantically linked to:
nasaa n[s] : to head into the interior of the land, country
tasaa [ts]38 x [s] : to travel, to cross a country, to head into the interior
of the land
asaa [s] : to be remote, to be situated at a great distance
This fact leads us to deduce that lasaa should be analysed as a form
incorporating the etymon {s,' } through initial incrementation of a
prefix l marking the reflexive meaning, in the same way as the n at the
beginning of the form nasaa.
For the other (B) senses:
S2 to prick somebody, to bite (of a scorpion or snake)
S3 to hurt somebody with malevolent words, with features of
satire
S4 to malign somebody
S5 F. IV to sow enmity between men
The etymon {l,s} that the radical develops is a realization of Matrix 8
organized around the notional field tongue as was earlier discussed at
the end of subsection 3.1.
Of these four senses,
S2 to prick somebody, to bite (of a scorpion or snake) is linked to S3.,
like:
lasaba : to prick somebody (said properly of a snake)
38
Tassa: F. II: to head into the interior of the land, country. The radical thus results
from the blending of the two etymons.
S3 to hurt somebody with malevolent words, with features of satire is

linked to S2 or S3, as is:
ladaga : to hurt somebody with ones tongue, i.e. with a biting
remark
S4 to malign somebody is linked to S2, like:
lasana : to give blows with the tongue, i.e. to malign somebody,
to tear him apart
S5 F. IV to sow enmity between men is in a cause > consequence
relation with S2 and S3. We can therefore show the lexicogenetic tree
which, for {l,s}, goes back to the matrix, and which, for the etymon {s,' }
remains at the etymonial level:
M8 M?
{[+lateral],
[+continuant]}
tongue
{l,s} {s,' } to head into the land
[ls] l[s]
lasa
4.3 lasafa and lasifa

This entry covers the following senses:
S1 to fit by placing one next to the other and one over the other
(e.g. stones in constructing a wall, in building)
S2 lasifa to be dried and stuck to the bones (of skin on a very
scrawny body)
S3 to wind a strap of sinew around the base of an arrow
S4 to shine, to gleam
These four senses do seem to be in homonymic relation, as it is difficult

to see how one could draw some sort of plausible semantic relation
between them.
For S1 to fit by placing one next to the other and one over the other
(e.g. stones in constructing a wall, in building), comparison with the
words below leads us to identify a common etymon {s,f } to arrange, to
organize:
saffa [sf ]f : to arrange in order
xasafa x[sf ] : to fit and join solidly
sannafa s[n]f : to compose, to make (a work, a book)
For S2, we recognize matrix 8 as in item:

1.3.1. consequence (1): to moisten and stick, as in:
lassa [ls]s F. VIII : to attach oneself, and to stick strongly
lasiqa [ls]q : to be stuck to the bones
lasiqa [ls]q : to be stuck
lasaa [ls] : to be dried and stuck to the bones (of skin or a
very scrawny body)
For S3 to wind a strap of sinew around the base of an arrow,

consider:
rasafa r[sf ] : to wind a solid strap or a flattened sinew around
the tip of an arrow to make firm the iron that has
been fitted
'afasa [ fs] : to wind a ifs around the mouth of a bottle
'asaba [sb] : to bandage, to wind a headband, bandage around;
to put a dressing on (the head, a member)
Examining the above, we may identify the etymon {s,f }, which is itself a
realization of Matrix 6, f [labial] and s [dorsal],39 of which the notional
invariant is curvature; and of which to surround and to wind around
are consequences.40
Finally, lasafa in sense S4 to shine, to gleam can be brought into

relation with:
walafa w[lf ] : to shine time after time with repeated flashes that
come in uninterrupted succession (of lightening)
jafala : to shine
39
Emphatics are characterized by the features [dorsal], [pharyngeal], [coronal].
40
See Bohas (2000: 115117).
The above enables us to identify the etymon lf to shine. As to the matrix

of which it is apparently a realization, it should contain the feature
[+approximant], as in:
balaqa F. VIII : to shine, to gleam
baraqa : to shine, to gleam, to be shiny
lamah a : to shine
lamaa : to shine
ramah a : to shine (of lightening)
In any case, as in walafa and lasafa, the composition {[+approximant],
[labial]}41 connected to the notional invariant to shine can be observed.
This is a complex matrix that has yet to be studied. Suffice it to say that,
in the present state of our knowledge, the etymon has been positively
identified, while the matrix has not. We can trace the lexicogenetic tree
thus:
M8 M? M6
{[+lateral], {[+approximant] ?} {[labial], [dorsal]}

[+continuant]}
tongue shininess curvature
{l,s} {l,f } {s,f }1 {s,f }2 to organize
[ls] f l[s] f l[sf ] l[sf ]
lasaf
4.4 lasaba and lasiba

S1 to prick somebody (of snake)
S2 to give somebody a lash with a whip
S3 lasiba to lick; to attach and to stick to something
41
The same remark as in footnote 28 is appropriate here. If gutturals appear to be a
part of this matrix, then they join the class of approximants, otherwise, [approximant]
will have to be restricted by the addition of [coronal].
Take S2 to start with. This form is related both semantically and

phonetically to:
saba a [sb] : to whip somebody with a whip until he bleeds
This fact leads us to deduce that lasaba should be analysed as a form
incorporating the etymon {s,b} through initial incrementation of the l,
which has no semantic value. The etymon {s,b} comes from Matrix 1
{[labial], [coronal]}, the notional invariant of which is to strike a blow,
with the specification, in this case, of the means with a whip, and other
specifications in:
rabasa : to strike with the hand
safaa : to strike, to give a blow (especially of birds, when, in
fighting, they give each other vigorous blows with
their wings)
sfa : to strike somebody with a sabre
nasama : to strike the ground with a foot
The two other senses come from the matrix {[+lateral], [+continuant]}
tongue, the first enters the heading:
3. The tip of the tongue: pointed, to be sharp, to become a point, and
thus to prick. As in:
lasaa : to prick somebody, to bite (of scorpion or snake)
lasana : to form a point, to give a pointed form, (e.g. to a shoe,
etc.)
: to prick (of scorpion)
salama : to prick somebody, to make a bite (of a snake)
F. IV passive : uslima: to be pricked by a snake
And the second the heading:

1.3. to lick and 1.3.1 (see the end of subsection 3.1 above)
as in:
lassa : to lick (a vase, a frying pan, etc.)
lasada : to lick (a vase)
ladasa : to lick
lahasa : to lick
Drawing up the lexicogenetic tree is therefore an easy matter:
M8 M1
{[+lateral], {[labial], [coronal]}

[+continuant]}
tongue to strike a blow
{l,s} {s,b}
[ls]b l[sb]
lasaba
4.5 lah afa

Whereas the analysis of lasaba presents no particular difficulty, that
of lah afa is formidably complex, which is why we have left it till last.
Indeed, this verb attests several senses, as follows:
S1 F. I: to lick something
F. IV: to pressure somebody, to trouble, to ask with insistence
S2 F. I: to envelop somebody in a sheet, a blanket
F. III: to help, to assist somebody42
F. IV: to clothe and envelop somebody in a garment
F. V: to envelop oneself in a sheet, in a cloth, in a piece of mate-
rial
F. VIII: to envelop oneself in a piece of material
S3 F. I P.: to suffer losses in ones belongings, flocks, etc. 43
F. IV: to do something bad to somebody, to cause him some
harm
S4 F. IV: to pull out (e.g. somebodys nail)
S5 F. IV: to burn, to have something consumed in fire44
Let us start with S1. The radical includes l and the fricative h , and can be
a realization of Matrix 8: {[+sonorant] , [+continuant]}
[+lateral]
42
This sense is linked to this block, see below.
43
P. indicates the passive.
44
Attested in Ass al-bala.
Notional invariant: tongue

to lick coming under the heading 1.3.45
The following forms develop the etymons {l,h} and {l,' }, that are
analogous realizations of this matrix:
lahika [lh ]k : to lick
lahisa [lh ]s : to lick
la'iqa [l]q : to lick
la' [l]w F. V : to lick
The meaning expressed by F. IV that we have placed under sense S1, to
pressure somebody, to trouble, to ask with insistence, seems to correlate
well with the idea of licking, exactly as in French in which coller, to
stick can mean46 to impose ones presence on somebody and collant
sticky: that which one cannot rid oneself of. Thus, we have the following
chain: lick >stick >insist, the sense that we find in other manifestations
of the etymon {l,h}:
lahh a [lh ]h : to insist on something to somebody, to persist in
asking him (for) something
hafala h [ f ]l F. V : to insist
halata [h l]t : to insist, to pressure, to scold somebody, and to
swear
mahala m[h l] F. V : to insist and to pressure somebody
For S.2, the emergence of the sense to envelop undoubtedly comes from
the blending of the two etymons: lf x h f, which can be clearly seen in:
laffa : to envelop, to twist, to surround with something
h affa : to surround somebody with something, to enve-
lop with something
Is it possible to go further and identify the matrix from which these
etymons stem? Without doubt, if we take into consideration the
following data:
laffa : (also means:) to gather, to collect from all sides
daffa : to gather, to bring together47
A hypothesis defended by Cantineau (1951), amongst others, is that in
protosemitic there was a lateralized emphatic that we shall write dl. In
Bohas and Janah (2000) it was argued that this dl had split in Arabic into
45
See above subsection 3.1.
46
See the Petit Robert
47
See also: mudf d[y] f : surrounded, attacked, encircled on all sides.
two phonemes: d and l. A word bearing one meaning had thus given
birth to two words with the same sense (modulo a few nuances). Thus
we would have:
dlaffa to bring together
daffa laffa
to gather, to bring together to gather, to collect from all sides

This means that the l of la a is not a true l, but lexically a dl and that,
as such, it has the feature [pharyngeal] of the emphatics. The pair:
laffa : to envelop, to twist, to surround with something
h affa : to surround somebody with something, to envelop
with something
is in fact a dlaffa/h affa pair and the etymons dlf/h f are realizations of
Matrix 3:
Matrix 3 {[ labial] , [ pharyngeal]}
Notional invariant: (a) tightening
The relation between to envelop, to surround and to tighten is merely
one of cause to consequence.
The sense expressed by F. III: to help, to assist somebody seems to be
accounted for in terms of metaphor: to help is to surround somebody
with assistance, protection or affection. As an argument in favour of this
relation, we have the verb h affa that explicitly shows this meaning, since,
in the Kazimirski, after to surround comes the sense to be constantly
around somebody, and to be attentive to serve or protect him.
S3 constitutes a single semantic block:
a : F. I P. : to suffer losses in ones belongings, flocks, etc.
b : F. IV : to do something bad to somebody, to cause him some
harm
Let us bring lah afa into relation with the forms which manifest the same
properties:
falla F. IV : to lose ones flocks
falaa : to lose, to reduce to nothing
faliya : to be cut, separated from the rest of the body
lafaa : to peel, to skin
wafala : to peel something by removing the bark
We introduce an etymon {l,f }; l is [coronal] and f [labial]. This etymon
can thus be a realization of Matrix 1:
{[labial], [coronal]}
Loss, harm comes under the heading B.3,48 global consequence, as in:
h afata : to destroy, to lose
talifa : to perish
As for S4 F. IV: to pull out (e.g. somebodys nail) it is a realization of

a Matrix under study:
{[+approximant], [+continuant]}
[coronal]
with the notional invariant: to bring something to oneself.
This manifests itself in the words quoted earlier in section 3.1. As f and
h are both continuants, it seems reasonable to consider the radical as a
blending of the two etymons that both realize this matrix: lh x lf, B type
blending.49
There remains sense S5 F. IV: to burn, to have something consumed in

fire. In this meaning, we can relate lahafa to:
lafaha : to burn, to cause harm through its intensity (of fire, or
a very warm wind)
fayh : heat caused by a star
sahafa : to burn, to have something consumed in fire
The above relation reveals the etymon { f,h}; the semantic relation with
to blow remains to be established in order to link it to Matrix 2.
As for the residual cases
F. II : to let the bottom of ones garment scrape the ground,
to wear it very long so that it trails, by extension: to
walk proudly
F. IV : to come to the foot of a mountain and lih f: the foot of
a mountain
At this stage in our research, all that can be noted is that the latter is
perhaps to be related to h ffatun: edge, margin, extremity, although we
cannot establish this with certainty. All these comparisons are of limited
interest, but it is worth remembering, as we said at the start, that certain
points are still unclear.
48
See Bohas (2000: 79).
49
See Bohas (2000: 50).
Consequently, we shall limit ourselves to those senses that are clear, in

order to construct the lexicogenetic tree:
M8 M1 M3 M7
{[+lat], [+cont]} {[labial], [coronal]} {[phar], [lab] } {[+approx], [+cont]}

tongue strike a blow tightening to bring to oneself
{l,h} {l,f } {h,f } {l,h} x {l,f }
[lh ] f l[h ] f l[h f ] [l,h ] x [l,f ]
lah af
5. Conclusion
Our method is thus distinct from that of the partisans of the

triconsonantal root. For them, it is sucient to identify the three
consonants in order to consider the analysis complete, even if
semantic incongruities and incompatibilities are evident, and even
if this identi cation provides no explanation of phono-semantic
links between words, such as homonymy and enantiosemy. What
does reassure them, however, is that they can pride themselves
on having reached a state of certainty.... that the root of maktab =
ktb is a certainty! Yet why should the root of istadaytu be dw rather
than dy, and, if it is dw, what phonetic motivation is there to be
found in istafaltu to justify the passage from w to y in istadaytu?
Indeed, this certainty is not as de nitive as they would have us believe.
With our approach based on argumentative reasoning, we might make
mistakes: one word might, perhaps, be matched up with another on
the basis of such and such a property without us having noticed this
relation. Given the explanatory results of our approach, which adopts
a heuristic point of view, this is a risk we assume.
The reader will have noticed that all our research has been carried
out taking the lexicon of Arabic as a synchronic whole. As we have often
repeated, in the study of the lexicon, it is vain to go back to a previous
biliteral stage in order to diachronically derive a triliteral stage. In other

words, the old debatebiliteral or triliteralis immaterial. The binary
composites (the matrices and the etymons) and the ternary composites
are there before our eyes: you only have to open a dictionary to find
them. Every radical is, according to the level of explanation, binary or
ternary, as the explanation of homonymy demonstrates all too well.
Admittedly, no one is obliged to overstep the conception, invented
by the Arab grammarians, of the tri- or quadri-consonantal root. How-
ever, to restrain oneself to this is to forego the opportunity to provide
explanations for the phono-semantic phenomena contained in the lexi-
con, phenomena that everyone may observe. To work with a theory that
posits that the Sun rotates around the Earth enables one to explain a
certain number of observable phenomena;50 but to work with a theory
that posits that the Earth rotates around the Sun enables one to explain
a greater range of phenomena, and fits in better with other knowledge
we have of the movement of the stars.
6. References
Ass al-bala = Ab l-Qsim Mahmd b. Umar az-Zamaxar, Ass al-bala,

Abdarrahm Mahmd, ed. Bayrt: Dr at-ti ba wa-n-nar.
Bahri, A. 2003. Lnantiosmie en arabe, Doctoral thesis. University Paris 8.
Bailly, A. 1950. Dictionnaire grec franais. Paris: Hachette.
Bohas, Georges. 1997. Matrices, tymons, racines, lments dune thorie lexicologique du
vocabulaire arabe. Paris: Peeters.
. 2000. Matrices et tymons, dveloppements de la thorie. Lausanne: Editions du
Zbre.
. 2006. De la motivation corporelle de certains signes de la langue arabe et de ses
implications. Cahiers de linguistique analogique, 3, 1141.
, and N. Darfouf. 1993. Contribution la rorganisation du lexique de larabe, les
tymons non-ordonns, Linguistica Communicatio, 5/12, 55103.
, and A. Janah. 2000. Le statut du dd dans le lexique de larabe et ses implications.
Langues et Littratures du Monde Arabe, 1, 1328.
50
We find in The Legend of Alexander: For the Sun is the servant of The Lord, which
interrupts its course neither day nor night. This idea that the stars are in the service of
God seems to date back at least to Bardesane (born in 154): Neither the Sun, nor the
Moon, nor the other beings which are superior to us in any thing have received power
over themselves, they are on the contrary subject to a law and, consequently, they do
what they have been ordered and never anything else. The Sun never says I shall not
rise at the given hour, nor the Moon I will no longer have phases, I will neither wax nor
wane, . . . All these creatures are servants and remain subject to a law: they are instru-
ments of the wisdom of The Lord, Who is infallible. (See Teixidor 1992).
, and M. Dat. 2005. La matrice acoustique {[dorsal], [pharyngal]} en arabe clas-

sique et en hbreu biblique, premire esquisse. Mlanges de lUniversit Saint-Joseph,
LVIII, 125143.
, and M. Dat. 2007. Une thorie de lorganisation du lexique des langues smitiques :
matrices et tymons, Lyon: ENS ditions.
, and A.R. Saguer. 2006. Sur un point de vue heuristique concernant lhomonymie
dans le lexique de larabe. In Edzard, L. & J. Watson (eds), Grammar as a Window
onto Arabic Humanism. A Collection of Articles in Honour of Michael G. Carter, Wies-
baden, Harrassowitz Verlag, 130154.
, and R. Serhane. 2003. Consquences de la dcomposition du phonme en traits.
In: Angoujard, J.-P. and S. Wauguier-Gravelines (Eds.): Phonologie. Champs et per-
spectives. Lyon: ENS ditions, 131-155.
Cantineau, Jean. 1951. Le consonantisme du smitique. Semitica, IV, 7994.
Dat, M. 2002. Matrices et tymons. Mimophonie lexicale en hbreu biblique. Doctoral
thesis, Lyon: Ecole Normale Suprieure Lettres et Sciences Humaines.
Dell, F. 1973. Les rgles et les sons: introduction la phonologie gnrative. Paris: Her-
mann.
Diab, S. 2005. La matrice {[coronal], [dorsal]}, Les tymons impliquant le jm, Masters 2
Research Paper. Lyon: cole normale suprieure lettres et sciences humaines.
Guerssel, M. and J. Lowenstamm. 1993. The Derivational Morphology of the Classical
Arabic Verbal System. ms. UQAM and University Paris VII.
Halle, M. 1991. Phonological Features. In W. Bright (ed.): Oxford International Ency-
clopedia of Linguistics. Oxford: Oxford University Press, 207212.
Hurwitz, S. 1913 [1966]. Root-Determinatives in Semitic Speech, a Contribution to Semitic
Philology. New York: Columbia University Press.
Joon, P. 1923. Grammaire de lhbreu biblique. Rome: Institut biblique pontifical.
Kazimirski, A. de Biberstein. 1860. Dictionnaire arabe franais, Paris: Maisonneuve et
Cie.
Kenstowicz, M. 1994. Phonology in Generative Grammar. Oxford UK : Blackwell.
Khatef, L. 2003. Statut de la troisime radicale en arabe: le croisement des tymons, Doc-
toral thesis. University Paris VIII.
. 2004. Le croisement des tymons: organisation formelle et smantique. Langues
et Littratures du Monde Arabe, 119138.
Lisn = Jaml ad-Dn Ab l-Fadl Muhammad b. Mukarram b. Al b. Ahmad b. Ab
l-Qsim b. H abqa Ibn Manzr, Lisn al-Arab, s.d. Abd Allh Al al-Kabr, Muhammad
Ahmad H asab Allh, Him Muhammad a-dil, eds. Cairo: Dr al-Marif.
Lipinski, E. 1997. Semitic Languages. Outline of a Comparative Grammar. Leuven:
Peeters.
Mansouri, W. 2006. La place du trait [sonorant] dans les matrices de larabe. Masters 2
Research Paper, Lyon, cole normale suprieure lettres et sciences humaines.
McCarthy, J.J. 1986. OCP Effects: Gemination and Antigemination. Linguistic Inquiry,
17,2. 207263.
Nyckees, V. 1998. La smantique, Paris: Belin.
Qms = Majd ad-Dn Muhammad b. Yaqb al-Fayrzbd, Al-Qms al-Muh t.
Bayrt: Muassasat ar-Risla.
Quiniou, Y. 2006. La mort scientifique de Dieu. Le nouvel observateur. Hors-srie,
3841.
Le Petit Robert, dictionnaire alphabtique et analogique de la langue franaise. 1967
[1993]. Paris: Dictionnaires Le Robert.
Saguer, A.R. 2000. Lincrmentation des prfixes dans le lexique de larabe. Le cas du n.
Actes du colloque Journes de linguistique arabe et smitique, Langues et littratures du
monde arabe, 1, 5782.
. 2002a. Lincrmentation des prfixes dans le lexique de larabe. Le cas du m.
Langues et littratures du monde arabe, 3, 2957.
. 2002b. Z hirat al-isbq f l-judr al-arabiyya. Agadir: publications de lUniversit

Ibn Zuhr.
. 2003. La matrice {[+nasal), [coronal)}, traction en arabe. Premire esquisse.
Langues et littratures du monde arabe, 4, 138183.
Segeral, Ph. 1995. Une thorie gnralise de lapophonie, Doctoral thesis. University
Paris 7.
Serhane, R. 2003. Etude de la matrice {[labial], [dorsal]} en arabe, Doctoral thesis. Uni-
versity Paris 8.
Teixidor, J. 1992. Bardesane dEdesse, la premire philosophie syriaque. Paris: Les ditions
du Cerf.
Yeou, M. and S. Maeda. 1994. Pharyngales et uvulaires arabes sont des approxim-
antes: caractrisation acoustique. 20me Journes dtudes sur la Parole, Trgastel,
409414.
THE PERIPHRASTIC BILINGUAL VERB CONSTRUCTION AS A
MARKER OF INTENSE LANGUAGE CONTACT.
EVIDENCE FROM GREEK, PORTUGUESE AND
MAGHRIBIAN ARABIC
Louis Boumans
University of Nijmegen
1. Introduction
I would like to start this paper with some terminological notes. I use
the dichotomy community language versus superimposed language
to refer to the typical unequal social-economic status of the bilingual
speakers languages. These terms refer to local as well as global power
relations and their sociolinguistic consequences. For instance, whether
in Portugal, Brazil or the United States, Portuguese speaking people
learn English in order to gain access to valuable information and upward
social mobility, i.e. education, media and employment.
I will use the term socially dominant for the language the individual
speaker is most exposed to in her daily life. This could be either the
community or the superimposed language, depending on the local situ-
ation. Thus English is more likely to be socially dominant for a particular
Portuguese/English bilingual member of the Portuguese community in
the US, while Portuguese will be socially dominant for most bilinguals
living in Portugal.
The terms matrix language and embedded language are grammati-
cal notions referring solely to local syntactic units of analysis in bilingual
speech. The higher order constituent is the matrix in which lower order
constituents are embedded. In mixed sentences higher and lower order
constituents are in different languages. In most instances the community
language functions as the matrix language and superimposed language
elements are embedded. However, the reverse occurs in a minority of
cases so the terms should not be confused.1
1
Since the early 1990s, Carol Myers-Scotton has been the most influential promoter of
the insertion approach to code-switching and the terms matrix and embedded language.
292 louis boumans
The present article deals with verbs from a superimposed language

that function as embedded elements in community language discourse.
The community language may or may not be socially dominant, and I
argue that this makes a difference for the way in which foreign verbs are
embedded.
2. The integration of foreign verbs
There are three ways in which foreign verbs are integrated in the matrix
language, two of which are common. One is the complete morphologi-
cal integration. Some basic form of the foreign verb, typically the verb
stem or the infinitive, is treated as the verb stem of the receiving matrix
language, and verbal categories of the latter are expressed by matrix
language morphology. Gloss (1) is a Moroccan Arabic/French example
showing the French verb stem montr- (from montrer [m ~tre] to show)
with an Arabic prefix and suffixes.2
(1) wa adi y-montr.i-w-l-ek . . .
Q FUT 3-show-PL-to-2SG
Are they going to show you . . .? MA3/French (Wernitz 1993, 308)
The insertion of foreign verb stems without overt morphological inte-
gration in matrix languages lacking verbal morphology, such as various
Austronesian languages (Van Staden 1999) can be considered as a sub-
category of the morphological integration strategy, even if the integra-
tion is not overtly expressed by ML morphemes.
I concur with the fundamentals of her original Matrix Language Frame (MLF) model
(Myers-Scotton 1993), except for the definition of the matrix language. In my view, all
syntactic constituents function as a matrix for lower-order constituents, whereas in the
MLF model only the Complementizer Phrase functions as a matrix. I refer to earlier
work for more details on this approach to code-switching (Boumans 1998, Boumans
and Caubet 2000, Boumans 2002).
2
The vowel i is not a proper suffix. Embedded French verbs are modelled on a class
of Arabic verbs ending in a vowel. This vowel is subject to a/i ablaut. Cf. Caubet (1993),
Boumans (1998), and Boumans and Caubet (2000).
3
The following abbreviations are being used: in the main text: MA Moroccan Arabic;
in the glosses to numbered examples: 1,2,3 first, second, third person; ACC accusative;
AGR agreement; ART article; AUX auxiliary; DEF definite article; FUT future tense;
IMPF imperfective; INF infinitive; M masculine; NEG negation; PASTPART past parti-
ciple; PL plural; PROGRPART progressive participle; PRT preterit; REL relative clause
marker; SG singular.
the periphrastic bilingual verb construction 293
The other common strategy is the periphrastic construction. This

strategy is equally common in the worlds languages and can also be
illustrated with Moroccan Arabic, this time in contact with Dutch:
(2) ba y-dir li-h aanvall-en
want 3-do to-3SG attack-INF
It [the bird] wanted to attack him. MADutch (Jamal, 16, Utrecht 2000)
In his book Bilingual Speech (2000), Muysken presents a typology of
verb integration with special attention for the periphrastic type. In the
periphrastic construction, he argues, the foreign verb can be a nominal-
ization, an infinitive or an adjunction. In the first case, the nominalized
verb is the complement of the helping verb, in the case of an infini-
tive the helping verb must be analysed as an auxiliary. In the adjunction
analysis, the foreign verb and the helping verb form a kind of verbal
compound.
Which of these three analyses is most appropriate does not only depend
on the language pair involved. As I discussed elsewhere (Boumans 1998;
2000), strategies and grammatical constructions differ among individu-
als belonging to the same bilingual community. Compare examples (2)
and (3). In Jamals utterance in (2), the patient participant of the Dutch
transitive verb aanvallen to attack is expressed as the MA indirect object
li-h to-him. The Patient role is expressed as an indirect object, because
the Dutch verb occupies the Direct Object position of the MA transitive
verb dar to do. This is in fact the common construction in the Dutch
Moroccan community. Now in Samirs example in (3), the patient of the
transitive verb controleren to supervise surfaces as the direct object suf-
fix -hom them. The construction in (2) can be described as a helping
verb plus a verbal noun object, while the one in example (3) is in fact
very similar to the monolingual Dutch auxiliary construction. Compare
(3) with its Dutch translation in (4). Note that the pronominal direct
object of the main verb cliticizes to the finite auxiliary, even if an adjunct
constituent (morgen tomorrow) comes between the two verbs.
(3) kun adi y-dir-hom controler-en?
who FUT 3do-3PL supervise-INF
Who will supervise them? MA/Dutch (Samir 20, Nijmegen 1991)
(4) wie gaa-t hun (morgen) controler-en

who go-2SG 3PL tomorrow supervise-INF
Who will supervise them (tomorrow)? Dutch translation of (3)
294 louis boumans
The variation between speakers, and even between utterances of the

same speaker, can be explained by the varying levels of competence in
both languages, and by the effect of grammaticalization and conven-
tionalization (Backus 1996).
The third, less common strategy for the incorporation of foreign
verbs deserves mentioning here for the sake of completeness. This stra-
tegy consists in inserting inflected verb forms rather than verb stems.
The inflected foreign forms are mapped onto the ML paradigm and
express ML verbal categories. A famous case in point is Mednyj Aleut,
also known as Copper Island Aleut (Thomason and Kaufman 1988,
2338). In this variety of Aleut, Russian inflectional patterns in finite
verbs replaced Aleut ones while most other grammatical subsystems
remained intact. Crucially, the Russian verbal inflections express Aleut
tense and aspect categories. Similarly, Igla (1991) reports on a dialect of
Romani nowadays spoken in a suburb of Athens in Greece. The speakers
of this dialect of Romani moved in from Turkey in the 1920s but no lon-
ger speak any Turkish. However there are still approximately 30 verbs of
Turkish origin that continue to be inflected with Turkish suffixes while
following the Romani verbal paradigm, cf. Table 1. That is, the Turkish
inflectional suffixes express Romani inflectional categories.
TABLE 1 THE PRESENT TENSE OF TURKISH VERBS AND OF VERBS

OF TURKISH AND NON-TURKISH ORIGIN IN AJIA VAVARA ROMANI
(FROM IGLA 1991, 52)
Turkish Ajia Vavara Romani
present to write to write to bring
sg 1 yaz-ar-m yaz-ar-um an-av
sg 2 yaz-ar-sn yaz-ar-sun an-es
sg 3 yaz-ar yaz-ar an-el
pl 1 yaz-ar-z yaz-ar-us an-as
pl 2 yaz-ar-snz yaz-ar-sunus an-en
pl 3 yaz-ar-(lar) yaz-ar-(lar) an-en
(5) i thagarni kurta-du len e rom-en

ART queen save-PRT REL ART gypsy-ACC.PL
(..) the queen saved (them) the gypsies. Romani/Turkish (Igla 1991, 53)
This latter strategy is actually very rare in language contact situations.
However, it may be more common in the case of two closely related
languages or varieties. For instance Malkiel (1986) draws attention to
the Spanish conjugation pattern of a number of Portuguese verbs. Also
non-tensed verb forms such as participles, infinitives and imperatives

are quite readily interchangeable between languages, cf. (6).
(6) Le estaba poniendo atencin qu estaba
3.IO was.3SG put.PROGRPART attention what was.SG
recorded
record.PASTPAR
He was paying attention to what was recorded. Spanish/English (Pfaff
1979, 300)
3. Explanations for the distribution of the two common strategies
Why do speakers opt for one strategy rather than the other? The two
common strategies in particular, morphological integration and peri-
phrastic constructions, allow us to compare the languages and linguistic
situations involved.
3.1 Characteristics of the matrix language

When one makes a list of all language contact situations and the attested
strategies for the integration of foreign verbs, it becomes clear that
genetic or areal typological factors play an important role. In the Indo-
Iranian and Turkic languages, for instance, the periphrastic construc-
tion appears in virtually all language contact situations. In connection
with this, Muysken (2000, 194) speaks of a large linguistic area, in
this case, stretching from Sri Lanka to Greece. On the other hand, most
western European languages seem to prefer the morphological integra-
tion strategy. A plausible explanation for the areal bias is that speakers
who are accustomed to a certain strategy of incorporating foreign verbs
will reuse this strategy in new contact situations. Some examples of this
kind of bilingual knowledge will be discussed below. Still, the two MA
examples cited in (1) and (2) show that the matrix language is not the
only factor deciding which incorporation strategy speakers will use.
3.2 Characteristics of the embedded language

One way to explain the difference between MA/French and MA/Dutch
code-switching is to look for differences in the superimposed embed-
ded languages. Heath (1989) argues that the phonological shape of the
French verbs facilitates their incorporation in Arabic. Bilingual speakers
associate the stressed final vowel /e/ of most French infinitives and other
296 louis boumans
verb forms with the final /i/ of the prefix conjugation (imperfective) of
a class of MA verbs.
Dutch infinitives typically end in an unstressed suffix -en that is pro-
nounced as a schwa. It is not obvious whether this makes them phono-
logically less similar to the MA finite verb in /i/ than French infinitives.
Firstly, MA phonology does not allow for the schwa in open syllables.
This may lead Moroccan listeners to ignore word-final schwas in Dutch,
and interpret the infinitives as consonant-final. Alternatively, however,
Moroccans may interpret the Dutch final schwa as a full front or back
vowel. Both tendencies can be observed in the speech of Moroccan
learners of Dutch.
Moreover, the vocalic ending in French verbs cannot be decisive,
since Spanish, Italian and English verbs are morphologically integrated
in North African varieties of Arabic in the same way as French verbs.
Many of these Romance and English infinitives end in a consonant.
Arabic/Spanish language contact is still common in the (formerly)
Spanish occupied northern parts of Morocco (cf. Heath 1989, Her-
rero Muoz-Cobo 1996). Cohen (1912) notes interesting observations
on the Jewish dialect of Algiers, where the Spanish (or Lingua Franca)
infinitive ending -ar is even extended to embedded French verbs. E.g.
refuzarit from French refuser [rfyze] in (8). Numerous verbs of Italian
and English origin are found in Maltese, another variety of Maghribian
Arabic (Aquilina 1965 [1987], Camilleri 1994, Mifsud, 1995).4 The stem
extension -jain Maltese verbs of English origin, as in (10), is a reflec-
tion of the Sicilian ~ Italian infinitive suffix -are.
(7) frin-ar-t ~ frin-ar.i-t
brake- INF-1SG ~ brake- INF.STEM EXTENSION -1SG (Sp. frenar)
I braked Tetouan Arabic/Spanish (Heath 1989, 184)
(8) refuz-ar-it
refuse-SPANISH INF-1SG (Fr. refuser)
I refused Jewish Arabic of Algiers/French (Cohen 1912, 432)
(9) ti-ppartiipa-w
3-participate-PL (It. partecipare)
you (pl) participate Maltese/Italian~ Sicilian (Camilleri 1994, 437)
4
As a matter of fact, the type of verb integration illustrated in (1) does not occur in
the Middle Eastern varieties of Arabic. Instead the periphrastic construction is more
common in that region.
(10) ni-bbli.ja-ha
1-bleach.STEM EXTENSION-3SG
I bleach it Maltese/English (Camilleri 1994, 443)
Thus, verb stems from various embedded languages and with diverse
phonological characteristics can be morphologically integrated into
Moroccan or Maghribian Arabic, and phonology does not seem to be an
explanation for the periphrastic construction in the case of MA/Dutch.
A second way in which the embedded language might influence the
selection of the verb integration strategy is the use of periphrastic verb
constructions in the embedded language itself. Following this line of
thought, the bilingual MA/Dutch periphrastic construction could be
inspired by periphrastic constructions with doen to do in standard or
non-standard Dutch.5 Again, the comparison with the other language
pairs speaks against this hypothesis. Periphrastic constructions with a
do verb are much more common in English than they are in Dutch,
while their abundance in French and Spanish may be similar to that in
Dutch.6 Therefore, the occurrence of do constructions in monolingual
Dutch likewise does not explain why the periphrasis strategy is chosen
to insert Dutch verbs in MA matrix clauses.
3.3 Characteristics of the sociolinguistic setting

In the case of the North African language contacts, neither the host lan-
guage nor the embedded language is the sole factor determining the way
in which the loan verbs are integrated in Arabic. For this reason I con-
jecture that the sociolinguistic setting in the Netherlands, where MA is
a minority language, is responsible for the periphrastic construction in
MA/Dutch. As a more general hypothesis, I suggest situations in which
language contact is more intense, like migration to an urban industria-
lised society, favour the use of the periphrastic construction.7
5
Jacomine Nortier and Roeland van Hout made this suggestion when we discussed
this paper at the SS14 workshop on borrowing in Gent, April 4, 2002, and on earlier
occasions.
6
With the exception of certain infrequent constructions, do-periphrasis in Dutch is
associated with non-standard regional varieties or child language. Cf. Nuijtens (1962,
15457), Giesbers (1984), and Cornips (1994).
7
One could argue that the morphological integration of French verbs is found both
in North Africa and in the North African diaspora in Francophone Europe and Can-
ada. In this particular case, however, code-switching strategies that were already estab-
lished in the homeland have simply been maintained in the diaspora.
298 louis boumans
In order to shed some light on this issue and to test the intensity of
contact hypothesis, the following two sections survey verb integration
in two additional language pairs for which both integration strategies
are attested: Greek/English and Portuguese/English. In both cases the
influence of English has been described for immigrant communities in
Anglophone countries as well as in the Greek and Portuguese speaking
homelands.
4. Greek/English
4.1 Greece
The morphological integration of Italian verbs consists of adding Greek
suffixes to the Italian infinitive in -are.8
(11) It. posare to pose
It. schizzare to sketch
Mainland Greek/Italian (Van Dijk-Wittop Koning 1963)
(12) fundro It. fondare to anchor

barkro It. barcare to board
kompletro It. completare to fill up [cargo]
Mainland Greek/Italian (Hartley 2001)
According to Swanson the mildly productive verbal suffix -aro entered
the Greek language in late Byzantine times (1958, 40). The Italian infini-
tive marker has been generalised to verbs of French and English origin
as well:
(13) v Fr. dbuter to make ones debut
Fr. lancer to launch
Mainland Greek/French (Contossopoulos 1978, 42)
(14) flertaro to flirt

stoparo to stop (a machine)
sutaro to make a shot (soccer, basketball)
Mainland Greek/English (Swanson 1958)
8
The authors on loanwords and codeswitching in Greek make use of different writ-
ing and spelling conventions. The spelling of the source publication is retained in the
examples cited here.
(15)
/parkaro/ to park

/manadzaro/ to manage

/triparo/ to trip
Mainland Greek/English (Apostolou Panara 1991, 50)
Interestingly, the recycling of a foreign verb marker like the Italian
-are to mark verbs from a third language is yet another example of how
bilingual experience influences the way in which speakers treat words
from foreign languages. Note the close parallel with -ar in the Jewish
Arabic dialect of Algiers (8), and the -ja stem extension after English
verb stems in Maltese (10).9
Most sources on loan verbs in Modern Greek mention only this
-ar- extension with complete morphological integration of the foreign
word. However, Apostolou Panara, while asserting that verbs all adapt,
i.e. are morphologically integrated, mentions that in some cases, along
with the single lexeme we attest a periphrasis with the Greek verb /kano/
(to do, to make) preceding a noun or a gerund (1991, 50), cf. the exam-
ples in (16). A few verbs occur exclusively as part of the periphrastic
construction (17), whereas the periphrasis is judged ungrammatical in
most cases (18).
(16) /flertaro/ besides /kano flert/ to flirt
/manadzaro/ besides /kano manadzing/ to manage
/stokaro/ besides /kano stok/ to stock
(17) /kano kambing/ but not */kambaro/ to camp

/kano serfing/ but not */serfaro/ to surf
(18) /triparo/ but not */kano trip/ to trip

/parkaro/ but not */kano park/ to park
Mainland Greek/English (Apostolou Panara, 1991, 51)
It is possible that a closer study of the English vocabulary in Modern
Greek would reveal regularities that govern the use of periphrasis.
Periphrasis with kno could be associated with particular types of events,
for instance habitual or durative activities. The periphrastic construc-
tion may be more common with recent English loans, or in the speech
9
In Dutch a similar situation obtains with the French infinitive marker -er [er] (Tref-
fers-Daller 1994), which is sometimes used for English loanwords as well, e.g. formatt-
er-en to format.
300 louis boumans
of speakers who are more exposed to English. kano kambing may be a

calque after the French expression faire du camping (cf. kno ski in (24)
below, possibly after French faire du ski). The available data do not allow
for such conclusions at the moment. Still, I think it is safe to conclude
from the literature that in Greece the periphrastic construction occurs
as a minor pattern with some verbs only, while morphological integra-
tion is the rule.
4.2 Cyprus
Cyprus has been under British rule for more than eighty years, during
which English was the language of administration. First, from 1878 on
Britain administered the island in agreement with the Ottoman Empire.
Then Cyprus was annexed by Britain when the Ottoman Empire enters
into World War I on the side of Germany, and subsequently the island
became a British Crown colony under the British rule. In 1960 Cyprus
gained independence, and (Modern Standard) Greek became the lan-
guage of administration. The English language remained influential
through the tourism industry and the large international community
on the island.
There are a number of studies dealing specifically with the influence
of English on Cypriot Greek. Papapavlou (1997) cites a list of English
loan words found in written sources and tape-recorded speech. His list
includes 23 verbs in the -aro conjugation. He makes no mention of the
periphrastic construction.
(19) flrdro to flirt
rejistrro to register
riskro to take a risk
arcro to change (money)
akhro to check (inspect)
Cypriot Greek/English (Papapavlou 1997)
Similarly, Evripidou, also on the basis of written sources and recorded
speech, lists 19 English-origin Cypriot Greek verbs, all of them inte-
grated in the -aro class (2001). Goutsos discusses Greek/English code-
switching among members of a middle-class Cypriot family in Limassol
(2001). This study differs from the other ones, as Goutsos does not list
English loan words but rather focuses on the discourse functions of
language choice in the conversations. He does not comment on inte-
grated verbs of the -aro conjugation. He does cite three instances of the
periphrastic construction. As in Apostolou Panaras data, there are two

types of construction, one in which the English verb is represented by
the infinitive (or verb stem), see (20) and (21), the other with the ger-
und (22).
(20) a kmno shower
I will shower
Cypriot Greek/English (Goutsos 2001, 203)
(21) i jti tan kmnis wash

no because when you wash
(22) . . . llun tus pu na rti i jaj su en kmni swimming

I was saying to them, when your grandma comes, she will swim.
The periphrastic construction is not a pervasive phenomenon in the
speech of Goutsos informants, nor is code-switching in general (Gout-
sos 2001, 216). The examples also show the absence of an established
convention on how to insert the English verb into the periphrastic con-
struction. Still it is noticeable that the do-construction is recorded on
Cyprus and not in mainland Greece. Goutsos comments: Compared
to SMG [Standard Modern Greek], CG [Cypriot Greek] draws more on
English resources for borrowing and creates mixed compound forms
with de-lexicalized verbs (2001, 204).
Although Apostolou Panara shows that the periphrastic construc-
tion does occur in mainland Greek, Goutsoss comment confirms the
impression that it is a marginal phenomenon. If it is indeed true that
the periphrasis is more common in Cyprus, it is attractive in the light
of the present discussion to explain this difference as resulting from the
greater impact of English on the language of the former crown colony.
Moreover, Goutsoss examples stem from a really bilingual setting,
whereas the loanwords collected by the other authors stem mostly from
written sources that do not presuppose a bilingual readership.
4.3 Mainland Greek in the Diaspora

The oldest study on American Greek mentions just one English-origin
verb: mouvaro to move (Lontos 192526). However, this particular
form probably goes back to an Italian form muevere, as Hartley (2001)
suggests in his paper on loan words in Greek nautical terminology.
302 louis boumans
A much more elaborated study of American Greek is provided by Sea-

man. He mentions the same form muvro (1972, 165), but it becomes
clear that the periphrastic construction with kno to do, make is the
productive way to incorporate English verbs in his data: In verbal loans
from English, /jno/ seems to be the auxiliary for passives and /kno/ for
actives (1972, 166). His examples of the passive construction with jno
to be, become are ambiguous, as they might also be analysed as cases
of a copula plus predicate, and there are only a handful of examples. On
the other hand, Seaman cites 47 examples of the periphrastic construc-
tion with kno. In 33 cases the embedded English element is unambigu-
ously a verb; 11 cases may involve either a verb or a noun, e.g. control,
welcome. Some examples are reproduced below (see also Muysken 2000,
212).10
(23) kno, cover up I cover up
knis, brush ta ndja su You (sg) brush your teeth
kane, punch s/he punched [a meal ticket]
American Greek (Seaman 1972, 16768)
Tamis found the same construction in Australia (1986, cited in Muysken
2000, 212). The construction is also known from Canada (24) and Brus-
sels (25). Compared with Seamans study, the information on the latter
diaspora communities is rather scarce. It is not clear whether in these
cases the periphrastic construction is really the productive strategy for
incorporating English or French verbs. The Canadian examples cited
in (24), involving computer terminology and sports, remind of the
use of the periphrastic construction in Greece described by Apostolou
Panara.
(24) kno ski
kno save [computer]
kno print [computer]
kno jogging
Montreal Greek/English (Muysken 2000, 213, based on Hatzidaki p.c.)
(25) kno dmnager to move house

Greek/French (Muysken 2000, 213, based on Hatzidaki p.c.)
10
In Table 7.7 on p. 212 of Muyskens book the words jno and kno have been
reversed.
4.4 Cypriot Greek in London

There are a number of publications (and many unpublished theses) on the
language of the Greek Cypriot community in London. Gardner-Chloros
(1992) mentions the occurrence of morphologically integrated verbs
like tekaro to check, which is not surprising as these words are found
in Cyprus (and Greece) as well. In addition, however, she points out the
periphrasis with kno as an innovative construction, occurring with Eng-
lish adjectives, nouns, participles . . . (Gardner-Chloros 1992, 127).
(26) kamno use to use
kamno respect to respect
kamno developed to develop
kamno spelling to spell
Cypriot Greek in London (Gardner-Chloros, 1992, 127)
Zarpetea compares the language use of three generations of London
Cypriots (1995). She notices the use of both integrated loan verbs and
the periphrastic construction in the first as well as the second-generation
immigrants. The periphrastic construction still occurs in the speech of
the third generation, young children who use only little Greek:
(27) wash cups, do you want me to do the plates now mum?
I have washed the cups, . . .
Cypriot Greek/English (Zarpetea 1995, 581)
As a final note on morphologically integrated English verbs in immi-
grant Greek, I would like to draw the attention to the fact that there is
much overlap between the lemmas mentioned in the various sources.
This is particularly striking in Zarpeteas paper. Out of the seven verbs
she cites, five also occur in other sources, cf. Table 2.
TABLE 2 OVERLAP BETWEEN LOAN VERBS IN ZARPETEA (1995)

AND OTHER SOURCES
Zarpetea 1995, 578 other sources
to cancel Gardner-Chloros 1992, 126, Papapavlou 1997,
232, Evripidou 2001, 24

to check Papapavlou 1997, 240, Evripidou 2001, 24
to charge Papapavlou 1997, 240
to park Apostolou Panara 1991, 50, Papapavlou 1997,
236
to move Lontos 192526, 309, Seaman 1972, 167, Hartley
2001
to book
to packet
304 louis boumans
There is no such overlap in the examples cited for the periphrastic con-
struction with kmno (mainland Greek kno).11 This indicates that the
morphologically integrated verbs found in London may well have been
coined in the more monolingual setting in Cyprus. As noted above,
to move is probably not even of English origin. This is a fur-
ther indication that in the Cypriot diaspora, the periphrastic construc-
tion replaces morphological integration as the most productive strategy
for the incorporation of new English verbs. The chances are that the
same development has taken place in the American Greek community
studied by Seaman (1972).
5. Portuguese/English
I have been able to trace surprisingly few studies on English loan words
in Portuguese, whether European or Brazilian. Compared with Greek,
there are also few studies on Portuguese speaking communities in
Anglophone countries. The picture that arises from the available data is
approximately the same as for Greek.
5.1 Portugal
The website of the Instituto Superior Politcnico de Viseu in Portugual
hosts a short list of English loanwords in Portuguese, compiled by two
students and their English teacher (Queiroz, Rodrigues and McKenny
1999, accessed June 2002). The authors are not explicit about their data
sources, except that visitors of the web page are invited to contribute
observations. As it concerns a Portuguese project, however, I assume
that most of their sources are likewise of Portuguese, rather than Brazil-
ian, origin.
Their list of English loan words contains four verbs integrated in
the Portuguese conjugation class ending in -ar (28), in addition to the
11
Apostolou-Panara, Goutsos, Gardner-Chloros and Zarpatea cite 14 different Eng-
lish verbs in the periphrastic construction. kmno wash is shared between Goutsos and
Zarpatea. kmno use is cited in the papers by Gardner-Chloros and Zarpetea. But these
two papers must be based on (partly) the same data. I infer this from the fact that the
example
very busy I know that you have (are) very busy is cited in both
(Gardner-Chloros 1992, 128; Zarpatea 1995, 578).
denominal form stressar (from stress) and the de-adjectival compactar

(to compress from compact, computer terminology).
(28) draftar to draft
driblar to dribble
linchar to lynch
snifar to sniff [drugs]
European Portuguese/English (Queiroz et al. 1999)
In addition this list contains six deverbal English nouns denoting activi-
ties in the domains of sports and information and communication tech-
nology: jogging, kickboxing, surf; scan, upgrade and zap. The web site
does not indicate how these words are used in Portuguese. Nicholas
Hurst (p.c. 18 Apr 2002) cites examples of periphrastic constructions
involving the Portuguese verb fazer to do (29). The examples show that
there is much variation: as in Greek contexts, the English verb is used in
either the citation form (infinitive) or the gerund. Moreover, the indefi-
nite article um may or may not be present.
(29) fazer zapping
fazer um scanning
fazer surf
fazer um upgrade
European Portuguese/English (p.c. Nicholas Hurst 18 Apr 2002)
5.2 Brazil
Kennedy (1971) and Harmon (1994) quote a number of English verbs
occurring in Brazilian dictionaries and other written sources. All are
integrated into the so-called first conjugation class ending in -ar:
(30) chutar to shoot [ball game]
drenar to drain
ranquear to rank
treinar to train
Brazilian Portuguese/English (Kennedy, 1971)
Harmon does not mention the construction with fazer. Kennedy cites
just one example of a somewhat enigmatic periphrastic construction:
fazer o footing, translated in English as take a stroll.
306 louis boumans
5.3 American Portuguese

There is a classical study of the language of Portuguese immigrants
in the US by Pap (1949). He gives many examples of morphologically
integrated English verbs. These are added to the most frequent en most
productive Portuguese conjugation class in -ar, with the necessary pho-
nological adaptations. In some cases the English verb stem is extended
with an -e- such that the infinitive ends in -ear.
(31) raid-ear to ride
canec-ar to connect
damp-ar to dump
s(e)leir-ar to slide
chinj-ar to change
raiv-ar to drive
North American Portuguese/English (Pap 1949, 95100)
While morphological integration of English verbs is the most common
pattern, Pap also lists a large number of instances of a periphrastic con-
struction with fazer. The nominalized English verb stem occupies the
direct object position. In contrast with the contemporary European
examples cited by Hurst, the foreign verb stem is preceded by the Portu-
guese masculine definite article o.
(32) fazer o boda to bother
fazer o spoil to spoil
fazer o save to save
fazer o find out to find out
fazer o give up to give up
North American Portuguese/English (Pap 1949, 105106)
6. Discussion
The data from North African varieties of Arabic in contact with Euro-
pean languages suggest that diaspora communities are more inclined
to use the periphrastic alternative for incorporating foreign verbs. The
studies on Greek and Portuguese in contact with English do not unam-
biguously confirm this. Morphological integration as well as periphras-
tic constructions are reported for all contact situations, whether in the
Greek and Portuguese speaking countries or in the respective immi-
grant communities in Anglophone countries.
Broadly speaking, however, the literature on Greek and Portuguese

does point in the same direction. Integration in the productive conju-
gation classes, Greek -ro and Portuguese -ar, is predominant in the
studies on English loanwords in Greece, Cyprus, Portugal and Brazil.
Do-constructions, on the other hand, feature prominently in the stud-
ies on migrant communities.
A caveat is in order: studies on the influence of English as a foreign
language are traditionally oriented toward loan words, i.e. words in
isolation, in written sources. These studies tend not to identify foreign
verbs in the periphrastic construction as verbs at all, which can easily
lead to the suggestion that embedded verbs are either absent or consis-
tently integrated into the morphology of the matrix language. For obvi-
ous reasons, studies on bilingualism in migrant communities focus on
oral communication. Thanks to the recognition of code-switching as
an interesting linguistic phenomenon, the latter studies also pay more
attention to the higher degrees of linguistic organisation like clauses and
utterances. Hence the emphasis on one verb strategy or the other may be
to some extent an artefact of the different research traditions.
Still, at least the Greek case supports my hypothesis, in particular
the studies by Apostolou Panara on Greece and Seaman on the Greek
community in the US. Both authors discuss the different types of verb
integration, and designate one as the default strategy: morphological
integration in Greece and periphrasis in the US. The studies on the Lon-
don Cypriots suggest that periphrasis is the most productive strategy
there as well.
With respect to the sociolinguistic factors in verb integration, I con-
clude that, all other things being equal, morphological integration is
the norm in homeland settings where the matrix language is socially
dominant, while the periphrastic construction is found in migrant com-
munities where the embedded language is dominant. The migrant com-
munities concerned here are actually undergoing language shift. Stated
in more general terms, the periphrastic construction characterises situa-
tions of intense contact with the language of the embedded foreign verb.
As mentioned above, other factors may override the sociolinguistic
ones, as in the case of matrix languages that opt for the same integration
strategy in no matter what sociolinguistic setting.
308 louis boumans
6.1 Automatization of superimposed language practices

The next question is why speakers more often revert to periphrasis
in situations of more intense contact. I can think of the following
explanations:
Firstly, automatization of superimposed language practices: In gen-
eral the degree of integration of foreign words is inversely proportional
to the speakers knowledge of the embedded language. With increasing
practise and competence, the linguistic habits characteristic of that for-
eign language become entrenched in the speakers brain, and can more
easily be reproduced. This even results in the disintegration of formerly
integrated foreign words, a process known as denativization (Haugen
1950). In Morocco, for instance, French-Moroccan Arabic bilinguals
now prefer to reborrow the same French terms in shapes closer to the
French prototype (Heath 1986, 114). Denativization has often been
described with respect to phonology of loan words from English, a lan-
guage whose influence is expanding (cf. Khubchandani 1968, Hasselmo
1969, De Reuse 1994, Van Ness 1994). I would like to stress this point,
because it is a commonly held belief that loanwords become, on the
contrary, more integrated over time (Nortier and Schatz 1988, Heath
1989).
The replacement of morpho-phonological integration by periphrasis
as the productive strategy for the integration of foreign verbs fits into
the denativization process, since the periphrastic construction entails
little or no phonological or morphological adaptation to the matrix lan-
guage. This type of change could be taking place in for instance Greece
or Portugal, where exposure to English is still increasing.
Denativization refers to a diachronic change in situations where lan-
guage contact increases over time. However, the underlying mechanism
of automatization of superimposed language practices also explains the
differences between immigrant and homeland bilingualism.
6.2 Loss of community language practices

Secondly, attrition or loss of community language practices: the mor-
phological integration of foreign verb requires the application of matrix
language morphological procedures. These procedures are typically the
same as those used for deriving verbs from nouns or adjectives. The
loss or erosion of community language morphology, a common effect
of decreased use of the language and incipient shift (Andersen 1982,
El Aissati 1997), cuts off morphological integration as a strategy for

embedding foreign verbs.
6.3 Collocational complements

Thirdly, triggering by collocational complements of the embedded verbs:
Many examples of periphrasis with foreign verbs involve a collocation
with a foreign direct object noun or other complement, cf. wash and
cups in Greek/English example (27) above and diep, gesprek and voeren
in the Arabic/Dutch example below.
(33) ma ne-qder eh n-dir
NEG 1-can NEG er 1-do
diep-e gesprekk-en voer-en
deep-AGR conversation-PL conduct-INF
I cant er carry out deep conversations.
MA/Dutch, Jamal, 20, Eindhoven 1991 (Boumans 1998, 245)
The periphrasis leaves the original embedded language collocation
intact. Therefore the collocation might trigger this construction. The
usage of foreign collocations presupposes a higher level of competence
in that language and hence more intense exposure to it.
7. Conclusions
In nearly all cases, foreign verbs are combined with matrix language
inflection in either of two ways: they are morphologically integrated and
inflected with ML affixes (if any), or they are embedded in a periphrastic
construction with an inflected ML verb. Some MLs, such as Magribian
Arabic, Portuguese and Greek, allow for both solutions. A comparison
of Magribian, Portuguese and Greek bilingual communities suggests
that the periphrastic integration of foreign verbs is favored in situations
of more intense contact with the superimposed language.
8. References
Andersen, R. 1982. Determining the Linguistic Attributes of Language Attrition. In

R. Lambert and B. Freed, eds. The Loss of Language Skills. Rowley, MA: Newbury
House, 83118.
Apostolou Panara, A. 1991. English Loanwords in Modern Greek: an Overview. Termi-
nologie et Traduction, 1.91, 4560.
310 louis boumans
Aquilina, J. 1965 (1987). Teach Yourself Maltese. Valetta: Hodder and Stoughton.
Backus, A. 1996. Two in One: Bilingual Speech of Turkish Immigrants in the Netherlands.
Tilburg: Tilburg University Press.
. 2001. Synchronic Register Variation in Monolingual and Bilingual Turkish (project
description).
Boumans, L. 1998. The Syntax of Code-switching. Analysing Moroccan Arabic/Dutch
Conversations. Tilburg: Tilburg University Press.
. 2000. Periphrastic Verb Constructions in Moroccan Arabic/Dutch Code-switch-
ing. In A. Fenyvesi and K. Sandor, eds. Language Contact and the Verbal Complex of
Dutch and Hungarian. Working Papers of the 1st Bilingual Language Use Theme Meet-
ing. Szeged: University of Szeged, Teacher Training College, 6784.
. 2002. Repetition Phenomena in Insertional Codes-witching. In A. Rouchdy, ed.
Language Contact and Language Conflict Phenomena in Arabic. London: Curzon,
279316.
, and Caubet, D. 2000. Modelling Intrasentential Code-switching: a Comparative
study of Algerian Arabic/French in Algeria and Moroccan Arabic/Dutch in the Neth-
erlands. In J. Owens, ed. Arabic as a Minority Language. Berlin: Mouton de Gruyter,
pp. 11380.
Camilleri, A. 1994. Language contact between Maltese and English: code-switching
and cross-linguistic influence. In D. Caubet and M. Vanhove, eds. Actes des premires
journes internationales de dialectologie arabe de Paris. Paris: INALCO, 43149.
Cohen, M. 1912. Le parler arabe des juifs dAlger. Paris: H. Champion.
Contossopoulos, N.G. 1978. Linfluence du franais sur le grec: emprunts lexicaux et
calques phrasologiques. Athnes: s.n.
Cornips, L. 1994. De hardnekkige vooroordelen over de regionale doen+infinitief-
constructie. Forum der Letteren, 35.4, 28294.
De Reuse, W. 1994. English Loanwords in the Native Languages of the Chukotka Pen-
insula. Anthropological Linguistics, 36.1, 5668.
El Aissati, A. 1997. Language Loss among Native Speakers of Moroccan Arabic in the
Netherlands. Tilburg: Tilburg University Press.
Evripidou, D. 2001. Lexical Borrowing: A Study of English Loanwords in the Greek Cypriot
Dialect. Lancaster: Lancaster University.
Gardner-Chloros, P. 1992. The Sociolinguistics of the Greek Cypriot Community in
London. Plurilinguismes, 4, 11236.
Giesbers, H. 1984. Doe jij even lief spelen? Notities over het perifrastisch doen. Mede-
delingen NCDN, 19, 5776.
Goutsos, D. 2001. A Discourse-analytic Approach to the Use of English in Cypriot
Greek Conversations. International Journal of Applied Linguistics, 11:2, 194223.
Harmon, R.M. 1994. Aspectos lingsticos dos emprstimos em portugus. Hispania,
77:3, 463469.
Hartley, A.H. 2001. Loanwords in Modern Nautical Greek. (2002).
Hasselmo, N. 1969. On Diversity in American Swedish. Svenska Landsml och Svenskt
Folkliv, 92, 5372.
Haugen, E. 1950. The Analysis of Linguistic Borrowing. Language, 26:2, 21031.
Heath, J. 1986. Hasta la Mujerra! and other Instances of Playful Language Mixing in
Morocco. Mediterranean Language Review, 2, 1136.
. 1989. From Code-Switching to Borrowing. A Case Study of Moroccan Arabic. Lon-
don: Kegan Paul International.
Herrero Muoz-Cobo, B. 1996. El rabe marroqu: aproximacin sociolingstica. Alme-
ra: Universidad de Almera.
Igla, B. 1991. On the Treatment of Foreign Verbs in Romani. In P. Bakker and M. Cor-
tiade, eds. In the Margin of Romani. Gypsy Languages in Contact. Amsterdam: Inst.
ATW, Univ. Amsterdam, 505.
Kennedy, J.H. 1971. The Influence of English on the Vocabulary of Brazilian Portu-
guese. Hispania 54:2, 327.
Khubchandani, L.M. 1968. The Gender of English Loanwords in Sindhi. In B. Krish-
namurti, ed. Studies in Indian Linguistics. (Professor M.B. Emeneau sastipurti volume).
Poona: Deccan College, 18088.
Lontos, S.S. 192526. American Greek. American Speech, I, 30710.
Malkiel, Y. 1986. A Spanish Conjugational Model Superimposed on Portuguese. Medi-
terranean Language Review 2, 5166.
Mifsud, M. 1995. Loan Verbs in Maltese. A Descriptive and Comparative Grammar.
Leiden: E.J. Brill.
Muysken, P. 2000. Bilingual Speech. A typology of Code-Mixing. Cambridge: Cambridge
University Press.
Myers-Scotton, C. 1993. Duelling Languages. Grammatical Structure in Codeswitching.
Oxford: Clarendon Press.
Nortier, J. and Schatz, H. 1988. Van nwoordwisseling naar ontlening, een vergelijk-
end onderzoek. Forum der Letteren, 29, 16178.
Nuijtens, E.T.G. 1962. De tweetalige menseen taalsociologisch onderzoek naar het
gebruik van dialect en cultuurtaal in Borne. Assen: Van Gorcum.
Pap, L. 1949. Portuguese-American Speech. New York: Kings Crown Press.
Papapavlou, A. 1997. The Influence of English and its Dominance in Cyprus: Reality or
unfounded fears? Journal of Mediterranean Studies, 7:2, 21849.
Queiroz, A., Rodrigues, J. and McKenny, J. 1999. Emprstimos da lngua inglesa para a
portuguesa. English loan words and phrases in Portuguese. Escola Superior de Edu-
cao de Viseu. 2002.
Seaman, P.D. 1972. Modern Greek and American English in Contact. The Hague:
Mouton.
Swanson, D. 1958. English Loanwords in Modern Greek. Word, 14, 2646.
Tamis, A. 1986. The State of the Modern Greek Language as Spoken in Victoria. Mel-
bourne, University of Melbourne.
Thomason, S.G. and Kaufman, T. 1988. Language Contact, Creolization and Genetic Lin-
guistics. Berkeley: Univ. of California Press.
Treffers-Daller, J. 1994. Mixing Two Languages. French-Dutch contact in a comparative
perspective. Berlin: Mouton de Gruyter.
Van Dijk-Wittop Koning, A.M. 1963. De continuteit van het Grieks: onderzoek naar de
herkomst van de woordvoorraad van het Nieuwgrieks. Zwolle: W.E.J. Tjeenk Willink.
Van Ness, S. 1994. Die Dimensionen lexicalischer Entlehnungen im Pennsylvanien-
deutschen von Ohio (USA). Zeitschrift fr Dialektologie und Linguistik, 61.3,
27997.
Van Staden, M. 1999. Where does Malay end and Tidore begin? In R.A.C. Dam, ed.
Perspectives on the Birds Head of Irian Jaya, Indonesia. Proceedings of the conference
Leiden, 1317 October 1997. Amsterdam: Rodopi, 691716.
Wernitz, C.J. 1993. Bedingungen und Voraussetzungen fr Sprachwechsel. Eine Untersuc-
hung zum Sprachwechsel bei bilingualen Marokkanern in Frankreich. Frankfurt: Peter
Lang.
Zarpetea, P. 1995. Code-switching and Lexical Borrowing (loanwords) in the Speech
of Three Generations of Greek Cypriots in London (Harringey). Studies in Greek
Linguistics. Proceedings of the 16th Annual Meeting of the Department of Linguistics.
Thessaloniki: Aristotle University, 57687.
FAULA, FAILA, FAALA :
DISPERSION ET RGULARITS SMANTIQUES
DANS LES TROIS SCHMES SIMPLES DU VERBE ARABE
Joseph Dichy
Universit Lumire-Lyon 2
Pour Kees Versteegh

Aut cuspis sic vita fluit, dum stare videtur.
1. Introduction1
Les schmes du verbe augment (mazd) en arabe sont traditionnelle-

ment associs des valeurs smantiques de base, telles que litration,
lintensivit, la causativit et la factitivit, la rflexivit, etc. Si les lis-
tes dexceptions qui accompagnent ces descriptions indiquent que le
problme, malgr des avances significatives, est loin dtre rsolu de
manire satisfaisante, la question des valeurs smantiques associables
aux trois schmes simples (mujarrad) ou non-augments faula, faila et
faala2 semble offrir encore plus de rsistance lanalyse.
La description de ces schmes dans les grammaires arabisantes est,
lorsquelle existe, bien plus sommaire que pour les verbes augments.3
1
Je reprends ici, sous une forme largement revue, le chapitre 3 de mon cours de pr-
paration lagrgation darabe du Centre national denseignement distance (CNED),
2002 et 2003. Par souci dexplicitation, jai prsent quelque peu en dtail le cadre concep-
tuel ncessaire au traitement de cette difficile question, et conserv des dfinitions ou
des indications qui pourront paratre videntes, mais ne le sont pas ncessairement.
Ce travail doit beaucoup au cadre thorique et aux travaux de Jean-Pierre Descls,
ainsi quaux sminaires que nous avons anims, et continuons danimer ensemble,
depuis 2001, sur les valeurs associes en contexte aux formes verbales en arabe et en
franais. Je remercie galement les membres de son quipe de recherche, notamment
Brahim Djioua, pour de nombreuses et fcondes discussions.
2
La convention qui prvaut dans la tradition linguistique arabe dsigne ce que nous
appelons les consonnes radicales 1, 2 et 3 (de la racine smitique) respectivement par
f, et l.
3
Ainsi pour les grammaires en langues occidentales : Neyreneuf et Al-Hakkak (1996,
3542) signalent des nuances de sens . . . pour chaque forme drive, mais non pour
les verbes simples (id., 2829). De mme, les descriptions consacres par Blachre et
314 joseph dichy
Il suffit dobserver quils demeurent dsigns dans la tradition arabi-

sante occidentale comme la Forme I. Cette dsignation relve dune
ide gnrale qui rpond, dfaut de dcrire les donnes linguistiques,
une cohrence interne. La Forme I est appele verbe primitif par A.
Sylvestre de Sacy, par rfrence au verbe nu mujarrad des grammaires
arabes (1831, I : 123). A cette forme primitive, Sylvestre de Sacy associe
les grandes divisions smantiques gnrales des verbes (id. 121122) sur
lesquelles il fondera ensuite son analyse des formes drives (id. 129143).
Mais il nidentifie pas de valeurs smantiques associes en propre aux
schmes faula, faila et faala, dans lesquelles il ne voit quune forme
primitive ou nue (mujarrad) partir de laquelle les formes drives ou
augmentes (mazda) sont construites.
Si la tradition arabisante a pu ainsi rapporter ces trois schmes une
seule forme, cest sans doute aussi pour une autre raison : ces derniers
apparaissent en effet comme instables, tant du point de vue formel, o lon
observe des modifications de la voyelle de la 2e radicale (section 2 ci-des-
sous), que du point de vue smantique, o les notions trs gnrales de
transitivit, de verbes daction ou dtat par lesquelles on a essay de les
caractriser rencontrent un nombre lev de contre-exemples. Cest ainsi
que les trois pages dindications essentiellement formelles consacres
cette question par la grammaire da-artn sont prcdes de cet aver-
tissement : On ne peut tablir [la forme des verbes relevant de] ces schmes
(awzn) quen ayant recours des ouvrages lexicographiques (1910, 11).
Faut-il donc renoncer ? Il y a de solides raisons de penser que non. Plu-
sieurs essais de mise en ordre et danalyse ont t tents. La synthse la
plus rcente de ces tentatives est, ma connaissance, celle de Pierre Lar-
cher (2003, chap. II), qui comporte en outre des propositions origina-
les. Signalons galement les analyses incontournables dAndr Roman
(1983, II : 886900). Dautres auteurs, au premier rang desquels Marcel
Cohen, Paul Joon, Jean Cantineau et Henri Fleisch ont t utiliss. En
outre, un nombre considrable dobservations et danalyses a t effectu
dans les textes des sciences linguistiques arabes mdivales. Je ferai lar-
gement usage de ces travaux, sans toutefois les discuter en dtail, pour
Godefroy-Demombynes (1952, 4970) aux formes drives contrastent avec labsence

danalyse de la Forme I. Belot, (1922, 15), Badawi, Carter et Gully (2004, 60) ne signa-
lent pour la forme primitive ou de base (al-fil al-mujarrad), de manire trs pru-
dente, que la transitivit ou son absence en fonction de la voyelle mdiane de faala,
faila et faula. On trouve en revanche des propositions de description du sens dans
Caspari-Uricoecha (1881, 3233), Wright (189698 I : 30), Brockelmann (1948, 35) ou
Boormans (1967, 10), ainsi, naturellement, que dans la tradition linguistique arabe (voir
la synthse de Nr ad-Dn 2002, 177186).
faula, faila, faala 315
viter dallonger indment cette tude. A lenttant problme pos par la

dispersion smantique des verbes de schme simple en arabe jai cherch
apporter une solution fonde sur une dfinition aussi rigoureuse que
possible des termes utiliss, ainsi que sur une conception, me semble-t-il,
renouvele de la relation entre sens et forme dans les schmes (au sens
que prend ce terme dans le domaine smitique) en arabe (Dichy 2003).
Les propositions qui suivent sappuient en outre sur un dpouillement
systmatique des donnes lexicales dans le roman de Tawfq al-H akm,
Yawmiyyt nib f l-aryf (abrg en Yawm.) dune part, et dans plu-
sieurs dictionnaires arabes de lautre (notamment Hans Wehr-Cowan
1979 et al-Mujam al-Wst). Jai fait largement usage de la base de don-
nes lexicale DIINAR.1 (Dictionnaire informatis de larabe, version
1 Dichy, Braham, Ghazeli et Hassoun 2002 ; Dichy et Hassoun 2005 ;
http ://www.elda.org), du concordancier informatis construit par
R. Abbs (2004), ainsi que de la liste indexe de 20.000 verbes rali-
se pour les besoins du Bescherelle des Verbes arabes (Ammar et Dichy
1999). Les rfrences ces dernires sources ne sont pas donnes, la
diffrence des textes arabes mdivaux, des travaux arabisants de lpo-
que moderne ou, au besoin, des dictionnaires ; Yawm est indiqu avec
le numro de la page suivi de la ligne lorsque le sens doit tre interprt
en fonction du contexte.
2. Les schmes du verbe non augment, aspects formels
2.1 Lalternance vocalique de la voyelle de la 2e radicale

Les trois schmes de base du verbe simple ont t identifis dans
les sciences mdivales arabes du langage partir de la voyelle de la
deuxime radicale ou voyelle mdiane du verbe au paradigme du md
(ou suffix),4 respectivement u, i et a. Les trois familles morphologi-
ques correspondantes sont numres sous forme dexemples (amtila)
4
Les termes daccompli et inaccompli, dachev et dinachev ou de perfectif et imper-
fectif, pourtant reus, comportent un problme, celui de dsigner la forme morpholo-
gique par un terme dnotant lune des valeurs aspectuo-temporelles que cette forme
peut prendre en contexte. Do le choix de dsigner ces paradigmes verbaux par leur
nom arabe ou par le trait morphologique qui les caractrise. Ainsi, le md (tradition-
nellement accompli), dans lequel le morphme de personne prcde la base du verbe,
correspond ici au suffix ; le mudri (traditionnellement inaccompli), dans lequel ce
morphme prcde la base, est dsign du terme de prfix (cf. Moscati, d. 1964, 1312 :
prfix-conjugation vs suffix-conjugation).
316 joseph dichy
par Sbawayhi (m. vers 180/796) ds lintroduction du Kitb (I : 12), et

reprises au cours de ce trait au moyen des conventions faala, faila et
faula (dsignes comme des formes construites, abniya) avec des indi-
cations morphologiques, syntaxiques et smantiques (Kitb IV : 567,
notamment 38). Je laisse de ct ici la question, qui a t pose, de savoir
si la forme sujet non exprim (dite passive) fuila ne constituerait pas
un schme part entire.5
Selon la reconstruction du proto-systme de la langue arabe, telle
quelle a t propose en diachronie (et en tout cas dans lorganisation
qui sous-tend le systme de la langue), la forme simple du verbe com-
porte trois schmes, reprsents dans le tableau ci-dessous (voir notam-
ment Cantineau 1950, 77 ; Fleisch 1957 et 1968, 115119 ; Roman 1983,
II : 894) :
Md. ( suffix) Mudri'( prfix)

Schme simple 1 faula yaf ulu
Schme simple 2 faila yaf alu
Schme simple 3 faala yaf ilu
FIGURE 1 LES SCHMES POSTULS DU VERBE SIMPLE DANS

LE PROTO-SYSTME DE LARABE
Selon Moscati, d. (1964 : 122), qui se situe dans la perspective des tudes
smitiques compares, lantiquit de ce schma vocalique trois ter-
mes en arabe est confirm par quelques-unes des manifestations les plus
anciennes du groupe smitique du Nord-Ouest, cest--dire, par lamorite,
lougaritique, et les gloses de Tell Amarna . . .6 Dans cette reconstruction,
comme le souligne Roman (loc. cit.), les schmes sont toujours identifia-
5
Voir sur ce point Cantineau (1950) ; Roman (1983 II : 897900 ; 2005, 33), qui voit
dans la premire voyelle la marque de la diathse (subjective ou objective) ; Fleisch
(1957 ; 1968, 246 ; 1979) ; Larcher (1996 ; 2003, 2628). Nr ad-Dn (2002, 185186)
signale des traits tardifs, dans lesquels la forme fuila est considre comme un qua-
trime principe (asl), i.e., un quatrime schme du verbe simple, et non une forme de
la conjugaison (comme dans , m. en 643/1245, arh al-Mulk : 30).
6
The antiquity of this threefold vocalic scheme in Arabic is confirmed by some of
the oldest manifestations in North-West Semitic, i.e. Amorite, Ugaritic, and the Tell
Amarna glosses. In the prefix-conjugation the variation in the second vowell is at least
partly paralleled : u or i corresponding to a, and a to i, while u generally remains (Mos-
cati, d. 1964, 122).
bles en fonction de la voyelle de la deuxime consonne radicale (en gras

dans la transcription), et ce, dans chacun des deux grands paradigmes
de la conjugaison. Dans ltat de langue arabe tel quil nous est parvenu,
ds les poques classiques, les trois schmes simples ci-dessus ne sont
plus directement identifiables, comme le rappelle le tableau suivant :
Md( suffix) Mudri' (prfix)

Schme simple 1 faula yaf ulu
Schme simple 2 yaf alu
faila
Sous-catgorie formelle yaf ilu
Schme simple 3 yaf ilu
Sous-catgorie formelle 1 faala yaf ulu
Sous-catgorie formelle 2 yaf alu
FIGURE 2 LES SCHMES DU VERBE SIMPLE ET LEURS SOUS-CAT-

GORIES FORMELLES, TELS QUILS NOUS SONT PARVENUS
Cette situation continue de permettre lidentification au md (suf-

fix) de trois schmes distingus au moyen de la voyelle qui affecte la
2e radicale, mais non au mudri (prfix). Lorsque lon traite daspects
formels, comme dans la conjugaison ou dans un dictionnaire, il est
donc ncessaire de caractriser le verbe par lalternance vocalique de la
voyelle mdiane aux deux paradigmes du suffix et du prfix, exemple :
daraba/yadribu frapper, dalternance vocalique a/i. (Jcrirai ici, selon
une convention reue : daraba i.)
2.2 Les zones de stabilit ou dinstabilit formelles de lalternance

vocalique
Comme le montre la figure 2, le schme simple 1 (alternance u/u) est
formellement stable. Dans les schmes faila et faala en revanche, on
trouve des sous-catgories formelles dans lesquelles la voyelle de la 2e
radicale varie. Ces variations sont souvent corrles la prsence, au
sein de la racine, dune cause de transformation (illa) :7 deuxime et
7
Sur le statut pistmologique de cette notion dans la tradition linguistique arabe
318 joseph dichy
troisime radicales identiques, consonne radicale /w/ ou /y/, ou encore

radicale // en premire, deuxime ou troisime position. Les relations
entre ces causes de transformation et les variations observes dans les
schmes du verbe simple ont t, sinon expliques, du moins largement
inventories dans les sciences linguistiques arabes mdivales (voir
rsum dans a-artn, 1910 : 1014 ; Qabbwa, 1998 : 8594).
2.2.1 La sous-catgorie formelle faila/yaf ilu du schme simple 2

La sous-catgorie formelle du schme simple 2 ( faila/yaf ilu) est corr-
le dans la plupart des cas une racine de premire radicale w ou y. Elle
ne concerne quun petit nombre de verbes. Trois dentre eux, repris de
Sbawayhi (Kitb IV : 3839) et rgulirement cits dans les traits lin-
guistiques arabes mdivaux ou par les grammaires, relvent de racines
sans cause de transformation : h asiba i ou a croire, estimer ; baisa i ou
a tre misrable ; naima i ou a tre doux, fin. Ils peuvent tre consid-
rs comme ngligeables, car ils supportent galement une ralisation a
de la voyelle mdiane du mudri prfix.8
2.2.2 La sous-catgorie formelle faala/yaf alu du schme simple 3

Une seule modification de la voyelle mdiane est suffisamment rgu-
lire. Elle est largement signale dans les ouvrages arabes mdivaux
(cf. par exemple, al-Zajjj, Jumal, 396397). Il sagit de la sous-cat-
gorie formelle 2 du schme simple 3 ci-dessus ( faala/yaf alu) : elle se
produit pour simplifier si la 2e ou la 3e radicale du verbe est lune
des six consonnes darrire (h urf al-h alq), i.e. appartient lensemble
{, h, h, , , x}. Cette rgle sapplique selon une gradation, de manire
hautement frquente pour les deux consonnes les plus antrieures (les
consonnes glottales et h), et nettement moins frquente pour les deux
mdivale, et notamment chez Ibn Jinn (m. en 393/1002), voir Guillaume 1984. La
liste et les combinaisons des causes de transformation (il peut y en voir plus dune dans
une racine donne) ont t prsents sous une forme modlise dans Dichy (1993) et
Ammar et Dichy (1999, 3134).
8
Voir par exemple Ibn Xlawayh, m. en 370/981, Laysa f kalm al-arab Point ne
se trouve dans le langage des Arabes . . . (sous-entendu hormis, lexception de), chap.
10 ; az-Zajjj, m. en 337/949, Jumal : 398 (plus cohrent que le prcdent). Sur le statut
de la classe h asiba/yah sibu, notamment chez Ibn Jinn, m. en 392/1002, cf. Guillaume
(1984, 427433).
consonnes les moins antrieures ( et x) (cf. Ibn Ya, arh al-Mufassa l

VII : 153154, repris in Ammar et Dichy 1999 : 24).
2.2.3 La sous-catgorie formelle faala/yaf ulu du schme simple 3

La sous-catgorie formelle 1 du schme simple 3 ( faala/yaf ulu) ne
parat pas, quant elle, pourvoir tre lobjet de corrlations systmati-
ques entre la voyelle de la 2e radicale et des traits phontiques affectant
les consonnes du verbe. On trouve ainsi un nombre non ngligeable de
verbes dont le prfix (mudri) admet les deux voyelles mdianes i et
u, ainsi : azafa/yazifu et yazufu (an) se dtourner (de) ; qafala/yaqfilu
et yaqfulu revenir (de voyage), rentrer, etc. La prsence, dans de tels
exemples, des deux ralisations met en dfaut toute corrlation entre les
traits phontiques des consonnes radicales et la voyelle mdiane dans la
langue arabe telle quelle nous est parvenue, avec ses normes stabilises
par les travaux des lexicographes et lautorit des grammairiens.
3. Aspects smantiques, concepts de base
Considrons maintenant les valeurs smantiques associables aux sch-

mes simples. Les deux notions qui ont t traditionnellement utilises
constituent un cadre conceptuel dont on verra quil ne peut suffire. Il
sagit du trait syntaxique de transitivit ( 3.1.1) et de la distinction entre
verbes daction et verbes dtat ( 3.1.2). Ces notions devront tre repri-
ses dans un cadre conceptuel largi (sections 3.2 3.4). Les descriptions
proposes ci-dessous sappuieront sur une srie dexemples, choisis pour
illustrer la plupart des types de valeurs smantiques qui caractrisent
ces schmes.
3.1 Les deux principales notions traditionnellement utilises pour

dcrire le sens des schmes simples
3.1.1 Le trait syntaxique de transitivit/intransitivit

La premire distinction que lon rencontre dans les grammaires arabisan-
tes comme dans les sciences mdivales arabes du langage est de nature
syntaxique. En fonction de la voyelle a, u ou i affectant la deuxime
consonne radicale, ces verbes sont dits :
320 joseph dichy
pour le schme faala, transitifs, bien que comportant un nombre

considrable dintransitifs,
transitifs et intransitifs pour le schme faila, et
toujours intransitifs pour les verbes en faula.9
Les traits syntaxiques de transitivit ou dintransitivit sont cruciaux

dans la thorie de la rection qui prvaut dans la tradition linguistique
arabe mdivale. Leur incidence sur le sens des verbes nest toutefois
pas directe : elle en est plutt une consquence quune cause. Une autre
question que la thorie de la rection laisse dans lombre est celle de la
transitivit indirecte, dans laquelle le complment du verbe est un syn-
tagme prpositionnel. La transitivit, envisage sous ses deux formes,
directe et indirecte sera naturellement incluse dans les valeurs smanti-
ques associes aux verbes des schmes simples, particulirement en ce
qui concerne le schme faala.
3.1.2 La partition traditionnelle entre verbes daction et verbes dtat

Les arabisants et les smitisants occidentaux ont gnralement fond
leurs analyses des verbes des schmes simples sur une partition gn-
rale en verbes daction, ou actifs, et verbes dtat ou statifs, les seconds
tant ncessairement intransitifs, et les premiers, transitifs ou intransi-
tifs selon les cas. Dans son principe, le schme faala aurait ainsi corres-
pondu des verbes daction (transitifs ou intransitifs), le schme faula
des verbes statifs toujours intransitifs exprimant un tat permanent
ou une qualit inhrente naturelle, et le schme faila des verbes gn-
ralement intransitifs indiquant soit un tat ou une condition tempo-
raire, soit une qualit purement accidentelle dune personne ou dune
chose (Wright 189698 I : 3638). Dans cette description, le schme
faila pose un problme, qui a t trait en recourant la notion de verbe
moyen dE. Benveniste (section 3.4 ci-dessous).
Si la notion de verbe dtat, une fois redfinie, rsiste lanalyse, celle
de verbe daction ne laisse pas, dans lusage qui en a t fait, dapparatre
comme indiffrencie. Que faire, par exemple de verbes comme zanna
u croire, saqata u tomber, halaka i prir, wajaba i devoir, amala u
englober ou ra a voir, qui relvent tous du schme faala, rput cor-
9
Cf. Sbawayhi (Kitb IV : 38), ainsi par exemple que az-Zamaxar (Mufassa l, 277
279). Lintransitivit des verbes dtat est rappele notamment par Wright (189698,
vol. I, 3638) ou Fleisch (1957 ; 1979). Pour lhbreu, voir par ex. Gesenius-Kautzsch-
Cowley (1910, 4143) ou Joon (1923, 4041).
respondre aux verbes daction ? Et que dire des paires akala u manger,
de schme faala vs. ariba a boire, de schme faila, ou ra a voir, vs.
samia a entendre ? La rponse consistant poser une catgorie inter-
mdiaire ou moyenne (Marcel Cohen 1929 ; Joon 1930 ; Fleisch 1957),
qui concerne au premier chef le schme faila tout en incluant des verbes
en faala, ne rsout que trs partiellement le problme. Les processus de
glissement smantique entranant le passage dun verbe daction un
verbe dtat, sils ont donn lieu quelques fines observations,10 nappor-
tent pas non plus de solution de porte gnrale. Mais le principal pro-
blme, comme la section 4.3 lillustrera, est que lapparente transparence
de la notion de verbe daction aura masqu la complexit des donnes
et la forte dispersion smantique du schme faala. On verra dans la
suite de ce travail quil nest de fait pas ncessaire de recourir aux verbes
daction pour dcrire les verbes en faala, malgr lintrt que prsente
par ailleurs cette notion.
Passons maintenant la dfinition des proprits fondamentales
ncessaires lanalyse des schmes simples : lagentivit (section 3.2), les
notions de procs, dvnement, de processus et dtat (sect. 3.3), celle de
verbe moyen ou de diathse interne (sect. 3.4).
3.2 La proprit dagentivit

Lagentivit se dfinit comme la participation de lagent au procs dnot
par le verbe, le critre de cette participation tant le contrle (Lyons
1978/90, 3.4 ; Descls 1994) de lagent sur ce procs. Il y a en arabe trois
degrs dagentivit :11 lagentivit pleine dans laquelle lagent contrle
effectivement le procs ; partielle, dans laquelle il nexerce quun con-
trle imparfait, et la non agentivit, dans laquelle il ne jouit daucun con-
trle (lagent existe, mais il ne concide nullement avec le sujet gramma-
tical). Ces trois degrs sopposent eux-mmes lagentivit neutralise,
qui correspond des procs dans lesquels il ny a aucun agent, ce qui se
10
Cf. par exemple pour lhbreu, Joon 1923, 41, pour larabe, Larcher 2003,
2728.
11
Cf. Roman (1990, 4243 ; 1999/2005, B-2.3.2). La dfinition ci-dessus de lagenti-
vit est en partie diffrente. Cf. galement lanalyse de Fleisch (1968, 116117), qui pose
une bipartition du verbe arabe : 1 le verbe agent (le sujet tant considr en tant
quagent) ; le verbe de qualit (le sujet tant simplement le qualifi), et distingue, dans la
premire catgorie lagent pur et simple de lagent intress.
322 joseph dichy
prsente typiquement dans les verbes valeur dtat caractristique (

4.1.1 ci-dessous). Exemples (verbes de schme simple) :12
(a) verbes dagentivit neutralise (valeur smantique tre x) ou non-

agentifs (valeur : devenir x) :
schme faula : karuma u, tre, devenir karm noble-et-gnreux ;
arufa u devenir, puis tre arf , i.e. noble, de premier rang social ;
saura u devenir, tre sar, petit ou : tre trs jeune ; latufa u tre,
devenir latf, subtil, fin, gracieux ;
schme faila : baliha a devenir ou tre ablah, faible desprit, stu-
pide ; salia a devenir ou tre chauve (asla) ; karia a tre ou devenir
rid, craquel (notamment : peau) ;
schme faala : xadaba i devenir, tre vert (plante, herbage, lieu . . .) ;
xaffa i devenir, tre xaff, lger (antonyme de schme faula : taqula
u devenir, tre taql, lourd) ;
(b) verbes non-agentifs (le sujet grammatical nest pas lagent du procs
dont il est le sige) :
schme faila : dasiqa a se remplir ( ras bords), dborder (bas-
sin . . .) ; marida a tomber malade ; qarima a avoir faim de viande
(par effet de privation) ;
schme faala : sakana u, se calmer (mer, vent, bruit . . .) ; waqaa a
tomber et galement : se trouver (ville, montagne, lieu . . .) ; halaka i
prir ;
(c) verbes partiellement agentifs :
schme faila : adiba a tre, se mettre en colre ; h azina a tre ou
devenir triste, prouver de la tristesse ; wamiqa i chrir ; aiqa a
tre ou tomber passionnment amoureux ; samia a entendre (sens
de loue) ;
schme faala : ra a voir (vue) ; amma u sentir (odorat) ;
(d) verbes pleinement agentifs :
schme faila : ariba a boire ; rakiba a chevaucher, monter (pour se
dplacer) ; h amida a louer (louange),
12
Certains des exemples donns dans ce travail existent aussi avec des formes corres-
pondant dautres schmes du verbe simple, quelquefois avec le mme sens. Ces autres
formes ne sont pas reprises ici, malgr lintrt quet reprsent la discussion.
schme faala : daraba i frapper ; jalasa i sasseoir ; da u revenir ;

dahaba a partir, aller ; tabaxa u (ou a) faire cuire ou bouillir, cui-
siner ; talaa a monter ; akala u manger ; qla u dire.
La sous-catgorie (a) inclut des verbes des trois schmes et les sous-cat-
gories (b) (d), des verbes des schmes faila et faala. Ces exemples, qui
anticipent par leur classement sur la suite de lexpos, montrent que la
proprit dagentivit, malgr limportance qui est la sienne dans la des-
cription smantique des schmes simples, ne peut, elle seule, consti-
tuer le principe qui en permettrait le classement.
3.2.1 Agent, patient, sujet grammatical

Il est essentiel, pour bien comprendre la notion dagentivit, de distin-
guer les termes dagent (et corollairement, de patient) du sujet gramma-
tical du verbe. Les formes verbales de larabe incluent en effet toujours
leur pronom sujet, qui correspond aux morphmes de personne suffixs
au md et prfixs au mudri, et concide avec le sujet grammatical du
verbe. Ainsi : jat l-amra correspond mot--mot : elle est venue la
princesse (Ammar et Dichy 1999, 1214). Cette structure est occulte
par lemploi indiffrenci, dans la tradition linguistique arabe, de fil
pour lagent comme pour le sujet grammatical.
Lagent ( fil) est ce qui effectue le procs dcrit par le verbe. Dans les
verbes dcrivant une action, lagent est lauteur de celle-ci. Dans les ver-
bes de perception ou de sentiment, par exemple, il est le sige de celui-ci.
Dans les verbes dcrivant un tat, le procs, comme on vient de le voir,
ne rfre pas un agent : le sujet grammatical inclus dans la forme verbale
est le sige de cet tat, quil sagisse dune personne comme dans raufa u
tre, devenir ou se montrer doux, compatissant ou dune chose, comme
dans daula u, devenir/tre troit ou petit (quantit). Dans les verbes
dagentivit partielle ou entire, lagent concide avec le sujet grammati-
cal. Ce dernier, la voix dite passive, concide avec le patient (qui subit
le procs dcrit par le verbe), exemple : suiltu jai t questionn. Dans
les formes non-agentives, le sujet grammatical subit galement le pro-
cs ou en est le sige, exemples :
saqatat al l-lard elle est tombe par terre (quil sagisse dune per-
sonne ou dune chose) ;
darasa r-rasmu la trace (du campement) sest efface : lagent est ici le
vent du dsert. (Ce thme est classique dans la posie arabe ancienne ;
324 joseph dichy
cf. la fin du 6e vers de la Muallaqa grande ode du pote emblma-

tique de lant-islam Imru al-Qays : Wa-hal al rasmin drisin min
muawwal, Sur traces effaces, qui sen irait pleurant ?)
3.3 Les proprits dvnement, de processus et dtat ( partir de

Descls 1994)
Le terme gnral de procs (auquel correspond en anglais celui de
situation)13 traduit le fait que le sens du verbe est inscrit dans une dure,
cest--dire, dans un intervalle de temps. Cette dure peut tre perue de
trois manires, correspondant aux notions aspectuelles dtat, dvne-
ment ou de processus. La dfinition de ces termes premiers est emprunte
directement J.-P. Descls.14 premire vue, ces dfinitions pourraient
apparatre comme dtournes : les trois termes aspectuels dvnement,
de processus ou dtat, qui dcrivent des phnomnes affectant des ver-
bes en en contexte sont rutiliss, dans ce travail comme dans ceux dont
la publication suivra, pour dsigner dans la smantique des schmes de
larabe des primitives incluses dans le sens grammatical associ aux sch-
mes simples et certains schmes augments. Mais y regarder de plus
prs, sagissant de valeurs smantiques primitives, il ny a rien dtonnant
ce quelles puissent galement tre observes dans les proprits lexi-
cales de familles morphologiques, telles que les schmes simples et les
schmes augments. La mise en lumire de ces observations constitue
une premire contribution thorique du travail prsent ici la dfini-
tion de ces valeurs primitives. Leur confrontation aux donnes de larabe
ma en outre conduit en proposer des extensions, particulirement en
ce qui concerne les sous-catgories de ltat ( 3.3.3).
13
Le terme de procs ne se traduit pas directement en anglais. Comrie (1985, 5),
Lyons (1978/1990, 3.4) ont recours au terme de situation pour recouvrir les tats dune
part, les vnements et les processus de lautre.
14
Mais voir aussi Comrie (1976, 1985), Lyons (1978/1990), ou, diffremment, la syn-
thse de Maingueneau (1994). La thorie des invariants cognitifs du langage et des pri-
mitives smantico-cognitives de J.-P. Descls est sous-jacente aux analyses prsentes
ici, notamment : lide que les primitives sont des oprateurs abstraits dtermins par
des proprits formelles, lhypothse selon laquelle elles sont ancres sur la perception
et laction, ainsi quun nombre important de points qui dcoulent de ces conceptions
(Descls 1990 ; 1994 ; Descls et Guentchva paratre). Le lecteur familier de ces tra-
vaux notera enfin que le terme de schme est employ ici la manire des travaux
arabisants et smitisants. On distinguera donc cet usage de celui des schmes sman-
tico-cognitifs (Descls, 1990 ; Abraham, 1995 ; Descls et al., 1998).
3.3.1 vnement
Laspect vnement [. . .] est lexpression dune occurrence qui apparat
sur un arrire-fond stable [. . . et] tablit une coupure entre lavant v-
nementiel (pas encore) et laprs vnementiel (ne plus). La dure de
lvnement est donc borne gauche par un avant et droite par un
aprs. Cet aprs peut concider avec un tat engendr par lvnement,
ou tat rsultatif (Descls 1994, 7375).15 Les exemples ci-dessous (verbe
concern en gras) illustrent quelques proprits fondamentales de las-
pect vnement :
m kidtu hd s-sabh a arifu finjna l-qahwa al maktab h att

waradat ira tilifuniyya . . . (Yawm., 123), peine, ce matin-l,
avais-je sirot le caf [pos] sur mon bureau, quun tlgramme tl-
phon survint . . . (mot--mot, lorsque apparut). La premire pro-
position de cette phrase (m kidtu . . .arifu peine . . . avais-je sirot)
constitue larrire-plan sur lequel se dtache lvnement h att wara-
dat lorsque apparut, vnement qui dtermine au sein de la situation
dcrite, un avant et un aprs.
wasala dayfun mundu sa notre invit est arriv depuis une heure :
de lvnement est arriv rsulte un tat, dont la dure englobe le
moment de lnonciation. Le locuteur rapporte lvnement de larri-
ve de linvit, et exprime par les mmes mots ltat dans lequel son
interlocuteur et lui se trouvent, et qui correspond laprs du procs :
leur invit est l depuis une heure. On dira, en simplifiant, que lv-
nement engendre ltat qui rsulte de lui.
ariqa r-rajulu lhomme sest noy (ou se noya) : lvnement dcrit
par le procs engendre ltat rsultatif noy, exprim en arabe par
ladjectif arq, associ ce verbe.
3.3.2 Processus
Laspect processus exprime un changement saisi dans son volution
interne. Tout processus exprime ncessairement un changement initial
(. . .) qui indique le dbut du processus. Le processus peut tre dcrit
15
Je fais dans ce travail un usage gnrique (ou hypronymique) du terme dtat rsul-
tatif. Ce dernier inclut ici ltat rsultant, qui est par ailleurs trs utile pour lanalyse des
valeurs aspectuo-temporelles en contexte, par exemple pour la description de ltat-pr-
sent, concomitant lacte dnonciation (Descls, 1994, 73). Ce choix, limit au prsent
travail, est d au fait que les valeurs smantiques associes aux familles morphologiques
constitues par les schmes ne sont pas sensibles cette distinction contextuelle.
326 joseph dichy
comme une succession dtats. Il peut, selon le cas, sorienter ou non vers
un terme explicite, qui correspond alors au dernier tat du processus ou
tat rsultatif (Descls 1994, 7580). Exemples :
Dans la phrase ci-dessus, le syntagme m kidtu arifu finjna l-qahwa

peine avais-je sirot le caf exprime un processus dont le terme est
spcifi : un moment donn, qui correspond ltat final du proces-
sus, le breuvage est entirement bu. La phrase dit mot--mot : . . . sirot
la tasse de caf ; la valeur smantique associe cette expression dans
le lexique de larabe implique que le contenu de la tasse a t bu en
entier.16
Dans arabati -amsu le soleil sest couch ou hal turdu an
tujanninan ? tu veux me rendre fou (folle) ? cest galement dans
la valeur lexicale du verbe que le terme du processus (le coucher du
soleil, la folie) est inscrit.
Le sens de amiya a perdre la vue comporte un terme, qui corres-
pond dans le lexique ladjectif am aveugle. Ce verbe dcrit, selon
le contexte, soit un vnement, soit un processus.
3.3.3 tat, types dtats et catgories drives ou corrles

Laspect tat exprime la stabilit de la situation rfrentielle reprsente,
cest--dire quil rfre une absence de mouvement et de changement.
Ltat dcrit un procs qui nest born ni gauche (en son commence-
ment) ni droite (le terme nest pas spcifi), et ne fait rfrence aucun
agent (Descls 1994, 7173).17 Il y a plusieurs types dtats. Deux dentre
eux, ltat caractristique ( 3.3.3.a) et ltat rsultatif ou tat acquis
( 3.3.3.b) intressent directement les schmes du verbe simple. Le second,
qui est produit par un changement dtat, correspond un glissement
smantique troitement corrl lusage des verbes dtat en discours.
Lanalyse ci-dessous dbouche sur ladjonction, la valeur smantique
de ltat caractristique, de celle de lacquisition dtat. Cette dernire
16
Ainsi, le dictionnaire glose lexpression raifa l-ina, mot--mot siroter un rci-
pient comme en aspirer le contenu jusquau bout (al-Mujam al-Wast : racine /r--f/) ;
de mme, ariba l-ina boire un rcipient correspond ariba kulla m fhi en boire
tout le contenu (Ilys et Nsf 1995, 164).
17
Comparer avec la dfinition de Maingueneau (1994, 64) : Les tats . . . nont ni
dbut, ni fin, ni milieu, ils ne supposent ni agent ni changement (ex. Luc est paresseux).
la diffrence de Comrie (1976, 1985), Lyons (1978/1990), Descls (1994) et du prsent
travail, cet auteur noppose pas tat, vnement et processus.
notion relve de celles, plus gnrales, dtat rsultatif et de changement

dtat. Lacquisition dtat constitue en effet la principale valeur smanti-
que corrle ltat caractristique ( 3.3.4).
3.3.3.a La notion dtat caractristique

Les donnes de larabe mont conduit proposer une sous-catgorie de
la notion gnrale dtat, celle dtat caractristique (Dichy 2002/2003).
Ce dernier est exprim par un verbe associ dans le lexique un adjectif,
traditionnellement appel adjectif assimil ou analogue [au participe]
(sifa muabbaha ci-aprs 4.1.2). Ltat caractristique par oppo-
sition ltat non caractristique , dcrit une proprit ou un attribut
de lobjet, pos, selon le point de vue, comme dominant, intrinsque,
inhrent ou inscrit dans la nature de celui-ci. Cet tat peut, comme on
le verra au paragraphe suivant, tre acquis lissue dun vnement ou
dun processus.
Le terme dtat caractristique, dont le contenu sera confront aux don-
nes du schme faula la section 4.1, est prfrable lemploi habituel
des grammaires arabisantes qui oppose, dans les verbes du schme sim-
ple, qualit permanente ou durable qualit temporaire (par exemple,
Wright 189698, I : 3638). La dure relative dun tat tout comme
celle dun processus ou dun vnement correspond en effet lune
des valeurs aspectuo-temporelles attribues une forme verbale donne
en fonction du contexte, et ne doit pas tre confondue avec une valeur
smantique morpho-lexicale, associe comme telle un schme donn.
Ainsi, h azina a tre ou devenir triste (de schme faila, adjectif cor-
respondant : h azn) est traditionnellement cit comme exemple de
qualit temporaire ou transitoire.18 Il rfre toutefois une qualit
durable dans les noncs suivants : fa-h azina il xiri umrihi Et il
fut triste jusqu la fin de ses jours ; inna lladna man . . . l xawfa
alayhim wa-l hum yah zann Ceux qui ont eu foi . . . aucune crainte
18
Ce verbe est cit par Caspari-Uricoecha pour illustrer un tat transitoire et passa-
ger ou bien une proprit ou une qualit qui naffecte lobjet du verbe que pendant peu
de temps (1881, 32) ; par Wright, comme illustrant a temporary state or condition,
or a merely accidental quality in persons or things (189698 I : 30) ; par Brockelmann
(1948, 35) comme exemple de verbes exprimant toujours une qualit ou une situation
accidentelle ou temporaire (Die Form faila steht durchweg fr zufllige, vorberge-
hende . . . Eigenschaften und Zustnde) ; ou Boormans (1967, 10), comme dcrivant un
tat passager.
328 joseph dichy
[ avoir] pour eux, il ne connatront [ternellement] aucune tristesse

mot--mot : et aucunement ne sattristeront (Coran 2 : 62). H azn sera
donc mieux dcrit comme un adjectif dtat non caractristique.
Inversement, le verbe xalua u mriter le nom de dprav, tre un
dprav (adjectif correspondant : xal), qui relve du schme faula,
rput renvoyer des verbes de qualit permanente ou durable
est glos dans le dictionnaire par lexpression taraka l-h ay wa-ra-
kiba hawhu, abandonner [toute] honte et se livrer ses passions
(al-Mujam al-Wast : racine /x-l-/). La qualit de xal dprav
(notoire) nest donc pas permanente. Xal peut en revanche tre
dcrit indpendamment toute rfrence une dure comme un
adjectif dtat caractristique.
Exemples de ces deux sous-catgories de la notion dtat :
verbes associs un adjectif dtat caractristique :

tla u tre tawl long ; hayifa a tre ahyaf , avoir la taille lance ;
awira a tre awar, borgne ; baxula u tre baxl, avaricieux (au
sens de tre avare par nature)19
verbes associs un adjectif dtat non caractristique :
farih a a tre farih , joyeux (lun des contraires de h azina ci-dessus) ;
yaisa a tre yis dsespr ; atia a tre atn ayant soif , avoir
soif ; marida a tre mard malade. . .
3.3.3.b Verbes dtat vs verbes de changement dtat produisant un

tat rsultatif
Pour H. Fleisch, cependant :
Le verbe de qualit nest pas un statif. Il signifie acqurir une qualit, autre-
ment dit : devenir tel (daprs la qualit en question) : karuma = devenir
19
Les proprits de lagentivit ne relvent pas dinvariants linguistiques, mais de la
reprsentation du procs vhicule par le vocabulaire de chaque culture. Ainsi, dans la
culture arabe mdivale, on peut, soit (a) se montrer avare envers quelquun dans une
situation donne, ou (b) tre avare par nature ( f t-tab) : cf. par exemple la prface du
Kitb al-Buxal Livre des avares dal-Jhiz, m. en 255/868 ; semblablement lavarice est,
lpoque de Molire considre comme pouvant tre inscrite dans le caractre de celui
qui, comme Harpagon, est un avaricieux. Le premier cas (a) est exprim au moyen du
verbe pleinement agentif baxila a al se montrer avare envers qqun, et le second (b),
par le verbe dagentivit neutralise baxula u tre un avaricieux (lavare est le sige du
vice inscrit dans sa nature).
karm (gnreux) ; ou bien, comme consquence de lacquisition : avoir

une qualit, soit : tre tel, cest un rsultatif, alors karuma = tre karm
(gnreux) (1968, 117).
Fleisch suit vraisemblablement ici la dfinition que donne al-Mubarrid
de la valeur smantique des verbes en faula :
innam huwa li-l-h l allat yantaqil ilayh l-fil . . . nahwu karuma wa-
zarufa wa-arufa elle (cette classe de verbes) dsigne uniquement ltat
dans lequel entre le sujet [du verbe] (mot--mot : ltat vers lequel se
dplace le sujet), exemples : karuma, zarufa et arufa (Muqtadab I : 209
traduction des exemples : devenir et donc tre noble-et-gnreux, deve-
nir, puis tre raffin et devenir, puis tre prminent, illustre, noble).
Ces deux descriptions font de ltat caractristique une valeur seconde
par rapport la valeur devenir qqch. Or Sbawayhi dcrit les verbes de
schme faula comme dnotant les caractres qui peuvent tre dans les
choses (al-xisl allat taknu f l-ay Kitb IV : 28). Ces termes ont
t par la suite largement repris dans les sciences linguistiques arabes
mdivales (cf. par ex. az-Zamaxar, Mufassa l : 279). Dans cette concep-
tion, la valeur fondamentale des verbes de qualit est stative, alors que
chez al-Mubarrid et Fleisch, elle correspond au changement dtat. Mais
faut-il vraiment opposer ces conceptions ? Est-il possible de dcider de
la valeur smantique qui serait premire par rapport lautre ?
Considrons les exemples suivants :
H asuna u tre h asan, beau, qaduma u tre qadm, ancien ou hayifa

a tre ahyaf , qui a la taille lance dsignent la base un tat carac-
tristique.
amasa i ou u devenir/tre ensoleill (jour) ; asaqa i devenir/tre
noir, obscur (nuit) rfrent fondamentalement un changement
dtat. Il sagit de verbes signifiant devenir qqch, dcrivant un proces-
sus qui aboutit un tat rsultatif non caractristique.
adiba a se mettre/tre en colre ou ja u avoir faim/tre affam (=
ayant faim) correspondent fondamentalement des verbes de chan-
gement dtat, dans lesquels le nouvel tat (non caractristique) qui
affecte le sujet grammatical rsulte du procs. Les adjectifs corres-
pondants, adib, adbn en colre et ji, jawn ayant faim sont des
tats rsultatifs non caractristiques.
Le verbe awira a devenir/tre borgne dsigne selon le contexte
soit un changement dtat (devenir qqch) dbouchant sur ltat carac-
tristique tre awar, borgne, soit cet tat caractristique lui-mme
(qui peut tre rsultatif ou non).
330 joseph dichy
Un verbe dtat caractristique correspondant fondamentalement

un tat, comme h asuna u tre beau ou hayifa a avoir la taille lance
ci-dessus peut aussi tre amen par le contexte dsigner le rsultat
dun vnement ou dun processus, exemple : hayifa l-na l-fat le
jeune homme est maintenant ahyaf , a maintenant une taille lance
(lnonc prsuppose : maintenant quil a grandi).
Ces exemples montrent qu la question pose (ce qui, de la valeur

smantique stative ou du changement dtat serait premier par rapport
lautre), il ny a pas de rponse gnrale, qui demeurerait valable dans
tous les cas. La rponse doit au contraire tre propose pour chaque
verbe, en fonction de son sens lexical (et des connaissances encyclop-
diques associes celui-ci). Cest ainsi que :
amasa i ou u devenir/tre ensoleill ou ja u avoir faim dcrivent

dabord un changement dtat (devenir qqch), et en second lieu, ltat
non caractristique correspondant.
Le sens des verbes dtat caractristique qaduma u tre ancien ou
jayida a avoir le cou long (signe de beaut) est peu compatible avec
un contexte dans lequel ces verbes prendraient la valeur devenir tel
(bien quun tel contexte demeure, comme on vient de lindiquer, pos-
sible).
Awira a, qui dcrit un tat caractristique, est compatible tant avec la
valeur stative tre borgne quavec celle du changement dtat (deve-
nir tel), sans quil soit possible de dcider de la valeur qui driverait
de lautre.
3.3.4. Lvnement de lacquisition dtat, cas particulier du changement

dtat et de ltat rsultatif
Mais il convient, pour comprendre la prsence dans ces verbes des deux
valeurs stative et dacquisition dtat, dinterroger plus avant la notion
dtat rsultatif. Ce dernier est engendr par un procs (vnement ou
processus) correspondant un changement dtat (Descls 1994, 71 et
suiv. ; Descls et Guentcheva paratre, chap. II). Dans les donnes sui-
vantes, cet tat met en jeu des notions dans lesquelles les proprits de
ltat, de lvnement et du processus se trouvent combines :
(a) tat rsultatif (caractristique ou non) engendr par un vnement :

Le verbe amura u galement ralis amira a est glos dans les
dictionnaires comme devenir mir, notamment par succession (al
Mujam al-Wast, al-Munjid, Hans Wehr-Cowan 1979). Ce verbe

dcrit, dans cette acception, lvnement dun changement dtat, et
le nom-adjectif associ amr, qui dsigne ltat caractristique mir,
prince, celui qui gouverne un mirat, correspond un tat rsultatif.
Mais il peut galement, comme nous le verrons au paragraphe sui-
vant, dsigner ltat caractristique correspondant, indpendamment
dun vnement ou dun processus qui laurait prcd.
H azina a tre ou devenir triste, sattrister dsigne ltat non caract-
ristique h azn triste ou lentre dans cet tat. Lvnement que consti-
tue cette entre peut tre caus par un autre vnement, par exemple
larrive dune mauvaise nouvelle : lamm samia xabara axhi, h azina
adda l-h uzni lorsquil entendit les nouvelles de son frre, il tomba
dans une profonde tristesse (mot--mot : il sattrista dune intense
tristesse). Ladjectif h azn, dcrit ci-dessus ( 3.3.3.a) comme un
adjectif dtat non caractristique peut donc en outre correspondre
un tat rsultatif, h azina relevant, dans ce dernier cas, des verbes de
changement dtat.
(b) tat rsultatif (caractristique ou non) correspondant au dernier tat
dun processus :
Comme rappel ci-dessus, un processus correspond une succession
dtats. Ltat rsultatif peut donc concider avec le dernier tat du pro-
cessus qui le produit. Exemples :
azuma anuhu ayan fa-ayan sa situation sociale est devenue peu

peu (ou devint peu peu) de grande importance (azm) ;
sawida a devenir ou tre aswad noir ; saura u devenir ou tre
(socialement ou quantitativement) sar petit ;
xadira z-zaru la rcolte a verdi ou verdit (Munjid, racine /x-d-r/ ;
Ilys et Nsf 1995, 113).
Ces donnes montrent que la notion dtat rsultatif est loin daller de
soi. Elles appellent deux remarques gnrales :
3.3.4.a Valeur aspectuo-temporelle du suffix (md) et vnement

engendrant un tat exprim ou non par un adjectif assimil
On connat en franais la valeur daccompli du prsent que peut pren-
dre le pass compos et qui indique dans une nonciation qui inclut
la fois le pass compos et le prsent de lnonciateur le rsultat
prsent dun procs accompli antrieurement (Maingueneau 1994,
332 joseph dichy
6768). Descls (1994, 73) dcrit ce rsultat comme un tat-prsent

concomitant lacte dnonciation. Comparons par exemple :
(1) Alors tu te dpches ? Je finis !

(2) Alors tu te dpches ? Jai fini ! (Leeman-Bouix 1994, 6162).
En (1), linterlocuteur est en train dachever quelque chose, le verbe

conjugu au prsent a donc valeur dinaccompli ; en (2), le pass com-
pos exprime un processus la fois accompli et achev (voir, pour cette
distinction, Descls 1994, 7680).
En arabe, la conjugaison au md (suffix) comporte galement la pro-

prit dengendrer un tat rsultatif, exemples :
saimtu taklifa l-h ayti . . . je suis las des fardeaux [dont nous charge]
la vie (Muallaqa grande ode du pote prislamique Zuhayr b. Ab
Sulm, 46e vers) : la forme conjugue au suffix (mot--mot : jai t
ou je fus las . . .) engendre ltat rsultatif tre las.
La traduction franaise : je suis las est contrainte dexprimer cet tat-

prsent (au sens de Descls 1994, 73), mais elle ne peut rendre par une
forme verbale lvnement qui lengendre. En arabe le md (suffix) de
saimtu englobe ces deux interprtations (cf. lexemple de wasala au
3.3.1) ;
atitu jai soif : la forme au suffix (mot-a-mot : jeus ou jai eu soif )

engendre ltat avoir soif (en traduction : jai soif mme commen-
taire que ci-dessus), qui pourrait tre glos comme : la soif sest main-
tenant installe en moi (ou sinstalla dsormais en moi).
Il est essentiel, pour comprendre le sens des schmes associs la notion

dtat, dobserver que ces exemples illustrent la prsence en arabe de deux
catgories diffrentes de verbes, selon que ltat rsultatif est exprim ou
non par un adjectif associ au verbe par la drivation morpho-lexicale.
Ltat rsultatif engendr par saimtu nest pas lexicalis : le cas est donc
comparable mutatis mutandis lusage du pass-simple franais.
Pour atitu, ltat qui rsulte du procs correspond une entre lexicale
(i.e. un adjectif assimil). Ainsi :
En reprenant les exemples (a) du 3.3.4, qui dcrivent des vnements :
Lnonc amura -iru (le pote est devenu/devint le prince

[rgnant] ou le pote rgne/rgna) peut tre glos comme : la posi-
tion de prince est maintenant (ou fut dsormais) occupe par le pote.
Le sens du verbe amura u est donc bien devenir, puis tre mir, et
non uniquement devenir mir, comme dans les dictionnaires cits
plus haut.
lamm samia xabara axhi, h azina adda l-h uzni lorsquil entendit
les nouvelles de son frre, il tomba dans une profonde tristesse cor-
respond : lorsquil entendit . . . ltat de tristesse sinstalla en lui.
Les exemples donns en (b), qui portent sur des processus, mritent un
mot dexplication : lachvement du processus parvenu son dernier tat
constitue lui-mme un vnement (Descls, 1994, 72, 7780) :
Dans xadira z-zaru la rcolte a verdi ou verdit cet vnement cor-

respond au moment o la rcolte parvient au degr de mrissement
attendu (ltat de verdeur).
Dans azuma anuhu ayan fa-ayan sa situation sociale est deve-
nue/devint peu peu de grande importance, cet vnement est celui
de lachvement du processus devenir peu peu, qui atteint alors son
dernier tat (celui de grande importance).
Dans tous ces cas, (1) il y a un avant et un aprs de lvnement du

changement dtat et de linstallation du nouvel tat, et (2) cet tat est
dsign dans le lexique par un adjectif assimil (sifa muabbaha). Lv-
nement pourrait de ce fait tre glos comme atteindre ou acqurir ltat
x, dans lequel x correspond un adjectif ou un nom-adjectif : atn
qui a soif ; amr mir ; h azn triste ; axdar vert ; azm de grande
importance.
3.3.4.b Les relations asymtriques tre x o devenir x et devenir x

tre x
La prsence dans le lexique dun adjectif assimil permet de mieux sai-
sir le lien entre tre x (valeur stative) et devenir x (valeur dacquisition
dtat) qui stablit entre les verbes associs un adjectif dtat caract-
ristique ou non caractristique dune part, et les verbes de changement
dtat de lautre. Ainsi :
334 joseph dichy
qadumat al-yawma dawlatuhum leur dynastie est devenue aujourdhui

ancienne peut tre glos comme : leur dynastie est devenue tant
ancienne (ou devint tant ancienne). Il sagit dun exemple de verbe
dtat caractristique (qaduma u tre ancien) dans lequel on observe
un glissement smantique en contexte de la valeur stative (tre qadm
ancien) celle dacquisition dtat (tre devenu tant qadm). Cest la
rencontre au sein de lnonc du sens lexical tre ancien et de la valeur
aspectuo-temporelle dachvement associe au md (suffix) elle-
mme renforce ici par le marqueur temporel al-yawma aujourdhui,
qui permet le glissement smantique tre x o devenir x observ.
adiba r-rajulu lhomme est (ou fut) en colre (adib) correspond
: ltat tre en colre sest accompli en lui. Cet exemple est celui dun
verbe dacquisition dtat non caractristique, dans lequel on ren-
contre un passage de la valeur de changement et dacquisition dtat
(devenir en colre) la valeur dtat rsultatif acquis (tre en colre),
la seconde tant engendre par la premire. Il peut tre glos comme :
tre devenu tant adib (en colre). Cest galement la rencontre en
contexte du sens lexical tre en colre et de la valeur dachvement
que prend le md (suffix) qui permet lengendrement de ltat acquis
tre x par la valeur de changement et dacquisition dtat devenir x.
Il y a donc deux cas gnraux de modification smantique corrls aux

verbes dtat. Les verbes dont la valeur de base est celle de ltat carac-
tristique connaissent un glissement smantique tre x o devenir x
(glissement de ltat au changement et lacquisition dtat). Dans les
verbes exprimant fondamentalement un changement correspondant
lacquisition dun tat (caractristique ou non), ltat acquis est engendr
par lvnement qui concide avec lachvement du processus dacquisi-
tion, ce qui peut tre reprsent schmatiquement comme : devenir x
tre x (dans lequel le symbole se lit engendre).
3.3.5 Les deux valeurs smantiques associes la notion dtat : ltat

caractristique et lacquisition dtat
Lanalyse ci-dessus conduit donc distinguer, pour les verbes associs
un adjectif dtat caractristique, dune part ltat caractristique (tre x)
et de lautre lacquisition dtat (devenir x), qui correspond un change-
ment dtat. Il leur est associ, respectivement, des adjectifs dtat carac-
tristique et des adjectifs dtat acquis.
Lexamen des donnes relatives aux verbes associs un adjectif dtat

non caractristique permet en outre dobserver que ces verbes relvent
tous du changement dtat : ils appartiennent de ce fait la catgorie des
verbes dacquisition dtat. A condition de tenir compte des glissements
smantiques dcrits plus haut (tre x o devenir x et devenir x
tre x), lanalyse de Fleisch et la conception dal-Mubarrid cites plus
haut ( 3.3.3.b), qui ne sapplique que partiellement aux verbes associs
un adjectif dtat caractristique, sapplique en revanche pleinement
aux verbes associs un adjectif dtat non caractristique, que la tra-
dition arabisante appelait tat temporaire. On en a examin ci-dessus
plusieurs exemples.
3.4 Benveniste revisit : la notion de verbe moyen (ou diathse interne) ;

le mdio-passif
Lautre grande distinction sur la dfinition de laquelle il convient de
revenir est celle de verbe moyen. On appelle verbe moyen selon la dfi-
nition dE. Benveniste, un verbe qui dcrit un procs dont le sujet est
le sige :20
Ici, le sujet est le lieu du procs, mme si ce procs, comme cest le cas
pour lat. fuor [jouir, avoir profit de] ou sanskr. manyate [prouver une
agitation mentale], demande un objet ; le sujet est centre en mme temps
quacteur du procs ; il accomplit quelque chose qui saccomplit en lui, na-
tre, dormir, gsir, imaginer, crotre, etc. Il est bien intrieur au procs . . .
(1950/1966, 172).
3.4.1 Ambigut de cette dfinition ; les verbes mdio-passifs

La formulation ci-dessus pose un double problme :
(a) Dans les verbes dtat aussi on pourrait dire que le sujet est le lieu du
procs, mme si lon voit bien quil ne peut en tre lacteur. Les verbes
dtat, dans lesquels lagentivit est neutralise, nont pas de diathse, et
ne constituent en aucun cas une sous-catgorie du verbe moyen. Cette
20
A propos de la notion de verbe moyen en arabe classique, outre M. Cohen (1929),
Joon (1930), Cantineau (1950) et Fleisch (1957 ; 1968, 116117), comparer la descrip-
tion ci-dessus avec P. Larcher (1995, 2956 ; 2003, 2226) ou, pour larabe dialectal gyp-
tien, de C. Audebert (2002).
336 joseph dichy
restriction, qui est implicite dans la dfinition ci-dessus, se dduit gale-

ment de cette remarque :
Il ne sagit donc nullement de faire concider la diffrence de lactif au
moyen avec celle des verbes daction et des verbes dtat (Benveniste
1950/1966, 172).
(b) Si le sujet est acteur du procs dont il est le sige, sil accomplit
quelque chose, cela suppose quil jouit dun certain degr dagentivit.
La proprit dagentivit est incluse de mme dans la notion de diathse
laquelle Benveniste a recours, notamment lorsquil propose en conclu-
sion de larticle cit, de substituer diathse interne (par opposition la
diathse externe) au terme de verbe moyen (par opposition au verbe
actif ou externe). Or le passage cit plus haut donne comme exemples
de ce type de verbes natre, gsir et crotre, qui sont non agentifs.
Pour traiter cette difficult, il convient demprunter au mme article

(mais non sur ce point lanalyse) dE. Benveniste la notion de verbe
mdio-passif, pour dcrire les verbes non agentifs hors verbes dtat
bien entendu comme nam u crotre, grandir, waqaa a f se trouver,
tre situ ou dans, mad passer (temps), mta u mourir, etc.21
3.4.2 Diathse interne vs diathse externe et proprits de transitivit/

intransitivit
Le verbe moyen soppose au verbe actif ou externe :
De cette confrontation [entre deux types de verbes] se dgage assez claire-
ment le principe dune distinction proprement linguistique, portant sur
la relation entre le sujet et le procs. Dans lactif, les verbes dnotent un
procs qui saccomplit partir du sujet et hors de lui.22 Dans le moyen, qui
est la diathse dfinir par opposition, le verbe indique un procs dont le
sujet est le sige ; le sujet est intrieur au procs (Benveniste, 1950/1966,
172). Lactif est une production dacte, rvlant plus clairement encore la
position extrieure du sujet relativement au procs ; et le moyen servira
dfinir le sujet comme intrieur au procs (op. cit. 173).
21
Je fais mienne ici une suggestion de Jean-Pierre Descls, communication orale.
22
Il est intressant de noter titre de comparaison que Sbawayhi considrait que
la transitivit fait entrer un verbe dans le chapitre des actions visibles ou audibles ( f
bb al-aml allat tur wa-tusma Kitb IV : 6). Cette description est reprise dans le
commentaire dal-Srf (en note de la mme page) et dans Ibn Ya, arh al-Mufassa l
VII : 157.
Le verbe actif ou externe est par dfinition un verbe transitif :

Ainsi se constituent partir du moyen des actifs [. . .] qui se caractrisent
toujours par ceci que le sujet, pos hors du procs, le commande dsor-
mais comme acteur, et que le procs, au lieu davoir le sujet pour sige, doit
prendre un objet pour fin (ibid.).
Rappelons que la transitivit doit sentendre, en ce qui concerne les don-
nes traites ici, comme directe ou indirecte (i.e. complment prposi-
tionnel). Exemples de verbes de diathse externe :
h akama ayan matriser qqch. ; aml al fulnin ayan dicter
qqch. qqun ; dafaa a pousser, repousser ou payer ; kataba u
crire ; qla u dire . . .
Les verbes moyens ou de diathse interne, dans la dfinition de Benve-
niste, peuvent tre aussi bien intransitifs que transitifs directs ou indi-
rects. Exemples de verbes du schme simple :
verbes moyens intransitifs :

baqiya a, rester, demeurer lorsque le sujet est humain ; farih a a se
rjouir, prouver de la joie ;
verbes moyens transitifs directs :
malla (dalternance vocalique i/a) tre ennuy par, prouver de len-
nui [ cause] de ; xa i craindre ;
verbes moyens transitifs indirects :
aara u bi-, sentir, ressentir ; raiba a f dsirer qqch, et son contraire
raiba a an dsirer viter ou spargner qqch.
4. Valeurs smantiques de base des schmes simples
Le cadre conceptuel tant pos, considrons maintenant les valeurs

smantiques associes aux trois schmes faula, faala et faila.
4.1 Les deux valeurs smantiques du schme faula
4.1.1 Les faula valeur dtat caractristique

Les verbes de ce schme sont traditionnellement dcrits comme expri-
mant un tat ou une qualit durable ou constante, par opposition aux
338 joseph dichy
verbes en faila, qui dnoteraient un tat temporaire ou instable (cf.

par exemple az-Zamaxar, Mufassa l : 230). A cette rfrence la dure,
jai propos de substituer la notion dtat caractristique, selon la dfini-
tion ci-dessus ( 3.3.3.a). Ces verbes dcrivent effet une qualit ou un
tat pos comme constitutif ou caractristique du sujet grammatical.
Wright dsignait ce dernier comme correspondant une qualit inh-
rente naturelle (a naturally inherent quality 189698, I : 3638),
et Sbawayhi, dj cit, comme les caractres susceptibles daffecter les
choses, mot--mot qui peuvent tre dans les choses (al-xisl allat
taknu f l-ay Kitb IV : 28).
Dans les faula valeur dtat caractristique, le procs tat est, comme
on la vu, dagentivit neutralise. Ces verbes sont, comme il est attendu,
intransitifs.23 Exemples (hors glissement smantique de type tre x o
devenir x) :
qabuh a u tre qabh laid (valeur physique ou morale) ; h asuba u
tre h asb noble, de haute naissance ; h aluma u tre h alm clment ;
sabuh a u tre sabh brillant, radieux (visage) ou beau, joli (garon,
personne) ; katufa u tre katf pais, dense.
Faula est le schme simple le plus stable dun point de vue formel. Nous
avons vu (figures 1 et 2) que la voyelle de sa deuxime consonne radicale
demeurait inchange au paradigme du prfix (mudri). Ce schme est
aussi le plus stable smantiquement. Sa valeur de base, ltat caractristique,
est toutefois lobjet dun glissement de sens qui lui est troitement cor-
rl, le changement et lacquisition dtat ( 3.3.3.b et 3.3.4 ci-dessus).
4.1.2 Les faula valeur dacquisition dtat caractristique

Ce glissement smantique, trs frquent, est illustr par lexemple :
ayra anna hdihi l-jama . . . d. a'ufa h udruh l-alaniyy . . . mundu
an . . . Mais la prsence publique de ce rassemblement . . . sest affai-
blie (mot--mot, est faible) . . . depuis que . . . (Majallat an-Nahr, 2
juillet 2006, 13, col. 1).
23
Cf. Sbawayhi : Il nexiste pas dans le langage des Arabes de verbe transitif en
faula (laysa f l-kalm faultuhu mutaaddiyan Kitb IV : 38).
Des exemples similaires ont t vus aux 3.3.4. a et b. Le contexte indi-

que clairement un dbut du procs (depuis que . . .) : le verbe daufa u
(adjectif dtat caractristique : daf faible) ne rfre donc pas un tat.
La forme conjugue au md exprime ici un vnement, qui concide
avec lachvement du processus dcrit par le verbe (laffaiblissement) :
ce processus a atteint son dernier tat, ce qui dtermine un avant et
un aprs. Cet aprs correspond un tat rsultatif que lon peut dcrire
comme un tat acquis : cette prsence publique est devenue faible ou
est dsormais faible.
On observe, travers les usages illustrs par cet exemple et ceux qui
ont t analyss plus haut, que la valeur smantique tre x incluse dans
celle de ltat caractristique se trouve trs souvent ralise en contexte
comme un changement dtat (devenir x). Ce changement a, son tour,
la proprit dengendrer un tat rsultatif. Un tel phnomne nest nulle-
ment propre larabe. On peut comparer, au moins partiellement, avec
les exemples suivants, qui sont emprunts au franais :
Ah ! Maintenant, tu es beau ! (dit une maman son fils, qui revenait en

tranant les pieds de chez le coiffeur).
Lucie est dsormais notre anctre.
Dans les deux cas, les prdicats tre beau et tre notre anctre paraissent
incompatibles avec un contexte qui leur assignerait un commencement
ou un terme. Or les marqueurs temporels maintenant et dsormais
prsents dans ces exemples assignent aux procs un dbut, ce qui a pour
effet doprer une translation des prdicats verbaux ci-dessus de la cat-
gorie de ltat vers celle du changement et de lacquisition dtat. Cest
comme si lon disait : tu es devenu beau (maintenant que tu as les che-
veux coups) et Lucie est devenue lanctre de lhumanit (depuis sa
dcouverte et les recherches qui ont suivi).
Le dictionnaire de Hans Wehr-Cowan (1979) traite le plus souvent

mais pas systmatiquement le glissement smantique de ltat caract-
ristique vers lacquisition dtat comme relevant des proprits lexicales
des verbes concerns, exemples :
(a) glissement smantique trait comme inclus dans le sens du verbe :

daufa u to be or become weak ; katura u to be . . . numerous suivi
de : to increase, augment ; saura u to be or become small (1er sens
de ce verbe) ;
340 joseph dichy
(b) glissement smantique non pris en compte dans le sens du verbe :

h asuna u to be handsome, beautiful ; qabuh a u to be ugly ; zarufu
u to be charming (. . .) witty ; saura u (2e sens de ce verbe) : to be
young ; qaduma u to be old, ancient.
Si ce dictionnaire semble hsiter, cest que la question de savoir si ce

glissement smantique est assez frquent pour tre inclus dans le sens
lexical du verbe nest pas toujours facile trancher. Les traductions pro-
poses dans Wehr-Cowan proviennent vraisemblablement des usages
attests dans les fiches des auteurs. Ces donnes montrent en tout tat de
cause que, pour un verbe donn du schme faula, les deux cas prsents
plus haut ( 3.3.4.b) se rencontrent. Voyons de quelle manire :
4.1.2a Le glissement smantique tre x o devenir x

Dans les verbes en faula dont le sens est fondamentalement celui de
ltat caractristique, comme dans h asuba u tre h asb noble, de haute
naissance, un contexte dacquisition de cet tat est presque toujours pos-
sible. Cette valeur doit alors tre considre comme smantiquement
seconde, quelle soit inscrite dans le lexique ou quelle soit limite un
nonc donn. Le glissement smantique observ correspond alors au
schma : tre x o devenir x, exemples :
h asuba l-na jruna notre voisin maintenant est devenu un noble

(nonc ironique) ;
karuma r-rajulu lhomme sest montr noble-et-gnreux.
Dans ces deux noncs, la prsence dun adjectif dtat caractristique

associ par les structures morpho-lexicales de la langue aux verbes en
faula suggre une valeur attributive : notre voisin a acquis le titre de
h asb, i.e. de noble de haute naissance (do lironie) ; lhomme a mrit
le nom de karm, noble-et-gnreux.
4.1.2.b Lengendrement de ltat acquis par lvnement ou le

processus du changement dtat (devenir x tre x)
Dans les faula dont le sens de base est celui de lacquisition dun tat,
comme dans badua u devenir incomparable, sans pareil (al-Mujam
al-Wast et al-Munjid, racine /b-d-/), le contexte associe au verbe un
tat acquis, qui est engendr, soit au terme du processus de changement
dtat (ltat acquis correspond alors au dernier tat du processus), soit
par un vnement. Exemple :
badua ibn sna bi-ilmihi xayran wa-arran Avicenne fut incompara-

ble par sa science, en bien comme en mal. Du vivant du philosophe-
mdecin, ltat acquis et pris une valeur dtat-prsent dont la dure
inclut le moment de lnonciation (Descls, 1994, 73), la traduction
tant alors Avicenne sest rvl . . . ou se rvle . . .
Comme dans les exemples du paragraphe prcdent, le verbe en faula

peut tre interprt comme prenant une valeur attributive dont le pr-
dicat est ladjectif dtat caractristique correspondant : Avicenne mrita
le nom de bad, incomparable et admirable. Ces exemples ralisent le
schma : devenir x tre x (dans lequel se lit engendre).
4.1.3 Schme faula et adjectifs assimils et/ou participes

Si lon carte le glissement smantique tre x o devenir x, le sens
des verbes valeur dtat caractristique exclut les participes actif (ism
al-fil) et passif (ism al-maf l). Les valeurs aspectuo-temporelles de
base des participes sont en fonction du contexte celles de lachve-
ment (qui peut engendrer un tat rsultatif ) ou du progressif (dont le
terme peut, ou non, tre spcifi).24 Ces valeurs ne sont pas compatibles
avec celles de ltat caractristique, dont la dure nest pas borne et qui
ne comporte aucune progression. En outre, le participe passif qui ne
peut, quant lui, tre driv que dun verbe transitif, est donc en quel-
que sorte doublement exclu.
Les adjectifs assimils ou analogues [au participe] (sifa muabbaha)
qui sont associs aux verbes en faula ainsi quaux verbes dtat caract-
ristique ou non relevant dautres schmes compensent en quelque sorte
labsence de participes. Comparer par exemple les phrases nominales :
hum kuram ils sont gnreux ; dans cette phrase, karm, plur.
kuram, est ladjectif assimil associ au verbe dtat caractristique
karuma u ;
(a) hum tlibn ils sont en train de demander/ils ont demand
ou (b) hum matlbn ils sont demands ; tlib et matlb sont les
24
Pour les deux valeurs de base des participes (dachvement et progressive), com-
parer avec Roman (1990, 3940). Pour une analyse dveloppe de la valeur smantique
des participes en contexte et notamment du progressif terme spcifi ou non, voir
Dichy (2002/2003, 5.2.2). En ce qui concerne la relation morphologique entre certains
schmes et ladjectif assimil ou le participe actif, comparer la discussion mene ici avec
Larcher (2003, 2526).
342 joseph dichy
participes actif et passif de talaba u demander, qui ne relve pas des

verbes dtat.
Cette division entre dune part les verbes dcrivant un tat et dots dun
adjectif assimil et de lautre les verbes auxquels sont associs des par-
ticipes, rencontre toutefois un problme. Au chapitre consacr ladjec-
tif assimil (sifa muabbaha), az-Zamaxar (Mufassa l : 230) associe
celui-ci la valeur smantique dimmutabilit man t-tubt (qui renvoie
ce qui est tbit immuable, tabli en quelque sorte hors nonciation du
temps). Il oppose cette dernire la valeur doccurrence dans le temps
man l-h udt, valeur quil associe aux participes. Or il nillustre pas
ces derniers par des verbes dcrivant des vnements ou des proces-
sus (comme je lai fait ci-dessus avec talaba u demander), mais par des
verbes associs un adjectif dtat caractristique. Le premier exemple
donn est celui de h sin, qui correspond un participe actif de h asuna u
tre beau (adjectif dtat caractristique h asan). Az-Zamaxar souligne
la valeur doccurrence dans le temps man h udt de la forme h sin en
insrant celle-ci dans les phrases :
huwa h sin al-na ou adan, mot--mot : il est en train de devenir/il

est devenu beau maintenant, ou il le deviendra demain.
Les exemples suivants sont : krim, participe actif de karuma u tre
noble et gnreux, til, participe actif de tla u tre ou devenir long
et diq, participe actif du verbe dtat caractristique dqa i tre ou
devenir troit (adjectif dtat caractristique dayyiq), pour lequel
az-Zamaxar cite le verset : wa-diqun bihi sadruka, alors que tu
es angoiss par cela, mot--mot, ta poitrine tant oppresse (rendue
troite) par cela (Coran 11 : 12).
Ibn Ya commente, dans ce verset, le choix du participe actif diq au

lieu de ladjectif dayyiq en indiquant que le premier rfre une troi-
tesse (i.e. une oppression) [correspondant ] un accident [se produi-
sant] dans le prsent et non un [caractre] immuable (dq rid f l-h l
ayr tbit arh al-Mufassa l VI : 83).
Si les verbes associs des adjectifs dtat caractristique peuvent tre
compatibles avec le sens des participes actifs, cest en raison :
(a) soit du glissement de sens qui permet ces verbes, lorsque leur sens
premier est celui de verbes dtat, de prendre celui de lacquisition dtat
(glissement tre x o devenir x),
(b) soit du fait que ces verbes rfrent, fondamentalement, lacquisition

dun tat caractristique, cest--dire, un processus ou un vnement
engendrant un tat rsultatif (schma devenir x tre x).
Les quatre exemples cits par az-Zamaxar relvent de ces deux cas.
Si h asuna tre beau relve de la premire catgorie, on a vu qual-Mu-
barrid, cit au 3.3.3.b, considrait karuma comme ayant une valeur
dacquisition dtat, et que le dictionnaire de Hans Wehr-Cowan oscillait
entre la valeur de base attribuer chaque verbe. La valeur attributive
se distribue, comme on la vu, sur les verbes relevant tant de (a) que de
(b). Il apparat clairement, en consquence, que lon ne peut rduire les
valeurs smantiques du schme faula au seul sens de verbe dtat : il sy
ajoute lacquisition dtat.
4.1.4 Deux consquences de lagentivit neutralise ou de la non

agentivit des faula
Lagentivit neutralise des verbes valeur dtat caractristique, et la
non agentivit des verbes dnotant lacquisition dun tat caractristi-
que entranent galement limpossibilit pour ceux-ci de limpratif :
ce dernier nest possible que si un certain degr de contrle exerc par
lagent sur le procs est envisageable (cest pourquoi les verbes pouvoir
et devoir nont pas dimpratif en franais). En outre, lagentivit neutra-
lise est incompatible avec la transitivit. Il sensuit que la conjugaison
des verbes correspondants exclut toutes les formes du passif (ce dernier
ntant possible quavec des verbes transitifs, dans le cadre de contraintes
smantiques prcises).25
Le schme faula est relativement peu frquent dans les textes, notamment
en arabe littraire moderne. On lui prfre sans doute les constructions
adjectivales correspondantes. Cela ne signifie nullement que ce schme
soit sorti de lusage ou en voie de disparition, comme le montre lexemple
de daufa ci-dessus, emprunt la presse actuelle. On trouve galement
assez souvent dans la prose contemporaine des emplois de construction
impersonnelle, la faveur, pour ainsi dire, de lagentivit neutralise :
l yah sunu an . . ., il ne convient pas que . . . [Yawm., 79, 18] ; yajduru,
il convient, suivi de an ou dune forme infinitive . . .
25
Les critres qui permettent, interdisent ou restreignent lapparition de limpratif
ou du passif en arabe ont t prsents dans Ammar et Dichy (1999, 1920).
344 joseph dichy
4.2 Les valeurs smantiques de base du schme faila

Le deuxime schme du verbe simple est dalternance vocalique i/a.
Comme on la vu au 2.2.1, une variante en i/i porte sur un petit nombre
de verbes : sur 15 ou 16 verbes inventoris par les traits de morphologie
arabe mdivale, huit sont attests en arabe moderne par le dictionnaire
de Hans Wehr (cf. liste dans Ammar et Dichy 1999, 24). Cette variante,
qui affecte essentiellement des verbes de 1re consonne radicale w ou y est
purement formelle, et nentrane pas de diffrence de valeur smantique
avec le schme faila dont elle relve. Exemples :
warita i hriter ; waliya i suivre ou succder .
On rencontre dans les verbes du schme faila tous les degrs dagenti-
vit (pleine ou partielle) ou labsence de cette proprit (non agentivit,
agentivit neutralise), ainsi que les diffrentes formes de transitivit ou
de son contraire. Do la partition des sens associables ce schme qui
a t observe par les auteurs qui se sont attachs le caractriser.26 La
valeur la plus gnrale est celle de verbes moyens ou de diathse interne
(au sens revisit prsent plus haut), ce schme comportant en outre un
nombre limit de verbes associs un adjectif dtat caractristique.
Il y a quatre grandes catgories smantiques de verbes en faila, dis-
tingues principalement au moyen des proprits dagentivit et de tran-
sitivit.
4.2.1 Les faila associs un adjectif dtat caractristique (verbes dtat

ou dacquisition dtat)
Un certain nombre de verbes dalternance vocalique i/a (schme faila)
relve des verbes associs un adjectif dtat caractristique. Ces derniers
incluent tous les traits prsents plus haut propos du schme faula, et
comportent comme eux les deux grandes valeurs smantiques troite-
ment corrles tre x et devenir x. Les verbes en faila relevant de cette
sous-catgorie sont intransitifs, dagentivit neutralise (valeur tre x)
ou non agentifs (mdio-passifs, de valeur devenir x), et ne peuvent avoir
de participe actif que dans les conditions, rarement ralises, prsentes
plus haut ( 4.1.3). Ils dcrivent pour la plupart un tat caractristique
qui correspond :
26
Outre M. Cohen (1929) et Joon (1930), voir notamment : Fleisch (1979, 220
suiv.) ; Ammar et Dichy (1999, 2324).
soit une particularit physique ou psychique (dfaut ou qualit)27

affectant un sujet humain ou un animal, domestique ou appartenant
un troupeau,
soit une particularit, souvent mais pas ncessairement connote
ngativement, affectant une entit concrte ou abstraite.
Ladjectif assimil est le plus souvent du schme af al, fm. fal.

Une srie limite de verbes de cette sous-catgorie relve de racines 2e
radicale w ou y. Jen ai dnombr 71, dont 18 de 2e consonne y dans Ilys
et Nsf (1995). Ces verbes ont pour particularit de ne pas appliquer
les rgles de transformation phonologiques habituelles.28 Exemples :
(a) Verbes adjectif dtat caractristique dcrivant une qualit ou un

dfaut le plus souvent physique (humain ou animal) :
jayida yajyadu avoir le cou long (signe de beaut) adj. ajyad, fm.
jayd ; awiza yawazu tre ou tomber dans le besoin (adj. awaz) ;
layisa yalyasu tre ou se montrer courageux, tenir sa position (dans
un combat) adj. alyas ; rawia yawau tre ou se montrer admira-
ble (humain) pour son intelligence, sa beaut, son courage, etc. (adj.
arwa) ; awira yawaru tre ou devenir borgne (adj. awar) ; qawida
a avoir le dos et le cou allongs (cheval . . .).
(b) Verbes adjectif dtat caractristique dcrivant une particularit
affectant une entit non anime concrte ou abstraite :
awisa yawasu tre ou devenir difficile, incomprhensible, abscons
(affaire, explication) adj. awas et aws ; awija yawaju tre ou
devenir tordu, incurv, tortueux (adj. awaj, fm. awj) ; ayisa
yayasu tre ou devenir doux, souple (adj. ayas) ; qawira yaqwaru
tre vaste (maison) adj. aqwar.
Un nombre galement limit, mais plus important, de verbes en faila

associs un adjectif dtat caractristique relve de racines sans trans-
formation, exemples :
27
Voir notamment : Sbawayhi, Kitb IV : 17 ; az-Zamaxar, Mufassa l : 278. La syn-
thse ci-dessus provient de mon propre examen des donnes.
28
Rappelons que dans xfa, yaxfu craindre, dalternance vocalique i/a, la 2e
consonne radicale w est lobjet de transformations en fonction de la voyelle qui la suit
et la prcde immdiatement, la diffrence des verbes de 2e radicale w ou y dont on
trouve ici des exemples.
346 joseph dichy
(c) Verbes dcrivant une qualit ou un dfaut chez un humain ou un

animal :
baliha a devenir faible desprit ou tre stupide, idiot (adj. ablah, fm.
balh) ; ramida a devenir, puis tre chassieux (il) ou (mot--mot)
devenir ayant, puis avoir lil chassieux (humain) adj. armad ;
darida a perdre ses dents, devenir ou tre dent (adj. adrad) ;
sawifa a (mot--mot) devenir ayant, puis avoir beaucoup de laine,
tre laineux (ovin ou caprin) adj. aswaf.
(d) Verbes dcrivant une particularit affectant une entit concrte ou
abstraite :
xadira a verdir, puis tre vert (adj. axdar vert) ; sawida a noir-
cir, puis tre noir (adj. aswad noir) ; xaliqa a suser, puis tre us
(vtement) adj. axlaq ; sawisa a se gter, tre rong par des vers
(viande, nourriture) adj. aswas.
(e) Verbes de cette sous-catgorie dont ladjectif nest pas de schme
af al :
yasira yaysaru devenir puis tre facile (adj. yasir et yasr).
La valeur smantique de base des verbes prsents en (c), (d) et (e) est
celle de lacquisition dtat, la diffrence des verbes des sous-catgories
(a) et (b). Dans leur sens de base, les premiers sont donc non agentifs, et
correspondent des verbes mdio-passifs.
4.2.2 Autres verbes mdio-passifs et non agentifs en faila

Dautres verbes du mme schme sont non agentifs et intransitifs. Ils
correspondent, non des verbes diathse interne ou moyens, comme
dans les deux paragraphes suivants, mais des verbes mdio-passifs.
Exemples :
verbes de sujet grammatical non-humain :

rasila a tre longue et pendre (chevelure) ; rawiya a tre irrigu,
arros (terre) ; xadiba a verdir (arbre, jardin, terre) ;
verbes de sujet grammatical humain :
ariqa a se noyer (au sens propre), mourir noy adj. assimil
arq ; marida a, tre ou tomber malade, adjectif assimil : mard.
Tous les verbes de cette catgorie ne sont pas associs, dans les diction-
naires, un adjectif assimil. Dans le cas des deux derniers exemples
ci-dessus, il est noter que le participe actif est attest, mais avec une
diffrence de sens du plus grand intrt pour la distinction entre les
valeurs smantiques du participe actif et celles de ladjectif assimil (ci-
dessus, 4.1.3) :
Le Munjid al-lua l-arabiyya l-musira (2000) donne lexemple de
anqaztu riqan jai sauv un noy (racine /-r-q/), et Roman : l
takul hd t-tama fa-innaka mridun in akaltahu Ne mange pas
de cette nourriture, car tu seras malade si tu en manges (1999, 71,
daprs le Lisn al-arab, XIIIe s., racine /m-r-d/).
Comparons ces emplois. Les deux adjectifs assimils arq et mard

correspondent des tats rsultatifs engendrs par lvnement ou con-
cidant avec la dernire tape du processus dcrit par le verbe (Descls
1994) : le processus se noyer a pour dernire tape ( !) tre mort noy,
dcrit par ladjectif arq, de mme que tomber malade entrane ltat
dcrit par ladj. mard. Par contraste, les participes actifs correspon-
dants riq et mrid ont une valeur de progressif terme spcifi (Dichy
2002/2003, 34, 36, 60) :
le procs de noyade na pas atteint son terme dans lexemple du Mun-

jid (le sauvetage a pu avoir lieu) ;
le terme (ou lachvement) du procs tomber malade est projet dans
le futur dans lexemple du Lisn cit par A. Roman. Ce dictionnaire
glose, significativement mridun la suite de la phrase ci-dessus par
ay tamradu cest--dire : tu seras malade.
Il est significatif ce quatteste notamment lexemple du Munjid al-lua

l-arabiyya l-musira que ces drivations restent vivantes en arabe
contemporain, mme si leur frquence est faible.
4.2.3 Les verbes de schme faila diathse interne (ou moyens) et

dagentivit partielle
Un nombre beaucoup plus important de verbes du schme faila corres-
pond des verbes diathse interne, ou verbes moyens : le procs affecte
le sujet, que le verbe soit transitif ou non (ainsi : rire de qqun ou rire).
Ces verbes sont transitifs indirects ou intransitifs et dagentivit partielle
348 joseph dichy
(lagent, ntant pas autonome par rapport au procs, ne conserve quun

contrle limit sur celui-ci). Ils peuvent soit admettre un participe actif,
mais non un adjectif assimil, soit tre associs dans le lexique un
adjectif assimil, et nadmettre de participe actif que dans les construc-
tions vues au paragraphe prcdent et en 4.1.3. Exemples :
(a) Verbes admettant un participe actif, mais non un adjectif assimil :

verbes intransitifs :
baqiya a rester, demeurer qq part (agent humain) [Yawm., 12, 12] ;
nma a dormir ; dah ika a rire (en emploi absolu, diffrent de : rire
de . . ., se moquer de . . .) ;
verbes transitifs directs ou indirects :
dah ika a min ou al rire de, se moquer de qqun ; nasiya a oublier
qqch ; samia a entendre qqch ; saima a sennuyer de . . . (qqch)
[Yawm., 50, 10] ; h asiba a anna . . . penser, croire, estimer que . . . ;
alima a anna . . . savoir que ou apprendre que . . . alima a bi-, avoir
connaissance de . . . .
(b) Verbes associs dans le lexique un adjectif assimil, mais nadmet-
tant de participe actif que dans les constructions lies loccurrence dun
procs (h udt) 4.1.3 :
verbes intransitifs :
atia a avoir soif (adj. atn) ; taiba a prouver de la fatigue, tre
fatigu (adj. taib) ; xafira a prouver un fort sentiment de timidit,
tre trs timide, notamment pour une jeune fille (adj. xafir);29 salima
a tre sain et sauf , chapper ( un mal ou un danger) adj. salm,
mais galement slim ;30
verbes transitifs directs ou indirects :
h afiya a bi- accueillir trs chaleureusement qqun (adj. h afiyy) ; adiba
a al tre, se mettre en colre contre qqun (adj. adib).
Dans les exemples que lon vient de voir, la catgorie morphologique

de ladjectif assimil ne correspond pas un tat caractristique, mais
29
Il existe un second adjectif assimil, mixfar grande timide : associ ce dernier,
xafirat tre une grande timide relve des verbes valeur dtat caractristique, dans
lesquels lagentivit est neutralise.
30
Dans des cas tels que slim ladjectif assimil est dit, dans la grammaire arabe tra-
ditionnelle, prendre la forme (sa) du participe actif. Autres exemples : yis dsespr,
bis misrable, mhir adroit, habile (artisan) . . .
plutt ltat rsultatif produit pas le procs (vnement ou processus) :

le verbe salima produit ainsi salm sain et sauf , mais galement slim
min . . . sauf de . . . (mal ou danger). Dans un verbe exprimant un pro-
cessus, comme dans adiba se mettre en colre, ladjectif adib dcrit
le dernier tat de ce processus ; si adiba se trouve, dans un contexte
donn, dcrire un vnement, ladjectif dcrit ltat rsultatif engendr
par cet vnement.
Ces verbes permettent de prciser la notion dagentivit partielle. Dans

les exemples ci-dessus, le contrle limit du procs par lagent peut sur-
tout tre glos ngativement :31
refuser de rester quelque part, voire dtre sain et sauf dun danger
(et agir pour sortir de ce lieu ou de cette situation) ; combattre ou sur-
monter un sentiment ou une motion (rire, ennui, timidit, chaleur
humaine ou colre envers quelquun), une sensation (sommeil, fait
dentendre, soif , fatigue), un mouvement de la pense ou proces-
sus cognitif (croire, oublier, savoir), etc.
On notera enfin que la relation entre le verbe et, soit un participe actif,
soit un adjectif assimil, est lexicalise (et non compositionnelle, comme
elle ltait sans doute aux plus anciennes poques de la langue arabe) : on
doit recourir aux connaissances lexicales pour la construire, et non aux
seules connaissances grammaticales ou une dduction partir du sens
du verbe.
4.2.4 Les verbes du schme faila diathse interne, transitifs et

dagentivit entire
Lautre grande partie des verbes du schme faila est constitue de verbes
galement moyens (ou diathse interne), dagentivit entire et de rgime
syntaxique transitif direct ou indirect. Dans cette configuration, le pro-
cs comporte un agent autonome contrlant effectivement (du fait de
lagentivit entire) lvnement ou le processus dcrit par le verbe et un
patient (du fait de la transitivit). Il englobe en outre dune manire ou
dune autre lagent ou se droule au moins partiellement en lui (diathse
31
Cf. Roman (1990, 4243), qui dcrit lagentivit partielle comme une agentivit
ractive.
350 joseph dichy
interne). Ces verbes ont un participe actif, et nadmettent pas dadjectif

assimil. Exemples :
ariba a, boire ; radiya a bi- tre satisfait de . . . ; lah iqa, suivre, pour-
suivre qqun, lah iqa bi-, suivre qqun au sens de le rejoindre [Yawm.,
66, 21] ; rakiba a, monter, chevaucher ; watiqa i bi-, faire confiance
(qqun).
4.3 Les valeurs smantiques de faala et leur dispersion

Le troisime schme de la forme simple du verbe est faala. Il comporte
comme on la vu une alternance vocalique de base ( faala i) et deux
variantes rsultant de divers conditionnements phontiques ( faala u et
faala a).
On pourrait poser que ce schme comprend majoritairement des ver-
bes relevant de la catgorie des verbes actifs ou externes qui soppose
structurellement celle des verbes moyens, au sens donn ces termes
dans larticle de Benveniste cit plus haut (section 3.4). Ce trait serait
alors ce qui distingue faala des schmes faila et faula.
Or il existe un nombre trs lev de verbes ne rpondant pas cette
dfinition de principe. Nous avons vu que ce schme connaissait une
importante dispersion sur le plan formel : lordonnance systmatique du
tableau des schmes de base du verbe simple (fig. 1) est principalement
rompue dans le tableau des schmes du verbe simple et de leurs sous-ca-
tgories (fig. 2) par les trois alternances vocaliques possibles du schme
faala. Les exemples ci-dessous montrent en outre que cette dispersion
se vrifie galement du point de vue smantique.
Considrons dabord les catgories manant des proprits dagentivit

et de transitivit :
Verbes pleinement agentifs et transitifs (directs ou indirects) :

amana u bi- avoir foi en ou dans ; qla u dire ; saala a interroger,
questionner ; sh a i bi- crier vers, en direction de, contre qqun, le
hler ; da u il revenir .
Verbes pleinement agentifs et transitifs indirects, auxquels est associ un
adjectif exprimant un tat rsultatif, ou tat acquis :
latafa u bi- se montrer gentil, prvenant envers qqun (adjectif assi-
mil, exprimant un tat rsultatif : latf cf. Coran 42 : 19) ; daxala u
f entrer dans un clan ou une tribu, se placer sous sa protection (adj.
exprimant un tat rsultatif : daxl) ;
Verbes pleinement agentifs et intransitifs :

makata u rester, demeurer (sujet humain) ; jalasa i sasseoir ; haraa
a se prcipiter (tte baisse) ; sakata u se taire (aprs avoir parl) ;
samata u se taire, garder le silence.
Verbes dagentivit partielle transitifs (directs ou indirects) :
ra a voir ; fza bi- gagner, obtenir qqch ; aara u bi- sentir, res-
sentir qqch.
Verbes dagentivit partielle intransitifs :
zahara a apparatre (intentionnellement) lorsque lagent est humain32
[Yawm., 32, 22] ; warada i, apparatre intentionnellement, arriver
(avec un agent humain).
Verbes non agentifs intransitifs :
mad i passer, se passer (laps de temps, vnements ou processus
dont on prend en compte la dure) [Yawm., 32, 16] ; bad u paratre
(avec un sujet non humain) [Yawm., 32, 17] ; h na i se produire acci-
dentellement [Yawm., 61, 15] ; araba u se coucher (soleil) ; mta u
mourir.
Verbes non agentifs transitifs (directs ou indirects) :
waqa a al tomber sur [Yawm., 32, 17] ; h aw i contenir, inclure,
comprendre.
Verbes dagentivit neutralise ou non-agentifs et intransitifs, associs
un adjectif assimil dcrivant un tat caractristique (verbes dacquisi-
tion dtat) :
dqa i tre ou devenir troit (adjectif dtat caractristique : dayyiq) ;
jda u tre ou devenir excellent (adj. jayyid) ; xaffa i tre ou devenir
lger (adj. xaff ) ; adda u tre ou devenir intense, dur ou violent
(adj. add) ; samala u devenir ou tre rigide ou sec (adj. smil et
saml).
Reprenons cette analyse en classant les verbes de schme faala selon les
grandes catgories prsentes plus haut. On a :
32
Lagentivit est susceptible de varier selon que le sujet grammatical rfre un
humain (qil) agissant intentionnellement ou non, ou un non humain (ayr qil).
Avec un sujet grammatical non humain, ou humain, mais dpourvu, dans le contexte,
dintentionnalit, zahara ou warada sont des verbes non agentifs. Un autre exemple (du
schme faila cette fois) est dans Yawm. (16, 15) : fa zafira n-nawmu bi-jufn, le som-
meil triompha de mes paupires. Dans cette phrase, le verbe zafira perd le trait dagenti-
vit quil aurait avec un sujet grammatical humain, et devient non agentif (le sommeil ne
disposant daucun contrle sur le procs, dont il nest dailleurs pas lagent). En revanche,
waqaa tomber demeure non agentif, que le sujet soit humain [Yawm. 16, 12] ou non
humain [Yawm. 16, 20 et 32, 17].
352 joseph dichy
(a) Verbes valeur dtat caractristique ou dacquisition dtat caractris-

tique repris en partie des exemples ci-dessus :
dqa i tre ou devenir troit (adjectif dtat caractristique : dayyiq) ;
lna i tre ou devenir souple (sens physique), ou tendre (sens psy-
chologique affectant un sujet humain) (adj. layyin) ; jda u tre
ou devenir excellent (adj. jayyid) ; sda u tre ou devenir le chef
(adj. sayyid) ; xaffa i tre ou devenir lger (adj. xaff ) ; qalla i tre
ou devenir peu nombreux (adj. qall) ; adda i et u tre ou devenir
intense, dur ou violent (adj. add) ; samala u devenir ou tre rigide
ou sec (adj. smil et saml) ; fasada u et i se corrompre, pourrir,
i.e. pour une viande, de la nourriture ou les murs dune personne,
etc. : devenir, puis tre corrompu (adj. fsid, fasd).
(b) Verbes diathse interne ou moyens repris des exemples ci-dessus :
Verbes associs dans le lexique un adjectif assimil exprimant
un tat rsultatif :
latafa u bi- se montrer gentil, prvenant envers qqun (adj. latf ) ;
daxala u f entrer dans un clan ou une tribu, se placer sous sa
protection (adj. daxl) ;
Verbes sans adjectif assimil :
amana u bi- avoir foi en ou dans ; makata u rester, demeurer
(agent humain) ; jalasa i sasseoir ; haraa a se prcipiter ; ra a
voir ; da u il revenir ; fza bi-, gagner, obtenir qqch ; zahara
a apparatre, se manifester intentionnellement (agent humain)
[Yawm., 32, 22] ; warada i apparatre intentionnellement, arri-
ver (agent humain).
(c) Verbes mdio-passifs repris des exemples ci-dessus :
mad i passer, se passer (laps de temps, vnement ou processus
dont on prend en compte la dure) [Yawm., 32, 16] ; bad u para-
tre (avec un sujet non humain) [Yawm., 32, 17] ; h na i se produire
accidentellement (vnement) [Yawm., 61, 15] ; araba u se cou-
cher (soleil) ; mta u mourir ; saqata a tomber, chuter ; h aw i
contenir, inclure, comprendre.
(d) Verbes diathse externe (transitifs) :
qatala u tuer ; daraba i frapper ; h asaba i compter (des objets) ;
madda u tendre, tendre ; sah ara a charmer, ensorceler ; sajana u
emprisonner ; xadaa a tromper, duper, leurrer ; fasala i sparer,
diviser ; qataa a couper ; fatah a a ouvrir ; ba i vendre ; qla u
dire ; saala a interroger, questionner ; talaba u min an . . . de-
mander qqun de . . . ; sh a i bi- crier vers, en direction de qqun, le

hler [Yawm., 61, 11].33
4.3.1 Les corrlats formels de la prsence de verbes adjectif dtat

caractristique dans le schme faala
Les verbes adjectif dtat caractristique (tat caractristique ou acquis)
du schme faala, dont on aurait attendu la prsence parmi les faula ,
sont associs des corrlats formels. Ils relvent :
soit des verbes concaves (ajwaf, i.e. de 2e consonne radicale w ou y),

pour lesquelles la ralisation de lalternance vocalique u/u ne semble
atteste que dans le cas de tla/yatlu adj. tawl (bien que lon
puisse galement voir dans cette forme un verbe en a/u) ;
soit des verbes redoubls (mudaaf, de 2e et 3e consonnes radicales
identiques), pour lesquels il nexiste quun tout petit nombre de ver-
bes en faula, qui sont signals comme des exceptions par Sbawayhi.34
Deux verbes redoubls en u/u sont attests en arabe contemporain
(Hans Wehr-Cowan 1979) : arra (1e pers. arurtu) u se monter ou tre
mauvais, malfaisant (adj. arr) et labba (1e pers. labubtu) u se mon-
trer ou tre intelligent (adj. labb autre alternance vocalique : i/a).
Ces deux verbes sont en outre donns par le dictionnaire dal-Fayrz
bd, dans la glose de h abba (1e pers. h abubtu) u il devenir un
proche [qui reoit des marques damiti] de qqun (adj. h abb), comme
tant, avec ce verbe, les seuls membres de cette classe formelle.35
33
Les verbes de parole du schme simple sont en arabe des verbes diathse externe.
Le passage une diathse interne se fait au moyen de schmes augments incluant ce que
Roman (1990 ; 1999/2005) appelle le morphme-cho t voir Ammar et Dichy (1999,
2728) ; Dichy (2002/2003, 4.2). Exemples : sala a interroger [schme simple, verbe
diathse externe] vs sala an interroger, questionner (qqun) sur, puis tasala an sin-
terroger sur [schme augment incluant le morphme-cho, verbe diathse interne] ;
qla u dire [schme simple, verbe de diathse externe] vs qawwala attribuer (des paro-
les) qqun, puis taqawwala prtendre, allguer (au profit, dans lintrt, en faveur de
soi-mme) [schme augment incluant le morphme-cho, verbe de diathse interne].
34
Sbawayhi indique quil ny a presque pas de verbes en faula de 2e et troisime radi-
cale identiques (ou redoubls, mina t-tadf Kitb IV : 3637). Ibn Xlawayh (Laysa . . . :
27), signale labsence de verbes redoubls en faula, lexception des deux cas de labba u
(dj cit par Sbawayhi, loc. cit.) et azza u avoir peu de lait (dit dune chvre).
35
Malgr lindication cite, on trouve dans le dictionnaire dal-Fayrz bd trois
autres verbes : azza (3e pers. fm. sing. azuzat) u voir (pour une chvre) son lait dimi-
nuer, avoir peu de lait (adj. azz), ainsi que damma (1e pers. damumtu) u devenir ou
tre trs laid, repoussant (adj. damm), et fakka (1e pers. fakuktu) u, se montrer ou tre
354 joseph dichy
Toutefois, il ne sagit pas dune rgle oprant dans les deux sens : tous les
verbes concaves ou redoubls de schme faala ne relvent pas, il sen
faut de beaucoup des verbes adjectif dtat caractristique. Les exem-
ples de samala u devenir ou tre rigide ou sec et de fasada i et u voir
4.3, exemples (a) montrent quil existe galement un certain nom-
bre de faala adjectif dtat caractristique qui ne relvent daucune des
deux catgories formelles que lon vient dindiquer. Leur sens de base est
toutefois celui de lacquisition dun tat caractristique plutt que celui
de verbes dtat.
4.3.2 Un glissement smantique partiellement responsable de la

prsence de verbes diathse interne et de verbes mdio-passifs
dans le schme faala
Les verbes diathse externe sont agentifs et transitifs : le procs ver-
bal affecte un objet externe lagent. Un glissement smantique trs
rpandu et observable dans plusieurs langues peut toutefois modifier la
diathse de ces verbes, qui devient alors interne. Deux cas se prsentent,
selon que la valeur smantique rsultant de ce glissement est agentive ou
non. Dans le premier, le sens produit est celui dun verbe moyen ou a
diathse interne : le verbe est agentif et le procs affecte lagent. Dans le
second, le sens produit par le glissement est mdio-passif : le procs, qui
est non agentif, affecte le sujet grammatical du verbe (lagent existe, mais
il nest pas dsign). Comparons les donnes de plusieurs langues :
Verbes agentifs diathse interne, partir de verbes diathse

externe :
En franais, le sens de base du verbe transitif plonger est celui dun
procs diathse externe dans lequel un agent plonge un objet
dans un liquide, ex. : Pierre plongea sa chemise dans leau de la les-
sive. Ce verbe a pour pendant une forme pronominale exprimant
une diathse interne, se plonger, dont le sens est mtaphorique,
ex. : se plonger dans un livre (mais non *dans la piscine). En arabe,
les sens de plonger et se plonger sont associs au verbe atasa i.
Joon donne un exemple similaire en hbreu : taval tremper et se
tremper (1923, 95). La diffrence est dans la prsence en franais
dune marque morphosyntaxique, celle des verbes pronominaux.
stupide ou faible, ces deux derniers verbes pouvant galement tre dalternance i/a. Cf.
aussi Ibn Xlawayh, Laysa . . . : 73 (note de lauteur).
Observons toutefois quen franais, de manire en partie similaire

taval en hbreu et atasa en arabe, il existe, outre les deux sens
que lon vient de voir, une forme intransitive moyenne plonger, de
sens physique, sans marque morphosyntaxique associe.36 Tous ces
exemples ont en commun dtre lexicaliss (la relation forme-sens
ne rsulte par dun phnomne de drivation compositionnel, mais
est inscrite dans le lexique).
Le verbe diathse externe daraba i frapper, battre en anglais,
to hit, to beat est cit dans tous les ouvrages, en langue arabe
ou en langues occidentales, pour illustrer sens du schme faala,
lexemple mille fois reproduit tant : daraba zaydun amran Zayd
a frapp Amr. Il comporte toutefois un sens dans lequel le procs
exprim par le verbe ne rencontre pas dobjet externe lagent, mais
affecte ce dernier (le verbe est donc diathse interne), comme
dans lexpression fige daraba f l-ardi parcourir la terre (ou une
contre, une rgion . . .). Un glissement smantique analogue se
produit en franais dans les expressions battre les chemins (sens
propre), ou battre la campagne (sens figur).
Verbes mdio-passifs (et donc non agentifs), partir de verbes dia-
thse externe :
Dans axada qalbuhu yadribu f sadrihi son cur se mit battre
dans sa poitrine, le verbe daraba prend une valeur non agentive ; le
procs affecte le sujet grammatical. Il est significatif que le mme
glissement smantique est observable en franais avec battre et en
anglais avec to beat dans les expressions correspondantes.
Le verbe diathse externe aqqa u fendre, briser (en anglais to
break) prend galement le sens de clore (bouton de fleur . . .),
cest--dire, fendre sa gaine, ou, pour une dent pousser. Ces
deux sens sont associs, respectivement, aux formes infinitives
(masdar) aqq et uqq. Comme dans tous les exemples ci-dessus,
le sens mdio-passif ou diathse interne drive du sens transitif
et diathse externe : il sagit dun phnomne de polysmie, inscrit
dans le lexique.
36
Toutefois, la diffrence des exemples emprunts larabe et lhbreu, le sens
physique de plonger intransitif et moyen cit ici comporte en franais un ajout
smantique (ide de plongeon), les donnes attaches ce verbe tant relativement
complexes. On a, par exemple, soit : plonger dans la piscine de son jardin, soit plonger
les mains dans leau froide du bain, mais on ne peut plonger (= faire un plongeon) dans
son bain . . .
356 joseph dichy
4.3.3 Les faala diathse interne ou mdio-passifs ne correspondant ni

un corrlat formel ni un glissement smantique
Il existe par ailleurs au sein du schme faala des verbes dont le sens est
ou en tout cas parat tre originellement celui de procs diathse
interne ou mdio-passifs, et qui ne relvent pas formellement des verbes
concaves ou redoubls. Exemples :
faala diathse interne :

dahaba a partir, aller, sen aller (agent humain) ; ma i marcher ;
daxala u entrer (agent humain) ; akala u manger (alors que ariba
a boire est, comme attendu, de schme faila) ; lamasa a toucher et
ra a voir (mais samia a entendre relve de faila) . . .
faala mdio-passifs :
zahaqa a prir, svanouir, disparatre ; waqaa a tomber ou se trou-
ver, tre situ (pour un lieu, une ville, etc.) ; dahaba a sen aller (sujet
grammatical non humain) ; fasada u et i se corrompre, pourrir (adj.
fsid, fasd).
4.3.4 Dispersion et rgularit dans le schme faala

Les exemples ci-dessus montrent quon trouve parmi les verbes en faala
toutes les catgories et proprits smantiques associes aux deux autres
schmes. la dispersion formelle prsente au dbut de ce travail semble
donc rpondre une dispersion smantique quivalente, dont on constate
lexamen quelle ne semble pas pouvoir tre rduite des corrlats for-
mels ou des glissements smantiques.
Une grande rgularit smantique peut toutefois tre observe : les
verbes diathse externe apparaissent exclusivement dans le schme
faala.
Cette zone de stabilit est, selon toute vraisemblance, rendue possible
par lorientation particulire du glissement smantique qui permet un
verbe diathse externe de devenir un verbe diathse interne ou un
verbe mdio-passif lorsque le procs verbal revient sur lagent ou sur
le sujet grammatical, ou, selon la formule de Benveniste, quand le sujet
devient intrieur au procs (1950/1966), exemples :
wadaa a poser, dposer est un verbe diathse externe : lagent

effectue laction dimprimer un objet le mouvement correspondant.
Le mme verbe prend le sens diathse interne de accoucher, qui
peut tre soit intransitif, soit transitif (Coran 3 : 36 et 46 : 15). Lagent
devient, dans ce cas, le lieu du procs. De mme, dans h amala i, le
sens diathse externe porter (un objet . . .) devient, avec une dia-
thse interne : porter [un enfant], devenir ou tre enceinte.
Le verbe wasala i joindre, unir une chose une autre (avec la prp.
bi-) est, dans son sens premier, diathse externe ; il est transitif un
double objet (de manire, respectivement, directe et indirecte). Au
sens de tre uni quelquun par un lien de parent, damiti . . . et de
frquenter assidment quelquun, lagent est en mme temps lun des
deux objets du procs : le verbe est alors diathse interne et transitif
( un seul objet).
Ces exemples, qui sajoutent ceux qui ont t prsents au 4.3.2,

illustrent un processus qui opre dans un sens, le sujet grammatical ou
lagent devenant le lieu du procs (comme dans wadaa et h amala) ou
lobjet de celui-ci (comme dans wasala). On peut en revanche formuler
lhypothse que le glissement smantique de la diathse interne vers la
diathse externe nest pas possible, compte-tenu des donnes de larabe :
une fois le sujet impliqu dans le procs, le verbe conserve cette pro-
prit smantique. Le passage de la diathse interne (exemple : jalasa i
sasseoir) une diathse externe sopre en construisant un autre verbe,
de mme racine, mais de schme diffrent, ainsi : ajlasa faire asseoir
(schme IV, af ala, de valeur causative-factitive).
Le passage dun verbe dtat caractristique un verbe diathse
interne se produit galement en changeant de schme, ainsi :
latufa u tre ou devenir gentil ou subtil est de schme faula ; latafa
u traiter quelquun avec gentillesse, de schme faala, est diathse
interne, la gentillesse affectant lagent.
Ces donnes dobservation permettent donc de formuler une hypothse

de porte descriptive.
5. Structure de la relation entre sens et forme

dans les schmes simples
La dispersion de la relation entre sens et forme dans les schmes du

verbe simple nest donc pas irrductible. Les raisons de cette disper-
sion ne sont pas, me semble-t-il, chercher dans la dtrioration dun
schma originel ou dun proto-schma qui entrerait en cohrence avec
celui des relations formelles entre les deux paradigmes de la conjugaison
objet de la figure 1. Sil parat raisonnable de partir dun proto-schma
358 joseph dichy
prsentant sur le mode hypothtique la relation entre sens et forme des

schmes du verbe simple dans son principe, il est essentiel de ne pas sar-
rter cette hypothse. Celle-ci est fonde sur lide que la relation sens
forme dans les schmes serait essentiellement, voire originellement,
de nature bijective. Or je montrerai plus loin que ce nest gure le cas.
5.1 Le proto-schma, reconsidrer, des relation entre sens et forme

dans les schmes simples
Le principe de ce proto-schma associerait, de manire exclusive, la
valeur smantique des verbes adjectif dtat caractristique au schme
faula, celle des verbes diathse interne ou mdio-passifs faila, et
celle des verbes diathse externe faala :
Schme Sens grammatical

faula Verbes adjectif dtat caractristique (tat ou
acquisition dtat)
faila Verbes diathse interne ou mdio-passifs
faala Verbes diathse externe
FIGURE 3 LES RELATIONS ENTRE SENS ET FORME DANS LE

VERBE SIMPLE, CONSIDRES DE MANIRE HYPOTHTIQUE
DANS LEUR PRINCIPE
Toutes les relations ci-dessus sont bijectives (ce qui est reprsent par
une flche deux ttes ). Il est peu vraisemblable quun tel systme ait
jamais exist en ltat, car, en sus dinluctables transformations morpho-
phonologiques, des glissements smantiques dune catgorie lautre et
divers effets de figement se sont ncessairement fait jour ds les toutes
premires poques de la langue. Un tel glissement est dj perceptible,
dans la figure 3, dans le fait que les verbes adjectif dtat caractristi-
que peuvent se raliser comme des verbes dacquisition dtat, qui sont
mdio-passifs. Or les verbes en faila peuvent galement correspondre
des mdio-passifs. Cest donc plus une reconstruction hypothti-
que du principe qui gouverne ces relations qu une reconstitution dia-
chronique que ce schma nous convie. Ce principe est reconsidrer
de deux points de vue : il masque larbre conceptuel des notions qui le
sous-tendent ( 5.2) ; il est mis en dfaut par les grandes lignes de glisse-
ments smantiques ou de modifications formelles observables dans les
schmes simples ( 5.3).
5.2 Larbre conceptuel des notions incluses dans cette structure

Conceptuellement, les grandes notions qui nous ont permis dtudier
les sens grammaticaux des schmes simples ne sont pas mettre sur
le mme plan : comme on la vu plus haut, les verbes diathse interne
(ou moyens) et mdio-passifs dun ct, et les verbes diathse externe
de lautre sont dans une relation dopposition. Or, les uns et les autres
peuvent correspondre un vnement ou un processus, mais non
un tat opposition fondamentale que masquait la distinction cou-
ramment rpandue entre verbes daction et verbes dtat. Lensemble
constitu par les verbes valeur dtat caractristique soppose dune part
aux verbes mdio-passifs qui incluent les verbes dacquisition dtat
caractristique ou non , et aux verbes et diathse interne ou externe
de lautre.
Lagentivit enfin opre elle mme une partition entre les verbes
valeur dtat caractristique, dans lesquels cette proprit est neutrali-
se, et les verbes mdio-passifs, qui sont non-agentifs. Par contraste,
les verbes diathse interne ou externe sont, selon le cas, entirement
ou partiellement agentifs. Ces oppositions sont rsumes dans larbre
conceptuel suivant :
notion aspectuelle dtat notions aspectuelles dvnement ou de processus
(agentivit neutralise) non agentivit agentivit entire ou

partielle
diathse diathse
interne externe
Verbes valeur Verbes Verbes Verbes

dtat caractristique mdio-passifs diathse diathse
( y compris interne externe
dacquisition dtat)
FIGURE 4 LES VALEURS SMANTIQUES DE BASE DES VERBES

DE SCHME SIMPLE, ARBRE CONCEPTUEL
Conventions : Les proprits smantiques sont en caractres gras ; les

catgories smantiques des verbes du schme simple sont en italiques.
On notera quen raison de phnomnes de polysmie, un mme verbe
peut appartenir des catgories smantiques diffrentes : les proprits
ci-dessus peuvent donc correspondre soit des catgories lexicales, soit
des occurrences en contexte.
360 joseph dichy
5.3 Glissements smantiques luvre dans la relation sens-forme dans

les schmes simples
Si lon part de laspect formel des schmes, on observe que :
(a) tous les verbes en faula correspondent un verbe adjectif dtat

caractristique (valeurs dtat caractristique ou dacquisition de
celui-ci) ;
(b) presque tous les verbes en faila relvent de lune ou lautre des
valeurs du verbe diathse interne ou du mdio-passif. Ce schme
inclut par ailleurs un sous-ensemble, rduit en nombre, de verbes
adjectif dtat caractristique ;
(c) les verbes en faala, cependant, nappartiennent que partiellement
aux verbes diathse externe.
Si lon part au contraire du sens grammatical, on observe que :
(d) tous les verbes diathse externe sont inclus dans le schme faala ;
(e) les verbes adjectif dtat caractristique sont en contrepartie repr-
sents dans les trois schmes ;
(f) les verbes moyens et mdio-passifs sont rpartis entre les schmes
faila et faala (dans le schme faula, les verbes dacquisition dtat
sont mdio-passifs).
Les deux relations (a) et (d) sont de la forme tous les . . . sont . . . ; la rela-
tion (b) est du type presque tous les . . . sont . . .. Ce qui prcde peut tre
reprsent par la figure suivante :
Schme Sens grammatical

faula Verbes adjectif dtat
caractristique
(tat ou acquisition dtat)
faila
Verbes diathse interne ou
mdio-passifs
faala
Verbes diathse externe
FIGURE 5 LES RELATIONS ENTRE SENS ET FORME DANS

LE VERBE SIMPLE
Conventions :
Les flches dont le trait est doubl ( ) indiquent une rela-

tion mettant en jeu la totalit des lments de lensemble considr. La
convention se lit : pour tout schme ou sens grammatical situ au dbut
de la flche, le sens ou le schme est celui qui est indiqu par la pointe de
la flche (cf. section 4.1).
La flche interrompue de trait doubl ( ) indique une
relation mettant en jeu la quasi totalit des lments de lensemble
considr, avec toutefois un ensemble dexceptions clairement identi-
fiable. La flche il ny en a quune se lit : pour presque tout schme
situ au dbut de la flche, le sens grammatical est celui qui est indiqu
par la pointe de la flche (voir section 4.2).
Les flches en pointills ( ) indiquent une relation qui ne
concerne quun nombre rduit et ferm de verbes, dans linventaire
duquel apparaissent des contraintes morphologiques, comme par
exemple dans les verbes dtat caractristique du schme faala
( 4.2.1 et 4.3.1).
La flche en trait plein ( ) signale une relation entre sens
grammatical et schme qui concerne un nombre important et en tout
tat de cause ouvert de verbes. Ces verbes sont produits par un pro-
cessus identifiable de glissement smantique ( 4.3.2 et 4.3.3).
6. Le dcrochage partiel entre sens et forme :

une relation non bijective
Le dcrochage partiel entre sens et forme illustr ci-dessus est d au jeu

de deux types de contraintes : formelles (incluses dans les flches trans-
versales en pointill) et smantiques (flche transversale en trait plein).
Pour comprendre la structure gnrale laquelle on est confront, il faut
tenir compte du fait que la relation entre les schmes et leur sens gram-
matical nobit par une relation bijective. On rencontre en effet, soit des
relations schme sens grammatical, comme dans faula et faila, soit
une relation schme sens grammatical, comme dans faala, mais non
la relation bijective ( deux sens) schme sens grammatical, comme le
laisserait croire une conception nave du langage.
Un tel dcrochage est frquent dans les langues : on le rencontre,
typiquement, dans la relation entre les actes de langage (Searle) et leurs
362 joseph dichy
ralisations dans une langue donne. En reconnaissance, il est gale-

ment trs frquent dans les valeurs aspectuo-temporelles associes en
contexte une forme verbale : une mme forme supportant plusieurs
interprtations, cest la mise en rapport de celle-ci avec divers indices
contextuels qui permet au processus de comprhension doprer (Des-
cls et al. 1998 ; Descls et Guentchva, paratre). Le reprage dindices
est particulirement important dans le processus de lecture en arabe,
lcriture courante tant dpourvue des signes diacritiques secondaires
dits de vocalisation.
Dans la constitution du lexique arabe, la relation entre les sens gram-
maticaux associs un schme donn et les sens lexicaux qui sadjoignent
un nom ou un verbe est soumise leffet du principe de figement
lexical (PFL), qui soumet le sens grammatical des schmes, ds le
niveau du mot, des carts smantiques (Dichy 2003, 204208). Dans le
verbe simple, des glissements smantiques tels que ceux qui sinstaurent
entre les valeurs de verbe dtat (tre x) et dacquisition dtat (devenir
x) ou celui qui permet certains verbes diathse externe de pren-
dre une valeur de diathse interne, soumettent les relations entre le sens
grammatical des schmes et leur forme de fortes pressions. Le proto-
schma de ces relations (figure 3) persiste toutefois dans le schma des
relations observables (figure 5), mais avec des connexions entre sens et
forme qui, ntant pas bijectives, sinscrivent dans la structure gnrale
des rapports entre le sens et la forme que lon observe dans le processus
de comprhension du langage humain.
7. Rfrences bibliographiques
7.1 Sources primaires

Ibn Xlawayh, al-H usayn b. Ahmad (m. en 370/981). Laysa f kalm al-arab, d. Ahmad
A. At t r. Beyrouth : Dr al-Ilm li-l-malyn, 1979.
Ibn Ya (m. en 643/1245). arh al-Mulk f t-tasrf, d. Faxr ad-Dn Qabbwa. Alep :
Al-Maktaba l-arabiyya, 1973.
. arh al-Mufassa l. Le Caire : Maktabat al-Mutanabb / Beyrouth : lam al-kutub,
10 tomes en 2 vols.
al-Mubarrid (ou al-Mubarrad), Ab l-Abbs (m. en 285/898). Al-Muqtadab, d.
Muhammad A. Udayma. Le Caire : Al-Ahrm, 1399/1978, 4 vols.
Sbawayhi (m. v. 180/796). Al-Kitb, d. Abd as-Salm M. Hrn. Le Caire : Al-Haya
l-misriyya l-mma li-l-kitb, 1977, 5 vols.
Yawm. = Tawfq al-H akm. 1937. Yawmiyyt nib f l-aryf. Le Caire : Dr Misr li-t -
tiba.
az-Zajjj, Ab l-Qsim (m. v. 340/952). Al-Jumal f nahw, d. Al T. al-H amad. Bey-
routh : Muassasat ar-risla, 1988 (4e d.).
az-Zamaxar, Ab l-Qsim. Al-Mufassa l f ilm al-arabiyya. Beyrouth : Dr al-Jl, s.d.
7.2 Sources secondaires

Abbs, Ramzi. 2004. La conception et la ralisation dun concordancier lectronique pour
larabe. Thse de doctorat en sciences de linformation, Lyon, ENSSIB/INSA.
Abraham, Maryvonne. 1995. Reprsentations smantico-cognitives des verbes (chan-
tillon des verbes de mouvement) : problmes et mthode. Thse dinformatique, Paris :
EHESS.
Ammar, Sam et Joseph Dichy. 1999. Les verbes arabes. Paris : Hatier (coll. Bescherelle).
Audebert, Claude. 2002. Verbes actifs et moyens dans le parler du Caire : une suite.
Annales islamologiques 36, Le Caire, I.F.A.O.
Badawi, Elsaid, Michael G. Carter, and Adrian Gully. 2004. Modern Written Arabic. A
comprehensive Grammar. London and New York : Routledge.
Belot, J. B. 1922. Cours pratique de langue arabe. Beyrouth : Imprimerie catholique.
Benveniste, mile. 1950/1966. Actif et moyen dans le verbe, in : Problmes de linguisti-
que gnrale. Paris, Gallimard, t. 1 (1966), 1950, repris in 1966, ch. XIV, 168175.
Blachre, Rgis et Maurice Godefroy-Demombynes.1952. Grammaire de larabe classi-
que. Paris : Maisonneuve (3e d.).
Boormans, Maurice. 1967. Grammaire darabe littral. Feuilles de travail. Rome : Institut
pontifical dtudes arabes (document mimographi).
Brockelmann, Carl. 1948. Arabische Grammatik. Leipzig : O. Harrasowitz.
Cantineau, Jean. 1950. La notion de schme et son altration dans diverses langues
smitiques. Semitica III. 7383.
Caspari, C.P. 1881. Grammaire arabe. Trad. E. Uricoecha. Paris : Maisonneuve.
Cohen, Marcel. 1929. Verbes dponents internes (ou verbes adhrents) en smitique.
Mmoires de la Socit de Linguistique de Paris XXIII : 4, 225248.
Comrie, Bernard. 1976. Aspect. An Introduction to the study of verbal aspect and related
problems. Cambridge University Press.
. 1985. Tense. Cambridge University Press.
Descls, Jean-Pierre. 1990. Langages applicatifs, langues naturelles et cognition. Paris :
Herms.
. 1994. Quelques concepts relatifs au temps et laspect pour lanalyse des textes.
Studia kognitywne 1, SOW, Varsovie. 5788.
, Valrie Flageul, Christiane Kekenbosch, Jean-Marc Meunier et Jean-Franois
Richard. 1998. Smantique cognitive de laction : 1. Contexte thorique. 2. tude
exprimentale de la catgorisation des verbes daction. Langages 132 (Cognition, cat-
gorisation, langage). 2847 ; 4868.
et Zlatka Guentchva. paratre. Aspectualit, temporalit : une approche cognitive
et formelle partir des langues. Document mimographi.
Dichy, Joseph. 1993. Knowledge-system simulation and the computer-aided learning
of Arabic verb-form synthesis and analysis. Processing Arabic Report, 6/7. T.C.M.O.,
Universit de Nimgue. 6784 ; 9395.
. 2002/2003. Structure de la drivation lexicale en arabe : sens et forme des verbes et
des drivs nominaux les plus immdiats, Cours de prparation au CAPES darabe,
session 2003, question de linguistique. Paris : C.N.E.D.
. 2003. Sens des schmes et sens des racines en arabe : le principe de figement lexical
(PFL) et ses effets sur le lexique dune langue smitique. In Sylvianne Rmi-Giraud
et Louis Panier, dirs. La polysmie ou lempire des sens. Lexique, discours, reprsenta-
tions. Presses universitaires de Lyon. 189211 (www.concours-arabe.paris4.sorbonne.
fr/cours/dichy.doc)
et Mohamed Hassoun. 2005. The DIINAR.1 ! Arabic Lexical Resource, an
outline of contents and methodology. The ELRA Newsletter, vol. 10, 2, April-June 2005.
510.
, Abdelfattah Braham, Salem Ghazali et Mohamed Hassoun. 2002. La Base de
connaissances linguistique DIINAR.1, (DIctionnaire INformatis de lARabe
version 1). In A. Braham, d. Actes du colloque international sur Le Traitement auto-
matique de larabe. Tunis : Universit de La Manouba. 4556.
364 joseph dichy
Fleisch, Henri. 1957. tudes sur le verbe arabe. Mlanges Louis Massignon. Institut
Franais de Damas. 153181.
. 1968. Larabe classique. Esquisse dune structure linguistique. Beyrouth : Dr al-
Machreq.
.1979. Trait de philologie arabe vol. 2. Beyrouth : Dr al-Machreq.
alyn, Musta f. 1912. Jmi ad-durs al-arabiyya. 19e d. revue par Muhammad A.
an-Ndir. Beyrouth, al-Maktaba l-Asriyya, 1994, 3 vols. en un.
Guillaume, Jean-Patrick (1984), Quelques aspects de la thorie morpho-phonologique
dIbn Jinn. propos des verbes glide mdian. In G. Bohas et J.-P. Guillaume, ds.
tude des thories des grammairiens arabes. I. Morphologie et phonologie. Institut fran-
ais de Damas. 338490.
Ilys, Jzf et Jirjis Nsf. Mujam ayn al-fil. Beyrouth : Dr al-Ilm li-l-malyn, 1995.
Joon, Paul. 1923. Grammaire de lhbreu biblique. Rome : Institut pontifical, rd.
1965.
. 1930. Smantique des verbes statifs de la forme qatila (qatel) en arabe, hbreu et
aramen. Mlanges de lUniversit Saint Joseph, XV : 1, 332.
Larcher Pierre. 1995. O il est montr quen arabe classique la racine na pas de sens et
quil ny a pas de sens driver delle. Arabica XLII. 291314.
. 1996. Drivation lexicale et relation au passif en arabe classique. Journal asiatique
284, 2. 265290.
. 2003. Le systme verbal de larabe classique. Aix-en-Provence : Publications de
lUniversit de Provence.
Leeman-Bouix, Danielle. 1994. Grammaire du verbe franais, des formes au sens. Paris :
Nathan.
Lyons, John. 1978/1990. Smantique linguistique, trad. fran. de la 3e d., revue et corri-
ge de Semantics II, Cambridge University Press (1978). Paris : Larousse, 1990.
Maingueneau, Dominique. 1994. Lnonciation en linguistique franaise. Paris : Hachette.
Moscati, Sabatino, d. 1964. An Introduction to the Comparative Grammar of the Semitic
Languages. Phonology and Morphology. Wiesbaden : O. Harrassowitz (2e d., 1969).
Al-Mujam al-Wast. 1973. Ibrhm Ans, Abdalhalm Muntasir, Atiyya as-Sawlih,
Muhammad X. Ahmad, eds. Le Caire : Dr al-Marif, 2e d. (1e d. 1960).
Al-Munjid f l-lua l-arabiyya l-musira. 2000. Subhi H amw, d. Beyrouth : Dr al-
Mariq.
Neyreneuf Michel et Ghalib Al-Hakkak. 1996. Grammaire active de larabe. Paris : Livre
de Poche.
Nr ad-Dn, Ism. 2002. Abniyat al-fil f fiyat Ibn al-H jib. Beyrouth : Dr al-Fikr
al-lubnn (1e d. 1982).
Qabbwa, Faxr ad-Dn. 1998. Tasrf al-asm wa-l-af l. Beyrouth : Maktabat al-Macrif,
3e d.
Roman, Andr. 1983. tude de la phonologie et de la morphologie de la koin arabe. Mar-
seille : Jeanne Laffitte, 2 vols.
. 1990. Grammaire de larabe. Paris : P.U.F. (coll. Que sais-je ?).
. 1999/2005. La cration lexicale en arabe, ressources et limites de la nomination dans
une langue humaine naturelle. Presses Universitaires de Lyon, 1999. 2e d. revue et
augmente, Presses Universitaires de Lyon et Universit de Kaslik, 2005.
a-artn, Rad. 1912. Mabdi al-arabiyya. Beyrouth : Imprimerie catholique (1e d.).
Sylvestre de Sacy, Antoine. 1831. Grammaire arabe lusage des lves de lcole spciale
des langues orientales vivantes. Paris : Imprimerie royale (2e d.). Rimpr. photomca-
nique. Paris : Institut du Monde arabe, 2 vols. s.d.
Versteegh, Kees. 1997/2003. The Arabic Language. New York : Columbia University Press,
1997. Trad. arabe : Al-lua al-arabiyya, trxuh wa-mustawaytuh wa-tatruh,
trad. Muhammad a-arqw. Le Caire : Al-Majlis al-al li-t -taqfa, 2003.
. 2004. Meanings of speech. The category of sentential mood in Arabic grammar.
In Joseph Dichy et Hassan Hamz, ds. Le voyage et la langue, Mlanges en lhon-
neur dAnouar Louca et Andr Roman. Damas : Institut Franais du Proche-Orient,

269287.
Wehr, Hans, John M. Cowan (ed). 1979. A Dictionary of Modern Written Arabic (Arabic-
English), edited by J. M. Cowan, 4th edition. Wiesbaden : O. Harrassowtiz.
Wright, W. 189698. A Grammar of the Arabic Language. Rimpr. Beyrouth : Librairie
du Liban, 1974.
FEATURING AS A DISAMBIGUATION
TOOL IN ARABIC NLP
Everhard Ditters
University of Nijmegen
1. Introduction
Featuring is the affixation of, general as well as language dependent,

second-level labels (such as gender1 or animate) to first-level non-
terminals (such as noun, verb or particle) in a formal description of
the syntax and semantics of a specific natural language. Such a formal
grammar may serve as input for the automated processing of the natural
language (NLP) concerned: Modern Standard Arabic.
About 98% of Modern Standard Arabic texts, whether in printed or
electronic form, are represented in a non-vocalized shorthand form.
The analysis of this kind of data, automated or otherwise, has to cope
with an exponential combinatorial ambiguity unless:
one considers the units of linguistic description beyond word level

such as constituent, sentence, paragraph or even text-level;
one combines this description with an adequate, coherent, consistent,
and as exhaustive as possible, featuring system for analysis purposes.
Therefore, our objective is: the design of a featuring system as a tool

for the disambiguation2 of undesired analysis results in the automated
processing of Arabic text data.3
1
Feature names and feature values are represented in the text in Italics. We use two
types of features: inherent and inherited features. The former represents the intrinsic
semantic value of the entry concerned. The latter follows as modifier the entry it modi-
fies according to language dependent concord and agreement rules.
2
We use the term disambiguation for attempts to obtain a single and most probable
syntactically and semantically correct analysis of input data in an automated processing
environment.
3
Language phenomena such as homonymy, polysemy and even antinomy should
also be controlled by means of adequate featuring.
368 everhard ditters
A comprehensive inventory of general or universal linguistic features

is not (yet) available for Modern Standard Arabic. This is mainly due to
the complexity of the description of semantic features. Usually, linguistic
theories and descriptive language models tend to become language
independent. However, we preferwhile exploiting current, language
independent linguistic theories and descriptive modelsto describe the
features that are specific for the language under consideration.
If the linguistic model chosen for the description may have some
impact on the organization of the featuring system, it remains possible to
use heuristics and to refine the formal description any time we consider
it worthwhile in order to avoid undesired ambiguities. So we are free to
insert or to combine features such as agreement and concord,4 in order
to account for regularities, relationships and dependencies occurring
between different constituents, or between different elements within a
constituent.
Considering Modern Standard Arabic language description, we
distinguish three layers of featuring: a morphological, a syntactic and a
semantic layer. They describe the form, the function and the meaning
of elements used in a, agreed upon by convention and expandable by
language evolution, system of information interchange, tailored to the
needs of users of any natural language system. This tri-partition roughly
coincides with traditional classifications into orthography, phonology,
morphology, syntax, stylistics, discourse, and even semantics.5
A description of the Arabic phonemes for analysis purposes does not
require a differentiation in terms of their production characteristics. A
listing of phonemes (26 consonants, three glides [semi-consonants or
semi-vowels],6 and three vowels [in a short and long variety]), together
with a number of graphemic alternatives, constitutes the basis for our
phonemic description of Modern Standard Arabic.
4
We distinguish between concord and agreement and reserve the latter to describe
regular patterns in relationship between an explicit agent and its predicate in a ver-
bal sentence (Sv) involving gender (Ditters 1992, 169, n. 13; Kihm 2006, 1415; cf. also:
Bahloul 2006, 4348). We use the term concord for feature-value sharing within the
noun phrase (NP) between the head and its modifiers, as well as in the nominal sentence
(Sn) for the marked relation between the topic and comment, involving, if applicable,
definiteness, gender, number, and case.
5
We are definitely not trying to introduce a formalized dynamic description of the
Arabic concept of the world. We rather follow a static semantic approach using finite
enumeration of pertinent (static) semantic features. See also subsection 4 below.
6
For coherency within our description, we adopt Sbawayhis inventory of 29 Arabic
consonants (Hrn 1982, 4, 431).
featuring as a disambiguation tool in arabic nlp 369
The introduction of a feature such as verbtype may well account for

alternative realizations of the finite forms of so-called weak verbs. A
feature like complement structure7 controls the number (argument-1
[arg-1], [arg-2], [arg-3] etc.) as well as the format (noun phrase (NP),
prepositional phrase (PP), complement clause (CCL) etc.) of possible
complements of a lexical verbal entry. By means of predicate-subject
and predicate-object matching features we are in a position to consider-
ably refine our formal description and to distinguish, for example,
between ordinary and metaphorical language use. However, for the dis-
ambiguation of analysis results our main task at this moment remains: to
draft a general pattern for nominal and verbal featuring; and to compose
a basic set of (semantic) features and feature-clusters embedded in an
adequate description of verbal complement structure.
A remark about the linguistic model chosen for the description may
be useful. We still opt for a Phrase Structure Grammar (psg) model
based on immediate constituency (ic), but enlarged with a second level
of description to account for prevailing relationships and dependencies
within a Modern Standard Arabic sentence.8 We even foresee the
introduction of a third level of description, grouping together pertinent
semantic features with information about their characteristic form of
realization at syntax level. Inspired by Saads (1982) syntactic-semantic
study of the Classical Arabic verb in Case Grammar (cg) terms, we
will use a number of Fillmores (1968) cg ideas as well as later
developments for the sub-categorization and patterning of verbs in
Modern Standard Arabic.9
Finally some words about the processing environment. We use the
agfl-system,10 a parser generator for research and applications in
7
To be elaborated upon later at syntax level when speaking about: tri-, bi-, mono-,
and intransitive verbs or the reduction to a one-argument realization used to uniquely
emphasize the semantic load of the verbal entry ; and at the semantic level while speaking
about: semantic feature hierarchy; the minimal and the maximal realization of comple-
ment structure; compulsory and optional complement structure realization; case roles;
and verb sub-categorization.
8
For other applications such as information retrieval (IR) or text-summarizing
(TS), we use a Dependency Grammar (DG) approach since it appears to express more
adequately semantic properties in terms of nodes and relations between nodes (Ditters
and Koster 2004).
9
Only recently we decided to include al-Saffr (1979) in our research for his seman-
tico-syntactic features in Case-Grammar terms.
10
For more information about the AGFL formalism, now in its 2.4th Windows ver-
sion, see Koster (1991) and www.cs.ru.nl/agfl. For more information about the formal
description of Arabic, see Ditters 1992; 2003a; 2003b; and forthcoming.
nlp. The acronym refers to Affix Grammars over Finite Lattices.11 The
parsers generated by the system are top-down recursive backup parsers
(Koster 1974, Meijer 1986), based on non-deterministic concepts and
the unification-principle (Nederhoff 1993). The agfl-formalism is
part of the family of two-level grammars: a context-free grammar is
augmented with set-valued features for expressing agreement between
syntactic categories. The formalism is, in principle, suited for describing
morphological and syntactic structure, as well as finite semantics in
terms of animate, concrete, human, volition and many others.
We discuss feature-sets at morphologic ( 2), syntactic ( 3), and
semantic level ( 4). We end with a conclusion ( 5) and a list of
references used ( 6).12
2. Morphology
Strictly speaking, morphology relates to the formal description of the

individual Arabic parts of speech: verb, noun and particle, and, if needed
or useful, of their differentiation into a further sub-categorization.13 In
a broader sense, morphology also comprises feature names and values
such as: aspect, case, definiteness, derivation, gender, number, person,
tense, voice, and many other features used at syntax level. Morphology,
finally, has to account for punctuation marks and other textual noise,
normally more conveniently stored in the lexicon module.
Literary Arabic has been described as a predominantly root and
pattern language type.14 Moreover, vowel-pattern variation combined
11
Here, the term affix, a variable with a finite set of values, has to be taken in its for-
mal and not in its linguistic sense.
12
Within the morphological, syntactic and semantic sections, we use the traditional
Arabic language parts-of-speech (POS) differentiation into noun, verb and particle as
headings for the subsections. In these subsections we only discuss the, for us, relevant
features.
13
Cf. Sbawayhis tri-partition (Hrn 1982, 1:12) and the subdivision of POS into 7
classes by alSq (1977, 214).
14
Cohen (1970, 49 ff.) has been one of the first to describe, for automated Arabic lan-
guage processing, a frame of mostly three, sometimes four and rarely five, consonants or
semi-consonants filled with a combination of vowels (including the absence of a vowel
at a certain slot) expressing semantic differentiations to the global meaning of the conso-
nantal root combination. Elements of a small subset of the phoneme inventory are used
to produce other derivates of the base frame, whether of the category verb, noun or
adjective, with their own specific variation on the global meaning of the consonantal
root combination.
with the insertion of a finite set of auxiliary phonemes triggers a more

or less regular, more or less predictable change in the semantic load of
the base root meaning.
The Arabic phoneme inventory has extensively been described else-
where.15 As far as the consonants are concerned, the description com-
prises a cluster of features referring to their way and place of production,
as well as to their marking for velarization and degree of voicing. As far
as the vowels are concerned, they have been described by their place and
the openness of that place, set out against the position of the tongue in
their production process.
For analysis purposes, we have the Arabic grapheme inventory with
variant scriptures for alternatives occurring in initial, median, end or
independent position. In order to account for fully, partially or non-
vocalized Arabic input data, the listing of the Arabic vowel graphemes
will include an empty vowel realization.
2.1 Verbs
Assigning morphological features and feature values to verbs, we
account for: the realization of finite verb occurrences on the basis of
number and place of the root consonants (radical, r1,16 r2, r3, and less
frequent: r4 and r5); the realization of the vowel (a, i, u) of the second
consonant (C2) of the base (Stem I) verb realization in perfect and mood
aspect (vowperf, vowimperf ); and the realization of prefixed or infixed
vowels (vow or sukn).
Morphological features, attached to elements of a finite verb realization
at syntactic level (see also 3.1 below), affect values concerning: aspect,
tense,17 voice, person, gender and number. For the description of a finite
15
Cf. Fleisch mainly used three, sometimes four, features to define the elements of
the consonantal system, which ends up in a 28 u 4 matrix (1961, 1:5665) or a 16 u 5
matrix (1968, 19); Versteegh (2001, 20) used a 9 u 6 matrix; Saad (1982, 6) used a 7 u
9 matrix for the consonants and a 4 u 2 matrix for the vowels. See also the phoneme
featuring of Bohas and Saguer (2007, 255 ff.) in this volume.
16
A listing of the Arabic phonemic (consonant and vowel) system is given in the lexi-
cal module(s). For analysis purposes, a single description of the r1 suffices. For genera-
tion purposes, features describing occurrence incompatibilities between an r1, r2 and r3
(r4, r5) should be provided for.
17
As Badawi et al. (2004, 362 ff.) do, we distinguish between a perfect and a non-
perfect (mood) value of the feature aspect of the verb. In combination with modal and
temporal verbs and/or adverbials a complete range of temporal and aspectual differen-
tiations (tenses) can be described (see on aspect and tense also Eisele 2006, 195201;
Bahloul 2006, 506; and Reese 2006, 5053).
verb form of the so-called weak verbs with alternative realizations in its
conjugation, a variable verbtype accounts for alternative verb realizations.
To account for the differentiation in base and derived stems we use the
feature derivation. With that we can list, in the form of a context-free
metagrammar, the following feature names or non-terminal affixes
(in capital letters) with their finite-set of values or terminal affixes (in
lower case), connected to a finite verb form:18
ASPECT :: perfect; MOOD.
MOOD :: indicative; subjunctive; jussive; imperative.19
TENSE :: temporal; modal.
VOICE :: active; passive.
PERSON :: first; second; third.
GENDER :: feminine; masculine.
NUMBER :: singular; dual; plural.
VERBTYPE :: r1r2r3; r2=r3; r1=w; r1=y; r2=a; r2=w;
r2=y; r3=w; r3=y; r1=w, r3=y; r2=w, r3=y;
r1=hamza; r2=hamza; r3=hamza.
DERIVATION :: base; DERIVED.
DERIVED :: ii ; iii ; iv ; v; vi; vii; viii; ix; x.
RADICALS :: three; four; five.
R1 :: alphabet.
R2 :: R1.
R3 :: R1.
R4 :: R1.
R5 :: R1.
VOW :: vowel; sukn.
VOWPAT :: VOW.
VOWIMPERF :: VOW.
VOWPERF :: VOW.
18
We will use the following conventions for the formal description of features and
the finite-set of values at the second level of description which closely follow the AGFL
convention:
feature names are written in upper case;
feature values are written in lower case;
the rewrite symbol is a double colon ::;
a single left-hand entry is rewritten in one ore more feature names
and/or feature values at the right-hand side;
alternative realizations at right-hand side are separated by a semicolon;
options in the right-hand side are separated by a vertical bar;
a rule will be closed by a period ..
19
The archaic energetic-1 and energetic-2 finite verb forms are not accounted for in
our formal description of Modern Standard Arabic.
2.2 Nouns
Features, at the morphological level assigned to elements of the category
noun, concern variables such as: person, gender and number.20 For the
description of nominal plural building, it is worthwhile to differentiate
between an external and internal formation. Besides person, gender and
number, the feature case is a pertinent part of the declension system,
in Modern Standard Arabic closely connected with the type of
declension.
For the description of a differentiation in base and derived stems for
deverbal nouns (active and passive participles, infinitives, nouns of
time, place, doing an action once, or referring to instruments used in an
occurrence), we use the feature derivation. By means of morphological
base rules, we describe the subset denominatives of the category noun,
such as: individuality nouns, multitude nouns, vessel nouns, relation
nouns, abstract quality nouns and diminutives.21
The feature variable determination with values as definite, indefinite
or morphologically neutral is another inherent or inherited (at syntax
level) characteristic of elements of the category noun. Elements of
the sub-classes demonstratives, personal pronouns, proper nouns, and
relative pronouns are inherently definite. Elements of the sub-classes
indefinite and interrogative pronouns definitely possess an inherent
indefinite value.
As far as the definite article is concerned, we follow Wrights inter-
pretation of the Arabic grammatical tradition in describing this article
on the basis of its deictic and not of its generic value as a sub-class of the
demonstratives.22 Furthermore, elements of the sub-classes adjectives,
common nouns, elatives, numerals, and quantifiers may, or may not,
receive a definite or an indefinite value depending on their function
and occurrence in phrases at syntax level. So we list the following
morphological feature rules for a noun:
CASE :: nominative; NOMINATIVE.23
DECLENSION :: invariable; diptote; triptote.
20
Since identical looking feature values are attached to distinct non-terminal names
(different category labels), we can reuse these feature names without any risk for unde-
sired ambiguities.
21
Cf. Wright 1974, 1:109177.
22
Cf. Wright 1974, 1:264270. See also Fleisch 1961, 1:339347.
23
The AGFL formalism allows for the use of logical markers such as +, -, and others
in combination with feature names and values.
DENOMINATIVE :: individuality; multitude; vessel; relation; quality;

diminutive.
DERIVATION :: base; DERIVED.
DERIVED :: ii; iii; iv; v; vi; vii; viii; ix; x.
DETERMINATION :: definite; indefinite.
DEVERBAL :: actpart; paspart; infinitive; time;
place; once; instrument.
GENDER :: feminine; masculine.
NOMINATIVE :: genitive; accusative.
NUMBER :: singular; dual; PLURAL.
PERSON :: first; second; third.
PLURAL :: external; internal.
2.3 Particles
According to the traditional description24 of particles we may distinguish
between: adverbs, conjunctions, interjections, and prepositions.25 A formal
description should include:
PARTICLE TYPE :: ADVERB; CONJUNCTION;
INTERJECTION ; PREPOSITION.
2.3.1 Adverbs
A for us still useful sub-categorization of adverbs, at morphological level,
is that in invariable, as far as case marking is concerned, and bound and
free forms, as far as the orthographic representation is concerned.
ADVERB :: bound; free; invariable.
2.3.2 Conjunctions
As far as the coordinating particles are concerned, a distinction should
be made in cumulative and selective particles because of the importance
of number value in concord and agreement phenomena at syntax level.
We then divide the selective particles into: alternative, consecutive,
explicative, exclusive, inclusive, restrictive, and successive elements.26 This
sub-categorization will be extensively used at the next higher level of
description.
24
Cf. Wright, 1974, 1:282296.
25
CF. Badawi et al. 2004, 174219; Cantarino, 19745, 2:253 ff.; El-Ayoubi et al. 2003,
1:2, 275460.
26
Cf. Ditters 1992, 222228.
For the conditional particles we prepare their application at syntax

level with a differentiation as to the nature of the condition in: possible,
real and unreal.
CONJUNCTION :: COORDINATING; CONDITIONAL.
COORDINATING :: cumulative; SELECTIVE.
SELECTIVE :: alternative; consecutive; exclusive; explicative;
inclusive; restrictive; successive.
CONDITIONAL :: possible; real; unreal.
2.3.3 Interjections
Elements of this sub-section of particles, rather close to discourse and
formalized language use, can, as far the morphological level is concerned,
be divided into vocatives and exclamations.27 However, we also have to
decide whether or not proverbs and frozen expressions should be
lodged among the interjections and labeled as adverbs or adverbials at
syntax level. In any case we also like to mention here the category of
formulaic greetings.28
INTERJECTION :: vocative; exclamation; f-expressions; proverbs;
greetings.
2.3.4 Prepositions
The standard differentiation of prepositions at this level is into a primary
and a secondary group. The secondary group consists of an open set of
noun-derived entries, marked for their function by means of a definite
accusative case value and, at syntax level, engaged in a construct state-like
link with the prepositional complement, itself marked with genitive case
value (NPgen). The primary group comprises a finite-set of non-derived
entries, some of which are, in their orthographic representation, directly
bound to the prepositional complement while others are unbound.29
PREPOSITION :: PRIMARY; secondary.
PRIMARY :: bound; unbound.
27
A still poorly described domain of frozen or set expressions like: greetings, insults,
proverbs and similar insertions should be included here. Cf. Bergman 2007, 136137.
28
Cf. Elzeiny 2007, 202207.
29
Cf. El-Ayoubi et al., 2003, I:2:574592.
3. Syntax
At syntax level, elements of the earlier mentioned word categories

and sub-classes are assembled into constituents, labeled: verb phrase
(vp), noun phrase (np), and particle phrase, sub-categorized according
to the element realizing the compulsory head function.30 These phrases
realize specific compulsory and optional functions within the two main
units of linguistic description at sentence level: the nominal (sn) and the
verbal (sv) sentence.31 It may be clear, that the combinatorial behavior
of elements within a constituent, and that of constituents at sentence
level, comprising mutual relationships and dependencies, is controlled
by features and feature-condition rules.
In Modern Standard Arabic text data one distinguishes within the
nominal sentence (sn) the compulsory topic and comment functions
(Figure 1a). In a verbal sentence (sv) we are dealing with an compulsory
predicate function and several optional complements (Figure 1b).
Optional sentence adverbials can occur in both sentence types.
A np with a nominative case value for the head (npnom) always32
realizes the topic function.33 The comment function is realized by: a np34
with nominative case value for the head (npnom); an adjp with nominative
case value for the head (adjpnom); a vp; a pp; or an adverb phrase (advp).
A vp always fills in a sv the predicate slot. Sentence adverbials (sadv) are
realized by: a np with accusative case value of the head (npacc); a pp; an
advp; a cp; or a ccl.35
30
As mentioned in 1, a particle phrase could be subdivided into a prepositional
phrase (PP), an adverbial phrase (ADVP), an interjectional phrase (IP), and (to remain
consistent) a conjunctive phrase (CP), or a complement clause (CCL).
31
Distinct concord and agreement phenomena mark a productive syntactic (and
semantic) distinction between a nominal and a verbal sentence type in MSA. Discourse
sensitive emphasis on the topic agent should be maintained, side by side with the pos-
sibility of emphasizing the action performance of, sometimes, the same agent in typical
VSO-oriented approaches. Cf. Ditters 2001, 3137.
32
The optional occurrence of a particle, like inna with an emphasizing semantic load
and governing its complement by an accusative case value, only represents an alternative
realization within the base structure.
33
An alternative topic realization is a CCL (complement clause), introduced by the
emphatic particle inna, governing the head of the following NP in the accusative case.
34
A sub-class of the nouns is constituted by different subsets of adjectives. We should
add the adjective phrase (ADJP) with a nominalized adjective in head position as pos-
sible alternative for a head or modifier function in the sentence.
35
For detailed structural descriptions of phrases in MSA, see Ditters 1992, chapters
III and IV, and for a formal description of sentence structures see Ditters forthcoming.
In Figure 1a are listed the function slots within the sn, as well as the
categories able to fill these slots. One may easily see how many different
realizations of a nominal sentence are accounted for in this diagram
where alternatives are represented within square brackets and optional
elements within parentheses.
FIGURE 1A SLOTS AND FILLERS WITHIN THE SN

Sn
Slots TOPIC COMMENT (Sadv)36
Fillers [NPnom]37 [ADJPnom] [ADVP]

[NPacc]38 [ADVP] [CCL]39
[CCL] [NPacc]
[NPnom] [PP]
Figure [PCCL]
2 [PP]
[VP]
In Figure 1b are listed the function slots within the sv, as well as the
categories able to fill these slots. One may easily see how many different
realizations of a verbal sentence are accounted for in this diagram
By means of variables as predicate-subject match (psmatch) and topic-
comment match (tcmatch) we describe regular patterns in concord and
agreement between the elements involved in a sv or sn, checking at the
same time the matching of a verb-argument-1 relation as well as the
compatibility of the elements involved in a topic-comment occurrence. In
the same way we control the occurrence of other argument realizations
of a verbal entry by means of a predicate-object match (pomatch).
36
The AGFL-formalism allows for free sequence variation at any level of description (this
means for us: at function and category level) of entries within the formal description and
processing. Therefore we do not need to list all mathematically possible realizations.
37
An alternative realization in the form of a CCL filler in a topic slot, as recorded from
Classical Arabic data (Qurn 2:184): an tasm (topic) xayrun la-kum (comment),
is, for its poor frequency in modern text data, not accounted for in our formal descrip-
tion of the Sn.
38
We include this alternative realization to account for an absolute negator-head
combination in topic position (see below 3.2).
39
In our formal description of Modern Standard Arabic we also include under the
variable name CCL the protasis (the condition posed) in conditional and hypothetical
sentences.
FIGURE 1B SLOTS AND FILLERS WITHIN THE SV

Sv
Slots PREDICATE (Sadv)
Fillers VP [ADVP]
[CCL]
[NPacc]
Figure 3 [PP]
3.1 Verb Phrase

At an early stage in the Arabic grammatical tradition, regular patterns
in syntactic behavior and the meaning of the elements involved in
verbal constructions brought grammarians to a general description
of the verbal complement structure40 and the grouping together of
verbs.41 Briefly summarizing this tradition we can say, that a finite-verb
realization (marked for perfect aspect, indicative, imperative mood or
their alternatives)42 functions as the compulsory head of a vp. In a
minimum vp configuration, an implicit agent provides values for gender,
number and person of the verbal head. Particles may precede the head
realizing optional pre-modifying (prem) functions such as a modal,
aspectual or temporal modifier, or carrying a negation value (neg). In
a maximum vp configuration, an explicit agent, objects (compl) as
well as other optional adverbial (adv pom) or peripheral (phr pom)
modifiers may follow the head. The general structure of a vp is presented
in Figure 2, where alternatives are included in square brackets and
optional elements in parentheses:
40
Within the grammatical tradition grammarians spoke about: mutaaddin (transi-
tive) and ayr mutaaddin (intransitive), further differentiated into and complemented
with: maf l bi-hi (direct object), maf l mutlaq (absolute object), maf l f-hi (object of
time or place), maf l la-hu or li-ajli-hi (object of cause or reason), maf l maa-hu (con-
comitative object), hl (circumstantial object), tamyz (object of specification). Cf. Sba-
wayhis al-Kitb. Hrn, 1:3454, 297310, 367384; see also Fleisch 1968, 177185.
41
We mention briefly: incomplete verbs (al-af l an-nqisa), the verbs of hope (af l
ar-ra), the verbs of beginning (af l a-ur), the verbs of the heart (af l al-qulb),
the verbs of praise and blame (af l al-madh wa-d -d amm), the verbs of approxima-
tion (af l al-muqraba), and the verbs of esteem (af l at-tafdl). Cf. a.o.: Ayoub 1980;
Cuvalay 1994, 1996; see also Wright, 2:4752.
42
The negation of the perfect aspect by means of the particle lam followed by a finite
verb form in the jussive or the negation particle lan governing a finite verb form in the
subjunctive.
FIGURE 2 GENERAL STRUCTURE OF THE VP43

VP
Slots (PREM) (NEG) HEAD (COMPL) (ADV POM) (PHR POM)
Fillers [ASP] [NEG PART] VERB [ADJPacc] ADVP [ADJPacc]

[MOD] [CCL] [NPacc] [NPacc]
[TMP] [NPnom] [PP] [PCCL]
[NPacc] [PP]
[PP] [VP]
[VP]
An active or passive voice value of the verbal head, in combination with

the variable transitivity44 (with values as intransitive or transitive), affects
the realization of a verbal complement structure, not only in the number
of compulsory or optional arguments, but also in the form of these
arguments as fixed in the lexically given DNA-structure of the verbal
entry concerned. With the term form we here refer to the argument
realization by means of a np with nominative case (npnom) for an
explicit agent, a np with accusative case (npacc) for direct object(s) or its
equivalent,45 and any subject or object attribute,46 a complement clause
(ccl), a co-referential vp for verbal extension,47 a pp as prepositional
object or its equivalent.48
Each finite verb occurrence in MSA, intransitive and transitive alike,
has in its complement structure a slot for an implicit agent or argument-
1 realization in the form of a suffixed pronoun (in past tense) or a
discontinuous pre- and suffixed pronoun combination (in present tense).
43
Ditters 1992, 286 ff.
44
We prefer the term transitivity, with the values: intransitive, transitive, bi-transitive
and tri-transitive to the term valency or valence since the latter is less apt to determine
non-optional core meanings.
45
A complement clause (CCL) introduced by the particle an or anna.
46
In de form of an NPacc or an indefinite adjective phrase marked for accusative case
value (ADJPindef,acc).
47
With the variable name verbal-extension we describe, in our formal grammar, the
occurrence of co-referential verb clusters, sometimes called auxiliaries, such as: incom-
plete verbs (kna or laysa, to be(come) or not to be); inceptive verbs (badaa, to begin
and the like); continuity verbs (istamarra, to continue and the like); anticipation verbs
(kda, to be on the point and the like). Cf. Haak 2006, Badawi et al. 2004, 422 ff.
48
A prepositional complement clause (PCCL) introduced by a preposition followed
by the particle an or anna.
Combined with an explicit agent in the form of a npnom, the finite verb
realization immobilizes into a 3rd person singular realization liable to
vary in gender value.
For monotransitive verbs with a direct object in the form of a npacc, in
arg-2 position, this argument will assume the arg-1 position in case of
a passive voice value of the verbal head. The same holds for bitransitive
or tritransitive verbs: the first npacc direct object will assume the arg-1
position in case of passive voice realization.
We may summarize the verbal complement structure in Modern
Standard Arabic in two tables. The first matches the features form,
function, and transitivity against the number of possible completive argu-
ments. Together with the semantic load of the verbal entry concerned,
we have the actors for the predicate-subject (psmatch) and the predicate
object matching (pomatch).
TABLE 1 VERBAL COMPLEMENT STRUCTURE (COMPLETIVE)
arg-1 arg-2 arg-3 arg-4 arg-5 arg-6
function agent v-extension object-1 object- 2 object-3 indirect object

transitivity intrans/trans intrans/trans monotrans bitrans tritrans intrans/trans
form [NPnom] [VPi] [NPacc] [ADJPindef,acc] [NPacc] [PPprep]
[PPprep] [NPacc] [ADJPindef,acc]
[CCLan/anna/inna] [PPprep]
[PCCLprep, an/anna]
The following table (Table 2) presents, in a general way, slots and

possible filler for (mainly circumstantial) elements occurring within
the complement structure of a lexically given verbal entry. As far as the
featuring is concerned, it is gradually being introduced in the form of
subscripts to the phrasal heads at syntax level.
TABLE 2 VERBAL COMPLEMENT STRUCTURE (CIRCUMSTANTIAL)

arg-7 arg-8 arg-9 arg-10 arg-11 arg-12 arg-13 arg-14
function situation absolute source goal time place instrument result

form [NPacc] [NPacc] [PPmin/an] [PPprep] [NPacc,time] [NPacc,place] [PPprep] [NPacc]
[CCLwa] [PPprep,time] [PPprep,place] [PPprep]
For the description of predicate-subject and predicate-object relations

in MSA (psmatch and pomatch), we have so far exploited the Arabic
grammatical tradition. Coming to Modern Standard Arabic, we like to
adopt, in the next section ( 3.1), Saads work (1982, 2026) in a Case
Grammar perspective for Classical Arabic.
Some of the features linked with a lexical verbal entry are listed
below:
AGREEMENT :: +human; human.
[PSMATCH :: ]49
[POMATCH :: ]
TRANSITIVITY :: intransitive; MONOTRANS; BITRANS;
TRITRANS.
MONOTRANS :: acc; prep.
BITRANS :: acc2; accprep; prep2.
TRITRANS :: acc3; acc2prep; accprep2; prep3.
3.2 Noun Phrase

We distinguish within the np one compulsory and several optional
functions. Only a noun (or a nominalized adjective) can realize the
compulsory head function. Among the optional functions we mention:
negator (neg), a mutually exclusive predeterminer (predet) and post-
determiner (postdet), postmodifier (pom), and, for some head rea-
lizations, a complement (compl).50 In the following general structure
of the np (Figure 3), optional functions are included in parentheses and
mutually exclusive functions in square brackets. 51
Relationships and dependencies within the np are specified and
controlled by features and feature-condition rules. In a neg-head
combination it is the negative particle l that imposes the unique
occurrence of an indefinite common noun in head-position provided
with a definite version of the accusative case value. Personal pronouns and
proper nouns, interrogative and indefinite pronouns (of the subset-type:
man and m) in head-position, are inherently marked as definite and
indefinite respectively, thus they may only occur with an optional pom.
49
These rules will be discussed later.
50
For example: a masdar in head function with a subjective complement in the slot
for postdeterminer (NPgen) and an objective complement in the slot for COMPL (NPacc),
both realizations in the form of a suffixed pronoun thus realizing a postdet and a
compl function: hubb--h = love-mine-her my love for her.
51
In Figure 3 we have included neg among the reciprocally exclusive function rea-
lizations. However, here we underline the incompatibility of a predet and a postdet
in a np. Cfr. the examples in Cantarino 1974, I:114; 1975, II:220222), Badawi et al.,
2004, 464466. El-Ayoubi et al. (20012003, I:12) do not mention a neg; neither in the
Vorfeld der Nominalgruppe, nor in the Nachfeld der Nominalgruppe.
FIGURE 3 GENERAL STRUCTURE OF THE NP52

NP
Slots ([NEG]) ([PREDET]) HEAD ([POSTDET]) (POM) (COMPL)
Fillers [NEG PARTl] [ART] NOUN [NPgen] [NP] [PP]

[ADJP] [NPacc]
[PP]
[REL CL]
Only (de)verbal nouns (those marked with acc, or acc2 for transitivity) as
well as some nominalized adjectives in head-position may occur with a
compl.53 Other relationships and dependencies concerning definiteness,
person, gender, number, and case, will be discussed below.
3.2.1 Definiteness
Elements of the noun subsets proper nouns and personal pronouns have
the value definite for definiteness with consequences for post-modifying
elements at constituent and sentence level. Elements of the noun-subsets
indefinite pronouns and interrogative pronouns have the value indefinite.
Common nouns, (de)verbal nouns, and nominalized adjectives receive
their value for definiteness by the occurrence of a predet or a postdet
modifier.
3.2.2 Person
Only elements of the noun subset personal pronouns vary in their value
for person ranging from first, second to third, while varying in number
and case feature values. All elements of other noun subsets bear the fixed
value third for the feature person.
3.2.3 Gender
Occurrences of the subsets proper nouns, personal pronouns, and common
nouns possess a lexically given value for the feature gender, whether this
52
53
For a more detailed discussion about adjectives in the context of the ellipsis of the
head of an NP and of nominalized adjectives see Ditters (forthcoming).
is marked morphologically by a default null-value, by one of the femi-

nine markers or not. Elements of the subsets indefinite and interrogative
pronouns bear the default gender value masculine, but may, in retro-
spective and supported by the unification-principle of the agfl-
formalism, receive an inherited feminine value provided by the context
in which they occur beyond sentence level. At syntax level we may have
to deal with the interference of semantic feature values such as animate,
concrete, human, volition etc., here mainly related to as: concord and
agreement phenomena.
3.2.4 Number
The singular and plural realization of elements of the subsets proper
nouns, personal pronouns, and common nouns are given lexically. A
dual number value realization is normally the result of the application
of a grammar-rule except for the personal pronouns. Only infinitives or
masdars always have a singular number value. Apparent evidence to the
contrary is caused by the exploitation of masdars for the creation of an
abstract lexicon.
3.2.5 Case
Case is a syntax dependent application feature. Its values (nom, gen, and
acc) are related to the function the element concerned has at phrasal or
higher levels of linguistic description. A null-value for case, any other
ambiguity concerning diptotic np-head occurrences, and gender issues
for a first person realization have to be disambiguated beyond phrasal
and sentence levels of description.
3.2.6 Trash
Earlier we mentioned feature-clusters. In some of them nps are involved
(concord, agreement, ps-match, po-match, and tc-match). In this section
we discuss the occurrence of a noun in the head-position of a phrase at
syntax level. Feature values of such a noun should be capable to match
with elements occurring within the same constituent or with feature
values of heads of other constituents within a given context.
Summarizing some conclusions at this point, we list the following
formal rules connected with a lexical noun entry at syntax level:
[AGREEMENT :: +human; human].54

[CONCORD :: full; partial.]
[PSMATCH :: ]
[POMATCH :: ]
[TCMATCH :: ]
CASE :: accusative; genitive; nominative; null.
DEFINITENESS :: definite; indefinite.
HEADREALIZATION :: demonstrative; personal; proper; relative;
indefinite; interrogative; adjective; common;
elative; numeral; quantifier; verbal.
3.3 Particle Phrase

As far as the occurrence of particles at syntax level is concerned: earlier we
distinguished between an adverb phrase (advp), a prepositional phrase
(pp), an interjectional phrase (ip), and a conjunctive phrase (cp).
3.3.1 Adverb Phrase

At the morphological level, we sub-categorized the adverbs into: invari-
able, as far as case marking is concerned, and bound or free forms, as far as
the orthographic representation is concerned. One feature of the bound
forms involves the temporal aspect of a verb realization itself marked by
a non-perfect or mood tense (sa, directly linked to a finite verb form,
then suggests a future temporal aspect).55 The interrogative a affects
the whole of a succeeding verbal or nominal sentence. The affirmative
la mainly occurs, in Modern Standard Arabic, as a premodifier to the
particle qad in an advp. At syntax level, the bound and free forms are
further differentiated into, among others, time, degree and manner
adverbials, attached to verbs and adjectives as well. Finally, we have to
distinguish between adverbs as head of an advp, on the one hand, and
other realizations of an adverbial function at phrasal or sentence level
such as pp, a npacc or a pccl at the other.
Formally, an adverb phrase (advp) can be described as a, at syntax
level occurring, constituent that, in its minimal configuration, has an
adverb in the compulsory head function. Optional functions in an advp
can be labeled as a pre- or postmodifier of the head. A premodifier may,
for example, figure as a discontinuous negative particle to adverbials as
54
These rules will be discussed later.
55
The so-called TMA (tense, mood and aspect) auxiliary particles have been discussed
by Kinberg (2001). See also Eisele 2006.
qattu never or faqat no more,56 or in the form of a preposition with

an head as in min qablu previously. An optional postmodifier can be
realized by a PP.57
FIGURE 4 GENERAL STRUCTURE OF THE ADVP 58
ADVP
Slots (PREM) HEAD (POM)
Fillers [PART] [ADV] [PART]

[PREP] [NPacc]
For the adverb we suggest the following combination of feature names

with corresponding values:59
ADVERB :: ARGUMENTATION; CONFIRMATION;
DISCOURSE ; FOCUS; FREQUENCE;
interrogative; LOCAL; MEASURE;
MODALITY; SEQUENCE; TEMPORAL.
ARGUMENTATION :: evaluative; comparative; CONCLUSIVE.
CONCLUSIVE :: positive; negative.
CONFIRMATION :: certain; affirmative; limitative.
DISCOURSE :: greetings; wishes; emphasis.
FOCUS :: specifying; inclusive; exclusive; cumulative.
FREQUENCE :: number; repetition; iteration; coinciding.
LOCAL :: position; dimension; direction.
MEASURE :: degree; quantity.
MODALITY :: expression; noun; adjective; nisbe.
TEMPORAL :: past; neutral; future; SUFFIXED; relative.
SEQUENCE :: coinciding; preceding; succeeding.
SUFFIXED :: idin; dka; pronoun.
3.3.2 Prepositional Phrase

Within a prepositional phrase we distinguish two compulsory func-
tions: a prepositional complement (pcompl) and a linker (plink) that
connects that complement to the next higher level of linguistic descrip-
tion. The complement function can be realized by: an adverb phrase
56
Cf. El-Ayoubi et al. 2003, I:2, 275460, and 406. Badawi et al. 2004, 161174.
57
In the discussion about the NP we already mentioned the use of an accusative NP
(NPacc) for adverbial purposes.
58
59
Cf. El-Ayoubi et al. 2004, I:2, 284 ff., and Wright 1974, I:282 ff.
(advp); a complement clause (ccl) introduced by the particle an or

anna (cclan|anna); a noun phrase with genitive case value (npgen); or
a verb phrase with a head marked for subjunctive mood (vpsubj). The
prepositional linker function is usually realized by an element of the
particle-subset of prepositions. However, there are also expanded con-
structions in which a preposition, followed by an infinitive or abstract
noun, realizes the header function,60 and prepositional clusters in which
an infinitive, marked with an indefinite accusative case value (npacc) and
followed by a preposition realizes the linker function.61
FIGURE 5 GENERAL STRUCTURE OF THE PP62

PP
Slots HEADER COMPL
Fillers [PREP] [ADVP]

[EXPAND-P] [CCLan|anna]
[P-CLUSTER] [NPgen]
[VPsubj]63
A good basis for a discussion about semantic features in the next section
certainly is the sub-categorization of prepositions into: time, place, and
ideal (or manner) features (Wright, 1974: II:129). With corpus based
evidence, forwarded by Cantarino (1974, II: 262 ff.) and El-Ayoubi
et al. (2004, I, 2:466 ff.) we are able to further differentiate this sub-
categorization of prepositions into features values:64
PREP :: IDEAL; PLACE; TIME.
IDEAL :: adversative; benefactive; causal; COINCIDING;
comparative; content; discourse; explicative; final;
HYPOTHETIC; instrumental; measure; modal;
partitive; possessive; specification; substitution.
COINCIDING :: comitative; simultaneous.
HYPOTHETIC :: concessive; conditional.
PLACE :: destination; direction; local; partitive; position.
TIME :: direction; partitive; temporal.
60
Cf. El-Ayoubi et al. 2004, I:2,574.
61
Cf. El-Ayoubi et al. 2004, I:2,584 ff.
62
63
Not to confuse with the inflectional phrase as in the Government and Binding
approach (cf. Fassi Fehri 1993, 16 ff.).
64
See also Badawi et al. 2004, 167 ff.
3.3.3 Interjectional phrase or clauses

Earlier ( 2.3.3) we differentiated the interjectional phrase (ip) into:
vocatives; exclamations;65 frozen expressions, proverbs and greetings. Two
of them, vocatives and exclamations, lend themselves easily for a general
structure description (see Figure 6).
Other optional ips ( f-expressions, proverbs, greetings, and insults) may
provisionally receive, in our formal description of Modern Standard
Arabic, the label of a phrasal or sentence adverbial since the syntactic
structure of such an insertion can easily be described by means of the
basic general syntactic patterns we already suggested. However, the
semantic value of their occurrence in a given context has still to be
evaluated and described.66
FIGURE 6 GENERAL STRUCTURE OF THE IP

IP
Slots (PREM) I-HEAD (I-COMPL) (TAIL)
Fillers [PART] [INTERJ] [NPnom] [PART]

[NPacc] [NPacc]
[Sv] [PPil|bi|al]
[PPli]
[Sadv]
A few comments may make this scheme more transparent. The line
of slots tells us about the compulsory function of an ip head (i-head)
with an optional tail in case of a discontinuous realization like w-h,
be it for Modern Standard Arabic a bit archaic, as in wzaydh Woe
upon Zayd! (Badawi et al., 2004:37). It further mentions the optional
occurrence of a premodifier (prem) or a complement (i-compl) of
the i-head. In the filler-section alternative realizations are listed, also
comprising a possible sv-sadv combination as man yaqtul yuqtal Who
kills will die killed. For this reason we included the term clause in the
heading of this subsection.
65
Cf. Firanescu 2007, 7981.
66
Cf. Bergman 2007, II:136137.
3.3.4 Conjunctive phrase (cp) or Clause (ccl)

Following the Wright tradition, at the morphological level, we made a
distinction between a separable and an inseparable occurrence of con-
junctions, as well as between connective and conditional conjunctive par-
ticles (Wright, 1974, I:290294). At syntax level, however, we also prefer
to distinguish between coordinating and subordinating conjunctions.
3.3.4.1 Coordination
The compulsory head-function in the cp is realized by an element of
the subset coordinators of the conjunctions. An example of an optional
premodifier would be the narrative connector wa preceeding the
adversative coordinator lkin but (Badawi et al., 2004:555). The com-
pulsory complement function may be realized by a variety of alterna-
tives (Figure 7).
FIGURE 7 GENERAL STRUCTURE OF THE CP
CP
Slots (PREM) C-HEAD67 COMPL
Fillers [PART] CONJ [NP]

[VP]
[ADJP]
[ADVP]
[Sn]
[Sv]
3.3.4.2 Subordination
The compulsory head function in a complement clause is realized by
an element of the subset subordinators, which may contain a null value,
for example, in caseof the occurrence of a purposive li followed by a
verb with a subjunctive mood value. In our formal grammar of Modern
Standard Arabic, the compulsory complement function may vary
between a sn and a sv.
67
The conjunctive head may be empty is case of asyndetic coordination (Badawi
et al. 2004, 539 ff.
FIGURE 8 GENERAL STRUCTURE OF THE CCL

CCL
Slots (PREM) HEAD COMPL
Fillers [PREP]68 SUB [Sn]

[PART] [Sv]
The variables and features we like to retain for the conjunctive phrase or
clause are:
CONJUNCTION :: COORDINATING ; SUBORDINATING.
COORDINATING :: cumulative; SELECTIVE.
SELECTIVE :: adversative; alternative; consecutive;
disjunctive; exclusive; explicative; inclusive;
restrictive; successive.
SUBORDINATING :: circumstantial; consequential; SINGLE-
CLAUSE; DOUBLE-CLAUSE.
SINGLE-CLAUSE :: comparative; complemental; purposive;
locative; temporal.
DOUBLE-CLAUSE :: CONDITIONAL; temporal.
CONDITIONAL :: possible; real; unreal.
4. Semantics
We discussed so far feature variables with distinct values at the morpho-

logical and syntactic level of description, variables such as: aspect, tense,
definiteness, gender, number, person, and case. Moreover, with feature-
clusters such as concord and agreement (involving: human, person,
gender, and number), we are able to describe regularities between a
predicate and an explicit agent (ps-match), between a predicate and
possibly realized complements (po-match) in a sv, as well as between
topic and comment realizations (tc-match) in a sn.
As far as our semantic level of description is concerned, we earlier
insisted to remain, in this paper, within the limits of static semantics as
used in almost all other applications of NLP.69 Table 3 summarizes our
68
In that case we can speak of a prepositional complement clause (PCCL).
69
We refer to the general domains of machine translation and information retrieval,
still using the basics on static semantics of Yngve (1966). See also Koster (2004).
TABLE 3
Levels Features
morphology morphological
morpho-syntactic
syntax syntactical
syntactico-semantic
semantics0 semantical0
static-semantical
semantics1 semantical1
dynamic-semantical
semantics 2 semantical2
etc.
working space ending at level-zero (semantics0) of static semantics. The

description of features at the levels listed in the squared box remains to
be done.
Thus, a comprehensive assignment of semantic features to all lexical
entries, which can occur as head element of one of the different phrasal
constituents, does not (yet) exist for automated analysis purposes of
Modern Standard Arabic. That may not be necessary when a heuristic
feature assignment for a ps-match, a po-match or a tc-match could
already meet our disambiguation objectives.
The more or less detailed sub-categorization of features for headers
in an advp and a pp ( 3.3.12) is a good example of our hypothesis.
This feature sub-categorization supplies such an advp with a specific
semantic load in the context of its occurrence. For the header of a pp, this
sub-categorization presupposes a matching with some corresponding
feature values of the head realization of the prepositional complement
for a grammatically and semantically correct analysis result. Therefore:
if the head-element of a prepositional complement shares the maximum
number of pertinent features with the prepositional header, alternative
analysis results should and could be discarded.70 If a noun like bustn
possesses a feature hinting at an open-air outdoors location we could
easily match the local feature values of the header and the prepositional
70
The fine-tuning of this matching occurs in the cyclic process of testing the descrip-
tion on new text data. Moreover, this matching produces optimal results when unifica-
tion provides contextual information. A sequence as min tahtihi remains ambiguous as
long as the anaphor hi is not disambiguated in context.
complement and interpret the sequence f l-bustni in the garden as an

attributive or adverbial adjunct with local value. As a matter of fact, we also
have to provide individual lexical elements of verb and noun categories
with a minimum set of pertinent feature names and feature values.
4.1 Verbs
In our formal description of Modern Standard Arabic for automated
analysis purposes, individual lexical verbal elements have to provide
information about:
compulsory and optional completives and circumstantials within the

complement structure of a given verbal entry (transitivity);
the combination of a verbal entry with elements of other word
categories to denote a particular meaning of such a verbal entry
(collocation);71
minimum qualifications for possible implied or explicit agents and
completives (in terms of case-roles and syntactic realization).
4.1.1 Transitivity
Within the framework of the diinar-project72 verbal entries have been
compiled according to the number and the syntactic realization of their
arguments. In this way the verbal lexicon is first split up in intransitive and
transitive verbs. The transitive verbs are then further divided into: mono-,
bi- or tri-transitive verbs. Finally, these verbal entries can be divided
into sub-categories governing a direct object(s) (npacc or ccl), and/or a
prepositional object (pp or pccl). A further differentiation concerns the
lexical realization of the preposition (ppli, ppmin, ppan etc.).
What is still missing in this approach is the semantic dimension as,
for instance, Levin (1993) did for English.73 She compiled 49 verb classes
with sub-categories. The members of each sub-class share the general
meaning of the class as well as a number of syntactic characteristics. So,
the members of the sub-class 10.2 of banish verbs (Levin 1993, 123):74
71
Cf. Hoogland 1993; El-Gemei 2006, 434439.
72
Information about the DIINAR lexical database can be found at: http://catalog.
elra.info. Information about DIINAR-MBC can be found at: http://sites.unin-lyon2.fr/
langues_promodiinar/accueil.htm.
73
See also Levin & Rappaport Hovav 1996.
74
These members are: banish, deport, evacuate, expel, extradite, recall, and remove.
relate to the removal of an entity, typically a person from a location. The

location argument is expressed in a prepositional phrase headed by the
preposition from. [. . . .] Unlike the remove verbs, these verbs allow to
phrases as well as from phrases, though not simultaneously.
The application of such an approach for the description of verbal entries
in Modern Standard Arabic is given in Table 4 where square brackets
point to mutually exclusive arguments and parentheses to optional
occurrences:75
TABLE 4
arg-4 arg-3.2 arg-3.1 arg-2 arg-1 entry
manner destination source object agent
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos76)] NPacc,+human (NPnom,+human)77 " #

(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) $
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) % $
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) % $#
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) &' (
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) ) * (
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) +) ,-
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human78 (NPnom,+human) . /
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) 0 ) (
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) 12
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) ) 12
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) 3 4'
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) + 56
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human| (NPnom,+human) 7 8 9
PPbi,+human
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) 8 :
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) ; 4<
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human)
: =
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) &: =
(PPbi) [(PPil|li,loc|pos)] [(PPan|min,loc|pos)] NPacc,+human (NPnom,+human) +>=
75
A vertical bar | here denotes alternatives in subscript variables and features.
76
The feature alternatives loc and pos comprise a physical location as well as a social
function.
77
An optional explicit agent has the same feature values as the implied agent
78
The object of this entry is typically realized by the collective nouns army, troops
or alternatives.
4.1.2 Collocation
At this point we question the way of featuring specific verbal head-advp,
verbal head-np, and verbal head-pp combinations resulting in a, for the
combination, specific semantic value. The problem is that the matching
of the individual feature values of the elements involved does not neces-
sarily point to the semantic value of the combination. A cheap solution
will be to list this kind of combinations as they are in the lexicon.
4.1.3 Case-roles and realization

According to Abdul-Raof (2006, 343), the use of case-roles may
adequately describe deep structure relations between noun phrases and
the verb. If this quote is equivalent to: feature matching between the
heads of a np and a vp (occurring as comment in a sn), or between the
heads of a vp and nps (occurring as dependent or independent nps, and
realizing a completive and/or a circumstantial function in a sv), we like to
explore the potentiality of Case Grammar for our purposes.
Winograd (1983, 311 ff.; 623624) describes the evolution of a case-
role based description of semantic relations within syntactic structures,
and lists some 65 case roles. For Arabic, we refer to Abdul-Raof (2006)
and LeTourneau (2006) on Case Roles and Case Theory respectively, but
we will exploit Saads monograph (1982) for further featuring in a CG
framework.
With 12 roles Saad easily represented our 14 columns in Table 1
and 2 in case grammar (CG) terms, as is shown in Table 5a and 5b
below.79 In the columns 5 (range) and 12 (comitative), we recognize the
circumstantials: al-maf l al-mutlaq (cognate object) and al-maf l
maahu (accompanying object):
TABLE 5A CASE ROLES IN MSA

arguments 1 2 3 4 5 6
case role agent instrument patient/ target range source

object80
abbreviation A I PO Ta R S
79
The row abbreviation has been inserted to introduce names of variables within
a representation of the complement structure of verbal complement structure in case
frame terms.
80
We prefer to differentiate between a +human object (= patient) and a human
object.
TABLE 5A (cont.)
arguments 1 2 3 4 5 6
verb class agentives agentives agentives; agentives; verbs of

verbs of a verbs of a change,
mental, mental, transformation,
physiological, physiological, experience,
emotional state emotional state action in
or process or process motion
or time
features +animate81 -animate
form NPnom NPacc|PP NPnom|NPacc NPnom|NPacc NPacc AJDPnom|
|PCL82 |PCL NPnom|PP
TABLE 5B CASE ROLES IN MSA

7 8 9 10 11 12
case role goal place time manner beneficiary comitative

abbreviation G P Ti M B C
verb class verbs of change,
experience,
transformation,
action in motion
or time
features local; time manner;
position degree
form PP PP NPacc|PP NPacc|PP PP NPacc|PP
With a paragraph on co-referential case roles, Abdul-Raof (2006,

345) gives a hint for the fine-tuning of verbal sub-categorization in
Modern Standard Arabic. He lists five cases wherein the role of Agent
is co-referential with the role of Object (= Saads patient), Source, Goal,
Experiencer (= Saads patient), and Beneficiary respectively.83
The verbal entries of Table 4 share an optional explicit subject84
(Agent) in the form of a np, marked for a syntactic nominative case
value and a semantic +human value (npnom,+human). Moreover, they share
81
Saads (1982, p. 20) suggestion for an alternative realization form (animate) to
describe astronomical bodies and semi-autonomous machines, gets a more elegant
foundation in Dichy (2005).
82
A vertical bar | here denotes alternatives at form level.
83
The co-referentiability between an Agent and more than a single other case role
remains to be studied.
84
An implicit subject of a finite verb form is always present or is neutralized and
overruled by an explicit agent.
an compulsory direct object (Patient) in the form of a np, syntactically

marked for accusative case value and semantically as +human
(npacc,+human). Above that, they share two optional, mutually exclusive
local circumstantials, expressing the origin (Source) and the destination
(Goal) of the action being or been performed (ppan|min,loc|pos or ppil,loc|pos).
Sometimes there is even mention of an optional circumstantial in the
form of a ppbi, marking the manner (Manner) in which the action is
performed (bi lquwa by force). Finally, one could witness the occurrence
of an optional circumstantial in the form of an advp or pp marking the
moment (Time) of the action performed (advptime or pptime).
Such a description can be represented by means of a frame with a
line for the lexical entry concerned, abbreviations for the cases with
realization forms between curled brackets ({}), while parentheses (()) and
square brackets ([]) denote optional occurrence and mutual exclusion
respectively. A vertical bar | separates alternative realization forms.
For our example of the sub-class of banish verbs, this results in the
following case frame for the members of this sub-category:
[__ A({NPnom,+human}), P{NPacc,+human}, [(S{PPan|min,loc|pos})], [(G{PPil,loc|pos})],
(M{PPbi}), (I{PPbi}), (Ti{ADVBtime|PPtime})]
During the analysis process of Arabic text data, a lexical verbal entry will
be recognized as far as its consonantal root and stem is concerned, while
the information about verbtype, derivation, aspect, tense, voice, person,
gender, and number are being stored. In the meantime, information
about the case frame of the verbal entry will be loaded from the lexicon
and matched against the string(s) following the finite verb form.
The next step in our description seems to comprise the drafting of
case frames for the Arabic equivalents of members of the other sub-
classes of the class of remove verbs. This predicts a cyclic process of the
drafting and testing of case frames for the Arabic equivalents of the
members of all the other 48 classes Levin (1993) listed for English.
We like to retain the following case roles in the description of the
complement structure of verbal entries for agent and object matching
with different kinds of completives and circumstantials: agent (a),
instrument (i), patient (po), target (ta), range (r), source (s), goal (g),
place (p), time (ti), manner (m), beneficiary (b), and comitative (c).
We like to retain the idea of pairing case roles by co-referencing, as
Abdul-Raof (2006, 345) gives with: Ai=Oi,85 Ai=Si, Ai=Gi, Ai=Ei, and
85
Abdul-Roaf only lists 8 case roles and does not employ the term Patient. Therefore
Ai=Bi. It certainly may refine the description of elements of different

verb classes and the complementstructure of verbal entries.
ACTORS :: MONO; PAIRS; TRIADES.
MONO :: agent; instrument; object; patient; target;
range; source; goal: place; time;
manner; BENEFICIARY; comitative.
BENEFICIARY :: client; recipient.
PAIRS :: A=PO; A=S; A=G; A=E; A=B.
[TRIADES :: ].
4.2 Nouns
At the end of the introduction we stated that the formalism for automated
processing is suited for describing morphological and syntactic struc-
ture, as well as finite semantics in terms of animate, concrete,
human, volition and many other. At the morphological and syntactic
level we already discussed a number of features attached to elements
of the category noun. Besides that, the differentiation of relevant
prepositional features in time, place, and ideal presupposes a matching
with corresponding features attached to the head of a prepositional
complement, in casu a noun. The question is: how to make an inventory
of the minimum number of required semantic noun features, and how
to set up the organization of this inventory.
A frequently used approach consists in the drafting of a conceptual
semantic tree. Hartmann (1974, 169) used Moravcsiks (1970) conceptual
tree starting with the node countable for her Transformational Gene-
rative Grammar of Arabic. Muhtaseb (1988, 65 ff.) developed (far too)
detailed conceptual trees describing the subject and action hierarchy for
his Natural Language Understanding System (NALUS). More recently,
Dichy (2005) proposed a conceptual tree of three layers starting with
concrete, using two other basic features: animate and human. This
tree generates the lexical categories: person, animal, perceptible
entity, and idea or non-physical experience. By a transfer of some
features, characteristic for +animate and +human entities, to animate
and human entities by assignment of specific values as intention,
+volition, or +motion, Dichy also generates the derived categories:
it is not yet clear how to harmonize his co-referential pairs Agent = Object (Ai=POi) and
Agent = Experiencer (Ai=Ei) with the set of 12 case roles of Saad.
pseudo-animated perceptible entity, collectivity, and pseudo-

animated perceptible entity.
We like to retain the following base set of semantic features (BASEF)
for elements of the category noun: concrete, animate, and human. For
the moment, we also like to incorporate Dichys sub-features: intention,
motion, and volition. Moreover, for elements of the category noun
we have to include features required in the matching with features of
elements of the categories verb and particle. In 4.1 we saw that those
features are required for predicate-subject (psmatch) and predicate-object
(pomatch) matching. In this subsection they are required in the topic-
comment (tcmatch) matching. In the following section these features
are needed in a preposition-complement ( pcmatch) matching.
In conclusion, the human value of a noun plays a role in agreement86
and concord87 phenomena involving, in the Sv, the elements of the pre-
dicate-explicit agent ( psmatch) matching or the elements of the predicate-
patient/object ( pomatch)88 matching, and, in the sn, the elements of the
topic-comment (tcmatch)89 matching.
Below we list our features rules for elements of the category noun,
easily expandable with other variables and values:
BASEF :: concrete; animate; human; place; time.
SUBF :: intention; motion; volition.
4.3 Particles
Concerning the vocatives we would like to discriminate at the semantic
level between occurrences related to +human and +animate, and human
entities. We prefer to further differentiate the exclamations into sub-
categories expressing: agreement, commands, dissent, enthusiasm, sorrow,
surprise, presentations, and quantitatives.
For our purposes it is meaningful to divide the group of adverbs by
means of a variable advtype into elements expressing: manner, place,
86
If an explicit agent (arg-1) is marked for +human, there is full gender matching
between predicate and subject, otherwise (if human), there is partial gender matching,
i.e. the verbal entry matches gender values with a subject marked for singular or dual
number, but it receives feminine gender value with plural subjects.
87
A NPnom,+human in the topic position of a Sn triggers full gender and number matching
with elements, liable for such a matching, in comment position.
88
In the case of a NPacc patient or object of a finite verb form marked for passive
voice.
89
In the case of a ADJPnom, NPnom or VP occurrence in comment position.
time, quantity, affirmation, negation, or interrogation.90 Another variable

may function as a basket for elements expressing: restriction, doubt,
wish, or exclamation.91
The semantic feature exclamation also may occur as one of the values
of the variable inttype of the interjections. Other values are: attention,
confirmation, support, and sociality. Values for the variable preptype and
conjtype can be borrowed from their differentiation at syntax level. The
preptype features play a role in the preposition-complement ( pcmatch)
matching.
INTYPE :: VOCATIVE; EXCLAMATION; f-expressions;
proverbs; greetings; insults.
VOCATIVE :: +human ; +animate,-human.
EXCLAMATION :: agreement; commands; dissent; enthusiasm;
presentations; quantitative;
sorrow; surprise; warning; wishes.
ADVTYPE :: DECLARATIVE; QUES; manner; place; quantity;
time.
DECLARATIVE :: affirmation; negation.
QUES :: doubt; exclamation; interrogation; restriction;
wish.
CONJTYPE :: cumulative.
PREPTYPE :: instrument; measure; place; time.
5. Conclusion
We discussed variables and values of features of importance for the dis-

ambiguation of analysis results from the automated processing of con-
temporary Arabic text data. We considered three levels of description.
Morphological feature variables and values serve as input for the syntax
level. Feature variables and values from the semantic level help to refine
the description of language structure at syntax level to arrive at a single,
most probable, syntactically as well as semantically correct, analysis result.
At syntax level we presented structural descriptions of sentence types
and basic sentence phrasal constituents in Modern Standard Arabic.
The general complement structure including optional and compulsory
90
We are here dealing with the interrogative adverbs hal and a marking Yes-No
questions.
91
A number of refinements to this description has been effectuated on the basis of
El-Ayoubi et al. 2003, 275460.
completives and circumstantials of a verbal entry were summarized in

the form of tables.
At the semantic level we suggested a description of Arabic verbal
entries on the basis of the number and the syntactic realization of their
arguments as well as a correspondence in meaning la Levin (1993) for
English. Arabic equivalents of banish verbs served as an example. Then
we started to translate the accumulated information into Case Grammar
terms using Saads (1983) twelve cases for the description of the role of
arguments within the possible complement structure of a verbal entry.
We started to complete the description with information about type of
verb classes, realization form and final semantic features values of the
complements. This exercise resulted in the presentation of a case frame
for the Arabic banish verbs.
In the next parts of this semantic section we discussed a base set of
semantic features for elements of the noun and particle categories to be
used together with case frames for Arabic verbal entries (sv), as well as
for topics complemented by comment realizations (sn). A draft for pre-
dicate-explicit agent, predicate-object, topic-comment, and preposition-
complement matching has been presented.
What remains to be done is:
to test the structural descriptions of sentence types and phrasal

constituents against new contemporary Arabic text data;
to apply Levins (1993) concept for verb categorization to the Arabic
verbal lexicon;
to test the results from this application against new contemporary
Arabic text data;
to translate the corrected results from that application into a
refinement of the minimum set of semantic features for elements
of the noun and verb categories;
to fill in the gaps and to expand the slots in figure 9.
FIGURE 9
concrete +
animated +
human +
adult +
male +
determinated +
Ahmad man Fayruz woman
6. References
Abdul-Raof, Hussein. 2006. Case Roles. In Versteegh, Kees et al., eds. EALL, vol. 1,
343347.
Ayoub, Georgine. 1980. Af lu l-qulb en arabe standard: lments pour une analyse.
In Bohas, Georges, ed. tudes Arabes: Analyses Thories. 1980, 1, 154.
Badawi, Elsaid, Michael Carter and Adrian Gully. 2004. Modern Written Arabic: A
Comprehensive Grammar. London and New York: Routledge.
Bahloul, Maher. 2006. Agreement. In Versteegh, Kees et al., eds. EALL, vol. 1, 4348.
. 2006. Copula. In Versteegh, Kees et al., eds. EALL, vol. 1, 506511.
Bergman, Elizabeth. 2007. Frozen Expressions. In Versteegh, Kees et al., eds. EALL,
vol. 2, 136137.
Bohas, Georges and Abderrahim Saguer. 2007. The Explanation of Homonymy in the
Lexicon of Arabic. In Ditters, Everhard and Harald Motzki, eds. Approaches to Arabic
Linguistics Presented to Kees Versteegh on the Occasion of his Sixtieth Birthday. Leiden:
E.J. Brill, 255289.
Cohen, David. 1970. Essai dune analyse automatique de larabe. In Cohen, David, ed.
tudes de linguistique smitique et arabe. The Hague: Mouton, 4978.
Cuvalay, Martine. 1994. Auxiliary Verbs in Arabic. In Elisabeth Engberg-Pedersen,
Lisbeth Falster Jakobsen, and Lone Schack Rasmussen, eds. Function and Expression
in Functional Grammar. Berlin: Mouton de Gruyter, 265283.
Cuvalay-Haak, Martine. 1996. The Arabic Verb: A Functional Grammar Approach to
Verbal Expression in Classical and Modern Arabic. PhD-thesis. Amsterdam: University
of Amsterdam.
Dichy, Joseph. 2005. Spcificateurs engendrs par les traits [anim], [humain],
[concret] et structures darguments en arabe et en franais. In Bjoint, Henri and
Franois Maniez, eds. De la msure dans les termes. Hommage Philippe Thoiron.
Travaux CRTT, Lyon: Presses Universitaires de Lyon, 151181.
Ditters, Everhard. 1992. A Formal Approach to Arabic Syntax: the Noun Phrase and the
Verb Phrase. PhD thesis Nijmegen University. Nijmegen: Luxor.
. 2001. A Formal Grammar for the Description of Sentence Structure in Modern
Standard Arabic. In Kaufman, Morgan, ed. Workshop Proceedings: Arabic Language
Processing: Status and Prospects, CRNSUniversit des Sciences Sociales Toulouse,
France, 3137.
. 2003a. Non-coinciding Phrasal Heads. In Chu, Hsing-Wei, Jos Ferrer, Tran
Nguyen, and Yuongquan Yu, eds. Proceedings of the Joint International Conference
on Computer, Communication and Control Technologies (CCCT03) and the 9th

International Conference on Information Systems Analysis and Synthesis (ISAS03).
Orlando, Florida, USA: International Institute of Informatics and Systemics, 5156.
. 2003b. An AGFL for the Description of Non-coinciding Phrasal Heads. In Callaos,
Nagib, William Lesso, Belkis Sanchez, and Elizabeth Hansen, eds. Proceedings of the
7th World Multiconference on Systemics, Cybernetics and Informatics. Orlando, USA:
International Institute of Informatics and Systemics, 107112.
. 2006. Computational Linguistics. In Versteegh, Kees et al., eds. EALL, vol. 1,
455465.
and Jan Hoogland. 2006. Corpus Linguistics. In Versteegh, Kees et al., eds. EALL,
vol. 1, 195201.
and Cornelis Koster. 2004. Transducing Arabic Phrases into Head-Modifier (HM)
Pairs for Arabic Information Retrieval. In Nikkhou, Maktab, ed. Proceedings of the
NEMLAR 2004 International Conference on Arabic Language Resources and Tools,
Cairo, Egypt: Magnet Creative Communications, 148154.
Eisele, John C. 2006. Aspect. In Versteegh, Kees et al., eds. EALL, vol. 1, 455465.
El-Gemei, Dalal Mahmoud. 2006. Collocation. In Versteegh, Kees et al., eds. EALL,
vol. 1, 434439.
Elzeiny, Nagwa. 2007. Greetings. In Versteegh, Kees et al., eds. EALL, vol. 2, 202207.
Fassi Fehri, Abdelkader. 1993. Issues in the Structure of Arabic Clauses and Words. Dor-
drecht : Kluwer Academic Publishers.
Fillmore, Charles. 1968. The Case for Case. In Bach, Emmon and Robert Harms, eds.
Universals in Linguistic Theory. New York: Holt, Rinehart and Winston Inc. 188.
Firanescu, Daniela. 2007. Exclamations. In Versteegh, Kees et al., eds. EALL, vol. 2,
7981.
Fleisch, Henri. 1961. Trait de philologie arabe, vol. 1: Prliminaires, phontique, mor-
phologie nominale. Imprimerie Catholique: Beyrouth, 5665.
. 1968. Larabe classique : Esquisse dune structure linguistique. Recherches publis
sous la direction de lInstitut de Lettres Orientales de Beyrouth, deuxime srie:
Langue et Littrature Arabes, volume 5. Beyrouth: Dar El-Machreq.
Haak, Martine. 2006. Auxiliary. In Versteegh, Kees et al., eds. EALL, vol. 1, pp.
216221.
Hartmann, Regina. 1974. Untersuchungen zur Syntax der Arabischen Schriftsprache.
Wiesbaden: Harrassowitz.
Hoogland, Jan. 1993. Collocation in Arabic (MSA) and the treatment of collocations
in Arabic dictionaries. In Dvnyi, Kinga, Tams Ivnyi, and Avihai Shivtiel, eds.
Proceedings of the Colloquium on Arabic Lexicology and Lexicography. Budapest: Etvs
Lornd University and Csoma de Krs Society, 7593.
Kinberg, Naphtali. 2001. Studies in the Linguistic Structure of Classical Arabic. Kinberg,
Leah and Kees Versteegh, eds. Leiden: E.J. Brill.
Khim, Alain. 2006. Adjective Phrase. In Versteegh, Kees et al., eds. EALL, vol. 1, 14.
Koster, Cornelis. 1974. A Technique for Parsing Ambiguous Grammars. In Siefkes, D.,
ed. GI-4 Jahrestagung, Lecture Notes in Computer Science 26 (1975). Springer, 233246.
. 1991. Affix Grammars for Natural Languages. In H. Alblas and B. Melichar, eds.
Springer LNCS 545. Heidelberg, 469484.
. 2004. Head-Modifier Frames for Information Retrieval. In Proceedings CICLing-
2004. Springer LNCS 2945, 420432.
LeTourneau, Mark S. 2006. Case Theory. In Versteegh, Kees et al., eds. EALL, vol. 1,
347353.
Levin, Beth. 1993. English Verb Classes and Alternations: A Preliminary Investigation.
Chicago, Ill.: University of Chicago Press.
Levin, Beth and Malka Rappaport Hovav. 1996. Unaccusativity at the Syntax-Lexical
Semantics Interface. Linguistic Inquiry Monograph 26. Cambridge, MA: MIT Press.
Meijer, Hans.1986. Programmar: A Translator Generator. PhD-thesis, Nijmegen: Univer-
sity of Nijmegen.
Moravcsik, Julius. 1970. Subcategorization and Abstract Terms. Foundations of

Language 1970 (6), 473487.
Nederhoff, Mark-Jan. 1993. A New Top-down Parsing Algorithm for Left-recursive
DCGs. In Programming Language Implementation and Logic Programming, 5th
International Symposium, Lecture Notes in Computer Science. Tallinn, Estonia:
Springer, Volume 715, 108122.
Saad, George Nehmeh. 1981. Transivity, Causation and Passivization: A semantic-syntactic
study of the verb in classical Arabic. Library of Arabic Linguistics. Monograph No. 4.
London: Keagan Paul International.
Al-Saffr, Abdul Emr Dhhir. 1979. A Semantico-Syntactic Study of Classical Arabic in
Case-Grammar Terms. PhD SOAS. London: University of London.
alSq, Fdil. 1977. Aqsmu l-kalmi l-arab min h aytu -akl wa-l-wazfa. al-Qhira:
Maktabatu l-Xnj.
Sbawayhi. al-Kitb. Hrn = Ab Bir Amr b. Utmn Sbawayhi. al-Kitb. 5 vols.
Abd as-Salm Muhammad Hrn, ed. 1977. Cairo: al-Haya al-Misriyya al-mma
li-l-Kitb.
Versteegh, Kees. 2001. The Arabic Language. 2nd revised ed. Edinburgh: Edinburgh
University Press.
, Mushira Eid, Alaa Elgibali, Manfred Woidich and Andrzej Zaborski, eds. 2006.
Encyclopedia of Arabic Language and Linguistics, Volume I, A-Ed. Leiden: E.J. Brill.
, eds. 2007. Encyclopedia of Arabic Language and Linguistics, Volume II, Eg-Lan.
Leiden: E.J. Brill.
Winograd, Terry. 1983. Language as a Cognitive Process. Volume I: Syntax. Reading:
Addison-Wesley.
Yngve, Victor. 1966. A framework for syntactic translation. In David G. Hays, ed. Read-
ings in Automatic Language Processing. New York: American Elsevier, 189198.
ARABIC ON THE MEDIA: HYBRIDITY AND STYLES*
Mushira Eid
University of Utah
1. Introduction
The purpose of media programming is to inform and to entertain. Some

programs can be classified as informative, others as entertainment, and
others as both. The language used on these programs is expected to
reflect these discourse functions. In a diglossic situation, such as that
found in the Arabic speech communities, the choice involves two (at
times more) relatively distinct varieties, or linguistic codes, of a lan-
guage: the high (standard variety) for information and the low (local
dialect) for entertainment. In Arabic these correspond to fush and
mmiyya, respectively. Programs classified as being both (information
and entertainment) raise a question as to what form of Arabic is, or
should be, used. These programs, like the broadcasting media in gen-
eral, represent hybrid contexts that mix, and at times merge, the public
and the private, the formal and the informal. They serve as bridges, or
in-between spaces, where linguistic differences and cultural identities
are negotiated in the production of a program or a performance.
The relationship between language and the media is particularly inter-
esting in the Arabic context due to the diglossic nature of its speech com-
munities. Two major forms of Arabic characterize these communities:
one is the spoken everyday language of communication (mmiyya ver-
nacular/colloquial/dialect), which is also written in certain contexts,1
the other a literary variety which is the medium of written communica-
* An earlier version of this paper was presented at the Center for the Advanced Study
of Language (CASL), University of Maryland, College Park, June 1, 2006.
1
Personal letters, for example, may be written in dialectal Arabic. In Egypt, cer-
tain literary genres are also written in Egyptian Arabic including plays, poetry, and a
few biographies (Awad 1965; El Assal 20022003). There is also one novel written in
Egyptian (Musharrafa n.d.). The practice of including some form of dialect in literature
mixed with fush is a relatively common practice, particularly when it involves dialogue
(Cachia 1990, 5975, chapter four The use of colloquial in modern Arabic literature;
Eid 2002).
404 mushira eid
tion and is also spoken in some formal contexts ( fush lit: eloquent,
literary/standard). The two varieties differ in linguistic form, manner
of acquisition, function or use, and social meaning/value. The definition
of diglossia as it was first presented by Ferguson (1959) is given in (1).
1. Diglossia is a relatively stable language situation in which, in
addition to the primary dialects of the language (which may include
a standard or regional standards), there is a very divergent, highly
codified (often grammatically more complex) superposed variety,
the vehicle of a large and respected body of written literature, either of
an earlier period or in another speech community, which is learned
largely by formal education and is used for most written and formal
spoken purposes but is not used by any sector of the community for
ordinary conversation (Ferguson 1959, Word 15:336).
The definition in (1) presupposes a relationship based on separation,
both linguistic and cultural, of the two varieties fush and dialect. It also
reflects the official discourse and predominant attitudes toward them.
Because of its being the language of divinity and heritage in Islam, the
fush is considered the language of high culture and prestige whereas
the dialects are local, highly divergent, and often not mutually intelli-
gible, at least at the outset. Fush is also considered a unifying force, the
pan-Arab national language, in the words of Shawqi Daif (2001) luat
ub al-umma jamcan the language of all the peoples of the [Arab]
nation, whereas the dialects represent the daily language of a single peo-
ple . . . understood only by its people hence divisive. (Quoted in Bous-
sofara-Omar 2006, 108; see also Haeri 2003 for further discussion.)
Social, political, educational, and technological changes in the 20th
century have brought about change in the relationship between the two
varieties, particularly in their contexts of use, their market value, and
their social meaning. Accessibility to fush increased, for example, due
to changes in the educational system. In Egypt the availability of free
public education through high school and higher education increased
the number of people who can use both varieties. Contexts in which
fush is used increased as well, the broadcasting media as a spoken con-
text for fush being one.2 As a result, the media serves as a liberating
force: by providing contexts that are not restricted to one variety, the
2
The media are often criticized for allowing the dialects to creep into domains origi-
nally reserved for fush . Such intrusions will eventually dismantle the ties that bond
the peoples of the umma [Arab nation] (Shawqi Daif 2001, cited in Boussofara-Omar
(2006) op. cit.). The broadcasting media, however, are not a domain originally reserved
arabic on the media: hybridity and styles 405
media allows users of Arabic access to many forms of Arabic, dialectal

and fush . This position contradicts the predominant discourse, as rep-
resented, for example, in the quotes from Shawqi Daif, that argues for
the separation of the two in order to maintain the purity of the fush
against intrusions from dialect (but see footnote 2).
The media, particularly the broadcasting media, has a unique role to
play in negotiating the relationship between the two varieties. Because
of its dual role, to inform and entertain, the media serves as a bridge
between the public and the private, the local and the global. It repre-
sents different forms of discourse: formal (speeches, lectures, news
broadcasts) and informal (conversations, story-telling, joking) and cre-
ates in-between spaces that serve as excellent sites for the negotiation of
identities. It does so by bringing public content into the privacy of the
home and taking private content to the public view to audiences that are
local and, when aired over satellite channels, global as well. For the Ara-
bic broadcasting media there is in addition a pan-Arab audience that
needs to be addressed. Some channels focus on that pan-Arab audience,
e.g. al-Jazeera. Others, e.g. the Egyptian Satellite Channel (ESC) and the
Lebanese Broadcasting Channel (LBC), focus more on local audiences.
The varieties of Arabic heard over these channels naturally reflect the
intended audience.
In this paper I start from the position that the broadcasting media
(henceforth just media) represent hybrid contexts, which for Arabic
means contexts that allow both fush and dialect. I focus on the inter-
view as a genre whose purpose is to inform and to entertain and whose
context is appropriate for either or both varieties of Arabic.3 I examine
two interviews aired over the Egyptian Satellite Channel. I show that the
language that emerges in these interviews is a mix, identifiable at times
as fush and at times as mmiyya (here Egyptian) but more often as an
in-between variety, a hybrid form of Arabic. Three major styles are
identified, corresponding roughly to Labovs reading, careful, and casual
styles. The styles vary depending on degree of formality and interaction,
topics, and participants (their perceptions of themselves, projections of
for fush . The statement must therefore be applicable to Arabic in the written media,
e.g. newspapers.
3
Eid (2004) shows that interviews and panel discussions from al-Jazeera are at times
conducted completely in fush , but they may include dialectal features depending on
speaker, topic, and degree of interaction. The programs analyzed, however, were limited
to different forms of political discourse.
406 mushira eid
identities, and perceived purpose of communication). For Arabic, they

also involve the alternating use of features of fush and dialect. I discuss
characteristics of each style in relation to the nature and source of the
hybridity. In concluding the article I briefly discuss hybridity in rela-
tion to code-switching and argue that they are different constructs.
1.1 Correspondences between fush and Egyptian Arabic

Table 1 illustrates four types of correspondences between fush and Egyp-
tian Arabic (henceforth, Egyptian) to be used in the subsequent analy-
sis of speech styles: phonological, syntactic, morphological, and lexical.
TABLE 1 SOME CORRESPONDENCES BETWEEN FUSH AND

EGYPTIAN ARABIC
Fush Egyptian Gloss
Phonological
q~ qalb alb heart
t ~ t ~ s aktar aktar aksar more, most

t ~ s tawra sawra *tawra revolution
d ~ d ~ z axad axad axaz he took
d ~ z kadlik kazlik *kadlik also, likewise
d ~ d ~ z du hr duhr zuhr noon
d ~ z du lm zulm *dulm injustice
diphthong~long vowel
ay ~ bayt bt house
aw ~ xawf xf fear
Syntactic
Relative markers alladi, allati, . . . illi who, whom . . .
Demonstratives hda, hdihi, . . . da, di, . . . this, . . .
Negatives laysa/l/lam/lan mi, ma- not
Complementizer an in/ that
Morphological
Passive (u-i/u-a ~ it-) kutiba it-katab was written
yuktab y-it-kitib to be written
Verbs: Stem IV
I N/A; rare asbah became
af al
faal ajab agab to please, like
Lexical
anf manaxr nose
lianna an because
Phonologically-related indi andi I have
kalima kilma word
Unspecified kitb kitb book
fannn fannn artist
On the phonological level, three major correspondences between fush

and Egyptian are illustrated: (1) uvular stop /q/ vs. glottal stop / /; (2)
the interdental fricatives / t d z/, which at times correspond to the stops
/ t d d/, at other times to the sibilants / s z z /; and (3) the diphthongs
/ay aw/ corresponding to the long vowels / /.4 Correspondences in
(1) and (3) apply across the board and therefore need no comment: if
a fush lexical item has features in (1) or (3), the corresponding Egyp-
tian item has the corresponding feature in (1) or (3).5 The interdental
fricatives series, however, require some comment. The fush fricatives
correspond in some words to the stops (aktar ~ aktar more), in others
to the sibilants (ulm ~ zulm injustice). At times both options are pos-
sible. For aktar there is a third alternative aksar, but not for du
lm since
*dulm is not an actual word. This distribution is problematic for the
classification of lexical items, a problem to be addressed briefly in dis-
cussing hybridity.
On the syntactic level, four features are chosen for analysis: relative
clause markers, demonstratives, negatives, and complementizers. They
all serve as functional heads of their respective clauses with clearly diver-
gent forms in the two varieties. Except for complementizers, they have
also been studied extensively in the literature of Arabic code-switching
(e.g. Bassiouney 2006, Boussofara-Omar 2006b, and Eid 1982, 1988.)
On the morphological level, two verb forms have been selected for
this study: passives and Stem IV verbs. Fush and Egyptian passive verbs
are formed by different morphological processes. In fush the passive is
marked derivationally by the vowel melody /u-i/ in the first and sec-
ond syllable, respectively, (e.g. kataba wrote ~ kutiba was written). In
Egyptian it is marked inflectionally by the prefix it- (e.g. katab wrote ~
itkatab was written). Stem IV verbs, formed on the pattern /a-CCVC/
like aqbal approach or agab like, are rare in Egyptian. They are some-
times replaced by verbs from different roots, e.g. gih to come instead of
aqbal, but more often they are replaced by Stem I (CVCVC) of the same
verb, e.g. agab instead of agab.
4
Following the transcription requirements for this volume, long vowels are repre-
sented as single segments / /. For reasons that have to do with syllable weight and
identification of minimal word, to be explained below, these should be understood as
sequences of two identical vowels /aa ee ii uu oo/.
5
The opposite, however, is not true. Not all words with sibilants, for example, in
Egyptian have a corresponding interdental fricative in fush ; some words may have a
sibilant in both, e.g., fush : sam sky and Egyptian sama.
408 mushira eid
Finally, lexical differences between the two varieties come from dif-
ferent sources. The examples in Table 1 illustrate two of such areas.
Some differences are the result of word usage, the association of a mean-
ing with two different lexical items, one considered fush usage and the
other Egyptian as, for example, anf and manaxr nose. Other differ-
ences are phonologically-related as, for example, the difference between
indi and andi I have. Because fush and Egyptian are varieties of the
same language, it is expected that an overwhelming majority of lexi-
cal items would be shared. Table 1 provides examples of words that are
identical in both varieties, e.g. kitb book and fannn artist. I refer to
such words as ambiguous or unspecified to mean non-distinct, or not
specified, as to language variety. I use the term both for words that
include features from both varieties, i.e. hybrid forms.
1.2 Hybridity
A hybrid is something that is mixed, and hybridity is simply the mix-
ture. The term has its origin in biology.6 In linguistics it refers to a word
parts of which come from two different languages. The term has become
central to major theoretical discussions among the discourses of race,
post-colonialism, identity, multiculturalism, and globalization. In theo-
ries of cultural studies, it is understood as recognition of two identities
or a mixture of identities but also as a refutation of assimilation into
a dominant culture. Bakhtin, for example, defines hybridity [hybridiza-
tion] as a mixture of two social languages within the limits of a single
utterance, an encounter, within the arena of an utterance, between two
different linguistic consciousnesses, separated from one another by an
epoch, by social differentiation, or by some other factor (1981, 358). The
term has also been used in relation to mixed language varieties that
result from code-switching among bilinguals (Hinnenkamp 2003).
In Arabic hybrid or intermediate forms, as they are sometimes called,
include features from both varieties fush and dialect and, therefore,
they cannot be clearly identified as belonging to one or the other.7
Table 2 illustrates Arabic hybrid forms with examples selected from the
6
It comes from the Latin: hybrida, a term used to classify the offspring of a tame sow
and a wild boar.
7
See Schulz (1981, 8789) for a discussion of hybridity versus mistakes.
interviews. In the following section, I also illustrate this with sample

texts from the interviews.
Here and in all other examples italics indicates Egyptian Arabic,
bold indicates fush , and unmarked typescript indicates material that
is unspecified as to Arabic variety, hence ambiguous or shared material.
The first example in Table 2 is a hybrid, because it includes the Egyp-
tian present tense marker, the prefix ba-, attached to a fush verb form
marked by the vowel melody /a-a/ of the Stem. The second example
includes the fush person marker prefix /yu-/, marked by the /u/ vowel
rather than the Egyptian /i/, and the fush mood marker, the final /-u/,
but it also includes the sibilant /s/ of Egyptian, instead of the fush /t/.
TABLE 2 HYBRID FORMS

Hybrid Fush Egyptian Impermissible Gloss
hybrids
1. bataammala-taammal b-[]a-tammil *bataammil I contemplate
2. yu-massil-u yu-mattil-u yi-massil, *yimattil *yi-massil-u he represents
3.a. salt ta talta talta *talsa/salta three
3.b. talsa *talta/talta
3.c. ? salsa ? salsa
The variation in correspondences between fush /t/ and Egyptian stops

and sibilants creates lexical possibilities for hybridity and ambiguity. The
last example in Table 2, with the different pronunciations of the word for
three, illustrates this point. The hybrid forms (3ab) allow the mixture,
or co-occurrence, of fush /t/ only with sibilants (e.g. talsa). Combi-
nations with Egyptian stops are impermissible hybrids (*talta). This
suggests that the sibilants themselves can be analyzed as in-between
spaces or ambiguous segments: at times they correspond to fush /s/ as
in the many shared words with sibilants (e.g. sitta six), at other times
they correspond to fush /t/ as in the example (2) in Table 2. More-
over, at times they correspond to Egyptian /t/ as in the last examples in
(3). The impermissibility of hybrids involving either the sibilants or the
interdental fricatives with the stops, as in /*talsa/ or /*ta lta/, suggests
that sibilants and interdental fricatives may be considered members of
the same category or group, since they can substitute for each other.
This distribution also raises a question as to the identification of sibilant
pronunciation, particularly in forms where the only Egyptian corre-
spondence to the interdental fricatives is the sibilant. Are the examples
410 mushira eid
(3ab) in Table 2 hybrids, as I classified them, or are they simply alterna-

tive pronunciations for the interdental fricative? It is an open question
with arguments for both alternatives.
The modus operandi adopted in earlier research contextualized the
decision (Eid 1982, 1988). In lexical items where speakers have all three
choices (t, t, and s), the sibilant pronunciation is counted as fush since
the speaker has an alternative, namely /t/ pronunciation, should she or
he have opted for Egyptian. Likewise, in lexical items where Egyptian /t/
pronunciation is not available, sibilant pronunciation of fush interden-
tal fricatives counts as Egyptian for similar reasons.8
The above discussion of hybrid forms illustrates the concept through
lexical items. In this article I consider hybridity to be a global feature of
text or discourse. It applies to forms of language that look more like a
collage, created out of two or more linguistic codes where the borders
between language varieties are blurred or no longer distinct.
2. Styles
Speech styles are forms of verbal self-expression and interaction. A style

is based on a speakers choice of linguistic form, discourse strategies,
and modes of interaction. Styles vary depending on degree of formality,
medium (spoken or written), and context or situation (conversation in
a coffee shop, for example, versus a lecture, speech, or an academic pre-
sentation). For Arabic, styles may also vary depending on language vari-
ety as well as the degree and nature of the mix between fush and dialect
features.
Three styles are identified in each interview : reading, formal, and
casual styles. While they correspond to Labovs classification of speech
styles, they are not identical to it. The difference lies in the formal and
casual styles, where for Labov casual is conversational and emerges in
relaxed situations when the participants become more comfortable with
each other and forget the context of the interview.9 For Arabic the styles
8
Alternatively, sibilant pronunciation may be considered ambiguous, or shared by
both varieties. The high frequency of sibilant pronunciation in fush contexts, e.g. read-
ing, speeches, and recitation of literary texts, supports this view. This pronunciation is
not permitted in recitations of the Quran, however.
9
Labov has a fourth style D: word lists, which is relevant as an elicitation technique
but not applicable to natural conversations as represented in these programs.
also correspond to the relative distribution of linguistic features from

fush and mmiyya. In this section I illustrate the three styles and pro-
vide a characterization of each based on a quantitative analysis of fush
and Egyptian features in them, as described in Table 1.
Eight segments are selected to illustrate the three styles, four segments
from each of the two interview programs. The interviewees in these pro-
grams are Yehia Haqqi and Anis Mansour. Both are represented by three
segments illustrating formal and casual styles. The reading style is repre-
sented by two other speakers, one from each interview.
2.1 Styles illustrated

The reading style can be easily identified. It occurs when an individual
is seen reading from a script, as in the Haqqi interview, or heard nar-
rating, for example through the use of voice-over, as in the Mansour
interview (see section 3). Segment 1 and 2 from the Mansour and Haqqi
programs, respectively, illustrate the reading style.
Segment 1 with the voice-over is pre-recorded in the studio. No evi-
dence can be found of any Egyptian features, with the exception of the
substitution of Egyptian /g/ for fush /j/, as in the word az-zawgat-u
for az-zawjat-u the wife. This substitution has become so common in
Egypt that it is considered an accepted fush pronunciation in almost all
readings except those of the Koran and other religious texts. Otherwise,
segment 1 can be considered pure fush , or as pure as it can get. Natu-
rally, this ignores the issue of lexical items and particles, or function
words, that are ambiguous or unspecified for variety: galla majesty,
sah fa journalism, f in, and, in if. Case markers are pronounced, for
example, and vowel patterns are those of fush , e.g. al- for the definite
article not il- as in as-sah fa.
SEGMENT 1: San Mansour (interviewer)

as-sahfa hiya z-zawgat-u at-tniya fi hayt-i l-ktib-i l-kabr ans
mansrin gz-a ha t-ta'br. bada"at 'alqat-u ans mansr bi shibat-i
l-galla as-sahfa fi 'm-i alf-in wa tisimiat-in wa sabat-in wa arban.
wa knat al-bidya fi gardat-i l-ass.
Journalism is the second wife in the life of the distinguished writer Anis
Mansour, if such an expression is permitted. Anis Mansours relation-
ship with her royal highness [journalism] began in 1947. The beginning
was in the newspaper al-ass.
412 mushira eid
Segment 2 read by Fouad Duwara in the Haqqi program has a few more
ambiguous lexical items; more importantly, however, is the Egyptian
pronunciation of the years 1962 and 1970. This is common in readings,
for example, of the news and on-the-air commentaries as is the case of
this reader, who is also reading on-the-air. Otherwise, fush is main-
tained throughout in vowel patterns, e.g. h wal not Egyptian h wil, and
other pronunciation differences.
SEGMENT 2: Narrator (Fouad Duwara in Haqqi interview)

wa fi ibrl 'm alf w tusumiyya itnn wi sittn 'uyyin-a yehya haqqi ra"s-an
li-tahrr-i magallat al-magalla wa zalla yatawalla mas"liyataha hatta
December alf w tusumiyya w sabn wa hwala tawl tilka s-sanawt an
yuhfiz-a li l-magalla ala i'raha alladi t-taxadathu li nafsih mundu
inih wa huwa sigil it-taqfat-i r-rafa.
And in April 1962 Yehya Haqqi was appointed editor-in-chief of al-
Magalla journal. He continued to be responsible for it until December
1970. He tried during all those years to preserve for the journal the
motto it had adopted for itself since it was established, and that is [to
be] a record of high culture.
Segments 35 are samples from Haqqis speech. The first, segment 3, has
all the characteristics of a reading style, although Haqqi was not read-
ing. The segment has all the features characteristic of fush including
case markers, vowel patterns, etc. and no features of Egyptian. It also
includes only a few words (6 total) unspecified for variety ( fann art, sir
struggle, maa with, zaman time, bir transient, and xlid immortal).
SEGMENT 3: Haqqi (on art)

al-fann 'ind mawqif-un drmiyy. "aql-u "anna l-fann sir maa z-zaman,
"annahu yurd-u "an yatasayyada l-bir li yuqayyidah, sir'-un maa
l-mawt. wa ma'a dlik hnam [indam] taqif "amma lawha ta'ur
"anna hda l-bir hda l-xlid allad xalladtahu innam huwa bir.
Art for me [is] a dramatic moment. I say that art [is] a struggle with
time, that it wants to capture the transient to preserve it, a struggle with
death. And despite this when [when] you stand in front of a painting
you feel that this transient, this immortal that you immortalized, is only
transient.
Segment 4 differs from segment 3 in that it has some evidence of Egyp-

tian Arabic features from syntax, morphology, lexicon, and pronuncia-
tion patterns: the relative marker illi, the tense/aspect prefix ba-, the
negative main malh, the word wayyit, and the pattern in baladna,
not baladina the expected fush pronunciation.
The segment also includes a relatively large number of words unspec-
ified for variety (27), almost equal to those specified for fush (26), and
a minimal (6) specified for Egyptian. This segment also includes the
hybrid form /bataammal/, reflecting and perhaps also constructing the
Egyptian and day-to-day context of the discourse. 10
SEGMENT 4. Haqqi (his hobby)

min asman il-hadya allati niltuha hdihi s-sra wa hiya min rasm
akbar musawwir s-sn f dlika l-waqt. il-huwya l-wahda illi xaragt
bha huwyat gami wayyit isyn ma-lh- ayy qma kbra lkin li kull
asya makna fi qalbi. wi bataammal kayfa axtr ihd hdihi l-usa. wa
hun aknu fi hla min al-istihz qallan wa urdu an atamid ala ay
fa axtr asan min agr-i misr min agr il-fallahn fi baladna.
One of the most precious gifts that I received is this picture and it is a
painting of one the best painters of China at that time. The only hobby
that I came out with is the hobby of collecting some canes that dont have
any value but each cane has a place in my heart. And I contemplate how
to choose one of these canes. And here I am being a little sarcastic and I
want to depend on something so I choose a cane from the trees of Egypt
from the trees of the peasants in our country.
Finally, segment 5 of the Haqqi sample includes more instances of lexi-
cal items marked for Egyptian (14) by comparison to previous segments.
The number is still the lowest by comparison to that of lexical items
marked for the predominant fush (36 instances) and those unspeci-
fied for variety (25 instances). Segment 5 is interesting from another
perspective: it is the only segment of the three from Haqqi that actually
has one or more sentences predominantly in Egyptian as, for example,
the last sentence. Finally, this segment also illustrates variation in the
use of fush interdental fricatives versus Egyptian sibilants through the
alternation muwaddaf ~ muwazza f employee. I have counted the form
with the sibilant /z/ as Egyptian since the alternative with the stop pro-
nunciation (*muwaddaf ) does not exist.
10
The hybrid form bataammal in this segment and yumassilu in segment 5 are
excluded from the count in each case for the obvious reason: each has features of both
varieties.
414 mushira eid
SEGMENT 5: Haqqi (on government employees)

fi hdihi l-fatra istatana an nungiz bada l-aml. rubbam l yahiqq-u
l an adkurha lkin l bas min an aql . . . fa knat tagriba arba gid-
dan xussan fi l-muwazzafn bi n-nisba li hdihi l-wazrt at-taqfiyya.
anta bayna namatayn min al-muwazzafn: muwazza f idri lkin sifr fi
l-fann wa muwazzaf fannn lkin sifr fi l-idra. timsiku min hina yigri min
hina wa il xirih. fa kna l-amal asr giddan. al-husl ala muwazza f
yumassilu xayra at -t arafayn an yakna fannnan wa fi l-waqt nafsu
yistah mil innu yuud ala maktab wi yira waraa wi yiraf yiktib gawb
kwayyis.
During this period we were able to accomplish some goals. Maybe I
dont have the right to mention them but its ok that I say . . . so it was a
very strange experience especially in the employees of these ministries
of culture. You are between two types of employees: an employee [who
is] administrator but zero in art and an employee [who is] an artist but
zero in administration. You catch him from here, he runs from there and
so on. So work was very difficult: the finding of an employee who is good
at both, to be an artist and at the same time tolerates to sit at the desk and
read a paper and know how to write a letter well.
The speech samples from Haqqi demonstrate that in this speech style
features from Egyptian are consistently much lower in frequency than
those from fush as well as those unspecified for variety. One sample
(segment 3) has no features of Egyptian and includes almost all the
characteristics of a reading style.
What, one might ask, would explain such differences? Topic is a very
likely explanation. In segment 4 with the most fush , Haqqi speaks
about art and what it means to him: a topic that lends itself to fush but
more importantly, a topic he has written and lectured about through-
out his long career.11 In segments 5 and 6 the topic shifts to relatively
more personal topics and situations: his hobbies and his experiences as
a government official, respectively. The style shifts accordingly. The last
segment includes the most dialectal features, perhaps because it is also
the one most related to everyday events and behaviors as a result of his
comments on government employees. This is not to say that it is neces-
sary that Egyptian be used in this case. Other segments in the program
show Haqqi not resorting to Egyptian under similar circumstances. In
fact, these two segments are the only ones in his speech sample that
include any features of Egyptian.
11
In an earlier segment Haqqi talks about the short story and the essay, stressing that
the purpose of the former is to entertain the reader (imt giving pleasure to) and the
latter is to inform (ilm) the reader.
Haqqis style can be said to conform to the discourse on diglossia by

relating Arabic variety to topic: fush for the abstract, informative, and
literary, but Egyptian for everyday situations. Haqqis speech sample
shows that fush can be used successfully and appropriately in descrip-
tions and narrations about everyday situations, although it would be
odd to use it in actual day-to-day interactions.
A different style is represented in segments 68 from the Anis Mansour
interview. Segment 6 consists of a question from the interviewer, Sanaa
Mansour, followed by A. Mansours answer. The question addresses
the issue of opposition to a mainstream position: What happens when
one goes against the tide by taking an unpopular or oppositional stand.
Mansours answer includes more instances of lexical items unspecified
for variety (28) than instances of either Egyptian (14) or fush (15). This
segment also shows an almost equal distribution of Egyptian and fush
lexical items.
SEGMENT 6: A. Mansour (Swimming against the tide)

Question 1. SM:
bi-ylu l yaksab katran man yasbah didd it-tayyr. hadritak kisibt
il-kitr walla dayman bi-tisbah didd it-tayyr.
They say: He does not gain much who swims against the tide. Have you
[sir] gained much or were you always swimming against the tide?
Answer 1. AM:
da yatawaqqaf ala anni tayyr. ahynan min al-maslaha l-ma aw
il-wata niyya annaki tasbahi didd it-tayyr iza kn it-tayyr inhill aw
fasd, tafha, adam axz il-ay bi giddiyya. hun yusbih min al-wgib
ala l-ktib innu yasbah didd it-tayyr, innu yimil tayyr mudi liannu
wagbu innu yuslih aw yartafi bi n-ns. fa da yatawaqqaf ala anhi tayyr
illi fh xf aw mahazr innu yusbah didd-u.
This depends on which tide. Sometimes it is for the public or national wel-
fare that you swim against the tide, if the tide immorality or corruption,
triviality, not taking matters seriously. Here it becomes the duty of the
writer that he swim against the tide, that he create a counter tide because
it is his duty that he reform or uplift the people up. So this depends on
which tide is fearful to swim against.
In segment 7, which is a follow-up on the question above, there is a clear
shift in style for both interviewer and interviewee. Sanaa Mansours
question here is an excellent example of what I call collage. It consists
of 9 words of which 4 are Egyptian, 3 are unspecified for variety, and 2
416 mushira eid
are specified for fush .12 She receives a one-word answer in Egyptian,
which she follows up with a question in clearly marked Egyptian syntax,
indicated by the WH-question word h what, and unmarked lexi-
cal items: kn and in-natga. Her question generates another collaged
response from A. Mansour, a response that is predominantly Egyptian
(16), minimally fush (2), with unmarked items in between (7).
SEGMENT 7: Mansour (Swimming against the tide, continued)

Q2. SM: gatlak fatart katra fi h aytak innak sabaht didd it-tayyr.
Did you have many times in your life that you swam against the
tide?
A2AM: kitr
A lot.
Q3SM: kn in-natga h
What was the result?
A3AM: wala h ga. w-adni id uddmik ahu. w-ixtalaft kasran w
ittafat iktr lkin madm inti muqtania aw amna f daawtik
aw fi bi wighit nazarik mayhimmik h ga aw mayhimmin
h ga.
Nothing, and here I am sitting in front of you. I have disagreed
often and agreed often but as long as you are convinced and hon-
est in your mission or in your point of view, you dont care for
anything or I dont care for anything.
The last segment 8 from Mansour is also predominantly Egyptian (35
instances), minimally fush (4), with unmarked items (20) in between.
Had it not been for the last sentence which includes a sentence, or part
thereof, in fush , this whole segment could have been considered Egyp-
tian. The unmarked lexical items, although a majority, have little impact
on the reader in terms of identification of variety. The syntax, phonology,
and lexicon of Egyptian predominate, and the identification as Egyptian
goes unchallenged.
SEGMENT 8. Anis Mansour (More about Kamel El Shenawi, ad h kan

zarf How nice he was!)
giddan yani maraf aullik nukatuh. kan yil ala Sawla H igzi allh
yirh amha innaha min adabha -idd ablima tiftah id-durg tixabbat alh
il-awwil. masalan kan andi arabiyya zuayyara kida baby Ford w-ana mi
12
The fush in her first question (segment 6) is the result of a quoted proverbial
statement.
saww kwayyis yani fa kn yil h in il-arabiyyt il-karro timi gamb

minni timil voo [li d-darag di] yani fa kull adt Kamel El Shenawi
marah wi xiffit damm w malib wi tayr fi k-kalm lkinnu abb w axx w
umm w nima s-sadq wa laysa lahu nazr fi h ayatna s-sahafiyya.
Very much [meaning] I dont know. I tell you his jokes. He used to say about
Salwa Higazi, God have mercy on her, that because of her politeness she
would knock on the drawer first before opening it up. For example, I used to
have a small car, a Baby Ford, and Im not a good driver [meaning] so he
used to say that karro carts would go beside me and sound voom. [laugh-
ter] to this extent [meaning]. So all his get-togethers were fun, humor, tricks,
and play with words but he was a father, a brother, a mother and the best of
friends and there is no one like him in our journalistic life.
Unlike the two previous segments, fush here comes at the end as a
sequence, a sentence or part thereof, and is not intermingled with the
earlier narrative. It serves as a conclusion to Mansours narrative on
Kamel El Shenawi and his recollection of his sense of humor. The con-
clusion, that there is no one like him, is what Mansour asserts and wants
the audience to remember. The code-switch, or style shift, to fush helps
him accomplish this goal. Mansour relies on his knowledge of code dif-
ferences to set the conclusion apart from previous narration, thereby
signaling the conclusion and highlighting its content. The switch, or
shift, is therefore meaningful at this particular moment and within the
context of the text. This, one might argue, represents the basic differ-
ence between hybridity and code-switching. Hybridity creates a speech
sample that resembles a collage of elements from one code or another.
In that sense its a global feature of a text. Code-switching as a conversa-
tional and stylistic strategy is meaningful in a specific context within a
text; in this sense its local within a text.
The three segments 6 through 8 from Mansour, like those from Haqqi,
show an increasing shift towards Egyptian, which may also be related
to topic. The shift in topic here is not as strong as it was in the Haqqi
samples. The difference between segment 6 and 7, for example, is not
so much in the topic as it is in locus. Segment 6 addresses the issue of
swimming against the tide on a general, more conceptual level whereas
the question in 7 shifts the locus to the person, to Mansour himself and
his experiences in swimming against the tide. The shift to the personal
tends to be accompanied by an increasing shift towards Egyptian. Like-
wise, the last segment (8) is located in the personal: reminiscing about
a friend and sharing jokes he used to tell with the audience to demon-
strate his sense of humor. Both are discourse functions most appropri-
ately conducted in Egyptian.
418 mushira eid
On the basis of these samples, two conversational styles are identified:

one is fush -based, the other Egyptian-based. The two styles incorporate
features from both varieties and include a significantly large number of
lexical items that are ambiguous or unspecified for language variety.13
The texts produced are often like a collage in which the borders sepa-
rating one variety from the other are often blurred making the prod-
uct look more like a hybrid text or discourse rather than one where the
boundaries are maintained and the switch from one to the other variety
is clear. The speech samples from Mansour fall into the first category,
those from Haqqi into the second.
The analysis in this section identified three styles and illustrates some
of their linguistic and discourse features. In the next section I present
a frequency-based quantitative analysis of speech samples from Haqqi
and Mansour to identify overall characteristics of the two styles they
represent.
2.2 Quantitative Analysis

The quantitative analysis is based on speech samples of approximately
the first 10-minutes from each speaker, Haqqi and Mansour. The results
are presented in Table 3.
The overall totals at the bottom of the table show clear differences
between the speakers styles. Haqqis is clearly fush -based (80%) and
Mansours Egyptian-based (74%). Although the two speakers have clear
preferences for different varieties of Arabic, they prefer one to the other
by about the same degree: a ratio for Haqqi of 80% fush to 20% Egyp-
tian and for Mansour 74% Egyptian to 26% fush .
Table 3 also provides a breakdown of the distribution of fush and
Egyptian features according to the selected phonological, syntactic, and
morphological features. The figures from Haqqis speech show a bias
towards fush in all areas, but particularly in the syntactic and morpho-
logical features with 96% and 100%, respectively. The highest percentage
of Egyptian features in his speech sample comes from phonology (28%).
Of the four phonological variables used in the analysis, the alternation
between fush interdental fricatives and Egyptian stops and sibilants is
the most significant; it is responsible for 85% of the Egyptian phono-
13
Since Schmidt (1974) very few studies, if any, have addressed the role of the lexicon
in Arabic mixed varieties and problems in the identification of lexical items by variety.
TABLE 3 STYLESHAQQI AND MANSOUR

Haqqi Mansour
Fush Egyptian Fush Egyptian
Phonological
q~ 70 4 20 61
t ~ t ~ s 4 20 3 16
d / d ~ d / d ~ z/z 29 21 2 30
diphthong ~ long vowel 20 3 2 10
TOTALS 123 48 27 117
72% 28% 19% 81%
Syntactic
Relative markers 9 1 0 10
Demonstratives 12 0 4 15
Negatives 6 1 10 20
Complementizer 20 0 5 13
TOTALS 47 2 19 58
96% 4% 25% 75%
Morphological
Passive (u-i, u-a ~ it-) 13 0 9 6
Stem IV: a-fal pattern 13 0 7 9
TOTALS 26 0 16 15
100% 0% 52% 48%
OVERALL TOTALS 196 50 62 181
80% 20% 26% 74%
logical features in Haqqis speech. In addition there is a strong differ-

ence in the distribution of the voiced and voiceless interdental fricatives
by comparison to their corresponding Egyptian pronunciation (stops
and sibilants). Compare the 29 instances of the fush voiced fricatives to
the 21 instances of Egyptian pronunciation and the 4 instances of fush
voiceless fricatives to the 20 instances of their corresponding Egyptian
pronunciation.
A very different picture emerges from the analysis of Mansours style.
While Egyptian is clearly the dominant variety for him, based on the
distribution of features in all three components (phonology, syntax, and
morphology), the contribution from fush is not as minimal as was the
distribution of features from Egyptian, the less dominant variety, in the
Haqqi sample. This is most striking in the morphology, for example,
where the distribution of fush and Egyptian features in Mansours
sample is almost evenly split (52% to 48%, respectively); in Haqqis
sample morphological features of the less dominant (Egyptian) variety
420 mushira eid
FIGURE 1 STYLESPERCENTAGES OF MIX HAQQI

MANSOUR
100
80
60
40
20
0
Egyptian Fush Egyptian Fush
Pronology 28 72 81 19
Syntax 4 96 75 25
Morphology 0 100 48 52
Haqqi Mansour
are nonexistent. Likewise, the distribution of syntactic features in Man-

sours style shows relatively more contribution form the less dominant
( fush ) with 25%; in Haqqis the less dominant (Egyptian) contributes
only 4% of the syntactic features. The phonology in Mansours style is
consistently Egyptian-dominant at 81%, stronger than the phonology of
Haqqis dominant variety at 72%. The strongest contribution from the
non-dominant fush in Mansours sample comes from the /q ~ / alter-
nation, with fush /q/ representing 25% of all instances of this variable.
In Haqqis style the Egyptian sibilant pronunciation of the interdental
fricative accounted for 85% of all Egyptian pronunciation features. A
graphic representation of these distributions is provided in Figure 1.
The focus of the discussion so far has been on linguistic features (of
phonology, morphology, and syntax) that are different in the two variet-
ies, hence serve to distinguish one from the other allowing speakers to
identify a word, a speech segment, or part thereof, as belonging to one
or the other variety. There is naturally a significant amount of overlap
in the three components. This overlap, or shared features, is typically
ignored in discussions of Arabic mixed varieties since similarity can be
assumed in the analysis of same language varieties.
The lexicon is an area perhaps ignored most in analyses of mixed Ara-
bic varieties, partly because of assumed similarity and partly because
of difficulties speakers encounter in categorizing lexical items accord-

ing to variety. Schmidt, for example, argues that speakers of Arabic are
often unable to classify lexical items as belonging to one or the other
variety, fush or Egyptian, and that the difficulty may be explained on
the basis of topic or semantic fields (1974, 6076). When speakers were
asked to provide fush and Egyptian equivalents in four semantic fields
(body parts, foods, arts, and politics), they had more difficulty with the
last two than they did with the first two. Results of this type can have
more than one interpretation. They can be taken to show a complex,
integrated, interconnected system whereby some areas of the lexicon
such as arts and politics are shared and others distinctevidence for
complementarity typical of subparts of a whole. They can also be inter-
preted as pointing to areas of deficiency, or incompleteness, in one or
the other language system and used to argue for the superiority of one
or the other variety. Regardless of interpretation, these subsystems of
Arabic are interconnected in ways that are yet to be discovered and sat-
isfactorily explained.
To get a sense of the distribution of lexical items by variety and the
impact the lexicon may have on identification of a base variety, I have
analyzed the lexical items in the 10-minute speech samples of Haqqi
and Mansour based on the presence or absence of clearly marked fea-
tures of either variety. The absence of such features makes the lexical
item unspecified for variety, or ambiguous. As explained earlier, hybrid
words would have features from both.14 Table 4 provides the overall dis-
tribution of lexical items from the 10-minutes speech samples of Haqqi
and Mansour according to the three categories: fush , Egyptian, and
unspecified.
The category unspecified refers to words (content morphemes) and
particles (grammatical morphemes)15 that cannot be distinguished as to
variety and can therefore be considered as ambiguous or shared items.
The results support the idea of a dominant variety underlying diglossic-
based stylistic variation. Haqqis lexicon is predominantly fush (74%)
with a smaller 23% of his lexical items unspecified, leaving a minimal 3%
marked as Egyptian. Mansours lexicon is unexpectedly only 45% marked
as Egyptian; but he has an almost equal amount (41%) unspecified and
14
The analysis of hybrid words in the speech samples of Haqqi and Mansour has not
been completed.
15
I adopt McCarthys and Princes (1990) definition of a minimal word in Arabic as
consisting of at least two moras.
422 mushira eid
TABLE 4 THE LEXICONPERCENTAGE OF MIX

Haqqi Mansour
Egyptian 25 (3%) 391 (45%)
Fush 683 (74%) 119 (14%)
Unspecified 214 (23%) 354 (41%)
TOTAL 922 (100%) 867 (100%)
smaller but sizeable percentage (14%) marked as fush . The major dif-
ference between the two styles is the proportion of the lexicon allot-
ted to the predominant variety by comparison to the unspecified lexical
items and to those marked for the non-dominant variety. In Mansours
style the discrepancy between the dominant and non-dominant variety
is not as marked, partly because the contribution from the unspecified
component of the lexicon is stronger than it is in Haqqis style (41% to
23%, respectively). And while the proportion of lexical items from the
dominant and non-dominant is higher in both styles, the difference is
much larger in Haqqis sample than it is in Mansours, 74% to 3% in the
former and 45% to 14% in the latter. A graphic representation of these
results is provided in Figure 2.
The graphic view in Figure 2 suggests a relatively more balanced
distribution in Mansours style than in Haqqis. The graph represent-
ing Haqqis style looks as though one component is suppressed (Egyp-
tian), another is allowed some presence (the unspecified), but the third
is asserted as an overpowering presence ( fush ). In Mansours there is
not one single category that is overpowering, but there is one ( fush )
less represented than the othersan unequal participant perhaps.
The graphic view also suggests that the linguistic boundaries are more
clearly defined in Haqqis style than they are in Mansours. For Haqqi
the domain is no doubt fush ; Egyptian may creep in only slightly, if
necessary. For Mansour, the domain is more inclusive.
The ambiguous, or unspecified, lexicon serves to mediate the differ-
ence between the two varieties. It creates a shared, or an in-between
space, consistent with the two distinct codes. In doing so, it contributes
to this linguistic collage by allowing smooth transitions from one code
to the other, thereby blurring the distinction between the two. Hybrid
forms, I would add, serve a similar purpose: they can be heard and
interpreted as one, the other, or both varieties. The result is a style that
does not sound too colloquial (dialectal) or too literate ( fush )a
balancing act that allows each speaker to accommodate the situation
FIGURE 2 THE LEXICONPERCENTAGE OF MIX

80
60
40
20
0
Haqqi Mansour
Egyptian 3 45
Fusha 74 14
Both 23 41
and create personas and identities that are sufficiently separate yet simi-
lar enough to be still viewed as one.
The question remains as to why these two speakers, both highly edu-
cated, intellectuals and writers, would choose styles so different from
each other in what appears to be very similar context: one conform-
ing to a great extent to the predominant (official) discourse of separa-
tion ( fush for public, formal contexts) and the other nonconforming
through its extensive use of Egyptian. The answer to this question
requires a more detailed analysis of the interviews themselves includ-
ing contexts and topics of conversation as well as speakers purpose in
communication.
3. The Interviews
An interview is a conversation or discussion between two participants

with the purpose of gathering information by the interviewer about an
individual, an institution, a topic, or some other issue. Since the inter-
viewee agrees to participate, s/he is expected by the interviewer to be
forthcoming and cooperative in this conversation. The interviewer,
having initiated the event, is likely to have his/her agenda as well. Both
have something to gain from participation in the event, thus a motive
for a successful performance.
An interview may be private or public and formal or informal,
depending on the purpose of the interview. Participants are expected to
424 mushira eid
follow certain rules, usually set ahead of time, regarding procedure and
at times topic as well. An interview on radio or television is in addition a
public performance intended to create the semblance of a conversation
whose purpose is to convey information to the audiences and to enter-
tain them as well. The ultimate success or failure of such an interview
depends on how the performance is staged and how participants interact.
In this section I describe the two interviews selected for this study
in terms of three components that define an interview: set-up, infor-
mation, and interaction. I relate these aspects of the interviews to the
styles identified above and to speakers linguistic choices and purpose in
communication.
3.1 The interviews compared

The two interviews are similar in set-up and type of information solic-
ited and conveyed. They differ primarily in forms of interaction and
interviewees purpose in conversation, more specifically in the identi-
ties, or persona, they negotiate and attempt to construct through the
interview. The interviewees, Yehya Haqqi and Anis Mansour, are both
famous men of letters, very much involved in the literary and cultural
scenes of their time. Yehya Haqqi (19051992) is a literary critic, essay-
ist and short story writer. He had tremendous influence on several gen-
erations of modern Arab writers and is considered one of the fathers
of modern Arab culture. Many of his works have been translated into
several foreign languages and are being taught in many academic insti-
tutions, especially his masterpiece Qindl Umm Hashim The Lantern
of Umm Hashim. Anis Mansour is a journalist, an essayist, and a story
writer. He is known for his daily column in al-Ahram mawqif Posi-
tions/Opinions and has written numerous books including his famous
h awla l-lam fi 200 yawm Around the world in 200 days. Both inter-
views are incomplete, with the early part missing from each. The amount
of recorded time available from them, however, is comparable: approxi-
mately the last 3035 minutes of each program.
3.1.1 Set-up
Both interviews take place on-location outside the studio in the homes
of the interviewees. In the Mansour program, the interview is conducted
outdoors by the swimming pool with the interviewer and interviewee
seated across a small table. In the Haqqi program, the interview takes
place inside his home. Some scenes are shot in his study, others in his
living room, around the dining room table, or in the entry hall.
3.1.2 Format
The Mansour interview adopts a question-answer format and is hosted
by Sanaa Mansour, an experienced interviewer and host of numer-
ous programs on the ESC including Good Morning, Egypt.16 In the
Mansour program she performs the interviewer role, asking the ques-
tions and directing the flow of conversation through topic introduction,
follow-up, commentary, and interruption where necessary. From this
perspective, the format of this interview is the conventional interview-
er-interviewee. But it actually is far from being so, as will be shown
below.
In the Haqqi program there is no interviewer on the screen. The
questions are not aired, only responses and comments. The program
also includes other participants: Haqqis daughter Nuha, his French wife,
and two of his colleagues, Fouad Duwara and Mohammad Rumeish; the
former is a critic and the latter a writer, as per information projected on
the screen. Both are friends and colleagues who had worked with Haqqi
during his term as editor of al-Magalla.
Despite the difference in format (one-to-one vs. multiple participa-
tion), one participant in each program performs the role of commenta-
tor and topic introducer. In the Mansour program this role is performed
by the interviewer, Sanaa Mansour, and in the Haqqi program by his
colleague, Fouad Duwara. In performing these roles they read short seg-
ments, thus illustrating the non-conversational reading style.
3.1.3 Textuality
Both programs rely on the visual, for entertainment and for reinforce-
ment of information. Images related to the topics under discussion are
often projected on the screen during the conversations. In the Haqqi
program, for example, close ups of paintings appear as Haqqi talks
about gifts he received during his visit to China as director of an Egyp-
tian film festival and about the significance of animals in Chinese art as
illustrated in the paintings. Likewise, pictures of the movie stars Fatin
16
To my knowledge, interviewer and interviewee are not related despite the same
last name.
426 mushira eid
Hamama and Omar Sharif are projected on the screen as Haqqi talks
about Fatin Hamamas popularity among Muslims in China. Pictures of
Yehya Haqqi himself, sometimes a younger Haqqi, also appear on the
screen showing him standing by a car in Paris, for example, as Rumeish
narrates a story about Haqqis trip to Paris to undergo a surgery.
In the Mansour interview, pictures of his books are projected on
the screen when a specific book is mentioned as happened with h awla
l-lam fi 200 yawm (Around the World in 200 Days); likewise images
of newspaper articles or quotes appear on the screen when mentioned
in the conversation. Each segment begins with scenes illustrating the
topic to be discussed with the interviewers voice-over introducing the
segment. To introduce the last segment on the intellectuals and the cafs
they frequented, for example, scenes of Cairo streets with various shops
were shown during the introduction. Later in this segment, scenes of
the Brazilian coffee shop, A. Mansours favorite, were shown when the
interviewer mentioned it in her questions. Pictures of Kamel El Shenawi,
Taha Hussein, Tawfiq al-Hakim, and Abbas al-Aqqad also appear at dif-
ferent points during the program as their names are mentioned.
This interweaving of materials from different text typesvisual, musi-
cal, writtenmakes both interviews multi-textual. It serves to create a
collage bringing into the main text of the interview other sub-texts and
sub-effects that contribute to the voice or voices represented in each
interview and ultimately influence the style speakers choose to adopt
when speaking in their voice and the voices of others.
3.1.4 Topics
Both interviews can be described as reminiscences, reflections on the
lives and times of these two professionals. Consequently the topics in
both interviews revolve around the personal and the professional.
In the 35 minutes available from the Mansour interview the discus-
sion is divided into four distinct segments. The first segment is personal
dealing with his views on such matters as having children and the nega-
tive stance he has taken towards women. The second revolves around
his life experiences as a journalist/writer, including reflections on his
relationship with leading intellectuals of his time (Kamel El Shenawi,
Taha Hussein, Abbas al-Aqqad, and Tawfiq al-Hakim). The third is
focused on his hectic (perhaps antagonistic) relationship with Abbas
al-Aqqad, and the fourth on cafs frequented by various intellectuals
including himself, a phenomenon the interviewer refers to as taqfat
a-ri (Street Culture).
Likewise, in the 32 minutes available from the Haqqi program, the

topics are focused on three areas: Haqqis artistic interests and his views
on literature and art (which for him transcend the literary and include
fine arts and music), his work as editor in chief of al-Magalla and as
government official in the Ministry of Culture, and his daily routines
and personal life as father and husband. These areas and the specific
topics they include are not as clearly bounded in the Haqqi program as
they are in the Mansour interview. Segments dealing with the personal
and private are interspersed with those dealing with the professional
and public. The sequencing in the first part of the program, for example,
illustrates this point. It starts with Haqqi talking about the short story
and the essay, the former being an art form very much like a painting.
Fouad Duawar follows commenting on Haqqis contribution to the per-
forming and fine arts. Haqqi follows to talk about what art means to
him. The topic then shifts to the personal through Nuha, Haqqis daugh-
ter, who talks about him as a speaker and a writer. Haqqi follows to talk
about subject matter and description in his work with comments on the
richness of the Arabic language, meaning fush . A switch to the per-
sonal follows again from his daughter and then his wife, and so on. The
segments flow one into the other and are not typically introduced with a
reading segment, for example, as is the case in the Mansour program.
This difference in topic organization and shift can be attributed to
production decisions on program set-up and format. But it has implica-
tions for the interviews and the speech styles chosen by the speakers. It
may, for example, have reduced, even eliminated, the on-screen interac-
tion among the participants allowing, for example, Haqqi and Duwara
to maintain their predominant fush style.17 The one-to-one set up in the
Mansour program allows for interaction and forms of discourse, such as
interruptions and corrections, that generate spontaneous speech, which
in turn triggers the more casual, conversational style which for Arabic
is dialect-based.
3.1.5 Voice
A major difference between the two interviews involves the issue of
voice: who speaks about what and in whose voice. Despite the differ-
ence in set up noted above, the two interviews are multi-voiced, but in
17
Duwara is included here although his speech style, other than reading, is not dis-
cussed in this article.
428 mushira eid
different ways. We hear Haqqi and Mansour speak in their voice, both
physically and metaphorically. They speak primarily in the I, not the
we, since neither is there to represent a group or organizational identity.
Mansour tends to represent others more so than Haqqi. When he speaks
in the name of the young generation of writers of his time, for example,
he speaks in the we. Haqqi brings in other fush voices by remembering
lines of classical Arabic poetry, for example, and reciting them for the
audience on the screen.
The one-to-one set up in the Mansour interview makes him the only
speaker, the interviewer being the other. Through these two speakers, we
hear other voices as well. The interviewer Sanaa Mansour speaks in her
own voice when, for example, she expresses an opinion, makes a com-
ment, or asks a question. But she also brings into the interview Man-
sours voice from the past by quoting his public pronouncements and
published views, at times challenging and questioning them, at other
times allowing him to elaborate and narrate. In so doing, she merges
the past and the present thereby adding to the multi-textual collage of
this program a time-based dimension. Although Anis Mansour speaks
in his voice, he brings in the voices of others through his narration. In
the story-telling segments of the interview, for example, he quotes and
reports the speech of others; we hear them but only through Mansours
voice and his perspective.
Through the multi-person set up in the Haqqi program, we hear and
see four other people, in addition to Haqqi, speak about his life and his
accomplishments. Haqqis comments, however, are limited to Haqqi the
professional: his views on the arts and his track record as an editor and
government official. This public persona is reinforced by his two col-
leagues, who provide their perspectives on Haqqi as a public figure, a
colleague, and a friend in the profession. Projections of Haqqis private
persona are left to his wife and daughter. As a result, the program as a
whole maintains the boundaries between the two personas: they remain
mostly separate as they come in different voices. Perhaps because of this
separation, segments dealing with the personal/private in this interview
are interspersed with those dealing with the professional/public allowing
the two to merge into one unified whole through program production.
3.2 Projections of Identities

Although the identities projected in these interviews fall into the gen-
eral categories of the personal and the professional, the persona created
by the main figure in each of these programs are very different. This dif-
ference is communicated through the speakers voice, i.e., the language
that expresses the persona. The style differentiation described above can
be viewed as representations of speakers voices. In this section I discuss
in relatively more detail the persona being created in the interviews,
relating it to topic, speakers voices, and the interview as a whole.
The Haqqi interview maintains the boundaries between the per-
sonal and the professional and, perhaps as a result, between fush and
Egyptian. The set-up and production of the program helps maintain
the boundaries by assigning participants in the program different roles
in the construction of Haqqis overall Persona. Haqqi himself addresses
only the professional, either as his own choice or as that of the producer;
his fush -based speech style reflects this choice. Others interviewed in
the program speak to the personal as, for example, his daughter Nuha,
or to both the personal and professional as does his colleague Meleish;
their styles vary accordingly and appear to be closer to Mansours than
they are to Haqqis.
The focus on the professional with Haqqi allows him to express
his views on literature, art, and literary criticism and to reflect on his
accomplishments in the public sphere as an intellectual, a government
official, and editor of a literary journal. He expresses his views on the
short story, for example, as an art form whose primary purpose is to
entertain, or give pleasure to, the reader (imt al-qri) as opposed to
the essay whose purpose is to inform (ilm). The voice we hear is Haqqis
throughout: his views on art, broadly defined to include literature, fine
arts, and performance arts, recollections of his three years administra-
tive experiences as Director of the Arts Department in the Ministry of
Culture, and his role as editor of al-Magalla. His voice is complemented,
and at times reinforced, by other participants. Haqqis reflections on his
role as editor of al-Magalla (that he made lots of friends through this
job) are picked up in both Rumeish and Duwaras comments. They talk,
for example, about Haqqi as being dedicated to his work and supportive
of other writers. Rumeish recalls his first meeting with Haqqi to submit
for publication his first short story, while Duwara comments on Haqqis
dedication as a journal editor and his travels to solicit manuscripts for
the journal.
On the personal, or private, domain Haqqi briefly mentions his
hobby of collecting canes from all over the world. He describes, and
shows the audience, a few of his favorite canes and explains how his
choice is often determined by his mood. His conclusion to this segment
430 mushira eid
relates the practice of using canes to the public domain, explaining how
it used to be a practice among ministers and politicians but is now a
thing of the past. All other aspects of his private persona are left for
others to project. His daugther, Nuha, speaks of Haqqi as a father and a
professional, merging the two identities a little. She gives the audience a
glimpse of Haqqi the writer and the public speaker through her eyes as
his daugther. She describes his eloquence as a speaker and the ease with
which words come out of his mouth when he speaks; but apparently
they dont come as easy when he writes. Nuha describes how more diffi-
cult and time-consuming writing is for him, comparing the writing pro-
cess to a complicated child birth (wilda mutaassira). She projects him
as a kind and nurturing father, describing their promenades in Maadi,
where he would urge her to observe the beauty in nature and to listen
to the birds, and their walks in Paris, where they would frequent art
museums. She contrasts his perspective of himself versus her perspec-
tive of him. While Haqqi says he is old at 80, she sees him as a young
man (b) still, more knowledgeable than anyone in her generation. A
more detailed description of his private life is left to his French wife to
construct for the audience. She speaks in French with a voice over in
Arabic, thus adding one more voice to the many voices in the program.
His wife describes their life together, their walks in Maadi and in Paris,
their visits to museums, and his eating habits (eats very little). As these
life events are narrated, we see the two of them at times sitting at the
dinner table, other times in their living room or in his study, and at
times joined by their daughter. His colleague Rumeish provides a com-
bined personal-professional account, for example, in his narration of
how Haqqi cancelled plans for surgery in Paris to return via Libya to be
with the Egyptian people during the 1967 war.
In the Haqqi program, then, the boundaries between personal
and professional identities are maintained as a result of participants
(assigned) roles in this overall construction of Haqqis identities. This
allows Haqqi to speak in his fush -based style and maintain the public
image he has created for himself throughout his career, a supporter of
the fush and its heritage. From this perspective the Mansour interview
is very different.
The Mansour program, unlike Haqqis, does not separate the profes-
sional and the personal. The boundaries between them are fuzzy, or
blurred, making the two identities appear merged into one entity, a col-
lage of the personal and the professional created primarily by Mansour
himself with the help of the interviewer. Since this program is set up
with Anis Mansour as the only speaker, it becomes his responsibility to

create this hybrid persona, the merger of the personal and the profes-
sional. This Mansour does through a speech style in which the boundar-
ies between the two varieties of Arabic are no longer maintained.
The interviewer Sanaa Mansour brings up for discussion positions
Anis Mansour has taken in public on certain issues. While these top-
ics all belong to the public and therefore professional domain, almost
all are handled on a personal level as narrative rather than as intellec-
tual debates on a more abstract level. The segment discussed earlier on
swimming against the tide is one example. Most of the questions are
handled on a concrete, personal experience level allowing Mansour to
engage in the narration of events in his voice but often reporting the
voices of other participants in past events.
In the Mansour interview, the personal is also public. In the first
segment, for example, the interviewer brings up Mansours views on
children, marriage, and women. Most, if not all, these views come as
quotations from, or paraphrases of, Mansours published articles, books,
or other public pronouncements. In the discussion regarding his nega-
tive views on women, the conversation at some point centers around
his wife. Despite his opposition to marriage and his negative views on
women, Mansour is in fact married and his wife is a public figure. The
interviewer wonders if his wife is then different from all other women?
Likewise in the discussion related to children, Mansour explains that
although he is opposed to having children (and gives his reasons based
partly on the professional), he does love children: all children in his
family have pictures with Mansour carrying them on his shoulder. Here
the professional justifies the personal and vice versa. The justification he
strongly voices in concluding this segment is based on the professional
and is expressed with a switch to clearly marked fush syntax and mor-
phology of the embedded that-clause: yihimmini "an "akna mufak-
kiran aw ktiban aw "adban giddan Its very important for me that I
be a thinker or a writer or a man of letters. This is the public persona
he projects of himself, but always together with the personal.
Differences in the interview set-up and the specific topics covered in
the interviews are partly responsible for the overall persona created of
the main figures. The bulk of the responsibility however lies with the
individuals being interviewed. The persona projected by Mansour is
clearly multi-voiced. He plays many roles and speaks in many voices. He
plays his own person as a writer, an intellectual, and a family person. But
he also plays the roles of others in his narratives: Salwa Hegazi, Kamel
432 mushira eid
Elshenawi, El-Aqqad, etc. In his narratives he assumes their roles by

quoting them, for example, directly or indirectly, by telling their jokes
to illustrate their sense of humor, and by reporting on their accomplish-
ments and modes of interaction as well. Throughout the interview his
voice is concrete, personalized, with a clear sense of humor. As a result,
he engages in different forms of discourse, e.g. telling jokes, narration,
and argumentation/persuasion, and takes on different personas, e.g. the
story-teller, the writer, the friend. He projects himself as a person who
values a sense of humor, friendship, nurturing, and being there for oth-
ers; these are the values he likes in others and praises them for having
such values. Haqqi, on the other hand, plays one role: the professional,
hence public, role of the intellectual, writer-artist, and administrator.
He speaks in one voice, that of this public persona. This single voice is
clearly reflected in his fush -based choice and the discourse of explica-
tion and narration. The personal he leaves to others, to his daughter and
his wife who complement this professional persona with the personal,
Haqqi the man: father, partner, and friend.
On some level, then, the two interviews turn out to be similar in topic,
voice, textuality, and overall informal setting. But they are actually very
different in terms of the roles played by the major figures. The overall
purpose in both programs is the creation of a public and private per-
sona for each interviewee. In so doing, the programs inform the audi-
ences about these individuals and entertain them as well. Entertainment
comes partly through the collage effect created during the interviewa
collage that extends to the language itself as reflected in the diversity of
speech styles and the many voices, images, and persona presented in the
programs, but not necessarily created by the two major figures. While
Haqqi plays one role and speaks in one voice, for example, the program
itself is multi-voiced. Likewise, while the Mansour program creates
boundaries separating its segments and topics, the speakers manage
to cross these boundaries merging the personal and the professional
through their conversational interaction and stylistic choices.
4. Conclusion: Hybrid people and hybrid contexts produce

hybrid language
The analysis presented in this article suggests that the media provides
a hybrid context whose purpose is to inform and entertain. Using the
interview as a genre that lends itself to this type of hybridity, I have

argued that the language used in this context is also a hybrid produced
by hybrid speakers, the degree of hybridity (mixture) being dependent
on a number of variables including purpose in communication, pro-
jections of identity, and the extent to which a speaker is comfortable
with this hybridity. The language variety produced in such contexts is
like a collage of features drawn from three linguistic spaces in Arabic:
one unspecified for variety, one specified for fush , and one specified
for Egyptian. This form of the language, like hybrids in general, can be
viewed as being neither fush nor Egyptian, since it does not conform to
either, or as both, since it includes features of both. I leave this question
open, for who is to say what, for example, a culturally-hybrid, bilingual
person would be? Is an Arab-American, for example, neither Arab nor
American, or is she/he both Arab and American? In concluding this
article I return instead to elaborate on the view of hybridity as a global
feature of text and code-switching as a local feature (meaningful on a
local level).
It is hard, if not impossible, to predict when or why a person would
switch from one language, or variety of a language, to another. A quick
glance at the sample texts included in section 2 provides enough evi-
dence to support this point. The discussion of these texts also shows that
sometimes the best one can do is identify areas (content, topics, criteria)
where people tend to move from one code to another more extensively
than other times. The end result is a performance where language variety
matches program goal as set by the satellite channel, speakers purpose,
topic, and intended audiences. At other times, however, it was possible
to identify a reason, stylistic or otherwise, for the movement form one
variety to another (e.g. Mansours segment 8).
I have used the terms hybridity and code-switching to capture this
difference. Others have used code-mixing instead of hybridity leaving
the latter to be inclusive of both (Hinnenkamp 2003). Either way two
levels or types of analysis are acknowledged: local analysis that focuses
on the specific motives for alternations, providing a rationale of local rel-
evance, and a global analysis that focuses on motives that can be found
in categories such as identity and group membership, providing a ratio-
nale of global relevance to the speakers involved. I have used hybrid-
ity in relation to global analysis to capture this cultural connection and
relevance to speaker. I hope to show in future research that for Ara-
bic, and possibly other alternations across codes of the same language,
434 mushira eid
the product may be better analyzed in terms of hybridity understood,

as suggested in this article, as a collage with fuzzy boundaries across
subspaces.
5. References
Awad, Luws. 1965. Mudakkirt tlib bita (Memoirs of a Student on a Study Abroad
Mission). Cairo: Muassasat Rz al-Ysuf.
Bakhtin, M.M. 1981. The Dialogic Imagination. Texas: University of Texas Press.
Bassiouney, Reem. 2006. Functions of Code Switching in Egypt. Evidence from mono-
logues. Leiden and Boston: Brill Academic Publishers.
Boussofara-Omar, Naima. 2006a. Diglossia. In Versteegh, Kees et al., eds. Encyclopedia
of Arabic Language and Linguistics, vol. 1, 629637.
. 2006b. Neither Third Language nor Middle Varieties but Arabic Diglossic Switch-
ing. Zeitschrift fr Arabische Linguistik/Journal of Arabic Linguistics 45:5580.
. 2004. Diglossia as Zones of Contact in the Media. Al-Arabiyya 37:101130.
Cachia, Pierre. 1990. An Overview of Modern Arabic Literature. Edinburgh: Edinburgh
University Press.
Eid, Mushira. 1982. The Non-Randomness of Diglossic Variation. Glossa. 16:1.
5484.
. 1988. Principles for Switching Between Standard and Egyptian Arabic. Al-Ara-
biyya 21, 5180.
. 1992. Directionality in Arabic-English Code-Switching. In Aleya Rouchdy, ed.,
The Arabic Language in America: A Sociolinguistic Study of a Growing Bilingual Com-
munity. Detroit: Wayne State University Press, 5071.
. 2002. Language is a Choice: Variation in Egyptian Womens Written Discourse.
In Aleya Rouchdy, ed., Language Contact and Language Conflict in Arabic. New York
and London: Routledge Curzon, 203232.
. 2004. Media Performances as Discourse Events. Arabic Media and Public Appear-
ance Forum, Center for the Advanced Study of Language (CASL), University of Mary-
land, College Park, June 810, 2004.
Haeri, Niloofar. 2003. Sacred Language. Ordinary People. New York: Palgrave Mac-
millan.
Hinnenkamp, Volker. 2003. Mixed Language Varieties of Migrant Adolescents and
the Discourse of Hybridity. Journal of Multilingual and Multicultural Development
24:1&2.
Mejdell, Gunvor. 2006. Code-Switching. In Versteegh, Kees et al., eds. Encyclopedia of
Arabic Language and Linguistics, vol. 1, 414421.
Muarrafa, Musta f. n.d. Qantara allad kafar. Cairo: Muassasat T ibat al-alwn
al-muttahida.
Schmidt, Richard. 1974. Sociostylistic Variation in Spoken Egyptian Arabic: A re-exami-
nation of the concept of diglossia. Ph.D. dissertation, Brown University.
Schulz, David. 1981. Diglossia and Variation in Formal Spoken Arabic in Egypt. Ph.D.
dissertation, University of Wisconsin-Madison.
Versteegh, Kees et al., eds. 2006. Encyclopedia of Arabic Language and Linguistics, vol. 1.
Leiden and Boston: Brill Academic Publishers.
THE USE OF MORPHOLOGICAL PATTERNS IN
ARABIC GRAMMARS OF TURKIC
Robert Ermers1
s-Hertogenbosch
1. Introduction
The use of meaningful (mor)phonological patterns2 (wazn, bunyat, pl.

awzn, abniyat) is very common in Arabic linguistic thinking. In this
contribution we examine the way these patterns were used in Arabic
grammars of Turkic, especially in the works of Ab H ayyn al-Andalus
(d. 745/1345) and Mahmd al-Kar (11th century).
2. Patterns in the Arabic linguistic tradition
In the Arabic grammatical tradition, each word is analyzed in terms of a

root that contains a certain number of base radicals. Most roots contain
three radicals, but there are also roots with two-, four or even five. For exam-
ple the verb karuma he was kind is built upon the three radicals k-r-m.
In lexicographical works k-r-m is often placed in the same three-radical
cluster as, say, r-k-m and m-k-r. A much later development is the alpha-
betical order in which k-r-m follows, e.g., k-r-, which starts with Ibn Fris
(d. 390/1000) (cf. Wild 1965, 66). Within a given entry the radical pat-
terns are further ordered according to the vowels they contain.
In order to make the vowels stand out more, the Arab scholars use the
paradigmatical root /f--l/ (cf. Wright 1986), which in our example kar-
uma yields faula. In this case faula is considered the wazn. In the lexi-
cographical sequence the pattern faula follows faala. For Arabic words
this system based on abstract patterns makes much sense. Verbs that
have the same pattern usually possess similar qualities, e.g. transitivity ;
verbs of the type faula are among other things usually intransitive and
1
r.ermers@hccnet.nl, P.O. Box 2176, 5202 CD, s-Hertogenbosch, Netherlands.
2
For the sake of brevity here referred to as morphological patterns.
436 robert ermers
most of them describe a state of mind or a quality (cf. e.g. Wright 1986,
I, 30, see also Versteegh 1992). Verbs with the same verbal pattern have
similar predictable morphological derivations, such as the formation of
the verbal noun masdar. For verbs of the type faula one possible form
of the masdar is falat e.g. karmat kindness. In regard to nouns of
the same pattern the forming of the plural is predictable in the same
way, the pattern filat may have the plural form fawil, e.g. qidatpl.
qawid base. Frb (d. 350/961) writes in his Dwn al-Adab that plu-
ral forms, derived from a pattern which usually yields such plurals, are
not included as entries (I:87). Also more or less predictable is the fact
that the pattern falat may refer to abstract nouns which are not usually
pluralized (similar examples in Irtif I 7397). In this respect the gram-
marians also indicate augmented radicals in Arabic patterns, e.g. af ala,
iftaala, istaf ala, in which -, -t-, and -stare augmented radicals which
add a specific meaning to the root.
Another obvious motive for collecting words in the same pattern, is
that this is handy when writing poetry. Words with the same pattern can
easily be used in rhyme schemes (cf. also Wild 1965, 66).
The use of morphological patterns thus appears to have two basic
objectives. The first is to determine which consonants in a given word are
basic and which are augmented, or more precisely: to determine which
ones are the basic radicals in a given word. This is for instance important
in establishing a words etymology. In the second place the pattern is
used for illustrating paradigmatic patterns, such as verbal conjugation,
the building of regular plural forms and some types of declension, in
which consonants (e.g. /w/ in /-uwna/ [m pl] and alif /"/ in /-a"t/ t [f
pl] are assigned special meanings (cf. Versteegh 1985). The scope of this
article is limited to the first objective.
For Turkic the general advantage of bringing together words with
similar patterns is evident too, but the morph(ono)logical arguments
do not apply. In Turkic languages meaningful elements usually have the
form of suffixes to a given stem, never infixes or prefixes, e.g. kas cut!,
kas-d he cut (3sg pt), kas-i-d they cut together, and qul slave
qul-lar slaves, qul-juq little slave, qul-juq-lar little slaves, etc. From a
given vowel sequence or consonant pattern no predictions as to plural
and verbal conjugations can be made and the same is true for words that
bear certain suffixes.3 In Arabic grammars of Turkic languages, some
authors make reference to morphological patterns too.
3
If we disregard the consonant assimilation and vowel harmony, that is.
morphological patterns in arabic grammars of turkic 437
3. Two Arabic treatises on Turkic
In the Arabic linguistic tradition the meaningful patterns are the basis
for the arrangement of lexicographical dictionaries. One important
work in this respect is Frbs (d. 350/961) lexicographical work Dwn
al-Adab (cf. Wild 1965, Haywood 1965). The pattern is also used in
Arabic grammars of Turkic. Especially, the Dwn Lut at-Turk (hence-
forth Dwn) compiled by the 11th century scholar Mahmd al-Kar
is set up in the same way as Frbs work (cf. Ermers 1999a, 19).
On top of the division in patterns, Kar, like Frb, uses a pri-
mary categorization of eight chapters (each of which is named Kitb
book), the final two (i.e. 7 and 8) being additions to Frbs division.
This superdivision reflects a primary interest in a morphological and
perhaps an orthographical arrangement. One item that is reflected in
the headings is the position of the glides, the socalled weak consonants,
and other non-regular consonants, such as hamza ().
The division of Dwn is as follows:
1. Kitb al-Hamz (29160): words with initial hamza,

2. Kitb as-Slim (160406): words having all sound consonants,
3. Kitb al-Mudaf (406445): words containing a geminate consonant
or two identical consonants,
4. Kitb al-Mitl (445493): words having an initial weak consonant,
i.e. y or w,
5. Kitb D awt at-talta (493535): words having a medial weak conso-
nant, i.e. y, w or alif,
6. Kitb D awt al-arbaa (535599): words having a final weak con-
sonant,
7. Kitb al-unna (599622): words containing [] or [n],
8. Kitb al-Jam bayn as-skinayn (622638): words containing clusters
of consonantal sounds that do not exist in Arabic, e.g. sirtl- climb
(635), td- turn down (633:15).
In this last chapter of his book Kar must have had in mind the cluster
/-ydt-/ in /tiydtiy/ he turned down, which includes the suffix -t for 3sg
past tense, which is the way Turkic verbs are given throughout Dwn.
Each chapter is further subdivided in separate sections on nouns and
verbs respectively. The more than 6700 entries are further collected in
minor paragraphs, often under a heading that contains a given morpho-
logical pattern (cf. Dankoff and Kelly 19825, Auezova 2005). Kar in
438 robert ermers
total lists some 109 main patterns, some of which contain subdivisions,
in total 143 patterns (see scheme 4 in appendix). The number of patterns
in Frbs Dwn al-Adab is much higher, about 1677.
Another scholar who deserves mention in this respect is Ab H ayyn
al-Andalus (d. 1345), who in his Kitb al-Idrk li-Lisn al-Atrk lists a
large number of meaningful morphological patterns according to which
in his view Turkic words are being construed. He mentions 91 nominal
patterns and 44 verbal ones (Idrk 101:10104:16), a total of 135. One
reason which makes it interesting to have a close look at Ab H ayyns
views on Turkic is that they can easily be compared to those on Arabic
through his oeuvre on Arabic grammar and Quranic exegesis. One of
his most prominent works on grammar is Irtif ad-D arab min Lisn
al-Arab, whose setup bears a strong resemblance with Idrk. In Irtif
(I:2072) there is a similar, but much more lengthy and elaborate sec-
tion on morphological patterns.
4. The application of morphological patterns
In his classification of verbal and nominal roots (asl) in Idrk, Ab

H ayyn differentiates between theoretically possible patterns (qisma
aqliyya) and those that are actually used (masm).4
In Idrk (101:13) Ab H ayyn writes that for biradical nouns there
are 12 theoretically possible forms, all of which are used. These are the
following: (1) fa like san you, (2) fi like kim who, (3) fu like yuz face,
(4) fau, (5) faa, (6) fai, (7) fuu, (8) fii, (9) fua, (10) fia, (11) fiu, (12)
fui. Obviously these can be divided into four groups of three sequences
of vowel patterns each, while after some reshuffling a more transparent
sequence results:
CVC (1) fa (2) fi (3) fu
CaCV (5) faa (6) fai (4) fau
CuCV (9) fua (12) fui (7) fuu
CiCV (10) fia (8) fii (11) fiu
SCHEME 1 PATTERNS OF BI-RADICAL NOUNS IN IDRK
4
Wild (1965, 37) mentions a similar distinction in Kitb al-Ayn between mustamal
used and muhmal not used, litt. neglected.
It is difficult to tell whether Ab H ayyn has a preference for a certain

sequence of the vowels. The regular sequence, though is a, u, i, since in
Arabic grammar a is considered lighter than u and i. The first three
items are listed in the sequence a-i-u, but Ab H ayyn refrains from
holding on to this from the second set onwards.
For nouns5 the following patterns are given:
CVC (1) fal, (3) ful, (2) fil,

"ard back kurt tree sirt back
CVCVC V (in Arabic (7) faal, (16) fuul, (18) fial
analysis CVCVCVC) (9) faal, (17) fual
(8) faal
CVCVC (19) faal, (22) fual, (25) fial,
(21) faul, (23) fuul (26) fiul,
(20) fail (24) fiil
CVCC V (in Arabic (4) fal, (13) ful, (10) fil,
analysis: CVCCVC) (6) fal, (14) ful, (12) fil,
(5) fal (15) ful (11) fil
SCHEME 2 PATTERNS OF TRI-RADICAL NOUNS IN IDRK
In scheme 1 and 2 Ab H ayyn describes nouns, the verbs being dealt
with at the end of the section (103:13). Note that in the second listing (cf.
scheme 2) Ab H ayyn is more consequent in holding to the sequence
a-i-u, which seems to evolve, than in the first. Further, it seems that
some of the patterns are incomplete, for example, the nominal pattern
fu-: there is no form fuil* which completes the set (16, 17). Likewise,
there is only one nominal triradical pattern that starts with fi . . . (18),
where one would expect three, the other two being fiul* and fial*. In
the third place, there is no noun that is formed according the pattern
fuil* (22,23). In sum, only four patterns appear to be lacking from this
overview. Nevertheless, it is easy to invent more patterns which would
fit in this triradical scheme, e.g. fual*, fual* and fual*.
5
Between parentheses the original sequence in Idrk.
440 robert ermers
For verbs Ab H ayyn gives the following patterns:

Number of radicals fa- fu- fi-
Uni-radical
CV (2) fa (3) fu (1) fi
Bi-radical
CVC (5) fa (4) fu (6) fi
CVCV (8) faa (12) fua (10) fia
(13) fai (7) fuu, (9) fii
(11) fui
Tri-radical
CVCC (14) fal (16) ful (15) fil
CVCVC (17) faal (22) fual (23) fial
(19) fail (20) fuul (21) fiil
(18) faul
CVCCV (24) fala (26) fula (25) fila
(27) fali (28) fuli
CVCVCV (32) faala (31) fuila (29) fiala
(30) fiila
Tetra-radical
CVCCVC (35) falal (37) fulal (38) filil
(34) falil (36) fulul
(33) falul
CVCCVCV (39) falula (43) fulila
(40) falala
CVCVCCV (41) faalla (44) fualla (46) fiilla
(42) failla (45) fuulla
Penta-radical
CVCVCCVC (47) faallal
CVCCVCCV (48) falalla (49) fililla
CVCCVCVC (50) falulal
SCHEME 3 PATTERNS OF TURKIC VERBS IN IDRK
Note that there are no long vowels in the verbs.
5. Primary and augmented radicals
The rules are, at least so it seems, not applied systematically. In the first
place augmented radicals are not reflected in the pattern and conversely,
radicals are presented in such a way that one thinks that they are con-
sidered augments. Ab H ayyn dedicates a large section to augmented
radicals (Idrk 111:17116:13), in which he incidentally investigates
the etymology of some compounds (Idrk 103:12), but the augmenta-
tion is not reflected in patterns in the way this is regularly done in the
Arabic linguistic tradition. The nouns and verbs he cites as belonging
to a given pattern are not analysed further in terms of basic radicals.
For example, "usurmaq fart could be analysed as fuul-maq*, since it is
an obvious derivative of the verbal stem "usur-. The same holds for the
noun ta arjuq sack, for which the pattern faal-juq* could be posited
(Idrk 103:2,5). Instead the former is attributed the pattern fuullal, and
the latter faallul, the regular way of dealing with hexa- and heptaradical
words, but neither of which gives any clues as to the status the radicals
of the respective suffixes possess.
Ab H ayyns formulates his goal in arranging words according to
patterns as follows: It is necessary to study the structures of each of
them, so that the primary radical (al-h arf al-asl) may be distinguished
from the augmented (zid). Only then can the primary radical be com-
pared with the primary radical and the augmented radical with the
augmented radical, (Idrk 104:15, translation Ermers 1999b). In our
introduction we already briefly touched upon the importance of being
able to differentiate between original radicals and augmented ones, e.g.
in af ala, iftaala, istafala, mafalat. The augmented radicals are called
h urf al-man particles of meaning (cf. Versteegh 1995, 120). One
would expect the Turkic augmented radicals in a similar way to be indi-
cated in the pattern, e.g. faal-dir* causative, faal-il* passive, faal-in*
reflexive form.
Kar takes a similar position in his Dwn. Much in the same way
Ab H ayyn does in Idrk he discusses in a separate section the aug-
mented radicals for nouns and verbs (Dwn 1316). In the nouns he
deals with the glides and hamza only, e.g. tar bag, "adir stallion,
along with -n, as in bazn /baza"n/ hammer, in which the added
meaning of the consonant alif /"/ is not obvious. The augments in the
verbs are dealt with in more detail. Kar mentions as many as eleven
augmented consonants, i.e. 1. alif, 2. t, 3. r, 4. s, 5. , 6. q, 7. k, 8. l, 9. n, 10.
l (lm-alif), 11. y, all of which are added because of [certain] mean-
ings ( fa-kull wh idin minh tuzd li-man, Dwn 14:2). Here we
suffice with a few examples, e.g. t in its causative meaning for the tran-
sitivity of the verb (li-t-tadiyat al-fil, 14:5) in "ari-t-t he dried, which
basic form is (asluhu) "ar-d it dried up. Another example is s, which,
according to Kar, expresses the meaning of wishing to carry out
that action/verb (man at-tamman li-iqmat dlik al-fil, 14:10 ), e.g.
suv "ij-a-sa-d he wished to drink ("ij-) water (suv). The -s- is also used
when the action of the verb is not actual (wa-lam yakun wuq dlika
442 robert ermers
al-fil minhu h aqqatan), e.g. "ul mank kul-um-sin-d ([uriya annahu]

yadh aku ilayya) [it seemed as if] he smiled at me (in which kul- means
laugh and kul-um-sin smile).
Even though patterns play a crucial role in Dwn, only very occa-
sionally the augmented radicals are made visible, to the extent that even
the ones dealt with in this section of the work are never referred to.
The most obvious example of a suffix that is visible, is -d, the 3sg past
tense suffix that Kar adds to verbs, e.g. faal-d (Dwn 305). The
use of this form is most likely a calque from the Arabic grammatical
tradition, where it is the form that is most unmarked (like karuma, lit.
he was kind), and one could therefore argue that /-diy/ is not really
considered an augmentin any case it is not mentioned in his listing
(Dwn 135).6 A more interesting example is the pattern fiinl (see no.
13 in scheme 4, in appendix, Dwn 82), which contains nouns like, e.g.
itind what is being pushed, and aqind (suv) running (water). Based
on this patternif we first disregard a in aq-f- stands for , and -- for
t/q, the stems are reconstructed as it- push and aq- run (of water)
respectively, both of which are common in Turkic, with a suffix -nd. But
in regard to the augments itself more can be said.
pattern f i i n l i y
Turkic 1 i t i n d i y
Turkic 2 a q i n d i y
All examples Kar supplies in this subsection end in -ind, which in

spite of the categorization as triradical, suggests that in his view /-l-/,
which then on a metalevel stands for -d-, is part of a triradical root and
that only -n- and - are augments. If /-l-/ is nevertheless considered a
literal part of the augment, he still has to explain under which phono-
logical circumstances /-nl-/ is realised as -nd. In Arabic grammar, the
phonological changes in consonants are typically explained by means of
qalb, the exchange of one consonant for another, based on a set of rules,
rather than insertion of meaningful consonants. In any case, a pattern of
the type find* would have been a more logical choice.
The patterns falal /fal1al2/ and fanal (Dwn 611) occur in one head-
ing under a section of tetraradical nouns in the book of Nasal Words.
This heading introduces nouns like tauj /tankuj/ something which
6
Perhaps for this reason Ab H ayyn does not include -d in his references to Tur-
kic verbs; he applies the patterns to the stem only, e.g. fu for "ub kiss!, fal for tart for
weigh!.
rises in the air (EDT7 [520] te), and qujr /qujnka"r/ ram. In
these nouns , which in both words is represented by /-nk-/, counts for
two consonants, e.g.:8
pattern 1 f a l1 a l2
pattern 2 f a n a l
Turkic t a n k u j
Other arguments for this reasoning are in the first place the fact that
according to the morphological pattern the augment is /-n-//-k-/
being perhaps more basiceven though Kar does not specify to
which one of both patterns the nouns in fact belong. In the second place
nouns like taik /tankik/ air and saak /sank1ak2/ cup and suuk /
sunk1uk2/ bone (Dwn 604) are considered to have a doubled con-
sonantif we take into consideration the directly preceding heading.
On a phonological level this suggests that /-k-/ is considered a separate
consonant, whereas from a morphological point of view it is probably
regarded as more basic than /-n-/. All this points to the assumption that
n and k are regarded as separate consonants.
In spite of this, nouns that are quite comparable to those above, like
taut (Dwn 603) /tankut/ name of a Turkic tribe and suqur /sunk-
qur/ falcon, kaa /kanka/ advice are assigned a triradical pattern of
the type faal, faul, fail.
pattern f a u l
Turkic t a nk u t
In those instances /--/ corresponds with the cluster /-nk-/; this never-
theless appears to be a more general morphology for words with .
In Dwn there are a few instances in which the problem is the other
way around: augments are made explicit when there is no apparent need
for it. The preceding case of fanal may represent such an instance, since
there is no apparent augmented meaning. A speculation along these
lines is that, perhaps Kar in fact was in doubt as to which pattern
to use here, so he wrote down both. Other examples are the pattern
faland that is applied to such nouns as udrund selected and afdind
7
EDT, here and henceforth, refers to: Etymological Dictionary of Pre-Thirteenth-
Century Turkish (Clauson 1972).
8
Clauson in EDT apparently supposes that, in view of the number of radicals, /-n-/
alone must stand for , whilein his viewa superfluous /-k-/ (for g) is added, which
results in the erroneous form teg.
444 robert ermers
collected (Dwn 84). If we for both nouns take /-l-/ in fal-and as the
final consonant of the stem, which seems to be the case, it follows that
the paradigmal suffix or augment must have the form -nd. The pro-
jected stems then must be udr-* and afd-* respectively.9 Here we have
a similar problem as in regard to the pattern fiinl, which we discussed
earlier. If instead Kar considered /-l-/ as the first consonant of the
suffix, one would have expected him to propose the pattern fa-dand*,
because otherwise the occurrence of -d- and -r- on the position of /-l-/
is not accounted for.
Kars discussion of the few internal augments in nouns he gives
is interesting too. Here the augments are placed between the radicals of
the stem, e.g. fawal for yuwlij lambswool (Dwn 456), fayal for nouns
like qaymaj lambswool (Dwn 522), both of which are considered
triradical. In these cases /w/ and /y/ respectively again, are presented
as augments, quite comparable to /y/ in fuayl, a regular Arabic pattern
for diminutives, e.g. fulays small coin (< fils). But the meaning of the
inserted glides here is quite unclear. If the comparison to fuayl would
hold, this suggests that these would be etymologically related to forms
like yulij* and qamaj*, which could not be found. In view of the fact
that the augment in these instances does not seem to carry a particu-
lar meaning, one could argue that the pattern faalal, or perhaps faalil
would have served as well, and the question remains why Kar did
not apply them here instead of proposing the insertion of augments.
There seems to have been a general degree of confusion as to the
applicability of morphological patterns in other works as well. Talmon
(1997, 172) in this respect notes that an authoritative source like al-Xall
(d. 791/175?) in his Kitb al-Ayn is not very consequent in the use of
patterns either: muatft, which according to one view takes the pat-
tern (wazn) of mufalt, and according to another mufalat. A similar
example, still according to Talmon, is andawat, which belongs to two
patterns: fanalat and faallwat. Some consonants, especially hamza, alif,
ww, y and nn are sometimes referred to as basic (asl) and others
as non-asl. In Irtif we came across a similar discrepancy, sanbitat a
period of time is assigned the pattern falatat, but Ab H ayyn acknowl-
edges that others apply fanalat instead: it is said that its pattern is . . .
9
Only in the first case this agrees with EDT [70] dr-, whereas for the second Clau-
son suggests evdin- pluck [7].
(wa-qla waznuh), obviously depending on how /-n-/ is interpreted. In

another instance here too the arbitrariness in assigning the patterns is
obvious (Irtif I:33). For example, firsan is attributed the pattern filan
(Irtif I:33), even though /-n/ cannot be considered an augmented con-
sonant. In this way, the use of the pattern does not give any clues for
determining which consonant is augmented. In Frbs Dwn al-Adab
similar examples are found, e.g. anbajn be great which follows the
pattern afaln (Dwn al-Adab I:280). In this example the form with
-n obviously is not a derivative from based on a form (anbaj), without
the suffix. Based on this brief survey we may conclude that Kars
approach is not very exceptional.
6. Other lacunae in the system
Earlier we mentioned that Kar uses some 109 headings in which

he mentions patterns, sometimes more than one. The total numbers of
patterns he refers to is 146. Not surprisingly, in Dwn the same pattern
recurs several times through the work. Within the different chapters
the same patterns almost inevitably recur. To mention only a few, af l
(see no. 4 in scheme 4 in the appendix) is the same as 14; 2 is identical
with 29 and 51; 5 with 32 and 60, more examples of similar patterns in
scheme 5 in the appendix).
Let us examine two occurrences of the same pattern. The pattern afl
(see no. 4) refers to "arqr mountain sheep (Dwn 71), but it also refers
to a noun like "ar fishing hook (see no. 14, Dwn 83).
In book I, Kitb al-Hamz, there is no apparent reason for giving any
pattern with initial hamza /-/. The chapter heading itself suggests that
the initial hamza /-/ is basic and thus should be represented by /f-/. The
only reason for quoting hamza could be that it is augmented, i.e. when
it is not a basic consonant. Yet seven (i.e. the items 3, 4, 11, 12, 14, 17
and 23see appendix scheme 4) out of the 24 morphological patterns
begin with -, e.g. (3) af al, (4) afl, uf l, if l, (11) afl, etc. where
one would have expected f- instead. Let us study two examples in more
detail:
A. "arqr 1. Book I, words with initial hamza, 2. nouns 3. vocalization

of middle radical (i.e. q), 4. Section on af l (and other patterns)
with a vowel on the middle radical, 5. R.
446 robert ermers
B. "ar 1. Book I, words with initial hamza, 2. tetraradical nouns, 3.

pattern af l, 4. a double sequence of the same radical within one
root (i.e. ).
B is explicitly placed under the heading of tetraradical words; it then

follows that A is considered triradical. Apart from being explicitly
quoted in the pattern, the initial hamza in A is also the only reasonable
candidate for being considered an augment, which is in contradiction
with the main chapter heading. It is difficult to speculate on the reason-
ing Kar follows here that would account for these two instances of
analysis.
There is also a small number of words that have a double entry in the
work in different places. One is kuk light shadow (EDT 753). Appar-
ently Kar has interpreted the pattern of this word in different ways:
A. 1. Book II regular words, 2. Nouns, 3. Words with added glide

between the second () and third radical (l), 4. The pattern fal and
the like, 5. (nouns with) K (Dwn 225:8; Auezova 2005, no. 2606)
B. 1. Book V words with a middle weak consonant, 2. [Nouns], 3. Sec-
tions on difficult (?) words, 4. The pattern fal and the like (871),
5. (nouns with) K (Dwn 521:12; Auezova 2005, no. 5500)
If we apply the patterns fal /faa"la"/ (1) and fal /faa"luw/ (2) to
kuk /kuiyka"/, in which /f-/ stands for /k/, /--/ for /--/, etc., the fol-
lowing scheme emerges:10
pattern 1 f a a '' l a ''
pattern 2 f a a '' l u w
Turkic k u i y k a ''
The application of either pattern suggests that -y- is not basic, but
inserted as an augment between -- and -k-. The goal here is of course
obvious: lengthening of /i/, but -y- is still not accounted for as a mean-
ingful augment. B is placed in the main division, the Book of words with
a glide as a middle radical out of these.11 If -y- is not considered a mean-
ingful augment, it must be an instance of prosodic lenghtening. This
same Book starts, as expected, with nouns like th th /ta"h/, but they
10
We will discuss below the fact that the vowels of the schemes do not correspond at
all with those in the Turkic, and that /-"-/ is apparently realized as y, and /-w/ as ".
11
If we correctly interpret dawt at-taltat, cf. Dwn al-Adab I 76,80.
are to our surprise considered biradical (t uniyya). In other words:

the alif is, in spite of the heading, not considered a basic radical, but
apparently a mere instrument for indicating lengthening. This suspicion
is reinforced by Kars following remark elsewhere the lenghtening
consonants may be elided when they are pronounced quickly; in that
case one says tah tah (wa-h urf al-ln suqita min inda surat an-nutq
bih, fa-yuqlu tah tah, Dwn 439:12). Thus the status of the glide still
remains, at least in these words, uncertain.
Finally there is the status of the vowels within the pattern. Ab H ayyn
is both in Idrk and Irtif quite accurate in his selection of matching
verbs and nouns to a given pattern, and the same holds for Frb in
his Dwn al-Adab. Whenever Kar presents a given pattern in a
heading, he nevertheless lists almost everywhere words which follow
an entirely different vowel pattern. To give but a few examples: fal is
applied to yij tree which in fact follows fil, yipr musk in fact fil,
yulr rein in fact ful (see item 60 in the scheme below; Dwn 456:3).
These instances are often covered with subtitles like f h araktihi literally
in the vowels (also 84:11) after the given pattern. In this way, patterns
containing /a"/ (indicating ) are also often applied to words with /
uw/ or /iy/, e.g. fal for yant response, which in fact has the pattern
fal (Dwn 456:3). It seems that Kar throughout his work uses the
patterns for indicating the mere position of the vowels and glides rather
than as a precise pattern.
In some instances Kars presentation is in contradiction with the
regular practice. When Kar introduces the pattern faal (cf. Scheme
4, no. 76), explicitly described as with the first and second consonant
vocalized (muh arraka al-f wa-l-ayn), and further specified as from
the weak words which contain a w (dawt al-ww) (Dwn 507:6).
However, w is not mentioned at all; Kar exclusively lists nouns of the
type (CC), such as qb /qa"b/ pot, tz /ta"z/ bald (Dwn 509:6). It is
unclear how Kar applies this pattern here, since alif ("), which must
be the second or middle consonant, cannot carry a vowel; it can only be
preceded by /a/.
7. The glides
The status of the glides is an important issue in morphological pat-

terns. The patterns were a great aid to the Arab grammarians for determin-
ing the status of the glides. In Arabic grammatical theory the glides, i.e.
448 robert ermers
/"/ ( alif), /w/ and /y/, are considered consonants. For example, bbun
door is derived from a projected form /ba1wa2bun/* of which the under-
lying pattern is /fa1a2lun/, by replacement of /w/ by //, which results
in /b1aa2bun/*, and a subsequent elision of /a2/: /ba1bun/. Even though
glides are used as instruments for indicating lengthening of the preced-
ing vowels, they are not considered vowels themselves. In this way, alif
(/"/), whose only function is expressing lengthening of /a/, is a con-
sonant as well. The glides, like a number of other consonants (e.g. /"/,
/n/, /t/, etc.), can be inserted or prefixed as augments. For example, the
verbal form qtala he battled is, according to the Arabic grammatical
tradition, derived from qatala he killed by inserting an alif /"/ accord-
ing to the pattern /fa"ala/, rather than by lengthening /a/ (cf. Bohas
1982, 168).
This does not mean that the concept of lengthening does not exist. In
a practical sense the concept of long vowel does exist in Arabic gram-
matical theory, but in the analysis a sharp distinction is made between
meaningful augments on the one hand, i.e. the insertion of phonemes,
and prosodic lengthening, i.e. lengthening for non-phonemical reasons
(ib) on the other. Or, in other words, the analysis of a given word is
based on the question whether a given long vowel is the result of either
ziydat (meaningful addition, or insertion) or of ib (prosodic length-
ening). If we think further along the lines of Arabic linguistic reasoning,
there is hardly a plausible reason to give for reflecting prosodic length-
ening in orthography. This is especially true for lexicography. Prosodic
lengthening typically occurs when the word is put in a context, not in
isolation and it is made explicit in poetry, not in regular prose.
The question of how to interpret lengthened vowels is also an issue in
our source material on Turkic languages. Ab H ayyn starts his expos
on this problem with the uniradical nouns s water, which he writes as
/suw/, and y /ya"/ bow, j /jiy/, moist (?). His almost casual remark
that the semiconsonants (h urf al-liyn wa-l-madd) [w], ["] and [y] form
no part of the root (asl, 101, 12), but rather arise from lengthening
of the vowels (nawi an ib al-h arakt) now gains importance. In
the case of s, y and j, Ab H ayyn says, not f /fuw/ or y /fa"/ is
intended, but rather fu+ and fa+, respectively (in which we use + for
prosodic lengthening). In this respect qis (qisa") short too must be
ranged under the pattern fia+. In spite of all this, the author still pro-
duces the patterns fal for words like barj all, faal for taraz scales
(see the patterns 418 in the listing given above). In these patterns the
glides, which, according to the author, here serve to indicate prosodic
lengthening, might as well be mistakenly regarded as a meaningful part

of the pattern. It seems that Ab H ayyn clearly was in doubt as to how
to reflect the morphological patterns.
The same ambivalence towards long vowels and their morphological
classification is found in Kars Dwn. Kar says that in general it
is possible to delete the glides (Dwn, 516:6) as in Arabic, in the way
zin decorate! is formed by eliding /y/ from the form /ziyn/*. The reason
he says so, no doubt is the fact that in Uygur script vowels are indicated
by means of the glides, without the possibility, as in the Arabic script, to
distinguish between long and short vowels (see Ermers 1999a, 130).
In regard of the previous one is inclined to think that vowel length is
unlikely to be considered a phoneme in Kars lexicon. This is not the
case. Kar takes either the long form or the short as an entry. In some
instances he gives the other as a dialectal exception to the general rule.
In other words, he deals with the short and long forms in separate para-
digmal forms. A nice illustration is the pair "aj- open (Dwn 92:4) and
"aj- be hungry (Dwn 95:17). Each of them has a separate entry in the
chapter on the biradical verbswhere // counts as the first consonant

but only "aj- be hungry is dealt with in the subdivision (lit.) Insufficient
(al-manqs). A similar example is "z, which has a number of meanings,
e.g. fat; deaf; self; valley; heart (Dwn 35:1), "z fat (Dwn 30:6) and
the verbs "z- cut (Dwn 93:1), and "z- pass first (Dwn 96:11), which
is also listed under the subheading al-manqs. In the case of "it/"t push/
bless (Dwn 95:10), which occur under the same pattern, lengthening
may have been used to indicate a different pronunciation altogether, e.g.
t- (cf. EDT 36). Even though of one and the same entry more than one
version does occur throughout the work, from the separate treatment it
is evident that for Kar length is a phoneme. One could even make
the assumption that the fact that length can be made explicit in Arabic
script gave Kar the opportunity to arrange his lexicon in a different
way than he could have in the Uygur script.
8. Summary
In this article we made a brief inventory of the way Arabic grammar-

ians applied morphological patterns to Turkic words. In Arabic linguis-
tic thinking, the patterns are a convenient instrument for indicating
which consonants in a word are basic and which are not. Non-basic
450 robert ermers
consonants are usually attributed a special meaning, such as, e.g., causa-
tivity or diminutive, which is added to the root. Furthermore, based
on the pattern words that contain glides, such as bbun can easily be
reconstructed in terms of regular patterns (*/bawabun/). An additional
motivation for lexicographical interests appears to have been bringing
together words with the same or a similar pattern. This comes in handy
for writing poetry, since words with the same pattern can be used in the
same rhyme scheme. Surprisingly, the two sources we examined, Diwn
Lut at-Turk (11th century) and Kitb al-Idrk li-Lisn al-Atrk (14th
century) hardly use Turkic meaningful consonants and suffixes in com-
bination with the patterns, even though the authors are clearly aware of
them. This appears to be a general feature of lexicographical works: the
patterns seem to be applied more or less arbitrarily. In Dwn Kar
goes one step further, in that he roughly indicates the position of the
vowels in a given pattern by means of /a/, while at the same time in those
instances he deals with words with u, and i, .
9. Appendix
9.1 An overview of the morphological patterns in Dwn Lut at-Turk

Scheme 4 below contains the patterns mentioned in the chapter and sec-
tion headings in Dwn. Sometimes more than one morphological pat-
tern is mentioned (e.g. in 1, 2, 4, 9, 28, etc.). The roman numerals refer
to the number of radicals mentioned in the closest preceding heading.
Book I, Kitb al-Hamz (29160): words with initial hamza

Nouns (Dwn 33) III (1) fal, ful, fil, (2) faal, faul, fail, (3) afal, (4) afl, ufl,
ifl, (5) fal, fal, (6) fal, fal, (7) fal, (8) fuul, (9)
fainl, (10) fal, ful, fil, (11) afl, (12) af aln,
(13) fiinl
IV (14) af l, (15) fal, (16) faall, (17) aflil, (18) fall,
(19) faland, (20) fuln,
V (21) faalal, (22) faaln, (23) afalal, (24) fulil, (25) falil
(Dwn 91)
Verbs (26) faal (Dwn 97), (27) falal-d
Book II, Kitb as-Slim (160406): words having all sound consonants
Nouns (Dwn 172) (28) fal, ful, fil, (29) faal, faul, fail, (30) fil, (31) fl,
(32) fal, fal, fal, (33) fal, fal, fal, (34) faln, fuln,
filn, (35) fal, (36) faln, faln, (37) faanl, (38)
fulul, (39) falal, fall, fulal,
V? (40) falil, V? (41) faalal, faall, (42) fulul, (43) fulund,
(44) faalal, (45) faalln,
VI (46) falall, (47) faalalal
Verbs (Dwn 305) III (48) faal-d
IV (49) falal-d
V (50) faalal-d
Book III, Kitb al-Mudaf (406445): words containing a geminate or two identical
consonants
Nouns III (51) faal
IV (52) fall
V (53) faalal
Verbs III (54) faal-d (Dwn 415)
IV (55) falan-d
Book IV, Kitb al-Mitl (445493): words with an initial weak consonant, y or w
Nouns III? (56) fal, ful, fil (Dwn 447), (57) faal, faul, fail, (58)
falal, (59) fawal, (60) fal, (61) fal, (62) fal, (63) faln,
(64) faln, (65) faanl
IV (66) falal, fall, (67) falil, (68) faall
V (69) faalal, (70) faaln, (71) falalin (Dwn 468)
Verbs (72) faal-d (Dwn 473)
IV (73) falal-d
V (74) faalal-d (Dwn 491)
Book V, Kitb D awt at-Talta (493535): words with a medial weak consonant, i.e. y,
w or alif
Nouns (75) fal, ful, fil (Dwn 494), (76) faal, (77) fil, (78) fal,
(79)
fal, (80) fal, (81) faln, (82) fayal, (83) falil
V (84) faalal (Dwn 523)
Verbs (85) faal-d (Dwn 526), (86) falal-d (Dwn 529)
Book VI, Kitb D awt al-Arbaa (535599): words having a final weak consonant
Nouns (87) faal, faul, fail (Dwn 540), (88) fil, (89) fal, (90) fal,
(91) fall
VI (92) falal, (93) faalll (Dwn 552)
Verbs III (94) faal-d (Dwn 565)
IV (95) falal-d, (96) fal-d, fal-d, fal-d
V (97) faal-d, V (98) falal-d (Dwn 597)
452 robert ermers
Book VII, Kitb al-unna (599622): words containing [] or [n]

Nouns (99) faal, faul, fail (Dwn 602) (100) fal; (101) fal,
(102) fal
IV (103) falal, fanal, (104) falal,
IV (105) faanll (Dwn 613)
Verbs (106) faal-d (Dwn 615)
IV (107) falal-d, (108) fall-d, (109) fal-d (Dwn 619)
Book VIII, Kitb al-Jam bayn as-skinayn (622638): words containing clusters of
consonantal sounds, does not refer to morphological patterns.
SCHEME 4 MORPHOLOGICAL PATTERNS EXPLICITLY
MENTIONED IN DWN
The patterns mentioned above recur in different places throughout the

work. The scheme below indicates which patterns are repeated:
pattern occurs in items no. in scheme 4
1 afl 4, 14
2 faal 2, 29, 51, 57, 76, 87, 99
3 fal 5, 32, 60, 78, 89, 100
4 fal 35, 96 ww
5 falil 40, 67, 83
6 fal 10, 80, 102
7 faalal 21, 41, 44, 53, 69, 84
8 faalal-d 50, 74
9 faaln 22, 70
10 faal-d 48, 54, 72, 85, 94, 106
11 faall 41, 68
12 fil 30, 77, 88
13 fal, ful, fil 1, 28, 56, 75
18 fal 6, 33, 61, 79, 90, 101
16 faln 34, 63, 81
17 faland 19 (noun), 55 (verb)
14 falal 35, 58, 66, 103
19 fall 39, 52, 66, 91, 108 (verb)
15 falald 27, 49, 73, 86, 95, 107
20 fulul 38, 42
SCHEME 5 DOUBLE OCCURRENCES OF MORPHOLOGICAL
PATTERNS IN IN DWN
10. References
10.1 Primary sources

al-Andalus, Ab H ayyn, Muhammad b. Ysuf. Irtif ad-D arab min Lisn al-Arab. (3
vol.) Musta f Ahmad an-Namms, ed. 198419871989. (I: Cairo: Matbaat an-Nar
ad-D ahab; II/III: Cairo: Mat baa al-Madan).
Kitb al-Idrk li-Lisn al-Atrk. Amet Caferolu, ed. 1931. Istanbul: Evkaf
Matbaas.
al-Frb, Ab Ibrhm Ishq b. Ibrhm. Dwn al-Adab. Ahmad Muxtr Umar, ed.
197478 (4 vol.) Cairo: al-Haya al-mma.
al-Kar, Mahmd b. al-H usayn b. Muhammad. Diwn Lut at-Turk. Facsimile edi-
tion of the MS; Ankara: Turkish Ministry of Culture. 1990.
al-Xall = Ab Abd ar-Rahmn al-Xall b. Ahmad al-Farhid. Kitb al-Ayn. al-Maxzm
and as-Smarr, ed. Beirut: al-Alam.

Auezova, Zifa. M. 2005. Diwan Lughat at-Turk. Translated (into Russian), with intro-
duction by Z.-A. Auezova; indices by R. Ermers. Almaty: Daik Press.
Bohas, Georges. 1982. Contribution ltude de la mthode des grammairiens arabes en
morphologie et en phonologie daprs grammairiens tardifs. Thse Universit de Paris
III (1979).
Clauson, Gerard. 1972. EDT = An Etymological Dictionary of Pre-Thirteenth-Century
Turkish. Oxford.
Dankoff, Robert and Kelly, James. 19825. Compendium of the Turkic Dialects. Trans-
lation of Mah mud al-Kars Dwn Lut at-Turk. 3 Vol.: I (1982); II (1984); III
(1985) indices to Vol. I and II. Harvard: University Press.
Ermers, Robert. 1999a. Arabic Grammars of Turkic. The Arabic Linguistic Model Applied
to Foreign Languages. Leiden: E.J. Brill.
. 1999b. Translation of Ab H ayyn al-Andaluss Kitb al-Idrk li-Lisn al-Atrk.
Leiden: E.J. Brill.
Haywood, John 1965. Arabic Lexicography. Leiden: E.J. Brill.
Owens, Jonathan. 1988. The Foundations of Grammar: An introduction to medieval Ara-
bic grammatical theory. Amsterdam: Benjamins
Talmon, Rafael. 1997. Arabic Grammar in its Formative Age. Kitb al-Ayn and its attri-
bution to H all b. Ah mad. Leiden, New York, Kln: Brill.
Versteegh, Kees. 1985. The development of argumentation in Arabic grammar: The
declension of the dual and the plural. Studies in the History of Arabic Grammar, ed. by
Hartmut Bobzin and Kees Versteegh, 152173. Wiesbaden: O. Harrassowitz.
. 1992. Grammar and rhetoric: Jurjn on the verbs of admiration. Studies in Semitic
Linguistics in honor of Joshua Blau, M. Bar-Asher et al., eds. Jerusalem: Hebrew Uni-
versity of Jerusalem. 113133.
. 1995. The explanation of Linguistic Causes. Az-Zajjjs theory of grammar. Intro-
duction, translation, commentary. Amsterdam: Benjamins.
Wild, Stefan. 1965. Das Kitb al-Ain und die arabische Lexikographie. Wiesbaden: Har-
rassowitz.
Wright, William. 1986. A Grammar of the Arabic Language. Translated from the German
of Caspari. Cambridge: University Press. (2 vols) (First edited in 1859/1862).
LEXICAL GAPS IN ARABIC LEXICOGRAPHY WITH
EVIDENCE FROM ARABIC DICTIONARIES
Jan Hoogland
Nijmegen University
1. Introduction
Between 1997 and 2003 we compiled with a dedicated group of specialists

at the Arabic department of the University of Nijmegen a twin set of
Arabic-Dutch and Dutch-Arabic dictionaries. Kees Versteegh, Manfred
Woidich, and myself were the responsible editors.1
It may be clear, that the compilation of these dictionaries not only
resulted in their publication, but also in the gathering of a lot of valuable
experience leading to some striking conclusions. One of these was the
observation that, in the Dutch-Arabic volume, many source (Dutch)
language units could not be paired with an equivalent in Arabic, since
they appeared to be non-existent in Arabic. As a matter of fact, almost
25% of all Dutch entries (words or expressions) could only be translated
by means of paraphrases. Paraphrased descriptions are, in most cases,
explanations and not direct equivalents of the source language word or
expression.
From this observation the question did arise: do these untranslatable
units represent lexical gaps in Arabic? In other words: do they represent
concepts without any lexicographically acceptable one-to-one translation
in Arabic?
It goes without saying that all dictionary compilers also benefit from
the work of others. Also the authors of the Nijmegen dictionary referred
to a large number of existing monolingual, bilingual and multilingual
dictionaries in order to single out adequate equivalents for the Dutch
entries selected to occur in the dictionary. Too often we concluded
that the other dictionaries also contained paraphrased descriptions or
definitions instead of one-to-one translations.
1
The whole process of compiling has been described on the project site (www.let.
ru.nl/wba).
456 jan hoogland
Upon this observation we came to the following hypothesis: when

a certain entry is translated into Arabic by means of a paraphrased
description or definition while all, or almost all, bilingual dictionaries,
included in this enquiry, translate the same entry in the form of a
description or definition, it is justified to conclude, that the entry
concerned does not have a one-to-one equivalent in Arabic, and that we,
therefore, are dealing with a lexical gap in Arabic. In order to verify this
hypothesis, a large sample of entries from the Dutch-Arabic dictionary,
with one-to-one source equivalents in various other dictionaries, was
gathered to compare their translations into Arabic.
The source language units can be divided in simple words (lexical units),
compound words and expressions. Compounds in Dutch are simply
written together as one word and consequently have to be entered as
independent lemmas in the dictionary.
This article limits a comparison of translations for simple words
from various bilingual dictionaries, since it turned out difficult to find
unambiguous translations in English, French or German for Dutch
compounds. The lack of those translations as a starting point for a
comparison of Arabic translations would result in data that would be
difficult to compare with each other.
2. Definition of description
As stated in the introduction, descriptions are, in most cases, not direct

equivalents of the word or expression in the source language, but
explanations, i.e. expressions of more than one word which provide
the dictionary user with an explanation in the target language. Since a
description is a combination that is not lexicalized, it is not entered in
the database as an expression in the target language.
Descriptions represent a unidirectional translation relation, i.e. they
are not included in the reversion process to produce the reverse part
of the dictionary. Assuming the description describes a concept that
represents a lexical gap in the target language, it is obvious this concept
could not be entered in the reverse part of the dictionary as an entry.
First of all, a sample of 25 Dutch simple words translated with
descriptions was gathered. These words were chosen randomly from the
underlying database. These words can be found in Table 1 below.
lexical gaps in arabic: evidence from dictionaries 457
In order to be able to look up these words in English-Arabic, French-

Arabic and German-Arabic dictionaries it was necessary to obtain
translations of the Dutch words from the sample in these languages.
The table below shows these 25 Dutch words with translations in
three western languages. Some Dutch words which were candidates
to be included in the sample, turned out to be untranslatable in one
or more western languages. The absence of an obvious translational
equivalent in one of these languages of course would cause problems
in using that language as source language for the comparison of Arabic
translations. For this reason, some randomly selected Dutch words,
translated with descriptions in the Nijmegen Dutch-Arabic dictionary
were not included in the sample.
TABLE 1 ENGLISH, FRENCH AND GERMAN TRANSLATIONS

OF THE SAMPLE
Dutch English French German
bestek cutlery couvert Besteck
francofiel francophile francophile frankophil
frankeren to frank affranchir frankieren
freak freak fana Freak
fresco fresco fresque Fresco
gletsjer glacier glacier Gletscher
glibberen slither glisser glitschen,
schlittern, rutschen
gniffelen snigger rire sous cape, rire schmunzeln, in
tout bas, rire dans sich hinein lcheln
sa barbe
log guest hte Gast, Logiergast,
Logierbesuch
logement lodging (house) auberge Gasthaus
lommerd pawnshop mont-de-pit Leihaus, Leihamt
loods pilot pilote, lamaneur Lotse
loops in heat/season en chaleur, en rut brnstig
lotgenoot partner (in compagnon, Leidensgenosse,
misfortune / compagne Schicksalsgenosse
adversity) dinfortune
ouderwets old-fashioned vieux jeu, dmod, altmodisch
pass de mode,
dsuet, surann,
archaque, prim
ouvreuse usherette ouvreuse Platzanweiserin
ouwel wafer pain ( m. ) azyme, Oblate
dautel, hostie non
consacre
458 jan hoogland
TABLE 1 (CONT.)
Dutch English French German
ribbel rib, cord cte Rippe

rif reef rcif Riff
riposteren riposte riposter ripostieren
roe rod verge Rute
rog ray raie Rochen
trampoline trampoline trampoline Trampoline
transcriberen transcribe transcrire transkribieren
traumatisch traumatic traumatique traumatisch
If the assumption that the concepts behind these words represent

lexical gaps in Arabic is correct, at least a part of these words have to
be translated with descriptions in other dictionaries too, if they were
included in these dictionaries at all. So in order to investigate this, the
translations of these words were checked in a number of dictionaries.
For this comparison the following dictionaries were consulted:
Al Mawrid English-Arabic
Oxford, Doniach English-Arabic
El Mounged English-Arabic
Al-Kamel Al-Kabir French-Arabic
Al Manhal French-Arabic
Schregle German-Arabic
Van Mol Dutch-Arabic
However, before presenting the results of this comparison of the seven

dictionaries mentioned, it seems useful to present the Arabic translations
from the Nijmegen dictionaries here first. Obviously, these should not
be part of the comparison, since the lack of translational equivalents for
these words in the Nijmegen Dutch-Arabic dictionary were the starting
point for the comparison. Inclusion of these translations would distort
the results of the comparison.
Tables 3a and 3b below contain the Arabic translations of the Dutch
words from the seven dictionaries introduced above.
TABLE 2 ARABIC TRANSLATIONS FROM THE NIJMEGEN

DUTCH-ARABIC DICTIONARY
DU_LU Eng_LU DA_Nijmegen
@ A' BC@ DEF ,H

? I@ #

bestek cutlery
francofiel francophile J LK M1 : J 6N .K O
!
frankeren to frank P "2 P Q @
freak freak RS O> T
! UV W
fresco fresco >K K5 X 1Y 5Z
gletsjer glacier (3[ \ ]) K _ 1 `
3[
glibberen slither A
4 a b@ &!
gniffelen snigger Idc ! e 5 = f g h g
log guest (BL ij[
a) eL Q
logement lodging (house) kLl m
n 6
lommerd pawn shop P op[ q b 1 +K O!

loods pilot (nautic) W 1
!, Jrs: ( W 1
!
loops in heat/season Aa@, m @, 1 t
4b $(J[ E )
lotgenoot partner (in 1 j5 6 ha1W
misfortune/adversity
ouderwets old-fashioned
(xn j@ y z C ]) {w 1C
va q!
ouvreuse usherette JrLS !
ouwel wafer | )
}~ 1
HWK 1
ribbel rib, cord w- >" k
rif reef - V
5 q! J (
riposteren riposte Jc 9B " ) -
roe rod } -4
j q! '
rog ray qjn: W 1 K O"
trampoline trampoline } J K5n !
transcriberen transcribe 1 #
&> Jc
7
1 # + >=
traumatisch traumatic Jc LK : = Jc ! 5 " AS T !
460 jan hoogland
TABLE 3A ARABIC TRANSLATIONS FROM 3 OF THE

7 OTHER DICTIONARIES
DU_LU Eng_LU EA_Mawrid EA_Oxford EA_Mounged
bestek cutlery :qjE>>>>>( o H4>>a ! 3@nZ I@#

J>"@ q>>! H
qjE>>>>>>( A>>'B!@ qjE( qjE>>>>>E
o IE>>W@ >b1j@
francofiel francophile @# JLM1: . >b >! + .O! DLM1: n!
JLM1: J6N >>>>M16 (nX#)
frankeren to frank "2 A>>>5a 1X# q>! &>>:'# (>sW#) +(-#)
(J!B{ @#) >>>>>a1 }@ >>>>>>a1>"
}# >>>>>s:a 1X# n" H>>(-
;6! a1 > A> &>>{
freak freak .s_' UV>>W
J
&>>>{ 1>>>a
>>>6 -=
?1j
fresco fresco %{)\1Y5 J!>(1! -> ->X v>(-
(7 @# }-\ &>>>>{ }" >o!
+d9 Rd_
6:X
gletsjer glacier 1>>>= s$ 1>>= s$ 1>>>=
s$
glibberen slither &>>>>> A4>>>>>a ->On! &>>{ A>>w A4>>= A>>w
A4>>>>a JsO>>E ew =1>>Z ew
gniffelen snigger fg) hpa :s: fg hg #1> I-) hOQ
(Id! e5= !T! h>>p ->>>>>>T:"
(4>T
log guest ( mn:") +a4<
logement lodging : q!3!:# ]) JW@1:! J61 ij" ]) JW@1:! J61
(house) ;! 1! ([ ij" 1X!@ ([
1X!@
lommerd pawn shop .T! :qb1 I>=b- +O! Ib- + qb1!
qjb1>>>>> eL>>>>>>> }E +"! 19)
@# (qb-
+"! 1>9
q>b-
TABLE 3A (CONT.)

loods pilot ]) q: W1! ] Jrs: W1! >W1!
(nautic) (61! @# n9 r# @# rLC +! Jrs:>>>>>>
(B! n9) -[' +>! >>6
@# >>>>>nL
->>>d' n#
n9
loops in heat/ Lb : (J[) i9@ >>6 .>- m@
season =# r' \ ] i[- :
J }L : (J>>>>>dE)
lotgenoot partner (in
misfortune/
adversity)
ouderwets old- w1 AL' w12 q! (e!) q!4 L{ &:'
fashioned :2 x( +" JL~ (-#) 9 .b8! Ja12)
6 - @# aw +" (Ja1=
|[) x()
ouvreuse usherette W1! :+L @- W1 T6 W1 >>>>>T6
{! -r vb' rL rL>>>>> @-
( yz! ]) vb>'
ouwel wafer q! -@! J99- }"1 }W1 }W1
H 6 | }"1
~1 ] |
( )
ribbel rib, cord J!{ 6- PQ Pp! ()
rif reef :1O[ L J>a1V>> W m6 - (
- ( (Js=>X1!) J: C (
C ( 19
riposteren riposte n' Xa 3a P>a1>>( - 3a P>a1>>>( -
>> J:2. ->>p &>>>{ ->p &>>{
J>a[ J>a[
roe rod ' .a 5' >5'
.>>>>>a>T
rog ray qjn:>>>>>> h 1 Dn:W qjn:>>>>>W
1>>>>O[ h=- 1>>>O>"
trampoline trampoline J5n! w) Dd!* [ w) Dd!*
}>>>>>>d h"w 1Y [ 4:a h"w 1Y
qjd!1>>>>>>> [' L{ 4:a (w[ [' L{
(w[
462 jan hoogland
TABLE 3A (CONT.)

transcriberen transcribe q! W q! +ra += >>>M + }K@ M
qY@ 3fW# J4! J">TE J"T C
+ra }E ) 1 q>! @# Ja{
@# ! &>>>> ka1>>W
J"T ! Ja{ J"E
(Ja
traumatisch traumatic >1>X >>Q- >Q- >1>X
TABLE 3B ARABIC TRANSLATIONS FROM THE REMAINING 4

OTHER DICTIONARIES
DU_LU Eng_LU FA_Kamil FA_Manhal GA_Schregle DA_Mol
bestek cutlery q!) oC Hw J! qj J!

JEW@ J! JE>>W@ JE>>>@ JE>>W@
q@ D(@ qj(@ J@ qj(@
( J26@
francofiel francophile M1: . .O! J>>>>>6N Aa
DLM1: n! >>>>>M1: J>s>>>>>>M1:
frankeren to frank H>>(- P6 %{ U PQ@
U>> J(- - q!) (- &>>>{ P"2
J(- &>>{ (AO! V
freak freak P! R>O>T! .5>T!
fresco fresco @# 1Y>>>>5Z 1>a>>>>5Z ko %{ =
->X v>(- v>(- ->X 16
-> o! -$
Jso! Ja-$
gletsjer glacier ) H-) L$ 1 1>>=
' .*! J! s$
] }a ] J(@ [d9
JL 3[\ (JL[ A2rC
A2rC ]@
1` (JL[
J! L$
glibberen slither w A4< Aw w A4< Aw
A,4 4
TABLE 3B (CONT.)
gniffelen snigger >>>6 h>>Q hg vZ 3 v"
1>( Js:
log guest eLQ '>! e>sQ esQ eLQ
vsa
logement lodging 34>= mn6 34>>= v! mn6 34<
(house) }
lommerd pawn shop qb1>> is" 7( hr" eL ij"
7!) Ib1 hr"
+"! eL qb1
o:"@ qb-
J@ !
(1:
loods pilot a }~- W1! W1! }~- >W1! >W1!
(nautic) rs:( +L
loops in heat/ .2) +o, .a
season (: .- (s)
s
lotgenoot partner (in >>>6 ha1>>W >n(a As6- qY19
misfortune 15 >b
/ adversity) >>a1>>>W
>>>" >>>>6
ouderwets old- q! 4 9 w1> q! w1> &>>{ AL'
fashioned AL' w12 va va 1T!
aw +" A>>>s' hT!)
- w1> (J9 L"
ouvreuse usherette :{! JrL! !{) * J
yz! ] !{ yzC
:L@ j( @# -p W1
-p W- (v($ q!#
{!
ouwel wafer 1j6 4 }~19 m>9-
JLg JL" (ia>>>>>l)
ribbel rib, cord PQ 2W PQ
= w@1
@# .=$)
(vX q! w-~
rif reef eL- J: 1) J: J>E # .>W J>dW
( :1 2W 19 [
( %{ Ja1 (1O[
C
464 jan hoogland
TABLE 3B (CONT.)
riposteren riposte - J{1>>l - @$
H>9 H>_b &>{ - (J{@)
>p! H>_" J:2 Jr"
(eL~ w-[C
.jp9)
roe rod (" a ' >5'
rog ray L h=-@ qjn:>>>>>W
1 Dn:W s 1>>O>"
?x(# R$)
(. . . Ja1
trampoline trampoline 4:! K
kn!
transcriberen transcribe - 1X - 1X d! >>>M

traumatisch traumatic >Q- >1>X >1X >>>d! >" >>Q-
>" >>Q-
So, in order to verify if the concepts behind these Dutch words do

represent lexical gaps in Arabic, it is useful to compare and to evaluate
the translations presented in these other dictionaries that provided the
same units with a description, if the word in question is included in the
other dictionary at all.
For all translations obtained from the 7 dictionaries mentioned above,
information was entered if the dictionary in question has included the
Dutch word or its English / French / German equivalent as an entry, and
if so, if this translation is a description. For this an evaluation system was
devised. All translations are evaluated with a code from the following
code system:
0= entry not included

1= entry is included, translation is an equivalent (not a description)
2= entry is included, translation is description
3= entry is included, translation is a neologism
4= entry is included, translation is a hyperonym
5= entry is included, translation is inaccurate or translation belongs to
a different or more specific meaning
6 = entry is included, translation consists of a neologism and a

description
Some remarks need to be made in relation to these values. For the sake
of clarity examples for each of these 6 categories will be presented. These
examples are taken from Table 3 a and 3b above.
Value 0 is obvious. If a word is not included in a dictionary, the value for

this word is 0. For example: glibberen (to slither) is not included as an
entry in D-A Van Mol, nor in G-A Schregle.
All other values (16) imply that the word occurs as an entry in the
source language of the dictionary heading the column.
Value 1 implies the translation is a translation equivalent. For example:

glibberen / to slither is included and its translation seems to be a real
equivalent in E-A Oxford (ew 1), E-A Mounged (ew A4< Aw),
and F-A Kamil (A,4 4 w A4< Aw).
Value 2 refers to a description. For example: glibberen / to slither

is included, and its translation is a description in E-A Mawrid (&
JL ).
This applies, first of all, to the real explanations which can be found,
for example, in Oxford English-Arabic at bestek / cutlery ( ox H4a !
IW@ A'B!@ DEF( q!) or logement / lodging (ij" ]) JW@1:! J61
q 1X!@ ([ ). However, one might as well define as descriptions
those combinations of two words that are not frequent. If we look at
the translations for francofiel / francophile for example, we find the
translation M1: . in both F-A Al Kamel and F-A Al Manhal. This is
a combination of words that is not lexicalized, and therefore should not
be entered in the database as an expression.
Value 3 refers to a neologism. For example: lommerd / pawnshop is

translated with a neologism in F-A Kamel (qb1 ij"). Some dictionaries
offer the user made up words in case of a lexical gap in Arabic. In Al
Manhal French-Arabic these neologisms are marked with an asterisk
466 jan hoogland
and an explanation comes with the neologism. See the examples of value
6 below.
Value 4 represents a hyperonym translation, i.e. the word in the target

language is actually the translation of a hyperonym of the source language
word. For example: transcriberen / to transcribe is translated as +=
M }K@ in several dictionaries. These translations do not express the
fact that to transcribe means decoding from one system and encoding
in another system. Therefore these translations express a more general
meaning and not the specific meaning of the source language word.
Value 5 as an evaluation means that the translation can be inaccurate or

wrong, or that it refers to a different, more specific meaning of the source
language word. For example: rif / reef is included, but its translation is
inaccurate in G-A Schregle (J# .W).
Value 6, finally, is a combination of 2 and 3, since it consists of a

neologism, with a description (mostly between brackets) added to it.
For example: gletsjer / glacier is included, but its translation consists
of a neologism and a description in F-A Manhal ([d9 H-
JL[ A2rC ] J(@) and F-A Kamil (3[\ ] }a ' .*! )
JL[ A2rC ]@ JL).
Before presenting all evaluations in a table, it is useful to introduce
the concept translation profile. This profile is a compilation of al the
evaluations, indicating the number of dictionaries in which a translation
was found, followed by an enumeration of all the evaluations. This profile
gives a quick indication of the number and types of translations found
in the various dictionaries.
For example 3:223 means a word was found in three dictionaries, and
in two of these dictionaries the word was translated with a description
(code 2), and in one with a neologism (code 3).
Below is a table containing columns with the Dutch words, the English
translations and 7 columns for all 7 sample dictionaries, and a final
evaluating column containing the translation profile of the concept.
3. Interpretation of Table 4
The translation profiles of the words should indicate if these words

represent concepts that can be considered lexical gaps in Arabic. The
higher the number of dictionaries in which these words were included,
and the higher the number of description translations or neologism
translations, the higher is the probability that the concept represents a
lexical gap in Arabic. So, profiles containing value 2 (description) and
value 6 (neologism + description) are the clearest indications.
Seven Dutch words have been included in all 7 dictionaries. These
words are: bestek / cutlery, gletsjer / glacier, logement / lodging,
loods / pilot, ouderwets / old-fashioned, rif / reef , and transcriberen
/ transcribe. A more thorough look at the translation profiles of these
words is useful in order to determine if the coding system seems reliable.
1. The word bestek / cutlery has the following profile 7: 2222222. It

seems justified to conclude this word/concept indeed represents a
lexical gap in Arabic since there is unanimity among all dictionaries
in this, and probably no other dictionary or other reference work
will mention an Arabic word for this concept. Obviously, we do not
have to go very deep into Arab or Islamic culture to conclude tat this
lexical gap can be explained by the fact that in the Arab world people
were and are still used to eat with their hands.
2. The word ouderwets / old-fashioned has the same profile:
7: 2222222. So, obviously this concept also represents a lexical gap
in Arabic.
3. The word loods / pilot has the following profile: 7: 1111222. The
Dutch word loods, as a person, is monoseme and can only mean
a pilot in shipping, and certainly not airplane pilot or any other
type of guide. The translation W1! is correct but requires a specific
context in order to distinguish this specific meaning from other more
general meanings of the word W1!.
4. The word transcriberen / transcribe has the following profile:
7: 2222444. As indicated above, there are presented description
translations and hyperonym translations.
5. The word rif / reef has the following profile 7: 2255666. Three
dictionaries have the word J: as a neologism, in all cases followed
or preceded by a description. However, this word was not found with
this meaning in the text corpus or on the internet.
6. The word gletsjer / glacier has the following profile: 7: 1116666.
468 jan hoogland
TABLE 4 EVALUATION OF THE TRANSLATIONS IN

7 DICTIONARIES
Dutch_LU Engl_Tr E_A_ E_A_ E_A_ F_A_ F_A_ G_A_ D_A_ Transl.
Mawr Oxf Moun Kamel Manh Schreg Mol Profile
bestek cutlery 2 2 2 2 2 2 2 7: 2222222

francofiel francophile 2 2 2 2 2 2 0 6: 222222
frankeren to frank 2 2 2 2 2 2 0 6: 222222
freak freak 2 0 0 4 4 0 0 3: 244
fresco fresco 2 2 2 2 2 2 0 6: 222222
gletsjer glacier 6 1 1 6 6 6 1 7: 1116666
glibberen slither 2 1 1 1 1 0 0 5: 11112
gniffelen snigger 2 2 2 2 2 2 0 6: 222222
loge guest 4 0 0 4 4 2 4 5: 24444
logement lodging 2 2 2 1 1 4 1 7: 2224445
(house)
lommerd pawnshop 6 2 6 3 2 2 0 6: 222366
loods pilot 2 2 2 1 1 1 1 7: 1111222
(nautic)
loops in heat/ 2 2 2 0 2 2 0 5: 22222
season
lotgenoot partner (in 0 0 0 0 2 2 4 3: 224
misfortune
/adversity)
ouderwets old- 2 2 2 2 2 2 2 7: 2222222
fashioned
ouvreuse usherette 2 2 2 6 6 0 0 5: 22266
ouwel wafer 2 2 2 5 2 4 0 6: 222245
ribbel rib, cord 4 2 0 4 2 0 0 4: 2244
rif reef 6 2 2 6 6 5 5 7: 2255666
riposteren riposte 2 2 2 2 2 0 0 5: 22222
roe rod 4 1 1 0 1 4 4 6: 111444
rog ray 1 1 1 1 1 0 0 5: 11111
trampoline trampo- 2 6 6 3 0 3 0 5: 23366
line
transcri- transcribe 2 2 2 4 4 2 4 7: 2222444
beren
traumatisch traumatic 1 0 1 1 1 1 1 6:111111
The equivalent L$ 1` which is mentioned in three dictionaries

seems to have become a lexicalized expression during the last years.
As for the neologisms presented by the other four dictionaries, the
fact that four dictionary compilers have added descriptions to these
neologisms illustrates that they are not completely at ease about the
understandability of the neologisms they present.
7. The word logement / lodging has the following profile 7: 2224445.
This profile seems to justify the conclusion that this word is a
problem word. Indeed, it can easily be translated with a hyperonym,

since it is a certain type of accommodation for staying overnight.
The Dutch Van Dale dictionary defines a logement as a place to stay,
a home where one can obtain a temporary stay (and food) against
payment (presently less in rank or appreciation than a hotel). Neither
is lodging the same as a hotel. So translations with mn6 or 34< should
be evaluated as hyperonym translations.
Another way of looking at the profiles is the following:

There are nine Dutch words that were translated with a description in
all dictionaries in which they were included:
bestek / cutlery 7: 2222222
ouderwets / old fashioned 7: 2222222
francofiel/ francophile 6: 222222
frankeren/ to frank 6: 222222
fresco/ fresco 6: 222222
gniffelen / to snigger 6: 222222
loops / in heat 5: 22222
ouvreuse / usherette 5: 22266
riposteren / riposte 5: 22266
These are the words of which the translation profile consists of code 2
and 6 only. It seems reasonable to sustain that words found in four or
more dictionaries (out of seven) and being translated with descriptions
(possibly in combination with a neologism) in all those dictionaries
indeed do represent a lexical gap in Arabic.
So the concepts expressed by the words bestek / cutlery, ouderwets
/ old-fashioned, francofiel / francophile, frankeren / to frank, fresco
/ fresco, gniffelen / to snigger, loops / in heat, ouvreuse / usherette
and riposteren / to riposte can be considered lexical gaps in Arabic.
There are five Dutch words that were translated with a description in
most dictionaries in which they were included:
lommerd / pawn shop 6: 222366
lotgenoot / partner in misfortune 3: 224
ouwel / wafer 6: 222245
rif / reef 7: 2255666
470 jan hoogland
trampoline / trampoline 5: 23366

Since most of the translations are descriptions, and the others should
be qualified as inadequate translations (neologisms, hyperonyms or
inaccurate), we may assume that these concepts also represent lexical
gaps in Arabic.
As a matter of fact, this system for encoding types of translations and
combining them in a translation profile seems to be a useful method for
comparing dictionaries and identifying lexical gaps in Arabic.
When we take a closer look at the Dutch words meeting with lexical
gaps in Arabic, it turns out, not surprisingly, that they can be categorized
as follows:
the source unit expresses a typical Dutch or non-Arabic concept

causing a lexical gap in Arabic: fresco / fresco, ouvreuse /
usherette, transcriberen / to transcribe;
the source unit expresses a technical concept, too rarely used
inside the Arab world in order to enter a dictionary and to receive
a general translation, thus this concept represents a lexical gap in
Arabic: frankeren / to frank, trampoline / trampoline.
the source unit is related to a non-Islamic religious concept,
for example: ouwel / wafer, coming from Catholicism within
Christianity.
4. Conclusion
4.1 Conclusion concerning the identification of lexical gaps in Arabic

By comparing the Arabic translations of twenty-five Dutch words,
we concluded that nine were translated with descriptions in all the
dictionaries in which they were included. This justifies the conclusion
that these words can be considered to represent a lexical gap in Arabic.
Five words were translated with descriptions in most of the dictionaries
in which they were included, which also indicates they should be
considered to represent lexical gaps in Arabic.
Therefore, the hypothesis formulated at the beginning of this
contribution may be considered to be confirmed by this conclusion.
4.2 Conclusions concerning lexicography in general

As a result of the comparison described above, some conclusions
concerning the lexicography of Arabic can be drawn as well.
The first conclusion concerns the use of Arabic neologisms in
dictionaries. During the stage of comparing the descriptions in various
dictionaries, it was noticed that in some dictionaries newly coined words,
which should be considered neologisms in Arabic, were presented as
translations.
However, when neologisms are introduced in a bilingual dictionary as
a translation equivalent for an unknown foreign language word, there is
no context in the users mother tongue. As a matter of fact the dictionary
user will only run into such a neologism by looking up a foreign word
in a foreign language-Arabic dictionary. When the user is confronted
with a neologism without any further explanation, he/she will probably
remain in confusion about the meaning of the foreign language word
as well as to the meaning of the Arabic neologism. Some dictionaries
provide neologisms with an additional explanation or description.
This will help the user both to understand the meanings of the foreign
language word and the neologism in his own language.
If the same dictionary is used by foreigners who use the dictionary
to produce Arabic, and these foreigners use the neologisms presented,
there is again the risk of the Arab receiver of the message who will not
understand the meaning of the neologism.
To conclude, the following question presents itself: why do dictionary
compilers decide to invent new words as translation equivalents for
foreign words / concepts? Is this related to a certain way of language
purism in order to prevent the use of loan words? Do they realize that
the dictionary users still will be left in confusion? The result of this policy
will be the occurrence of ghost words, i.e. words that do only occur in
dictionaries and not in authentic texts.
A second conclusion concerns the occurrence of inaccurate trans-
lations in the dictionaries. As indicated in Table 4, some inaccurate
translations have been met during the process of comparing
the translations. Inaccurate translations are represented by value 5 in
the table. Fortunately, the total number of translations, qualified as
inaccurate, is rather limited, since value 5 occurred only three times on
a total of 142 translations in Table 4.
No dictionary is free of inaccuracies, and no one knows this better
than the present author who was involved in compiling the Dutch-
472 jan hoogland
Arabic dictionaries. Some descriptions in the Nijmegen Dutch-Arabic

dictionary could have been replaced with real translational equivalents.
As stated before, some Dutch words in the sample turned out to be
translatable with equivalents in at least some of the dictionaries. This
implies first of all that the concepts behind them certainly do not represent
lexical gaps in Arabic. And secondly, these words should not have been
translated with descriptions in the Nijmegen Dutch-Arabic dictionary.
This, in a certain way, can also be considered a series of inaccuracies
in the said dictionary. Would the authors involved in translating these
specific words have consulted all the dictionaries available, this might
have led to the insertion of a translational equivalent in stead of a
description. Not only would this have enriched the Dutch-Arabic part,
but also the reverse part, since translational equivalent relations are
bidirectional and consequently would have been entered in the Arabic-
Dutch part as well.
Some dictionary makers rely heavily on their colleagues or
predecessors. Some similarities between certain pairs of dictionaries
in the comparison above, however, are striking. If we, for instance,
look at the translations for logement / lodging, lommerd / pawn
shop, loods / pilot, loops / in heat, ouvreuse / usherette, ouwel /
wafer, riposteren / to riposte, in the dictionaries E-A Oxford and E-A
Mounged, we find a number of literally identical translations. This is so
obvious, that one may conclude, that the compiler of the most recent one
of the two dictionaries (Mounged English-Arabic) depended heavily on
his predecessor (Oxford English-Arabic), without mentioning this in
the introduction of his dictionary, and even without entering the Oxford
dictionary in the list of sources consulted. The possibility that both of
the dictionaries were based on a common third ancestor requires further
research.
5. References
Balabakk, Munr. 1981. Al Mawrid: A modern English-Arabic dictionary. Beirut: Dr

al-Ilm li-l-Malyn.
Doniach, Nakdimon. 1972. The Oxford English-Arabic Dictionary of Current Usage.
Oxford: Oxford University Press.
Hoogland, Jan, Kees Versteegh, and Manfred Woidich, eds. 2003. Woordenboek Arabisch-
Nederlands Nederlands-Arabisch. 2 Vols. Amsterdam: Bulaaq.
Idriss, Souheil. 1996. Al Manhal Dictionnaire Franais-Arabe. Beirut: Dar Al Adab.
Mol, Mark van. 2001. Leerwoordenboek Nederlands-Arabisch. Amsterdam: Bulaaq.
Reda, Youssof M. 1996. Al-Kamel Al-Kabir plus, dictionnaire du franais classique et

contemporain, franais-arabe. Beirut: Librairie du Liban.
Schregle, Gtz. 1977. Deutsch-Arabisches Wrterbuch. Beirut: Librairie du Liban.
Theodory, Constantin. 1996. Al-Mounged English-Arabic. Beirut: Librairie Orientale.
MASDAR FORMATION
Joost Kremers
University of Cologne
1. Introduction
Arabic morphology includes a nominal form, traditionally called

masdar, whose meaning and form is related to that of a verb. The form
relation, although clearly visible, is complex, the meaning relation is
straightforward: the masdar is a complex event nominal, in terms of
Grimshaw (1990), and names the action expressed by the verb, retaining
the verbs argument and event structure.1
The masdar has been analyzed among others by Fassi Fehri (1993)
and Kremers (2003), who both base themselves on Abneys (1987) anal-
ysis of English gerunds. These analyses are purely syntactic, however.
Neither discusses how the morphology of these forms functions, the
tacit assumption being that the masdar form is derived by some post-
lexical process that does not play a role in syntax.
Ackema and Neeleman (2004) discuss so-called mixed categories,
such as English gerunds, in the framework of their theory on word for-
mation and morphological structure. They argue that the peculiar syn-
tactic properties of such structures follow from the way in which they
are formed morphologically.
Arabic masdars share the peculiar syntactic properties of other mixed
categories, but Ackema and Neelemans analysis does not provide any
direct insight into the reason why this should be so. Their analysis cru-
cially depends on the concatenative nature of morphology in the lan-
guages that they discuss, and since masdar formation in Arabic uses
non-concatenative morphology, the analysis at first sight does not carry
1
The same nominal forms can also have non-event meanings, making them result
or simplex event nominals, in Grimshaws terms. These lack the argument and event
structure of the corresponding verb, and have no systematic meaning relation to it. See
Fassi Fehri (1993) and Kremers (2003) for some discussion. Traditionally, these nouns
are not called masdars.
476 joost kremers
over straightforwardly. As I show in this paper, however, nothing needs

to be changed about Ackema and Neelemans analysis if we adopt a pro-
sodic model for Arabic morphology, along the lines of McCarthy and
Prince (1990b, 1996).
2. Abneys analysis of English gerunds
In his influential PhD dissertation, Abney (1987) presents an analysis of

the English gerund that tries to account for the fact that the gerund can
be used in at least three different constructions:
(1) a. Johns singing of the Marseillaise
b. Johns singing the Marseillaise
c. John singing the Marseillaise
In all three constructions in (1), the gerund singing is an event nominal,
in the sense that it retains the argument and event structure of the under-
lying verb to sing (Grimshaw 1990). In other words, all three phrases
describe an event, and the subject and object of the event are obligatorily
expressed.2 Abney observes (as others have done before him) that there
is a decreasing degree of nominality in the structures (1ac): (1a), the
so-called Ing-of construction, is the most nominal, in that it expresses
both the subject and the object with nominal means: the Saxon geni-
tive s and the dummy preposition/case marker of, respectively. (1b),
the Poss-ing construction, is more nominal, in that the object is marked
with accusative case, rather than with of. Yet, the subject is still marked
with the Saxon genitive. Lastly, (1c), the so-called Acc-ing construction,
marks the object with accusative, and the subject as well. This is there-
fore the most verbal of the three constructions, in that there is no nomi-
nal marking present anymore.
Crucially, the fourth logical possibility, the subject being marked with
accusative and the object with of, does not occur:
(2) *John singing of the Marseillaise
2
If one or both of the arguments are not expressed, e.g. in Johns singing, the gerund is
no longer a complex event nominal, but rather a simplex event or result nominal, which,
as Grimshaw (1990) shows, have markedly different properties.
masdar formation 477
This fact can be accounted for if we assume, as Abney does, that gerunds
start out as V projections, and change into an N projection somewhere
along the way. Until the point where the change takes place, arguments
can be licensed through case, but after the change, only nominal licens-
ing mechanisms (s and of ) are available. Once the change has taken
place, it cannot be undone, which accounts for the impossibility of (2).
The analysis is supported by the observation that (1bc) allow adverbs
but no adjectives, while (1a), the purely nominal gerund type, allows
adjectives but no adverbs:
(3) a. Johns constant/*constantly singing of the Marseillaise
b. Johns *constant/constantly singing the Marseillaise
c. John *constant/constantly singing the Marseillaise
The analysis that Abney proposes assumes that there is an affix -ing,
which attaches to a verbal category, changing it into its corresponding
nominal category. It can attach at three levels: at V, creating an N, at VP,
creating an NP, and at IP, creating its corresponding nominal projec-
tion DP. Crucially, -ing is not a head, i.e., it does not project a syntactic
phrase of its own. It just attaches to a projection, changing its category.
The tree structures that Abney proposes are the following:
(4) Ing-of:
DP
Johns D
D NP
N PP
-ing V of the Marsaillaise
sing
478 joost kremers
(5) Poss-ing:
DP
Johns D
D NP
-ing VP
V PP
sing the Marsaillaise
(6) Acc-ing:
DP
-ing IP
John I
I VP
V DP
sing the Marsaillaise
There is of course one unattractive aspect to these structures: it is not

clear how the verb stem sing and the suffix -ing combine to form the
gerund singing. In fact, it is not at all clear what it really means for an
affix to attach to an XP.3
3. Ackema and Neelemans treatment of mixed categories
Ackema and Neeleman (2004) retain the idea that what is different about
the different gerund structures is the level at which the nominalizing
3
Abney notices this problem himself. In order to solve it, he argues for an elaboration
of X theory, which basically makes a distinction between below X0 and above X0 syntax
which is not altogether convincing.
affix attaches in syntax. However, they place this idea in the context of a
much broader theory on word formation, which enables them to show
what exactly it means for an affix to attach to different levels of projec-
tion. Before we look at how they deal with gerunds, it is necessary to
discuss some of the aspects of their theory.
3.1 Short outline of the theory

Ackema and Neeleman argue for a view of the language faculty that is
inspired by Jackendoff (1997, 2002). In this view, the language faculty
contains three generative systems, one for syntax, one for semantics and
one for phonology. The output of these generative systems are linked to
each other by mapping rules.
In this model of the language faculty, a lexical item is not just a con-
glomerate of the semantic, syntactic and phonological properties of a
word. Rather, the three types of features are essentially separate, func-
tioning in separate components of the language faculty, linked through
mapping rules. A word such as tree has the syntactic representation
N[+count,sg], which is used in the syntactic module, and it has the
phonological representation /ti:/, which features in the phonological
module. If we then abbreviate the semantic concept as tree, we can
represent the lexical item tree as in (7), where the double arrows indi-
cate two-way mapping relations:
(7) tree N[+count,sg] /ti:/
Affixes function the same way: they have, apart from a semantic rep-
resentation, which does not concern us here, a morphosyntactic one,
which Ackema and Neeleman represent in small caps, as affix, and a
morphophonological one, which they represent with slashes, as /affix/.
The point is that what is traditionally seen as a single affix actually con-
sists of three separate elements, linked by mapping principles.
There are different types of mapping principles. Ackema and Neele-
man argue that there are (at least) three general mapping principles, of
which two concern us here.4 The first is Linear Correspondence:
4
The third mapping principle is Quantitative Correspondence which states that no
element in the morphosyntax is spelled out more than once.
480 joost kremers
(8) Linear Correspondence

If X is structurally external to Y, then F(X) is linearly external to F(Y).5
Linear Correspondence states that a morphosyntactic representation
such as [[root sf1] sf2], in which sf1 is first attached to root, and sf2
then to the complex [root sf1], is mapped onto a morphophonological
representation in which the order of the two suffixes is retained, i.e., /
rootsf1sf2/. It can in principle not be mapped onto a structure in which
both suffixes have switched places: */rootsf2sf1/.
The second general mapping principle is Input Correspondence:
(9) Input Correspondence
If an affix selects (a category headed by) X, then F(affix) takes F(X) as its
host.
Input Correspondence states that if an affix in morphosyntax attaches
to an element X of a specific category, or to a projection of X, then the
morphophonological form to which affix is mapped must attach to
the morphophonological element associated with X. In other words, the
phonological form of a nominal affix cannot attach to a phonological
form that is associated with a syntactic V head; it must attach to some-
thing that is associated with an N head.
Note that both Input and Linear Correspondence only apply when
F(affix) has an overt phonological form. When F(affix) is empty (i.e.
//), they do not apply (or apply vacuously). Since both principles regu-
late the distribution of phonological material, it stands to reason that
they do not apply when there is no phonological material to distribute.
Apart from these general mapping principles, there are also lexical
mapping principles, as in (7). Note that not only words, but also affixes
can be described with lexical mapping rules. For example, the English
agentive suffix er is normally mapped onto the phonological form //:
to write writer. This fact is recorded in the lexical entry for the affix:
5
The notation F(X) refers to the phonological structure onto which the syntactic
structure X is mapped. It is equivalent to the slash notation /x/ that Ackema and Neele-
man use, but in my opinion less confusing. I define it as in (i):
(i) a. D(X): the subtree that has X as root
b. F(X): the phonological material onto which D(X) is mapped.
(10) agentive noun er[N,sg] //

There are, however, specific cases in which er receives an idiosyncratic
mapping. One example is the case of the verb to type, of which the agen-
tive noun is typist, not *typer. In other words, we can say that there is an
idiosyncratic mapping rule of the form in (11):
(11) [[type] er] /taip/ /ist/
This rule says that a syntactic structure of the form [[type] er] is
mapped onto a phonological structure /taip/ /ist/. Note that mentioning
the phonological form of the stem in the rule is in fact redundant: there
is no reason to assume that the lexical entry for type cannot provide its
phonological form in (11). Therefore, I adopt a slightly different nota-
tion for such idiosyncratic mapping rules. Instead of (11), I will write
the following:
(12) er/type /ist/
(12) expresses that er, when attached to type, is mapped onto /-ist/. 6
Affixes are well-known to have selectional restrictions. For example, the
English agentive suffix -er must attach to a word (i.e., it cannot attach
to a phrase), and moreover, this word must be a verb. The point that
Ackema and Neeleman make is that these selectional restrictions are in
fact of non-uniform nature, and must therefore lie in different modules
of the grammar.
By representing a suffix such as -er as is done in (10), it becomes pos-
sible to express this point: Ackema and Neeleman state that the syntactic
affix er has the requirement that it must attach to something of cat-
egory V, while the phonological affix // has the requirement that it must
attach to something that is a (phonological) word.
6
Note that I use a hyphen here in the affix /-ist /. In Ackema and Neeleman's original
formulation, this is not necessary, because the mapping rule explicitly mentions the stem
and with that, the position of the affix with respect to the stem. In my reformulation,
some way is needed to indicate that the affix is actually a suffix. Note also that although
I use a double arrow in (14), this is not meant to indicate that this is a one-to-one map-
ping; the phonological form /-ist / does not always signal an agentive noun, cf. words
such as communist, guitarist. Here, Ackema and Neelemans formulation may seem to
have an advantage, since their equivalence is one-to one. As explained in footnote 30,
however, this is not the case in every kind of idiosyncratic mapping rule that we need.
482 joost kremers
This analysis means that er does not require that it be attached to a

head; instead, it can attach to any projection of category V.7 In this way,
Ackema and Neeleman account for the existence of agentive nouns such
as in (13):
(13) spring-in-t-veldje
jump-in-the-field.dim
little madcap
The phrase spring-in-t-veld is a VP, containing an intransitive verb and a
locative PP. In order to derive its meaning as an agentive noun, Ackema
and Neeleman assume that the Dutch er suffix is present in the syntactic
structure, and that it is idiosyncratically mapped onto //:8
(14) [[spring [in t veld]] er] /spri/ /int/ /vlt/ //
That is, there is a mapping rule of the form:
(15) er/VP //
In (14), the suffix er attaches to the VP [spring [in t veld]]; being a
syntactic suffix, er does not require a V head, it only requires a structure
of category V. However, mapping er onto // would violate Input Cor-
respondence, which requires that F(er) take F(V) as its host, even when
it attaches to VP. As mentioned above, Ackema and Neeleman argue
that the alternative mapping onto // voids Input Correspondence.
Therefore, (14) is possible.
3.2 English gerunds

Having discussed the essentials of Ackema and Neelemans theory, we
can now look at their analysis of English gerunds. First, Ackema and
Neeleman assume that the actual nominalizing affix in gerunds is not
-ing. The reason for this is that -ing does not just derive nominal catego-
ries, it also derives the (verbal) present participle, which indicates that it
is probably best categorized as a non-finite verb suffix. Therefore, there
must be another affix in the structure that is responsible for the nomi-
7
That is not to say that the attachment is completely free. Some requirements do
apply. Since er saturates the external argument of the verb it attaches to, it cannot attach
to a VP of which the external theta role has already been assigned.
8
I have ignored the diminutive suffix -je here. Although it is attached to the noun
veld phonologically, morphosyntactically, it attaches to the entire structure. (It is not the
field that is small, but the madcap.) See Ackema and Neeleman (2004) for details.
nalization. Because no such overt affix exists, Ackema and Neeleman

assume that its phonological form is null.9
As in Abneys analysis, the nominalizing affix is attached to differ-
ent levels of projection. Note that if the affix had an overt phonological
form, the syntactic structure could not be mapped onto a phonological
structure without violating either Input Correspondence or Linear Cor-
respondence. Suppose, for example, that the affix attaches to VP, creat-
ing a Poss-ing construction:10
(16) Poss-ing:
DP
John D
D NP
VP AFFIX
Adv VP
constantly V-ing DP
singing the Marsaillaise

Note that the nominalizing affix is assumed to be a suffix in (16). It
attaches to a projection of category V. If its corresponding F(affix) were
overt, Input Correspondence would require it to attach to the /word/ that
the head of this projection is mapped onto, i.e., singing. That, however,
would violate Linear Correspondence: in the syntactic structure affix
is external to the VP, and therefore Linear Correspondence requires its
phonological form to be external to F(VP). However, in a phonological
realization such as /Johns/ /singing/ /affix/ /the Marseillaise/, this would
not be the case: the Marseillaise here follows the /affix/, meaning that the
/affix/ is internal to F(VP).11
9
The same conclusion is reached by Yoon (1996).
10
Ackema and Neeleman do not specify to which levels of projection the affix
attaches in English, they only do this for Dutch nominal infinitives. I simply assume that
English behaves the same. Note, by the way, that Ackema and Neeleman only discuss
Ing-of and Poss-ing, not Acc-ing. I assume the reason for this is that Acc-ing is more
likely a participial structure than a nominal one, as argued also by Reuland (1983).
11
Note that this situation would obtain if we were to assume that -ing is the nominal-
izing affix.
484 joost kremers
A realization with the /affix/ after the object DP would comply with
Linear Correspondence, but would violate Input Correspondence,
because the /affix/ can no longer attach to /singing/.12 However, on the
assumption that the /affix/ has no overt form, neither Input Correspon-
dence nor Linear Correspondence applies (or they apply vacuously),
and hence no mapping principle is violated.13
4. Arabic masdars
4.1 The data

Let us now turn to the Arabic masdar. First, as observed by Fassi Fehri
(1993) and Kremers (2003), Arabic masdars are typical mixed catego-
ries. They can assign (overt) accusative to their objects, as in (17a), but
the object can also be licensed through a preposition, as in (17b):
(17) a. aqlaqa-n -ntiqd-u -r-rajul-i -l-mar-a
annoyed-me criticizing-nom the-man-gen the-project-acc
the mans criticizing the project annoyed me
b. aqlaqa-n -ntiqd-u -r-rajul-i li -l-mar-i
annoyed-me criticizing-nom the-man-gen to the-project-gen
the mans criticizing of the project annoyed me
I will refer to constructions of the type in (17a) as masdar + acc, and to
constructions of the type in (17b) as masdar+li. Note that in both cases,
the subject of the masdar is marked with genitive case, making the struc-
tures equivalent to the Poss-ing and Ing-of constructions, respectively.
Furthermore, (17a) allows an adverbial expression, while (17b) allows
an adjective:
12
Note, by the way, that it is really immaterial whether we assume that the affix
adjoins to the left or the right of the VP. We would run into the same problems with the
adverbial constantly.
13
The analysis makes a strong prediction: if the nominalizing suffix is overt, a dever-
bal noun cannot assign accusative to its object if this object follows the verb. Ackema
and Neeleman show facts from Norwegian and Quechua that suggest that this predic-
tion is borne out.
(18) a. aqlaqa-n -ntiqd-u -r-rajul-i

annoyed-me criticizing-nom the-man-gen
bi-stimrrin -l-mar-a
with persistence the-project-acc
the mans persistently criticizing the project annoyed me
b. aqlaqa-n -ntiqd-u -r-rajul-i -l-mustamirr-u
annoyed-me criticizing-nom the-man-gen the-persistent
li -l-mar-i
to the-project-gen
the mans persistent criticizing of the project annoyed me
In contrast, masdar + acc cannot be modified by an adjective:14
(19) *aqlaqa-n -ntiqd-u -r-rajul-i -l-mustamirr-u
annoyed-me criticizing-nom the-man-gen the-persistent
-l-mar-a
the-project-acc
the mans constant criticizing the project annoyed me
Fassi Fehri (1993) analyzes masdars much along the lines of Abney
(1987). He argues that there is an (abstract) masdar affix that can attach
at the V level or at the VP level, turning the verbal projection into an N
or an NP. In Kremers (2003), I present a somewhat different analysis.
Instead of having an affix attach to a verbal projection at different levels,
I argue that during the derivation, a nominal functional category can
take a verbal category as its complement. This analysis is based on the
idea that there are certain parallels in the nominal and the verbal projec-
tion lines. Essentially, I argue that at any point in the verbal projection
line, a verbal functional head can be replaced by its nominal counter-
part, thus deriving the mixed nature of the construction.
However, as remarked above, neither of these approaches takes into
account the morphology of masdar forms. Within Ackema and Neele-
mans framework, however, we can do better: it becomes possible to see
how masdar formation works given the syntactic analysis. Before we go
into that question, I will first discuss the morphology of masdars.
14
Somewhat unexpectedly, however, masdar+li can be modified by adverbial
phrases:
(i) aqlaqa-n -ntiqd-u -r-rajul-i bi -istimrrin li -l-mar-i
annoyed-me criticizing-nom the-man-gen with the-persistence tothe-project-gen
the mans constantly criticizing of the project annoyed me
As suggested in Kremers (2003), this may be due to the fact that the adverbial in (i) is
a PP.
486 joost kremers
4.2 Masdar formation

Arabic lexical items generally have a root and a stem. The root is a
sequence of (usually) three consonants,15 and is not in itself pronounce-
able. From a root, stems can be derived by applying prosodic morphemes
(McCarthy and Prince 1996). A stem thus formed is a pronounceable
form, and is the basis for further conjugation and derivation, by means
of pre- and suffixes or by additional prosodic morphological processes
(McCarthy and Prince 1990b).
The original analysis holds that stems are derived by applying CV-
templates to a root. For example, the perfective stem of the base verb is
formed by applying the template CVCVCV to a root (McCarthy 1981).
More recently, however, it has been argued that this approach cannot
account for certain facts. In particular, it allows templates that do not
occur (McCarthy and Prince 1990a, 1990b). Therefore, a newer approach
is developed, in which templates are expressed in prosodic terms.16 The
perfective stem of the base verb is then expressed as a prosodic template
of the form , i.e., a structure of two short syllables.17
It is important to note that a root consists of just three consonants,
and unlike much derivational morphology in Arabic does not have any
prosodic specification. That is, the root does not specify in which posi-
tions in the syllable the consonants appear. For example, the initial con-
sonant can appear in the onset (20a) or in the coda (20b):
(20) root: /ktb/
a. ka.ta.b to write
b. ak.ta.b to dictate
15
There are some quadriliteral roots, and there are categories that can be considered
biliteral roots. These will not be discussed here.
16
Prosodic here refers to the prosodic hierarchy (Selkirk 1980, Truckenbrodt in
press), i.e. the hierarchy of prosodic constituents that constitute a phrase, i.e. Utterance >
intonational phrase > phonological phrase > prosodic word > foot > syllable > mora. For
the present discussion, only syllables () and mor () are relevant. A syllable consists
of an onset (the initial consonants) and a rhyme (the vowel and any final consonants).
The onset is extramoraic, the rhyme contains either one mora or two, for light and heavy
syllables respectively. A light syllable is represented as , i.e. a syllable of a single mora,
while a heavy syllable is represented as . In Arabic, the first mora of a syllable is always
associated with a vowel. The second mora can be associated with a consonant or with the
same vowel as the first mora, when this vowel spreads (i.e. when it is long).
17
The third syllable in a form such as kataba he wrote is not part of the template.
Stems in Arabic always have a final extrametrical consonant (McCarthy and Prince
1990a), and it is usually the third root consonant which fills this position. This con-
sonant normally resyllabifies when an ending is added, in this case the -a of the third
person masculine singular.
The same goes for the second consonant, although it rarely happens in
the verbal system that it appears in a coda. In the nominal system, this
is not uncommon, however:
(21) root: /f l/
a. fa.a.l to do
b. fi.l action, act
Note that (21) shows that a root does not even specify whether two con-
sonants appear in the same syllable or not. This is solely specified by
the prosodic template of the particular stem formed from the root in
question.
Table 1 lists the perfective stem and masdar forms of triliteral verbs.18
TABLE 1 ARABIC MASDARS

Stem perfective stem masdar
I KaTaB (irregular)
II KaTTaB taKTB
III KTaB muKTaBa(t)
IV aKTaB iKTB
V taKaTTaB taKaTTuB
VI taKTaB taKTuB
VII inKaTaB inKiTB
VIII iKtaTaB iKtiTB
IX iKTaBB iKTiBB
X istaKTaB istiKTB
XI iKTTaB iKTTB
XII iKTawTaB iKTiwTB
XIII iKTawwaB iKTiwwB
XIV iKTanBaB iKTinBB
XV iKTanBay iKTinBy
18
As discussed by Wright (1981), Classical Arabic had alternative masdar forms for
most verb stems. Nonetheless, the forms in Table 1 are the standard forms, which is why
I confine myself to them here.
488 joost kremers
In this table, the capitalized consonants KTB belong to the root, any
other non-capitalized consonants are stem affixes.
Looking at Table 1, one quickly notices that one particular vowel pat-
tern dominates the masdar forms: /i/, although on some occasions
(when the corresponding verb stem has a long vowel, such as in III and
XI), the /i/ is also long, yielding //.
Some more analysis is possible, though. McCarthy and Prince (1990a)
argue that the last vowel in a verbal form is always short in finite forms
and often long in non-finite forms (masdars, but also passive partici-
ples of stem I, and some deverbal instrumental nouns). They therefore
analyze vowel quantity of the final vowel as a template suffix indicat-
ing (non-)finiteness. With that modification, the masdar is no longer
marked by /i/ or //, but simply by /ia/, with vowel length deter-
mined by other factors. We can therefore say that the Arabic masdar
contains two morphemes: a nominalizer with the form /ia/, and a non-
finite suffix -.
This template suffix, - for finite forms and - for non-finite forms,
combines with a base stem template, which is monosyllabic.19 So the
stem I template for finite forms exists of the stem base - plus the finite
suffix -. The stem II template is formed from the stem base - plus
the finite suffix. In addition to these morphemes, stems can also have a
prefix, such as the tain stems V and VI, n- in stem VII and st- in stem
X, etc.20
We can now analyze a form such as /infil/, the masdar of stem VII
of the root /fl/, as containing four distinct morphemes: the consonantal
root, the stem VII marker, the non-finite marker and the nominalizer:
(22) root: /f l/
stem VII: (n)
nominalizer: /ia/
non-finite: -
Putting these together requires that we analyze phonological structure
as consisting of several layers, or autosegmental tiers (Leben 1973). The
basic tier is the segmental tier, which contains the segments (phonemes)
19
Although it may contain a detransitivizing prefix /n-/ or /t-/, as in stems VII
and VIII, which consists of an extrametrical consonant that resyllabifies into a coda
position.
20
The stem VIII infix -t- is analyzed as a prefix as well, after which a metathesis rule
swaps the initial consonant and the prefix.
of the utterance. For Arabic, we need to assume an additional syllabic

tier. The root morpheme in (22) maps onto the segmental tier, as does
the nominalizer. The non-finiteness morpheme maps onto the syllabic
tier. The stem VII marker is mixed: it maps onto the syllabic tier, but at
the same time fills a slot in the segmental tier. When all four morphemes
of (22) are put together, we obtain the following representation:
(23) () ()

n f i a l
The top row in (23) represents the syllabic tier: the first two syllables
are given by the stem VII marker. The first syllable is extrametrical (it
is later syllabified by the insertion of an epenthetic /i/), and its coda
position is filled by the /n/. The third syllable is the non-finiteness suf-
fix, which is heavy (i.e. has two mor), as indicated. The final syllable
is again extrametrical, and is added by default, since every Arabic stem
ends in an extrametrical syllable.
The bottom row represents the segmental tier. Crucial is of course
the question how the various slots in both tiers are associated with each
other. The initial /n/ is straight-forward: the stem VII marker speci-
fies that it is associated with the first (extrametrical) syllable. Just as
straightforward is the final root consonant /l/. Because a stem must end
in an extrametrical syllable, the third consonant must always take this
position.
The remaining segments, /f/ and // of the root, and /a/ and /i/ of
the nominalizer, are associated through the principle of Left-to-Right
Association (Leben 1973). The root consonants fill the coda positions of
the second and third syllable, and the vowels of the nominalizer fill the
peaks. Because the third syllable has an additional mora, and because
there is no segmental material anymore to fill it, the vowel /a/ spreads to
the second mora, which results in a long vowel.
Most of the masdar forms can, mutatis mutandis, be analyzed in this
manner: for the masdar forms of stems IV and VIIXV, all that needs to
change in (23) is the stem template.21 As is clear from Table 1, however,
21
Note that the masdar of stem IX /iKTaBB/ is formed on the base of the underlying
form /iKTaBaB/. The gemination of the third root consonant is the result of deletion of
the /a/, a common process in this context.
490 joost kremers
things are different for the remaining forms: they appear to be quite
idiosyncratic. In spite of the apparent complexity, however, we can show
that they all make use of the same stem template that the corresponding
verb forms use. Where they differ from other masdars is the form of the
nominalizer and the non-finite morpheme, which is sometimes absent.
Let us look at the various forms. First, masdars V and VI have an
idiosyncratic nominalizer, that takes the form /au/ rather than /ia/.
They also lack the non-finite morpheme , since the forms are not
*/taKaTTB/ and */taKTB/. Here, we can either say that they have an
idiosyncratic non-finite marker -, or that the lack of a second syllable
in the template causes not just the stem affix but the entire verbal stem
template to be used as a basis for the masdar. Either way, we obtain the
same result:
(24) stem V/VI: -
t
nominalizer: /au/
I will assume that the nominalizer does not associate with the prefixed
syllable in the stem template. Therefore, the /a/ of the nominalizer asso-
ciates with the second (heavy) syllable of the stem template, and the /u/
with the third. Presumably, the /a/ of the first, prefixed, syllable is a copy
of the /a/ of the second syllable.22
Stem II masdars can be dealt with in a similar way. On the face of it,
the stem II masdar does not seem to be formed on the template of the
stem II verb, as it contains a prefix /ta/ which the verbal template lacks,
and its vowel pattern is /ai/, not /ia/, as the default nominalizer speci-
fies. Furthermore, it does not show gemination of the second consonant,
which seems typical for the verbal template.
However, McCarthy and Prince (1990b) observe that Arabic pho-
nology does not distinguish between CVV and CVC syllables: all that
counts is their prosodic status as heavy syllables. The templates of stems
II and III (and likewise stems V and VI) are therefore identical: their
first syllable is , without any specification how the heaviness of the
22
One fact that supports this assumption is that Classical Arabic had an alternative
masdar form for stem V verbs, namely /tiKiTTB/. On the assumption that the vowel
in the prefix is a copy of the vowel in the next syllable, this form is completely regular:
it contains the masdar morpheme /ia/ and the non-finite suffix , as indicated by the
fact that the final vowel is long.
syllable is obtained. An additional gemination rule operates on stem II

and V verbs, but not on stem III and VI.
Suppose now that we say that the gemination rule does not apply in
stem II masdars (unlike stem V masdars), and that the nominalizer of
stem II is idiosyncratic. We then obtain the following:
(25) stem II:
nominalizer: /tai/
non-finite: -
Here, the stem II template is not specific to the masdar, it is the same
template that forms finite verb forms of stem II. In order to see how
these morphemes yield the masdar form /taktb/, let us see how they are
associated with the template. First, we add the non-finite suffix to the
stem base, and associate the nominalizer with the resulting template:
(26)
t a i
The first syllable here has two mor because this is specified in the stem
II template, and the second syllable has two mor because it is the non-
finite suffix. The vowel /i/ cannot be associated with the second mora of
the first syllable because Arabic does not allow two vowels in a single
syllable.23 When the root is intercalated into the template, the second
mora of the first syllable can be associated with the first root consonant.
The second root consonant can be associated with the onset of the sec-
ond syllable, and, as usual, the last consonant is associated with a (newly
created) extrametrical syllable. At the same time, the /i/ associated with
the first mora spreads to the second, creating a long vowel:
(27)
()
t a k t i b
23
Long vowels and diphthongs are either the result of lengthening or because the sec-
ond mora is filled by a semi-consonant, not because two vowels end up in one syllable.
492 joost kremers
In this way, we can derive the stem II masdar by just assuming an idio-
syncratic nominalizer, in the same way that we have analyzed stem V
and VI masdars.24
The stem III masdar is quite idiosyncratic as well. Its form is
/muKTaBa(t)/, where the final /-a(t)/ is the regular feminine ending.
This masdar is a so-called masdar mmy or m-masdar. The m-masdar is
an alternative masdar formation found in Classical Arabic, in which the
feminine form of the passive participle is used as masdar. This masdar
formation has mostly disappeared, but in stem III, it is the dominant
one.25 I will not go into the details of participle formation in Arabic
(see McCarthy 1981 for some discussion), but like masdar formation,
it is templatic, with an /m-/ prefix that marks the participial form and a
vowel pattern that indicates voice.
Summarizing, we can say that there is a regular nominalizer /ia/,
which applies in the majority of cases. This affix is combined with a non-
finite suffix . Stems II, V, and VI have an idiosyncratic nominal-
izer, and stem III has an idiosyncratic masdar formation. Stems V and
VI lack the non-finite suffix, and instead get a default as second syl-
lable. Lastly, stem I masdars are all idiosyncratic. In the next section, I
will discuss the syntactic and phonological processes that underlie the
masdar formation in more detail, and discuss the mapping rules needed
to account for them.
4.3 The masdar affix

As we have seen, the syntactic properties of masdars are essentially
identical to those of gerunds. Therefore, the starting point for the analy-
sis should be that the syntactic structure of masdars is the same as well:
there is a syntactic masdar affix that attaches to V or to a projection of
V.
If masdar attaches to V, we obtain the equivalent of the English Ing-of
construction: masdar+li. Mapping of such a structure to phonology
24
Note that the /ta/ element in the masdar morpheme is not a prefix, contrary to the
/t/ element in stem V and VI forms. If it were, a form /taKaTTB/ would result. Instead,
it must be part of the nominalizer.
25
Classical Arabic had three other stem III masdars, /KiTB/, /KTB/ and /KiTTB/,
but none of these were as common as the m-masdar.
is unproblematic in any event, as we have seen above, because V and

masdar are sisters in the tree. Input and Linear Correspondence are
easily satisfied.
However, Linear Correspondence is only apparently satisfied. As we
will see, there is no way in which masdar in a masdar+acc construction
can adhere to Linear Correspondence. We must therefore assume that
masdar is not subject to it at all, so that even in masdar+li construc-
tions, adherence to Linear Correspondence is only apparent.
(28) shows the tree of a masdar + acc construction. masdar attaches
to VP, which yields a structure in which the object can be assigned accu-
sative case:
(28) DP
D NP
MASDAR VP
Subj V
V Obj
However, on the assumption that masdar is subject to Linear Corre-
spondence, the mapping to phonology would violate at least one map-
ping principle. The reasoning is identical to the one discussed in section
3.2 for the English gerund constructions. masdar attaches to the VP, so
Input Correspondence requires that /masdar/ attaches to /v/. Doing so
would violate Linear Correspondence, however: masdar is external to
the VP, but having /masdar/ attach to /v/ leaves it internal to it, as it then
occupies a position between /subject/ and /v/.
The analysis that made this configuration unproblematic for English,
saying that the nominalizing /affix/ is phonologically null, is not avail-
able for Arabic. As we have seen above, the Arabic masdar formation
uses an overt nominalizing morpheme. Another possible solution that
easily comes to mind is to adjust the structural relations in the tree in
such a way that masdar and its intended host V are adjacent, so that /
masdar/ can attach to /v/ without violating Linear Correspondence.
Such structural rearrangement obviously implies movement. As
argued in many works (e.g., Ritter 1991, Kremers 2003), in possessive
constructions in Arabic and Hebrew (of which (28) is one, because the
494 joost kremers
subject is assigned genitive case), N moves to D. If we apply this move-

ment to (28), the V head26 and masdar would end up adjacent.
Such an analysis might seem attractive at first sight, because the order
in masdar constructions is always NSO. Therefore, the V head must
move to a higher position: the tree in (28) could not derive the correct
order. Moving V to a position adjacent to mas d ar in (28) would enable
Linear Correspondence to be adhered to while at the same time deriv-
ing the correct word order.
There are some problems with this analysis, however. First, there is no
real consensus that N-to-D movement really takes place: Borer (1999)
argues against it, for example. Furthermore, even if one assumes that it
can take place, N does not always move to D in the Arabic noun phrase.
If there is no genitive-marked DP in the noun phrase, N-to-D movement
presumably does not take place. Therefore, it cannot be guaranteed that
/masdar/ and /v/ will always, under any conceivable construction, end
up adjacent.
The root of these problems, however, is the tacit assumption that
masdar is a prefix.27 The phonological shape of the masdar affix, how-
ever, suggests that this assumption is incorrect: as discussed above, much
of Arabic morphology consists of prosodic templates that map onto a
prosodic tier in phonology (see, e.g. McCarthy 1981). The masdar affix
is such a prosodic template, as we have seen in section 4.2. It is neither a
prefix nor a suffix, and hence does not appear before or after the verbal
root. Rather, one would say it appears simultaneous with it.
At first sight, this seems a problematic conclusion. masdar is attached
to a syntactic structure, and syntactic structures are at some point lin-
earized. The common assumption is that linearization is total: it applies
to every terminal element in the tree. That is, the linear structure that
results from a syntactic tree contains all the terminal elements in that
tree, and for any pair of distinct elements x,y, a linear order is defined,
either x>y or x<y (Kayne 1994). Linear Correspondence is basically a
formalization of this assumption.
26
Of course, there is no N head in (28), so we would have to assume that V moves
to D.
27
Note that the entire solution would be impossible if /masdar/ were a suffix: the
object would then always end up between /v/ and /masdar/, no matter where V moves
to.
However, as mentioned above, Ackema and Neeleman assume that

Linear and Input Correspondence do not apply when a syntactic element
maps onto // in phonology. That is, the mapping principles are sensi-
tive to the phonological form of the elements they apply to. It should be
no surprise, then, that given the prosodic nature of the masdar affix, it
is also not subject to Linear Correspondence.
More precisely, the masdar morpheme maps onto the prosodic tier,
and therefore does not (and in fact cannot) require linearization with
respect to the material on the segmental tier. However, it does require
association with material on the segmental tier. This is obtained through
Input Correspondence: because masdar attaches to the VP, F(masdar)
must attach to F(V), which is the root morpheme. Since the root mor-
pheme consists of segmental material, the required association is
obtained.
Having said that, let us look at how masdar formation takes place
in detail. In section 4.2, I have argued, following McCarthy and Prince
(1990b), that the morphosyntactic process of masdar formation involves
two morphemes: a nominalizer, which I will indicate with noml, and
a non-finite suffix n-fin. Above in this section, on the other hand, I
have assumed a masdar affix that must be able to attach to a structure
in phrasal syntax, in order to derive the Arabic equivalent of Poss-ing
constructions. The question we must ask, then, is what is the relation
between masdar on the one hand, and noml and n-fin on the other.
Obviously, noml and n-fin in some way make up masdar. I propose
the following structure:
(29)
NOML
NOML N-FIN
This is a morphological structure, which is then inserted into the syn-
tactic structure at the position of the masdar affix:
496 joost kremers
(30) DP
D NP
masdar VP
noml Subj V
noml n-fin V Obj
vii root
I assume that the V head is composed of the root and the verb stem
marker, here stem VII. The masdar affix, as indicated, is composed of
the nominalizer and the non-finiteness suffix. When the tree in (30) is
mapped onto phonology, Input Correspondence will make sure that
F(masdar) is properly associated with F(V). The formation of the
masdar form /infil/ will then proceed as described above.
Note that whether a syntactic element is subject to Linear Correspon-
dence is not a function of its morphological form alone. F(V) itself con-
sists of autosegmental morphemes, but cannot be exempt from Linear
Correspondence: the entire masdar form consist of autosegmental mor-
phemes, and at least one of them must be subject to Linear Correspon-
dence, otherwise the form could not be linearized with respect to the
other terminal elements in the structure. The natural assumption is that
the root, which is not a syntactic affix, unlike all the other morphemes,
is this element. Note that the root is a phonological affix, because it can-
not form a stem of its own. Syntactically, however, it is not, because it
does not require adjunction to a structure of a specific category. This
syntactic difference between the root and the other elements is presum-
ably the result of a semantic difference: the root is a lexical item (in the
traditional sense of the word) and as such member of an open semantic
class. All the other morphemes are functional or derivational, and part
of closed semantic classes. It therefore seems reasonable to assume that
members of open classes must always be mapped onto the segmental tier
in some way, and are thus always subject to Linear Correspondence. The
result of this assumption is that the masdar form is linearized according

to the position that the head V takes in the syntactic structure.28
Stem V/VI and stem II masdars differ very little from regular masdars.
All that is required is to assume the following idiosyncratic mapping
rules:
(31) stem V: noml/V /ua/
n-fin/V
stem II: noml/II /tai/
That is, noml in the context of a stem V template is mapped onto /ua/,
and n-fin onto -.29 Furthermore, when attached to a stem II template,
noml is mapped onto /tai/ rather than the default /ia/.
The system can also account for the stem III masdar. Recall that this
masdar is a so-called m-masdar, i.e., it is not formed with some genuine
masdar affix. Rather, a more or less suppletive strategy is used, which
employs the feminine passive participle. Let us say that there is an idio-
syncratic mapping rule of the following form:30
28
Note that this means that V must move, because masdar constructions have the
order NSO. At first sight, it might be tempting to argue that the NSO order is derived by
spelling out the masdar in the position of masdar, but this would not work for masdars
that license their objects with the preposition li: the analysis states that in such masdars
the masdar affix attaches to V rather than VP, which would predict a surface order of
SNO for such constructions. The actual surface structure is NSO, however, the same as
for masdars that assign accusative.
29
The stem V rules obviously also apply to stem VI, and, although not discussed
here, to the second quadriliteral stem as well. The rule may in fact refer not to the stem
templates but to some other property: as McCarthy and Prince (1990b) discuss, these
three verb stems share properties with each other that are not found in other verb stems,
indicating that they form a class of their own. Presumably, the mapping rules refer to
this class.
30
I mentioned in footnote 6 that Ackema and Neelemans formulation of idiosyn-
cratic mapping rules cannot be read as a one-to-one mapping in all cases. (32) is one
such case. Ackema and Neelemans rules differ in that they mention the phonological
form of the host as well as of the affix. But doing that in (32) would still not establish
that a form consisting of F(prt.pass.f)+F(III) is equivalent to a stem III masdar, as it
could (obviously) also be a feminine passive participle of a stem III verb. Note that this
is indicative of a general asymmetry between syntax and phonology : a syntactic struc-
ture is always mapped onto one particular phonological structure, but a phonological
structure may have more than one syntactic equivalent. In other words: phonological
structures can be ambiguous.
498 joost kremers
(32) [noml noml n-fin]/III F(prt.pass.f)
I will not go into the formation of the participle here. What is relevant
is the fact that an idiosyncratic mapping rule of the form in can exists.
Rather than specifying the phonological material that the syntactic
structure under consideration is mapped onto, the rule specifies a dif-
ferent (morpho)syntactic form whose phonological mapping must be
applied.
5. Conclusions
Ackema and Neelemans (2004) theory on word formation can give a

straightforward account of so-called mixed categories in languages
where morphology is concatenative. Non-concatenative morphology,
such as that of Arabic masdars, at first sight does not yield to an analysis
in terms of Ackema and Neeleman. If, however, we adopt the common
analysis of non-concatenative morphology in terms of autosegmental
tiers, we find a natural way to exclude the masdar morpheme from Lin-
ear Correspondence, which is the greatest obstacle to the application of
Ackema and Neelemans analysis to masdars: Linear Correspondence
only applies within an autosegmental tier in phonology.
At the same time, Input Correspondence still applies to the masdar
morpheme, accounting for the fact that it takes the verb as its host, and
not some random root in the rest of the structure.
6. References
Abney, Steven. 1987. The English Noun Phrase in its Sentential Aspect. PhD thesis, MIT:
MIT Working Papers in Linguistics.
Ackema, Peter and Ad Neeleman. 2004. Beyond Morphology: Interface conditions on
word formation. Oxford: Oxford University Press.
Borer, Hagit. 1999. Deconstructing the construct. In Kyle Johnson and Ian Roberts,
eds. Beyond Principles and Parameters. Dordrecht: Kluwer Academic Publishers,
4390.
Eid, Mushira and John McCarthy, eds. 1990. Perspectives on Arabic Linguistics II.
Amsterdam: John Benjamins.
Fassi Fehri, Abdelkader. 1993. Issues in the Structure of Arabic Clauses and Words. Dor-
drecht: Kluwer Academic Publishers.
Grimshaw, Jane.1990. Argument Structure. Cambridge, Mass: The MIT Press.
Jackendoff, Ray. 1997. The Architecture of the Language Faculty. Cambridge, Mass: The
MIT Press.
. 2002. Foundations of Language: Brain, meaning, grammar, evolution. Oxford:

Oxford University Press.
Kayne, Richard. 1994. The Antisymmetry of Syntax. Cambridge, Mass: The MIT Press.
Kremers, Joost. 2003. The Noun Phrase in Arabic: A minimalist approach, PhD thesis,
University of Nijmegen. LOT Dissertation Series 79.
Leben, William. 1973. Suprasegmental Phonology. PhD thesis, MIT.
McCarthy, John. 1981. A prosodic theory of nonconcatenative morphology. Linguistic
Inquiry 12(3). 373418.
McCarthy, John and Alan Prince. 1990a. Foot and word in prosodic morphology: the
Arabic broken plural. Natural Language and Linguistic Theory 8. 209283.
. 1990b. Prosodic morphology and templatic morphology. In Eid and McCarthy,
eds. Perspectives on Arabic Linguistics II Amsterdam: John Benjamins. 154.
. 1996. Prosodic morphology 1986. Ms. University of Massachusetts, Rutgers Uni-
versity.
Reuland, Eric. 1983. Governing ing. Linguistic Inquiry 14(1). 101136.
Ritter, Elizabeth. 1991. Two functional categories in noun phrases: Evidence from
Modern Hebrew. In Susan Rothstein, ed. Perspectives on phrase structure: Heads and
licensing. San Diego: Academic Press. 3762.
Selkirk, Elizabeth. 1980. The role of prosodic categories in English word stress. Lin-
guistic Inquiry 11(3). 563605.
Truckenbrodt, Hubert. In press. The syntax-phonology interface. In Paul de Lacy, ed.
Cambridge Handbook of Phonology. Cambridge: Cambridge University Press.
Wright, William. 1981. A Grammar of the Arabic Language. Beirut: Librairie du Liban.
Yoon, James. 1996. Nominal gerund phrases in English as phrasal zero derivations.
Linguistics 34. 329356.
MTHODOLOGIE LINGUISTIQUE :
ORGANISATION DE LA LANGUE ARABE
ORGANISATION GNRALE DES LANGUES
Andr Roman
Lyon
1. Introduction
Archimde (~ 287~ 212), ayant tabli la thorie du levier, aurait

demand pour soulever le monde un point dappui .
Lon peut voir dans le levier le symbole des mthodologies. Dans le
point dappui , le symbole de leur premier point dapplication.1
Un exemple admirable, clbre, de cette double dmarche est donn
par Ren Descartes dans son Discours de la mthode (1819 ; 29 ; 32) :
Je crus que jaurais assez des quatre [prceptes] suivants [. . .]
Le premier tait de ne recevoir jamais aucune chose pour vraie, que
je ne la connusse videmment tre telle [. . .]
Le second, de diviser chacune des difficults que jexaminerais, en
autant de parcelles quil se pourrait, et quil serait requis pour les mieux
rsoudre.
Le troisime, de conduire par ordre mes penses, en commenant
par les objets les plus simples et les plus aiss connatre, pour mon-
ter peu peu, comme par degrs, jusques la connaissance des plus
composs ; et supposant mme de lordre entre ceux qui ne se prcdent
point naturellement les uns les autres.
Et le dernier, de faire partout des dnombrements si entiers, et des
revues si gnrales, que je fusse assur de ne rien omettre.
Ces longues chanes de raisons, toutes simples et faciles, dont les gom-
tres ont coutume de se servir, pour parvenir leurs plus difficiles dmon-
strations, mavaient donn occasion de mimaginer que toutes les choses,
qui peuvent tomber sous la connaissance des hommes, sentre-suivent en
mme faon, et que, pourvu seulement quon sabstienne den recevoir
aucune pour vraie qui ne le soit, et quon garde toujours lordre quil faut
1
Le symbole est la peinture dune mtaphore.
502 andr roman
pour les dduire les unes des autres, il ny en peut avoir de si loignes
auxquelles enfin on ne parvienne, ni de si caches quon ne dcouvre. Et
je ne fus pas beaucoup en peine de chercher par lesquelles il tait besoin
de commencer : car je savais dj que ctait par les plus simples et les plus
aises connatre.
Au contraire [des sceptiques] tout mon dessein ne tendait qu massurer,
et rejeter la terre mouvante et le sable, pour trouver le roc ou largile.
Et remarquant que cette vrit : je pense, donc je suis, tait si ferme et si
assure, que toutes les plus extravagantes suppositions des sceptiques
ntaient pas capables de lbranler, je jugeai que je pouvais la recevoir, sans
scrupule, pour le premier principe de la philosophie que je cherchais.
Ce problme du premier point de lapplication dune mthodologie,
Ferdinand de Saussure lvoque dans une question, Unde exoriar ,
Do commencer ? (2002, 281). Sa question nest pas, semble-t-il, une
citation. La langue latine, dans laquelle il la formule, est la langue de la
premire voie qui sest ouverte lEurope, la voie romaine.2 La langue
latine est la langue de la culture europenne.
1.1 Une langue et lhomme

Voici le linguiste devant une langue et devant lhomme. Il peut com-
mencer son tude par lhomme car lhomme est, dans le monde, le seul
animal qui parle. La capacit propre lhomme, lhomme seul, qui fait
de lui un homo loquens, un h ayawn ntiq,3 est, selon toute vraisem-
blance, reflte par les langues que lhomme parle.4
Si cela est,
ou bien la capacit de parole de lhomme, cette capacit constitutive

de lhomme, tant reconnue, la constitution gnrale de la langue,
qui est une actualisation de cette capacit, pourra tre reconnue dans
cette capacit ;
ou bien la constitution gnrale de la langue tant reconnue, la consti-
tution de lhomme, qui en est le crateur, pourra tre saisie dans la
langue.
2
Voir R. Brague, 1992.
3
Al-Maydn cite dans son Majma al-Amtl 2 : 291, n3958, ladage : m l-insnu
law l l-lisnu ill sratun mumatta latun aw bahmatun muhmalat .
4
Il ne sagit pas ici des capacits physiologiques de lhomme, ses capacits auditives,
articulatoires, car celles-ci ne portent que sur les sons que lhomme doit entendre, doit
produire comme il parle, et non pas sur la grammaire de la langue.
mthodologie linguistique : organisation de la langue 503
Faut-il commencer par lhomme ? Faut-il commencer par la langue ?
Traditionnellement les savants qui se sont intresss aux langues ont

tudi les langues et non pas lhomme. Ou, exceptionnellement, quand
ils se sont intresss lhomme, cest pour affirmer que la langue a t
donne lhomme par Dieu, ou, comme Noam Chomsky, pour avancer,
axiomatiquement, que la langue est inne dans lhomme.
1.2 Mthodologie
Lexemple de Descartes incite rechercher dabord une mthodologie.
La premire esquisse de la mthodologie requise peut se rsumer dans
les deux oprations suivantes :
classer ;
dfinir.
Classer , cest--dire constituer en un ensemble des entits runies sur

quelque trait commun, reconnu, un trait propre, prsent ici, absent l,
qui les diffrencie, les oppose aux autres entits. Les plantes . Parmi
les plantes, les arbres . Parmi les arbres, le palmier . . . Les fruits
des arbres . Les fruits rouges , la fraise , la framboise , la gro-
seille . . .5
Dfinir , cest--dire tablir, dans un ensemble donn par un classe-

ment, une certaine identit de chacun de ses lments. La plante est
un vgtal . L arbre est une plante dune certaine hauteur . . . Clas-
ser est une opration humaine, immdiate, constante. Dfinir est
tout autant une opration humaine, immdiate, constante.
La casuistique est une manifestation, universelle, clatante, de mise

en uvre, voulue objective, de ces deux oprations.6 Le jugement de
valeur , si commun, est une autre mise en uvre, subjective, de ces
5
Les langues ngro-africaines ont un systme de classes nominales qui repose
sur une rpartition des tres et des objets et, postrieurement sans doute, des abstrac-
tions en un certain nombre de catgories , in Les langues du monde, 2, 740741.
6
On a soutenu avec vraisemblance que la tradition grammaticale arabe avait
emprunt la casuistique le cadre gnral de sa dmarche (Carter 1972, particulire-
ment). Versteegh (1980, 1114 ; 1993, 3336) a replac cette hypothse dans une per-
spective plus assure.
504 andr roman
deux oprations. Classer et dfinir apparaissent comme des op-

rations binaires, simples.7
Tous les chercheurs ont class. Exemples de classement des gram-
mairiens, les parties du discours : le verbe ; l adverbe . . . ou, dans
le discours, la phrase . . . Et ils ont dfini.8 Comme tous les hommes,
toujours.
Mais, la diffrence de lhomme naf, le chercheur, doit dfinir en
attribuant chaque lment dun ensemble donn une identit tablie
de telle sorte que cet lment puisse tre, constamment, reconnu par
une diffrence raisonne, irrductible, persistante, qui loppose rguli-
rement ce qui nest pas lui.
Et le chercheur, la diffrence de lhomme naf, doit rviser ses clas-
sements au fur et mesure du progrs de ses dfinitions. Lidentit du
dernier classement auquel il aboutira ne pourra tre que la mme iden-
tit des lments quil a identifis, qui le composent.
Ce faisant, le chercheur doit inscrire lobjet de sa recherche dans le
temps, car rien sur terre na dexistence hors du temps. Il lui faut donc ne
pas mconnatre lhistoire de lobjet de sa recherche.
2. Sbawayhi
Les dfinitions donnes par Sbawayhi (al-Kitb, I : 12) des trois parties
du discours sont brves : 9
[La particule] est [dans la langue] pour un sens qui nexiste ni par le nom,
ni par le verbe. Le nom, cest homme, cheval.10 Le verbe, ce sont les
7
Schuler (1990, 252) a pu prsenter la dmarche gnrale de la plaidoirie dune
cause, de sa premire position sa dernire position dprcatoire , sur un algo-
rithme exactement binaire.
8
Voir infra les dfinitions des parties du discours donnes par Sbawayhi dans son
Kitb. De mme al-Frb, lun des plus minents et des plus clbres philosophes
musulmans [. . .] surnomm le second matre, le premier tant Aristote [. . .] , mort
Damas, en 339/950, dans son Ih s al-ulm, p. 5. Le nom h add, singulier de h udd,
signifie communment limite et manire . Il a, chez Sbawayhi, o il est un terme
de mthodologie, gard son sens de manire (Troupeau, 1976, s.v. h add). Cest, sem-
ble-t-il, dans le Kitb al-h udd du grammairien et philosophe mutazilite ar-Rummn,
mort en 394/994, quil a pris, suivant sa pente, le sens de dfinition . Semblablement, le
terme orismos, dfinition , employ par Aristote, est rapprocher de horos, bord .
9
Voir sur la dfinition des parties du discours Versteegh (1995, 2242).
10
Certains manuscrits du Kitb, la somme de la tradition grammaticale arabe
naissante, livre imparfaitement dit, ajoutent mur h it).
schmes drivs de lexpression phonique des vnements lis aux noms et

qui sont construits pour ce qui est pass, ce qui sera, ce qui est encore. 11
fa l-kalimu : ismun wa-filun wa-harfun ja li-manan laysa bi-smin wa-l
fil fa-l-ismu : rajulun wa-farasun. wa-amm l-filu fa-amti latun uxidat
min lafzi ahdti l-asmi wa-buniyat li-m mad wa-li-m yaknu wa-
lam yaqa wa-m huwa kinun lam yanqati.
La seule partie du discours recevoir dans ce court paragraphe une dfi-
nition est le verbe. Le verbe se conjugue. Sa conjugaison offrait une prise
Sbawayhi. Mais, remarquablement, ce que Sbawayhi dit ici du verbe,
il ne le dduit pas du verbe. Lexpression du temps quil prte au verbe,
est lexpression du droulement du temps quil constate dans le monde.
Du nom, il sest born donner deux exemples. De fait, dans la langue
arabe de son temps, le nom nest pas construit. Il nest saisissable que par
sa relation ce quil nomme. Il faut donc, pour le reconnatre, regarder
dans le monde lentit quil nomme.12
Au demeurant, les grammairiens arabes, avant mme Sbawayhi,
avaient class, faussement, le relatif h aytu, l o , parmi les noms.
Linclusion dans lensemble des noms dun lment qui nest pas un nom,
rendait impossible la dfinition du nom. Il fallait sortir h aytu de len-
semble des noms, recomposer le premier ensemble, encore mal assur,
des noms.
De la particule, qui ne lui apparat pas construite non plus, Sbawayhi
relve quelle assure, complmentairement, les tches smantiques, rf-
rentielles, que le verbe et le nom nassurent pas.13 Ces tches sont les
tches de relation et de localisation, que lexprience immdiate montre
comme tant ncessaires.
Ce texte, qui sera exactement repris, perptuellement, apparat
comme une image du monde : ce qui est , le nom ; ce qui est dans
11
La constatation que les modus nexistent que dans ou par les res.
12
Un autre grand grammairien, Ibn Fris, mort Rayy en 395/1004, a, dans son livre,
as-Sh ibi f fiqh al-lua wa-sunan al-arab f kalmi-h, reproch Sbawayhi, p. 85,
davoir, pour le nom, donn non pas une dfinition mais des exemples (tamtl).
Lui-mme a retenu, comme tant plausible, la dfinition suivante : Le nom est ce qui est
implant sur le nomm comme on le mentionne et qui lui reste attach. , dukira l an
badi ahli l-arabiyyati anna l-isma m kna mustaqirran al l-musamm waqta dikri-ka
iyy-hu wa-lziman la-hu wa-hd qarb . Mais cette dfinition nest rien dautre que la
reconnaissance du fait que le nom est rapport immdiatement au nomm, cest--dire
sans le truchement du systme de la langue.
13
En fait plusieurs particules sont analysables ; exemple la ngation, /law/, du mode
rel, faite de la ngation /l/ et du morphme /w/ (< /u/) du mode rel ; voir A. Roman
1998.
506 andr roman
le temps , le verbe ; avec, entre lun et lautre, le lien , indispensable,

verbalis par la particule qui les situe, qui est diffrente, par son rle,
du verbe et du nom .
Ici, la tradition grammaticale arabe a choisi, semble-t-il, dobser-
ver dans la langue le monde, non pas lhomme donc et non plus, tout
dabord, la langue. Or la langue ne peut reflter limmense complexit
du monde.14 Cette porte du monde qui ouvrirait sur la langue est une
fausse porte.15
Quant la quatrime unit fondamentale de la langue que compose-
raient ensemble le nom, le verbe, la particule, cette unit qui serait leur
finalit dans la langue, cette unit, la phrase , les grammairiens non
arabes en ont prsum lexistence. Ils ont chou la dfinir. 16 Les gram-
mairiens arabes, eux, ne se sont pas attachs cette unit incertaine.
3. Discussion
Sil faut commencer ltude des langues smitiques par un ensemble

taill dans ces langues, pourquoi ne pas commencer par un ensemble,
caractristique de ces langues, lensemble des suites ordonnes de trois
consonnes sur lesquelles les units de nomination de ces langues appa-
raissent construites, le plus souvent.
14
Les systmes matriels qui composent le monde, dune part, sont ouverts, la
diffrences des systmes linguistiques ; dautre part, la combinatoire qui structure les
systme linguistiques est binaire voir infra , tandis que les combinatoires multiples
qui structurent les systmes matriels sont n-aires, avec n suprieur 2 .
15
La proposition de Thom, rapporte dans Petitot, Entrevue avec Ren Thom, est,
senzaltro, plus pittoresque. J.P. : Votre hypothse est [. . .] que les actions archtypales
comme capturer, prendre, couper, lier etc. sont devenues par ritualisation les matrices
de toutes les structures syntaxiques R.T. : Oui. Elles ont captur les structures plus
complexes. La meilleure preuve cest quil nexiste pas de verbes de valence suprieure
quatre. Cest la rgle des phrases de Gibbs . Ce passage ici repris de P.M. Lavorel 1980 :
I, 475).
16
G. Mounin, 1960, a recens, semble-t-il, toutes les dfinitions de la phrase. Aucune
nest opratoire. G. Kleiber, 2003, qui rfute le nouveau dcoupage du discours en clauses
et en priodes, concde qu on a tout gagner tenir compte de la dimension mmori-
elle introduite par Berrendonner, qui tait totalement absente des analyses tradition-
nelles et dont la nouveaut consiste montrer que les relations de discours narticulent
pas deux segments textuels, mais un segment textuel et une information en mmoire
discursive . Mais comment voir une pice de la langue dans cette dimension mmori-
elle ou dans la priode dfinie hypothtiquement comme une unit ergonomique
(A. Berrendonner, confrence, Lyon, 23.10.2003) ?
Ainsi, pour reprendre lexemple de toutes les grammaires, la mme

squence, ktb, se retrouve dans : /katbun/, crire (nominatif) ;
/kitbatun/, temps dcriture, criture (dun manuscrit) (nomina-
tif) ; /aktubu/, Jcris (mode rel) ; /ktibu-n/, crivant (nomi-
natif) ;17 /maktabun/, objet dcriture, bureau (nominatif) .
Cette squence de trois consonnes, la tradition grammaticale arabe
la nomme, tardivement, asl, souche .18 Mais ce nom prometteur res-
tera lettre morte faute dune hypothse sur le rle de la squence quil
nomme, faute de lhypothse dun plan gnral, 19 faute de la prise en
compte du temps.
Aprs la tradition arabe, la tradition orientaliste a vu dans cette
squence la racine de ces formes toutefois sans prendre en considra-
tion ni son rle, ni lorganisation de la langue quelle implique. En cons-
quence elle posera la question du nombre des consonnes composant ces
squences non pas dans le cadre dune organisation globale reconstruite
mais uniquement en diachronie, la recherche donc dun nombre pri-
mitif et non pas dun nombre dtermin systmatiquement ou, sinon,
rfrentiellement : elle cherchera tablir si les racines primitives ont
compt deux consonnes ou trois consonnes.
Ainsi la tradition orientaliste a, dans son tude de la racine ,
pris le temps en compte mais non pas lhypothse dune systmatique
gnrale.
Les deux traditions, arabe et non arabe, accepteront donc des racines
de deux consonnes, de trois consonnes, de quatre consonnes et mme,
parfois, de cinq consonnes. Ce faisant, elles auront empch la recher-
che dune organisation de la nomination. En effet, une telle irrgularit
des signifiants de racines autrement comparables exclut la rgularit
essentielle tout plan densemble.
17
Sens secondaires : secrtaire, crivain .
18
Dans le Kitb, asl dsigne le principe (dune loi de la langue), ltat primitif dun
h arf (i.e. dun son de la langue) hors de tout conditionnement ou encore dun schme
qui est donc non anomal. Dans le Sh ib (29), le asl englobe ce qui traite de la constitu-
tion de la langue (mawd al-lua), de sa materia prima (awwaliyya), de sa construc-
tion (mana) puis des rgles suivies par les Arabes dans leurs communications (rusm
al-Arab f muxtabti-h), des ressources multiformes de lloquence (iftinn) qui sont
leur disposition sur les deux modes rel (tah qq) et figur (majz).
19
Ibn Fris est original par sa proposition, qui semble tre sans autre exemple dans la
tradition grammaticale arabe, dune saisie totalisante de la langue. La langue est, dans le
Sh ib, prsente, dentre de jeu, dune part, dans son plan gnral et, dautre part, dans
sa double relation un Dieu crateur et la crature humaine que ce Dieu a dou dune
parole qui est la matrice de sa pense.
508 andr roman
Dans toutes les langues du monde, les racines apparaissent comme les
squences de phonmes ou, sinon, de syllabes,20 qui chanent les units
de nomination des entits et des expriences que lhomme a inventes
dans le monde : /katb/, /kitbat/, /aktubu/, /ktib/, /maktab/ . . .
Les racines qui regroupent les entits, les expriences, reconnues par
lhomme comme des units parentes, sont la premire nomination de
ces entits, de ces expriences.
Les racines sont leur premier tablissement dans les langues.
Le premier rle des racines est de leur donner dans les langues une
forme et une dimension telles que les langues puissent les manier.
Le nombre des lments qui composent les racines doit rpondre
cette exigence de maniabilit.
Les racines, condition de leur maniabilit, ne compteront que quel-
ques lments. Cest l une premire rduction, trs forte, impose par
les langues la nomination du monde.
Une deuxime rduction ncessaire est ralise par labstraction qui
aboutit donner, en effaant leurs diffrences, des entits diverses un
mme nom commun , des expriences diverses un mme verbe
commun .
Les racines, porteuses de ces sens abstraits, cest l leur deuxime
rle, devront compter, chacune, assez dlments pour satisfaire aux
besoins de nomination des entits et des expriences que lhomme veut
nommer.
Dans les langues smitiques, les racines des noms communs et des
verbes communs comptent, rgulirement, trois consonnes parce
que la combinatoire de trois consonnes est la premire combinatoire
mme de produire en nombre suffisant les arrangements qui seront
leurs signifiants.21
20
Les langues smitiques sont les seules avoir construit leurs systmes de nomina-
tion sur des racines de consonnes. Les langues tons les ont construits sur des racines
de voyelles. Les autres langues, sur des racines de syllabes. Les systmes de nomination
des langues smitiques et des langues tons vont se rorganisant sur des racines de syl-
labes.
21
Manifestement, les racines produites par la combinatoire de deux seules consonnes
sont en nombre insuffisant. Les racines produites par la combinatoire de trois consonnes
sont, par contre, en nombre plus que suffisant. Ce sont donc ces racines que la langue
arabe a utilises, rgulirement, dans son systme de nomination. Jamais elle na utilis
dans son systme de nomination de racines de deux consonnes. Les racines de qua-
tre consonnes, secondaires, C1C :2C3C4, ont t inventes dans les formes du verbe
racines de trois consonnes et modalit ditration ; le signifiant de cette modalit tait
la longueur de la deuxime consonne, C :2 ; cette consonne longue a t rinterprte
Ce nombre trois , suffisant, est un nombre trs bas. Il est donc satis-
faisant.22
La tradition arabe et la tradition non arabe ne poseront pas non plus la

question du statut des autres consonnes qui flanquent ces racines. La
reconnaissance par elles de //, dans /aktub-u/, comme tant la premire
personne du verbe ne touche que le signifi de cette unit de nomina-
tion. De mme lidentification par elles du tanwn /n/ comme un arti-
cle indfini dans les autres formes cites. Quant lidentification de
/t/ dans /kitb-a-t-u-n/ comme un suffixe , de /m/ dans /maktab-u-n/,
comme un prfixe , elle est encore plus imprcise : ces deux conson-
nes, /t/ et /m/, sont, pour elles, des zawid, des lments ajouts , des
formans dans des wazn, des schmes , filat pour /kitbat/, maf al, pour
/maktab/. Mais ces wazn, ces schmes, ne sont que les mimes des formes
quils prtendent analyser.23
4. laboration en schmes
4.1 Schme 1
Il est incontestable :
que les racines de la langue arabe sont, gnralement, des racines de

trois consonnes ;
que les units de nomination construites sur ces racines reoivent,
gnralement, dans les phrases, une voyelle dsinentielle, brve, qui
est soit le signifiant dun mode, dans les seuls verbes, soit le signifiant
comme une consonne double ; cette consonne double a, par dissimilation dabord, donn
naissance deux consonnes diffrentes ; exemple : /faqqaa/ > /farqaa/, craquer qqc. .
Naissance dun nouveau paradigme. Quant aux racines de cinq consonnes, ce sont des
chimres. Voir Roman, 2005, le chapitre Une brve histoire de la langue arabe .
22
Il est remarquable quune morphologie de larabe toute construite sur des racines
de deux seules consonnes est possible. Mais elle ne suffirait pas aux besoins de la nomi-
nation. Au demeurant la nomination smitique, par racines de consonnes, a succd
une nomination par racines de syllabes, dj nombreuses ; voir A. Roman, op. cit., loc. cit.
23
En ralit, ces wazn, ces schmes sont produits en trompe-lil par le sous-sys-
tme syllabique de la langue. En effet, ses deux seules syllabes, CV et CVC, produ-
isent mcaniquement des squences rgulires ; mais chacune de ces squences nest le
plus souvent que la figure dun sens global, la figure dune unit de nomination constru-
ite sur une racine syllabique, cest--dire sur un radical .
510 andr roman
dun cas dans les autres units de nomination (la voyelle /u/ dans
chacun des exemples donns) ;
que jamais une consonne na un signifi modal ou casuel.
Ces constatations suggrent un premier schma de la langue arabe :
SCHMA I DE LA LANGUE ARABE24
CCC V
4.2 Les schmes II et III

Hypothtiquement une langue peut tre dfinie comme lintercon-
nexion de deux systmes interdpendants : un systme de nomination
produisant des units de nomination, et un systme de communication,
sa syntaxe, dans le cadre duquel ses units de nomination entrent en
relation pour dire une exprience.
Dans le premier schma prsent comme lesquisse dune organisa-
tion gnrale probable, CCC reprsente, videmment, le systme de
nomination ; V , le systme de communication :
SCHMA II DE LA LANGUE ARABE

CCC V
systeme de nomination systeme de communication
La raison de cette rpartition des tches entre les consonnes, seules

pouvoir entrer dans les racines de la langue, et les voyelles, seules
signifier les cas et les modes, se dcouvre dans la structuration du sys-
tme syllabique de la langue : en arabe classique, tout arrangement de
consonnes et de voyelles qui, dans le cadre dune syllabe, ne serait ni
CV , ni CVC , constitue une syllabe a-systmatique produite par
une contrainte phontique ou par une pause syntaxique ; un tel sys-
tme syllabique, , dtermine, dans le fonctionnement de la langue
la disjonction du sous-ensemble des phonmes consonnes, {C}, et du
sous-ensemble des phonmes voyelles, {V} ; en effet deux squences de
syllabes CV et CVC ne peuvent composer les signifiants de deux
units de nomination qui, elles-mmes, constituant une paire lgitime,
sopposeraient, en bonne rgle, par lopposition simple dune consonne
24
est pour racine ; C , pour consonne ; V , pour voyelle .
et dune voyelle ; ce que montre sommairement le colonage ci-dessous

de deux squences diffrentes :
CV C.CV
CV.C VC
dont lopposition {C vs V} est non pas simple mais double.25
Cette disjonction, ds lors que les consonnes et les voyelles peuvent tre
utilises indpendamment les unes des autres, permet effectivement
lattribution systmatique de tches diffrentes aux consonnes et aux
voyelles.
Cest ainsi que sest trouve constitue la structure fondamentale de la
langue arabe, des langues smitiques.26
Cette raison syllabique trouve, la rgularit de son effet implique
la rgularit du plan gnral qui a t suppos.
SCHMA III DE LA LANGUE ARABE
= {CV, CVC} {C} {V} =
Systme Systme
de nomination de communication
5. Systme de nomination
5.1 Le temps
Dans lensemble des units du systme de nomination, quel lment
choisir qui serait susceptible de lordonner ? Le temps, sans doute.
En effet, le temps est essentiel aux expriences. En consquence il est
essentiel aux langues. Cest grce sa saisie et sa mmoire du temps,
25
En franais, diffremment, les paires { aorte vs porte }, { pote vs porte }
opposent une voyelle une consonne ; en franais, consonnes et voyelles peuvent occu-
per, dans les formes, les mmes positions.
26
Invitablement, les consonnes et les voyelles, disjointes par le sous-systme syl-
labique, sont conjointes dans les syllabes mais les racines slectionnent dans les syllabes
les consonnes qui sont radicales.
512 andr roman
que lhomme a pu crer ses langues. Heureusement, le temps peut tre

saisi, navement, comme une donne floue sans doute mais linaire et
simple.
Lespace, lautre coordonne du monde, est toujours ralis dans des
entits multidimensionnelles, souvent irrgulires, que leur complexit
a mises, demble, hors de porte des ressources dont les langues dispo-
sent mme si telle ou telle forme, telle ou telle silhouette, peut tre ab-
straite en un type : les dimensions trop nombreuses de lespace dbordent
les possibilits des langues.27
Les langues se sont donc constitues, avec la materia prima de la voix,
dans une premire classification, labore sur lopposition rfrentielle :
[+ temps] vs [ temps]
Elles ont, en consquence, produit, ncessairement, leurs units de

nomination,
soit comme des units inscrites dans un droulement apparent du

temps, dont le temps est lune des composantes, leurs modus exem-
ple : /ktib/, crivant ;
soit, contre la ralit du monde, comme des units trangres au
temps, leurs res exemple : /ayx/, cheikh .
5.2 Le temps et la racine

Selon toute vraisemblance, le systme de nomination se sera construit,
sur les deux fondements reconnus : le temps et la racine.
En arabe, aucune unit de nomination, res ou modus, ne peut com-
mencer par deux consonnes implosives. Le patron syllabique impose
leur sparation par une voyelle :
27
Le nom commun na pas t dtach de lespace. Il tait concret. Ou, plutt, son
caractre concret, ainsi dsign, par hypallage, est apparu quand lhomme a invent des
entits abstraites, des entits dtaches de lespace et par l-mme du temps, partir
donc, chaque fois, dun modus. Mais ce modus rifi, ce nouveau nom, reconnu comme
le nom dune entit noccupant aucun espace, ce nouveau venu, dsign comme un nom
abstrait , par la mme hypallage, na pas t signifi comme tel par un nouveau mor-
phme : la langue, pour le discriminer na invent aucun morphme dnotant la prsence
ou labsence de lespace dans les entits nommes par ces noms. Aussi la premire langue
aura-t-elle t vocale, le geste tant une criture sur un espace suggr, sur une donne,
donc, non prise en compte par la systmatique de la langue.
*/CC . . . / > /CVC . . . /28
La premire voyelle des res communes construites sur trois consonnes

radicales, cette premire voyelle que le systme syllabique impose, sem-
ble avoir t le signifiant de lanimit.
Exemples :
/kalb/, chien vs /milh /, sel
Nulle voyelle entre les deuximes et troisimes consonnes radicales. Le

genre et le nombre, qui sont les autres dterminants propres aux res,
sont, systmatiquement suffixs.
Exemples :
/kalb-a-t/, chienne vs /kalb--t/, chiennes
La premire voyelle des modus communs galement construits sur des

racines de trois consonnes, cette mme voyelle impose, tait, est res-
te irrgulirement, le signifiant de la diathse, subjective ou objective.
Paradigmatiquement, les modus ne pouvaient se distinguer des res que
par la prsence dune voyelle entre les deuxime et troisime consonnes
radicales. Cest effectivement cette voyelle qui les a distingues des res.
Exemple :
/kalab/, rage vs /kalb/, chien
Cette voyelle, la h arakat al-ayn de la tradition grammaticale arabe, est,

dans /kalab/, le signifiant de la prsence du temps. Elle est le premier
signifiant du temps. Un temps non prcis.
/kalab/ est, dans la tradition grammaticale arabe, un masdar.
Le terme masdar, source, origine , a t choisi parce que certains

grammairiens, les grammairiens dal-Basra, les premiers semble-t-il, ont
fait de cette forme, hors plan, la forme mre du verbe.29
28
Ibn Fris, as-Sh ib, p. 3839 : inna l arab-a [. . .] l t-ajmau bayna skin-ay-ni
wa l t-abtadiu bi skin-i-n .
29
Voir Ibn al-Anbr, Al-Insf f masili l-xilf bayna n-nahwiyyna l-basriyyna wa
l-kfiyyn vol. I, chapitre XXVIII ; al-Frb, op. cit., p. 6. ; Fleisch, 1979 : 149 ; 1995.
514 andr roman
La tradition na donc pas dfini le masdar en lui-mme.

Il apparat ici comme un modus infinitif.
Lopposition du modus infinitif, /kalab/, la res anime, /kalb/, est un

exemple de la relation biunivoque, gnrale, , qui seffectue, dans le
systme de nomination de chaque langue, comme une opposition, vs ,
entre deux paradigmes ; exemple :
2 1 1 2
/katab + t a/ vs /t a + ktub(u)/
Tu as crit vs Tu cris
ou entre deux schmes ; exemple :
/katab + t a/ vs /katab + t i/
Tu as crit (homme) vs Tu as crit (femme)
Dans le systme de nomination, cette relation biunivoque reste une,

car, tant elle-mme opposition , elle ne peut gnrer, par opposition,
aucune autre relation, diffrente, qui serait univoque.
5.3 Des racines dune seule consonne

Les racines de trois consonnes ne sont pas les seules racines des langues
smitiques. En effet, ct des units de nomination commune, qui
sont des images diffrencies du monde, et sont donc trs nombreuses,
il existe dautres units de nomination non plus commune mais gnrale
ou banale qui sont des images indiffrencies du monde, peu nombreu-
ses donc.30
Le grand nombre des units de nomination commune a impos leur
construction sur des racines de trois consonnes.
Le petit nombre des units de nomination gnrale ou banale a per-
mis leur construction sur des racines dune seule consonne.
Les res banales sont :

la premire et la deuxime personne, remployes, tour tour, dans
chacune de leurs occurrences ; exemples : //, je , de racine , la
30
Banal , qui se met la disposition de tout le monde , Littr.
racine de la premire personne ; /ka/, te, de toi , de racine k, la

racine de la deuxime personne ;
les res montres, remployes semblablement, tour tour ; exemple : /
t/, celle-ci , de racine t, la racine des res dostension ;
les res reprsentes, remployes semblablement, tour tour ; exem-
ple : /hu/, lui, le, de lui , de proto-racine *c, la racine des res de
reprsentation.31
Les res gnrales sont :

les res vagues, exemple : /m/, quoi ? , de racine m, la racine de la
res gnrale ; t, racine du temps gnral, homophone de la racine t
des res dostension ;32 n, racine du lieu gnral.33
Les modus de racines monoconsonantiques sont :

le modus /f/, avatar de lancienne racine, p, du modus gnral ;34
le modus /s/, galement ralis //, faire , avatars de lancienne
racine *c ;35
le modus /yy/, tre , de proto-racine *c ;36
le modus d assertion , de proto-racine *m ;37
31
La proto-consonne */c/, occlusive medio-palatale sourde, se transformera en per-
dant, par lnition, son occlusion : */c/ > *// > // > /s-/ > /h/ > /-/ ; ou, sa sourdit : */c/ >
*/ / > /y-/ /-yy/. Seules les consonnes qui sont affectes dun tiret sont encore employes
dans la langue arabe historique en tant que pices de son systme. La branche sourde est
termine par locclusive glottale // qui a remplac, contre-courant, la constrictive /h/
dont elle est issue. En effet // est moins diffrente que /h/ des voyelles devant lesquelles,
gnralement, elle se trouve. Quant /yy/, cest la mme consonne vocalique /y/, mais
elle a t allonge pour assurer sa dfense contre les conditionnements des voyelles car
elle est toujours, en fin de forme, en position intervocalique ; /y/, diffremment, est tou-
jours en dbut de forme.
32
Lhomophonie de certaines racines monoconsonantiques est annule par leurs dis-
tributions diffrentes.
33
la diffrence de /m/, dans /m-aktab/, de /t/, dans /kitba-t/ (chacun de ces deux
morphmes appartient la forme dans laquelle il se trouve), /n/, est une autre unit
de nomination en fonction dexpansion annective de /maktab/. Unit de nomination
non spcifie, /n/ est la tte du paradigme des expansions annectives, spcifies, de la
langue arabe ; /maktab-u-n/ signifie, littralement, bureau dun lieu indtermin >
un bureau ; do lidentification traditionnelle de /n/ comme un article indfini ;
do son nom traditionnel, tanwn, cest--dire nounation , prsence de /n/ .
34
Exemple dans /kayfa/, Comment ? . Ce modus gnral, concurrenc par les deux
modus tre et faire tait sans emploi ; il a t cr par le jeu des oppositions binaires
constitutif de la langue. Il nexiste que dans /kayfa/.
35
Exemples dans les verbes de paradigmes / + af ala/ et /(i)s + taf ala/ : / + ahadta/,
Tu as fait que qqn tmoigne ; /(i)s + tahadta/, Tu as fait que toi-mme sois tmoin .
36
Exemple : /m + aktab i + yy/, bureaucratique ; littralement : tre bureau .
37
Exemple : */allh - u + m/ > /allh + u/, Allh.
516 andr roman
le modus d invocation , de proto-racine ;38

le modus d exclamation , de proto-racine .39
La combinatoire des consonnes dans le cadre des racines de trois conson-

nes est double par une combinatoire de racines.
Exemples :
/ - a - ktub u/, Jcris , construit sur les deux racines et k-t-b ;
/kitbat/, temps dcriture , construit sur les deux racines k-t-b
et t ;40
/m a ktab/, objet dcriture, bureau , construit sur les deux raci-
nes m et k-t-b ;
/m a ktab i yy/, bureaucratique , construit sur les trois racines
m, k-t-b et *c > /yy/.
Ainsi le systme de nomination exploitait deux combinatoires :
une combinatoire de consonnes radicales ;

une combinatoire de racines associant :
soit une racine monoconsonantique une autre racine monoconso-
nantique ;
soit une ou deux ou trois racines monoconsonantiques une racine
triconsonantique.
5.4 Units non construites sur des racines

Face aux units de nomination construites sur des racines, la langue
a produit des units non construites sur des racines. Ces pices de la
machinerie de la langue, ses modalits , sont dans le systme de
nomination les partenaires des racines. Elles jouent par rapport elles
un rle de dterminants .
la diffrence des racines, les modalits sont les images linguistiques
dentits du monde qui peuvent tre apprhendes, navement, non pas
comme des entits communes mais comme des entits universelles, le
temps, la vie . . . 41
38
Exemple : /a + ayx-u/, Cheikh !
39
Exemple : */allh - a + / > /allh + a/, Allh !
40
Exemples dans le verset II :280 : /fa nazirat-u-n il maysarat-i-n/, Alors attendre
le temps [quil faudra] jusqu ce que [le dbiteur] soit dans une aisance [suffisante] .
41
Navement, une entit peut tre perue comme universelle si elle est apparat tou-
jours prsente. Sa prsence constante implique sa prsence partout. Luniversalit ainsi
Les modalits de temps sont employes particulirement dans les

conjugaisons. La modalit de vie a t employe comme une modalit
d animit , mle, femelle . . . 42
6. Les res, les modus et le temps
Les langues dtachent les res du temps.

Elles rattachent les modus au temps, or les res nexistent pas hors du
temps,
et les modus nexistent que dans ou par les res.
Les langues ne peuvent donc garder les res hors du temps et elles ne
peuvent davantage garder les modus hors des res. Il faut donc que, dans
le discours, elles accouplent les res et les modus.
Le premier couple {res modus} construit le noyau de chaque phrase
structure.43 Ses deux composantes sont ainsi relies, lune lautre, ipso
facto, par une relation intrinsque, biunivoque, galitaire : .
cette relation biunivoque, galitaire, de dclaration du lien, indis-
sociable, entre deux units de nomination,44 sopposent, ventuellement,
deux relations univoques, les deux seules relations mme dentrer en
opposition avec elle :
lune galitaire, la coordination , + ;

lautre non galitaire, la subordination , .
entendue lie le temps et lespace, indissociablement. Et elle nimplique avec lespace et

le temps que la vie, qui nat du temps, qui est avec le mouvement dans lespace lautre
mesure du temps. Le temps et la vie . . . sont signifis par des modalits ; lespace nest
jamais dnot que par des noms .
42
Les modalits aspectuelles spcifient diffremment non pas la relation des modus
au temps mais, par hypallage, le droulement intrinsque des modus.
43
Les modalits aspectuelles spcifient diffremment non pas la relation des modus
au temps mais, par hypallage, le droulement intrinsque des modus.
44
Les phrases non structures sont les onomatopes, units unaires trangres
lorganisation de la langue. Les langues animales sont unaires.
518 andr roman
SCHMA IV DES RELATIONS BINAIRES
[BIUNIVOQUE] [UNIVOQUE]
[HIRARCHISE] [NON HIRARCHISE]
DCLARATION SUBORDINATION COORDINATION

+
Les trois relations ainsi dfinies, ds lors quelles sont les seules relations
possibles, sont communes toutes les langues.
Ds lors quelles sont communes toutes les langues, le systme quel-
les constituent est universel.
Ainsi le systme de communication sest tabli pour toutes les phrases
de toutes les langues du monde sur le plan suivant :
PLAN UNIVERSEL DE LA PHRASE

noyau
+
{x (+ e . . . ) } {y (+ e . . . ) } extensions
extensions
+
{e (+... )} {e (+... )}
extensions
{x (+ e . . . ) } {y (+ e . . . ) }
. . . .
. . . .
. . . .
Dans ce plan, les units de nomination x et y sont, in prsentia

ou in absentia, les deux lments structurellement ncessaires du noyau
de la phrase, ses lments fondamentaux. La relation biunivoque, ,
qui les solidarise, les constitue en une structure unique : un duo de
deux voix .
Dans ce plan, x , y ..., e , e . . . sont des units de nomina-
tion ventuellement appeles par le locuteur, selon son besoin ; ce sont
des extensions, des complments sans aucune ncessit structurelle.
Un tel systme est simple.
Sa simplicit tient au trs petit nombre des oppositions en jeu, deux

seulement.
Sa simplicit tient aussi au caractre abstrait des trois relations nes
de ces oppositions.
Commun toutes les langues, ce systme est diffrent dans chaque
langue, superficiellement, par les morphmes spcificateurs propres
son systme de communication.
Le recours ces morphmes est rendu ncessaire par le caractre ab-
strait des trois relations en jeu. Et cest ce mme caractre abstrait qui le
rend possible.
Les morphmes spcificateurs propres la relation galitaire, biuni-
voque, de dclaration , sont principalement :
La diathse subjective ou objective ;

Le mode rel ou potentiel ou irrel ;45
Les modaux affirmatif, ngatif, interrogatif, impratif . . .
Les morphmes propres aux relations univoques sont :

les morphmes dfinisseurs de la relation univoque, galitaire, + ,
qui sont les coordonnants ;
les morphmes dfinisseurs de la relation univoque, non galitaire,
, qui sont les subordonnants , la modalit exceptive.
Ainsi les morphmes qui spcifient la relation biunivoque et ceux qui

spcifient les relation univoques sont complmentaires.
7. Lorganisation de la langue arabe
Ont t constates :
dans le systme de nomination,
la prsence dune structure radicale dans les units flchies ; lab-
sence de structure radicale dans les units amorphes ;
la prsence du temps dans les modus ; labsence du temps dans les
res ;46
45
La tradition grammaticale arabe na pas reconnu les modalits de mode.
46
Cette opposition, fondamentale, {res vs modus} nest plus vivante dans les langues
historiques.
520 andr roman
la relation, biunivoque, dopposition, entre les paradigmes eux-

mmes et les schmes qui les composent ;
dans le systme de communication,
les relations biunivoque, de dclaration ; univoques, de coordination,
de subordination.
Ces quatre oppositions, binaires, retraces dabord, dans la langue arabe

primitive rvlent un plan de la langue arabe exactement binaire.47
ARBRE DE LORGANISATION GNRALE PRIMITIVE DE LA LANGUE

ARABE
[SIGNIFIANTS] [SIGNIFIS]
[NOMINATION] [COMMUNICATION]
[TEMPS]

[RACINES] +
phonmes modus aspects res animit
syllabes
8. La structuration binaire de la langue
La reconstruction prsente de la systmatique de la langue arabe ne

saurait surprendre. Cest cette mme reconstruction quaboutirait une
tude qui commencerait, non plus par lobservation de la langue, mais
par lobservation de lhomme.
En effet, la parole de lhomme est convenue, ncessairement : il ne
saurait parler au hasard.
Si la convention qui rgle sa parole tait brute, cest--dire amorphe,
chaque entit, chaque exprience du monde, serait dite, immdiatement,
absolument, par une onomatope.48
47
La relation biunivoque, , est d opposition dans le systme de nomination ; elle
est de dclaration dans le systme de communication.
48
Ouille ! , sil faut un exemple, ne peut que crier une douleur de linstant.
Ouille ! ne peut crier le souvenir dune douleur passe, la crainte dune douleur future.
Les limites, limprcision, dune telle langue sont manifestes.

Une telle langue ne peut tre la langue de lhomo sapiens, faber.
La convention qui rgle sa parole nest donc pas amorphe mais
structure.
La premire structuration qui existe est la structuration binaire ; au
de, cest--dire hors structuration, le son profr est unaire ; il est
ralis comme une onomatope ; au-del de la structuration binaire la
structuration n-aire (avec n suprieur 2) tait et reste inaccessible
lhomme sans autre recours que son corps. Lhomme a forcment struc-
tur ses langues binairement, ds lors que la structuration binaire est la
premire de toutes les structurations possibles et quelle est la seule quil
puisse matriser.
9. Conclusion
La systmatique de ses langues a t impose lhomme par sa capacit

de combinatoire binaire, son langage naturel, sa premire langue com-
mune, la premire langue de ses conventions.49
Le plan primitif de lorganisation gnrale des langues humaines
naturelles, ne de la capacit binaire de lhomme, ne peut tre diffrent
du plan gnral de la langue arabe.
49
Les langues se sont constitues, par pas de deux , chaque signifiant propos
par le systme des sons trouvant, ventuellement, son signifi . Exemples de pas
de deux : C1V1C2C3 est le schma des res ; C1V1C2V2C3 est le schma des modus ;
C1uC2C2R3 est objectif ; C1aC2V2C3 est subjectif ; V2 = /u/, si le sujet nest que le lieu
du modus ; V2 = /a/, si le sujet produit le modus par son action ; V2 = /i/, si le sujet
produit le modus par sa raction ; si le sujet produit le modus par son action, la transitiv-
it ainsi ralise est soit afficiente, alors V1 = /a/ ; soit dficiente, alors V1 = // . . . Jeux
dopposition qui produisent les diffrences ncessaires mais aussi jeux analogiques qui
produisent les paradigmes ncessaires et, parmi eux particulirement, les jeux iconiques
qui mlent les analogies abstraites et les analogies concrtes ; exemple : laspect achev est
signifi par lordre {C - CCC} des racines ; laspect inachev par lordre inverse, {CCC
- C}.
522 andr roman
ARBRE DE LORGANISATION GNRALE DE LANGUES

CAPACIT BINAIRE
[SIGNIFIANTS] [SIGNIFIS]
[NOMINATION] [COMMUNICATION]
[TIME]

[RACINES] +
phonmes modus tenses res animit
syllabes aspects
Lorganisation primordiale retrouve par la dmarche qui commence dans

ltude de la langue et par la dmarche qui commence dans lobservation
de lhomo loquens est lorganisation binaire, simple et puissante.
La capacit de combinatoire binaire est le seuil de lhominisation.
10. Rfrences
Brague, Rmi. 1992. Europe, la voie romaine. Paris : Criterion.

Carter, Mike G. 1972. Les origines de la grammaire arabe, Revue des tudes islamiques,
40, 6997.
Descartes, Ren. (15961650). 1987. Discours de la mthode, texte et commentaire par
tienne Gilson. Paris : Vrin.
al-Frb (m. 339/950). 1350/1931. Ih s al-ulm, dition tablie et prface par Utmn
Muhammad Amn. Le Caire.
Fleisch, Henri. 1961. Trait de philologie arabe, Vol. 1 : Prliminaires, phontique,
morphologie nominale; 1979, vol. 2 : Pronoms, morphologie verbale, particules.
Beyrouth : Dar El-Machreq.
Ibn al-Anbr (513/1119577/1181). s.d. Al-Insf f masil al-xilf bayna n-nahwiyyna
l-basriyyna wa-l-kfiyyn (publi avec le Kitb al-Intisf min al-Insf de son diteur
Muhammad Muhy d-Dn Abd al-H amd), 2 vols.
Ibn Fris (m. 395/1004). 1382/1963. as-Sh ibi f fiqh al-lua wa-sunan al-arab f kalmi-
h, d. Musta f uwaym. Beyrouth: Badrn.
Kleiber, Georges. 2003. Faut-il dire adieu la phrase ?. LInformation grammaticale,
98. 1722.
Lavorel, Pierre Marie. 1980. Aspects de la performance linguistique Contribution
neurolinguistique et psycholinguistique lanalyse des systmes langagiers, Thse dtat.
Lyon : Universit lyon II.
al-Maydn Ahmad (m. 518/1124). s.d. Majma al-amtl, d. Muhammad Muhy

d-Dn Abd al-H amd (2 vols.). Damas-Beyrouth : Manrt Dr an-Nasr.
Meillet, Antoine et Cohen, Marcel. d. 1952. Les langues du monde, nouvelle dition,
CNRS, 2 vols. Paris : Atlas.
Mounin, Georges. 1960. Dfinitions rcentes du langage. Diogne 31, 99112.
Roman, Andr. 1998. Les particules dcomposes ou la reconnaissance des com-
posantes des morphmes de ngation de la langue arabe . In D. Leeman et A. Boone,
eds. Du percevoir au dire. Hommage Andr Joly, Paris : LHarmattan, 8796.
. 2005. La cration lexicale. 2e d. Lyon-Kaslik : Presses Universitaires de Lyon-
Presses Universitaires de Kaslik.
Saussure, Ferdinand de (18571913). 2002. crits de linguistique gnrale, texte tabli
par S. Bouquet et R. Engler. Paris : Gallimard.
Schuler, Bernard. 1990. Rhetorica. Rhetorica, 8, 229254, University of California
Press.
Sbawayhi (m. circa 180/786). 13851397/19661977. Kitb, 5 vols. Abd as-Salm
Muhammad Hrn, ed. Cairo.
Troupeau, Grard. 1976. Lexique-Index du Kitb de Sbawayhi. Paris : Klincksieck.
Versteegh, Kees. 1977. Greek Elements in Arabic Linguistic Thinking. Leiden : E.J. Brill.
. 1980. The Origin of the term qiys in the Arabic grammar. Zeitschrift fr arabische
Linguistik, 480, 730.
. 1993. Arabic Grammar and Qurnic Exegesis in Early Islam. Leiden : E. J. Brill.
. 1995. The Explanation of Linguistic Causes : Az-Zajjjs theory of grammar Amster-
dam : John Benjamins.
DIALECTS
HOW TO BE KOOL IN ARABIC WRITING:
LINGUISTIC OBSERVATIONS FROM THE SIDE LINE
Gert Borg
Nijmegen University
1. Introduction
Students of Arabiccertainly those of a former generationhave often

been led astray by a fatal fallacy: the Arabic language they study, in its
grammar and syntax, shines out as a neat structure of blissful regular-
ity. They learn by heart a set of rules with a high rate of predictability,
that govern the language that was used to convey Arab thought and cre-
ativity for 15 or 16 centuries. One might say that just like the Orient
is the dream of the Orientalists, Arabic seems to be the dream of the
Arabists.
But most Arabists are confronted with the harsh reality of the lin-
guistic situation in the Arab world during their first visit: a reality called
diglossia. It is bitter in a way to have to learn a second Arabic language
in order to be able to survive in everyday situations.
The actual situation of the foreign learner in the Arab world is not very
helpful either: if he or she gives the impression of being able to speak
some Arabic, the addressee may well decide to speak a higher variety
of the language out of politeness and respect, or start rattling in ver-
nacular assuming that for a foreigner to speak MSA would imply even
more strongly that he or she has an easy command of lower language
levels. As a result the linguistic confrontation between native speaker
and foreign language learner in the Arab world is hardly ever a natural
process.
Nevertheless the gifted and dedicated learner of Arabic can achieve
astonishing results in mastering this complicated linguistic situation.
Diglossia of course has some consequences for the teaching process
of the Arabic language; fundamental choices have to be made. Do we
teach Modern Standard Arabic, one of the modern dialects or some
kind of mix between the two? Must this mix be presented synchronic-
ally or should it be taught diachronically, in this case leaving open the
question with which variety of Arabic to begin.
528 gert borg
The choices and options become aggravated by growing pressures to

make university programs more time efficient and productive. Andas
if this were not enoughtraditional academic ambitions aim at a wider
study of history of the Arab world, of culture and of religion as well,
apart from the obvious purpose, that students of Arabic Studies are
expected to know their way around in Arabic and in the present day
Arab world.
Although we usually consume Arabic in written form, we soon come
to realize, that the linguistic reality of the Arab world is by no means
as static as the written material suggests: among Arabs new needs and
concepts emerge, new answers and solutions to the diglossic situation
are implicitly or explicitly formulated and this momentumshaped by
the mobility and flexibility of actual peopleis unfolding in a typically
linguistic way: anarchy.
It is obvious, that something static as a language course, a grammar or
even recently collected audiovisual material cannot possibly follow the
volatile dynamics of this process.
2. Ih n, sawt jl bi-h luh
In this contribution I will discuss a single phenomenon that represents

the present stage (August 2005) of this flexibility and anarchy in written
Arabic in Egypt: the monthly Ih n, sawt jl bi-h luh is a glossy maga-
zine in its first year. It is printed and distributed in Cairo where it sells
for LE 5 (= around 0.75). The names of the main editors are d arf
al-Jazr, Jhn Mahmd and arf al-Alf. The number of contributing
editors (muh arrirn) is 13.
It is not easy to identify the intended readership, but on the basis of
the various contributions and the goods advertised in it (mobile phones,
cars, beer, traveling and leisure) one might assume roughly that the
readers would be between 18 to 30 year olds, well to do and with a wide
range of interests from pop music, sex and vintage cars to (anti-Amer-
ican) politics, recent Egyptian history and backgrounds of hot news
items in Egypt and abroad.
The general character of this magazine can be qualified as:
critical open minded
easy to read fairly independent
cultural humorous and satirical
how to be kool in arabic writing 529
The political sympathy of the editors lies apparently with the Kifya
popular movement in Egypt.
The most significant feature of this magazine is its use of two lan-
guage levels in various ways: contributions in MSA, others in Egyptian
colloquial (mmiyya) and some with a mixed use of language.
I conducted a basic round of questioning about the opinion of the
reading public and the following image emerged: it is immensely popu-
lar, also in circles that are not intended, like 14 year old school girls and
their mothers, who appreciate the easy reading of this magazine. Some
readers however regret the use of colloquial whichin their opinion
hampers children in learning real Arabic, but they enjoy the reading
nonetheless.
In a lecture at the Netherlands-Flemish Institute the Egyptian lin-
guist Madha Doss emphasized, that the editors of this magazine use
Egyptian colloquial, not because they do not know how to write MSA/
fush they actually seem to publish in MSA elsewherebut that they
choose to use colloquial according to the character of this particular
magazine.
To illustrate the use of modern colloquial in Egypt and to give an
impression of the purposes for which it is used I selected one issue of
this magazine to be discussed here: nr. 8, published in August 2005.
2.1 Fragment 11
Let us first have a look at a contribution that is purely in MSA. It is the
beginning of a contribution entitled: ahr al-irhb al-aswad2 . . . black
month of terror, about the terrorist attacks in London and the murder
of the Egyptian Ambassador in Iraq:
(yawm al-irhb al-aswad) kna hd huwa l-unwn ar-ras li-jardat
al-ahrm as-sdira sabh yawm al-juma [blank] yliy wa-knat
al-jarda bi-hd al-unwn turu il l-ah dt al-irhbiyya llat waqaat f
l-yawm as-sbiq, h aytu knat iddat infijrt qad waqaat f qalb al-sima
al-bartniyya london mdiya bi-h ayt al-aart wa-kna tanzm
al-qida f bild al-rfidn qad alana an qatl as-safr al-misr f l-irq
hb arf . . . ,
1
By the nature of this contribution many quotes are from texts in which it is not
always clear whether the Arabic should be understood as MSA or mmiyya. Therefore
the transliteration may sometimes be inconsistent.
2
Cf. pp. 1011. All loose page numbers refer to this specific issue of Ih n.
530 gert borg
(a summary of al-Ahrms report of the terrorist attacks in London and

the killing of the Egyptian Ambassador in Iraq).
Up to this point the choice of language and register may have been dic-
tated by the source that is quoted, al-Ahrm, but after this the article
continues:
ill anna al-irhbiyyn lam yaktaf bi-hd al-yawm li-l-qiym
bi-hajamtihimi llat stamarrat tiwl ahr yliy mustahdifah (sic: h)
amkin mutafarriqa min al-lam. Fa-qad istayqaza al-misriyyn sabh
yawm as-sabt 23 yliy amalan f l-istimt bi-ajzt at-tawra qabla an
yufja bi-anna bosla at-tafjrt al-irhbya qadi ttajahat il madna
arm a-ayx al-misriyya . . . (the author turns to the recent terrorist attacks
in Egypt).
In his own words the editor continues the language and register as set
by the caption, referring to the terrorist attack in Sharm el-Sheikh. The
remaining part of this article is more of the same, a language register
that I would qualify as just above the level of average Media Arabic. The
level even rises a little towards classical in a context in which the editor
compares these erring groups of Muslims ( firaq islmiyya dlla) with
the Xawrij, quoting extensively from H adt and Qurn.
The choice of language in this contribution is quite predictable:
commentary on news item o Media Arabic or higher

religious/historic context o Classical Arabic
2.2 Fragment 2
The Ih n-editor al-Alf conducts an interview with an elderly man who
is fishing from the kbr ag-gmia. The photographer he has with him
is named Huseyn.
ajba h usayn f taraddud la, mi hanistann li-h adde m samaka titla.
ibtasam ar-rajul wa-ml li-yaftah antatah wa-axraj minh samaka
ka-annah muidda xiss san li-t-taswr wa-abakah f xutt f as-sinnra
wa-rh yahuzzuh h att tazhar ka-annah sh iya wa-btilab amm
al-kmr
(Huseyn refuses to wait to shoot his photographs because of the failing
light:) No, were not going to wait till a fish pops up. The man smiles
and turns to his bag, producing a fish as if prepared for the photo, hooks
it to the rod again and starts shaking it till it looks like its alive, playing
for the camera
saaltuh d nh h y ustd h asan?. anm. anm?. anm . . bi-

l-mm. amm?. anm. anm . . . ah . . .
(the interviewer asking about the kind of fish: a very realistic conversation
indeed).
2.3 Fragment 3
A truly remarkable article is wijha nazar,3 point of view, a contribu-
tion physically surrounded on its page by a quotation from the Qurn:
rabban l tuxidn in nasn aw axtan . . . (Q. 2/286). The title of
this contribution is y rabb . . . anta rif (O Lord, You are aware . . . )
qla llh: wa-id saalaka ibd ann fa-inn qarbun ujbu dawata
d-di id dan (= Q. 2/186) y rabb anta qarb minn, lh ana bad
annak? y rabb anta rif kulle h ga, wa-rif (q)adde h ana muhtglak
wa-rif (q)adde h gahl wa-duf byixalln abad annak wa-amil h gt
tzaallak aw mamal al-h gt ell turdk aw amilh bass mi bi-t-tar(q)
a ell turdk. Ana rif ana (q)adde h wih i mak, wa-rif enna l-h gt
wi-niam ell anta addthl akbar min ayye h ga a(q)dir amilh
alanak, lh ba(q) l-whid byistashal ennuh yamil (q)alle h ga wa-
huwa fkir ennuh kedah xals amal kull al-matlb? . . . lh al-duny dalma
kedah quddm al-wh id wa-lh al-h ayt saba? Ana mi rif wa-mah ad
rif lh kulle h ga btih sa l f l-duny btih sa l kedah lh . . . I feel guilty about
all my wrongdoings in the light of your benevolence, o Lord, and I dont
understand the world anymore . . .
This text is very personal in tone, it is a monologue with God, almost
like a prayer and it is in mmiyya throughout, except verbal forms like
turdk.
What we have here is a religiously inspired text, in the context of
Qurnic Arabic, but written for a purpose that isapparentlybest
served in colloquial.
2.4 Fragment 4
H ukkm l yahbna amrk, an article about world leaders who defy
the US and its politics,4 is also mainly in MSA and Media Arabic. Typical
markers for this second type of Arabic are for instance passages like:
3
Cf. p. 18.
4
Cf. pp. 2526.
532 gert borg
tumma dahab5 Castro il qatar li-tazz al-alqt bayna l-baladayn

xssa tan wa-anna l-alqti llat tarbut qatar bi-l-wilyt al-muttah ida
watqa. tumma dahab il sriy wa-wasafa ziyratahu li-dimaq bi-annah
muhimma wa-annahu yaqif il jnib sriy . . . . . . .etc.
. . . then Castro travelled to Qatar to strengthen the relations between the
two nations, especially because Qatars relations with the US are solid.
Then he went to Syria and described his visit to Damascus as important
and (stated) that he stood next to Syria . . .
The type of language corresponds neatly to the subject of this contribution.
2.5 Fragment 5
Announced as a grain of seriousness (h abba jadd) we find another
contribution in MSA on the daring subject of the role of sex in matrimo-
nial relations: m hiya h udd al-alqa l-h amma bayna l-azwj? 6 (what
are the boundaries of the intimate relationship between married part-
ners?). With captions like an-nazar il awrat az-zawja (looking at your
naked wife), h urriyyat al-wad al-jasad atn al-jim (freedom of body
positions during intercourse), it does not shy away from daring subjects.
The whole article is written in MSA except for some anonymous reac-
tions by boys (ray a-abb) and girls (amm l-fatayt fa-tah addatna
ka-t-tl). These reactions follow separate patterns:
in the boys section the introductory remarks are in MSA, which

gradually turns into mmiyya. Finally the boys are cited in full
mmiyya:
wa-al l-rami mimm qad yazunn al-bad min annahum l yatahaddatn
f tilka l-mawd . . . . ad-dukr arab an rayihim bi-bagh a7 wa-ka-
annan gn lahum a-ag-garh . qla ahaduhum: ih n makbtn8
wa-madtn min (!) al-mugtama . . . . although some expected them not
to talk about such subjects, the men expressed their view with impudence
as if we were touching a sore point. One of them said: we are oppressed
by society . . .
5
The use of the verb dahaba is extremely rare in Media Arabic: compare the Nijme-
gen Arabic corpus by Jan Hoogland.
6
Cf. pp. 2829.
7
A typical masdar for mmiyya, not found in classical and MSA dictionaries, but in
Badawi-Hinds.
8
A typical mmiyya word.
the girls section is less predictable:

wa-ih d l-fatayt qlat f tamallul: wa-h-yifd bi-h marifat ray f
al-awd ag-ginsya . . . whats the use of knowing my opinion about sex-
ual positions!
wa-law yiz tifham nafsiyyat al-bant mila izzy f l-mawd dah f flm
woman on top fa-huwa yah k qissa t fatt rmnsiyya tazawwajat bi-bb
wasm (if you want to know how womans psychology works in this matter,
than look at the film woman on top; it tells the romantic story of a girl
that married a handsome boy . . . )
Then the story of the film is being told in MSA.
2.6 Fragment 6
A constant mix between MSA and colloquial is a contribution by Yumn
Bassiouni, called al-istina bi-mumatti ln kmbrs min ajl az-ziyrt
ar-rasmiyya,9 calling in extra representatives at official visits for help,
a hilarious story about the hiring of extra patients to accommodate
the official opening of a hospital wing for children on the 4th floor of
the Ab R Academic Hospital. The first line of this story is indica-
tive for its mixed character: wa-alan uakkid annahu mi mugarrad
kalm, hunka mitln h adat muakka ran yuakkidn al mawd
al-istina . . . to make sure that its not only a rumour, here you have
two examples that occurred recently, confirming the topic of calling
in . . . .
From a linguistic point of view the use of mmiyya in this article is
interesting: a remarkable colloquial factor is the use of mmiyya con-
junctions instead of MSA ones: apparently this contributes to the fluent
style of writing:
. . . h att az-zar (!) byitl alan yith att f makn ziyra rasmiyya tnya
(for li . . . )
. . . wa-lkin inn al-mawd yawsal (for lkinna)
. . . wa-alan uakkid . . . (for li . . .)
. . . - zayye m binf f t-tilifizyn(for kam)
. . . kullu deh wa-l-umr m zlat tabd tabiyya li-l-ya . . . (for hd
wa-)
9
Cf. p. 31.
534 gert borg
The mmiyya can also be used to indicate direct speech without using
quotation marks. Doing this makes the scene lively and realistic, because
we can hardly do anything else than picturing the usual hanger around
in such buildings, all the more so because typically the floor that is said
to be still closed is the floor that was meant to be inaugurated from the
beginning:
. . . f d-dr ar-rbi ell lissa muftatah nuh . . . on the fourth floor that
they didnt open yet.
2.7 Fragment 7
Ih n apparently did some investigative research into the behavior of
female singers on the various MTV-like channels and the impact it has on
the audience in an article with the title man tufaddil min al-muanniyt?,
Which of these singers do you like most?.10 Some youths give their
commentary in which the choice of language seems significant for
emphasizing the point of view. A conservative 23 year old boy says:
masxara al-an ntija an s ahwl al-mujtama wa-laysat al-sabab fh
wa-l-ns mudnibn h aytu annahum hum yuajjin tilka al-an . . . the
cause of those ridiculous songs is the deterioration of society, not the other
way around; people have to blame themselves because they are encourag-
ing these songs . . .
A 25 year old girl sounds a bit more committed:
al-fdiy klb ba isff wa-axjal min muhadat qanawt al-an amm
wld video-clips are just pulp; I feel embarrassed to watch them when
my father is around
This 24 year old girl is very dismissive:
hirt hd az-zamn asbah na yulinna an anfusihinna f t-tilifizyn f
akl fdiy klb the whores of this age have come to express themselves
through the video-clip
And this 27 year old girl sounds critical:
law nrakkaz al urbitn mumkin namil h gt original bass al-Arab
(ih n yan) dyn if we concentrate on our Arab identity we could come
up with something original, but the Arabsthat means weare lost
10
Cf. pp. 3233.
A 36 year old male:

bs samr law m amalit kedah mh ad hayiufh if Busi Samir wouldnt
act that way, nobody would notice her
The background of the speakers shifts from:
conservative and intellectual to

sympathetic but dismissive
conservative and denouncing
conservative and nationalistic
realistic
and with these attitudes the language variety shifts correspondingly from:
MSA to
popular (mmiyya)
MSA
street wise (mmiyya)
popular (mmiyya)
2.8 Fragment 8
Walid Irfas contribution is a review of Paulo Coelhos book The Zahir,
which was popular at the time. The language level chosen for this article
is close to Classical Arabic.
2.9 Fragment 9
A special page is reserved for reviews of DVDs. These are all in MSA. The
writer, Marwn Qadr, also comments on the Broadcast and Television
Festival in July 2005.11 On this subject his criticism is harsh: wa-ka-da
kull al-qimn alayhi (sc. al-mahrajn) faal al-mahrajn faalan daran
min h aytu t-tanzm wa-l-iftith wa-l-xitm . . . as usual with all organiz-
ers the festival failed completely in organization and during opening and
closing sessions. Only once Qadr deviates slightly from MSA: muzam
ad-duyf wa-l-mutarikn lays . . . , most guests and participants were
(plural) not . . . , but this is far from abnormal in Media Arabic. When
11
Cf. pp. 3637.
536 gert borg
the author voices his indignation about the unavoidable balad/fellh

dancer performing at these occasions he shifts to mmiyya:
l in anh f ayyi htifl h att wa-law kna tahr ibn al-jrn wa-llad
yablu min al-umr 7 sanawt. Lh .. lh al-raqsa l-fallh ? H arm alk
zihiqn gaddad, huwa int mi btitfarrag al ayye stirdt aw h aflt
xitm aw iftith f ayye h itta barra masr. Kulluh km wa-l-wufd aw
bi-man asah h ruas al-wufd km tn . . . there is no avoiding her
at any festive occasion, even if it be the neighbors son circumcision whos
only seven years. Why . . . why this peasant dancer? Shame on you, were
fed up with it, find something new(?), its because you havent seen any
festivities of openings or closures anywhere outside Egypt. All that is one
thing, but these delegations or more specifically the delegation leaders are
another!.
2.10 Fragment 10
In an article with the title bsbr axdar m yisw12 (a green passport
isnt worth anything) Yumn Bassiouni complains about the first class
treatment visitors to Egypt receive from the authorities if they carry for-
eign passports. As witnesses to this practice she cites a few personal sto-
ries, parts of which are in mmiyya. One account is by Mohamed Sami
(22) who traveled in Egypt in the winter season with a relative of his
who carries an American passport:13
ya ibn xlat f amrk wa-ndiran m yat il misr . . . . . wa-kunn
rjina min xarga14 bi-l-layl mutaaxxir(!) jiddan . . . wa-awqafan ad-
dbit wa-kna kik(!) fn li-anna ibn xlat kn lbis rt f izz al-bard!
wa-awdaha muhammad anna ad-dbit akka f annahum mutatn
h ga muxaddara muxalliyhum mi h ssn bi-nafsihim wa-lbsn kedah,
xssa tan wa-anna lukna qarbih f l-arab mi salma wa-sammam ad-
dbit annahum yatl a al al-qism. al-h ga l-wah da ell anqadatn
annan gibn li-d-dbit gawz safar ibn xlat al-amrk wa-hadadnhu
annan hanatakh f s-sifra l-amrkya an al-bahdala ell h asalit/hasalat
la-n wa-bi-t-tl mumalat ad-dbit itayyarat tamman wa-tadar kitr
alan m namil makil! . . . yan ibn xlat lamm kn byistamal
al-bsbr al-masr maa dalika d-dbit , m la r s mumla wa-awwal
m azhar bsbruh al-agnab kn lahu al-ihtirm a-add wa-l-adab f
l-mumala
my cousin lives in the US and rarely comes to Egypt . . . we were return-
ing from Kharga very late at night . . . and a policeman stopped us and
12
Cf. p. 41.
13
Passages in colloquial are in italics.
14
Written is xurga which seems improbable.
suspected us because my cousin was wearing shorts in this extreme cold!

Mohamed made it clear that the policeman was wondering whether they
had taken some drug that left them not feeling themselves while they were
dressed that way, especially because the accent of his relative in Arabic was
not ok. The policeman decided that they had to go to the police post. The
only thing that saved us was that we handed the policeman my cousins
foreign passport and we threatened him that we would complain about
him at the American Embassy about the illtreatment that we encountered.
All of a sudden the policemans behavior changed completely and he apol-
ogized deeply so that we wouldnt cause him any problems! So when my
cousin used his Egyptian passport (in dealing) with that policeman, he
encountered maltreatment and as soon as he waved his foreign passport
he was treated with deep respect and decency.
In this passage the story changes rapidly from account to direct speech
and the language changes accordingly.
2.11 Fragment 11
In the first lines of the contribution tanfsah15 (reassurance?) we find a
remarkable blend of MSA and mmiyya:
Kun Anta. Ql iraf nafsak dah ahamm wa-asab wa-aktar al-ns
l yarifn!! Amm d h ga arba kulle m inn abuss f bitqat arif ana
mn wa-ahdat at-taxarrug wa-ahdat ag-g wa-al-bsbr . . . Be your-
self. They say: Know yourself. That is the most important and the most
difficult thing, while most people dont!! That is something strange: every
time I look at my ID I know who I am . . . , my degree, my army card and
my passport . . .
But then the mmya takes over almost completely:
Huwa deh ana wa-l deh elli itfarad alayy enn aknuh . . .
Amil ell inta yizuh . . . yiz arabiyya bi-rmt bitru mayyah (unzur
namdaj 2)16 yiz tilab mazk! Yll nirh ri muh ammad al bukra
mumkin yigibak gtr wall darabkah . . . yiz tiqr!?wa-llhi fikra mi
batt laindin maktabt fh kull al-mawdt ell titxayyilh, mamak
fuls?! inzil sq17 al-ezbekya fa (!) al-ataba . . .
That is me and not the one who imposed on me that it is me . . . Do what-
ever you want . . . You want a car with remote control that sprays water (see
15
Cf. p. 42.
16
Referring to a photograph.
17
The text reads sr, but that seems improbable.
538 gert borg
example18 2). You want to play music!19 Lets go to Muhammed Ali Street
tomorrow. Maybe youll like a guitar or a darabukka . . . you want to read
sure, not a bad ideawe have bookshops with all subjects that you can
imagine, you dont have money?! Go to al-Ezbekiyah and Ataba.
In this fragment MSA is used for the general introductory remarks; the
more subjective fictitious monologue is in colloquial.
2.12 Fragments 1216

A few remaining articles about (muscle) cars, sporting exercises, qual-
ity food, gadgets and the unavoidable horoscope are all completely in
MSA.20
2.13 Fragments 17 and 18

In a magazine that is so obviously playing with the possibilities that lan-
guage levels offer it is only natural that the reader would find contribu-
tions about language itself. In this issue of Ih n we find two, both under
the common and significant title Arbk slng.21 We find two lists of
say and dont say items (qul and l taqul), here represented as
opposites:
Say Dont say22 Say Dont say
matar natar dafar dfa/ir (?)
silsila sinsila muhandisn muhandizn
gurnl gurnn bsbr bzbr
bartamn batramn23 isbniy izbniy
fingn fingl dalwaq/t dawaq/t
And under the heading qms Ih n (Ih ns dictionary) we find a qua-
si-classical explanation of the word sakalnsun (sic):
kalima taxdum jam al-ard al hasb al-man l-murd: yumkin
istixdmuh bi-man kullih f t-tamm kulluh f s-sakalns, wa-yum-
kin al-ira li-l-by frnd aw al-grl frnd al annahum as-sakalns.
18
Example refers to a photograph accompanying the text.
19
Mark the calque for azaf.
20
Cf. pp. 4450.
21
Cf. p. 6.
22
These lists seem to be provocative to me; please go ahead and say gurnn!
23
Badawi-Hinds mentions both but apparently prefers batramn, a glass or plastic
jar with a lid, a jam jar.
kam yumkin al-qawl bi-anna axs (m f t-tarwa?) bi-annahu f

s-sakalns. s-sakalns tustaxdam f l-qahw li-tasmiyat al-kktlt:
qahwa al yzabd bi-l-fawkih . . . lx. f l-matim, as-sakalns tur
il sndwitt al-fl bi-t -tamya bi-l-bayd aw sandwitt as-samak bi-g-
gambar bi-s-sabt (?) bi-t-tahna wa-llat turaf aydan bi-sandwitt
al-fiygr. sakalns tamsr li-kalima aksalns . . .
an all purpose word depending on the intended meaning: it can be
used as a generic term: its all sakalns. It can refer to your boy friend
or girl-friend: they are sakalns. One can also say that someone . . . (?) is
in sakalns. sakalns is being used in coffee shops to indicate cocktails:
coffee on tea, cream on fruit etc. In restaurants sakalns refers to bean
sandwiches with ta mya and egg or fish sandwiches with shrimps with ?
with ta hna that are also known as viagra sandwiches. sakalns is an egyp-
tianisation of the word excellence.
A wonderful play of words and styles.
3. When is mmiyya being used in this magazine?
The use of mmiyya seems neither inconsistent nor indiscriminate in

this magazine; I will try to clarify the preference of the contributors for
using mmiyya in its various contexts:
1. mmiyya is used for representing direct speech and to add a realistic

flavor
la, mi hanistann li-h adde m samaka titla (fragment 2)
also for fictitious direct speech: yiz tiqr!?wa-llhi fikra mi
batt la(fragment 11)
2. for adressing the reader
zayy m binf f t-tilifizyn (fragment 6)
3. for contexts with humorous, hilarious or ironical purposes
f d-dr ar-rbi ell lissa muftatah nuh (fragment 6)
h att z-zar byitl alan yith att f makn ziyra rasmya tnya
(fragment 6)
4. for intimate and personal expression
fragment 3: the prayer
5. for clarifying the social status or political point of view of the speak-
ing individual
the interviews in fragment 7
540 gert borg
6. for emotive contexts, like anger or indignation

the way indignation that is expressed about the baladi dancer in
fragment 9
fragment 5: ih n makbtn wa-madtn min al-mugtama
4. Orthography
It is obvious that for an Arabic periodical in most Arabic speaking

countries to use written mmiyya is like walking a minefield: it will be
frowned upon by the conservative cultural and intellectual elite and the
periodical might very well be unable to penetrate a wider Arab market
although in the case of Egyptian colloquial this risk is limited. It opens
however a thrilling space of new and unexplored opportunities. It has to
be said though that this experiment faces some practical problems, one
of which is orthography.
For Classical and Modern Standard Arabic the orthography is fairly
straightforward. Even in written Media Arabic the orthographical prob-
lems are mainly confined to transliteration of loan words and foreign
names. But for written colloquial the situation is completely different:
the Arabic graphemes do obviously not cover the mmiyya stock of pho-
nemes and no consistent convention has been developed yet. If this trend
of writing colloquial will persist, it would mean, that new orthographical
standards will emerge. It will be hard to continue the standard of one-
to-one equivalents as in written CA and MSA. For written colloquial
the orthography of the future might well be less stable and predictable.
In a few instances we can already get an impression of some of the
difficulties: do we spell as we speak or spell as the conventional spelling
of MSA tells us to spell: for example the (ayn) is clearly disappearing
from the urban Cairene dialect. As a consequence Ih n writes byarif
for the MSA yarifna. To do this on the cover even seems to be meant
as an elegant provocation, because the same phenomenon ( for ) can-
not be witnessed in the remaining articles except for the article that the
cover announces: istift ih n: 95% min a-abb byarif (with final
alif ) man bs samr . . . inquiry by Ih n: 95% of the kids know who
Busi Samir is (fragment 7).
In other issues of Ih n, not quoted here, I noticed a tendency to
represent stressed vowels by lengthening them, using matres lectionis,
which may be confusing for the unaccustomed eye.
Furthermore we find inconsistencies for instance in prepositions with

the first person suffix: liy, alayy, but biyyah.
The conjunctions wa and fa are increasingly represented and
therefore considered as separate words, not only in advertisement, but
also within running contexts like these: kull an-ns knit rfa maslaht
aktar minn: mm wa bb, al-mudarrisn, . . . everyone knew better
than me what was in my interest: ma and pa, the teachers, . . . (from
fragment 11); inzil sq al-ezbekya fa al-ataba . . . go to the Ezbekiya
market or Ataba (fragment 11).
5. Conclusions for teaching
For the present and coming generations of students of Arabic it might

be important to become acquainted with the notions and fashions of
their own generation in the Arab world. For them to learn about dif-
ferences and similarities will probably contribute to motivation and
mutual understanding.
These notions and fashions are for a large part expressed in language.
The present state of the Arabic language and its day-to day develop-
ments cannot however be covered by course books and vocabularies
because these cannot possibly keep up with the tempo of modern urban
life, the media or the internet.
The pace of changes in the Arab world as reflected in the quick devel-
opment of the written Arabic language confronts us, teachers of the
Arabic language and culture, with the question whether we can still
convey the concept of the written Arabic as a continuum; or should we
actually split the current language acquisition curricula into the catego-
ries classical and modern? The implications of options like these are
far reaching.
There is at least one practical consideration: teachers of modern Ara-
bic can hardly do without regularly being in touch with the Arab world
in order to adequately follow the linguistic developments in the Arab
world. This can be achieved either by following the (satellite) media
orbetter stillby regular visits to Arab countries to witness the fast
developments firsthand.
542 gert borg
6. References
Hinds-Badawi: Hinds, Martin and El-Said Badawi. 1986. A Dictionary of Egyptian Ara-
bic: Arabic English. Beirut: Librairie du Liban.
Ih n, sawt jl bi-h luh. 2005. August issue.
Nijmegen Arabic Corpus: a collection of MSA texts from the written media compiled by
Everhard Ditters and Jan Hoogland.
HELLO, I SAY, AND WELCOME! WHERE FROM,
THESE RIDING MEN?
ARABIC POPULAR POETRY AND POLITICAL SATIRE:
A STUDY IN INTERTEXTUALITY FROM JORDAN1
Clive Holes
University of Oxford
1. Introduction
In Arab literary studies, popular poetry, that is, poetry composed in a

non-standard form of the language, remains a relatively unexplored res-
ervoir of creative activity. There has been a tendency for native and west-
ern critics alike to ignore it, or at best pigeonhole it as folklore, devoid
of literary value, and written in a debased form of the language. One
of the few great Arab writers to stand out against the prevailing opinion
was the 14th century historian, sociologist and polymath Ibn Xaldn,
a writer who in this, as in so much else, was sui generis. His words are
worth quoting, since he puts his finger squarely on the main reasons for
the prejudice against popular poetry (ignorance) which applies with as
much force today as it did six centuries ago:
Most contemporary scholars, philologists in particular, disapprove of
these types (of poems) when they hear them, and refuse to consider them
poetry when they are recited. They believe that their (literary) taste recoils
from them, because they are (linguistically) incorrect and lack vowel end-
ings. This, however, is merely the result of the loss of the habit (of using
vowel endings) in the dialect of the (Arabs). If these (philologists) pos-
sessed the same (speech) habit, taste and natural (feeling) would prove
to them that these poems are eloquent, provided that their (own) natu-
ral dispositions and point of view are not distorted. Vowel endings have
1
The present paper is based on data gathered in the course of a field-based investiga-
tion into the practice of Bedouin poetry in Jordan and Sinai, involving the collection of
poetry composed over the last 50 years, and particularly over the last 20, direct from the
poets themselves. The fieldwork was carried out by Dr Said Salman Abu Athera, a Bed-
ouin originally from Beersheva but now resident in Jordan, and the translation, glossing
and annotating of the poems was done by me. Our book Y Kundlzza Rys! Politics
and Popular Poetry In the Contemporary Arab World is to be published by Brill in 2007.
544 clive holes
nothing to do with eloquence. Eloquence is the conformity of speech

to what one wants to express and the requirements of a given situation,
regardless of whether the u-ending indicates the subject and the a-ending
the object, or vice versa. (Ibn Xaldn, tr. Rosenthal 1958, 3:4145).
With the beginnings of western interest in Arabic literature, from the
mid-19th century until the early decades of the 20th, there was a good
deal of academic effort devoted to collecting examples of Arabic popular
poetry, but with a mainly philological rather than ethnographic or liter-
ary inspiration (e.g. Wallin 1851; 1852; Socin 19001; Landberg 1901;
Meissner 1903, though Musil 1928 is an exception). In recent years there
has been a fresh burst of interest, particularly in the contemporary Bed-
ouin poetic tradition (Sowayan 1985; Abu-Lughod 1986; Ingham 1986;
Bailey 1991; and Kurpershoeks monumental five-volume study of the
Bedouin poetry of Najd, 19932005). These later studies have focused
on the practice and social functions of poetry, and on its ethnographic
value as both the product and record of Bedouin societies and modes
of thought that are now rapidly passing into history. There is evidence
to suggest, however, that the tradition of popular poetry in todays Bed-
ouin society is far from dead; on the contrary, it seems to be transform-
ing itself into a voice of socio-political commentary and criticism which
transcends purely local tribal concerns and addresses regional and even
international issues. During the 20th century, Arabic popular poetry
was first deployed by urban poets, in the absence of any other public
forum for protest, as a potent political weapon, the two prime examples
being Mahmd Bayram at-Tniss barbs against Egypts British colo-
nists during the early part of the century (Booth 1990), which led to his
expulsion from Egypt, and his heir, Ahmad Fud Nigms coruscating
attacks on Anwar Sadat and the policy of economic and cultural infith
pursued by the Egyptian government during the 1970s (Abdul-Malek
1990), a campaign which landed him in prison on several occasions.
Popular Bedouin poetry, on the other hand, has until relatively recently
been little concerned with politics and society outside those of the tribal
dra; only with post-Second World War political independence, and
a perceived encroachment of the state on the concerns of their com-
munities have Bedouin poets gone down the same path as their urban
counterparts and turned their poetry into an instrument for expressing
communal popular protest.
But what is meant, in the early 21st century, by the term Bedouin in
the phrase Bedouin popular poetry, since virtually no one now leads the
arabic popular poetry and political satire 545
Bedouin life of nomadism? As a statement that today refers to member-

ship in a genealogy-based social structure, rather than to a way of life,
ih na badu, or more frequently ih na arab, is still a proud boast. Individu-
als whose families may have been settled for many generations continue
to use this phrase to identify themselves, however far their life-style has
deviated from the Bedouin stereotype of yesteryear. In countries like
Jordan, Iraq and Saudi Arabia, the concentric circles of family ties, clan
membership and tribal affiliation into which self-proclaimed Bedouin
are born continue to exercise a strong pull on their imagination and pro-
vide the backdrop against which their social relationships are conducted
and political attitudes forged, even in large cities.2
Contemporary Bedouin poetry is composed in a stylised form of
the poets spoken dialect, but this does not mean that it is a purely oral
form, or that the poet has had any formal training. Poets come in all
shapes and sizes, and not from any particular class or background: he
(or she, as there are some skilled female poets) can be a tribal elder, a
supermarket owner, a smuggler, or a doorman;3 what counts is a gift for
composing, honed through years of listening to other poets. The oldest
generation of poets now in their 70s and 80s are almost without excep-
tion illiterate, and use a vocabulary and forms of poetic diction now
difficult for younger generations to understand.4 Nowadays, however,
many of those who would describe themselves as Bedouin poets, like
the one discussed here, are literate in Modern Standard Arabic, but, for
a variety of audience- and topic-related reasons, some of which I will
discuss below, choose to compose in dialect, and use a vocabulary closer
to that of everyday speech. Written versions of their poems not infre-
quently appear in newspapers and are published in locally and cheaply
produced dwns.
2
In Egypt, in the mouths of the settled farmers and town-dwellers of the Nile valley,
arab is a term of opprobrium, signifying backwardness and stupidity; but for the Sinai
Bedouin, resentful of the power over their lives which the settled population of Egypt
now exercises, these farmers and town-dwellers are contemptuously referred to as ban
firn the sons of the Pharaoh, an insult that alludes to their supposed slave origin and
habit of subservience.
3
Poetry was collected from poets fitting all these descriptions in the course of the
present project.
4
A good example is the poetry of the illiterate Unz Ab Slim at-Turbn of south-
ern Sinai, considered by many the premier Bedouin poet of the region, who spent many
years in Egyptian prisons for smuggling offences, and died in his 80s in 2000.
546 clive holes
2. Ghassn Surr a-baylt ( Ab Surr)
One of the Jordanian poets from whom we have collected a large number
of unpublished poems, though many of them have circulated by word
of mouth and on cassette, is Ghassn Surr a-baylt (Ab Surr).
Ghassn was born in az-Zarq, north Jordan, in 1954, the son of the
paramount sheikh of the al-Uwayst section (ara) of the Ban H asan,
a north Jordanian tribe. He now lives in a large house in al-Mafraq. He
describes his social background with the tongue-in-cheek epithet ladu:
an abbreviation of l filh wa l badu neither cultivator nor nomad, a
term used to describe people historically of Bedouin origin who are now
fully settled. Ghassn was the eldest son in his family, with seven broth-
ers. When his father died, he was around 14 years old. His father left him
virtually nothing (300 JD), having sold off or squandered almost all the
family wealth. As a result, Ghassn was forced to leave school immedi-
ately and find work to support his siblings and mother. At that time, in
the late 1960s, the Iraqi army still had units in Jordan following the 1967
war with Israel. This presented opportunities for smuggling. At that
time the Iraqi army was using Russian weapons and ammunition, but
the Bedouin in Iraq still carried British weapons from the days of Brit-
ish influence before the revolution of 1958, just as the Bedouin of Jordan
still did. When Iraqi units rotated home, the young Ghassn would sell
them British weapons and ammunition, which they smuggled across the
border and sold on at a profit to Iraqi Bedouinthere were no border
checks on troop movements. This ammunition and weapons smuggling
subsequently broadened into a more extensive illicit tradecigarettes
and other goodsand extended also to Syria, which was a much nearer
border. By the early 80s, as a young man, Ghassn had begun to travel
regularly to West Germany and the countries of the eastern bloc, from
where he brought back cars and other goods to sell in Jordan. Eventu-
ally, the profits from this enabled him to buy land and start businesses,
including supermarkets. He is now, at around 50 years of age, a rela-
tively wealthy man. That is the bare bones of Ghassns journey from
childhood rags to adult riches. He is at pains to point out that none of
what he owns and has achieved was through family influence, but rather
through thirty years of his own efforts and honesty with business part-
ners: what he has achieved has been, as he put it, bi s-sadga wa l bi
l-garba (through friendship and not through family ties).
Ghassn is a complex character. Despite his lack of formal educa-

tion he has read quite widely, and is in the habit of quoting Plato in
his conversation. It became obvious from talking to him that he has a
thorough knowledge of the history of the Levant. He also, despite (or
perhaps because of the means by which he acquired) his wealth, has a
well-developed social conscience. This is strongly reflected in his poetry,
most of which was composed during the 1980s when he was struggling
to make a living and continually coming up against economic and social
barriers. His constant poetic targets are: the economic inequalities of the
Jordan of this period; the incompetence, hypocrisy, cupidity, corruption
and arbitrary powers of government ministers and officials; the nepo-
tism and snobbery endemic in Jordan at all levels; and perhaps most
emphatically, the loss of the sense of honor and dignity which, in his
view, no longer underpins the fabric of Jordanian society in the way it
once did. All of these he had to battle against to achieve what he has.
The satirical poem presented here, as well as many others, could not
be published in Jordan, and the poet ran into serious problems with
the intelligence services at the time when his poetry first began to cir-
culate informally some twenty years ago. How did, and does, Ghassns
poetry circulate? Most often a finished poem or poems would be recited
in a gathering in the poets majlis, and recorded there on a cassette. This
would be copied and distributed to local shops and sold for a very low
price. Bus drivers would often play such cassettes to their passengers. The
fax machine, when it became common in the 1980s, was another means
of circulation of the poem in written form. The latest method is via text
messaging on mobile phones. Although Ghassn still composes poetry,
he has turned away from the provocative and often personal attacks
of the 1980s, and stays behind what he describes as al-xutt al-h amr
(the red lines) that must not be crossedin particular anything that
could be construed as critical of the Hashemite royal family. His mission
remains, however, to speak up for the little people of Jordanian society,
in particular the Bedouin, whose voice and concerns can rarely be heard
above what he describes in another poem (al-Jarid the Newspapers)
as the sycophantic blather of the government-controlled media.
Ghassn commented thus on the poem presented below:
A famous Jordanian song, popular in the 1960s, was y marh ab, y hal,
minn ar-rakb, minn? The song praised the heroes of our nation and
encouraged manliness (rugla), gallantry (ahma), self-respect (karma)
and freedom (h urriyya). But peoples ideas have changed, as have their
dealings with each other. People have become so preoccupied by earning
548 clive holes
a crust and making ends meet that they have no time to think about any-
thing else, and if they do they get distracted and lost.
3. Rashd al-Kln in Abdh Mss musical setting
The words of the original song (Ghawnmah 2002, 1501) are by Rashd
al-Kln. A famous Jordanian gypsy singer and rebec player of the time,
Abdh Ms (19271977), made them famous by putting them to
music. Ms was well known for his popular nationalistic songs prais-
ing the army, national heroes, and the Hashemite dynasty, as well as
sentimental ditties for religious occasions. Ghassn takes many of the
original lines of Abdh Mss song and, while preserving the original
meter and rhyme scheme, alters the words to provide a series of poetic
snapshots of the state of Jordan in the early 1980s. The effect of insert-
ing such material into the structure of the song, whilst at the same time
preserving much of the original bombastic phraseology, is pungently
satirical and bathetic. The impact of Ghassns poem on a local audience
depends precisely on the fact that Mss song from an earlier, more
confident period of Jordanian history was so well known to ordinary
Jordanians from its frequent airing in the local media.
Rashd al-Klns original words are as follows: 5
1 y marh ab, y hal, mnn Hello, I say, and welcome! Now

ar-rakb, min wn? 6 where from, these riding men?
agbal aln d-dih y zna The mornings come upon us,
gblah see how beautifully, again!
2 h inn dr7 al-id, tallba li Against the foe8 were ruthless;
d-dn our debts we neer disclaim,
5
Each verse of the poem is divided into two hemistiches, all the opening hemis-
tiches as a group, and all the closing hemistiches as a group, being metrically identical,
and each group having a different rhyme, -n (or -n) for the opening hemistiches and
-lah for the closing ones. This is a common arrangement. Unfortunately it is almost
impossible to imitate in an English translation, so I have rendered the poem into English
rhyming couplets, with fourteen syllables to each hemistich, and tried to be as faithful as
possible to the meaning and tone of the original.
6
A standard Bedouin greeting.
7
dir pl dr ruthless, remorseless.
8
Lit aggression < CLA ad.
wa l-gr m yigbalah illa For only he accepts abuse whose

r-rad xlah menfolk have no shame.
3 xd il-marik lin min ymn The fray we enter swaggering,
salfn9 our heads we hold up high,
wa l-hm di llan wa r-rh The Hashemites provide our
fadw lah shade10for them we live or
die!
4 wa t -t r krin11 lina, y t rn Revenge is second nature; for
bi t rn each wrong we take back two,
y sbin h aggan l budd m From any who usurps our
nnlah rights, we wrest them back
anew!
5 nizh af ala ll ba wa xn al- We march against oppressors
ahd wa d-dn who break their word and creed,
wa nds ala ll ta wi nigzh We stamp upon the tyrants
b af lah head, and punish his misdeeds!
6 h ayyhum nim l-watan, So salute our homeland brave
h ayyhum gund ih sn hearts! So salute them, Husseins
pride!
rab il-kaff l-h umur wa Theyre the ones who wear
l-ugul mayylah red head-cloths, head-ropes
slanting to the side!
7 y mgana12 bi n-nay13 kaff O wife who grieves in widows
dum il-n weeds, weep not, nor solace
seek,
kaff dum il-as a l-xadd Weep no more tears of sorrow,
sayylah no tears trickling down your
cheek.
8 g k n-nim, laf,14 wi Our gallant lads have come to
tbair y zn you, theyre here, be of good
cheer!
9
salf pl salfn brave, swaggering. Cf. Musil 1928, 561 salf a strong gust of wind, Clas-
sical Arabic salif vainglorious, boastful.
10
I.e. protect us.
11
kr pl. kiyr custom, habit. Cf. Bailey 1991, 451.
12
mgana woman wearing a guna = womans black head covering reaching down to
her behind.
13
nay sadness, dejection. < Classical Arabic n-w-. Cf. na bi l-h iml to groan under
a burden.
14
lafa (yilfi) come, arrive.
550 clive holes
igbn fg il-higin, w usd Like eagles riding camels, like

xayylah fierce lions, those cavaliers!
9 add l-bandig zurug They wield machine guns,
rah saffn15 double-clipped, the metal
glinting blue,
u mudarratin uhub nrn And armoured cars, their colour
gattlah grey, death-dealing bullets
spew!
10 madfi mwallafa, tirm ala Their guns co-ordinated, and
l-gln16 they let fly to each side,
saly17 l-qanbil raad u burg A thunderous volley of shellfire,
alah it lights up the countryside.
11 wa nsr gu ww s-sam And vultures19 hover in the
mitlaf a bi l-n18 sky, wrapped round by cotton
clouds,
tixw 20 ala l-mitad lil l-mt They plunge down on the
ayylah enemy, and bring him his death
shroud!
12 tixw ala l-mitad mixlbh They plunge down on the
bi h addn enemy, their claws as sharp as
nails,
m dg tam as-salma min And none who dares to
nisat glah challenge them has lived to tell
the tale.21
13 aswad sabh al-id ymin Black morn for the aggressor,
alhum n the day him ill befell!
u ubr yim da f il-galb an By dust his weak heart blinded,
h lah his fate he could not tell!
14 min kull dabbbtin ummt There comes from every battle
ganzrn tank, twin caterpillar-tracked,
tuhdur hadr il-bah ar, a A roar deep as the oceans,
l-gm22 sayylah gainst the foe they launch
attacks!
15
saff cartridge magazine or clip of a firearm.
16
g l pl. gln side, direction.
17
saly blazing volley (of shots, cannon fire).
18
Sc. m clouds. The change to n is a poetic licence to maintain the rhyme.
19
I.e. fighter pilots.
20
xawa to plunge, dive.
21
Lit: None tasted safety that went in their direction.
22
Normal Bedouin term for enemy.
15 u fhd bn il-h ufar tuhgum And darting tween the battle

ala rigln lines23 two-legged cheetahs
growl,
gannsa li l-id w inmr And hunt down the aggressor,
gawwlah just like panthers on the prowl.
16 h ayyhum nim l-watan, So salute our homeland brave
gu nd ih sn hearts! So salute them, Husseins
pride!
rabi il-kaff l-h umur wi Theyre the ones who wear
l-ugul mayylah red head-cloths, head-ropes
slanting to the side!
4. Text comparison
In his version, Ghassn takes the sentimental nationalism of Mss

refrain:
h ayyhum nim l-watan, So salute our homeland brave

h ayyhum gu nd ih sn hearts! So salute them, Husseins
rab il-kaff l-h umur wa l-ugul pride!
mayylah Theyre the ones who wear red
head-cloths, head-ropes slanting
to the side!
(line 6 of the original, repeated in line 16)
and transforms it into a variable refrain of his own in which he in turn

ridicules Jordans political leaders:
h ayyhum iyx il-arab il So salute our Arab leaders, on their

takstuh24 bi antnn cars two aerials fixed!
rab it-taks l-h umur wa Theyre the ones who drive red
d-daym bi iglah limos, head-ropes full of despots
tricks!
(line 5 in Ghassns version).
23
Lit: ditches, i.e. defensive trenches.
24
taks (in construct phrase and when suffixed takst-) pl. taks car, limousine (as
well as taxi).
552 clive holes
its well-paid but incompetent officials:
u h ayy wizrat il-il il alh agatn And salute the High Price Ministry
bi d-dn! thats saddled us with debts!
rab it-tawbig h agar u amwl Theyre the ones in stone-built
sayylah condos, with their liquid cash
assets!
(line 13 in Ghassns version)
returning more wistfully at the end to Mss subject, the army:
u ym id-dgt wn rh in-nim Where did they go, now times are
wn? hard, those brave heart lads with
pride?
rabi il-kaff l-h umur wi l-ugul The ones who wear red head-cloths,
mayylah . . .? head-ropes slanting to the side . . .?
(line 19 in Ghassns version)
The confident vainglory of Abdh Mss opening lines, in which he

brags about the pride, dignity and valour of King Husseins Bedouin
troops, becomes in Ghassns poem an abject confessional put in the
mouths of ordinary Jordanians who have learnt to accept incompetence
and oppression and to keep their mouths shut and look out for them-
selves. Compare the original:
y marh ab, y hal, mnn Hello, I say, and welcome! Now

il-rakb, min wn? where from, these riding men?
agbal aln d-dih y znat The mornings come upon us, see
gblah! how beautifully, again!
h inn dr al-id, tallba li d-dn Against the foe were ruthless; our
debts we neer disclaim,
wa l-gr m yigbalah illa r-rad For only he accepts abuse whose
xlah menfolk have no shame.
xd il-marik lin min ymn The fray we enter swaggering, our
salfn heads we hold up high,
wa l-hm dillan wa r-rh The Hashemites provide our shade
fadw lah for them we live or die!
wa t- t r krin lina, y t rn bi Revenge is second nature; for each

t rn wrong we take back two,
y sbin h aggan l budd m From any who usurps our rights,
nnlah we wrest them back anew!
(lines 14)
with Ghassns:
y marh ab, y hal, mnn Hello, I say, and welcome! But now
il-bal min wn? wheres this scourge come from?
agbal aln l-il y maba Price rises came upon us, and
igblah! 25 theyve hit us like a bomb!
h inn gh id-duwal w il We act like world-class donkeys,
nartad bi l-bn26 with our cruel fate content,
wa l-kull rd l-gr h att Each one of us accepts abuse, even
-aham xlah27 those of high descent.
wa l-amr tanfdah lin u min We do as we are ordered to; its
drn xann not our place to ask,
wa l-gir sr ar-rabb wa r-rh Money rules our lives; our souls are
fadw lah mortgaged to that task.
wa s-samt krin lin wa Keeping quiets second nature; we
a l-h agg m nin dont stick up for peoples rights,
wa l-h agg min il-mursil28 Our rights we beg from minions,
wallah m nnlah then give back without a fight.
(lines 14)
The symbolic widow of lines 78 of the original, weeping for a husband

killed in battle, and comforted by the arrival of the army of the nim
(gallant lads) of the army:
25
< y m ab igblah lit: how awful its coming!
26
= w illi nartadi bi l-bn those who are content with adversity.
27
Ironically echoing the sentiment of the popular Bedouin saying dawwir li wil-
dik xl find a maternal uncle for your son, i.e. find a brave man and marry his sister
because the bravery will run in the family. Nowadays, the poet is saying, the once noble,
independent Bedouin are ordered about by moral pygmies, and money has become the
yardstick of a persons worth.
28
mursil servant, messenger-boy.
554 clive holes
g k n-nim, laf, Our gallant lads have come to you,

wi tbair y zn theyre here, be of good cheer!
igbn fg il-higin, w usd Like eagles riding camels, like fierce
xayylah lions, those cavaliers!
is told in Ghassns version (line 7) to cry even more, as there is no lon-

ger any comforting presence to reassure her:
rh n-nim u madaw, Our gallant lads have gone for

l tafrah y zn good, theres nothing left to cheer,
b syf il-h arb u m dall They sold their swords off long ago;
xayylah their cavalry disappeared.
Instead of the exultant heroism of:
madfi mwallafa, tirm Their guns are at the ready, and

ala l-gln they let fly to each side,
saly l-qanbil raad u burg A thunderous volley of shellfire,
alah it lights up the countryside.
(line 10 of the original)
the widow, and all other ordinary Jordanians can expect no protection at
all; on the contrary, they have become the targets of (economic) attack:
awmrah mh add a rah Hes got his orders ready, and he lets
yirm ala l-gln fly to each side,
saly l-awmir raad u A thunderous volley of memos,
alt gattlah death-dealing stuff inside!
(line 9 of Ghassns version)
the orders in this case being those of the Prime-Minister Mudar Badrn
to raise the price of the basic necessities of life beyond the point where
they can afford them.
The original song concludes (lines 815) with a stirring picture of the
derring-do of the Jordanian army and air force; the corresponding lines
in Ghassns poem depict a sleazy world of fawning senators selling off
the national patrimony, accepting bribes, lining their own pockets, and
hanging around night-clubs and brothels (Ghassns lines 1117):
il-wisk lah wi l-z lah Hes got his whisky, young roast
wa li h sbah ha-l-girn lamb, hes got our money too,
wa -aab b h amdullah The people feed on wheat-chaff
m yiba nxlah29 and just thank the Lord they do.
fawwad ard l-arab He gave away our tribal lands to
li s-sr luh ahrn folk just two months here,
li l m yinirif ammah To people with no lineage, no
u l yinirif xlah pedigree thats clear.
u h ayy wizrat il-il il And salute the High Price Ministry
alh agatn bi d-dn! thats saddled us with debts!
rab it-tawbig h agar Theyre the ones in stone-built
u amwl sayylah condos, with their liquid cash assets!
din yxk30 u sin31 Were lost, my brother, wander-
wn il-harba wn? ing . . . oh where is our escape?
u h litn b hal-balad By God, the nations state is dire:
wallah mi h lah no order here, no shape.
wisiln dr il-h add min To blame? Our sheikhs an
xn wa l-n32 senators; the mess were in is foul:
ill ala abwb id-dant33 They spend their time in sleazy
inmr gawwlah dives, like panthers on the prowl.
l yiurrak kubr i-anab Dont be fooled by big moustaches,
wa l-fard maa mitn guns with double cartridge clips,
bi d-dabdaba34 yibd u bi l-madh What they write begins with
yinh maglah weasel words and ends with
flattering quips.
iyh tt lah gu ww z-zarf Our rulers slip them money, in
h ukkmn girn brown envelopes its brought,
u min agal hal-girn yanh ar Theyd kill their kids to get one, slit
lak iylah their throats without a thought.
Ghassn then rubs it in by concluding his message with a variation of

the opening lines:
29
nxla poor quality wheat left after sieving.
30
A so-called bi-polar address form, also common in Eastern Arabian dialects. See
Yassin 1977.
31
s to wander, go aimlessly from one place to another. sin u din were wandering
lost. Cf Classical Arabic sa to scatter (people, things).
32
I.e. a member of the maglis il-ayn the Senate, all appointed by the King.
33
I.e. brothels and drinking dens.
34
dabda b to talk meaninglessly or ambiguously, utter weasel words.
556 clive holes
y marh ab y hal! imnn SoHello, I say, and welcome! But

il-il min wn? now wheres this price rise from?
agbal aln l-bal, y maba A scourge is come upon us, and its
igblah! hit us like a bomb!
(line 18)
and a prediction of even worse to come from an incompetent government:
win xilsat il-arkt, yallah But when the battles over, and the
is-salmah situations calm,
itxalln il-wh id min id-dill You lot will mess things up again;
h ltah h lah youll do us yet more harm!
(line 20)
Ghassns poem is four lines longer than the original, and the formal
correspondences of the original lines to those of Ghassns version (i.e.
where Ghassns lines mimic the diction of the originals) are as follows:
Rashd al-Klns lines correspond to Ghassns lines

14 14
5
68 57
9
8
10 9
1114
1014
15 15
1617
1 18 (reprise of 1)
6, 16 19
20
This is Ghassns complete (1982) text:

1 y marh ab, y hal, mnn il- Hello, I say, and welcome! But
bal min wn? now wheres this scourge come
from?
agbal aln l-il y maba Price rises came upon us, and
igblah! theyve hit us like a bomb!
2 h inn gh id-duwal w il We act like world-class donkeys,
nartad bi l-bn with our cruel fate content,
wa l-kull rd l-gr h att Each one of us accepts abuse,
-aham xlah even those of high descent.
3 wa l-amr tanfdah lin u min We do as we are ordered to; its
drn xann not our place to ask,
wa l-gir sr ar-rabb wa r-rh Money rules our lives; our souls
fadw lah are mortgaged to that task.
4 wa s-samt krin lin wa a Keeping quiets second nature; we
l-h agg m nin dont stick up for peoples rights,
wa l-h agg min il-mursil wallahOur rights we beg from minions,
m nnlah then give back without a fight.
5 h ayyhum iyx il-arab il So salute our Arab leaders, on
takstuh bi antnn their cars two aerials fixed!
rab it-taks l-h umur wa Theyre the ones who drive red
d-daym bi iglah limos, head-ropes full of despots
tricks!
6 y mgana bi n-nay zd dum O wife who grieves in widows
il-n weeds, weep on, the futures
bleak,
zd dum il-as a l-xadd Weep yet more tears of sorrow,
sayylah let them trickle down your cheek.
7 rh n-nim u madaw, l Our gallant lads have gone for
tafrah y zn good, theres nothing left to
cheer,
b syf il-h arb u m dall They sold their swords off long
xayylah ago; their horsemen disappeared.
8 kull nadil h kim bihum h ukm Now any numb-skull scoundrel
il-and i-n treats them like his abject tools,
u xall wgh il-balad bi s-sg Makes porters out of highborn
attlah men who once this country
ruled.35
35
The reference is to Mudar Badrn, the Jordanian Prime-Minister of the time. There
is also a whiff here of a sense of lse-majest: Jordanians of pure Bedouin descent regard
the Badrn family, which has its roots in urban Syria, with disdain.
558 clive holes
9 awmrah mh add a rah yirm Hes got his orders ready, and he
ala l-gln lets fly to each side,
saly l-awmir raad u alt A thunderous volley of memos,
gattlah death-dealing stuff inside!
10 saww tnakit il-kz36 bi h dd A jerry-can of kerosene put up to
dnrn two JD,37
wa a h sb g in-ns titkt ar To make his stacks of cash pile
amwlah up, the people go hungry.
11 il-wisk lah wi l-z lah wa li Hes got his whisky, young roast
h sbah hal-girn lamb, hes got our money too,
wa -aab b h amdullah m The people feed on wheat-chaff
yiba nxlah and just thank the Lord they do.
12 fawwad ard l-arab li s-sr luhHe gave away our tribal lands to
ahrn folk just two months here,
li l m yinirif ammah u l To people with no lineage, no
yinirif xlah pedigree thats clear.
13 u h ayy wizrat il-il il And salute the High Price
alh agatn bi d-dn! Ministry thats saddled us with
debts!
rab it-tawbig h agar u amwl Theyre the ones in stone-built
sayylah condos, with their liquid cash
assets!
14 di n yxk u sin wn il- Were lost, my brother,
harba wn? wandering . . . oh where is our
escape?
u h litn b hal-balad wallah By God, the nations state is dire:
mi h lah no order here, no shape.
15 wisiln dr il-h add min xn To blame? Our sheikhs and
wa l-n senators; the mess were in is foul:
ill ala abwb id-dant inmr They spend their time in sleazy
gawwlah dives, like panthers on the prowl.
16 l yiurrak kubr i-anab wa Dont be fooled by big
l-fard maa mitn moustaches, guns with double
cartridge clips,
bi d-dabdaba yibd u bi l-madh What they write begins with
yinh maglah weasel words and ends with
flattering quips.
36
A twenty-litre jerry can of kerosene, used for cooking on primus stoves.
37
I.e. (in 1982) very expensive.
17 iyh tt lah gu ww z-zarf Our rulers slip them money, in

h ukkmn girn brown envelopes its brought,
u min agal hal-girn yanh ar Theyd kill their kids to get
lak iylah one, slit their throats without a
thought.
18 y marh ab y hal! imnn SoHello, I say, and welcome! But
il-il min wn? now wheres this price rise from?
agbal aln l-bal, y maba A scourge is come upon us, and
igblah! its hit us like a bomb!
19 u ym id-dgt wn rh Where did they go, now times
in-nim wn? are hard, those brave heart lads
with pride?
rabi il-kaff l-h umur wi l-ugul The ones who wear red head-
mayylah, cloths, head-ropes slanting to the
side . . . ?
20 win xilsat il-arkt, yallah is- But when the battles over, and
salmah the situations calm,
itxalln il-wh id min id-dill You lot will mess things up again;
h ltah h lah youll do us yet more harm!
Thus where the original song was a jingoistic, bombastic romp, in

Ghassns hands it is transformed into a mordant diatribe against the
Badrns governments economic mismanagement, its cavalier treatment
of ordinary Jordanians, and the corruption and cupidity of its officials.
It is precisely the parodying of the diction of the original that gives the
poem its punch as a piece of political satire.
The relationship of language level, poetic form and subject matter is
worth pondering. Historically, Bedouin popular poetry was concerned
solely with the local tribal milieu: with vaunting the merits of ones tribe
over those of a neighbour, with praising heroic warriors and generous
shaikhs, with ridiculing enemies, against a backdrop of descriptions
of desert flora and fauna and the movement of the tribes in constant
search for water and good pasture. In this, as has often been observed,
Bedouin popular poetry is the direct descendant of the old pre-Islamic
poetry, and is similarly restrictive in matters of meter and rhyme. For
all these topoi, the only possible linguistic vehicle was the vernacular,
both because it was the only form of Arabic the poets and their audi-
ence knew, but also because of the organic nexus between locale, sub-
ject matter and language. But now that the traditional subject matter for
the exercise of poetic talent has all but disappeared, this ancient poetic
560 clive holes
tradition has shown its protean nature: it continues to be used, even by

Bedouin poets who are literate, to serve many of the same functions,
though now on a much broader social and political canvas. If one wants
to emoteto praise, to damn, to ridicule, to gloat, to cajole, to bewail, to
lovehow much more effective is the vernacular because of its infinitely
greater allusiveness, colour, concreteness and idiomaticity for the ordi-
nary person, compared with the sanitised and institutionalised bland-
ness of the modern standard language.
As I write this article, I have to hand the latest in a long-running
series of poems written since 2003 by a south Jordanian Bedouin poet,
Muh ammad al-Fanti l al-H ajy on the American-British invasion of
Iraq, some of which have been published in local newspapers and maga-
zines. The poems recycle standard Bedouin poetic tropes, such as the
journey (similar to the Classical rah l), whereby a rider, after first having
had the noble qualities of his camel38 described by the poet, is sent with
a message (the poem) to kinsmen encamped far away; or the ascent
whereby the poet climbs to the top of a nearby hill (rigm), there to wres-
tle with the cauldron of his emotions (higs, hgs, hawgs) and finally
be seized by poetic inspiration to write his poem39 . . . except that in these
poems, it is leading international politicians who are thus depicted! For
example, the poem Fadas wa Fads40 opens with George Bush making
an ascent in his home state of Texas to ponder on his woes during the
2004 Iraq insurgency:41
araft rigmin f marb takss Ah climbed atop a Texas peak, out

west in our proud nation,
rigmin tawl min al-baar A peak so hah no folks live there, to
r mns fahnd some isolation.
38
Nowadays, it is quite common for the poet, without a trace of irony, to describe
the qualities of a modern steeda Toyota Landcruiser, for instancein terms similar
to the ways a fine riding camel would once have been described: its strong chassis, its
smooth gear-change, its fine performance, etc. See, e.g., Kurpershoek 1999:110114 for
examples from Najd. We recorded similar descriptions of cars in Bedouin poetry from
Sinai.
39
Kurpershoek 1999:3134.
40
These pet names are often given by the Jordanian Bedouin to dogs (rather like
Rex and Rover in English). In the poem they are insultingly applied to the pet rulers
of unnamed Gulf States who, in the poets view, treacherously provided bases for the
Americans at the time of their invasion of Iraq.
41
And speaking in a Texas accent, which I have tried to imitate in my translation!
u darabt min hamm xms Ah brooded deep upon mah woes,

bi l-asds Ah felt at mah wits end, when
u h asst f rs kam targ ngs Ah heard a bell toll in mah head
what did that bell portend, then?
u gadt wagt bn higsin u hgs My worries crowded in on me, my
state it was concernin,
wi l-galb tgad bih kam nr Mah heart aflame as if it were a
fns lanterns oil-wick burnin!
Another character who has had poems put in his mouth by this poet
is Saddam Hussein, in prison in Iraq, but stillpoeticallyshaking a
defiant fist a the Americans. And to George Galloway, founder and sole
representative of the UKs Respect political party, he has dedicated a
typical piece of Bedouin madh , praising Galloways opposition to Tony
Blair and what the poet sees as his dogged, unpopular but heroic sup-
port for the Iraqi people.
Popular poetic commentary on contemporary events such as has
been exemplified in this paper is nothing new: its just that the focus
has broadened. Before the birth of the nation state in the Arab World,
and even in the period since, Bedouin tribal poetry was one of the main
means by which partisan communal sentiments could be articulated. A
good example of this was a dispute between the H uwt t, a Jordanian
tribe, and the Ban Atiyya of Saudi Arabia, which rumbled on into the
late 1980s in the form of an extended poetic debate of claim and counter-
claim. The dispute centred on Tubayq, an area in the far southeast of
Jordan which the Jordanian government ceded to the Saudis in 1964
in return for a stretch of Red Sea coastline which would enable Jordan
to improve its naval facilities away from the prying eyes of the Israelis.
Initially, this exchange of land and redrawing of the border caused no
problems, and the H uwt t on the Jordanian side continued to drive
their animals to seasonal pastures on the Saudi side, exactly as they had
always done. They incidentally benefited from the fact that many goods
were cheaper in Saudi Arabia, and border controls were lax or non-exis-
tent. A number of events, however, changed all this. During the Jordan-
ian civil war of 1970, Russian machine guns became easily available, and
enterprising H uwt t began a lucrative smuggling operation into Saudi
Arabia. Drugs were also smuggled in. This led to a clampdown by the
Saudis, and the blacklisting of many H uwt t. Then, after the shock-
ing siege of the grand mosque in Mecca by Islamic fundamentalists in
1979, the Saudis put even stricter border controls in place. One measure
562 clive holes
was the digging of a ditch, 3m wide by 3m deep along the border with
Jordan in Tubayq. No camel or car could cross it except at designated
control points situated far apart. H uwt t who had been used to driving
their animals 10 miles to pasture were now faced with driving them 100
miles to find the nearest border post and 100 miles back again. Customs
controls became much stricter. A system of registration documents
was introduced for the family members who moved with the migrating
flocks. There were cases of H uwt t mothers who had given birth while
in Saudi Arabia being detained there because, when they tried to return
to Jordan, the number of family members did not correspond with
the number on the registration document. There were intimate body
searches of all females at the border checkpoints. This was bad enough
for the socially conservative Bedouin, but it also closed off the last
avenue for smuggling. On the other hand, the Jordanian borders
remained open to Saudis without let or hindrance, a fact that caused
huge resentment. This unequal treatment, coupled with what was per-
ceived by the Jordanian Bedouin as the central governments lack of
economic help to the people of the south, was one of the factors that
precipitated rioting in Maan and Al-Jafr in 1989.
A well-known H uwt poet, Barrk Dish Ab Tyih, wrote several
emotional poems about the dispute that the Ban Ati yya, on the Saudi
side of the border, did not like. Ban Ati yya poets replied with poems
which were recorded and passed back to the H uwt t. The initial skirm-
ish might have ended there. However, it was followed by a long and
insulting poem from a young H uwt poet, Nad Tmn Ab Tyih. The
poetic tit-for-tat then escalated, with ten poems by Ban Atiyya poets in
reply, releasing much pent-up personal vituperation. This caused an out-
raged reaction among the H uwt t, who had always regarded the Ban
Atiyya as inferior to themselves. Seventy years ago, the result would
undoubtedly have been a tribal war; on this occasion, the dispute was
finally defused in 1990 by mediation, and the signing of an agreement
that neither side would write any more poems on the subject of Tubayq.
The key point here is that poetry was, and is, regarded by the Bedouin
as a suitable vehicle for airing important issues of the moment, rather
than a letter to the provincial governor, still less lobbying a remote and
seemingly uncaring central government.
5. References
Abdel-Malek, Kamal. 1990. A Study of the Vernacular Poetry of Ah mad Fud Nigm.
Leiden: E.J. Brill.
Abu Athera, Said Salman and Holes, Clive. Forthcoming 2007. Y Kundlzza Rys!
Politics and Popular Poetry in the Contemporary Arab World. Leiden: E.J. Brill.
Abu-Lughod, Lila. 1986. Veiled Sentiments: Honor and Poetry in a Bedouin Society.
Berkeley: University of California Press.
Bailey, Clinton. 1991. Bedouin Poetry from Sinai and the Negev. Oxford: OUP.
Booth, Marilyn. 1990. Bayram al-Tunisis Egypt: Social Criticism and Narrative Strate-
gies. Exeter: Ithaca.
Ghawnmah, M. Abdh Ms: Ridan wa-Mubdian, Amman, Dr al-Kind, 2002.
Ibn Khaldn, tr. Rosenthal, Franz. 1958. The Muqaddimah. An Introduction to History.
London: Routledge and Kegan Paul.
Ingham, Bruce. 1986. Bedouin of Northern Arabia: Traditions of the l-D hafr. London:
Kegan Paul International.
Kurpershoek, P. Marcel. 1994. Oral Poetry and Narratives from Central Arabia I: The
Poetry of al-Dindn. Leiden: E.J. Brill.
. 1995. Oral Poetry and Narratives from Central Arabia II: The Story of a Desert
Knight. The Legend of lwh and other Utayba heroes. Leiden: E.J. Brill.
. 1999. Oral Poetry and Narratives from Central Arabia III: Bedouin Poets of the
Dawsir Tribe. Between Nomadism and Settlement in Southern Najd. Leiden: E.J.
Brill.
. 2002. Oral Poetry and Narratives from Central Arabia IV : A Saudi Tribal History.
Leiden: E.J. Brill.
. 2005. Oral Poetry and Narratives from Central Arabia V : Voices from the Desert.
Glossary, Indices, and List of Recordings. Leiden: E.J. Brill.
Landberg, Carlo le Comte de. 1901. tudes sur les dialectes de larabie mridionale.
Leiden: E.J. Brill.
Meissner, Bruno. 1903. Neuarabische Gedichte aus dem Iraq. Mitteilungen des Semi-
nars fr orientalische Sprachen zu Berlin VI. 57125.
Musil, Alois. 1928. The Manners and Customs of the Rwala Bedouin. New York: Ame-
rican Geographical Society.
Socin, Albert. 19001. Diwan aus Centralarabien. Leipzig: Teubner.
Sowayan, Saad Abdallah. 1985. Nabati Poetry: The Oral Poetry of Arabia. Berkeley: Uni-
versity of California.
Wallin, Georg August. 1851. Probe aus einer Anthologie neuarabischer Gesnge in der
Wste gesammelt. Zeitschrift der Deutschen Morgenlndischen Gesellschaft 5. 123.
. 1852. Probe aus einer Anthologie neuarabischer Gesnge in der Wste gesam-
melt. Zeitschrift der Deutschen Morgenlndischen Gesellschaft 6. 190218, 369378.
Yassin, M.A. 1977. Bi-polar terms in Kuwaiti Arabic. Bulletin of the School of Oriental
and African Studies 40. 297330.
NOTES ON THE DIALECTS OF THE LGT AND H AMD AH
OF SOUTHERN SINAI
Rudolf de Jong
ACLC, University of Amsterdam
1. Introduction
The subject of this contribution1 is the dialect of the Lgt (or Ulayqt),
a bedouin tribe who live in the western central part of southern Sinai
(see map below).2 In addition, notes on the dialect of the H amda h have
been included. Some texts recorded among the Lgtwith additional
information in footnotesmay serve for further illustration.
The H amda h are only few, and are often regarded as a clan (or fam-
ily) of the Lgt, although various sources claim that they were present
in Sinai before the Lgt.3 We shall see that there are some notable dif-
ferences between these two varieties of speech.
1
With great pleasure I dedicate this contribution to Kees Versteegh. My dedication
is with deep respect for his stature in our field of Arabic Studies, with gratitude for the
inspiring thoughts he has shared with our community and with fond memories of the
(too few) occasions I had the honor to work with him.
2
The material used for this article was collected in the framework of my own research
into the bedouin dialects of southern Sinai. This project is funded and supported by The
Netherlands Organization for Scientific Research (abbreviated in Dutch as NWO) and
the Amsterdam Center for Linguistics and Communication (ACLC) at the University
of Amsterdam. I am sincerely grateful to both organizations, as I am thankful also to
Manfred Woidich for his advice and support during my researches. I am no less grateful
to my desert guideand now dear friendId Abuw Silm (al-Atra at-Turbniy), who
has been my ally in this project and without whose contributions this research would
not have been possible.
3
Murray (1935, 291), for instance, writes that the H amdah [. . .] are now only about
twelve tents strong, and live mostly near Serabit el-Khadim and Bir Nasb, where they are
reckoned as part of the Aleiqat. For the location of Sarbt al-Xdim, see Google Earth
at (appr.) 29 00 05 N33 28 01E.
At -Tayyib (1997, 480481) mentions that the H amdah joined the Lgt under the
Lgiy x at that time (A.D. 1914) Mdaxxal Slmn of the Lgiy clan named Awld
Silmiy. Von Oppenheim (1943, 162, 164) writes that the H amdah were angegliedert to
the Olkt (in his transcription).
On the webpage http://members.nova.org/~lroeder/alegat.htm (authored by Larry
Roeder), under History of Hamada, it is reported that the H amdah [. . .] probably
[joined the Alegat] shortly after the capture of Nuweiba. We do know that from then
until the 1880s this was a sub-tribe (or section) of the Alegat. Then in 1880 internal
566 rudolf de jong
Reported to have arrived in Sinai at some time in the fourteenth

century,4 the Lgt today number around 1,000,5 and are found predom-
inantly in an area named ar-Ramlah, near where Jabal Sarbt al-Xdim6
stands, situated just south-west of the Th escarpment, and north of
Wdiy Frn and in and to the east of the town of Abuw Znmah.7 Also
farther north, to the south of Rs Sadr and in Wdiy arandal (not far
north of H ammm Farawn) families of the Lgt are said to reside, as
well as farther south, near the town of at-Tr.8
Approximate distribution of bedouin tribes in southern Sinai 9
politics forced a split when the Hamada requested permission to live under the protec-
tion of the Muzeina. By 1935, they had become integrated enough in Muzeina affairs to
be considered an integral part of that tribe.
4
See Bailey 1985, 48. For more information on subdivisions of this tribe, their xs,
history, territories, etc., see at -Tayyib 1997, Part 2, 475489.
5
Such numbers are of course approximations.
6
Sarbt al-Xdim, some 40 kilometres east of Abuw Znmah (on maps usually
spelled as Abu Zinima), is famous as the site of turquoise mines operated since early
pharaonic times, and the temple of Hathor, which is the only pharaonic temple built
outside Egypt proper. See also the webpage about the Sinai at http://www.arcl.ed.ac.uk/
arcx/remot esense/sinai/.
7
Also in Wdiy Isla. The Lgt are reported to form an alliance in Sinai with the
and Mznah, see At -Tayyib (1993, 705) and Bailey (1991, 5). At- Tayyib (1993,
H amdah
706) reports that their territory stretches from ar-Ramlah to Wdiy arandal. One of my
own Lgiy informants mentioned these areas too, but said their drah stretches up until
the area named ar-Rynih, north of Rs Sadr. For a map locating their (there transcribed
as Algt) drah, see Bailey (1991, 4).
8
Wdiy s Sahaw, mntigit Mbajjmah, Br anNasb and Wdiy Lihyn were mentioned
to me by a Lgiy informant as parts of the H amdiy drah.
9
Abbreviations: Ah = Ah aywt, Tr = Tarbn, H w = H wtt, Db = Dbr (see remark
below), Ty = Tayhah, Lg = Lgt, Bd = Badrah, Jr = Jarjrah (see remark below),
dialects of the lgt and h amd ah 567
1.1 General
In the course of the ongoing research into the dialects of southern Sinai,
a picture has been emerging in which dialects of southern Sinai appear
to constituteto an as yet undefined degreea homogeneous group.
This group is typologically related to the dialects of group II in the north
of Sinai: that of the Agylah, andeven more clearly sothat of the
Samnah.10
As a group, the dialects in the south stand separate from the dia-
lects of group I, or the Negev-type of dialect (to which the dialects of
the Ah aywt and Tarbn belong, and also those of the Tayhah and
Jarjrah).11 The southern group also shows important differences with
the dialect of the Mznah, who live in the eastern part of southern
Sinai and its southern tip.12 We shall see that of the two dialects treated
here, H amdi y appears to have the most in common with this southern
Sinaitic group, while Lgiy occupies a position typologically somewhat
nearer to group I.13
2. Phonology
2.1 Consonants
As bedouin dialects spoken in southern Sinai, the phoneme inventories
of Lgiy and H amdi y do not hold many surprises. As far as consonan-
tal phonemes are concerned, all three interdentals /t/ , /d/ and /d/ (the
latter as a reflex for both *d and *d) are present. A voiced and unaffri-
cated /g/ reflects Classical Arabic *q and a voiced affricate /j/ reflects *j
H m = H amda h, Sw = Sawlh ah, Gr = Garrah, AS = Awld Sad, Mz = Mznah, Jb =

Jbliyyah, BW = Baniy Wsil.
I have met Jarjrah in the area (named Malbad, see Google Earth 29 29 29.50 N.,
33 05 36.50 E.) as marked on this map. Murray (1935, 263) also refers to the Jara-
jira as a family of the Aleiqat. A tribe also indicated on this map are the Dbr, whose
presence was mentioned to me by sources in the field. They are said to be related to the
Masd, who live in and around the village of Jilbnah in the northwest of Sinai.
10
As described in De Jong (2000, chapter 2).
11
Being based on research in progress, this conclusion is tentative.
12
For a preliminary report on research into the dialect of the Mznah, see De Jong,
The Dialect of the Mznih of Southern Sinai (to appear in Estudios de Dialectologa
Norteafricana y Andalus (EDNA) 8 in honor of Peter Behnstedt).
13
As described in Blanc (1970) and De Jong (2000). Murray (1935, 263 ff.), writes
that the Lgt [. . .] are said to talk like the Terabin.
568 rudolf de jong
(while the fricative allophone [] appears to be as current as in e.g. the

dialect of the Mznah in the southeast of Sinai).
There are also exceptions to the interdental reflexes of Classical Ara-
bic interdentals: e.g. (z for *d) bizr seed and aza ~ iza if (compare
Classical Arabic *id). In the loan masalan for instance, s reflects *t.
Although z as a phoneme is rare (and marginal as such) in Sinai be-
douin dialectsthe current reflex for both *d and *d being the interden-
tal emphatic d it is regular in lexemes (loaned from Modern Standard
Arabic or Cairene Arabic) like zabbat, yzabbit arrange; do properly,
bz, ybz go bad (of food) (from the Turkish root boz-mak).
Also, like in many southern Sinaitic dialects (though not all), a mini-
mal pair such as btk your (m. sg) housebtk your (f. sg.) house yields
the two (unaffricated) phonemes (heavily velarized) /k/ and (other-
wise neutral) /k/.14
Another feature of Lgiy and H amdi y is its regular progressive
assimilation of initial h of suffixes to preceding voiceless consonants,
e.g. fattte # (< fatt + ha)15 he shredded it (f. sg.) (said of bread, when
one prepares fattah), maslah tta (< maslah t + ha) her affairs, arssa
(< ars + ha) her groom, dabah h a he slaughtered her, msikka hav-
ing taken hold of her and a very current (in both varieties, as well as
in many other dialects of Sinai) reciprocal assimilation of + h in e.g.
mih h a (< mi + ha) with her.
2.2 Vowels
As for vowels, phonemic opposition of short vowels /i/ and /u/ is as lim-
ited as it is in other bedouin dialects of Sinai, but the minimal pair Xidr
male given namexudr green (c. pl.) will isolate these vowels as sepa-
rate phonemes in Lgiy and H amdi y. Finding a minimal pair to isolate
/a/ as a phoneme is not a problem in Sinai; for Lgiy a pair like xa he
enteredxu enter! works like in any other Sinai bedouin dialect.
Long vowel phonemes are //, // and //, and additional (uncondi-
tionally monophthongized diphthongs *ay and *aw) // and //. Result-
14
The opposition is about as widely used in Lgiy as it is in other dialects of southern
Sinai. In H amdi y, however, the suffix -kiy is also used instead of -k for the 2nd p. f. sg.,
while V(C)uk (CC-uk) is regular for the 2nd. p. m. sg. in both H amdiy and Lgiy.
15
The spelling here with triple t is for the sake of morphological transparency. The
pronunciation is, however, not noticeably different from doubled t (IPA [(t)]). See also
fn 90.
ing monophthongs and tend to be realized lower when they are

preceded by some of the back spirants (notably x, and h and ), or
by certain (primary or secondary) emphatics: da f guest, and also xat
thread. ar must, and I have also heard Farn (Wdiy) Frn and less
(but still) lowering in nuss n two halves and masta rnih ruler; other
than, h l (in which // is near, or just above IPA [ :]) year, but also quite
low // in rdin get (f. pl.) water. When h preceedes, however, there is
no noticeable lowering, as in hl cardamon and also IPA [o] in Abuw
lHl Blue Hole.16
In some cases a diphthong will remain unchanged, as in e.g. (loan
from English) dayf dive, lay (~ l) to me, aw or, law if (loan from
Modern Standard Arabic), or when Systemzwang is operative, as in
mawjd present, in the imperative form awuw watch out (m. pl.) and
also in the f. sg. form h awly (m. sg. ah awal) cross-eyed, where reten-
tion of the diphthong prevents a (near) homophonic clash with h l my
year.
2.2.1 Raising the short vowel a

Short a is raised in a variety of positions preceding primary stress. In
the cases cited below, such raising is concluded to be optional, since it
does not occur always. Raising of a tends to be towards [u] when it pre-
cedes C, or when followed by w, and towards [i] in otherwise neutral
environments.
Lgiy ilh (~ alh) on him (notice the short base,17 but in H amdi y
more regularly ilh), dibht I slaughtered and in sandhi also i- mb
on the side.
Short a preceding stressed : gibyil tribes, iml north and ibn
hungry, kisln lazy. Short a preceding stressed : kitr much, bid far,
digg flour, ijn dough. Although such raising is predominant, forms
with unraised a also appear, e.g. targah method, kabr large, and even
more regularly so when h , precedes a, as in h add iron, ars groom,
xarf autumn.
16
Abuw lHl, lit. The Sphinx, but here as a folk etymology for The Blue Hole, which
is a popular dive site about 12 km north of D ahab.
17
This must have developed in analogy to such variation as luh (~ much less frequent)
lh to him. Comparable forms (alh and alh) were also recorded in the dialect of the
Tarbn of the north and that of the Rmlt (respectively), see De Jong (2000, 181).
570 rudolf de jong
Short a preceding stressed : unb (~ anb) south, isSudiyyih

Saudi Arabia, gud camel that has not yet cut its canine teeth,18 lugh
impregnated (of a she-camel) and uby (~ aby) my father. Absence
of such raising was also recorded, as in ar emperor fish.19 Such raising
appears to be less regular when h precedes, as in h amlih animal led to a
party to be slaughtered for those present,20 ars bride, azz pole. The
gahawah-vowel (see below) a preceding in an open syllable remains
unaffected, e.g. maxarm pierced, maarf known, mah att placed.
Preceding stressed : iluk on you, fidt I sacrificed and middt I
stretched, xit I entered, istinnt I waited. Raising towards [u] pre-
ceding w: in muwwl poems, ruwynah well-watered, suwwt I did.
In some cases comparable raising has led to morphological restruc-
turing.21 Examples are irbt I drank, nist I forgot, ligt I found. Notice
that in such cases underlying |a| no longer has a surface form a.22
Not only when preceding primary stress, but also when preceding
secondary stress (and following primary stress), a may be raised in
certain positions. Examples are (Lgiy): nwikal he was eaten, ttifag
he agreed, jjimal the camel and lmutar the rain. Raising of a in
these previous four examples is optional, but raising of a has become
steady (i.e. it has become morphologically restructured, although it has
remained underlying |a|) in open syllables of imperfect forms such as
(underlying |a| underlined) ynbisit he rejoices, yttifig he agrees.23
2.2.2 Final ah and *- ( )

The fem. morpheme tends to be raised, and not only in pause. The height
of raising is from [e] to [i] (e.g. sayyrih # small, madrasih # school,
mirfih # acquaintance, h jih # thing, zibdih # butter) provided such
18
See gad, Bailey (2004, 457) (glossary).
19
See (spelled shaour) webpage http://www.saudi-fisheries.comArabic/fish_
product_2.htm
20
In other dialects h amlah is used in the meaning of clan, see Blanc (1970, 114).
21
a no longer appears at the surface as a, but as i. This i, however, is not dropped
in open unstressed syllables (e.g. iribt, not ribt for I drank, the latter of which can
be heard e.g. in the dialect of alAr, see De Jong (2000, 515)), and is therefore to be
regarded as still underlying |a|, see also following fn.
22
In Negev dialect, on the other hand, the |a| reappears as a in closed syllables, e.g.
arbit she drank, cf. Blanc (1970, 134).
23
In closed syllables however, underlying |a| does indeed reappear as a, e.g. yinbstu w
they rejoice, yittfgin they (f.) agree. Notice that in the dialect of alAr, this vowel i is
also underlyingly |i|, which may be concluded from its elision in a form like yniwkil it
is eaten and (not ynwikil), see De Jong (2000, 521).
raising is not inhibited by phonetic factors, such as preceding emphat-

ics (e.g. h tah #, ajarah # tree, H amda h # name of the tribe/clan) or
(some) X (e.g. srh ah # taking out to graze, ladah sting, bite, but notice
raising in tisih # nine.
Extreme raising of final *- () may be heard in the f. sg. *fal pattern
for colors and bodily deficiencies, e.g. arjy limping, h awly cross-eyed,
aby dark-colored, (a gahawah-form) ah aby light-colored. When
velarized consonants precede, raising remains absent (and often a glot-
tal catch will be audible): bd white, h amr red, safr yellow, zarg
blue, black, xadr green and also tarm having a gap in ones teeth.
Other cases of raising of - () include: salt i evening prayer, ti
summer, btti her house, ilf i diy this viper (although lf iy was also
recorded), ngatti h h i we cut it (f. sg.) to pieces and also ji he came.
Although the final glottal catch presumably originated as a pausal fea-
ture, it may often be heard in (especially lento forms in) sandhi as well,
keeping such endings separate from forms ending in t marbtah.
However, when a directly precedes in open syllable, and/or velariza-
tion is effective, such raising tends to remain absent, e.g. sma sky,
lada the lunch, d wa he returned home before sunset, ma he
went, fda he sacrificed, but also biyffa he sees her, (a gahawah-
form) taxarizha24 you stitch it (f. sg.).
3. Stress and phonotactics
As for ordering of rules, the rules for elision stress and anaptyxis are
executed in that same order, i.e. like in other bedouin dialects of Sinai.
3.1 Stress
A sequence CaCaC is usually stressed CaCC in Lgiy, while H amdi y
much more regularly stresses CCaC. The article is often stressed in
both dialects, as in lCaCaC (e.g. ljimal the camel, lmutar the rain,
llah am the meat, aar the hair), but is in a limited number of cases
unstressed in H amdi y alCCaC, e.g. alfrah the wedding ceremony,
ilnam the goats, ilbsal the onions. Comparable stress patterns occur
24
Raising of final in the pronominal suffix -h in neutral environments appears to
occur at random; I have not been able to discover a pattern.
572 rudolf de jong
in the perfect forms of verbal measures VII (or n-1) and VIII (or 1-t):
they tend to be stressed inCCac and iCtCaC in H amdi y, but nCa-
CaC and CtaCaC in Lgiy, while both varieties stress ynCiCiC and
yCtiCiC. Examples are (Lgiy) nkabas, ynkibis be jinxed, stawa,
ystiwiy become ripe/cooked (H amdi y) inbsat, ynbisit rejoice and
ittfag, yttifig agree.25
3.2 The gahawah-syndrome

As has already become apparent in several examples cited above, the
gahawah-syndrome has left its traces in Lgiy and H amdi y as well.
Some additional examples are ghawah coffee, h awal cross-eyed,
th alib she milks, yazlin they (f.) spin. Notice that when suffixed, the
fem. suffix -ah in gahawah-forms becomes -at, irrespective of whether
the suffix is vowel-initial or consonant-initial, e.g. gahawat my coffee,
gahawathin their (f.) coffee.26
3.3 Stress in gahawah-forms

In verb forms, the gahawah-vowel a has a tendency to behave like
an ordinary anaptyctic, in the sense that it remains unstressed, where
it would otherwise receive stress, (gahawah-vowels underlined) e.g.
th ar tuw you (m. pl.) plough and ya zlin they (f. pl.) spin;27 compare
these to forms like (anaptyctic high vowel underlined) tgu duw you
(m. pl.) sit and contrast with stress in e.g. saltuw you (m. pl.) asked.
3.4 Elision and anaptyxis

Generally, elision of high short vowels i and u (except those that are
underlying |a|) takes place like elsewhere in Sinai.28
25
Notice that group II (i.e. the dialects of the Samnah and Agylah) dialects in the
north have stress patterns inCCaC, yinCCiC. On the intermediate position between
groups I and II in the north of the stress-type inCCaC, ynCiCiC, see De Jong (2004,
159).
26
See remarks below.
27
Comparable forms with vowel-initial suffixes were heard in group I, see De Jong
(2000, 105).
28
See De Jong (2000, chapters IIII, 2.4).
4. Morphology
As bedouin dialects do, Lgiy and H amdi y too differentiate between

the masculine and feminine in plural forms of verbs and nouns.
4.1 Nominal morphology
4.1.1 Pronominals
The personal pronominal for the 3rd person masc. pl. in Lgiy and
H amdi y is hum m a, while h in Lgiy and huwwa in H amdi y tend to
be used for the 3rd person masc. sg.29 The suffixed form of the 3rd p.
masc. sg. is -u(h) in both dialects. The 3rd p. m. pl. pron. suffix is usually
-huw in Lgiy, but -hum in H amdi y. Both dialects have hinna and -hin
for the f. pl.
The 2nd person masc. pl. is often intum in H amdi y, but intuw in
Lgiy, the latter of which is also heard in southern Sinai dialects farther
east. Similarly, the suffixed pronominal directly elicited in H amdi y was
-kum, while I only heard -kuw in Lgiy, the latter of which is also heard
in dialects farther to the east. Both dialects have intin and -kin for the
f. pl. The 2nd p. sg. suffixes -k and -k, which are typical for many south-
ern Sinai dialects,30 are also present in H amdi y and Lgiy.
Like almost all dialects of Sinai, Lgiy and H amdi y have stressed 1st
p. c. sg. suffixes -n (obj.) and - (poss.).31
4.1.2 The article

The vowel of the article tends to vary. The article is il- or al-, which
appears to be mainly dependant on the vowel in the first syllable of the
following nominal. If this syllable contains (surface) a, the article tends
to be al, assamm the poison, assaddah the dam, assa fr the yellow,
attamr the dates, alamaliyyah the affair, almayyah the water, alh all
the small cattle and also aldra h the corn.
29
Some central and eastern dialects of southern Sinai have huwwa for the m. pl.,
while h is used for the masc. sg.
30
See De Jong (2004, 163164).
31
See Blanc (1970, 130131), De Jong (2000, 675, map 38) and ibid. (2004, 163, with
remark *** on 164165).
574 rudolf de jong
In other cases, the article il- is much more likely to be used (though
al- occurs in such positions as well), e.g. ilwild the boys, issin the goat-
skin (used for churning butter), iddims the stone, ilkibdah the liver,
iddinya the world.
In stressed positions (i.e. in Lgiy), the article is almost invariably
al- (or assimilated allomorphs with initial a-), e.g. ssa h an the bowl,
ljimal the camel, ssalag the hunting dog, lh atab the firewood and
a h a the morning, but ti (~ mati) the winter.
dd
Apart from assimilating to sunletters, l of the article will also often
assimilate to j, e.g. jjimal the camel, ajjawlig the carpets, ijjihhl the
youngsters.
4.1.3 T in construct state

The fem. morpheme T (-ah) becomes -at, when aC directly precedes,
e.g.: ragabatha her neck, ragabat my neck, and (gahawah-form) gaha-
watha her coffee,32 gahawat my coffee and also sanatn two years.
Examples in sandhi are: h arakat stn something like two hours,
maslah at alh urmah the affair of the woman.
Otherwise, T becomes -it, whose high vowel is dropped in open syl-
lables, e.g.: (in closed syllables) jurrithe her traces, rh tuk your smell,
ngitkin your (f. pl.) she-camel, ilbtk your (f. sg.) pack/can, wijitha
her pain (i.e. the pain she causes) and arabiyythum their (m. pl.) car.
In sandhi: wayyit rizz a bit of rice, rih lit ats a diving trip, aklit ixd r a
vegetable dish
Instances with dropped i (often anaptycticshere underlinedare
inserted to eliminate a resulting cluster) ilibt my pack/can, arabytuk33
your car, and in sandhi u l t almayyah of (gen. exponent) the water. rht
ilbirdign the smell of the orange, giblt iliLgt the tribe of the Lgt.
4.2 Verbal morphology

A notable feature of verbal morphology is the vowel harmony in pre-
formatives of the imperfect forms, which is common in many Sinai dia-
32
In this respect, Lgiy differs from dialects of the Negev, where one will hear (also)
gahawat(+), but gahawithin, see Blanc (1970, 142). Other dialects in Sinai show yet
other ways of treatment of T in gahawah-forms, see De Jong (2000, various chapters,
3.1.10.3). Treatment of T in Lgiy is much like what was described for Smniy (i.e.
the dialect of the Samnah) of group II in the north, see De Jong (2000, 279280).
33
Notice that the form is not arabiyytk! Contrast with the form arabiyythum listed
above.
lects, but in our dialects discussed here the first p. c. singular is included
in this rule (so that in measure 1 of regular verbs the 1st p. c. sg becomes
homophonic with the m. sg. imperative form), e.g. iktib I write, udr ub
I hit and arab I drink, but also ugm I rise, il I carry, anm I sleep
and jiy I come.34
Another characteristic of bedouin dialects in the south of Sinai
(including Mzniy), though not of all, is the apocopation of 2nd p. masc.
sg. forms of tertiae infirmae verbs. This occurs in all measures (indi-
cated in Roman numbers here), e.g. in Lgiy: (I) tim you go, (I) talg
you find, (II) tsaww you make/do, (III) tlg you find, (IV) tit you give,
(V) taa you have dinner, (VI) talg you meet (with),35 (VIII) titir
you buy and also (with shortened base vowel) tij (~ tjiy) you come.
In a similar fashion, masc. imperatives of tertiae y verbs are usually
apocopated, as in the examples ijr run, ans forget, saww do/make!, it
give! etc. Also when suffixes are appended, the apocopated forms are
used, e.g. ansuh forget him!, sawwha make it (f. sg.)!, ituh-yyh give it
to him! and itwarrh-iyyh you show her to him.36
5. Typological positioning of Lgiy and H amdi y inside Sinai
Lgiy may not very convincingly be part of the group I or Negev dialect-
type (as illustrated below in a comparison with the group I dialect of
the Tarbn around Rs Sadr, who are the northern neighbours of the
Lgt), but it is still nearer to group I than H amdi y.37 In this respect the
34
Forms like atlub and aktib (like in dialects of northern Sinai, see De Jong (2000,
3.2.1.2 of chapters 1V)) were also recorded, but the forms with vowel harmony in
the 1st p. c. sg. turn out to be much more regular than I had previously noticed (contrast
De Jong (2004, 166), where only atlub and aktib are listed).
35
Another feature typical of many southern Sinai dialects is the reduction of initial
tt- (> t-, sometimes erroneously referred to as a haplological drop of ta-) in ta- initial
measures (V and VI).
36
The origin of such apocopation may lie in a reasoning by analogy (extrapolation
leading to paradigmatic levelling within the verb system): if forms like e.g. imperfect
tgudiy and tkitbiy (or imperative gudiy and kitbiy) are fem. forms for the 2nd per-
son singular, and the forms tugud and tiktib are used for the 2nd p. sg. masculine, then
dropping the -iy ending from a 2nd p. f. sg. forms like e.g. titiy will yield the form for
the 2nd p. masc. sg.: tit.
37
Largely illustrating Murrays remark (1935, 263 f, see fn. 13 above). The dialect of
the Tarbn of Rs Sadr (not very different from other varieties of Turbniy) is taken
here as representative for group I (or Negev-type) dialects. For the latter type, see Blanc
(1970) and De Jong (2000, chapter 1).
576 rudolf de jong
following dialectal featuresfor the greater part in recapitulationmay

be considered (this list is not exhaustive):3839
Lgiy and H amdiy* Turbniy
Phonology
phoneme */k/ /k/ and /k/38 only /k/
allophones of /j/ [d] ~ often [] [d], rarely []
diphthongs *ay, *aw // and // phon. conditioned ay
and aw
raising of a preceding middt maddt
(extreme) raising *- () ti, lai tiy, laiy
velariz. in pl. of katr ktar (no velarization) ktr (velarized)
Stress CaCC (Lg) CaCC

CCaC (~ few CaCC)
(H m)
lCaCaC (Lg) lCaCaC
alCCaC ~ lCaCaC
(H m)
Parts of Speech
Article and relative al-/-il & alliy / illiy only al- & alliy
pronoun
Demonstratives: singular m. (h)da, f. (h)diy m. hda, f. hdi y
coll., plural dill (ih) (~ hdil)(Lg) hdal
/ hd
l

dill (ih) (~ hdl
)
(H m)
Negated pronoun 3. sg. mh, mh (~ mha) mh, mh (~ mha)
Pronoun suffix 3. m. sg -uh -ah /-ih
3. m. pl. -huw (Lg), -hum -hum
(H m)
2 m. sg. C-uk C-ak
2 f. sg. V(C)-kiy, CC-ik39 invariable -kiy
Suffixed prepositions muh mah
ilh (Lg), ilh (H m) alh
fh, f uk fah, fak
Verb impf. 1st p. c. sg. iktib, udrub,
arab, aktib, adrub,
arab,
ugm, uxu, etc. agm, axu, etc.
Apocopated impf. 2 m. sg. tim, talg, tlg timiy, talga, tlgiy,
tit, taadd, etc. titiy, taadda, etc.
38
But this phonemic opposition is not as widely used in Lg.
39
This is not entirely certain; my H amdiy informant produced btk your (f. sg.)
house, but also naxal tkiy your (f. sg.) date palm.
Verb to come i (pl.) um ~uw (sg.) ja (pl.) jaw

perf. 3.m. (sg.)
impf. 3.m. yiy, yuw yjiy, yjuw
Suffixed prepositions: muh mah
ilh (Lg) alh
Vowel impf. to eat, to take ykul, yxud (H m) ykil, yxid
ykil, yxid (Lg)
Vowel perf. 3rd p. f. pl. impf. katabin, glin, kataban, gl an
yarabin yaraban
-m in 2nd m. pl. perfect katabtuw (Lg), katabtuw
katabtum40~-tuw
(H m)
Interrogatives
when? imtn mat (~ wagt)
who? mn min
Adverb
here nih(-niy) hniy
Selected lexicon
mother um m am m
baking sheet z sj
* Where not specified, the listed form is heard in both dialects.40
6. Conclusion
The comparison drawn above shows a number of differences between

Lgiy/H amdi y and Turb niy. Notwithstanding the characteristics
shared only by H amdi y and Turb niy, however, Lgiy still appears to
be typologically nearer to Turb niy than H amdi y.
7. References
at-Tayyib, Muhammad Sulaymn. 1997. Mawsat al-qabil al-arabiyya, Part 2, Cairo:

Dr al-Fikr al-Arab.
Bailey, Clinton. 1974. Bedouin Weddings in Sinai and the Negev. Folklore Research
Center Studies, vol. 4, Jerusalem Magness Press, 105132.
. 1985. Dating the Arrival of the Bedouin Tribes in Sinai and the Negev. Journal of
the Economic and Social History of the Orient 28, 2049.
. 2004. A Culture of Desert Survival. Bedouin Proverbs from Sinai and the Negev. New
Haven: Yale University Press.
40
Verb forms with final -m were only recorded during direct elicitation. Such forms
also occur in group II of the north, see De Jong (2000, 298299).
578 rudolf de jong
Behnstedt, Peter. 1979. Die nordmittelgyptischen bukara-Dialekte. Zeitschrift fr ara-

bische Linguistik 3, 6295.
Blanc, Haim. 1970. The Arabic Dialect of the Negev Bedouins. Israel Academy of Sci-
ences and Humanities, Proceedings, vol. 4, 112150.
de Jong, Rudolf. 2000. A Grammar of the Bedouin Dialects of the Northern Sinai Litto-
ral. Bridging the Linguistic Gap between the Eastern and Western Arab World. Leiden:
E. J. Brill.
. 2004, Characteristics of Bedouin Dialects in Southern Sinai. Approaches to Arabic
Dialects. A Collection of articles presented to Manfred Woidich on the occasion of his
sixtieth birthday, ed. by Martine Haak, Rudolf de Jong, Kees Versteegh, Leiden: E. J.
Brill, 151175.
Hava, J.G. 1982. Al-Farid ad-durriyyah, Arabic-English Dictionary (fifth edition). Bei-
rut: Dar el-Mashreq.
Lavie, Smadar. 1990. The Poetics of Military Occupation. Berkeley: University of Cali-
fornia Press.
Marx, Emanuel. 1999. Oases in South Sinai. Human Ecology 27, 341357. New York:
Kluwer Academic.
Murray, George W. 1935. Sons of Ishmaela Study of the Egyptian Bedouin. London:
Routledge.
Oppenheim, Max Freiherr von. 1943. Die Beduinen, Band II: Die Beduinenstmme in
Palstina, Transjordanien, Sinai, H edjz. Leipzig: O. Harrassowitz.
Stewart, Frank Henderson. 1990. Texts in Sinai Bedouin Law, part 2 (Texts are in Ara-
bic). Wiesbaden: O. Harrassowitz.
Tckholm, Vivi. 1974. Students Flora of Egypt. 2nd ed., Beirut: Cairo University (Coop-
erative Printing Company).
8. Texts
The speaker is a member of the Lgt, who was 38 years old at the time
of recording. He was born in Sarbt alXdim (about 40 km east of
Abuw Znmah) in the area called ar-Ramlah, where he had lived until
his 29th. He has had no school training. (S) = Speaker of Lgiy, (R) =
interviewer.
8.1 Hunting rabbits
8.1.1 Transcript
1. (R) hatkallimni fi sayd ilarnib? 2. (S) a:ywah. alarnib dilleh41 . . . fi
nnahr bitnm. fi nnahr . . . bitnm m btarta.42 3. ibtarta ar h
billl . . . ugb almaarib . . . ibtusrub. iza nymah fi h ajr, fi ajarah . . .
41
dill (-ih) is the c. pl. demonstrative pronoun for near deixis. Notice the absence of
velarization.
42
The single negation m + verb form is regular.
ibtusrub h fi lll bitdawwr alma. 4. idda ll43 h srbeh, iw srh ah . . .

l assu bih . . . . la:44 lfajir. alfajir xals giddm aam la dd rr45 b iwayyeh . . .
u
5. bitdawwir lhe . . . ajarah . . . bitdawwir lhe jibl, bitdawwir lh-ayyi
h jeh, iw bitgtir fha bitnm. 6. iygl luk46 alarnab nymeh. alarnab
dd a h a, w issu bih . . . tulgutha . . . iw h nymeh. 7. ibyalguw jurritte,
yalguw jurritta ind ijar da ind ijar da ind ijar da. 8. iw
biygtiruw mi jurritte . . . lam biyrawwh uw ind . . . manmha.
8.1.2 Translation
1. (R) Are you going talk to me about hunting rabbits? 2. (S) Yes. These
rabbits . . . they sleep during the day. They sleep during the day and they
dont graze. 3. They only graze at night . . . after sunset. They run out,
if they sleep in the rocks, into the bushes . . . They run out at night to
look for food. 4. They keep running . . . grazing till the morning, (all
the time) until daybreak. Daybreak, (after that its) over, just before the
sun rises . . . 5. It looks for a bush . . . it looks for a mountain, it looks for
anything, and goes (in) there to sleep. 6. Hell tell you the rabbits are
asleep. The rabbits in the forenoon, which is the morning . . . you seize
them . . . while they are asleep. 7. They find their tracks, they find its tracks
near this bush, near that bush. 8. And they follow its tracks . . . until they
come to . . . its sleeping place.
8.2.1 Transcript
9. fh . . . ns ibyulgutha . . . b duh. iw fh ns biyannin alhe. biyannin
alha yaniy bitkn imh47 ash . . . igsayyreh. 10. ibrn kdiy . . . hda
ym ffa48 nymah fi ajarah, iyannin alha w yuxbutha b ilash . . . 11.
iw linnha49 tbbah fhe.[. . .] iw fh ns . . . induh salg. 12. ssalag hda
43
a l: assimilation of initial t to follwing d in tdall.
dd
44
Such prosodic lengthening of the vowel is often used to indicate the long duration
of time.
45
dd u rr, see fn 43.
46
The suffixed preposition l: (sg.) 3.m. luh, 3.f. lha, 2.m. luk, 2.f. lkiy (~ seldom lik),
1.c. lay ~ l and (pl.) 3.m. lhum , 3.f. lhin, 2.m. lukum, 2.f. lkin, 1.c. lna.
47
The suffixed preposition m(i): (sg.) muh, mih h a, muk, mik (~ frequent mikiy, i.e.
like in group I), m and pl. mih h um , mih h in, mikum , mikin, mina.
48
ffa: fha one of many examples of assimilation of h to a preceding voiceless
consonant.
49
The particle linn or lann is often used to present a sudden or unexpected turn in a
narration, see Blanc (1970, 145).
580 rudolf de jong
. . . biynaggizhe . . . biynaggizh-ssalag, 13. w imarabbh . . . ind ssalag i

mna . . . biyda lliy . . . ssalag yizatti . . . lamma yulgutha. 14. m ykilha
lamma yjbha la sh buh. iygm sh buh dbh h e. hda larnab. (R)
aywah . . . 15. (S) iw fh ns biyannin alha tann bi lash biyirmha.
16. iw fh ns ibyulgutha. imh abtuh, imh bantuh muh
h jah lm ffa fi ajarah, ramha alha w lagatha. hda larnab.
8.2.2 Translation
9. There are people who seize her with the hand, and there are people
who take aim at it. They take aim at it, that is, theyll have a small stick
with them. 10. Of two spans of the hand (in length) . . . this (person)
when he sees it asleep in a bush, he creaps up on it and hits it with the
stick. . . . 11. And lo it (i.e. the stick) hits it (i.e. the rabbit). And there
are people who have a hunting dog. 12. This hunting dog will make it
jump around. The hunting dog makes it jump around. 13. And he has
taught it . . . with the hunting dog . . . the hunting dog keeps following it
closely . . . until it grabs it. 14. He wont eat it until it has brought it to its
master. His master then slaughters it, this rabbit. 15. And there are peo-
ple who take good aim at it with the stick, which he throws. And there
are people who seize it. 16. Hell have his cloak with him, (or) his jacket
with him, or something, (and) when he sees it in the bush, he throws it
at it and seizes it. This is (what they do with) the rabbit.
8.3.1 Transcript
17. (R) iw fh bardu h . . . biysawwuw faxx? (S) biysawwuw. alfaxx da
grayyib, alfaxx da grayyib byistagduw50 alh nns mi zamn. 18.
igrayyib byistagduw a lfaxx. yibnh,51 w iyh uttu w fh . . . iyh uttu w fh
giirit52 birdagneh, iyh uttu w fh ayyi h jah lfaxx. 19. iw h btjiy a
53
rhtuh . . . hda byibnh ar ugb almaarib. ym lanam tidw iy, w
alh all dwa xals . . . m f h all, ibyibnh. 20. m byibnh fi nnahr.
50
A measure X (or ista-1) verb of the root q-d-y: istagda, yistagdiy (ala) take up as a
new habit by following an example.
51
yibnh lit. they build it, here they set it (i.e. the trap).
52
giirit: a bukara-form of girit.
53
da wa, yidwiy
go home in the evening, see Stewart (1990, 214) (glossary). In
Turbniy this verb is measure IV, e.g. albant midiwyt the girls are going home (just
before sunset). The root d-w-y
is probably related to CA d-w- light; using the last day-
light to go home.
iw tniy h jeh byibnh fi lgd . . . wagt assiyyl ym assiyyl, fh barm iw

h l, 21. ibyibnh kamn fi irg54 assiyyleh. fh barm biyth h bitrawwi
diy55 fi lgd, fi ahar assabah w ihr issittah w ihr dillah lgd. 22. fh
barm biyth fi ssiyyl, talgha56 tamalliy fi ssiyyl alarnib bitrawwih . . .
8.3.2 Translation
17. (R) And are there . . . (people) who make a trap? (S) Yes, they do.
This trap is (a) recent (thing), this trap is something recent, which
people have imitated, not people of the old days. People have started
copying the (method of hunting with a) trap. They set it, and they place
in it . . . they place the peel of an orange in it, they put anything in the
trap. 19. And it (the rabbit) is attracted to its smell. This (person) sets it
only after sunset. When the small cattle come back home. And when it
has come back home . . . thats it . . . there is no small cattle (left roaming
around), (then) they set it. 20. They dont set it during the daytime. And
besides this they set it in summer . . . the time when the acacias bloom,
when the acacias have fruit57 and stuff. 21. They also set it next to the
acacia tree. There is fruit that falls down. It comes normally in summer,
the month of July, the months of June and those of summer. 22. There
is fruit that falls from the acacias, (so) youll find rabbits going to the
acacias all the time.
8.3.3 Transcript
23. ibyibnuw lha fa . . . alfaxx f-assiyylah diy, iw kamn fi ssiyyl d k,
fi ttalah58 diy byibnuw lhin. 24. iw yulugtha f- alfaxx . . . yulgutha
min ragabatta min . . . min rjilha min h jeh. 25. assu bih ym yjiy l
alfaxx iw lannha malgtah. 26. iw fh ns ibyugud . . . igrayyib lha . . . fi
lll . . . yasma xabit alfaxx. 27. ym ytulguh byasmauh (R) aywah . . . (S)
54
fi irg, litt. in the root (of) has been grammaticalized as a preposition meaning
next to (presumably next to some standing object), also e.g. fi irg alh tah next to the
wall.
55
Assimilation of bitrawwih + diy.
56
talgha: apocopated 2nd p. m. sg. imperfect of the verb ligiy, yalga find, followed by
suffix -ha.
57
Lane (1863, part 1, 195): [. . .] the fruit of the talh [or acacia gummifera, which is
of the trees called idh] [. . .]
58
talah (with t!) is a water course between two mountain peaks, and can be used as
a pass between mountains.
582 rudolf de jong
ibyarf alfaxx sdha . . . ibylgutuh . . . fi lll 28. iw fh ns ibyibnh . . . iw

biygtir ibld bideh . . . zayy xamsah kluh aarah kluh ssu bh ibyjiy luh.
29. ym yjiy luh linnuh, linnuh59 lgitha. hda larnab. 30. amma lazl
hda, m kull anns ibtugnus luh, alizln dillah m biyth in fi l . . . fi
lwtiy, ar fi jjbl, fi jjbl albideh . . . 31. (R) ar tatla fg . . . (S) ar tugnus
lhin fi lijbl albideh . . . alazl. 32. w alazl law arwh uk60 . . . ibyun-
guz min induh. 33 law arwah rh tuk . . . yaniy jt a rrh w arwah , xals
biygtir min induk. hda lazl. 34. alliy h lbadan w alh jt dillah
kulluh. hda . . . hda wn ilinsn ibylugtuh.
8.3.4 Translation
23. They set a trap for it in (under) this acacia, and also in those (other)
acacias, in this watercourse they set (traps) for them. 24. and they catch
it in the trap. . . . they catch it by its neck . . . by its leg . . . by something. 25.
In the morning, when he comes to the trap, there it is, caught. 26. And
there are people who sit (and wait) . . . near it . . . at night . . . hell hear the
snapping of the trap. 27. When it releases hell hear it . . . (R) Yes . . . (S)
and he knows that the trap caught it . . . (and) he seizes it . . . at night 28.
and there are people who set it . . . and they go away to a far place61 . . . like
five kilometres, ten kilometres (and) in the morning he comes (back) to
it. 29. When he comes to it, lo, lo, it has caught it. That is the rabbits.
30. As far as these gazelles are concerned, not all people hunt them.
These gazelles dont come down to the lower areas, (they stay) only in
the mountains . . . 31. (R) You have to go up . . . (S) You have to hunt them
in the far mountains . . . Gazelles. 32. And gazelles, when they smell
you, they jump away from where they are. 33. When they smell your
scent. . . . that is, if you come down with the wind and he smells (you),
thats it, he gets away from you. These are the gazelles, 34. which are the
ibex and all these things. This . . . this is where a person catches them.
59
See fn. 49.
60
The (superscript) anaptyctic vowel is voiceless.
61
bld lit. land, here place.
8.4 Snakes
8.4.1 Transcript
62
35. (R) fh bardu h, rif fi ddd? (S) iddd allafiy? (R) aywah . . . (S)
allafiy dilleh . . . hda l . . . ayyi insn ibyuktilhe. 36. hdiy ygl luk
almzyeh, in kaltuk bidduk,63 ir kn64 daktr walla bidduk, aza
f-albarr kamn m h waluk daktr ir kn insn h wiy. 37. fh ns
f-larab . . . biykn h wiy. [. . .] fh hwytuh xabit, iw fh ihwytuh lah s,
alliy hda biyrudd assamm. 38. hda ddd, alliy h lagrab . . . assa fr diy-
m m dl . . . iw fh lfiy . . . alliy h . . . h tti bn. hda la . . . hda lam kaltuk
r kn insn h wiy walla daktr. 39. fh ns katr mtuw minha. imn
alamaliyyah diy. (R) mi ssamm . . . (S) mi ssamm, hda ssamm . . . 40.
(R) tab w ilh wiy biysawwiy h? (S) alh wiy biyrudd issamm . . . biyrudd
assamm . . . 41. (R) kf biysawwiy? (S) alh wiy byaxabtuk xabit . . .
yaxabtuk . . . talat xabtt . . . ib rguh . . . (R) ib rguh . . . (S) hda biyruddd
assamm. lamma trawwh addaktr, aw mumkin kamn m . . . m trawwh
addaktr ihwytuh jiyydeh, bitkaffiy xals. 42. hda b innisbah la . . . la
lagarab65 w alfiy w atti bn . . . w alh jt dillah.
8.4.2 Translation
35. (R) Are there also . . . do you know about creepy crawlies? (S) Vipers?
(R) Yes. . . . (S) Yes, these vipers . . . these . . . anyone will kill them. 36. This
one will say to you the advantage, if it bites you, you need, by necessity
a doctor or you need, if it is in the desert, and there is also no doctor
around (you), a hwiy person. 37. There are people among the bedou-
ins . . . who are hwiy. Hitting is part of his hwiy-activity, and licking is
part of his hwiy-activity, which brings back (i.e. out of the body) the
poison. 38. This is the creepy crawly, which is the scorpion, this yellow
one with a tail . . . and there is the viper . . . which is the snake. This one,
when it bites you, you need a hwiy or a doctor. 39. There are many
people who died from it, from this business. (R) from the poison . . . (S)
62
dd (n.u. -ih, pl. ddn) is used for any crawling animal, like snakes, scorpions, and
lizzards.
63
bidd, rather than widd (as in group I), is used to express want or need. Here it us
used to express necessity from the perspective of the speaker, like English should, as in
he should see a doctor, see De Jong (2000, 239).
64
The context calls for words like ayr kn. In the recording I hear what has been
transcribed above, but perhaps it should read (strongly reduced) er kn.
65
ag araba bukara form for agrab.
584 rudolf de jong
From the poison, this poison. 40. (R) Okay, and what does the hwiy do?
(S) The hwiy brings back the poison, he brings back the poison . . . 41.
(R) how does he do it? (S) The hwiy gives you a good slapping. He slaps
you three times . . . with his saliva. . . . (R) With his saliva. (S) This brings
back the poison until you get to the doctor, and its also possible that
you dont (have to) go to the doctor. If his hwiy-work is good, thats
enough, thats it. 42. This in reference to the scorpion and the viper and
the snake . . . and these things.
8.5 Falling in love
8.5.1 Transcript
43. (S) lwalad . . . lwalad biygm . . . aza srih . . . biyf albint . . .
44. (R) bifha wn? (S) biyfha fi masrah h a w h srh ah b anamhe. 45.
jabatuh . . . iw h l . . . biygm . . . biyxarrf um m uh, h m biyxarrf ubh,
biyxarrf um m uh. 46. ummu h bitxarrf ubh. bitgl lwalad . . . ryid bint
ifln . . . imn larab iliflniyyah. . . . 47. biygm ubh . . . biyrawwih l ub
ha . . . iw . . . biyrawwih l ubha w biyxarrfuh. 48. w um m uh . . . bitrawwih
l um m albint, iw bitxarrifhe. gl abuw lbint mh maxatbeh . . . iw bin-
jawwizkuw . . . hda fh kalm tniy. 49. gl abuw lbint maxatbeh66 . . .
hda fh kalm. gl abuw lbint kamn yizha walad ammha . . . hda
fh kalm tniy. 50. iw fh ns biyffa . . . min tanabh.67 tanabh alliy
induh biyf bint ibtijbuh . . . biyxarrf um m uh. 51. biygl ana yz ilbint
lifln diy. 52. bitrawwih l um m ha w bitxarrifha w tasalha w itgl h
lbint maxatbah walla bint ass ibnt walla kidiy walla kidiy . . . 53.
bitxarrifha. baadn ubh . . . biyrawwih l arrijjl . . . abuw lbint . . . iw biyx-
arrfuh. 54. biyxarrfuh gl h biyjawwzuk xals albint mh maxatbah,
iygl luk xals. 55. iza lbint isayyrih biygl luk xalluk ugub sanah.
xalluk ugub xamis t-uhur, xalluk ugub sanatn . . . 56. biygm byadbah
alhe . . . biyjb tuh gl luh xalluk ugub sanah, sanatn . . . albint
isayyrih. 57. biygm biyjb tuh w yadbah h a. dabah h -ath algasalah.
58. min ym ath lgasalah xals irif hdiy h rumtuh, ib sinnt All h w
rasl-ath . . . gsalatuh . . .
66
maxatbah, gahawah-form of maxtbah.
67
tinb, tanaba member of the same encampment, see Stewart (1990, 273274,
glossary).
8.5.2 Translation
43. (S) The boy . . . the boy then goes . . . when he is out herding . . . he sees
the girl . . . 44. (R) Where does he see her? (S) He sees her in her herd-
ing place where she is herding her goats. 45. If he likes her. . . . and in
case. . . . he then goes . . . and speaks to his mother. He doesnt speak to his
father, he speaks to his mother. 46. His mother speaks to his father. She
says The boy wants (to marry) so-and-sos daughter . . . from so-and-sos
family . . . 47. His father then goes to her father . . . and . . . he goes to her
father and speaks to him. 48. And his mother . . . goes to the girls mother,
and speaks to her. If the father of the girl says she is not engaged . . . and
we will marry her into your family . . . (then) theres more talk. 49. If the
father of the girl says Shes engaged . . . (then) there is a talk. If the girls
father says Her cousin also wants (to marry) her, (then) theres more
talk. 50. And there are people who see her, of their neighbours . . . one
of his neighbours that he has sees a girl that he likes . . . 51. hell (then)
speak to his mother. Hell say I want (to marry) so-and-sos daughter.
52. She goes to her mother and speaks to her and asks her and she says
the girl is engaged or she is still an unbetrothed virgin or thus or thus...
53. After that his father speaks to her . . . he goes to the man . . . the girls
father, and he speaks to him. 54. He speaks to him. If he says that hell
marry the girl you, she is not engaged, hell say to you okay. 55. If the
girl is young, he says to you wait (until) after one year, wait (until)
after five months, wait (until) after two years. 56. He then slaughters
for her. He gets his sheep, if he says to him wait (until) after one year,
two years . . . the girl is (still too) young. 57. He then brings his sheep
and slaughters it. When he has slaughtered it, then they have betrothed
her.68 58. From the moment they have given him his twig, thats it, he
knows that this (girl) is (going to be) his wife, in the tradition of God
and His Prophet69 they have given him . . . his twig.
68
ath gasalah, lit. they have given him a twig. The twig is given to the groom in
betrothal ceremonies, who is then mxid gasalatha holding her twig, i.e. she has been
betrothed to him, see Bailey (1974).
69
The phrase b sinnt Al l h w rasluh in the tradition of God and His Prophet may
often be heard in additionused almost as an excuse in pre-emptionto the descrip-
tion of traditions, of which the islamic origin is doubtful (or non-existent).
586 rudolf de jong
8.6.1 Transcript
59. iw h l . . . gm xallha w ugub sanatn . . . zabbat70 lh atab, iw jb
idda byih , iw jb ibyt aar. 60. iw azam anns, iw gl lhuw lfarah
ind . . in All h jjimah jjyah jjimah lliy drha, iw banna lbyt . . . ym
alxims . . . assu bih . 61. iw sawwa. . . . atta h wiy. iw juw anns d kull
wh id ib h amltuh.71 kull wh id ib tuh . . . kull wh id ib h amltuh
w h rumtuh muh. 62. ym alfrah ind ifln [linna] imrawwh n
nah da ruh, kull wh id ib h amltuh. iw rawwah uw, iw ugub salt
alasir kamn . . . dabih . 63. dabah iw saww-laa. iw fh ns bitsaww-
alfitr . . alfarh 72 . . . arrjl w alih ra yyim biysawwin. 64. iw gassam . . . kurr
rijjl 73 firh ah iw h iggtuh lh amatuh fha . . . gassam alhuw . . . kullhuw.
65. iw ugm a74 bat iam baadn . . . ugub salt i . . . h awalha . . .
gm at iddah h iyyih.75 66. fh ns bitsawwiy dah h iyyih, iw fh ns bissaw-
wiy marbah.76 67. almarbah lliy h arrafh iy.77 w iddah h iyyah-lliy
h . . . alxumus. iddah h iyyah kull wh id biygl min induh . . .
8.6.2 Translation
59. And then the situation is . . . he leaves her be,78 and after two
years . . . he arranges the firewood, and brings animals for slaughter, and
brings tents. 60 And he invites people, and says to them the wedding
party is at my house, God willing, coming Friday, Friday next week
and he has set up the tents on Thursday . . . in the morning . . . 61. And
70
Although z as a phoneme is rare (and marginal as such) in Sinai bedouin dialects
(the current reflex for both *d and *d is interdental emphatic d), it is regular in lexemes
(loaned from MSA or Cairene Arabic) like zabbat , yzabbit arrange; do properly (Stew-
art (1990, 285286, glossary) transcribes z), n(i)zm system and z(u)rf circumstances,
see De Jong (2000, 60).
71
Although h amlah is listed by Blanc (1970, 114) in the meaning of clan, in dialects
of other tribes it means animal led to a festive occasion to be slaughtered as a present
for the host.
72
farh , sg. firh ah large flat (and thin) round bread baked on the sj.
73
kurr rijjl: assimilation of kull rijjl every man.
74
ugm a: assimilated ugb + ma.
75
dah iyyah a type of bedouin dance, see Bailey (1991, 436, glossary). During
dah iyyah, men stand in a line, while women dance in a line opposite to them.
76
In a marbah older men stand in a square facing each other.
77
It was later explained to me that the rafh iy is a dance unlike marbah: during
rafh iy, in which younger people take part, the boys will sing their rhyme, while the
girls dance in front of them (one at a time, often while holding a stick), which is much
more in line with the meaning associated with the root r-f-h , greet (new spouses) in
Hava (1982).
78
The verbal perfect forms used by the speaker have here been translated into imper-
fect forms.
he has prepared the food (to be served before the wedding ceremony),
and the people have come, everyone with his slaughter animal, everyone
with his sheep. Everyone has his slaughter animal and his wife with him.
62. On the day of the wedding ceremony with so-and-so we go over
(to him) and bring him over. Everyone (goes) with his slaughter ani-
mal and goes, and also after afternoon prayer . . . (there is) slaughtering.
63. He slaughters and prepares dinner. And there are people who make
flat (unleavened) bread . . . farh . . . the men and women make them.
64. And he distributes . . . (to) every man a flat bread and his portion in
which there is his meat. . . . he distributes it to them . . . all of them. 65.
And after the sun has set, after that . . . after the evening prayer . . . round
about it (i.e. that time) . . . the dancing starts. 66. There are people who
do a dah iyyah, and there are people who do marbah. 67. The marbah
is the same as the rafh iy. And the dah iyyya is . . . the clapping of hands.
(during) The dah iyya everyone says his own lines (of poetry) . . .
8.7.1 Transcript
68. hda xams t-infr aarah . . . xamistar nafr biysawwuw, w
arraggsah h urmah wih dih . . . ibturgus alhuw. hda b innisbah l
addah h iyyeh. 69. w almarbah hdiy . . . m . . . biyrawbuw fh ar anns .
. . ijjihhl79 . . . alliy hinna lbant . . . ibyrugsin . . . w ilwild. 70. ijjihhl hdil
biysawwin . . . almarbah. 71. assmir da biysawwh ar aiyyb. anns
alkibir hdil bysumruw. ibysumruw a jl . . . ismuh smir. 72. amma
ddah h iyyah hdiy . . . biysawwha . . . kitr . . . fi Sna nih biysawwha
wjdah ddah iyyah. 73. ym alfar[h awt], iw ym azzawyir, iw ym
ald . . . biysawwuw ddah h iyyah. hda h . . .
8.7.2 Translation
68. This, five, ten . . . fifteen persons do this, and the dancer is one
woman . . . who dances before them. This as far as the dahiyyah is con-
cerned. 69. And this marbah . . . only young people take part in the
marbah . . . which is the girls . . . who dance . . . and the boys. 70. These
youngsters take part . . . in the marbah. 71. In the smir only older peo-
ple take part. These old people take part in the smir. They do a smir
on the fringe (of the main festivities) . . . it is called smir. 72. As far as
this dahiyyah is concerned . . . they do it . . . a lot . . . Here in Sinai they do
it a lot . . . the dahiyyah. 73. When there are weddings, and when there
79
jhil lit. ignorant is commonly used in Sinai for young or child.
588 rudolf de jong
are visits (to sheikhs tombs), and when there is the feast . . . they do a
dahiyyah. Thats it . . .
8.8.1 Transcript
74. (R) w alars . . . w al . . . (S) ilurs fi lbt (R) fi lbt . . . 75. (S)
ywah . . . alurs . . . itsawwiy zzaffih humma , iw jbha b ilarabiyyt iw
waddha . . . albt. 76. alurs hadiy btugud saba t-iyym, iw h fi lbt.
77. m btatla minnuh wala biyhiddh bt alfrah . . . mabniy . . . 78. sabi
t-iyym mn ajjimeh . . . l ajjimeh. 79. iw h gdih80 lars . . . batnuh . . . iw
irssa indaha. l(a:) jjimah jjyih. 80. ajjimah jjyih ssawwiy sib,
dabah luh h tn . . . 81. kamn h idi r alliy yiz yah da r,81 w alliy
mh yiz mh yiz . . . 82. iw ugub kidiy diy[t] waddha druh.
aza druh . . . hlih . . . waddha-yyha. mh hleh druh, iw h l . . . 83.
bitrawwih ind ummh a. ibtasrah ib anamha. 84. lam biysawwiy druh,
iw h l . . . iw hdiy h rumtuh induh. 85. hda b innisbah l ijjawzeh zayy
kidiy.
8.8.2 Translation
74. (R) And the bride and the groom. (S) And the bride is at home (R)
At home . . . 75. (S) Yes . . . the bride, they organise a wedding procession.
And they have brought her with cars and brought her over to the house
(tent). 76. The bride spends seven days (while she is) in the tent. 77.
She does not go out of it (i.e. the tent), nor do they take it down the
wedding tent . . . stands (lit. is built) 78. For seven days from Friday till
Friday. 79. And she sits inside it, the bride . . . while her groom is with
her. (all the time) until the next Friday. 80. The next Friday you celebrate
the weeks feast, he slaughters one or two sheep (for himself) . . . 81. And
also whoever wants to attend is present, and whoever does not want,
does not want . . . 82. And after that he then takes her to his house. If
his house is . . . ready, he takes her to it. If it is not ready his house, 83.
then the situation is. . . . she goes to her mother. She takes her cattle out
to graze. . . . 84. When he prepares his house, then the situation is . . . and
this is then his wife with him. 85. This is with reference to the wedding,
like this.
80
in gdah is quite high; when in neutral environmentsin the CCiC pattern
of the active participle tends to be realized near IPA [:].
81
he is present, notice the absence of the gahawah-syndrome here.
yah dar
8.9.1 Transcript
86. (R) bardu h alars ibturud? (S) ha? (R) fh ns biygluw inn alars
ibturud . . . (S) ibturud. iza mh rydtuh . . . ibtuurud innuh. 87. itgl
mn yztuh tgtir, tudxul ibyt bid . . . an um m ha w ubha tgtir
bad . . . 88. itgl mn yztuh. biyjbha luh tniy w biy . . . asil min
ym mh yiztuh turud . . . marra h marrtn taltih xals lzim
iytallighe . . . xals mh yiztuh. 89. hdiy lars. hda mitjaww . . .
mxidha asib . . . hda lzim iytalligha . . . 90. yiztuh, ibtugud imn
awwil l turud wala h jeh. 91. min ym yizt lwalad da xals. hda
lars . . . biysawwiy btuh, aza ind um m uh, aza ind arbnuh, aza ind
nasbtuh . . . taba art. 92. irt ind awwil ma jawwazh. aza gluw
la lzim tuskun indina . . . xals, biysawwiy btuh induhuw.
8.9.2 Translation
86. Does the bride flee as well?82 (S) What? (R) There are people who
say that the bride flees . . . (S) She flees. If she doesnt want him . . . she
flees from him. 87. She says I dont want him, she goes away and
enters houses83 far away . . . from her mother and her father, she goes far
away . . . 88. She says I dont want him. They bring her back to hem again
and he . . . because when she does not want him she flees . . . Once, twice,
three times, and thats it, he has to divorce her . . . Thats it, she doesnt
want him. 89. This is the bride. This one (man) married . . . has taken
her by force . . . this (person) must divorce her . . . 90. If she wants him,
shell sit with him from the start and wont flee or anything. 91. When
she wants the boy, thats it. This groom . . . sets up his house, be it with his
mother, be it with his clan, be it with his kin (of fathers side) . . . accord-
ing to the stipulations. 92. Conditions (that were agreed on) from the
moment they married him (to her). If they say No, she has to live with
us . . . then thats it, he sets up his house with them.
82
With some tribes it was customary for the bride to flee from her groom, see Murray
(1935, 182183).
83
The verb daxal, yudxul here actually means to enter a house to seek refuge (i.e. as
a daxl).
590 rudolf de jong
8.10 Agriculture
8.10.1 Transcript
93. (R) bardu h . . . anns ibyazrauw batt x? (S) ibyazrauw batt x . . . 94.
wagt lmatar fh ns ibyazrauw batt x . . . -lmatar. (R) kf biysaw-
wiy? (S) biysawwiy saddeh . . . imsawwayih min zamn . . . 95. ibyah aritta
-jjimal . . . ym lmatar biytx . . . iw biynaggitha . . . 96. (R) b ilijml, b
iljamal, fh fard? (S) aywah byah aritta . . . hda l algamih . . . 97. (R) iw
fh fard walla bitgluw h? (S) aywah, farid . . . farid . . . (R) iw alha bg?
(S) aywah, marbtah f-ljimal . . . iw msikka . . . arrijjl mn-wara. 98.
hda . . ibyah arit l algamih . . . hda . . . h ittih l algamih , iw xalla h ittih
l al . . . l albatt x. 99. albatt x da . . . nggatuh b ilfs. 100. bizruh fi jam-
buh, iw yah afir ingr a tl b ilfs iw h biynaggit. 101. albatt x hda
indma tili diy byatla -lmatar. min ym assaddih ruwynih byatla.
ibyunur . . . 102. biytaabbak fi baadu h biysr batt x wjid . . . bass
anniswn m tjh wala h add . . . 103. min ym ibybudruh biygtruw
annuh [ba] a jl. ibynkibis 104. in nkabas bz84 albatt x. hda . . . l
albatt x. 105. w algamh hda . . . ibyatla ym rabbna biyrd. 106. algamh
hda lliy binsawwiy minnuh l . . . addagg w alh l da, hda h. w attibin
. . . alliy h biyjbh l ajjml . . .
8.10.2 Translation
93. (R) Do people also grow watermellons? (S) They grow watermel-
lons . . . 94. At times of rain, there are people who grow watermel-
lons . . . on rain (water). (R) How does he do (it)? (S) He makes a dam . . . It
was made long time ago . . . 95. He ploughs it with the (plough drawn
by a) camel . . . When the rain falls . . . and he sows it. . . . 96. (R) With
camels, with the camel, is there a plough? (S) Yes, he ploughs it . . . this
wheat . . . 97. (R) And is there a plough, or what do you call it? (S) Yes, a
plough . . . (R) And with a (large trumpet-shaped) funnel on it? (S) Yes,
tied to the camel . . . and holding it . . . is the man at the back end. 98. This
(man) ploughs for (sowing) the wheat, a piece (of land) for the wheat,
and he leaves a piece (of land) for the . . . for the watermellons. 99. These
watermellons, he sows them with the hoe. 100. He keeps his seed by his
side, and he digs holes with the hoe while he sows.85 101. These water-
84
bz, ybz go bad is a loan from Egyptian; therefore z.
85
Lit. let drip, here in the meaning of dropping one or two seeds in a hole at a time.
mellons, when it comes up normally, it grows on rain (water). When the

dam86 is well-watered it comes up and spreads out 102. It becomes inter-
twined, and the watermellons become plentiful. But the women dont
come there . . . not one. 103. From the moment he sows it, they go away
from it (and stay) on the side. (if not) It is jinxed. 104. If it is jinxed the
watermellons will go bad. This . . . is with reference to the watermellons.
105. And this wheat . . . comes up when our Lord wills it. 106. This wheat
is what we make flour from and this situation, thats how it is. And the
straw . . . which they take (as fodder) to the camels . . .
8.11.1 Transcript
107. (R) bardu h indukuw rh iy? (S) ha? (R) walla bitgluw , rh iy walla
rh h? (S) arrh h . . . [. . .] arrh h lliy h btath an algamih . . . hda-smha87
rh h h ajr . . . 108. (R) aywah. w alh urmah alliy btath an? (S) aywah
lh urmah btath an, iw fh irjl ibyath anuw. 109. (R) aywah, biysawwuw
? (S) biysawwuw fh ns ibyath anuw . . . aldra h. aldra h hdiy biysaw-
wuw minha . . . biysawwuw minha . . . libbih. 110. (R) ismuh ? ildra h?
(S) aldra h88 hdiy biysawwuw minha libbih. allibbah hdiy biysawwha
mn aldra h . . . ibyath anha a rh h.
8.11.2 Translation
107. (R) Do you also have hand mills? (S) What? (R) Or what do you
say? rh iy or rh h? (S) hand mill . . . The hand mill which grinds the
wheat . . . this is called the stone hand mill . . . 108. (R) Yes. And it is the
woman who does the grinding? (S) Yes, the woman does the grinding,
and there are men who grind (as well). 109. (R) Yes, what do they do?
(S) They do, there are people who grind . . . the sorghum. This sorghum
from which they make . . . they make libbah89 from it. 110. (R) What is
it called? ildra h? (S) From this corn they make the libbah. This libbah
they make it form the corn . . . they grind it on a hand mill.
86
Interestingly, saddah here means the area covered by water, rather than the bar-
rier blocking the water or dam itself.
87
ismha (with Cairene type of stress) instead of simha.
88
In group I dialects the form is more typically dra () (or with velarization marked in
the interdental, dr a()) and with article ddr a , see De Jong (2000, 82). For a remark on d
as a reflex for *d, see De Jong (2000, 332333).
89
Stewart (1990, 245, glossary) lists libbih as a small round of bread. A libbah, often
in other dialects referred to as gurs, is what men typically prepare themselves when they
are travelling, and women are not present. It is baked in glowing embers in clean sand.
592 rudolf de jong
8.12.1 Transcript
111. an l makn wala h jah byath an . . . w jjimal kamn, aza induh
jiml walla . . . jamaln walla burn, 112. aza m f h urmah fdy ih...
int ibtath an . . . biddi90 . . . biddi. biddi . . . yaniy aldra h . . . itxallha
mi dagg . . . itxallha majrih jari. 113. hda l alburn . . . l azzamil . . .
itain.91 hda kull ym . . . lzim tujuru lhin . . . imn aldra h diy. 114.
algamih hda . . . hda byath aninnuh duriy a rh h . . . dagg . . . dagg hda
zuwwdeh. 115. itdibb almazwad . . . alh l hda . . . addagg imn algamih .
aldra h hdiy . . . l alburn biyjirha. 116. [kull ym] hda lh l ulitta
btath an algamih , w ibtath an aldra h, iw kulluh. wi . . . w attibin hda . . . l
ijjml a tl. m biyath anh walla h jih bass ibyidirsh daris . . . ibydir-
suw lgamih minnuh . . . iw biywaddh a lj[l] . . . 117. (R) iw byidirsh b
h? fh lh ah kidiy walla . . . (S) aywah . . . (R) ismuh h? (S) fh ns ibydir-
suh bi l . . . b jjimal . . . biysawwiy luh madrs, ibydirsuh. 118. iw fh ns
halh n ibyidirsh b lmakan. (R) rif kilmit ilhjal?92 (S) alhjal . . . (R)
bitgl yaniy alhjal? (S) aywah . . . [. . .] alhjal hda biysawwuw minnuh,
ibydirsuw minnuh b jjimal. iw ar insn rif iysawwh . . .
8.12.2 Translation
111. Because there were no machines or anything, he grinds . . . and also
the camels, if he (a man) has a camel or . . . two camels or (more) cam-
els. 112. If there is not a woman free (to do it) for them . . . you grind . . .,
you grind coarsely, you grind coarsely. You grind coarsely . . . that is, the
sorghum . . . you do not turn it into flour . . . but you turn it into coarsely
ground corn. 113. This is for the camels . . . for the camels . . . to feed them
(at dinner time). This (you do) every day . . . you have to coarsely grind
(some) of this corn for them (i.e. the camels). 114. This wheat . . . this they
(f.) grind normally on the hand mill . . . flour . . . flour, this is (for) provi-
sions. 115. You fill the sack . . . (in) this case . . . with flour from wheat.
This corn is for the camels, they grind it coarsely . . . 116. [Every day] This
90
ddi, assimilated dd < td in tdi; da, ydi grind corn coarsely, see Stewart (1990,
211, glossary).
91
tain: an apocopated 2nd. p. m. sg. form (ta) of measure 2 verb aa, yaiy
feed dinner, with initial h- of the suffix -hin (camels are usually referred to in the f. pl.)
assimilated to the preceding voiceless consonant. The spelling here with triple is for
reasons of morphological transparency. The pronunciation is, however, not audibly dif-
ferent from doubled (IPA [( :)]).
92
hjal: a threshing board with sharp stones in its underside, on which the man
stands while it is being pulled by a camel.
is how it goes (lit. its situation), you grind the wheat, and you grind the
corn, and everything. And this straw . . . goes (lit. is) straight to the camels.
They dont grind it or anything, they only thresh it properly. They thresh
the wheat from it . . . and they put it aside . . . 117. (R) And with what do
they thresh? Is there like a (wooden) board, or . . . (S) Yes . . . (R) What is
it callled? (S) There are people who thresh it with . . . with camels. . . . hell
make (for himself) a threshing floor. 118. And there are people now-
adays who thresh it with machines. (R) Do you know the word hjal? (S)
the hjal. . . . (R) Do you call it a hjal? (S) Yes . . . [. . .] This hjal (people)
do with it, they thresh with it with camels. And one has to know (how)
to do it . . .
CLASSICAL AND COLLOQUIAL ARABIC ARCHAISMS
Alan S. Kaye
California State University, Fullerton*
1. Introduction
This essay1 argues that modern spoken Arabic dialects sometimes

retain very archaic Semitic features. In fact, they may even preserve
Proto-Semitic forms that have been lost in Classical Arabicanother
indication that Classical Arabic is not to be regarded as their ancestral
proto-language (see Kaye 1976 for fuller explication of this idea). Rather,
there were other Arabic dialects spoken alongside Classical Arabic all
throughout history that served as the ancestral inputting ones to the
contemporary picture of Arabic dialects.
The orthographic representation of Hebrew l no as <l>2 may best be
interpreted as evidence for the reconstruction of Proto-Semitic *la no.
No matter what vowel preceded the glottal stop, I do not think any Semi-
tist can deny that the Proto-Semitic word ended in a glottal stop.3 Final
* Shortly after Alans untimely death we received the proofs of his article. In deep
respect we insert his paper in this volume as it was at his passing on in appreciation for
what he contributed to our common field of interest (the editors).
1
This was originally presented to the joint session of the 216th meeting of the Ameri-
can Oriental Society and the 34th annual North American Conference on Afroasiatic
Linguistics on March 17th, 2006, in Seattle. I am thankful for the stimulating discussion
by the audience participants. I also wish to express my gratitude to Vit Bubenik, Gideon
Goldenberg, Wolfhart Heinrichs, Bob Hoberman, Jonathan Owens, Adrian Mcelaru,
Gary A. Rendsburg, Judith Rosenhouse, Aaron D. Rubin, Avi Shivtiel, Laurence J. (Taw-
fiq) Surfas, Rainer Voigt, and Bill Young for useful comments on a preliminary version.
2
According to Koehler and Baumgartner (1958, 466) and (1998, 466), this spelling
occurs 466 times in the Old Testament, while the plene spelling <lw> occurs only 35
times. Gary A. Rendsburg [p. c.] notes that the plene spelling occurs especially in Jer-
emiah, and that there are more instances of the interrogative and presentative /hall/
spelled plene than there are defectiva.
3
Let me comment on the notion of a pho(a)ethematic glottal stop, which occurs
in some Arabic dialects in final position; e.g., /mia/ he went. These are clearly inno-
vations within Arabic (noted by the medieval Arab grammarians as well) and are pa-
ralleled by similar occurrences in Neo-Aramaic dialects (thanks to Bob Hoberman,
p. c.). Neo-Aramaic /la/h/ is paralleled by Israeli Hebrew /lo/ and, in many ways, by the
English voiceless unreleased bilabial stop in yep and nope.
596 alan s. kaye
graphemic aleph in Biblical Hebrew4 /q r / he read presents evidence

for an earlier */q r / or perhaps an even earlier */q ra/ < */qara/
(cf. /q r they read), which is equivalent to Classical Arabic /qaraa/.5
Similarly, the Hebrew form <l> points to an earlier *l < earlier */la/6
= colloquial Arabic la, la, or laa (Hinds and Badawi 1986, 775). The
Biblical Hebrew variant spellings <lh> (Koehler and Baumgartner 1998,
474) and <lw> (Koehler and Baumgartner 1998, 475) prove that the
glottal stop was no longer pronounced. Still, the ubiquitous spelling with
the glottal stop must be accounted for. Moroccan Arabic (so too Alge-
rian Arabic) even has lla no and lla7 no way!; absolutely not! (Sobel-
man and Harrell 1963, 130 for the former alternant) demonstrating that
4
Hebrew refers to Biblical Hebrew unless otherwise noted.
5
The problem of stress is not a relevant consideration here and its discussion is thus
left unmarked. Let me take up the matter of the Hebrew final graphemic glottal stop.
Could this merely be an orthographic convention to distinguish two common words:
no, not and to him, originally spelled <lh> (= Classical Arabic <lh>), making it homo-
graphic to to her, and later spelled <lw>? The answer is clearly in the negative, since the
Proto-Semitic form contained a glottal stop.
6
Cf. Proto-Semitic ra head > Classical Arabic ras colloquial Arabic rs =
Hebrew r and Proto-Semitic *kas = colloquial Arabic ks = Hebrew ks (spelled with
aleph), although Gideon Goldenberg remarks that Anton Spitaler thought the glottal
stop in the last example was a hypercorrectiona theory that I find farfetched. Bob
Hoberman [p. c.] also notes the Hebrew verbs ymar he will say and ykal he will eat,
etc., that are spelled with aleph but where its phonetic realization as a glottal stop disap-
peared. Let me hasten to add that when it comes to a glottal stop, it is important to point
out that Hebrew /yir/ he will see retains it, whereas Classical Arabic /yar/ he will
see does not. Cf. Israeli Hebrew /lo/ absolutely not!. Although it is conceivable that the
glottal stop might go back to Proto-Hebrew times, it is much more likely that the latter
form is evidence of history repeating itself. Although the final glottal stop in this word
may originally be due to sound symbolism (cf. English yep and nope that are marked for
informal register, corresponding to more formal yes and no), the fact remains that it is
Proto-Semitic. Proto-Hebrew glottal stop in final position disappeared only to surface
again in Israeli Hebrew (see below) as a result of a linguistic cycle to give the root struc-
ture more than a mono-consonantal baseas a kind of Systemzwang. This is parallel to
Modern English ask < Old English aks, which is also once again the form (aks) in several
modern English dialects.
Concerning the Israeli Hebrew development, Judith Rosenhouse [p. c.] notes that it
seems to be a recent phenomenon and appears to be restricted to young adults, espe-
cially females, and seems to indicate abruptness or vehemence, particularly as an answer
to a yes-no question in order to stop someones nagging. When I suggested that this
might be a result of colloquial Arabic influence on Israeli Hebrew, she replied that this
was not possible, since it was used by native speakers of Hebrew who do not have direct
contact with Arabic speakers. Still, I pose the following question in rebuttal: does one
have to have direct contact to have influence? Bob Hoberman [p. c.] notes that the final
glottal stop in Israeli Hebrew is older than Rosenhouse thinks and it is not restricted to
just young adults and females, and is not mainly confined to nagging. Arabic full
investigation of this is, in my view, a desideratum.
7
The form with the geminated l actually gives a tri-consonantal body to this root: ll.
classical and colloquial arabic archaisms 597
Arabic dialects do not like to tolerate a mono-consonantal root. The first

cited Moroccan form at least points to a bi-consonantal or geminated
root with double /l/, although such roots are also not characteristic of
Semitic languages in that C1C1C2 roots are unknown.
Before continuing the discussion of the importance of the aleph in
the Hebrew word for no, not, it is necessary to point out that Classical
Arabic, although usually very conservative in matters of phonology, is
sometimes innovative. I will now discuss two significant innovations.
First, many contemporary spoken Arabic dialects preserve the Proto-
Semitic imperfect vowel preformative (the so-called taltalah) */i/ rather
than */a/ = Classical Arabic /a/. Colloquial Arabic /yifqid/ corresponds
do Akkadian /ipqid/, Aramaic /nepqud/, Hebrew /yipqod/, Geez /
yfqad/, and Classical Arabic /yafqid/ (Proto-Semitic fqd = to seek).
Most Semitic languages and most modern Arabic dialects have /i/ or
a vowel clearly originating from it, as opposed to Classical Arabic /a/.
Although Moscati et al. (1964) postulate the vowel */a/ as Proto-Semitic,
they also remark (1964, 143): Some scholars, however, regard the vowel
i of the prefix as primary, alongside a. I believe there is much more evi-
dence in favor of Proto-Semitic */i/ than */a/.
Second, the Classical Arabic voiced alveo-palatal affricate ( jm) is
clearly secondary deriving from Proto-Semitic */g/. Although I recon-
struct a voiced lamino-palatal fricative *// for Proto-Colloquial Arabic,
the entire story of the development in Arabic dialects is not germane to
this investigation (see Kaye 1972 for the details).
Looking at the cognate Semitic languages, Ugaritic l (variant spell-
ing l,8 syllabically spelled la-a indicating l) does not support the col-
loquial Arabic archaic forms with glottal stop, but matches up exactly
with Akkadian l and Classical Arabic l. Aramaic, Old South Arabian
and Geez lend further support to the earlier glottal stop with their gra-
phemic alephs, although the Geez all is problematic from a compara-
tive Semitic point of view. (Hebrew does have a negative al for negative
imperatives with cognates in Ugaritic, Phoenician, Biblical Aramaic,
Geez, and possibly Akkadian ul ).9 Thus, any notion that colloquial Ara-
bic la() might be secondary is, I believe, an erroneous perspective due
8
The Ugaritic spelling with a glottal stop is erroneous, although Koehler and Baum-
gartner (1998, 466) cite it. Gideon Goldenberg [p. c.] informs me that the spelling with
aleph was originally postulated by G.R. Driver; however, others thought that this was a
verb meaning to be weakened or gleaming.
9
See Koehler and Baumgartner (1998, 45).
598 alan s. kaye
to the graphemic evidence from the ancient Semitic languages. Rather,

Classical Arabic l (which might represent /la/ since the grapheme for
the hamza is latesee further below) should be viewed as the later devel-
opment < Proto-Semitic */la/. When the final glottal stop disappeared in
this Classical Arabic word, the previous vowel was lengthened as a result
of compensatory lengthening. It should be noted that this form without
the final glottal stop breaks with root bi-consonantalism or tri-conso-
nantalism in favor of mono-consonantalisma very rare, if not impos-
sible, situation in the classical and modern Semitic languages, especially
for words (prepositions and other function words, such as b- in or l-
to; for, are mono-consonantal, although the latter mono-consonantal
form is certainly related to the bi-consonantal root l in Hebrew and the
tri-consonantal ly in Classical Arabic (Koehler and Baumgartner 1998,
48).10 Moreover, it is not impossible that <l> in the very early stages of
Classical Arabic could have represented /la/, but that the hamza was not
written on top of the alif.11 This spelling represents the scripta defectiva,
much in the way the hamza of Modern Standard Arabic is often omit-
ted in /uktub/ write! (masc. sg.) as it should be in /waktub/ and write!
(masc. sg.). This hamzatu lwasl should, strictly speaking, be written in
word-initial position, at least in some varieties of Modern Standard
Arabic.12 I can only speculate that the Akkadian graphemes might also
represent an earlier pronunciation in a pre-Akkadian dialect or in early
Akkadian with a final glottal stop before the general loss (in East Semitic
as a whole) of the pharyngeals and laryngeals.
Since one is confronted with the fact that Proto-Hebrew *-a > (as
in /q r / he read discussed above), one may be forced to hypothesize
that the Proto-Semitic word for no has a long , thus yielding a long
via the Canaanite vowel shift. It is important to point out that *qara
he read (previously discussed) does not yield qar or the like; thus the
inputting original forms had to be different because of the very different
10
There are other prepositions in the Semitic languages which clearly have bi-con-
sonantal roots, such as Hebrew m- from < *min (Koehler and Baumgartner 1998, 535)
usually assimilated to the former form, and Hebrew al on < ly.
11
Jonathan Owens notes [p. c.] that the hamza was a post-hoc addition to Arabic
orthography, which is why the orthographic rules relating to it are so complicated.
12
Gideon Goldenberg [p. c.] comments that in early Modern Standard Arabic (e.g.,
Blq Press, Cairo), fully vocalized Modern Standard Arabic was printed with the hamza
sign on the alif only when the glottal stop is phonemic, such as in urdu I want. He
maintains: The rule to write the hamza where the glottal stop is a positional alternant
was adopted by the Egyptian Ministry of Education, then became rather popular.
outputs of the development. Bob Hoberman [p. c.] suggests that /a/ >
// was only in closed syllables where the // was the syllable coda. The
relative chronologies of the various sound changes are also important
factors to consider affecting vowel quantity.
2 Biblical Hebrew13 and Comparative Semitic Evidence for Proto-Hebrew 14

*// in Final Position Proving Proto-Hebrew */l/ < Proto-Semitic */la/
no
1. /b /15 enter; come

Classical Arabic /ba/, Akkadian /bu/, Ugaritic */bw/,
Phoenician /b/, Old South Arabian /bh/ (Koehler and
Baumgartner 111)
2. /b r / create
Old South Arabian /br/ build (Koehler and Baumgartner
14647)
Classical Arabic /baraa/ is not in Koehler and Baumgart-
ner (1998).
Wehr (1974, 49) lists it (create).
3. /gb/ ditch
= Akkadian /gubbu/ < */gubu/ (Koehler and Baumgart-
ner 163)
4. dr /der n/ abhorrence
= Classical Arabic /daraa/ repel (evil) (Koehler and
Baumgartner 217)
5. /d/ grass
Akkadian /diu/ spring, Biblical Aramaic /dit/, Syriac
/t/ (according to Carl Brockelmann via Rainer Voigt
[p. c.], /te/ ~ /ta/), Classical Arabic /adiya/ be moist,
Old South Arabian /d/ season of grass; spring (Koehler
and Baumgartner 220)
13
The Biblical Hebrew data in Section 2.0 of this essay have been taken from Koehler
and Baumgartner (1998), an updated and revised version of Koehler and Baumgartner
(1958). These 40 roots offer the evidence from other Semitic languages of roots ending
in aleph, corresponding to Proto-Semitic */la/ no, not.
14
It makes little difference whether this happened in Proto-Hebrew or a pre-Hebrew
dialect.
15
It is important to point out that the aleph is pronounced in various conjugated
forms of tertiae-aleph roots; e.g., /b / they entered, /y b/ they will enter.
600 alan s. kaye
6. /h/ he
Classical Arabic /huwa/ but Bedouin Arabic /h/. Cf.
the // in Geez /wt/ (Koehler and Baumgartner
226). /h/
she works the same as /h/. Cf. Geez /yt/. Koeh-
ler and Baumgartner (1998) do not cite Bedouin Arabic
*/h/, but it probably exists (Koehler and Baumgart-
ner 226). Koehler and Baumgartner (226) reconstruct
Proto-Ethiopian Semitic */hatu/ he and */hati/
she. Gideon Goldenberg [p. c.] notes that Qumranic
Hebrew /ha/ he and /ha/ she parallel the Classi-
cal Arabic forms with /-a/.
7. hb /nhb / hide oneself
Classical Arabic /xabaa/ hide (Koehler and Baum-
gartner 270)
8. /h g / shame; confusion
Classical Arabic /xajaa/ look ashamed (Koehler and
Baumgartner 275)
9. /h t / miss the mark
Classical Arabic /xati a/ make a mistake; sin Wehr
(1974, 245)
Koehler and Baumgartner (1998) cite Classical Arabic
/xati ya/. Ugaritic /xt /, Biblical Aramaic /xt /, Old
South Arabian /xt/, Geez xt (Koehler and Baum-
gartner 288)
10. /h l / fall ill
Classical Arabic /xalaa/, Old South Arabian /hl/ fall
ill (Koehler and Baumgartner 298). Gary A. Rends-
burg [p. c.] notes that this root is a by-form of /hly/
become weak (Koehler and Baumgartner 300) =
Middle Hebrew and Jewish Aramaic /hl/.
11. /hm (h)/ sweat; butter
Ugaritic /xmt/; Jewish Aramaic /xm ()/ (Koeh-
ler and Baumgartner 308)
12. /y re/ fear
Ugaritic /yr/ (Koehler and Baumgartner 399); Avi
Shivtiel [p. c.] adds Classical Arabic waraa to repel,
listed in Lane Vol. 8 (1893, 2933).
13. /k l / restrain; keep from
Classical Arabic /kalaa/, Egyptian Aramaic, Jewish Ara-

maic, Syriac, Geez kl (Koehler and Baumgartner
436)
14. /kisse/ seat
Akkadian /kuss/, Ugaritic /ks/, Phoenician /ks/,
Aramaic /k rse/, Syriac /kursy/ with plene written
/u/, Classical Arabic /kurs/ < */kursi/ (Koehler and
Baumgartner 446)
15. /ks/ headdress of the moongod at the time of the full moon;
full moon
Akkadian /kusu/, Phoenician /ks/, Syriac /ks/ ~
/kas/ (Koehler and Baumgartner 446)
16. /l bo/ East of Jordan (Gary A. Rendsburg [p. c.] translates
entrance; on the way to.)
Assyrian /labu/, Egyptian /rbw/ (Koehler and Baum-
gartner 470)
17. /y s / go out
Phoenician, Ugaritic /ys/, Old South Arabian /wz/,
Geez ws ~ wd (Koehler and Baumgartner 393)
18. /l b/ lion
Akkadian */labu/ (Koehler and Baumgartner 472 state
that they follow Benno Landsberger), Ugaritic /lbt/, Old
South Arabian /lb(t)/. Avi Shivtiel [p. c.] adds Classical
Arabic /labwa/ lioness, not in Koehler and Baumgartner,
but listed in Wehr (1974, 857). Vit Bubenik [p. c.] alerts
me to the alternant /labua/ lioness (Madina 1973, 591).
The Proto-Semitic root is lb. The form with the semi-
vowel is secondary.
19. /m le/ be full
Classical Arabic /malaa/ fill; /malia/ be full, Old Ara-
maic, Biblical Aramaic, Ugaritic /ml/, Akkadian /mal/,
(Koehler and Baumgartner 523); Geez /mala/, not in
Koehler and Baumgartner, added courtesy of Aaron D.
Rubin, [p. c.].
20. /m s / meet; find
Classical Arabic /ant/ < /amt/, Jewish Aramaic /ms /,
Biblical Aramaic /mt /, Syriac /mt/ ~ /mtay/, Old
South Arabian /mz/, Geez ms (Koehler and Baum-
gartner 553)
602 alan s. kaye
21. /m r / feed
Classical Arabic /maria/ agree with (of food), Ugaritic /
mr/ (Koehler and Baumgartner 563)
22. /mass / Masu
= /masu/ name of a North Arabian tribe (Koehler and
Baumgartner 570)
23. /n / half done
Classical Arabic /na/ be raw, uncooked (especially of
meat) (Wehr 1974, 1014) (Koehler and Baumgartner 570)
24. nb denominative of /n / prophet
Classical Arabic /nuba/ prophecy, /tanabbaa/ (Form
V) to prophecy (Wehr 1974, 937) (Koehler and Baum-
gartner 586)
25. nd drive (cattle); detach, remove from
Geez nd drive cattle
26. /n / lift up
Classical Arabic /naaa/, Phoenician, Moabite, Ugaritic
/n/, Old South Arabian /n/, Geez ns (Koeh-
ler and Baumgartner 638) Biblical Hebrew /nm/
clouds; damp; fog, Classical Arabic /na/ hovering
clouds (Koehler and Baumgartner 638)
27. /n / lay claim; lend
Middle Hebrew ny, Classical Arabic /nasaa/ grant
credit (Wehr 1974, 959) (Koehler and Baumgartner
638)
28. /s l (h)/ participle /msull m/ be paid pointing to sl
Classical Arabic /salaa/ pay promptly (Koehler and
Baumgartner 658)
29. /pl/ miracle
Classical Arabic /fal/ good omen (Koehler and Baum-
gartner 759)
30. /pr/ zebra16
Classical Araic /fara/ wild ass; onager (Wehr 1974, 701).
The Akkadian cognate is purmu. (Koehler and Baum-
gartner 775). Vit Bubenik [p. c.] states that Akkadian
par mule is closer phonologically than is purmu.
16
Although Koehler and Baumgartner (1998, 775) translate zebra, this is erroneous
for onager (thanks to Wolfhart Heinrichs and Gary A. Rendsburg for this correction).
31. /s b / army

Akkadian /sabu/, Old South Arabian /db/, Geez /
sb/. Loanword in Egyptian /dbi/ (Koehler and Baum-
gartner 790)
32. /s me/ be thirsty
Classical Arabic /zamia/ (Wehr 1974, 583)
(Koehler and Baumgartner /zamiya/ seem to be errone-
ous.)
Ugaritic /zm/, Akkadian /samu/, Old South Arabian
/zm/, Geez sm (Koehler and Baumgartner 806)
33. qy /q (h)/, impf. /y q/ vomit
Classical Arabic /qa/, Geez qy vomit, Akkadian
/qu/ [sic] for /qu/ excrete (Egyptian q vomit)
(Koehler and Baumgartner 83637) (also Koehler and
Baumgartner 1958, 83637)
34. qn /qinne/ be envious of
Classical Arabic /qanaa/ become intensely red, /qni/
blood-red, deep-red (Wehr 1974, 791) (Koehler and
Baumgartner 84243)
35. /q r / read
Classical Arabic /qaraa/ read, Ugaritic /qr/ read
Koehler and Baumgartner (851) relate Classical Arabic
read with /qr/ II encounter, befall and not with /qr/
I call, recite > read (Koehler and Baumgartner 849).
In my opinion, this view is erroneous. Their theory is
also expressed in Koehler and Baumgartner (1958,
849). They further note that Biblical Hebrew qry = /
q r (h)/ encounter (Koehler and Baumgartner 1958,
853) = Classical Arabic /qaraa/ to go about (not in
Wehr 1974) and Classical Arabic /qar/ receive as guest
(Wehr /qar, yaqr/ receive hospitably; entertain (1974,
761)
36. /r p / heal
Classical Arabic /rafaa/ mend; darn, Old South Ara-
bian /rf/ repair, Geez rf stitch together (Koehler
and Baumgartner 903)
37. / ne/ hate
Classical Arabic /anaa/ hate (Wehr 1974, 487)
604 alan s. kaye
(Koehler and Baumgartner /aniya/ for Classical Arabic

appears to be erroneous). Ugaritic /n/, Old South Ara-
bian /n/ (Koehler and Baumgartner 925)
38. /b / Sheba
Classical Arabic /saba/, Sheba Akkadian /sabe/ man-
kind (?), Old South Arabian /sb/ Sheba (Koehler and
Baumgartner 939)
39. / w/ a bad thing
Classical Arabic /sa/ be evil; bad (Koehler and Baum-
gartner 951)
40. /t / room
Akkadian /tu/ room; guard chamber (Koehler and
Baumgartner 1015)
3. Conclusion
The basic thesis of this work is to affirm that Classical Arabic should
stop being referred to as the preserver of all, or nearly all, original Proto-
Semitic phenomena. This paper suggests that certain Classical Arabic
forms are indeed secondary, as e.g., /l/ no discussed above. Consider
also Classical Arabic qalb heart < * lbb heart. The latter root is also fully
preserved in Classical Arabic lubb heart = Hebrew leb and Akkadian
libbu < Proto-Semitic *libbu (= Proto Afro-Asiatic */lib-/ ~ */lub-/ (Orel
and Stolbova 1995, 362). Moreover, Egyptian ib supports Proto-Semitic
*libbu, while the vocalism in Classical Arabic lubb can be explained as
the result of regressive labial assimilation. Thus, Classical Arabic q- in
qalb is a remnant of a prefix of some kind or root determinative17 used
in an old Arabic dialect or an earlier Semitic language antedating Clas-
sical Arabic (cf. Classical Arabic qadima or qadama to gnaw; compress
the lips (Wehr 1974, 544 < damma bring together further suggestive of
q- as a root determinative of some kind). Additional comparisons along
the lines of the present investigation of Classical Arabic and colloquial
Arabic dialects with other Semitic languages will undoubtedly further
17
Harsusi /helbb/ and Mehri /hewbb/ have /h-/ before the root /lb/ heart, which is
apparently lost in Soqotri /elbeb/ and Sheri /b/. Mokilko (East Chadic) /ulbo/ heart
looks as though it displays */q/ > //, as occurs in many eastern Arabic dialects.
demonstrate that spoken Arabic dialects occasionally preserve more

original forms than does Classical Arabic.
4. References
Hinds, Martin and El-Said Badawi. 1986. Dictionary of Egyptian Arabic: Arabic-English.
Beirut: Librairie du Liban.
Kaye, Alan S. 1972. Arabic /iim/: Arabic Synchronic and Diachronic Study. Linguistics
79, pp. 3172.
. 1976. Chadian and Sudanese Colloquial Arabic in the Light of Comparative Arabic
Dialectology. The Hague: Mouton and Co.
Koehler, Ludwig and Walter Baumgartner. 1958. Lexicon in Veteris Testamenti Libros.
Leiden: E.J. Brill. (Abbreviated Koehler and Baumgartner 1958.)
. 1998. A Bilingual Dictionary of the Hebrew and Aramaic Old Testament. Leiden:
Brill. (Abbreviated Koehler and Baumgartner 1998.)
Lane, Edward William. 1893. An Arabic-English Lexicon. London: Williams and Nor-
gate. (Reprinted 1968, Beirut: Librairie du Liban.)
Madina, Maan Z. 1973. Arabic-English Dictionary of the Modern Literary Language. New
York: Pocket Books.
Moscati, Sabatino, Anton Spitaler, Edward Ullendorff, Wolfram von Soden, eds. 1964.
An Introduction to the Comparative Grammar of the Semitic Languages (Porta Lin-
guarum Orientalism). Wiesbaden: Otto Harrassowitz.
Orel, Vladimir E. and Olga V. Stolbova. 1995. Hamito-Semitic Etymological Dictionary.
Leiden: E.J. Brill.
Sobelman, Harvey and Richard S. Harrell, eds. 1963. Arabic Dictionary of Moroccan
Arabic. Washington, D.C.: Georgetown University Press.
Wehr, Hans. 1974. A Dictionary of Modern Written Arabic. 3rd ed. by J. Milton Cowan.
Ithaca, N.Y.: Spoken Language Services.
DO THEY SPEAK THE SAME LANGUAGE? LANGUAGE USE
IN JUBA LOCAL COURTS
Catherine Miller
IREMAM-CNRS of Aix en Provence, France
1. Introduction
1.1 Language Description and Language Uses: the unavoidable

discrepency
K. Versteeghs book (1984) on Pidginization and Creolization in Arabic
was one of the first attempts to link two previously separated domains,
Arabic studies and Pidgin/Creole studies. The controversy that followed
publication of his book contributed to diffusion of information about the
main contemporary Arabic based Pidgin-Creole varieties of the South
Sudanese basin, namely Ki-Nubi and Juba Arabic (and former Turku).
No serious Arabicist can now ignore the three above mentioned names,
and a number of publications have been devoted to the description of
these varieties. To recall but a few: B. Heine (1982), X. Luffin (2005),
U. Mahmud (1979), C. Miller (1984, 1992), J. Owens (1991, 1997),
M. Tosco (1993, 1995), R. Watson (1984), I. Wellens (2005), E. Yokwe
(1995).
Most of these studies provide a systemic description of Ki-Nubi or
Juba Arabic, and compare them with other Pidgin-Creole languages,
with other Arabic vernaculars, or with local African languages in order
to establish their specificity and autonomy and to postulate hypothe-
ses about their genesis and development. Although still considered as
the orphan of the orphans (Owens 2001), marginalized in both the
Creole Studies and the Arabic studies, Juba Arabic and Ki-Nubi start
to be rather well described, even if many more need to be done. One of
the impacts of all these studies is the acknowledgement that Ki-Nubi
and Juba-Arabic are indeed autonomous and specific varieties that can-
not be confused with any other Arabic vernaculars. They are identified
through their specific name; as it is well known, to name something is
to provide it with a specific identification.
An interesting development of this linguistic achievement is the fact
that, since Ki-Nubi and Juba Arabic have been identified as specific
608 catherine miller
languages, the local speakers (Nubi people from Uganda or Kenya and
all Sudanese Southerners) are expected to speak what the linguists have
described as Juba Arabic and Ki-Nubi. This is what is happening for
example in the case of language tests applied to asylum seekers in a
number of European countries. If somebody claims to be a Southern
Sudanese but speaks an Arabic variety close to Northern Sudanese col-
loquial varieties instead of the Juba Arabic described by the linguists,
he cannot be a true Southerner. Asked to provide some counter-exper-
tises, I realized that the way of speaking of these doubtful Southerners
was sometimes rather similar to some cases I recorded long ago in some
local courts of Juba, with well-attested Southerners!
In many contexts, most Sudanese Southerners constantly shift from
a more Creole level of Juba Arabic (or basilectal level) to a level more
influenced by Northern Sudanese Colloquial Arabic (mesolectal level).
The coexistence of different levels of Juba Arabic was first highlighted by
U. Mahmud (1979), who applied the Creolistic concepts of continuum
and basilectal and mesolectal varieties to describe the variation found
in the verbal system. I dedicated a number of papers to emphasize the
degree of variation and diversity included under the generic label of
Juba Arabic (Miller 1984, 1987, 1989).
In the continuum approach, each pole (basilectal/mesolectal) is char-
acterized by a number of specific features (isoglosses), while the inter-
mediate levels of the continuum are characterized by different degrees
of occurrence and melting of the various features. It appears, however,
that it is extremely difficult to draw an implicational scale that will be
characterized by a regular acquisition of dialectal features along the
continuum scale. In fact, each speaker tends to have its own way of mix-
ing the various features, some focusing more on phonological features,
others on lexicon or morphological features.
I myself, insisted on the fact that the evolution of Juba Arabic was far
from linear, and was not automatically leading to a process of decreoliza-
tion. Different trends of change and restructuration were simultaneously
recorded in a city like Juba and different influences were operating on
the daily language; one leading towards a rapprochement to Khartoum
Arabic, the other toward what I have called a process of vernaculariza-
tion (Miller 1987).
This diversity and variability of Juba-Arabic raise the following ques-

tions: where are the boundaries of Juba Arabic? On which criteria can
we decide that a person speaks or does not speak Juba Arabic? Are the
language use in juba local courts 609
speakers conscious that they are mixing different features and variables
in their speech? Do the speakers and auditors have the same perception
than an outsider linguist?
1.2 Language Uses in Juba Local Courts

In order to illustrate the diversity prevailing under the generic label of
Juba-Arabic, I choose to present complete samples of interaction, instead
of short extracts given as examples. These samples were recorded in
1981 and 1984 in two local Courts of Juba, the capital-city of Southern
Sudan. Local courts were of particular interest because very different
kinds of people were coming to present and defend their cases. Taking
into account that more than twenty years have passed since the mid
1980s, and that almost nothing is known about the present linguistic
reality of Juba, such samples could be considered as a kind of linguistic
testimony of what speaking Juba Arabic meant in the 1980s.
The term local court and its Arabic equivalent makma ahliyya ~
makama ahliya was applied to Southern Sudanese A and B Courts at the
levels of the village, the district or the city. Various languages were used
in these local Courts.
Outside Juba, the A Court was the villages Court headed by the
villages chief, assisted by two or three elders. The language of communi-
cation was, most of the time, the local African vernacular (Bari, Kakwa,
Zande, Moro, etc.). The B Court was the district Court headed by the
paramount Chief assisted by the village chiefs. It dealt with matters not
solved in the A Court. In multilingual districts of Equatoria, such as Yei
B Court, the languages spoken were local vernaculars (Kakwa, Moro,
Madi, Avokaya), Juba Arabic, Bangala (the local variety of Lingala) and
a few English.
In Juba, the only A court was the Garawiyya, located in the center
of Juba, near the main market, and dealing with low-level criminal cases
(robbery, neighbors quarrels, etc.) involving any person living in Juba,
whatever his ethnic affiliation and period of stay. The court was headed
by an appointed local judge (Ramadan, 49 y. old) and two assistants. The
main language of communication was Arabic (Juba Arabic up to North-
ern Sudanese Arabic (NSA). A few speakers used an African vernacular
and were assisted by a translator. The summary of the case was always
pronounced in Arabic but written in English.
The Kator B court was the Bari local court (the Bari being the main
local group of the Juba-Rejaf area), located in Kator district. It was
dealing with personal status cases involving at least one Bari participant.
The President of Kator B court was a Bari (Stephen, a retired police-
man), assisted by two or three Bari elders. The language of communica-
tion was mainly Juba Arabic, Bari and a few English.
The localization and specialization of each court influenced the
language use. In Juba, the Garawiyya court was far more exposed to
the influence of NSA than the Kator B court. In the Garawiyya, only 4
speakers, not resident in Juba, spoke in their native vernaculars, while
35 speakers spoke only in Arabic (with 16 speaking a more mesolectal
variety and five speaking NSA). In Kator, 10 speakers spoke mainly in
Bari against 24 in Arabic (with none speaking NSA and 14 more or less
mesolectal):
FIGURE 1 LANGUAGE USES IN KATOR AND GARAWIYYA

LOCAL COURTS
Courts Speakers* Non-AR Vernaculars Juba Arabic NSA English
Kator B 34 10 Bari 24
Garawiyya 39 3 Bari 30 5
1 Dinka
Total 73 14 54 5 0
* Number of speakers does not include the Judges
The above figure classifies each speaker according to the use of one
dominant language (English, NSA, JA and Non Arabic Vernaculars). In
fact, the delimitation between each language was not that clear-cut and
there were many cases of language mixing, as will be evident in the Cor-
pus presented in 5. Appendices, the texts: 5.1, 5.2, and 5.3.
2. Presentation and Brief Analyse of the Corpus
The samples presented in the Appendix are taken from three cases, con-
sidered to be representative of three types of speaker. The participants
involved in Case 1, belong to the important Nilotic Dinka-speaking
group, a non-local Juba ethnic group. Not well-educated, they speak
Juba-Arabic as a second language. In Case 2, the participants belong to
a small ethnic group from Western Southern Sudan (Bay from Bahr al
Ghazal Region), are educated, work in the police or the army, and speak
a level of Arabic very much influenced by NSA and WSA (Western
Sudanic Arabic). In Case 3, speakers are all Bari Juba citizens and speak
Bari or a vernacularized Juba Arabic, very much influenced by Bari
features.
In both courts, the procedure was rather similar. The case starts with
an oath (on the Bible, the Qurn, or a spear according to the partici-
pants religion),see (s.9) in Case 1 and (s.14) in Case 2. Then the plain-
tiff tells his/her story; after some questions, the judge summarizes the
story and asks the defendant to present his/her own version. Additional
questions follow; sometimes, additional testimonies are required. After
listening to all parties, the Court deliberates and announces its verdict.
While most cases took between 30 to 40 minutes in Juba, they could
take hours in the villages. Cases tended to be shorter in the Garawiyya
than in Kator.
Speakers are presented as follow: J.= Judge S. = Courts Secretary. T. =

Translator, P. = Plaintiff, D. = Defendant and W. = Witness. Elements
in italic are considered to be average basilectal Juba Arabic. Elements
in bold are considered to be more mesolectal. Underlined elements
are borrowings from the local languages, mainly Bari. Plain words are
English borrowings. Many elements are common to both basilectal and
mesolectal Juba Arabic. They have been put in both italic and bold, but
the distinction between more basilectal or more mesolectal remains
very fluid. The transcription of the Bari sentences have been done in
1987 by a Bari student, who followed the standard Bari written system
used in Equatoria.
Mesolectal features taken into consideration include a) phonologi-
cal features (consonants , x, , t, d, s, , h, long vowels and geminates),
b) morphological features (object suffix pronouns, definite article, plu-
ral and gender affixes, TMA and person verbal affixes) c) lexical items.
It must be noted that the insertion of mesolectal features lead often to
mix forms, which are neither Northern Sudanese Arabic (NSA), nor
Creole Juba-Arabic. African vernacular features taken into consider-
ation include:
a) phonological features (p, , , , , ", , j strong palatalization and
affrication);
b) lexical items and idiomatic expressions. Syntactic features will not be
discussed here.
The brief analysis that follows intends to higlight the individual level as
well as main examples of mixing with focus on mesolectal insertion. But
readers are invited to go to Appendix 1, 2 and 3 in order to evaluate the

degree of mixing within each individual speech.
2.1 Text 1, The Dinka Case, Garawiyya Juba, 1981. (Transcript in

Appendix 1)
Summary: A Dinka woman coming from a village accuses a Dinka man
living in Juba to have rob her a sac of sorghum when she came down
from the ferry. She speaks only Dinka and her speech is translated by
another Dinka from Juba. The case started the day before but was post-
poned in order to look for an additional witness who could prove the
identity of the defendant. The woman affirms that he is called Maluk
while the defendant claims to be called Deng. At the beginning of the
second audience, the secretary summarizes the facts (s.1) and then the
Judge is looking for the witness (s.2 to s.9) who finally arrives and sup-
ports the plaintiff against the defendant that he knows since 1975 (s.10
to s.20); the later is finally recognized culprit by the Court (s.21).
This case illustrates a type of language use rather common in the
Garawiyya Court, that I will define as a formal basilectal non vernacular-
ized type of Juba-Arabic. This type of speech is very common among non
L1 speakers of JA (rural people, older generation) in formal situations
and lead to many mix forms. Although it contains most of JA basilectal
features (such as verbal invariable forms), it is characterized by numer-
ous instances of phonological variation (s/; k/x; j/z; a/, etc.) and by
the frequent dropping of the verbs final epenthetic vowel (kelem vs. JA
kelemu speak, gal vs. JA gale say, baaref vs. JA bi-arifu know, etc.). A
number of lexical items are taken from NSA like der want (NSA dyir/
JA aoju), bet house (NSA bt/JA jua), particle aw or (JA wela), etc.
Some old terms inherited from the Ottoman military language remain
like dosoman troubles, fight (cf. s.8) and appear to be part of the court
vocabulary. One notes the irregular insertion of a number of mesolectal
morphological features borrowed from NSA such as definite article al-
(l/el) the, verbal suffixes and prefixes as well as affix pronouns. In Case
1, all speakers speak this formal non vernacularized JA, with various
degree of mesolectal insertions.
The two Dinka participants (the translator and the witness) use inde-
pendent personal pronouns and invariable verbal stems like Translator
(T.) in sentence (s.3) nina m dr mi we dont want to go or sentence
(s.5) ya na klem le ede I talked to him like this. The pronuncia-
tion shows the influence of the Dinka Mother Tongue, particularly con-
cerning the realization of the phonemes // /s/ and /z/ realized as [] (a
frequent pronunciation among Nilotic speakers) like in en what (JA
senu /NSA en); ol somebody (NSA zl/ JA jol), etc. Vernacular influ-
ence is also noticeable in the realization of the vowels, with a tendency
to realize /e/ like an open [] and /o/ like an open [ ]: k (s.3), le (s.5),
bls (s.12), barf (s.13), nugasr (s.12), b and ka (s.16), etc.
The Dinka witness (W) realizes a number of inflected verbal forms
and affix pronouns:
(s.12) nugasr (JA ana bigeseru/NSA agaar)to limit, shorten, (s.20) axk
(JA aku btak) your brother, gutta lek (JA ana gale le-ita) I told you,
negabel (JA. ana bigabalu/NSA agbil) I meet, nesalem (JA ana biwodi
salam to/NSA asallam) I great, neddu (JA ana wodi leo/NSA nadd) I
gave him). It may be noted that 1st sg. imperfective form is n- + verb
(similar to WSA) and not a- + verb (= NSA).
The secretary (S) has a JA pronunciation: (s.1) asrin twenty (NSA
arn), asan because (NSA an), jl person (NSA zl) but izebu he
brings (JA jibu/ NSA ijib).
The Judge displays many instances of phonological variation: z/j (s.2)
win zl el-brefu jl da (with dominance of [j] cf. (s.4); x/k (s.6) axu
brother, (s.15) kemsa five, (s.17) inta bikutu you put; /s (s.4) aan/
asan because, (s.17) musu isnt it.
He keeps many invariable verbal forms: cf. (s.4) nina il we take, (s.8)
nina balasu we stop, (s.21) inta bigul sehi you say right, etc. and many
independent personal and possessive pronouns: (s.9) el-mahkma nadi
eta the Court calls you; (s.15) inta baarfo aglaK bitao you know his
behavior? (s.21) weled btak suker winu where is your small boy?
The mesolectal variables include the 2nd sg. personal pronoun inta
you (JA ita), pl. deictic del this (JA de), definite article l (JA de) and
a number of conjugated verbal forms: (s.9) twori you show, (s.21)
kizibta you lied, sarkta you stole, etc. The conjugation of verbs leads
to some cases of malapropism such as (s.9) nasaltak we ask you with
the coexistence of n- 1st pers. imperfective and -t 1st or 2nd pers. per-
fective. Another mix form is the verb gul in (s.4) bas iji wri gl he
just come to tell that (JA bas bija wori gale/ NSA iji igl or iji iworri
innu). The insertion of mesolectal features increases at the end of the
case (s.22), when the Judge is enouncing his verdict and is using a num-
ber of technical terms such as tazwir falsification, jiza faults/penalty.
Note however that the Judge always alternates basilectal and mesolectal
realizations like the alternation sarkta/ita sereG in (s.22).
A number of mix JA/NSA realizations recorded in Case 1, were almost

permanent in the Judges speech of the Garawiyya and appear also in
Case 2. They seem to indicate that, indeed, this type of mix level was
considered as the appropriate level in the Garawiyya Court.
2.2 Texte 2, The Bay Case, Garawiyya Juba, 1981. (Transcript in

Appendix 2)
Summary: a young Bay police-lady (Hawa) from Bahr al Ghazal accuses
two Bay guys to have tried to attack her the night before. She recalls
that she was going back to her home when the two young men started
to insult her and beat her. Fortunately she was able to call for help and
a soldier came to rescue her and, after many events, managed to bring
the two boys to the police office. The two defendants contest the ladys
version and claim to be victims of a confusion, i.e. the soldier took them
instead of the real culprit. The lady and the two boys agree that they
never met before, in spite of the fact that they all belong to the same
small tribe (Bay). The two Judges argue between them concerning the
importance of this common ethnic background. The soldier is called as
a witness and makes a very long testimony, which supports the police-
womans story. After many discussions and controversies, the two boys
are recognized culprits and have to pay a fine of 36 Sudanese pounds.
This case illustrates a more mesolectal level, mastered by the persons
who have been educated in Arabic and particularly those coming from
the region of Bahr al-Ghazal. However, each speaker has his specificity,
and while the young lady is the most influenced by colloquial features,
the men alternate between basilectal and mesolectal features and cre-
ate many mix forms. A number of lexemes appear to be shared by all
the participants such as der (JA auju) want, bet (JA jua) house, gul (JA
gale) say, fat (JA futu) pass, darab (JA dugu) heat, suf/uf (JA ainu) see,
le-raet-ma (JA lakadi) until, etc.. At the level of the verbal system, one
notes many occurrences of participial forms such as jay coming, gaad
staying, el working and the frequent use of TMA auxiliaries/verbal
particles like kan, bikun, gam/gum, gaad, lisa, biga/baga, ja, aoz.
Hawa (s.15 and s.17), the young policewoman speaks a level very close
to NSA or rather to Western Sudanese Arabic (WSA). She pronounces
many long vowels, as well as velar fricative /x/ and sibilant // but she
realizes few pharyngeal and emphatic consonants: i.e. (s.15) saa watch
(NSA saa), saba morning NSA ab), talta I came (NSA talat) but
end of (s.17) sa'a fekka the watch dropped. She conjugates almost all
verbal forms: cf. (s.15) jit, saketta, zitta, hum zdu and (s.17) gumta,
wageftu, gulta, tetfahim, netfahim, maet, etc. Object pronouns and
possessive pronouns are systematically suffixed: (s.15) garas-ni fi-sulb-i
he pinched me in my buttocks; (s.17) dagga-ni, axwn-i, masak-ni,
xam-ak, bedugg-k, etc. She uses a number of TMA auxiliaries, like
gam and ja to mark inchoative:
(s.15) ana gumta zitta katwa btay I accelerated my walk
(s.17) da tawli masakn ja garasni He immediately sized me and
start to beat me
waled de ja gam darabni bunya The boys started to beat me
strongly.
Hawa masters the morphological rules of NSA and shows very few
occurrences of malapropism or mix form. Gender and plural agreement
however remains irregular: (s.15) ana jay I came-masc., ana mya
I went-fem.; (s.17) arabiya aba yagif the car (f.) refused (masc.) to
stop (masc.), awld der iyamal mayi mukel boys (pl.) want (sg.) he-
makes with me problems/ the boys want to make me problems.
Defendant 1 (John Gabriel, s.28, s.30, s.34) starts answering both Judges
with mesolectal features:
(s.7) J2 sgol btk ? Your work D1: m el/ tleb Not
working/student
(s. 24) samtu I heard-it
(s.28) ana kunt jay min hay kmryal li-l-bT I was coming from district
Commercial to the house.
Then he alternates between few inflected verbal forms and many
invariable verbal forms as well as between affixed and independent
pronouns:
(s.30): ma blis gaadn isugg zl de / uman ft giddm/ ana ja wara/ nama
ana wosol mbil/
/he and police stay-pl they-drive person this/ they pass (sg.) in front/ I
came (inv.) behind/ when I arrive (inv.) Mobil/
He and the police officer were driving this person, they passed in front,
I was coming behind when I arrived at the Mobil Station.
(s. 38) bass darab bitoman ana ja katal /only fight of-them I came (inv.) kill
(inv.)/
I just came to stop their fight.
Defendant 2 (William Peter, s.45, s.47, s.51) realizes many inflected ver-
bal forms and affix pronouns:
(s.52) gabodni zamn gale nemi-l-nukta niuf/ min wisolta unak an
nera mukila bta dn kif
/they-size-me (3rd perf.) time say I-go (1st imp.) the post I-see (1st
imp)/from I-arrived (1st perf.) there for I-see (1st imp) problem of
John how/
They captured me when I went to the police station to see/since I
arrived there to assess the problem of John . . .
Like the Dinka witness in Case 1, William uses the WSA n- 1st sg. imper-
fect pers. pronoun instead of NSA a-. Note also the use of verb gale say
to introduce an embedded sentence (zamn gale time that).
The witness (W. the soldier) starts with a formal mesolectal level:
(s.75) ana kunta aal fi-l-jawazt, ana sken fi Muluk I was work-
ing in the passports, I lived in Muluk. But very quickly he alternates
between basilectal and mesolectal features: ana jit ja wosolu fi medresa
/ana ft/ fialan fatet.. I arrive near the school/ I passed/ of course I
passed; baadin ana jt ana gul when I arrived I say, ana gulta tayib
ya axwna matakum sakal bit fi tarka zede I said good oh my broth-
ers, dont quarrel a girl in this way (note JA negative imperative marker
matakum dont!).
He ends up speaking almost only Juba-Arabic, including at the pho-
nological level ( >s, x > k). Unlike the speakers of Case 1, his level of
Juba Arabic is more grammaticalized. It includes many verbal TMA
markers as well as embedded sentences:
(s.75bis) d tawli gum amol musakl fo mbili li raat ma aoz kaser mrat bta
arabiya zatu
He immediatly started making problems in the Mobil station until
he was almost breaking the mirror of the car
ana bija ft fi jawazt ya neselem kelem kabr li jamaa fi taakir el ana
akr line zaman
I was going to the passeport (office) in order to bring the news to
the people about the delay that I delayed before.
Compared to Case 1, Judge 1 (Ramadan) shows the same level of pho-
nological variation (s/, x/k, z/j) but uses more lexical and morphologi-
cal mesolectal features (see s.25, s.37, s.52, s.74, s.79). It may be noted
that in (s.25) when Judge1 is summarizing the testimony of the police-
woman, he makes more gender agreement than she did and employs the
2nd and 3rd fem. personal pronouns inti and iya:
(s. 25) el-bint de gl iya maya The girl said (masc.) that she was going
(fem.);
inti jya min wn you (fem.) coming (fem.) from where?;
saa btao fakkat waga watch (fem.) of-him drop (fem.) fall (mas.).
But he also mixes with more basilectal JA and consistently uses the
invariable form gul say:
(s. 25) intakum gul inti ermuta sakeT you-pl say you (fem.) prostitute only
(s. 52) nker u gl woket dak kalam inti gul ma hasal klu-klu
deny and say time-that word you (f.) say not happen all-all/
He denied and said (that) at that time what you say never happened.
As in Case 1, the conjugation of verbs and auxiliaries leads to cases of
malapropism or mix forms.:
(s.52) asa nafarn de badi sumit kalamu /now person-Dual this after I-lis-
tened word-his/
now after I listened the word of these two persons (sumit vs. JA asuma
and NSA samit)
fi tarka gmo daribtu bi daraba adid /in way they-stand you-beat
with beat strong/ On the way you started to beat her strongly (gmo
vs. NSA gumtu).
Compared to Judge 1, Judge 2 (see s.66, s.70, s.76, s.78), sticks to a more
basilectal pronunciation ( > s, x > k, etc) and grammar:
(s. 66) musu barau rtan bta bay keda wonos /Neg alone language of Bay
like-this spoken/
Isnt it a specific Bay language which is spoken? (note the impersonal
structure rendered by shift of stress to the final syllable of wonosu).
Case 2 highlights the coexistence of different levels (types?) of (Juba)
Arabic and different degrees of mixing. The level of the police-lady and
the intermediate level of Judge 1 and D2 could be an indication that the
mesolectal level is indeed the prestigious urban formal level that people
speak or try to speak in formal context like the Court. This hypothesis is
however not sustained by the data from Kator B Court, which indicates
another type of urban use.
2.3 Texte 3, Kator B Court, Judge Stephen, 1984. (Transcript in

Appendix 3)
Summary: This case starts in a noisy atmosphere and the Judge Stephen
asks for silence from the audience (s.1). The secretary and the police-
officer try then to stop a person who wants to talk (s.2 to s.5). The case
concerns the non-payment of a dowry. The plaintiff Gabriel (from s.10
to s.30) accuses the defendant (Santino) to have married his sister dur-
ing the first civil war and to have never paid the dowry. The sister had
died and Santino did not take care of the four children, and one died in
an accident. Gabriel now takes care of the children and is asking for the
payment of the dowry. From (s.31) to (s.33) there is another interrup-
tion from a woman. In (s.35) the defendant explains why he was in the
incapacity of paying the dowry. A number of relatives come to give their
testimony (parts not reproduced in the Appendix). At the end, the fam-
ily of the defendant agrees to pay the remaining of the dowry and the
Judge Stephen (s.36) asks everybody to reconcile.
The plaintiff is speaking in Juba Arabic while the Defendant is speak-
ing in Bari. The Judge Stephen, as well as the police and the secretary
shift between Juba-Arabic, Bari and a few English (see s.30 for an exam-
ple of code mixing Bari-English).
This Case illustrates a type of language use very common among
the Bari population of Juba and surrounding areas, with a deep inter-
penetration between Bari and Juba-Arabic, which certainly helped the
vernacularization of JA.
When the Judge Stephen and the plaintiff are speaking in JA, one
notes the phonological influence of the Bari languages with realizations:
f>p pi in (NSA fii); pogu on (NSA fgu),

z>j julumin spoiled (NSA mazlmn)
> aat banaat girls (JA banat/NSA bant), maat die (JA mutu/
NSA mt)
e>" gdiyat case (JA gediya/NSA gadya) (/"/ is noted in Bari
script and by Bari speakers).
A number of words are common to both JA and Bari like mali dowry
from Arabic maal money, or kurju cultivate from Bari kuruju. Some
expressions are idiomatic translations from Bari expressions such as
(s.38) tusu bujak ~tufu buzaK spit saliva, i.e. give benediction.
When the defendant is speaking in Bari, one notes the presence of a

number of Arabic (sometimes from Ottoman military Arabic) words like
temporal markers kan was/before, badin then/at that time, asa now,
discourse markers taban of course kalas OK, numerals (all numbers
are in Arabic) and words linked to administration or political events
such as hawadis civil war, hukum judgement, korbaat whip.
Inter-penetration between the two languages goes over phonological
and lexical transfers. It includes also syntactic similarities like the use of
verb say (gale in JA, adi in Bari) to introduce embedded clauses (Miller
2001).
(s.7) taban ana sufu zede gal kan uwo ro ptisu mal . . . .
/naturally I saw like-this say if him go look dowry . . ./
I realized that if he was going to look for the dowry
(s. 36) kirut adi monye sarji adi ukum nagwon kju monye a ukumbe adi asa
man dek pitn kwe
Then her father sent a message that the judgement passed before was
that now I want the remainder of my money.
Note that almost none of the mesolectal features, so common in the
Garawiyya, were recorded in this case. I noticed that the Judge Stephen
never tried to imitate a mesolectal level, even if some of the young par-
ticipants were talking in mesolectal JA. If he had to alternate and adapt
to the speaker, it was between JA and Bari and not between JA and
mesolectal. Stephens way of speaking is still a rather formal and juridi-
cal way of speaking and cannot be associated with an informal level of
JA. However his speech (particularly s.37), as well as that of Plaintiff
Gabriel, was rather similar to the level of JA broadcasted by the radio of
the Sudan Council of Churches and is rather similar to what linguists
will consider as Juba Arabic (see Watson (1984) in particular).
3. Conclusion
The three cases illustrate different levels/types of Juba-Arabic. Of

particular interest was the attitude of the two Judges, Ramadan in
the Garawiyya and Stephen in Kator. Ramadan realized a number of
mesolectal features and used some Arabic Juridical terms. However, he
was adaptating his speech level to his interlocutors (cf. compared degree
of mesolectal insertion between Case 1 and Case 2) and he never tried
to assert his Authority through the exclusive use of a high linguistic
norm. He also never commented about the language use of the other
participants. As for Stephen, he always spoke a basilectal vernacularized
Juba Arabic and was playing between Bari and Juba Arabic. He often
made some comments about the Southern identity (ena fi januub here
in the South, kalam bta januub words/languages of the South, arabi
bitana bta januub our Southern Arabic) and once criticized a young
man talking like a Northerner.
The language use of the Judges indicate that while the Garawiyya
Court was influenced by its surrounding Arabized Malekiyya neighbor-
hood, the Kator B Court was influenced by the Bari surrounding, even
if most speakers prefer to speak Arabic rather than Bari. Two types of
urban models were present here. On the one hand, the Northern Suda-
nese Arabized urban model symbolized by the merchant community
(both Arabs and non Arabs living near-by), which still had an influence
upon part of the Southern population (old settlers as well as newcomers).
On the other hand, the East African urban model (Nairobi, Kampala)
brought back by the returnees and supported by the local Churches and
the political Southern activists. In 1984, at the breaking of the second
civil war, these two urban models were still coexisting together with
more rural traditional ways of life. Language diversity, as well as religious
and ethnic diversity, were considered natural components of the city.
How far did somebody like Hawa, the young Bay policewoman,
and Stephen, the former policeman or Gabriel consider that they were
speaking the same language (Juba-Arabic?), or that they were speaking
two different languages? This was a question that I did not ask at the
time of recording and that I cant answer. I later worked with some Bari
informants in Khartoum. They could very easily reproduce Creole/basi-
lectal Juba Arabic features, when asked to do so. For those who knew
NSA, they were perfectly able to distinguish between the two systems.
They had therefore a clear consciousness of what JA was, as a distinctive
linguistic system. I noticed, however, that when I asked some of them
(students trained in linguistics) to transcribe some of the tapes recorded
in Juba local Courts, they tended to systematically transcribe them in a
basilectal Juba Arabic phonology, without reproducing the mesolectal
variations. It was as if, for them, people from Juba were speaking one
language. I did not discuss with them the reasons of their attitudes and
I dont know if they were not aware of these variations.
The recording of natural corpora help to better grasp the natural
diversity. But it makes the linguistic analyses more complex and renders
the concept of autonomous linguistic system rather problematic. As it
is also the case with Standard and Colloquial Arabic, it is possible to

describe the two poles of the continuum as two different systems. But
it is hardly possible to decide where the boundary is between these two
languages within the continuum.
4. References
Heine, Bernd. 1982. The Nubi Language of Kiberia. An Arabic Creole. Berlin: Dietrich
Reimer.
Luffin, Xavier. 2005. Un crole arabe: le kinubi de Mombasa, Kenya. Muenchen: Lin-
com.
Mahmud, Ushari. 1979. Variation and Change in the Aspectual System of Juba Arabic.
Ph.D. thesis, Georgetown University.
Miller, Catherine. 1984. tude socio-linguistique du dveloppement de larabe au Sud
Soudan. Thesis, University of Sorbonne Nouvelle-Paris III.
. 1987. De la campagne la ville. volution fonctionnelle de larabe vhiculaire
en Equatoria (Sud Soudan). Bulletin du Centre dEtude des Plurilinguismes (Nice) 9.
126.
. 1989. Kelem kalam bitak: langues et tribunaux urbains en Equatoria. Matriaux
Arabes et Sudarabiques (Paris) 2. 2358.
. 1993. Restructuration morpho syntaxique en Juba-Arabic et Ki-Nubi: propos du
dbat universaux/superstrat/substrat dans les tudes croles. MAS-GELLAS Nouvelle
Srie 5. 137174.
. 2001. Grammaticalisation du verbe dire et subordination en Juba Arabic. Leons
dAfrique. Filiation, rupture et reconstitution des langues: un hommage G. Manessy,
ed. by Robert Nicola, 455482. Leuven: Peeters.
Owens, Jonathan. 1991. Nubi, Genetic Linguistics and Language Classification.
Anthropological Linguistics 33, 130.
. 1997. Arabic-based Pidgins and Creole. Contact Languages. A Wider Perspec-
tive, ed. by Sarah G. Thomason, 125172. Amsterdam-Philadelphia: John Benjamins
Publishing.
. 2001. Creole Arabic: the Orphan of all Orphans. Anthropological Studies 43,
3. 348378.
Tosco, Mauro and Jonathan Owens. 1993. Turku: A Descriptive and Comparative
Study. SUGIA 14. 177268.
Tosco, Mauro. 1995. A Pidgin Verbal System: The case of Juba Arabic. Anthropological
Linguistics 37, 4. 423459.
Versteegh, Kees. 1984. Pidginization and Creolization: the Case of Arabic. Amsterdam-
Philadelphia: John Benjamins.
Watson, Richard. 1984. Juba Arabic for Beginners. Juba: Summer Institute of Linguistics.
Wellens, Ineke. 2005. The Nubi Language of Uganda. An Arabic Creole in Africa. Leiden:
E.J. Brill.
Yokwe, Elisai M. 1985. The Diversity of Juba-Arabic. Studies in African Linguistics,
Supplement 9. 323328.
5. Appendices
5.1 The Dinkas case: Garawiyya Juba, 1981

(1) S. ombre gedya fi mantr tultu mya tnn u asrn / tabn ombre azim
asan mra de ibu ns/ jl el ga-ref asn zdu bta jl de / gal yani
de wa gl mlk/
Yesterday the case in the register (was recorded as) three hundred
twenty two. Of course it has been reported so as the woman could bring
people, somebody who knows him [i.e. the defendant] and could tes-
tify. Because the man said he is called Deng and she said he is called
Maluk.
(2) J. win sud/ zl el-brefu jl da / inta
Where is the witness? The one who knows this person? You?
(3) T. ama ol igl en ya wktu / gl ns tmon fi k k / lkan mai
lomon / oman gl zl de armi / mumkin fi jerma tni nina m dr mi
/ umon kfu b / m dr msi/
Listen she says what Wakotu (??) /she says that their people are in
Konyo Konyo/ if she goes to them / they say this person is a thief / maybe
there are other crimes / we dont want to go / they are afraid of him, they
dont want to come.
(4) J. Nna bas l haj-e-en/ nna ndin jl /wo biji an ije-wri isim ta jl de
bs / ms kalm tni/ bas iji wri gl ya mahkama na brefu jol de / sim
t filn / asn wo indi tnn isim ena / jl de biwri yat el-numro biyref
fg / bes kalm tni m f/
We take him for what? / we called the person / (so that) he comes to
show the name of the person only / Nothing else / He just comes and
tells Court I know this person / his name is so / because he has two
names here / The person tells which number (name) he knows / there
is nothing else.
(5) T. Ya na klem le ede b"t nagfa / gl l/ lkn intkum dr kalam/
klita ol ir wodna ma bl fi k k
Indeed I talked to him like this but he refused / he says to me / if you want
words / lets somebody go take us with the police in Konyo Konyo.
(6) J. La la / tyeb hwa indu xu / jl bita f / rajel bita wn ?
No, no / OK does she have a brother / does she have somebody / where
is her husband?
(7) T. ay f fi bra
Yes he is outside.
(8) J. Tyeb / kan uwo ra ma axu bitao / ma rajel bitao / uwa birh wri nas
del / yajeb nas del / ma fi kalam dosoman / keli nasma blis lokan fi blis
Good / she should go with her brother / her husband / she goes to tell
this people / she brings this people / there will be no troubles / lets listen
to the police if there is a policeman.
[interruption] The policeman refuses to go with the woman and says:
/ you can not approve a sick man / you cannot approve it a all / because
we blis ma negder/. ./
(8bis J.) its up to us / lakin de ya ma besma min kalm tay aja zede / de fikra ana
kelem ta sakeT maales / kan keda kwayis / izakan u ma jebu nas de / jol
/ nas el beld el ja wri isem / nina balasu-l-gedya /
Its up to us / but this one does not listen to me / this was a simple
idea / no matter / if so ok / if she does not bring the people / the per-
son / people of the village who come to show the name / we will end
the case.
[the case is suspended, the woman goes outside and comes back with a
witness]
(9) J. inta bahalef harba wala bk / bibl / asma el-mahkma ndi eta aan
haja / nasaltek aja / twori kalam mosbt inta baaref / eta ma tddisu
/ mara de gl aslu inta baarfu jl de / jl de gabel keda besuf ?
You swear on the spear or on the Book? / the Bible? / Listen the Court
call you for something / we ask you something / you tell the right thing
you know / you dont hide / this woman say that you know this person /
this person did you see her before?
(10) W. jol de / ana baaref yisim bita / isem t maluk / maluk aywen
This person / I know his name / his name is Maluk / Maluk Aywen
(11) J. inta arfu da mitin
Since when do you know him?
(12) W. Ya zl taban ana bigul neya /asa jina asa al ma ksel ydi / fa wd
de ma al u bls masa alna / u wd de taban u barefu zol de taban /
lakin u mubleK / badelma ana baref zol de / ana ma mumken nugasr
gediya bita l / interruption /
O naturally I will say the truth / actually I work as a cleaner / and
the boy came to me and the police came to me / and this boy of course
he knows this person / but he is the plaintiff / since I know him (not
clear if it is the plaintiff or the defendant) I cannot shorten(?) his case
if
(13) J. tegriban jo kem sanawt eta barfo?

Approximately, for how many years do you know him?
(14) W. aywa s / min kema u abain
Yes true / since 1975.
(15) J. inta mudda de / min kemsa u sabain li haddi nahar de / jl de min de
inta brfo uwo / hl inta baarfo aglaK bitao / u jl kwayis wala kan
indu jerima bta nas biil haja bta nas
You, during this period / since 1975 until now / this person since you
know him / do you know his behavior? / Is he a good person or someone
who has crime of people stealing things of people?
(16) T. s kalm ya ayed el-reys / inta tabyan ta-aref zol wa inta yikun
kunta gyad fi bet wd / inta ma botaaref tani jol de / bref b u binum
wn / inta ainu fi ka / teselem ydu u bref taban / u lekin inta m
ref baaml enu aw baaml enu
You are right Mr. the President / Of course you know a person if you
are staying with him / you dont know else / this person you just know
where he sleeps / you see him on the way / you great him you I know
him of course / but you dont know what he is exactly doing.
(17) J. m batl / Deng Mayen inta bikutu isomu bitak musu Deng Aywen /
lakin dk bigul isom btk Malk Aywen / indak sual le / asal /
Not bad / Deng Mayen didnt you put your name as Deng Aywen?
/ But this one says that your name is Maluk Aywen / Do you have a
question for him? / Ask!
(18) D. uwa d / ana indu sual wa gidamu / uw akyuT / ana nam seta yom
/ uwa ma adana akel wala mya / de lujal jr / [laughter in the room]
/ mara de dr fgri ana bi bi b-l-awanta
This one / I have one question for him / Is he a brother? / I slept six days
/ he did not give me food nor water / Hes a ?! / This woman wants to
make me poor by her tricks!
[The witness and the defendant start to argue between themselves in Dinka]
(19) J. mutarjem winu where is the translator?
(20) W. nerja le kalam de / uw gal enu / izakan ana axk ita addni futur aw
ita adni ay / u gal keda / lekin ya assi ana gutta lek jebel / l-zol de
ana baaref u-binum wn / wa baamol enu fi-l-yom u baamol !enu ana ma
baaref /negabel fi ska nesalem yeda / fa neddu l -akel/ m indi akwn
rabto kulu
Shall I return to you the words? / What did he say? / if I am your brother
you give me breakfast or tea / He said like this / but I told you before /
this person I know where does he sleep / and what does he do during the
day I dont know / I meet him on the way, I great him / and I would give
him food? / Didnt I have brothers that I all raised?
(21) J. to W. izakan ita kan kazb / inta begul sehi lama kan l-mahkma sala
/ lokan gal ana aarfa kalam btau uwa jol batl / lakin inta bet-
kelem ez-zn / l-haG / inta gul ma baaref el aja uwo baamol /
bes inta ligo fi seka bes / de mus kwayis? / tb ya mara / weled btak
suker winu
If you were a liar / you would have say true when the Court ask you / if
you had say I know hos words he is a bad man / but you speak the right /
the truth / you say you dont know what is he doing / you just meet him
on the way / isnt it fine? / Ok woman / where is your small boy?
(22) J. to D. delwokti maakama ligak bi galata galatatn / alla inta kizibta
/ inta gulta isma bitaki Deng / wa hini akwana btk igulu l de
ms isema btak / da tazwir / isem btak Maluk Aywen / de barau
jiza / jiza numro itnin / sarkta dura bta mara de / inta sarkta
dura li"ana inta gr isma bitak / lokan ita ma serG inta ma mum-
kin bitager isma btk /. . . [the story continues again and again..] /
fa keda /nina binsuf nas / daiman mujrimin ketr / nama gabod
bikr isem /. . ./ an keda kalam de nihna gul inta muznb ala-
t-tnin / fa kalam el-mahkama / inta biyemsi sita suur fi sijin /
izakan inta ma jibtu talta u taltin gine bta mara de / w jiza
btk an inta xin l-aman / inta bmsoku talata suur / kulu juma
s-sijin btak tesa sar / izakan ma dafa grs / kan dafa grus inta
tla bara min sijin
Now the Court find you with two faults / today you lied / you said
your name is Deng / and your brothers say no its not your name / this
is falsification / your name is Maluk Aywen / this alone is a fault / the
fault number two / you stole the sorghum of this woman / you stole
the sorghum because you changed your name / if you did not steal you
would not change your name / [the story continues] and like this / we
see people / many criminals always / when they are taken they change
their names / because of this we say that you are culprit for two things
/ and the decision of the Court / you go six months in jail / if you dont
bring the 33 pounds of this woman / and your fault that you are a per-
jurer / you take three months / the total of the jail is nine months / if you
dont pay the money / if you pay you will got out from the jail.
5.2 The Bays case

/. . ./ indicates a cut.
/. . ./
(4) J2. smu mnu D1. John Gabriel

Your name?
(5) J2. abla ? D1. Bay
Tribe
(6) J2. sken wn D1. Hay is-sinema
Living where?
(7) J2. sgol btk D1. ma el/tleb
Your work? Im not working/
student
(8) J2. mrk D1. tnn u irn
Your age? Twenty two
(9) J2. mutm numro tnin smu mnu D2. welyam pitr
(10) J2. abla D2. bay
(11) J2. sgol btk D2. ma el/tleb
(12) J2. student / ah yr ed D2. omr/ sabaater sna
(13) J2. yur ed by ingliz/ eta mus asa tleb wela kf
Your age in English / arent you student or what?
(14) J1. ya askri inta bitalfa / ah an iglu l-hG / kut yidek na / tabn
mesya mu kda / gl wl al-azm / kitb el-mukaddes / agl el-hG
/ kl-el-haG / wel y gr el-hG / kda wor el-mdu li-l-mahkma
/ nafern da mal lk nu
soldier you take an oath / ah in order to say the truth / put your hand
here / of course Christian isnt it ? / say Lord the Great / Sacred book
/ I say the truth/ all the truth / nothing else than the truth / so tell the
matter to the Court / what did these two-persons make to you?
(15) P. wellay ombreh hawli saa saba fi-l-misa ana jay bi tark bta-l-m-
drasa tijriyya de / ana jt / f askri mi giddmi lbes rasmi /
baadin el-awld bard kaman fi giddmi / ana jt talta bi jambum
/ el-be henk gl le ya axi salm m f / el bi-jay bardu kaman gl le
salm m fi / ana saktta / aslu m gutta hja / ana mya / baadin
ana gumta zitta katwa btay / hum bard zdu katwa waray / el-be
henk ja garasn fisulbi / badin el-bi jay ja daggan
By God, yesterday / around seven oclock in the evening I was coming
by the road of the Commercial School / I came / There is a soldier walk-
ing before me, wearing his official clothes / then the boys also are before
me / I came near them / this one told me of my brother, no greetings? /
the other also said no greetings? / I kept silent / I did not say any thing /
I was walking / then I accelerated my walk / they also accelerated their

walk / this one started to pinch my buttocks / and then the other started
to beat me . . .
(16) J1. b-l-yedd wela bi kur
With the hand or with foot?
(17) P. bi-yddu garasn fi sulb bi-ydd / baadin el-bi-jay ja daggan
keda bi-ketfa / ana gumta wagftu / gulta mlkum ya axwni / gl le
mlkum kf / ma titfahim ma'an / ana gult nitfahim ma'akum
agl makum en / gl le inta askut / inta m ermta u bass / ana
gulta lem ukrn / ana met / el-bi jay da tawli masakn ja garasni
/gulta le inta mlek asa / gl li axlaK xamak asa ana beduggk
/. . ./ tawli gumta nadt askri / gutta ya askari taal haj f el-ja-
maa del al-ajamn min / tawli hu / waled de ja gam darabni
bunya / askari jire ja / ja gl intum mlkum ya awld /. . ./ gulta hum
ajyomni / an keda ana korokta lek / askri da gl xls xls de
mawdua bast /"arahkum / woled de lisa der amol maaya mukel
/. . ./ sukna maan l-merkez tawli /. . ./ askari da / f arabiya tli
keda / askri tawli wgef arabiya / arabiya aba yagif / mai wgef
musfa keda /. . ./ lamm- jere mi li-l-arabiya henka / kullum
itnn tawli jmu foK / de gm bdrd fK / gaadn nesru / baa-
din de gm bdrbni bi wara kamn /. . ./ sa' fekka min yddi / ma
ref kn wa ya lu / wela kan wg wn / na ma rfu /. . ./ ya
gumna jna lahadd-l-nkta / sbbo le / inta en / nahl dnek wa
inta ermta u bass / inta mi riya skeT /. . . . /
With his hands he pinched my buttocks with his hand / then this one
slapped me / I stopped / I said whats the matter my brother / he said
what the matter how? / let you understand us ! / I said I understand you
I tell you what? / He said to me you shut up / you are just a pimp / I told
him thank you / I went / this one sized me directly and pinched me / I
told him whats the matter / he said close your mouth now I will beat you
/. . ./ I immediately called the soldier / I said soldier come and see this
people who are attacking me / immediately he / the boy started beating
me badly / the soldier run /asked what the matter boys / I said they
attacked me / this is why I shout after you / the soldier said ok ok this
is a simple matter / lets go ! / the boy still wanted to make me problems
/. . ./ we brought them to the post directly / / the soldier / there was a
car coming like this / the soldier stopped the car immediately / the car
refused to stop / it went stopping at some distance / when he run after
the car there / both of them immediately attacked me again / this one
beated me again / we were fighting / then this one hit me from behind /
my watch came off from my arm / I dont know if he took it / or if it fall
somewhere / I dont know /. . ./ so we came up to the police-station / they

insulted me / what are you / Curse on your religion you are just a pimp
/ you go in the street for nothing . . .
/. . ./
(24) J1. intum samt kalm da D1. samtu
Did you hear these words? I heard
(25) J1. el-bint de gl iya maya w intakum ta-azmtlu skeT / gl dik ya
ermta / u de gerres fislbu hene / w-inta bardu daggest bi ketf
/ u badin gl mlkum / intakum gul inti ermuta skeT / inti jya
min wn / fi tarka gmo daribtu bi daraba adid / wa gam korrk
l-askri ja / w bardu baad el-asker / kamn intu gwum darbtu
baad / asn askri mi an ijib / yemsi kelem lel-l-sowwaK el-kan
waggofu / fat giddm / enta azemtolu darobtu li-raet ma saa btao
fakkat waga / wa wahed halaK kamn / da amn / entu gabelt fi
ska / inta /. . ./
The girl said she was walking and you attacked her for nothing / this
one said pimp / and that one pinched her buttock here / and you also
slapped her / and then she said whats the matter / and you said you are
just a pimp / from where are you coming / on the way you start to beat
her badly / and she cried the soldier came / and also after the soldier /
you start to beat each other / during the time the soldier went to bring /
went to talk to the driver who stopped / he went in front / you attacked
her and beated her until her watch came off / and a hear-ring also / this
is sure / you met in the way / you /. . . ./
(28) D1. el-bit-de / ana kunt jay min hay kmryal li-l-bT / fi wad gidmi
/ ana ma baarf gbel / ma bit de wara / towones m bit de k /
umon gam fi akla / an na zeyde / ana askate umon /
The girl / I was coming from the Commercial District / there is one
in front of me / I dont know him before / with the girl behind / he dis-
cusses with the girl / they start quarreling / because Im like this / I make
them silent.
(29) J1. da awli s kem
This about what time?
(30) D1. awli s sbaa / fi jay bls / fi jay / bit krk le bls / fialn el-blis
j / woddit len mukla klu / ana mi le-beT / yalla ez-zl el-
daraba-l-bit de / ma blis gaadn isugg zl de li-nukta / uman
ft giddm / ana ja wara / nama ana wosol mbil / uman lisa / jl
el-dagga bit u blis / ana ja ligm lsa fi mukla / dl fi mukla unk
/ min kede nama blis ainu na / el-bit gl aywa na gibel hink / ana
gl filn ana gibel hink / akala hsel ana f / ya al-amal min ene
ana ma rif / blis masako jl de u jl de ft / ya ja bas masakan
whed be-rejlan /. . ./
Around seven oclock / a policeman came / the girl called after the police
/ of course the police came / and brought us the all problem / I was going
home / Then the person who hit the girl / he and the police they were
driving this person to the police-station / they went in front / I was com-
ing behind / when I arrived to the Mobil station / they were still, the one
who beat the girl and the police / I found them still quarrelling / from
this when the police saw me the girl said yes I was there before / I said
of course I was there before / the quarrel happened I was there / then I
dont know what happened / the police sized the person and the person
went / then one took me by the legs /. . ./
/. . ./
(34) D1. away / badaln talta / na / blis u bt da / sillna fi nkta / amdulla
na lig waed min az ma mbil enk / baaref ns bT / na gl
kwayis / kan keda ami klem ns bt / gl na fi nkta / il-ana fi
nkta / minnak axuy de / nama nsma kede / r fi nkta enk / nama
anu axuya ja / gl ya tni f / kutt tawli jowa / u badin sab / bit
de gam j len / gl saa woddur u en / u filan ana ma uftu saa
klu klu /
Yes / we stayed three / me, the police and the girl / they took us to the
police station / Amdullah I found one I wanted in the Mobil station /
he knows the house / I said good / if its so go and tell the people of the
house that I am in the police-station / they took me in the station / from
there my brother / when he heard like this / he went to the post / when
he saw my brother coming / he said the other came / they put him inside
immediately / and then in the morning / the girl came and told us that
her watch was lost and what / and of course I did not see her watch at
all
(35) J1. tyib / lama kn wled de /el-biyamml makil maa bt de / inta
ma mait / inta ajiju / inta kan barak wela ma wald da /
Good / when there was this boy / who was doing the trouble with the
girl / you did not go / you support / you were alone or with this boy?
(36) D1. na bara
I was alone
(37) J1. inta bara / tyeb badin hinay de / lama blis gabdu weled dk
mi bi / u gamaa kal da / inta ajijtu maa el-bint-da / blis
lamma j / hl giddm el-b lis eta amoltu makil maa bint da
You were alone / good and then / when the police captured that boy
and went with him / and the group quarrel / you supported the girl / the
police when he came / did you in front of the police make quarrel with
this girl?
(38) D1. ana ma awmol makel / bass darab bitman na ja katal
I did not make quarrels / I stopped their fight
/. . ./
(45) D2. el-zl de ombare ma biill nihna / ana gaad fi-l-bet / ya rasel wdd
da min el-mbl / wodd de maa unk / gl ya ns el-bet John gabad
/ ya ma kan nemi yisf li-l nkta
The person of yesterday did not take us / I was at home / and he sent
the boy from the Mobil station / the boy went there / he says people of
the house John has been captured/ so I had to go and see at the police
station.
/. . ./
(50) J1. nta ma fs / u gabodk mittn
You were not present / where did they capture you?
(51) D2. gabodni zamn gale nemi-l-nukta niuf / min wisolta unak an
nera mukila bta djn kif / bass waed askri gl ya lzem bikn
wahed minhum / bas ja fi hene / tni ma barja bt / u fialan jabni
fi-l-arsa / ya sab bit de maa unk / djn gal ya bit bta ombare
enay jabni fi arsa unk / /
They captured me when I went to the police station to see / when I
arrived there in order to assess the problem of John / one soldier said he
must be one of them / he came here / I didnt come back to the house /
and of course they took me in jail / and in the morning the girl came /
John said this is the girl from yesterday who brought me in jail /. . ./
(52) J1. ya bit / asa nafarn de badi sumit kalmu / da numro wahed de
nker / u gl woket dak kalam inti gul ma hasal klu-klu / gl uwa
mi liga inta kunta akal maa wahed tni / w uwo mi aan iyaziz-
kum / iyazizkum bas inta maa zl dak / el-wokT inta ndit askri
ja / w-el-askri gam sgo / sG el-waled maaki / inta mi u bardu
waratum kaman / kddam giddm henk intum wageftu kaman bit-
kelem kalmt keda / lama u masa wosolu lekum henk / bls gl xls kan
izan keda ta'ali inta kaman maa el-jamaa del / gam xala dk u alu uwa
girl / these two persons after I heard their words / number one this
one denied / and said at that time, the story you tell did not happen at
all / he was going and met you quarreling with somebody else / and he
came to support you / support you against this person / when you called
the soldier came / and the soldier took the boy with you / you go and he
is behing you also / in front in front there you stopped you talked / when
he arrived near you / the police say ok if it is like this come you also with
the people / he let that one and sized him
/. . ./
(66) J2. /. . ./ keda ita indu rtn / musu barau rtan bta bay keda wonos
Do you have a language / isnt it a specific Bay language which is
spoken?
(67) D1. ana borton ndogo lakin ma fi rtan bay
I speak Ndogo but there is no language Bay
(68) J2. kulu ndgo intum trtn ndgo
You all speak Ndogo?
(69) P. ana borton ay / norton kalam enu lakin
Yes I speak / but what shall I say?
(70) J2. de bigl uman beroton enay de rutn / de bigul l / ma boroton ndgo
/ beroton bay / nina azin rfu uman sei bay aw /. . ./ :
This one says that they speak their tribal language / this one says no /
he doesnt speak Ndogo / he speaks Bay / we want to know if there are
really Bay!
(71) J1. tayyb / izakan bay / el-fakka bikun fi enu / lokn ma bay bikun fi
enu / nesma el-jarma bass
Ok / if Bay / the difference would be in what? / If not Bay what would
happen?/ we listen to the crime only!
(72) J1. hata kan abla tni bara ma tam / jol el-arfu jol amel ma ajam /
biwori gul ya de / kan gabil wela ma bay biwori / bas el muhim inta
jeb el-blis el-kn ahall al-mawdu /. . ./
Even if it is another tribe it doesnt matter / the person who knows the
person who attacked him / he shows that it is him / if from the Bay or
not he shows / the important you bring the policeman who solved the
matter!
/. . ./
(74) J1. asma / al-sakya da ixtarrat gl inta shid / lama kan hsel beinakum
maa nafarn del maskal / eta keda gul s el eta bitaarefi bi zabti / eta
jt kif / eta legt maamal enu keda / gul len / le-l-mahkma
Listen the lady (?) decided that you are the witness / when the prob-
lems happened between you and this two persons / say what you know
exactly / how did you come/ what did you find? / tell us / the Court!
(75) W. taban siytu / ana kunta saal fi kart / fi-l-jawazt / mal el-jensi-
yat / badin ana sken fi muluk / lama ana jit ja wosolu fi medresa
Komeriy l / ana ft / taban al-bit de askri ana ma baaref / lbis
maleki / ana ma biyaref askari aw ma askri / faaln ftt / giddm
swiya / baad-ma fatt / al-bit de gum bikor bi-isma-l-blis / gal ya
blis ya blis tal agaod / fialan f gann btatna / bta al-blis / ay jol
bikrek bi-isim el-bls ma mafrd tafgo / lajem ita-ajri suf fi senu fi
senu / tekusu / fialan ana jay / ana jay legtum del itnn klu biajem
el-bit de / baadin ana jt / ana gul ya jamaa mlkum fi sen / gal
el-bit de sottemuni u ft keda / ana gulta tayib ya axwna matakum
sakal bit fi tarka / bi tarika zede / keda tosf el-jisim btna w-el jisim
bitkum wa jisim el-bit / taajem el-bit fi tarka bi-l-ll zede m sh /
wa lisa bikroko / lsa bikroko / fa ana asala-l-bit / hasala enu maak
/ gul nas del ya fat keda / ns del ajemu fi tarik / agru sulbu bitao /
u nama kelem umon setemu / fialan uman setemu /. . ./ bad-ma ana
rajo enka / ana raja / ana ja legitum lisa bisakal ala bit / nama ana
raja min mahal taban ana ft keda / baad-ma ana asma korokoro
waray b-ism el-blis / ana jire tawli ja wara w ana ja legitum l-itnin
del kullu / uman yau / fa filan kelemtum liw / ana gul ya akwnna
nina janbiyin ma mafrd neskel maa badna / hajt zede ma kwayis
/ el-muskila de best tkelwokT // el bit de je gum kelem gl ya askri
il tawasolna fi nukta //
Of course Sir / I was working in the cards / in the passports / the place
of nationalities / then I am living in Muluk / when I arrived at the Com-
mercial school / I passed / of course I dont know that the girl is a soldier
/ she wore normal clothes / I dont know if she is a soldier or not / of
course I passed / a little bit in front / after I passed / the girl started to call
by the name of the police / she said police police come in / of course
in our laws / of the police / anyone who call after the police you cant
let him / you must run see what is happening / you search / of course I
came / I came and found them both attacking the girl / I came / I said
people whats the matter / the girl said they insulted me and went / I said,
good my brothers dont quarrel a girl in this way / by this way / you see
my body and your body and the body of the girl / you attack the girl by
night like this its not correct / and they still shout / they still shout / and
I asked the girl / what happened with you / she said this people passed
like this / this people attacked her on the way / they pinched her buttock
/ and when she spoke they insulted her / of course they insulted her /. . ./
when I came back / I came back / I found that they still quarrel on the
girl / when I came back from the place I went there / after I listened the
cries behind by the name of the police / I run directly back and found
them both / there they are / and I spoke to them / I told them brother
we are southerners no need to quarrel between us / things like this are
not good / the problem is small until now /. . ./ the girl spoke and said
soldier you must bring us to the police office /. . . ./
(75bis) W. ma f maal hukuma keda gerb fogo /. ./ fi arabiya gaad j /. . ./ ana
gum jire bi-wara arabiya de bi-ztu /. . ./ min ana jire henak / ana
asma baga el-bit de biga bikore / uman tni keda bidugu el-bit bi-
wara /. . ./ filan ana jere kede kede kede / ana ja gabodtu da / xalas
amdullah / fi fi mujra waga fi jua mujra /. . ./ negedem kede kede
kede likaat nina wosolu mobli /. . ./ d tawli gum amol musakl
fo mbili li-raat-ma aoz kaser mrat bta arabiya zatu / aoz amol
muskil fi jua arabiya ztu / baadu sd el-arabiya uf kede m f /
da gum maragu bara / ana bardu gum maragu /.../ tni arabiya
ja / rakebna fog le-raat wosoluna fi nokta / baad-ma wosoluna
fi nokta / ana gul jol da jere kals lakin axu da biyaref mahal de /
d nker gul m biyaref da / d maal btao barao u d barao / u ma
biyaref klu klu / ana gul kf / lajem de aku btao li"anu uman hader
al-muskel giddmi ana /. . ./ el-hamdulilay lam fatna balK /
mowdu negla askri de masi mustafa / ana biga ft fi-jawazt ya
neselem kelem kabr li jamaa fi taakir el ana akr lina zaman / ana
msi wodi kabar li-hum enay / lama fi rojyu bitay / el-hamdullay d
kamn gum wasal / ana gal ya de bs ya de bs / uman l-itnn el
ajam ala bit ya de bs /
There is no governmental place nearby /. . ./ a car came / I run after the
car /. . ./ when I run there / I heard the girl shouting / they were again
beating her from behind /. . ./ of course I run like this like this . . ./ I
captured this one / ok thanks to God / there is a sewer canal / he fall in
the sewer /. . ./ we accompany like this like this until we arrived at the
Mobil station /. . ./ this one started to make problem in the station until
he was near to break the mirror of the car / he almost make problem
inside the car / then the owner of the car see its not possible / this one
went out / I also went out /. . ./ another car came / we went in until we
arrived at the police station / after we arrived in the station / I said this
one run away but his brother must know the place / he denied he said
he doesnt know him / this one has his own place and this one also / and
he doesnt know him at all / I said how? / It must be his brother because
they attended the troubles in front of me /. . ./ Thanks to God when we
opened the case / the matter of the transport of the soldier to the hospi-
tal / I went to the passports to bring the news to the people there about
my delay before / I was going to bring the news to them there / when I
was coming back / Praise God this one also arrived / I said this is him
this is him / the two who attacked the girl, this is him /. . ./
/ story continues . . . ./
(76) J2. Tayib intakum ref nefsu kalatnin
Good do you recognize that you are faulty?
(77) D1. la la la kalm ke ana wosol dak ana baray / kalam de kadb / ana ma
jay maa jol / ana jay baray
No no this words I arrived there I was alone / these words are lie / I
didnt come with somebody / I came alone
(78) J2. ya de suhud kelem kalam bta / keda wonosu ma suhud / asalo /
ind"k sual yeslo
But the witness spoke / so discuss with the witness / ask / do you have
a question for him?
[Discussion between the defendant and the soldier /. . ./]
(79) J1. intum kalatanin / mahkama gul "annu intu aemtu bit da fi-t-terk
bidun ay munasaba / u atemtu u gul ermuta / u into garestu-ha
li-raet-ma sa"a bitwdda / tamenu xemsa u arbain gine / w indu
wahed fardeK halla wodur / tomon sabaa gine u nus / de klu bi
kuss maskel btakm maao fi-t-teriK /. . . continue /. . ./
You are faulty / the Court says that you attacked this girl on the way
without any reason / and you insulted her and said pimp / and you pinch
her until her watch got lost / its price was 45 pounds / and she had ear-
rings which got lost / their price seven pounds and half / all because of
your quarrel with her on the way/. . . ./
5.3 Kator B Court 1984, Record 3/ 1984

Judge Stephen to the public
(1) J. . . ./ numro tnin koreraK / izakan nasi bikore bi-ataku ktir ma bikli
makama ymsi / izakan nas biasma kalam de bmsi bi-r / u gdiyat
bemsi bi sra / gowm / mumkin nihina baamolu isrin gediya pi yom
/ izakan ma pi pogu ijaat / ajaat bi sikil de /. . ./ arjuk sabab wa intum
nas mutaimin mutamain / el ja asan asma makama / intum ma bija

ena sakiT / eta jet asan testepiT fi makama / asan kul wa baadin
bukura bisala kalam pi bet / aw mara kaman banaat kan ja asma
asan ma baamalu kalaT pi biyutum / sukran / istamir gediya tni /
Number one, shoutings / if people shout laugh a lot it does not let the
Court going / if people listen the words go quietly / and the case goes
quickly / fast / we can do twenty cases in one day / if there is no troubles /
things like this /. . ./ I praise young people and you the defendants / those
who came to listen the Court / you didnt come here for nothing / you
came to benefit in the Court / because each one later tomorrow solve
his problem at home / or women, also girls they come to listen in order
not to make mistake in their house / thank you / continue next case!
(2) S. Taban Sabastian ../ Alexi Lado
S. to the J. lisa ma ja
Not yet arrived!
to a man nearby) kede belay atla awal haja / ihina fi gdiya / wa mata
lakbat / ita yau ga-asma kalam ini / ita ja bikelem / istena lakad bad
gdiya / msi
Please go away from here first thing / we are in a Case / dont interfere
/ you have heard the words here / you will speak / wait until after the
case / go!
(3) J. to policeman kli jol de yeji hina / korbaaT btak win ? /
Lets this person going here / where is your whip?
(4) Police yau le Wani Its with Wany
(5) J. ilu min Wany / u de jatu barfu urkali ? / masi tow gene-gene bita enay /
Take it from Wani / and this one does he know the Courts police? /
go and sit there!
(6) J. mublek minu / Mogga lo munu D. ana
Who is the Plantiff? / Whos Mogga? I
(7) J. a isim bitak D. Gabriel Morbe
Ah, your name?
(8) J. Barinit wen D. lobunuk
Bari from where?
(9) J. u eta majlum knyo u knyo.
From what are you victim?
(10) P. de taban kan pi ukti keda / kan juju ukti fi jaman hawadis / badin ana
ruwa nadi ml / / gal mal maap / taban jaman de hawadis / taban ana
sufu zede gal kan uwo ro ptisu mal / mumkin boro ligo harib / taban pi
jaman de pi aja de /
Of course there was my sister / he married my sister during the civil
war / then I went to ask for the dowry / he said there is no dowry / of
course it was during the civil war / I realized that if he was going to look
for the dowry he would maybe find the war / of course at that time there
were such things
(11) J. hawdis pi sanatu km (9) P. tamanya u sitin
Troubles of which years? Sixty eight
(12) J. badin ada ta awadis mitin
Then the end of the troubles when?
(13) P. tnen u sabain / badin kida pi itnin sebeyin / taban ina kaman nadi mal
/ gal mal maap / ila nina amul aja de / dgit mrisa / anina aju wodi
le uwo / gali kasara ita jibu / ana ma indu haja tani / badin uwo amulu
karama / badin dgig mrisa de nina wodi le aja de / le uwo / ma amul
lena haja tani / badi swiya yaba ja mat /anina amulu karama ta yaba
wkit maat / wa asa ukti de / ukti de mat / wokan ma wdi le ana ml
/ bes teletin jine eli kan wdi le nina
Seventy-two / then in seventy-two / we asked again for the dowry / he
said that there is no dowry / only we do this thing / beer flour / we want
to bring it to him / he said you bring it for nothing / I have nothing else /
then he made a ceremony / then this beer-flour we brought it to him / he
did not do anything else for us / later on our father died / we made a cer-
emony for our father when he died / and now my sister / my sister died /
and he did not give us the dowry / only 30 pounds that he gave us.
(14) J. mt mitin When did (s)he died?
(15) P. mat aja de fi tnin u tamanin / saar idaser
(S)he died in seventy-two / in November
(16) J. indu iyal Does (s)he have children?
P. indi iyl He has children
(17) J. kam How many?
(18) P. fi tegriban arba / waiT kan ge dowru ma rokuba / uwo ma wdi baal
pogo / ja waga fi nr / gam ja mat
Approximately four / one was walking on his knees / he did not take
care / he fall in the fire / he died
(19) J. pi talta P. pi talta ay
There is three? Yes there is three.
(20) J. talta de fi bant fogo P. fi wae bit u badin tnin yl / tnin woled
Among the three There is a girl and two
there are girls? boys/two boy.
(21) J. eli mutu de woled The one who died was a boy?
(22) P. la / bineya / wa asa itnin iyal de uwo ma biwdi baal fogo / asa itnin
de ge geni le ana
No / a girl / and now the two boys he does not take care of them / now
these two are staying with me.
(23) J. ge gene le eta / kalam kwes / aah /

They stay with you, good, aah?
(24) P. wa min de taban ina julumin / bineya tanina ma mijawaz / bes ina
akudu teletin jine fakat / u bineya kaman maat / iyal kaman ana lisa
yau ge akilu
And from there of course we have been spoliated / our daughter is
not married / we just took 30 pounds only / and the girl died / and the
children I am still feeding them.
(25) J. itakum yau ge raba
You (pl.) are the ones who raise them
(26) P. nina yau ge raba / yau ana majlum ma kalam de
We are the one who raises them / I am spoliated by this story
(27) J. ita masi leo pi bet / le bet btomon gale malu ita keli iyal de kede . . .
Did you went to him in the house / to their house to ask why you left
the children like this
(28) P. walay insala ana masi aslu ma bijibu kabar
By god, even if I go he will never inform me
(29) J. ma ja kurju
He did not cultivate (for you)?
(30) P. bikurju lakin oslu ma bowodi le iyal de
He cultivates, but he never gives to the children
/. . . . . ./ Interruption with another case
(31) a woman : nan kan a tu i diit na kora na torobo na a ilu utu ti mrri
mrri yi ko wate
I went at the time of the distribution of the furniture and then he
fought with me and his wifes
(32) J. ina kju kdyo kiyang baligga / kede kiyang sina / maybe nan kan a
mistake / ado yie gwodam nyo Wani ? ma tini wuni ilo korobat / ti
ko yege o kunen ko doggu kito kadi / bonggwat ban le /
She had reported the case earlier / wait first of all like this / I might have
been mistaken / why do you stand still Wani ? you go receive that whip /
let him carry these, carry these furniture / are there no clothes?
(33) Woman: bonggwat ma a kekeren Clothes! He torn them

(34) J. Santino makme ko Santino there it is
(35) D. Walayi / a hal diri bijab i yemba nio nagwon nan yembi kiyasir nanyit
na i diit na hawadis / a de i inu diit ni kan ukum a itinasar jine
badin kayayu a saba jine / badin kjulu anan medde sona anan di min
taab kulu anan kan lunggi luaser baba uba a ko doggi male kwe kunu
uba nyu / kirut ni a kine mali kirut taban jur kaman ktir bayn a nan
/ kalas utu adi kalas lo utu taban bubul yemba makune mali lep-
eng aje dippa salet / loutu gwe a komonit ti poki ibang / nyena diri i
gwe i dida ko ina uro / taban ko yaba ko atu a nan bubul tindu o
nagwon kata / wu hal kajelu ni a baba lo twane / badin ne a monye lo

twane / kirut adi monye sarji adi ukum nagwon kju monye a ukumbe
adi asa man dek pitn kwe ti gurut kune ko na utu nan dek sitin
jine /
By God / really during the time of my marriage when I married his sis-
ter during the civil war / by then the Judgement here was twelve pounds
and ours was seven pounds / At that time, on seeing this, and in order to
avoid these troubles, I decided to call my fathers brother from Juba so
that he could go to collect my dowry from Juba there / Then from there
this dowry, of course the village is large, not me alone / ok, people said
ok, this man can marry, here is the dowry, he has prepared a cooking
place / He has become an in-law, let him be in the house / at this point
I started to stay with this woman / of course, if her father comes, I can
give him what ever possible / and at that time my father died / then
her father died too / then her father (i.e. an uncle) sent a message that
the judgement passed before was that now I want the remainder of my
money of this woman I want sixty pound /. . . . ./
[The case continues in Bari and JA with many different participants. At

the end the Defendant and his family agree to pay the remaining of the
dowry]
Conclusion from Judge Stephen:

(37) S. dgiga / dgiga / ah bagi mal el kede eta bi silu lau pi bet le nsibat inak
/ ah / u wdium kabar gal nina jain / asan bisilu iyl de kli ruwa tusu
bujak / uman kaman biptisu kruP kwes kida / u jama biji yesrub / wa
nas umu yani makasutin / nas klu / kulu jene beji intum bestenu / jebu
gdiya tani
Minute / minute / ah the rest of the dowry you take it to them at home
to your in-law there / and give them news that we are coming / so that
they take the children for spitting saliva (benediction) / they also look
for a good sheep like this / and the people come to drink / and the moth-
ers people are happy / the maternal uncles people / every child you wait
for (??) / bring another case!.
PARADIGMATIC STABILITY AND FINAL LARYNGEALS IN
NIGERIAN ARABIC, OR WHY HISTORY REPEATS ITSELF,
WITHOUT ACTUALLY DOING SO
Jonathan Owens
CASL, University of Maryland
1. Introduction
Arabic presents a rich, if underutilized, domain for examining processes

of linguistic change. At one extreme are the challenges presented by the
rapid and dramatic restructuring witnessed in the Creole Arabic variet-
ies of the southern Sudan and East Africa (Versteegh 1984, 2004). At the
other are the mundane, day-to-day, decade to decade, century to cen-
tury processes which move slowly through the different varieties, some-
times leading to marked changes, sometimes not. In this paper I would
like to document one such change, which has happened not once, but a
number of times in the history of Arabic, and relate the phenomenon to
a more global interpretation of Arabic language history.
At issue is the conceptualization of Arabic language history. The
reigning paradigm today, indeed one established some 150 years ago,
is that a classical language, or Old Arabic, by various processes of
simplification passed into the modern dialects, or Neo Arabic. One of a
number of problems with this model is that it leads one to collapse what
are often internally differentiated developments into a common mold,
which are then assumed to substantiate the Old Arabic/Neo Arabic
dichotomy. Why this militates against a properly nuanced reading of
Arabic language history is that what may lie behind linguistic changes
are a combination of diverse local developments and fundamental
linguistic principles. There have indeed been many changes in the many
varieties of Arabic spoken throughout Asia, Africa and in former times,
Spain. The fact of change in one variety, however, does not justify an
historical model based on the Old/New split.
This point will be illustrated on the basis of verbs ending in a glottal
stop, or in some cases, a final voiceless glottal fricative /h/. Classical
Arabic, of course, has a glottal stop, whereas nearly all modern dialects
do not.
640 jonathan owens
Thus, in Classical Arabic forms such as bada begin and ja come

appear in most dialects as bada and j. For purposes of this paper I
will assume that there has been a change of , whereby verbs such
as bada fell together with verbs like ban he built.1 In contemporary
dialects these will typically belong to the same class of weak-final verbs,
as the following sampling indicates:
(1) Algerian bd-t, bn-t I began, I built, Nigerian bad-t/ban-t,
Eastern Libyan bid-t/bin-t, Mardin (Turkey) baday-t/banay-t, etc.2
A related development is attested in contemporary Western Sudanic
Arabic comprising NE Nigeria, northern Cameroon, Chad and the
western part of the Sudan (see Owens 2006, chapter 5). In the rest of
the paper I will describe the development in the western Sudanic area
on the basis of a relatively detailed corpus-based analysis of Arabic from
Maiduguri in NE Nigeria, and then address the question of its signifi-
cance for the history of Arabic.
2. Nigerian Arabic3
Nigerian Arabic is unremarkable vis vis other Arabic dialects in its

verbal structure. Like all other varieties of Arabic it has two basic verb
forms, perfect and imperfect. The perfect stem is marked by suffixes
indicating person, number and gender, in the imperfect these being
indicated by suffixes and prefixes. Both perfect and imperfect stems
have two conjugations based on whether the stem vowel is high or low.
In (2) strong verbs are illustrated, with the verbs katab write and libis
1
The proviso for purposes . . . leaves open the possibility that the modern glottal
(stop)-less dialects go back to original glottal-less dialects in Old Arabic (see Rabin
1951). Certainly the glottal-stop-less varieties go back to pre-diasporic Arabic, as
attested in their widespread distribution across the Arabic-speaking world, and their
attestation in Old Arabic sources. If there are proto-forms of Arabic without the glottal
stop, the thrust of the paper would need to be reorientated. A few Yemeni dialects have
a glottal stop (Behnstedt 1985, 43).
Even in classical times, there was a variety without the glottal stop that a glottal stop-
less variant which was prominent enough that in the Koranic reading tradition (qirt)
recitations are fully allowed without the phoneme (Ibn Mujahid ).
2
The value of the suffix, -t, -t, -ayt is an issue independent of the status of the glot-
tal stop.
3
Research support for this work was provided by the German Research Council
(DFG).
paradigmatic stability and final laryngeals 641
wear, which belong to opposite conjugations in both the perfect and

imperfect:
(2) Low stem vowels: perfect /a/ katab; imperfect /a/ bi-lbas he dresses
High stem vowels: perfect /i/ libis; imperfect /u/ bu-ktub he writes4
By and large the assignment of verb stems in imperfect and perfect to
the two classes is lexically determined, and as seen in the example, one
and the same verb stem can, indeed as a rule is, assigned to a different
class in the perfect and imperfect tenses.
Final weak verbs in Nigerian Arabic also follow the same classificatory
schema, as can be seen in the following examples in (3):
(3) Low stem vowels: bana he built, bi-lga he finds
High stem vowels: ligi he found, bi-bni he builds
2.1 Final laryngeals //, /h/ in Nigerian Arabic, some data

So far as the basic paradigmatic facts go, Nigerian Arabic is fundamen-
tally identical to other Arabic dialects, and broadly similar to classical
Arabic. One phonological change involving two sounds has, however,
complicated the distribution of lexical forms.
The two sounds are h and . Historically speaking, NA h has two sources,
OA (or pre-diasporic), /h/ as in ahar month < OA *ahar, and OA *h
as in hilim dream, < h ilm. NA // derives from //, as in irif he knew,
< OA arifa. The change of OA /h / and // to /h/ and // is one common
to Arabic in Chad, Nigeria and parts of the western Sudan. It happens,
however, that /h/ and // themselves are weak sounds, and liable to
variation of different kinds. The range of variants includes the following:
They may be kept: biarif he knows, ahar month, whid one and ahamar
red.5
They may change to the semivowel /y/ next to an /i/: biyarif, wyid.
Lastly, they may be deleted altogether: biarif, shr, wid.
4
As in many dialects, there is no phonemic contrast between short high front and
back vowels. In this dialect, however, /i/ and /u/ must be lexically specified, as they are
unpredictable. As far as verbs go, given a lexical stem specification, the pre-formative
vowel is usually determined by vowel harmony rules: if the stem vowel is /u/, the prefix
vowel is /u/. Otherwise it is /i/.
5
Via the so-called gahawa-complex, whereby an /a/ is inserted in the sequence of
guttural C + C, in this case < *ahmar < ah mar.
642 jonathan owens
The variation spoken of here is at the individual level, determined by

contextual and socio-linguistic factors (see Owens 2006, 242 for statisti-
cal treatment); one and the same speaker may in one place for month
give ahar, and in another r.
The question I will investigate in this paper is what happens to
verbs like simi hear, ga#a 6 cut and karah hate, verbs with final
/h. There are two reasons for concentrating on the laryngeals in this
position. First it is only in this position that a change in the laryngeals
is nearly categorical, as will be explained.7 Secondly, it is here that the
phonological change has direct morphological consequences. If they
are deleted in final position, what effects, if any, the deletion has on the
morphological structure of the language need to be specified. Given a
form such as tismao you.M.PL hear, if the final glottal stop is deleted,
the form tismao would automatically arise. Tisma-o, with final V-V is
not a paradigm otherwise attested in the dialect, however. Alternatively,
it could collapse with the already existent weak final paradigm, tilga you
get, tilg-o they get (see (3) above), which would yield tism-o. Which
alternative emerges is described in this section.
The data for addressing this question is of two types. One is textual
data, which will be summarized in 2.2 below. First I present the results
of a test in which 8 Arabs from Maiduguri were presented orally with a
series of test sentences in which one word was left out. The respondents
had to complete the sentence with the correct form. All of the answers
involved verbs with a final // or /h/, the purpose of the test being to see
in which contexts the final laryngeals were kept or deleted. A sample
question was the following, involving the verb simi hear, the first
sentence requiring a perfect verb, the second an imperfect:
(4) hu simi an-ndim haw ana kula .
he heard the man and I also
hi tisma l-kalm haw hinna kula .
she hears the sentence/matter and they F also
6
The /#/ is emphatic and implosive.
7
It is my impression that the deletion in this position is statistically more common
than in word initial or medial position, though I have not checked this.
The most obvious response for the first was heard him, the Arabic for this
being either sim-t-a heard-I-him if the final // is deleted, or simi-t-a
if it is kept, while for the second either b-isma-ann-a 3-hear-FPL-it
(if kept) or bism-ann-a (if deleted). Note that in the imperfect a final
V-initial suffix (-an FPL in the second example) replaces the final stem
vowel if the final laryngeal is deleted.
The respondents were all under 30 and most are educated. It was found
that those older than 30 had trouble concentrating on the questions, so
unfortunately age differential could not be taken into account.
A total of 16 sentences were asked. The anticipated responses were
distributed among different morphological classes of verbs, both basic
and derived verbs for example, and among different inflectional contexts,
though only subject suffixes were tested for.
Questions were asked for both perfect and imperfect verbs, as
illustrated in (4) above. Two types of suffixes can be distinguished here,
those that begin with a C (e.g. 1SG -t) and those that begin with a V
(e.g. FPL -an). In the 16 test sentences, the answers divided into forms
involving the following suffixes:
(5) Perfect
verbs with subject person suffix -t N = 58
3MPL -o N = 3
3FSG -at N = 1
Imperfect
3 MPL -u/o9 N = 5 (3FPL -an, N = 2)
total N = 16
There were 128 total responses (16 8). In all but 2 cases the responses
conformed to the anticipated answers. The two deviant answers are
ignored here, so in all there are 126 responses. The results are summa-
rized in (6). In (7) three typical responses are given.
(6) Stems appearing with:
8
This stands for any subject suffix that begins with a -t, including for example 1 sg,
sim-t I heard, 2 FSG sim-ti you F heard and 2 MPL man-tu you MPL prevented.
9
In the imperfect the MPL suffix is -u after a high stem vowel, bim-u they go from
the stem bimi, and -o after a low stem vowel, bilg-o they find, from the stem bilga,.
644 jonathan owens
(6) Stems appearing with

(a) (b) (c)
/h = , stem vowel kept /h kept
(= weak verbs) (mixed stems) (= strong verbs)
Perfect
-t 36 0 2
-o 11 1 12
-at 2 0 6
Imperfect
-u/o 14 1 25
-an 8 8 8
Total 71 2 53
before C-initial suffixes:
36 0 2
before V-initial suffixes:
35 2 51
(7) (a) (b) (c)

/h = mixed stem /h kept
Perfect
-t ga#-t I cut ga#a-t
Cf. weak verb ma-t I went cf. strong verb katab-t
-o sim-o they heard daba-o they killed sim-o they heard
cf. ma-o they went
Imperfect
-u/o bism-o they hear bisma-o bisam-o
cf. bilg-o they get
The verbs in column (a) behave like weak-final verbs, examples of which
are given in brackets in (7). The verbs in column (c) behave like strong
verbs, verbs with 3 consonantal roots. The mixed stem in column (b)
has attributes both of stems with deleted final laryngeal and of those
with the laryngeal maintained. Like the former the laryngeal is deleted;
like the latter the final stem vowel is maintained.
There is a fundamental contrast defined by the variable C- or V-initial
suffix.
Final /h disappear almost categorically before a C-initial subject suffix
(36 cases where /h = (column a), 2 where they are kept (column c)).
The verbs are then conjugated like weak-final verbs, that is, verbs with a
CVCV stem, e.g. ga#-t I cut < *ga#a-t , cf. ma-t I walked. Otherwise
the maintenance of /h is slightly dominant statistically, though only in
one instance is the dominance overwhelming, namely in the form b-at
she sold with the 3 FSG suffix -at. Whether the conditioning factor here
is the suffix itself or the verb stem (CC-, which I suspect is the case) is
not apparent in the data since only one -at frame was used.
As with C-initial suffixes, usually before a V-initial suffix if /h are
not used, the stem is shifted to the weak-final class, e.g. ga#-o they
cut, budb-o they M slaughter. In two cases, however the stem vowel
was kept: daba-o they m slaughtered, bitba-o they M follow. Also, in
three cases where /h are not used, stress was irregularly shifted to the
penultimate syllable, bukr-u they M hate.
None of the respondents categorically used or disposed of the final
/h, even in a discrete sub-class of forms (e.g. perfect stems, excepting
the C-initial conditioning factor). Table 1 summarizes the global scores
for individuals.
TABLE 1 INDIVIDUAL RESPONSES
Respondent 1 2 3 4 5 6 7 8
/h kept 4 2 9 2 9 13 9 4
/h dropped 12 14 7 14 6 3 7 11
For 6 of the respondents there is textual material against which their test
scores can be compared. These texts include standard interviews as well
as less formal situations. Of the six, one has no tokens of final /h in his
text, and one has only 1. The total scores from the texts, classified into
morphological context, are given in Table 2.
TABLE 2 TEXT SCORES OF SIX INDIVIDUALS
Perfect Imperfect AP Total

t/n -o at u/o an V,
/h 0 6 5 10 3 0 0 24
10 7 5 5 0 6 2 35
The contexts are the following; in the perfect: suffix -t/n 1 SG, 2, 1PL,
MPL -o, FSG -at, in imperfect MPL -u/o, FPL -an, 3 MSG with object
suffix, AP with plural suffix. An example of each (taken from texts) is as
follows, where relevant giving one example with the final laryngeal kept,
one with it absent.
dabt I slaughtered, daboh-o they slaughtered, dab-a they M
slaughtered it F (these 2 tokens from same speaker); wagaat she fell;
simata she heard him, bism-o they M hear; tugo you M fall; bimba-an
646 jonathan owens
they F are sold; basem I hear him; tbn (AP, active participle)
following-MPL (<tbi). Note that the AP and object-suffix contexts
were not used in the test frames.
In two cases the sample in the textual material is not very representative.
The FSG -at suffix is represented only on two lexemes, one speaker with
five tokens of wagaat she fell, another with five of sim-at she heard
(etc.). Also, the FPL -an occurs only in one speakers text.
In most respects the textual and test material agree: in texts h/ are
categorically dropped before the subject suffixes beginning with -t/n,
and they are similarly nearly categorically absent in the test. In both sets
of data the stems shift to the weak-final class. Similarly, in both there is a
greater degree of variability before the MPL suffix (-u/o) in both perfect
and imperfect verbs, the ratios of kept/dropped tokens being roughly
the same. In Table 3 the percentages are calculated by dividing the total
tokens of stems where /h are kept by the total of laryngeal stems with a
MPL suffix. The token count is given before the percentage.
TABLE 3 TEXT COUNT: /H KEPT BEFORE MPL

SUFFIX/DROPPED, % KEPT
perfect imperfect
kept dropped % kept dropped %
test 11 12 48 14 25 64
text 7 6 55 10 5 66
The texts also indicate that before object suffixes there is a categorical
shift to the weak-final paradigm, regardless of whether or not the suffix
begins with a V or C; thus daba-hin he slaughtered them F, budb he
slaughters it M, rather than dabah-hin/budbaha.108
There is also a fair degree of agreement between the test scores of
individual speakers and their treatment of /h in natural speech. Speaker
4, for instance has the lowest percentage retention of /h in the test, and is
also lowest in the text count, while speaker 6 has the highest percentage
retention in both. Only speaker 7 has a lower percentage retention in the
text than in the test.
10
After strong verbs object suffixes are suffixed directly to the final -C, with no other
changes occurring, e.g. katab-a he wrote-it, katab-hin he wrote-them. After verbs end-
ing in a final -V the final vowel lengthens before a suffix, ligi-hin lig-hin he found
them. Before object suffixes (which were not tested in the frames like (4)) the laryngeal-
final verbs shift to the weak-final class.
TABLE 4 TOTAL RETENTION/DELETION OF FINAL /H

IN TEXTS OF 5 SPEAKERS
Speaker 2 4 6 7 8
/h kept 0 4 9 6 5
deleted 1 11 2 11 8
For the present sample of speakers it can then be said that verbs with
final /h belong to a mixed paradigm: before C-initial subject suffixes
they belong to the paradigm of weak-final verbs; before V-initial subject
suffixes they are treated variably, in cases even by the same speaker,
sometimes as CVCVC stems, sometimes as CVCV, weak- final stems.
2.2 Larger sample

Looking at the larger sample of texts, as summarized in Owens (1998),
the extent of deletion in the pre-vocalic context was indicized over three
sample groups, Maiduguri interviews (N = 58), Maiduguri group con-
versations (N = 61) and village interviews (N = 52). Aspects of the cor-
pus can be briefly summarized here. In the following, an index of 0%
means that the laryngeal is never deleted, of 100% that it always is. Table
5 groups all speakers together, while Table 6 divides them according to
place and sex. The following indices pertain only to the context V /h-V,
i.e. V-initial suffixes.
TABLE 5 TOTAL SAMPLE
Table 5a: Place Table 5b: Sex Table 5c: Age
Village 71% Male 79% <32 83%

Maiduguri 79% Female 71% 3249 83%
>49 61%
TABLE 6 PLACE X SEX
Male Female
Maid 81% 75%

Villages 74% 54%
More so than the data examined thus far, the overall scores point to a loss
of final laryngeals in pre-vocalic position, the loss being most pronounced
648 jonathan owens
among younger, male residents of Maiduguri, with Maiduguri well ahead

of the village overall among both males and females. The text scores
show a higher degree of deletion overall than do the test scores described
above. This is interesting, if the test scores indicate a higher degree of
monitoring (the reading list effect in the classical, early sociolinguistic
studies), it would indicate that the laryngeals are somehow there to a
higher degree than the test scores would show. The lower female scores
would deserve comment, if adequate comparative data were available,
which it is not. In western studies, females have consistently been shown
to lead changes. However, even if western sociolinguistics attempts to
draw far-reaching conclusions from such tendencies (Chambers 1995,
139), the present data cautions against overgeneralizing, to say the least.
Looking at the data overall, a change can be said to have occurred
in the apportionment of final verbal /h verbs: before C-initial suffixes
they merge with weak final verbs whereas before V-initial suffixes
they variably merge with weak-final verbs. Factors speak both for and
against the variable merger going to completion. In favor thereof, urban
dwellers, especially younger ones, have the highest degree of merger.
Rural dwellers, however, have a lower degree, and at this point in the
history of Arabic in Nigeria, they constitute the overwhelming majority.
Moreover, there are structural factors favoring retention: in explicit tests
the final laryngeal was retained more often than in spontaneous speech.
Moreover, in other positions, V-/h-V sequences, such as word-internally
(ahar month), are generally maintained, so at this point at least, the
high degree of variability is most marked at the right morphological edge.
Moving to the next section, the current situation will be represented as
having a split paradigm, with the possibility, that in the future laryngeal-
final verbs will merge completely with weak-final ones.
Before proceeding to the final section, it is relevant to note that
an identical treatment of laryngeal-final verbs is found in Abbeche
Arabic in eastern Chad (Roth-Laly 1979, 810). This shows that the
split of laryngeal-final verbs into two morpho-phonological classes is
old enough to be a pan-western Sudanic Arabic trait. Again, however,
without more detailed study of the situation in eastern Chad it is not
possible to say more than that. If the loss of laryngeals is more advanced
in Chad than in Nigeria, one would predict that they would also be lost
eventually in Nigeria. If, on the other hand, the same complementary
treatment of final laryngeals is found as in Nigeria (governed by form of
suffix), one would rather see the split as stabilized.
3. Morphological Regularity, paradigmatic stability
Before continuing, it is relevant to examine the manner in which the

laryngeal-final verbs have split into two classes. As seen in (7), the laryn-
geal-final verbs either end in a laryngeal (column c in (7)) or in a vowel
(column a). They thus either remain in the class of C- final verbs or they
shift to the class of weak-final verbs:
(8) strong verb (C-final) weak verb (V-final)
ga#aan they F cut ga#ti you FSG cut
cf. katab-an they F write cf. ma-ti you FSG went
What happens only rarely is that a new morphological class is created,
one with properties of both the strong and weak verbs. In (6) there are
two such tokens, attesting to the fact that such forms do in fact occur.11
They are given in column b; both tokens occur before V-initial suffixes.
A form like daba-o they M killed derives from dabah-o with deletion
of the final laryngeal. Rather than switch to another stem class, which
is what usually happens when the final laryngeal is deleted (dab-o), the
final stem vowel has been maintained. In essence, a new morphological
class has been created. This class, however, has not become widespread
or established in the speech community.
Looking at the development structurally, the deletion of the final
laryngeal leads to the splitting of /h-final verb stems into two pre-
existing morphological classes, the distribution of these being broadly
defined by morpho-phonological context. What did not happen is that
the loss of the final laryngeal in one context led to the development of a
new morphological class based on the stem CVCV-suffix. Morphological
stability takes precedence over forms, which would be derived by the
logic of phonological rules.
The WSA developments would ostensibly appear to support the idea
of linguistic history repeating itself: final laryngeals in Arabic verbs
tend to drift into weak-final ones. This repeats a similar process, which
occurred in pre-diasporic Arabic, as illustrated in (1) in section one
(bada-t/bad-t). Under one interpretation of Arabic language history,
one could this add drift to the catalogue of features characteristic of New
Arabic (allowing for the cautionary note 1 above).
11
In the texts from speakers from the present sample no mixed forms occur. In other
texts, however, they are attested.
650 jonathan owens
However, the data allows for a more nuanced description. The change
described for WSA can be broken down into two parts. On the one hand,
what may be termed a principle of paradigmatic stability, one aspect of
morphological stability, can be invoked:
(9) Principle of paradigmatic stability: do not create new paradigms
In its categoricality this statement is clearly too strong, and one can imag-
ine adding many conditions to it, but it serves present purposes. Both
the Old Arabic and the WSA data obey the principle: the loss of a final
laryngeal in both cases did not lead to the creation of a new paradigm.
Rather, the laryngeal-less forms simply collapsed, or in the case of WSA,
are still in the process of collapsing, into already-existing paradigms.
The other part is the phonological change that creates the condition
for the collapse into pre-existing paradigms. This change is one of
happenstance. In WSA, /h/ became /h/ and // became //. This type of
change is not unique among varieties of Arabic. // has moved to // in the
Tihama, /q/ appears as // in Cairene, Damascene, as well as elsewhere.
Maltese presents a complicated picture of its own. and // merged in
//, which in turn was lost, leading, as in WSA, to the merger of *//-final
verbs with weak finals.129
(10) sm-ayt I heard
tf-ayna we threw (< df)
Why the changes occurred in WSA is, frankly, not clear at this point, as
is the question, why in some dialects and/or h continue on to . One
12
In Maltese the final -a of the suffix is conditioned by the historical pharyngeal, and
hence contrasts with, say, bn-eyt I built, where the suffix goes back to the diphthong
*ay-t.
In fact, the historical phonology of Maltese remains to be worked out in detail. Not
least is the problem of a not inconsiderable dialect variability, with its potential impor-
tance for historical reconstruction (cf. Owens 2006, chapter 7 on imla in Maltese).
Mifsud (1995, 308-9) explains the final /y/ in sm-ayt etc. as a change of the historical
pharyngeal trace to /y/. This analysis is interesting in and of itself, but probably deserves
an article of its own. Briefly, while Mifsuds analysis still maintains the paradigmatic
stability principle, it is on an a priori basis more complicated than the treatment offered
here. Mifsud notes that in general verbs with historical final voiced pharyngeals merge
with weak-final verbs (e.g. nitfa we throw, like ninsa we forget). The current analy-
sis sees the merger as having occurred throughout all inflectional paradigm members,
allowing for the underlying phonetic conditioning of [a] due to the pharyngeal trace.
Mifsuds analysis would split the paradigm of voiced pharyngeal-final verbs. In the pro-
cess this creates an otherwise unattested CCay-C stem, where the /y/ represents C3 of the
root, and hence would contrast with the split paradigms-analysis offered above for WSA,
since in the present analysis laryngeal-final verbs split into two existent paradigms.
can, of course, appeal to sub-stratal influence. This is quite plausible in

the case of WSA. However, why should the shift have occurred in WSA,
but not in another notable Sprachinsel where Arabic is a remarkable
minority language, Uzbekistan?13 In short, the phonological shift to /h/
and // and the further loss of these sounds, is history in its truest sense,
contingent on particular developments in particular places, in particu-
lar times. This contingency is in evidence in WSA in the very variability
of the process. It is unfortunate that longer term trend studies cannot
be followed through on to trace the further development of this phe-
nomenon. At what point, if ever, does the change go to completion, and
why?
In any case, once these sounds are given up, the paradigmatic stability
principle takes over. This obtains in all varieties of Arabic that have (1)
above, in WSA, as described here, and in Maltese as well, as suggested
around the discussion of (10) above.
In short then, the sequence of developments is as follows:
(11) loss of final laryngeal
(12) obeisance of the paradigmatic stability principle (see (9) above)
In other words, given the loss of the laryngeal, collapsing into the weak-
final paradigm follows automatically. In the current framework, only
(11) is history proper. (12) is suggested to be the instantiation of a gen-
eral linguistic principle in Arabic verbal morphology. In a sense, so long
as Arabic exists, (12) will be operative in the way described here. Its
reappearance at different points in the history of Arabic, however, is not
compatible with a conceptualization of Arabic as having changed from
and Old type to a New type. To the contrary, because the same struc-
tural forces are at work in pre-diasporic times as are at work in 2006, no
change has occurred.
4. Appendix
Verbs used in test frames (given in the form of a possible answer): dbaho
they slaughtered; simt I heard; wjt I faced; karaho they hate; ga#t
I cut; garti you FSG stopped; mantu you MPL prevented; binfaan
they F are useful; budbaho they M slaughter; bisaman they F hear;
13
Moreover, in WSA emphatic sounds are fully maintained, but they are lost in
Uzbekistan Arabic.
652 jonathan owens
tuwjuhu you MPL face; bikrahu they M hate; bitbao they M follow;
bat she sold; bibu they M sell; ga#ao they M cut.
5. References
Behnstedt, Peter. 1985. Die nordjemenitischen Dialekte. Teil 1: Atlas. Wiesbaden:

Reichert.
Chambers, Jack. 1995. Sociolinguistics Theory. Oxford: Blackwell.
Ibn Mujhid, Ahmad b. Ms. 1972. As-Saba f l-qirt. Ahmad awqi D ayf, ed. Cairo:
Dr al-Marif, 1972.
Mifsud, Manwel. 1995. Loan Verbs in Maltese. Leiden: E.J. Brill.
Owens, Jonathan. 1998. Neighborhood and Ancestry: Variation in the Spoken Arabic of
Maiduguri, Nigeria. Amsterdam: Benjamins.
. 2006. A Linguistic History of Arabic. Oxford: Oxford University Press.
Rabin, Chaim. 1951. Ancient West Arabian. London: Taylors Foreign Press.
Roth-Laly, Arlette. 1979. Esquisse grammaticale du parler arabe dAbbeche. Paris:
Geuthner.
Versteegh, Kees. 1984. Pidginization and Creolization: the Case of Arabic. In Amsterdam:
Benjamins.
. 2004. Pidginization and Creolization Revisited: The Case of Arabic. In Haak,
Martine, Rudolf de Jong, and Kees Versteegh, eds. Approaches to Arabic Dialects:
A collection of articles presented to Manfred Woidich on the occasion of his sixtieth
birthday. Leiden: Brill, 343358.
SOME ASPECTS OF DIGLOSSIA AS REFLECTED IN THE
VOCABULARY OF LITERARY AND COLLOQUIAL ARABIC
Judith Rosenhouse
Swantech Ltd., Haifa, Israel
1. Introduction
1.1 Preliminary remarks

This paper focuses on diglossic differences between literary and collo-
quial Arabic mainly from the morpho-phonological and lexico-semantic
aspects.
The lexical part of the language is probably the area that most clearly
distinguishes between these varieties of Arabic. Our survey deals with
phonology, morphology and lexicon/semantics.
After the introduction, the second part deals with phonological
features that distinguish literary Arabic from colloquial Arabic (e.g.,
consonants such as /q, j/ and /t/, the vowel of word-final imla, and the
emphatics). The main differences between the two varieties of Arabic
reflect the well-known typical segmental features of colloquial Arabic.
Next, morphological or morpho-phonological comparison between the
phonological system of cognate lexical items in colloquial Arabic (as
spoken in Israel) and literary Arabic is undertaken.
Literary Arabic uses a considerable number of lexemes in their
foreign forms often adapted to Arabic morpho-phonology ; in such cases
colloquial and literary Arabic lexemes are often cognate. We look briefly
into the lexical areas where these phenomena occur.
These features can be described within language interference or
language contact as terms of one linguistic approach, or as code mixture
according to another. From this point of view, these features do not differ
much from phenomena in other Arabic dialects (Boucherit, 2000, Heath
1989, Owens, 1998, Taine-Cheich, 2000) or languages (e.g., Ritchie and
Bhatia, 2004).
654 judith rosenhouse
1.2 Introductory remarks

The centuries-long development of modern literary Arabic, elsewhere
also Modern Standard Arabic, and colloquial Arabic, the two faces
of the Arabic language, and the differences between literary Arabic
and colloquial Arabic, have been the topic of much linguistic debate
especially since Fergusons (1959) well-known paper on diglossia (e.g.,
Blau, 1977, Blanc, 1960, Bousaffara-Omar, 2005, Diem, 1974, Fernandez,
1993, Fck, 1950, Haeri, 2003, Holes, 1995, Kaye, 1994, 2002, Mejdell,
2002, Meiseles, 1975, 1980, Monteil, 1960, Stetkevych, 1970, Versteegh,
1997). As this area is also of much interest for Professor Kees Versteegh
I dedicate this paper to him.
Literary and colloquial Arabic reveal differences on all linguistic
levels, including phonology, morphology and syntax, in addition to
vocabulary, as described in Ferguson (1959). Our recent work on
trilingual dictionaries (Rosenhouse, 2001, 2004) has given us a genuine
and tangible feel for this situation from the lexical aspect of Arabic
diglossia in Israel, and the present paper is partly based on items in
those volumes.
Some random examples of such differences are shows in Table 1
below. We take into account the relevant literature, and rely on a lexical
search of several Arabic newspapers from which we gathered examples
for the studied points.
TABLE 1 EXAMPLES OF LITERARY AND COLLOQUIAL ARABIC

NON-COMPATIBLE LEXEMES
Literary Arabic Colloquial Arabic Gloss
dumya lube Puppet

ihd tarh Abortion
ihtada libes el-kundara To put shoes on
zujj qazz Glass
kura tbe Ball
zawraq flka Boat
The situation of the Arabic language has recently been described as

follows in Ayyoub: In societies where a substantial segment of the
population is literate, and two variants of language usage exist, the gap
between the literary/written and spoken Arabic tends to diminish. With
the gap disappearing, a continuum emerges, offering the speaker a rich
array of possibilities (2002, see also Kaye 2002). Such a mixed language
diglossia as reflected in the vocabulary of arabic 655
can be examined at present from oral material, or from written texts

dating from the 7th century to the present (e.g., Blau 1977,1 and see
Chapter 1 in Mazraani 1997, for a review of relevant literature). For
Arabic, this mixture is known also as middle Arabic (al-lua al-wust,
Badawi 1973).2
The importance of the lexical aspect in the distinction between literary
and colloquial Arabic is partly due to the sheer number of the lexical
items, which naturally far exceed the number of grammatical structures
and phonological categories.
Our research hypothesis and therefore starting points are based on
two assumptions: 1. More literary Arabic lexemes (and more loanwords
from various languages) are used in colloquial Arabic, than colloquial
Arabic lexemes are used in literary Arabic (cf., e.g., Al-Falay 1996,
chapter four). This apparently is because literary Arabic often uses loan
translation (calque) or new terminology coined by official authorities
such as academies of Arabic language in Egypt and elsewhere (e.g.,
Shraybom-Shivtiel 2005) instead of loanwords. 2. When foreign lexemes
are used in literary Arabic or colloquial Arabic, they usually adapt to the
Arabic system phonologically, and also morphologically when necessary
and possible.
1.3 Literary and colloquial Arabic cognate words

Literary and colloquial Arabic cognate words vary mainly in phonologi-
cal and/or morphological features and not semantically (Cadora 1979).
Most of the colloquial Arabic examples referred to below reflect col-
loquial Arabic as spoken in Israel. The picture we get from this coun-
try may differ somewhat from that in other countries. In fact, literary
Arabic is not uniform either, and literary Arabic lexemes vary in different
1
Several views of middle Arabic do not consider it a separate variety or a variant of
the language, for it lacks a grammatical coherence: we simultaneously find correct forms
of the literary Arabic and deviations from these forms. We even find hybrid forms that
are neither literary nor dialectal (Ayyoub, 2002. Cf. e.g., Kaye, 2002, Mejdell, 2002).
2
Diglossia is also related to bilingualism, since two language systems are dealt with
(cf. Fasold, 1984, Rosenhouse and Goral, 2004). Thus, when speakers use colloquial
Arabic lexemes in a literary Arabic context, or vice versa, they may do it either for spe-
cific functional (stylistic) purposes or as an unintentional process of code switching.
Although we do not deal with code switching in this paper, I believe that the use of liter-
ary Arabic words in a colloquial Arabic context is not to be considered code switching
or code mixing if the literary Arabic word is already integrated in colloquial Arabic and
used in it as the only accepted lexeme for that specific notion.
Arabic-speaking countries, in particular in the fields of technical

and scientific vocabulary (Holes 1995, Badawi 1997). For simplicity,
we refer to literary and colloquial Arabic as distinct and independent
systems.
The following sections of the paper focus on a comparison of
contemporary literary and colloquial Arabic vocabulary as reflecting
Arabic diglossia, and not on their diachronic development. After a short
description of phonological and morphological features of literary and
colloquial Arabic vocabulary ( 2), three aspects relating to semantics
are examined:
(1) the use of foreign lexemes in literary and colloquial Arabic ( 3.1);
(2) the use of colloquial Arabic words in literary Arabic and literary
Arabic words in colloquial Arabic ( 3.2);
(3) semantic differences between cognate literary Arabic / colloquial
Arabic words ( 3.3). Section 4 concludes this essay.
2. Phonological and morphological differences between literary

and colloquial Arabic
The brief description of phonological and morphological differences

between literary and colloquial Arabic lexical pairs in this section is
intended to demonstrate processes that distinguish literary from col-
loquial Arabic. These processes appear in the oral articulation of literary
and colloquial Arabic and mark them clearly. Ferguson (1959) consid-
ered such lexical literary / colloquial Arabic pairs as an important fea-
ture of diglossia. For a detailed phonological and morphological analysis
of colloquial / literary Arabic lexical items in the discourse of speakers
from three countries (Egypt, Iraq and Libya) see Mazraani (1997).
2.1 Phonology
Phonological differences between literary and colloquial Arabic in Israel
and elsewhere are among the most studied and noted features in the
literature of Arabic diglossia (e.g., Altoma 1969, Blanc 1960, Meiseles
1980, Fischer and Jastrow 1980, Holes 1995, Mazraani 1997, Rosenhouse
1984, Versteegh 1997). Usually, speakers of a colloquial Arabic dialect
use the same consonants for literary Arabic speech as for colloquial
Arabic. The emphatics /s, d, t, / and the somewhat language-specific
pharyngeals /, h /, as well as plosives, fricatives, lateral and semivow-

els are pronounced in the same manner in both (spoken) literary and
in colloquial Arabic (at least in Israel), and therefore are not discussed
here. That is, most speech sounds are shared by the phonological sys-
tems of literary and colloquial Arabic. In Israel as in other dialects (see
Kaye and Rosenhouse 1997), the phonological differences between liter-
ary and colloquial Arabic refer mainly to uvular /q/, the three interden-
tals /t, d, z/, and the allophones / - j/ for Semitic *g. The phoneme /k/
has a palatalized variant // used in certain rural and Bedouin dialects
in Israel, but its use seems to be somewhat dwindling (at least in Israel)
due to the effects of prestigious urban colloquial Arabic (in Israel) dia-
lects and the spread of literary Arabic due to school education. Another
case is that of the emphatic /d/ which merges with // in certain (rural)
dialects. Sometimes such dialect speakers do not merge these phonemes
in literary Arabic. The vowel system of many colloquial Arabic dialects
(in Israel) includes two vowels that do not exist in literary Arabic: /e, o/,
and the parallel long monophthongs /, / which usually reflect the liter-
ary Arabic diphthongs /ai, au/ respectively. When speaking in literary
Arabic, the traditional literary Arabic articulation of these diphthongs,
as diphthongs, is usually retained in Israel.
Let us discuss now the problematic phonemes of the system as they
occur in Israel. Mainly Muslim, Christian, and Druze speakers of rural
dialects pronounce the phoneme q as uvular. In urban dialects this
phoneme is articulated as glottal // and in Bedouin dialects and some
rural dialects its pronunciation varies, often under certain conditions,
between /g/ and /j/. However, Bedouins and urban speakers do
articulate the phoneme /q/ as uvular when reading the Qurn or when
certain literary Arabic lexemes are used in words such as al-qurn the
Koran, qadiyye problem, issue, qawmiyye nationalism, or dimuqrtiyye
democracy3 (cf. Mazraani 1997, Haeri 2003). A similar process refers to
the other phonemes under discussion here: the colloquial Arabic dialects
vary in their articulation between the urban /d-d-z/ for /z/ and /s, t/ for /
t/, and the respective traditional literary Arabic articulation /d/ and /d/
which simultaneously characterizes rural and Bedouin dialects. The four
basic emphatics /s, d, t, z/ (especially at word-final position) are often
3
In non-phonetic dictionaries the transcription usually hides such dialect differ-
ences. Also in our dictionary only one letter transcribes q in colloquial Arabic.
pronounced as non-emphatic phonemes in some colloquial Arabic

dialects (in Israel), mainly urban ones, as in /s/ for /s/ in colloquial Arabic
/sand/ for literary Arabic/sundq/ box (see Rosenhouse 2002). These
phonemes are, however, transcribed in colloquial Arabic dictionaries
(in Israel) without consideration of the option of this non-emphatic
articulation. Colloquial Arabic (in Israel) has, on the other hand, a
few emphatic consonants that do not exist in literary Arabic, and are
usually phonetic variants of non-emphatic phonemes. The phenomenon
is marginal in rural and urban colloquial Arabic in Israel and is more
prevalent in Bedouin dialects (in Israel and elsewhere). But it may form
phonemic distinctions in certain words such as /b$ba/ pope vs. /b$%ba/
her door, or /a&&a/ God vs. /al-la/ he said to her.
The vowel /a/ of the feminine ending in noun and adjective forms,
which in modern literary Arabic is pronounced /a/, may be /a, e/ or
/i/ in colloquial Arabic (in Israel), with differences depending on
dialect features and phonetic factors. The phonetic condition of urban
and rural dialects in this region is that if the consonant preceding
the suffix is pharyngeal, laryngeal, emphatic, or uvular, the vowel
remains /a/; cf. /sa/ hour, clock; otherwise it bends upward (imla
of classical Arabic), as in e.g. /sane, sini/ year. The /a/ is also a feature
characterizing (colloquial Arabic) Bedouin dialects. Thus this suffix
uses the vowel /a/ in formal (literary Arabic) and Bedouin speech, but
also in (non-Bedouin) colloquial Arabic when preceded by a back or
guttural consonant. The difference in the use of /e/ or /i/ in this ending is
less easy to define, because although it is sometimes apparently dialect-
dependent (see examples in, e.g., Geva-Kleinberger 2004) it often reflects
free variation and is sometime also a more central /i/. Consequently,
four contexts of articulation of the t marbta as /a/ (rather than /e/
or /i/) can exist: (1) formal, literary Arabic; (2) a Bedouin dialect; (3)
phonetically-conditioned urban/rural colloquial Arabic (in Israel); or
(4) a mixed/switched utterance using literary Arabic pronunciation
within a colloquial Arabic structure (in Israel).
The features briefly described here for colloquial Arabic (in Israel) are
similar in principle to those in many other colloquial Arabic dialects,
including Mesopotamian q'ltu and g'l't dialects (Blanc 1964, Jastrow
1978).
Unlike consonants, the vowel system differs in colloquial Arabic
from literary Arabic almost in all the Arabic dialects. This is expressed
by expansion of the phonetic vowel system on the one hand and the
merging of several vowels to a smaller set of phonemes on the other
hand (the latter is conspicuous in North African dialects). The wider the
differences between the traditional literary Arabic phonological system
and that of daily colloquial Arabic, the more literary Arabic articulation
errors may be found in the speech of such colloquial Arabic speakers.
This also is a typical feature of the code-mixing phenomenon. Examples
abound in any recorded colloquial or literary Arabic text (e.g., Blanc
1960), and even presidential speeches (see Mazraani 1997).
2.2 Morphology
Morphological differences between literary and colloquial Arabic (in
Israel) exist in various categories. The pronominal system exhibits well-
known differences in the free and bound pronouns in the nominal and
verbal systems. See examples in Table 2.
TABLE 2 EXAMPLES OF MORPHOLOGICAL DIFFERENCES
BETWEEN LITERARY AND COLLOQUIAL ARABIC
nah nu nih na, ih na we

antum intu(m) you, pl. m.
druki drek your sg. f. home
drukunna drku/drkun/drken your pl. f. home
katabtu katabt I wrote
taktubna (b)tukutbi/(b)tikitbi you sg. f. write/will write
The suffixed object bound pronouns also differ: literary Arabic talabtuka
vs. colloquial Arabic (in Israel) talabtak I requested you, m. sg., literary
Arabic fatah ahu vs. colloquial Arabic (in Israel) fatah o he opened it, etc.
In the nominal system we note literary Arabic af al vs. ifal in several

local dialects of colloquial Arabic (in Israel), e.g., literary Arabic aswad
colloquial Arabic (in Israel) iswad black in the colors groups. Another
example is the female form of some adjectives, which in literary Arabic
are of the fal pattern whereas in colloquial Arabic they are regular,
i.e., with the t marbta, e.g. literary Arabic sakrn colloquial Arabic
(in Israel) sakrne drunk. Differences between literary and colloquial
Arabic (in Israel) also occur in plural patterns4 as the examples in
Table 3 show.
4
We refer to those lexemes where the singular is identical or at least cognate in both
literary and colloquial Arabic.
TABLE 3 EXAMPLES OF DIFFERENCES BETWEEN PLURAL FORMS

IN LITERARY AND COLLOQUIAL ARABIC (IN ISRAEL)
Literary Arabic Colloquial Arabic (Israel) Gloss
br byra wells

ahur, uhr (u)hr months
numr nmura tigers
nuqs nawqes defects, faults
bah h rna bah h ra seamen
Such differences are, of course, in addition to basically phonetic

differences of cognates, existing in singular as well as in plural forms
see Table 4.
TABLE 4 PHONETIC DIFFERENCES BETWEEN COGNATE LITERARY

AND COLLOQUIAL ARABIC LEXEMES
Literary Arabic Colloquial Arabic (Israel) Gloss
ism usum name

far fr mouse
mintaqa mantiqa region
ibar ubar needles, injections
asnn snn teeth
daqiq daqyeq minutes
astida astze gentlemen, professors
In the verb system also certain verb patterns vary between literary and
colloquial Arabic (in Israel). See regular verb patterns such as literary
Arabic qadaracolloquial Arabic (in Israel) idir was able, literary
Arabic arafacolloquial Arabic (in Israel) iref knew, and their
respective non-past patterns: literary Arabic yaqdirucolloquial Arabic
(in Israel) yidar will be/is able, literary Arabic yarifucolloquial
Arabic (in Israel) yiraf will know/knows. Differences are also found in
verbs of C1//, e.g., literary Arabic yaxudu- colloquial Arabic (in Israel)
yxud/yxud/yxed will take/he takes/, C2w/y/ pattern, e.g., literary
Arabic xiftucolloquial Arabic (in Israel) xuft I was afraid, C3w/y
pattern, e.g., literary Arabic baqiyacolloquial Arabic (in Israel) baqa/
biqi he remained, literary Arabic qaraacolloquial Arabic (in Israel)
qara, qiri he read, and many more.
Differences of these kinds are well known from the literature, and in
a dictionary presenting both literary Arabic and colloquial Arabic (in
Israel) (such as Rosenhouse 2001, 2004) one at once sees the abundance
of these forms. However, the main goal of a dictionary study is word

meanings, and to this issue we now turn.
3. Semantic considerations of the differences between literary Arabic

and colloquial Arabic lexemes
3.1 The use of foreign lexemes

Like other languages, Arabic has absorbed many lexical items from
foreign languages from its classical past to the present. Since the sec-
ond half of the 20th century the major source of foreign loanwords in
Arabic has been the American English language, which has become the
modern lingua franca of the world. Before that, French, Italian, Spanish,
Turkish, Persian, Latin, Greek and others made lasting impacts on Ara-
bic vocabulary (Rosenhouse, to appear).
Literary Arabic was revived in the 19th century through the diverse
efforts of writers, journalists, scientists and linguists. These efforts led
to the establishment of academies for the Arabic language in Damascus
(1919), Cairo (1934), Baghdad (1947) and Amman (1976). Coining new
terminology for modern notions is one of the declared tasks of these
academies, as it is of many language academies elsewhere. The linguistic
methods applied include borrowing terms and integrating them into
the language phonologically and morphologically; extending meanings
of existing (Arabic) words; analogical creations based on existing
roots; translation of foreign words into Arabic (calque); and blending
(Stetkevych 1970, also quoted in Suleiman 2005, Shraybom-Shivtiel
2005). Such Arabicizing of foreign terminology goes on in other Arabic-
speaking countries too (e.g., Al-Qahtani 2002, for Saudi Arabia, and
Kharbush 2002, for Jordan see Badawi 1997). This activity has added
thousands of new words to the literary Arabic vocabulary, although
there is no full agreement between the various language institutes
(whatever their titles) in Arab countries and although many of these new
words remain buried in dictionaries without enjoying general public
use (Badawi, 1997). Whatever the distribution of lexical innovations
in literary Arabic vocabulary, this process clearly contributes to the
decrease in the number of foreign terms in literary Arabic, particularly
in the semantic fields of modern technology and science.
Unlike literary Arabic, colloquial Arabic is not subject to any official
language policy and it develops spontaneously, albeit under the effect of
internal and external circumstances. Accordingly, foreign terminology
flows freely into colloquial Arabic and is integrated in keeping with

the needs of its speakers (see Rosenbaum 2000/2). As noted, this
terminology now comes mainly from English, but residues of the effects
of other languages from earlier periods can still be found.
In Israel the situation is somewhat different from Egypt (as Rosenbaum
ibid. describes it), since in Israel Hebrew is the dominant language
(Rosenhouse forthcoming). Since the establishment of the state of Israel
in 1948 native speakers of Arabic in Israel have been in contact with
Hebrew officially and spontaneously. Since they use Hebrew daily, the
latter has a strong impact on their Arabic mother tongue. The fact that
they study Hebrew at school from grade three to the end of high school
reinforces their Hebrew proficiency. Hebrew has been the major source
of borrowing of foreign lexemes for colloquial Arabic in Israel for the
last five decades or so. English lexemes (including American-English)
also penetrate colloquial Arabic directly, or through Hebrew that often
serves as mediator (see, e.g., Abdeen 2004, Amara and Spolsky 1986,
Koplewitz 1989). A simple dictionary search soon reveals many cases
where literary Arabic uses Arabic words (often modern innovations)
while colloquial Arabic (in Israel) uses foreign words for the same items
(see Table 5). Literary and colloquial Arabic also share identical foreign
loanwords. Some of these words in colloquial Arabic (in Israel) may be
transferred directly from the foreign language, or indirectly via literary
Arabic, in particular when they refer to higher and modern lexical
domains such as politics, religion or other cultural areas, e.g., ittisl
communication, often by phone, ijr:t procedures (in court), asm
muddda antigens, isti niyy artificial, ihtiyt reserves, ih s iyyt
statistics, awz safar passport (see more examples in Table 6). The
best sources for such examples are newspapers and weekly magazines,
which usually use the literary Arabic variety nearest to middle Arabic.
On the other hand, a large share of lexemes of basic daily colloquial
Arabic (in Israel) vocabulary refers to proper names of products and
place names, which are not usually translated into Arabic, e.g., jeans,
coca cola, jeep.5
5
Usually, such words are not translated into any other language, though they may be
adapted to them phonologically and morphologically, as described above.
TABLE 5 EXAMPLES OF ARABIC LEXEMES IN LITERARY ARABIC

AND FOREIGN LEXEMES IN COLLOQUIAL ARABIC
(FOUND IN ROSENHOUSE 2001)
Literary Arabic words Foreign colloquial Translation

Arabic words
art an-nr fyz (electric) fuse

atra sandw sandwich
amat al-itil bgiyye plug (in a car)
tabtba rakt racket (in tennis game)
ajala ihtiytiyya spr spare (tiyre in a car)
miqwad strin steering (in a car)
muh arrik matr motor
mujhir mikroskb microscope
funduq otl hotel
TABLE 6 EXAMPLES OF IDENTICAL FOREIGN LEXEMES IN

LITERARY AND COLLOQUIAL ARABIC (FROM ROSENHOUSE 2001,
AND AL-AHRM, 4/3/2006)
al-aksada al-aksade oxidation

influenza at-tuyr Id. bird influenza
al-interbank ad-dlriyy Id. The Dollar Interbank
arf istitl ar-ray Id. archives of public
opinion polls
aristuqrtiyya Id. aristocracy
bank Id. bank
bra Id. beer
brtn Id. protein
btagz Id. gas stove, oven
daktr Id. doctor
ad-diktafn Id. dictaphone
dmuqrtiyy Id. democratic
elektrniyya elektrniyya electronic
mashariyya fza vase
anarl Id. general
hmglbn Id. hemoglobin
istd istd stadium
ihz al-mbil belefn6 mobile phone
6
Based on the Colloquial Arabic term for this device.
TABLE 6 (CONT.)

malyn Id. million
malynr Id. millionaire
mawqi gugl Id. Google site
milyr Id. milliard/billion
mda mda fashion
miknkiyy mekanki mechanic
msq klsk msq klskiyye classical music
tmtk tmtki automatic
al-tizm Id. autism
ar-rtr Id. Rotary (club)
sln br Id. saloon bar
sekretra Id. secretary
sigr Id. cigar
simfnia Id. symphony
sinryu Id. scenario
slkn silikon silicon
sirmik Id. ceramics
as-sundq al-aswad es-sandq el-iswad the black box in a vehicle
amb Id. shampoo
slaidt Id. slides
fniyya Id. Chauvinism
klta uklta chocolate
talifn (also htif) talafn telephone
zarfa zarfe giraffe
3.2 Literary Arabic words in colloquial Arabic and colloquial Arabic

words in literary Arabic
Diglossia has been often described in rather idealistic terms, as com-
prising two varieties, literary Arabic and colloquial Arabic. However,
more and more voices (in the 20th century) describe Arabic on a con-
tinuum of various degrees and types of intermingling elements of these
two poles or varieties (see Ayyoub 2002, Badawi 1973, Blanc 1960, Diem
1974, Ferguson 1996, Fischer and Jastrow 1980, Meiseles 1975, 1980,
Mitchell 1986, Kaye 1994, Holes 1995, Versteegh 1997, etc.).
Since literary Arabic is the prestigious variety of Arabic, while collo-
quial Arabic is hardly deemed (by its native speakers, even today) to
merit any official attention or reference, colloquial Arabic speakers will
naturally tend to borrow from literary Arabic more than literary Arabic
(writers or speakers) would tend to borrow from colloquial Arabic.7 This

prestige factor is also well known in language communities that borrow
from other languages (e.g. Owens 1988, Thomason and Kaufman
1988/1991, Thomason 2001, Weinreich 1967).
Moreover, since communicating in literary Arabic is obviously more
difficult for colloquial Arabic speakers than communicating in colloquial
Arabic, both lexical and grammatical elements from colloquial Arabic
slip into these speakers literary Arabic utterances by code-mixing and
codeswitching processes (cf. Haeri 2003, Meiseles 1975, 1980).
The varieties developed in Arabic (typical of the 20th century) are
entitled al-lua al-wust middle Arabic (Meiseles 1975, 1980, Holes
1995, Kaye 2002, Versteegh 1997), and educated spoken Arabic
(Badawi 1973, El-Hassan 1978, Mitchell 1986), among others. It has
been observed that the modern version of Middle Arabic establishes
bridges between literary Arabic and dialects (Ayyoub 2002). Moreover,
the deliberate use of dialect within the written text has resulted not in
the dialect as a distant variant but as a level of literary Arabic (Ayyoub
2002; see also Rosenbaum 2000/2, 2004, and Somekh 1991, 1993).
A major field of differences between literary and colloquial Arabic
vocabularies concerns the semantic fields of the items. literary Arabic
has been used for high scholarly fields since classical times, and still
today it is the carrier of modern scholarly terminology in the realms
of religion, culture, science and technology, namely areas not typically
considered part of the colloquial Arabic world (see examples in section
3.1). However, with the integration of these modern topics and concepts
in the speakers daily life, their lexical elements (from literary Arabic)
penetrate the daily discourse, via the mediating stage of code mixing
until they integrate in colloquial Arabic and become an inseparable part
of its vocabulary.8
An important part of colloquial Arabic oral communication and
vocabulary is often described as dealing with emotions and personal
needs (also in Holes 1995, Versteegh 1997). This is complemented by the
fact that personal topics are not normally discussed in literary Arabic.
7
That colloquial Arabic borrows and uses literary Arabic items more than literary
Arabic borrows from colloquial Arabic is readily seen in any dictionary.
8
For more examples of such lexemes see Diem 1974, 4647, section 9.54, and Abd
Al-ls introduction (1971, 7) to his dictionary of Colloquial Egyptian Expressions
Whose Origins are Arabic (i.e., cognates).
Although literary Arabic certainly has ways of emotional expression, as

written literary Arabic prose and poetry show, the general attitude to
emotion in literary Arabic, even in modern writers, is still generally not
as warm as their attitude to the flowing and rich colloquial Arabic style.
The difference between the serious, preaching style of literary Arabic
and the spontaneous colloquial Arabic speech has been demonstrated
and explained, for example, through the simultaneous use of literary and
colloquial Arabic in three Arab leaders speeches (Mazraani 1997). Arab
writers often complain that literary Arabic vocabulary and structure
need to be facilitated and made more natural, supple and personal,
like colloquial Arabic. For example, Tawfiq al-Hakim and Mahmud
Taymur write in the beginning of the 20th century about their struggle
to make their works natural by using colloquial Arabic words and
expressions (see for example Diem 1974, 117 ff.),9 and Mikhail Nuayma
(among others) puts in quotation marks the colloquial Arabic words
he uses in his early stories (e.g., sanatuha al-jadda, sat dahab), thus
marking their difference from the main story style. As a result of this
distinction, however, many Arab writers end up using colloquial Arabic
in their worksin drama, in novels and short stories, and always in
caricaturesto make their expressions more realistic as well as lively. In
Egypt this spontaneous development has reached the level of a writing
system (Rosenbaum 2000/ 2002, 2004).
Literary Arabic has developed and is now much more fluent and
easy than it was at the turn of 19th century, when the revival movement
was gaining power (Haeri 2003). Still, the general attitude to colloquial
Arabic and literary Arabic has remained largely unchanged.
3.3 Cognate literary/colloquial Arabic words and semantic differences

between them
The basic and largest part of the vocabulary in Arabic is shared by liter-
ary and colloquial Arabic, and includes pronouns, kinship terms, body
parts, social structure, animals, geographical objects, basic activities
(verbs), etc. These lexemes are cognate items, namely the same words
(expressed in the same consonantal roots), used in different forms
9
From another angle, Frayha (1955) describes his sons frustration and problems in
answering a simple question in literary Arabic. Frayha (1959) later suggested a method
for accomplishing the simplified language (although later on he stopped publicizing this
position).
(inflections, derivations) in literary and colloquial Arabic. The main dif-

ferences between literary and colloquial Arabic cognates seem to be not
semantic but phonological or morphological (see section 2 above), and
lexemes where the semantic notion is not shared by literary and collo-
quial Arabic are relatively few. Cadora (1979) analyzes the lexicon in the
Syro-Lebanese region, to which colloquial Arabic (in Israel) belongs,
and we therefore refer to it here. Cadora realizes that Swadeshs list
(1952), on which his work is based, includes words whose meanings are
so general that cultural innovation or discontinuation does not affect
them greatly (26), which reduces their validity. Still, he finds (27) in the
Syro-Lebanese varieties checked against this vocabulary list, that 151
(75.5%) of the 200 lexical items have non-contrastive compatibility (i.e.,
more or less, lexical similarity) This rate is even higher (165 or 82.5%) if
the dialect of Dr Ez-Zr is removed from the analysis. The comparison
of literary Arabic and Syro-Lebanese colloquial Arabic varieties reveals
an even higher compatibility, which on the average reaches 91% (29).
Comparing the studied Syro-Lebanese dialects (except Dr Ez-Zr)
among themselves, Cadora finds 96% non-contrastive compatibility
between each pair of dialects (32). The differences between these dialects
and more distant ones (Casablanca, Cairo, Jidda, Baghdad) are larger, of
course, but lets keep to our region.
Cadora prepared another list, based on Swadeshs list (1952), and an
unpublished list by Ferguson and Said (1958), with his modifications of
them. This new list finally had 101 contrastive compatible items. From
the results of this work, on the whole, he finds that Arabic tends to use
cognate words in literary and colloquial Arabic, while different words
(rather than cognates) occur in cases where the meanings differ between
literary Arabic and colloquial Arabic.
Differences between literary and colloquial Arabic cognates may
arise when the literary Arabic items have several semantic fields, not
all of which are used in modern colloquial Arabic lexical items. The
colloquial Arabic items may refer to one of the several semantic fields of
the literary Arabic item. Such cases may be due to the fact that modern
literary Arabic draws on a huge dialectal inventory from the past, which
is not known in all the colloquial Arabic dialect regions. Among the
most famous examples, is f, Literary Arabic (rarely used) to polish vs.
colloquial Arabic he saw (it is probably famous also because it appears
in Ferguson 1959). Other examples more specific to the region of
colloquial Arabic in Israel are ujra, in literary Arabic wages, colloquial
Arabic taxes (wages in colloquial Arabic in Israel is more often ajar,
ma or rteb, which are also used in literary Arabic); muallim in literary

Arabic means a teacher or instructor, but in colloquial Arabic teacher,
as well as a boss in a workshop, an office etc.10 Kabs(a) is in literary
Arabic pressure, attack, colloquial Arabic heavy (rain) (parallel to the
English expression its raining cats and dogs); fatina (literary Arabic)
/ fiten (colloquial Arabic)literary Arabic be clever, smart, colloquial
Arabic remember; qalbliterary Arabic heart, colloquial Arabic ones
internal part of the body, used also for the heart and entrails.
4. Conclusion
This chapter has focused on literary and colloquial Arabic vocabulary.

An important feature of colloquial Arabic vocabulary is that it is more
affected by literary Arabic lexemes than literary Arabic is by colloquial
Arabic lexemes (as expected). The reasons for this have been studied
elsewhere and we did not analyze them here. Suffice it to say that this
picture is largely due to the prestigious status of literary Arabic com-
pared to colloquial Arabic (see Rosenhouse forthcoming).
We have also seen that phonological and morphological features mark
many auditory / articulatory differences between literary and colloquial
Arabic. Our semantic comparisons between literary and colloquial
Arabic lexemes have shown both diverging and converging trends of
development in each of these varieties.
A relatively small part of the sum total of Arabic lexical items is
composed of foreign loanwords that penetrate Arabic due to the
cultural trends of modernization / Westernization. In this section of
the vocabulary we also note that colloquial Arabic uses more foreign
loanwords (in Israel sometimes borrowed via Hebrew) than literary
Arabic does. This state is apparently due to the fact that in many cases
literary Arabic uses translations (calque) and newly coined terms rather
than the foreign ones used in colloquial Arabic (see Abu-Haidar 1992,
Shraybom-Shivtiel 2005).
We did not find many colloquial Arabic items in Israeli newspapers.
Those found occurred mainly in advertisements and caricatures
where foreign loanwords relating to politics, shopping and economics,
10
Wehr (1971) gives also master of trade as one of the meanings of this word.
sports and entertainment, etc., are used, as in many other language

communities in the world. On many hoardings (especially in Arabic-
speaking townships or villages in Israel), however, colloquial Arabic (of
Israel) vocabulary and structures play an important role, even if they
are not the norm. A considerable measure of linguistic influence on the
general public comes from mass media such as radio and TV, where
advertisements are orally and visually published. These recurring ads
affect the public subconsciously so that this source of mixture between
literary and colloquial Arabic, which was on the increase all through
the 20th century, has been going on up to now. Abu-Haidar (1992, 104)
sums up the effect of modern standard Arabic on dialect convergence
(in Baghdad) by saying that certain lexemes (in their MSA forms)
are now part of Muslim Baghdadi and Christian Baghdadi speakers
everyday experiences (105). She adds: salient features of Muslim
Baghdadi and Christian Baghdadi have for a long time constituted
closely defined boundaries separating the two dialects from each other
and from Modern Standard Arabic. At the present time, however,
these boundaries are shifting as some salient features become obso-
lete and more unassimilated Modern Standard Arabic forms are diffused
into both dialects (p. 105). In her conclusion Ayyoub (2002) states:
From its journey across the 20th century, the Arabic language carries two
remarkable traits: never in the past have the written and spoken languages
been intricately intertwined; never before have the centrifugal forces of
standardization affected the dialects as much as now . . .
A separate question is, however,
what will become of the Arabic language in the future? If the past century
has sensibly modified the modalities of functions between literary and
dialectal Arabic, present for more than one and a half millennia, multiple
paths remain possible (Ayyoub 2002).
Literary Arabic is a source of fascination (for researchers and native
speakers alike, if we may add) because of its relation to both the secular
and the sacred (Haeri 2003). Moreover, economic constraints and
globalization alike favor literary Arabic, as a common language for a vast
geographic area, more advantageous than multiple languages (Ayyoub
2002). Ayyoub (2002) poses additional questions about the future of
literary Arabic and colloquial Arabic in the 21st century: Will a form
of Middle Arabic be the future? Will the dialects become the national
languages of tomorrow?
We see that the diglossic makeup of Arabic is changing and evolving

all the time, including the structure and vocabulary of this language.
In spite of the generally unfavorable attitude to code mixing/switching
between some foreign language and colloquial Arabic, these processes are
apparently inevitable under the modern global village circumstances.
These foreign lexical elements penetrate colloquial Arabic, the receiving
language (or matrix language, according to Myers-Scotton 1993), and
usually become an integral part of it. A similar process takes place in
literary Arabic, although apparently at a smaller (slower?) rate. Since
this process existed also in the past (see, for example, Al-Falay, 1996,
chapter four) without lessening the vitality of the Arabic language,
we assume that the similar process of modern times will not lead to
an entirely different future development. In spite of the huge number
of lexical items (including borrowed and innovated words) and their
semantic fields in a language like Arabic, lexemes are an open class and
may be added, exchanged, renewed or discarded according to need.
When thinking of structural changes of Arabic due to code mixing or
code switching phenomena, we do not consider the lexical developments
of literary and colloquial Arabic as described here to be the straw that
breaks the camels back. Moreover, the features and examples in this
chapter do not differ much, in principle, from those found in other
Arabic dialects (Boucherit 2000, Heath 1989, Owens 1998, Taine-Cheich
2000). We therefore assume that the Arabic language will continue its
basic path sweeping along all the side trimmings noted in the changes
of its vocabulary. We also support the view that the relation between
the studied linguistic varieties in Arabic (Arabic dialects and literary
Arabic) is similar in many respects to the relations between different
languages in bilingual situations (Fasold 1984).
The above analysis may be seen within the framework of phenomena
of language interference or language contact according to one linguistic
approach, in addition to bilingual code mixing according to another.
We hope the examination of the intertwining vocabulary of literary
and colloquial Arabic will contribute to the attempts to improve
our understanding of the Arabic language. Since the paths that the
Arabic language will take do not have to be identical to the paths of
the past, further attention to the issues raised here should prove to be
interesting.
5. References
Abd al-l, Abd al-Munim Sayyid. 1971. Mujam al-alfz al-mmiyya dt al-usl
al-arabiyya. Cairo: Maktabat Nahda al-Misriyya.
Abu-Haidar, Farida. 1992. Shifting Boundaries: The effect of modern standard Arabic
on dialect convergence in Baghdad. In Broselow et al., Perspectives on Arabic
Linguistics IV, 91106.
al-Falay, Ibrh$m Salh. 1996. Izdiwjiyyat at-lua: An-nazariyya wa-t-tatbiq. Ar-Riyd
King Sad University.
Al-Hassan, S.A. 1978. Educated spoken Arabic in Egypt and the Levant: A critical
review of diglossia and related concepts. Archivum Linguisticum, 8(2):112132.
Altoma, Salah J. 1969. The Problem of Diglossia in Arabic: A comparative study of classical
and Iraqi Arabic. Cambridge: Harvard University Press.
Arnold, Werner, and Hartmut Bobzin, eds. 2002. Sprich doch mit deinen Knechten
Aramisch, wir verstehen es! 60 Beitrge zur Semitistik Festschrift fr Otto Jastrow zum
60. Geburtstag. Wiesbaden: O. Harrassowitz.
Ayyoub, Georgine. 2002. An Odyssey of Words: Evolution of the Arabic language in the
20th century. Al-Jadid Magazine 8:40 (Summer).
Badaw$, Sa$d Musta f. 1973. Mustawayt al-lua al-arabiyya al-musira f misr. Cairo:
Dr al-Marif.
Badawi, Mohamed. 1997. Probleme des Fachwortschatzes im Arabischen dargestellt
insbesondere an der Terminologie der Teleinformatik. Hildesheim: Georg Olms Verlag.
Bhatia, Tej. K., and William. C. Ritchie, eds. 2004. The Handbook of Bilingualism. New
York: Blackwell.
Blanc, Haim. 1960. Stylistic Variation in Spoken Arabic: A sample of interdialectal
educated conversation. In C.A. Ferguson, ed., Contributions to Arabic Linguistics,
81156.
. 1964. Communal Dialects in Baghdad, Cambridge, Mass.: Harvard University
Press.
Blau, Joshua. 1977. The Beginnings of the Arabic Diglossia. A study of the origins of
Neoarabic. Afroasiatic Linguistics 4:428.
Boucherit, Aziza. 2000. Reflexions sur le contact de langues partir du cas dAlger. In
Manwel Mifsud, ed., Proceedings of the Third International Conference AIDA, Malta,
8388.
Bousaffara-Omar, Naima. 2005. Diglossia. In Kees Versteegh et al., eds., Encyclopedia
of Arabic Language and Linguistics, vol. 1, 62937.
Broselow, Ellen, Mushira Eid, and John McCarthy, eds. 1992. Perspectives on Arabic
Linguistics IV, Papers from the Fourth Annual Symposium on Arabic Linguistics.
Amsterdam/Philadelphia: J. Benjamins.
Cadora, Frederic. 1979. Interdialectal Lexical Compatibility in Arabic: An analytical study
of the lexical relationships among the major Syro-Lebanese varieties. Leiden: E.J. Brill.
Diem, Werner. 1974. Hochsprache und Dialekt im Arabischen: Untersuchungen zur
heutigen arabischen Zweisprachigkeit. Wiesbaden: F. Steiner.
Elgibali, Alaa, ed. 1996. Understanding Arabic: Essays in Contemporary Arabic Linguistics
in Honour of El-Said Badawi. Cairo: American University of Cairo Press.
Fasold, Ralph, W. 1984. The Sociolinguistics of Society, Cambridge: Blackwell.
Ferguson, Charles Albert. 1959. Diglossia. Word, 15: 325340.
. 1996. Epilogue: Diglossia revisited. In Alaa Elgibali, ed., Understanding Arabic: Essays
in Contemporary Arabic Linguistics in Honour of El-Said Badawi, 4967.
. (ed.). 1960. Contributions to Arabic Linguistics. Cambridge, Mass.: Harvard
University Press.
, and Majid Said. 1958. Lexical variants in Arabic dialects (unpublished).
Fernandez, Mauro. 1993. Diglossia: A Comprehensive Bibliography, 19601990 and

Supplements. Amsterdam: J. Benjamins.
Fischer, Wolfdietrich, and Otto Jastrow. 1980. Handbuch der arabischen Dialekte.
Wiesbaden: O. Harrassowitz.
Furayha, An$s. 1955. Nawa arabiyya muyassara. Beirut: Dr a%-%aqfa.
. 1959. Tabst qawid al-lua al-arabiyya al usl jadda: iqtir wa-namdij.
Beirut: American University of Beirut.
Fck, Johann. 1950. Arabiyya: Untersuchungen zur arabischen Sprach- und Stilgeschichte.
Berlin: Akademie-Verlag.
Geva-Kleinberger, Aharon. 2004. Die arabische Dialekte der Stadt Haifa in der ersten
Hlfte des 20. Jahrhunderts. Wiesbaden: O. Harrassowitz.
Haeri, Niloofar. 2003. Sacred Language, Ordinary People: Dilemmas of culture and politics
in Egypt. New York: Palgrave Macmillan.
Hary, Benjamin. 1996. The importance of the language continuum in Arabic
multiglossia. In Alaa Elgibali, ed., Understanding Arabic: Essays in Contemporary
Arabic Linguistics in Honor of El-Said Badawi, 6990.
Heath, Jeffrey. 1989. From Code Switching to Borrowing: Foreign and diglossic mixing in
Moroccan Arabic. London: Kegan Paul International.
Hetzron, Robert, ed. 1997.The Semitic Languages. London: Routledge.
Holes, Clive. 1995. Modern Arabic: Structures, functions and varieties. London: Longman.
Jastrow, Otto. 1978. Die Mesopotamisch-arabischen qeltu Dialekte. Vol. 1: Phonologie und
Morphologie. Wiesbaden: O. Harrassowitz.
Kaye, Alan S. 1994. Formal vs. Informal Arabic: Diglossia, triglossia, tetraglossia, etc.,
polyglossia- multiglossia viewed as a continuum. Zeitschrift fr Arabische Linguistik,
27:4766.
. 2002. Diglossia. The state of the art for the new millennium. In W. Arnold and
H. Bobzin eds., Sprich doch mit deinen Knechten Aramisch, 37988.
, and Rosenhouse, Judith. 1997. Arabic dialects and Maltese. In R. Hetzron, ed.,
The Semitic languages, 263311.
Koplewitz, Emanuel. 1989. The use and integration of Hebrew lexemes in Israeli spoken
Arabic. Multilingual Matters 71.181195.
Levin, Arieh. 1995. Grammar of Spoken Arabic in Jerusalem. Jerusalem: Magnes Press
(in Hebrew).
, and Friedmann, Yohanan, eds. 2004. Studies in Honor of Moshe Piamenta.
Jerusalem Studies in Arabic and Islam 29, Jerusalem: Magnes Press.
Mazraani, Nathalie. 1997. Aspects of Language Variation in Arabic Political Speech
Making. Richmond, UK: Curzon.
Mejdell, Gunvor. 2002. Features of lua wustmixed discourse in spoken Arabic of
Egypt. In A. Youssi et al., eds. Aspects of the Dialects of Arabic Today, 31728.
Meiseles, Gustav. 1975. Oral Literary Arabic: Its main characteristics in speech and
reading, Ph.D. thesis, The Hebrew University in Jerusalem (in Hebrew).
. 1980. Educated spoken Arabic and the Arabic language continuum. Archivum
Linguisticum 11:118143.
Mifsud, Manwel (ed.). 2000. Proceedings of the Third International Conference AIDA,
Malta, 29 March2 April 1998. Paris: Association International de Dialectologie
Arabe.
Mitchell, Terrence F. 1986. What is educated spoken Arabic? International Journal of
the Sociology of Language 61, 732.
Monteil, Vincent. 1960. Larabe moderne. Paris: C. Klincksieck.
Myers-Scotton, C. 1993. Duelling Languages: Grammatical structures in code-switching.
Oxford: Clarendon Press.
Nuayma, M$x$l. 1961.6 Kna y m kna: Short Stories, Beirut.
Owens, Jonathan. 1998. Neighbourhood and Ancestry: Variation in the spoken Arabic of
Maiduguri, Nigeria. Amsterdam/Philadelphia: John Benjamins.
al-Qahtani, S. 2002. Arabization and the Theory of Language Planning: An Applied Study
of Terminology Arabization in Saudi Arabia. Beirut: Center for Studies of the Arabic
Union.
Ritchie, W.C., and Tej K. Bhatia,. 2004. Social and psychological factors in language
mixing. In T.K. Bhatia and W.C. Ritchie, eds. The Handbook of Bilingualism, 336352.
Rosenbaum, Gabriel M. 2000/2. Do you parler Arabi? Mixing colloquial Arabic
and European languages in Egyptian literature. Materiaux Arabes et Sudarabiques
(Nouvelle Serie), 10:1147.
. 2004. Egyptian Arabic as a written language. In A. Levin and Y. Friedmann, eds.
Studies in Honor of Moshe Piamenta, 281340.
Rosenhouse, Judith. 1984. The Arabic Bedouin Dialects: General problems and a close
analysis of North Israel Bedouin Dialects. Wiesbaden: O. Harrassowitz.
. 2001. Trilingual Practical Dictionary: Hebrewliterary Arabic-colloquial Arabic.
Rosh Ha-ayin: Prolog.
. 2002. Phonetic trends of colloquial Arabic dialects in Israel. In W. Arnold and H.
Bobzin, eds., Sprich doch mit deinen Knechten Aramisch, 599611.
. 2004. Trilingual Practical Dictionary: Literary Arabic-Hebrewcolloquial Arabic ,
Rosh Ha-ayin: Prolog.
. Forthcoming. The English language and its impact on Arabic in Israel. In. J.
Rosenhouse and R. Kowner, eds., Globally Speaking.
, and Mira Goral. 2004. Bilingualism in the Middle East and North Africa: A focus
on the Arabic-speaking world. In T.K. Bhatia and W.C. Ritchie, eds. The Handbook
of Bilingualism, 835868.
, and Rotem Kowner, eds. Forthcoming. Globally Speaking: Motives for Borrowing
English Loanwords in World Languages. Clevedon, UK: Multilingual Matters.
Shraybom-Shivtiel, Shlomit. 2005. The Revival of the Arabic Language as a Mission of the
National Ideology in Egypt. Jerusalem: Magnes Press (in Hebrew).
Somekh, Sasson. 1991. Genre and Language in Modern Arabic Literature. Wiesbaden:
O. Harrassowitz.
. 1993. Colloqialized fush $ in modern Arabic prose fiction. Jerusalem Studies in
Arabic and Islam, 16:17694.
Stetkevych, J. 1970. The Modern Literary Arabic Language: Lexical and stylistic devel-
opments. Chicago: University of Chicago Press.
Suleiman, Yaser. 2006. Arabiyya. In Kees Versteegh et al., eds. Encyclopedia of Arabic
Language and Linguistics, vol. 1, 173178.
Swadesh, Morris. 1952. Lexico-statistics dating of pre-historic ethnic contacts with
special reference to North American Indians and Eskimos. Proceedings of the
American Philosophical Society, XCVI:452463.
Taine-Cheich, Catherine. 2000. Reflexions sur le statut des emprunts dans les langues
mixtes partir du cas mauritanien. In M. Mifsud, ed. Proceedings of the Third
International Conference AIDA, 10712.
Thomason, Sara Grey. 2001. Language Contact: An Introduction. Washington D.C.:
Georgetown University Press.
, and Terrence Kaufman. 19881/19912. Language Contact, Creolization and Genetic
Linguistics. Berkeley: University of California Press.
Versteegh, Kees. 1984. Pidginization and Creolization: The Case of Arabic. Amsterdam-
Philadelphia: J. Benjamins.
. 1997. The Arabic Language. Edinburgh: Edinburgh University Press.
, et al., eds. 2006. Encyclopedia of Arabic Language and Linguistics. Vol. 1. Leiden:
E.J. Brill.
Wehr, Hans; J. Milton Cowan, ed. 1971. A Dictionary of Modern Written Arabic.
Wiesbaden: O. Harrassowitz.
Weinreich, Uriel. 1967. Languages in Contact: Problems and findings. The Hague:
Mouton.
Xarb&, A. 2002. (arakat at-tarb f l-urdunn. Amman: Ministry of Education.
Youssi, Abderrahim, Fauzia Benjelloun, Mohamed Dahbi, and Zakia Iraqui-Sinaceur,
eds. 2000. Aspects of the Dialects of Arabic Today: Proceedings of the 4th Conference
of the International Arabic Dialectology Association (AIDA), Marrakesh, April 14.
Rabat: Ampatril, 317328.
EVERYTHING YOU ALWAYS WANTED TO KNOW ABOUT
L, YIL TO SAY IN EGYPTIAN ARABIC*
Manfred Woidich
University of Amsterdam
1. Introduction
The idea of contributing to the Festschrift for our esteemed friend and
colleague Kees Versteegh with an article on the verb l, yil (gl, yigl)
to say in Egyptian Arabic dialects and its various idiomatic uses, came
to me when I was reading a draft of his lemma Serial Verbs, which he
had written for the Encyclopedia of Arabic Language and Linguistics.1 In
this lemma, he reconsiders my view of items like (01) as an originally
paratactic but now grammaticalized construction in order to introduce
a pseudo-complement,2 and prefers to regard it as serialization, that is,
as a serial verb construction, albeit not without hesitation.
(01) huwwa raddi alk allak ?
huwwa radd al-k l-l-ak
he answered to-you said-to-you what
What did he answer you? LAB 118,8
Unfortunately, I do not share this view and I shall reaffirm my position
in section 4 below. Since this pseudo-complementation is not the
only case for l, yil to appear in contexts and functions that involve
grammaticalization, it seemed appropriate to give an overview of several
other cases where l, yil clearly has lost its original lexical meaning
by a process known as bleaching and acquired new meanings and
developed into a function word. After making some brief remarks on
the syntactic behaviour of l, yil as a lexical verb in section 2, several
steps and ramifications in the history of l, yil will be described in
* I should like to thank Rudolf de Jong for going through an earlier draft of this arti-
cle and giving me some valuable hints. Needless to say, any remaining errors are mine.
1
To appear as Versteegh 2007.
2
Woidich 2002, in particular pp. 183184.
676 manfred woidich
section 3 based on data collected over the last 40 years mostly from
written sources but also from some recordings of rural dialects. The
starting point of these developments apparently is the use of l, yil by
the speaker to introduce reported speech3 on discourse level4 within a
pragmatic strategy, namely introducing a direct or an indirect quote, be
it the speakers own words or the speech of somebody else, in order to
show rejection and non-acceptance of an utterance, to give reasons for
acting in a certain way, to explain intent, to make a comparison between
two things or for other reasons. The final sectionsection 5deals
with a lexical aspect and gives some examples of the use of l, yil with
vocatives and in delocutive derivations.
2. On propositional level: direct vs. indirect speech
Here I will briefly describe a syntactical point, that is, the introduction
of direct and indirect speech. Direct or quoted speech follows l, yil
as an asyndetic sentence without a complementizer:
(02) ana mu ayllik hti )a)a
Did I not say to you, bring an ashtray? RUH 112,-1;
(03) alitlaha tali uudi maya
She said to her, sit down with me! FWQ 107,-7
(04) ns kitr bau ylu f-nafsuhum w ana mli
Many asked themselves, what business is this of mine? BAHN 122,9
A sentence reporting indirectly what was said, however, may either
be introduced by the complementizer inn (05, 06) or be connected
asyndetically (07, 08):
(05) iddakatra lu inn abak tabna wayya
The doctors said that your nerves are a little bit off HAM 95,7f
(06) l innu aynm andina llla diyyat wi bass
He said that he would sleep with us only that night LAB 16,8
(07) **a kkallim wi l aybt ba++a
Daddy called and said that he would spend the night out of the house
LAY 113,3
3
Following Gldemann et al. (2002, viii), I use the term -reported speech as a gene-
ric term for both direct and indirect speech.
4
In many languages, the verb -say constitutes a source of various pathways in gram-
maticalization that can lead to distinct types of function words, see Heine et al. (1993),
and Heine et al. (2002, 261 ff ).
l, yil to say in egyptian arabic 677
(08) dil kn l ayxud agza nnaha+da w ayfu aa f-Alma

dil has said, he would take the day off and they would have a look at
the apartment in Alm'a WAZ 374,12
3. l, yil (gl, yigl) on discourse level
Reported speech introduced by l, yil (gl, yigl) serves in various ways
and on different levels for text constitution and as a discourse device.
In narration it is commonly used for telling a story (3.1) by means of
reporting it as a dialogue. In a similar way, reasons and intentions may
be presented to the hearer in the form of reported speech (3.2), that is,
in the form of the speakers own words. Whereas the last two uses do
not exhibit any syntactic differences in comparison to the normal one
in a simple proposition (see 2), there are phrases formed with l, yil
that deviate in certain respects from the normal usage and suggest that
the items concerned areor at least have started to begrammaticalized.
So, ulti serves as a turn-giving device asking for a positive reaction
(3.3), l marks a short-cut (3.4), l serves as a marker of incredulity
(3.5), tiul as a comparison marker (3.6), baullak as a turn-
taking device (3.7), biylu as a reference to hearsay and general
knowledge (3.8).
3.1 Narrative device: story-telling by means of reported speech

A very common way of telling a story is to present it as a sequence of direct
utterances, that is, as a constructed dialogue in which each utterance is
introduced by quotative l, yil (gl, yigl). The phrases are juxtaposed
asyndetically and are not connected by means of wi and. Speakers make
extensive use of this means, as can be seen in any collection of texts of
Arabic dialects recorded in the field. For the audience, a story structured
step by step in this way is more insightful and easier to follow than a
purely descriptive narration, since the narrator gives the story a structure
that conveys liveliness and persuasion and involves the hearer. 5 A short
passage from a Bri text may illustrate this:
5
Involvement of the hearer is considered one of the most important pragmatic stra-
tegies in Georgakopoulu et al. (2004, 136f).
678 manfred woidich
(09) glat isma&&a alk y-w+ad, isma&&a alk mlak. gultilha-na

anaalam wu nji an nafsi anijlib min ilba+bax an nissabba,
rut wagat min fg ilmanma. galatli kaamt rijli ya wldi. gultilha
ana gultluk ijilbi wa+y? wu baadn galatli +aa-txallni nmt
kamn wu rijli maksra, gultilha lyi li. galatli jaddak Abda&&a
lGum mt huwwa w rijla maksra. gultilha min ? galatli lamma
knaw aytiu fi a&amt, gultilha k ytiu fi a&amt galatli
l nta mi rif ilikkwa? gultilha l ma-ufth wala xabir ja.
galatli badri, )irg issuww diyya . . .
She said: May God keep you safe, my boy, may God protect you!
Whats wrong with you? I said to her: I was dreaming and [in my
dream] I jumped involuntarily from the footbridge to have a swim.
So I fell from the sleeping roof. She said to me: You sprained my
ankle, my boy! I said to her: Did I tell you to jump after me?
And then she said to me: You are going to let me even die from
a broken leg. I said to her: It was not my fault. She said to me:
Your grandfather Abda**a lGum died from a broken leg. I said to
her: How come? She said to me: When they were pulling on the
colossoi.6 I said to her: How come they pulled on the colossoi? She
said: Why? Dont you know the story? I said to her: I did not see
it, nor do I know anything. She said to me: In former times, this
tourists way . . . (Woidich 1980, 236,2731)
This use of directly reported (quoted) speech introduced by l, yil
(gl, yigl) is characteristic of the narrative style and gives a dialogue
structure to the whole narrative or parts of it. The narrative becomes
like a theatre play delivered by the speaker, and indirect speech and
complicated subordinate structures are avoided. There is no sign of
grammaticalization, and l, yil (gl, yigl) is subject to inflection
following the logic of the story and remains in its position in front of
the quoted speech.
3.2 Argumentation by quoted speech

In a similar way, comments on a statement may be given by the speaker
to the hearer in the form of direct speech, mainly as the speakers own
words, presenting the argumentation as an inner monologue. Such self-
quotations are introduced by ult I said and follow a statement. Speakers
report to the hearer the reasons for decisions or conclusions they made,
6
ia&amt *nm (literally the idols) is the Bri word for the Colossoi of Memnon
on the West Bank of Luxor.
or the reasons they had for acting in a particular way. 7 This comment
explains either why what is described in this statement occurred (10,
11) or what the intention behind it was or is (13, 14, 15). In other cases
it tells the hearer what follows for the speaker from what is described
in the statement (12). Again, presenting the argumentation as internal
dialogue makes it more insightful to the hearer and easier to follow and
involves him directly.8 Using ult in this way is very common in standard
Egyptian Arabic, in both its spoken and its written form. If an intention
is involved, the verb is in the y-imperfect.
(10) ruti +af a idayya )awwli mdarya wii, ulti la-ykn alh afrt ismu
i/rab
I immediately lifted my hands up to protect my face. I said, it may have
got an evil spirit on it, I am afraid, whose name is i/rab [hit!] LAB
214,4 (reason)
(11) ana grt ulti yimkin miawwi ga kida walla kida min wa+a-mmi
I hurried up, I said, perhaps he has somehow saved something behind
the back of my mother HAM 28,1 (reason)
(12) laetkum )ayyibn ulti ya bitt uudi
I found you to be good people, I said, stay here, girl! RUH 30,5
(consequence)
(13) mi gayyili nm ult atmaa
I could not sleep, so I said, let me go for a walk SIB 89,3 (intention)
(14) il0ayya btizaf kamn ala irka. ulti nsibha abli ma ti+a
The water is creeping towards the company, too. I said, let us leave
before it is flooded FWQ 50,1 (intention)
(15) lammt uwayyit xaab kasri ala ittitn awli wi ult awalliha addaffa bha
I collected a little bit of broken wood and some corn cobs and said, let
me set fire to it to warm me up MHR 7,6 (intention)
(15) shows that a syndetic connection with wi and is also possible. (16,
17) provide evidence that this discourse strategy is not confined to a
quotation of the speaker himself using the 1st person but may be applied
in the same way in narratives in the 3rd person as well. While we cannot
see in (16) whether direct or indirect speech is involved, the use of a 1st
pl. verbal form in (17) signals direct speech:
7
For similar functions of self-quotation in German (i.e. reporting on decisions, rea-
soning, intention), see Golato 2002. Further development of items meaning ;saying
i.e. original quotation markers to a grammatical markers expressing causality oris
widespread in the languages of the world, see e.g. Lord 1993, 177 for se say in Twi and
for the Turkish gerundium diye saying, see Kissling 1960, 191; for Bengali bole having
said see Chisarik et al. 2003.
8
See Georgakopoulu et al. 2004, 136f.
680 manfred woidich
(16) bai a kkanaba lwa++aniyya ma-lta ayyuha ga, zaar a kkursiyyn

iluddamiyyn iltahum ya mawlya zayyi ma xalatini, l tiba an)a
lli wa+a
He looked at the rear seats, but did not find anything, he glanced at the
front seats, but found them empty, he said, it must be the boot LAB
59,4 (conclusion)
(17) fa xad ba/tih wu , a+abiyya mi lminim gl nifu (ilmi Ysif yikn
iann
He set out and what? [he took] a car from the mines, he said, let us have
a look at ilmi Ysif, perhaps hes gone mad (Ba<ariyya: Manda)
(intention)
3.3 ulti what did you say? as a discourse routine

ulti ,literally What did you say?immediately follows a statement
conveying a suggestion or presenting an astonishing fact to the hearer
(21). ulti is connected to this statement by the intonational contour.
At first glance, the perfect ult makes it look like a request to repeat
something that has been said, pretending that a comment has been
overheard. But this is not the case since ulti is closely connected to
the statement with no pause and there is no time for the hearer to show
any reaction to this statement. Rather, it functions as a turn-giving
discourse marker, which asks for a positive comment on the statement
that is presented hereby as surprisingly good news to the hearer. For
that reason, it should be translated as How about that! which should
be understood as Arent you surprised and am I not to be praised?
Examples:
(18) tikkaffil inta bi maawr il-iyl wi ddurs w ana astannh, ulti ?
You are put in charge of the errands for the children and lessons, and I
wait for him, (so) what do you say now?! BTR 11,3
(19) wala yhimmak ya (sn xui ala lbutk xud illi yigibak wi kullu bi ttas)
tidfa tultumt gin dilwati wi lbi bi ttas) ilmirayya ala irn ah+i
xamas t-alf gin, ulti ?
Dont mention it, isn, go into the boutique, take what you like and
pay for everything in instalments; you pay 300 pounds now and the rest
in easy instalments, 5000 pounds over 20 months, so how about it?!
RAS 13,5
(20) tir ilwat illi yigibak, fi nnr, fi //alma. illi tfu, ulti ?
You go whenever you like, with daylight, in the darkness, just as you see
it (whenever it is convenient for you), isnt this a good suggestion? SIK
106,10
(21) ana tnzilti an ai fi kkaka li x Ima, ultu baa?
I gave up my rights to the cake in favor of x Ima, what do you say
now? MAX 93,4
Given this meaning, we would expect a continuous form of the verb

such as til or bitl , rather than of the perfect ulti . But replacing
the perfect here with an imperfect, ulti would mean losing its special
effect, which originates in a violation of the discourse record principle:9
the speaker acts as if his statement had already been commented on,
thereby constructing a reality that contradicts the speaker/hearer record
of the discussion so far and attracts the attention of the hearer and makes
it more relevant to him. At the same time, the rhetorical intonation of
ulti suggests that the hearer is assumed to consider this statement
as positive and to react accordingly. Apparently, this discourse routine10
did not yet lead to grammaticalization and to the forming of a particle,
since ulti still follows the rules of concord, as can be seen in (21). (22)
shows that giving the turn may be done not only by the speaker but also
by a third participant in the discourse:
(22) : ana mumkin anm uddm ilbb ilxargi laddi ma lmukila titall,
bass tiwidni nni ma-f addi yxui min innaya ttanya li lbt: ulti
ya Ar? A: ittafana.
: I can sleep in front of the outer door until the problem is solved.
You only have to promise me that nobody will enter the house from the
other side.: What did you say [what do you mean], Ar?A:
Agreed! MUW 134,4.
Again, the use of perfect by in (22) makes no sense and ulti has to be
seen as a turn-giving discourse marker.
3.4 l say! as a discourse marker

l say!the masculine imperative of l, yilmay be used in two
different ways. Within a text it introduces what a longer explanation or
description boils down to and serves as a short-cut in order to finish a
topic and to switch to another one. It corresponds to in short or to cut
a long story short in English, as in:11
(23) )iliti ba+a wi lwd iabi huwwa lli tabas bidli, a&&h yir/a annu, l
amadti +abbina lli )liti b gildi
9
See Schwenter et al. 2005 for a case of discourse record manipulation regarding
English too.
10
This is why I prefer to call it a routine, i.e. a discourse technique to reach a certain
goal, and not a marker, which in my view would include some grammaticalization as is
the case with l say!; see 3.5.
11
There are other markers with a similar function such as ur ikkalm, ilad,
nahaytu.
682 manfred woidich
I was found not guilty and the apprentice was jailed instead of me, may
God bless him, in short, I praised God that I had come away safely DAR
71,10
(24) wi msik sikkna yiz yiui bha ba)ni. girt minnu. )r wa+ya. l inns
ah
With a knife in his hand, he wanted to rip my belly open. I ran away,
he went after me. To cut a long story short, people held him back DAR
93,8
There is no evidence for female li being used in this way, but apparently
the plural lu serves the same purpose when several persons are
addressed. So l in this case should be considered not as a marker but
as a routine:
(25) inns yadb simit fh markib mal, risyit ala baladhum wi dl gary
an yitiru. lu awwil markib itbit fi sat
The people had only just heard that a salt ship had moored in their town
when they came running in order to buy. In short, the first shipful was
sold within hours MAL 36,8
(26) aadit ah+i w ah+i w ah+. lu iblit
She stayed a month and a month and [another] month. In short, she got
pregnant MAL 39,33
Another use of l say! derives from a discourse routine, that is, self-
correction in the form of a request to the hearer to correct a piece of
informationvery often numerical informationadduced by the
speaker. As the speaker seeks confirmation from the hearer, he suggests
that he is not sure about his estimate and that this information is not
100% reliable. Within a sentence l may thus indicate an approximation,
and correspond to perhaps; like, for example like the English: the whole
affair lasted, say, 10 minutes, as can be seen from (27). l underwent
grammaticalization and is frozen as a particle, since neither the feminine
nor the plural form are possible here and it is moved into the sentence,
standing for example in front of the direct object (29) or an adverbial
expression (30). More often than not, it combines with ygi about as a
reinforcement of this approximative meaning (28, 29, 30):
(27) ilamaliyya atxudlaha talat t-iyym l a+baa
The operation will take three days, perhaps four YUN 72,9
(28) km ya Nabawiyya km?l ygi irn gin aw akta+
How many, Nabawiyya, how many?Perhaps about twenty pounds or
more RUH 137,7
(29) iggamustn ayilibulna kulli ym l ygi mt klu laban
The two buffalos will give us about 100 litres of milk a day HAM
90,4
(30) wi lamma ifyu aadti ala fariti l ygi sa, itnn

And when it had finished, I sat down on my bed for an hour or two
RUH 7,7
3.5 Discourse marker for rejection and non-acceptance l and l

The entry l in Hinds et al. (1986, 722b) gives l as a modal of
incredulity with just imagine! fancy that! can you credit it! as
translation equivalents. And ll nonsense! baloney! humph!,
which introduces and closes a word or sentence as an exclamation of
jeering disbelief.12 I shall try to give a more precise description of its use
and at the same time sketch the historical development.
As the examples below show, an l introduces or follows the reported

utterance made by somebody else and urges the hearer to interpret this
utterance as rejected and not acceptable to the speaker. It functions thus
as a discourse marker and is fully grammaticalized as it is no longer
inflected (31, 33, 35). As becomes clear from the examples, it is indirect
speech here, not direct speech.
(31) Munra: bal li a/ritik ana mi ataal andi add, ana mi xaddma.
Bahga: mi xaddma! samn! l ma-h xaddma
Munra: I am telling you, I shall not work with anybody, I am not a maid.
Bahga: Not a maid! Did you hear! Not a maid, incredible! NAS 81,4
(32) ilminayyil ibn i))abbxa bn Ummi Anwa+, l gayy yux)ub Gamalt
This damned son of the cook, Umm Anwars son, incredible! He is
coming to ask for the hand of Gamalt! FWQ 90,12
(33) lat ilumm bitaadni wi btiddni darsi fi linniyya.ilinniyya?l ana
mi fa/ya l-binti wi an kida hiyya bitr bt zimilha da an ummu
hiyya-lli anna alha
It happened that the mother let me sit down and gave me a lesson in
affection.Affection?As if I would not be free for my daughter and
this is why she goes to the home of her colleague because his mother
always shows affection towards her SIG 131,3
In both (31) and (33), the context leaves no other possibility than that
it was a female speaker who made the incriminated utterance, even
though l remains masculine. l may follow the reported speech (34)
and may even occur in both positions at the same time (35). The latter is
particularly common with single words, as in (36):
12
For a brief discussion, see Woidich 1995, 265.
684 manfred woidich
(34) wi biylu ma-bnitaali l

And they say, can you believe it, [that] we do not work! WAR 20,5
(35) daxalit tialli l bitalli l )ayyib xallha talli
She went in to pray, what she calls pray, well, let her pray then! NAS
130,12
(36) dukham! sama biylu ala abbahathum ? dukham! l dukham l!
Those (people)! Did you hear what they said about her parents? Those!
Imagine, those! GIL 142,3
Grammaticalization is further corroborated by the fact that l acquires
increased mobility and may be moved into the sentence itself (37, 38)
and occur within subordinated sentences (39):
(37) dl fi Mari l biyiksabu bi lmt alfi gn
In Cairo they earncan you believe!a hundred thousand pounds!
SMB 93,5
(38) glu w xawat fi lBaw)i, illi lbalad ilmidna di. yiglu imu+uk1n. w
gl xadu ax mirsm dk, dahab. bassi da kalm 2 , wala xadu wala amilu,
hayugudu yibitu lxawat? 2
They said that foreigners came to ilBaw>i, which is the main town. They
say Americans. They came, allegedly, they took [with them] a golden
statuette in the shape of a cock. This is babble: they neither took anything
nor did anything. Would foreigners go around digging up things?
(Ba<ariyya: Manda)
(39) lit imbri innak l ma-bitru ilmadrasa
She said yesterday that youincredibly enough!do not go to school
TAY 38,11
In all cases from (31) through (39), the rejection of real speechthat is,
utterances that were really madehappens by reporting it, introduced
by l and pronounced with the appropriate intonation. In first instance,
it is this typical intonation of indignation and annoyance that makes it
clear to the hearer what the speaker means: disapproval and rejection.
Reporting it alone would not be sufficient for this purpose; intonation
must be an integral part of the construction. The semantic content of
this intonation materializes in grammaticalized l. This means that
disapproval and rejection now belong to the semantic content of l,
which in this way is recruited as a pragmatic marker for disapproval
and rejection.13 This makes it possible that the development goes further
13
Indeed, for at least some speakers/writers l seems to be no longer associated with
the verb ql in this meaning. In MRR and BAHN, for example, we find it written with
Alif Mdda as 3 whereas l as a verb preserves the original orthography and is written
with qf 39 throughout.
and l no longer needs to refer back to a real utterance or to an earlier

part of the discourse (see 31, 35, 36), but rather to a general saying as
in (40). l thus loses its function as a discourse marker and undergoes
a development into something like a sentence adverb, but its semantic
content remains unchanged. As both (40) and (41) show, the sentence
introduced by l was not uttered earlier:
(40) yiba ++gil lbis ya w akitta w alha ban)u w lfif ala rabtu kufiyya
f wi byirtii mi-lbardi wi llabwa m+tu walla bintu maya gambu mal)i
wala hiyya sala f-bardi walla f-talg. l irriggla yita))u wi nniswn
yitarru
So the man wears a collar, a jacket, and over it a coat, and wraps his
neck in a woollen scarf, and still he is shivering from cold, and his wife,
the bitch, or his daughter, walks naked beside him not bothered by cold
or snow. Unbelievably, the men cover themselves and the women strip
themselves DAR 163,3
(41) ilfustn )ili dayya alha, l batahli albisu!
The dress was too tight for her, incredible, she sent it to me to wear!
FWQ 92,5 (the person who sent the dress does not show up personally)
A further extension we notice when l introduces a sentence starting
with yani, an, or biylu. Just as in the examples above, l no longer
introduces a quote or reported speech, but expresses disapproval of a
fact or, more properly, the inference of the earlier statement. More often
than not, a sense of irony is involved. So, in (42) the first statement by A
invites the inference that A, as a matter of course, does not wet himself,
an inference that is rejected by B, ironically by using l together with
yani, which itself could convey a similar meaning.14 l serves here as
a reinforcement of this inference and adds the disapproving semantic
element. In the same vein, in (43)which is a quote from Bayram
at-Tniss isSayyid wi m+tu f-Barsthe pretentious behavior of the
umda described in the first part of the sentence and its inference, is
rejected and ridiculed by introducing this inference in the second part
of the sentence with l.
(42) A: ti+af ya ammi Wad iligli da biyuxxi ala nafsuB: l yani inta-lli
ma-bituxxi?
A: Do you know, Uncle Wa<d, this calf wets itself!B: Baloney,
would you pretend not to pee?! RUH 40,4
14
For a possible role of Sudan Arabic yani in the grammaticalization of Beja miyaad
3fait de dire by language contact, see Vanhove 2004, 149.
686 manfred woidich
(43) (after having listened to a short text in unintelligible newspaper talk)

yim ilumda yubrum anabu wi yhizzi dmu l yani fhim
The mayor twists his moustache and shakes his head, thus pretending
that he had understood SMB 84,10.15
l may be deleted in this case, since yani alone already conveys a
similar meaning and connects the two utterances in much the same
way, albeit without the sense of irony, cf. ya sitt ina wakln fl yani
mbarimn Madam, we have eaten fool, that means, we are full RAS
15,3 and i))ulumba mumkin titaal ala a+abiyyitn, yani mumkin
taman a+abiyyt yimlu banzn fi nafs ilwat The pump can serve two
cars, that means, eight cars can refuel at the same time NUS 35,3.
In (44) below, it is not so clear if rejection of the inference and irony,
which can clearly be understood from (42) and (43), are present and we
may even note here a further step in the semantic development towards
the sense of as if , just as in (45):
(44) nazln yisallimu laslia di l-ns fi mar, a))n ilmanga foha a)a l yani
humma tugg+ fakha
They went to deliver these weapons to people in Cairo, they had put
mangos on it as a cover, thus pretending that (~ as if) they were fruit
merchants BAHN 116,11
(45) lustz Ibdu baa txall masalan yidhin awaln buu min igginb kirm
abya/ l yani +awi
let, for example, Ustz Ibdu smear white cream on the side of his
mouth, as if it were foam ABM 206,-5
Similar semantic extensions may be stated for l an and l biyl, in
both cases l could be deleted:
(46) wi lingilz gum l an yimu mar
And the English came in order towhat they calledprotect Egypt
MAL 79,28
(47) ana ftakartu sak+n wi biyxa++af l biyl inn ilad+a )ilitlu wi huwwa
mi gamb sr ikkinsa
I considered him drunk and raving, when hebaloneysaid that the
Virgin appeared to him when he was walking besides the wall of the
church RUH 114,5
(48) da ddinya hayga wi lukma alba ddinya hnk l biylu lau ia+t
hink biyiftau
The world is upset and the government is turning everything round
there, they saybaloneythat they found traffic signs there that open
the road RUH 108,4
15
See SMB 85,4 for a similar passage.
This grammaticalization of l apparently started when it was used as

an exclamation with a strong tone of disapproval, introducing reported
speech16 uttered by somebody, but not approved by the speaker. The
following step must have been the switch from report to the inference
of the utterance, and finally from there to independent, generally
known facts, not specifically uttered previously. The connection with
a real utterance made by somebody was thus lost. This semantic
developmentnamely semantic bleaching and reinterpretationis
accompanied by a morphological reduction of l, that is, the inflection
is lost, and a syntactic flexibility appears: it may occur within sentences
and in subordinate clauses, and may be repeated producing a kind of
circumfix around a rejected phrase (35, 36).
The discourse marker l has l as an equivalent, which is nothing
other than the common rhetorical question formed with , a type of
discourse strategy that abounds in narrative contexts in Egyptian and
other dialects of Arabic.17 An utterance like (49) biyl alki di bitibbini,
l inti bitibb He says about you: she loves me. Nonsense, you and
loving him! FWQ 90,6 can be transformed into a narrative question,
as in (49) biyl alki di bitibbini, l , inti bitibb with the same
disapproval, disbelief and incredulity as the exclamation. Again, there
is lack of concord (5053), that is, l no longer refers to any referent.
Grammaticalization as a discourse marker is further evidenced by the
possibility of incorporation of l into the sentence, and, conversely,
by the fact that l may be dropped from sentences (5153) below
without disturbing the syntactical structure:
(50) xadit minni baiyyit ilirn illi fa/ln kamlit iah+, l atixzin bilibf
wi mallabt wi abn wi makaront
She took the rest of the few remaining piasters for the month, allegedly
she would store corned beef, tins, soap and pastas GIL 120,4
16
All my examples show indirect speech, not direct quotes, as far as this can be seen
from the reference of the pronouns. Direct quote as the original structure from which
the further development started cannot be excluded and is even probable. The lack of
examples could be seen as proof of early grammaticalization with loss of the starting
point structure. This view is corroborated by the fact that no inflected examples of l in
this sense could be found.
17
Any sentence may be split up into a kind of cleft sentence with a first part ending
with what? and the rest of the sentence as answer to this question: m wid minhum
, wxid blu mi lamaliyya, m zanna , ala aibna m aibna )aban , badi ma
allu ana +yi dort il0ayya, xa+ag ala ba++a One of them then what? He realized what
was going on, he hemmed in what? our friend, our friend of course what? After he had
told him I go to the loo, he headed out the door [Cairo: recorded text]. See Woidich
2006, 50.
688 manfred woidich
(51) knu biylu nni arb ilistinzf di l ma-kani laha lazma

They used to say that this war of attritionjust baloneywas not
necessary BAHN 78,3, = knu biylu nni arb ilistinzf di ma-kani
laha lazma
(52) ittit binti mafa dikit alayya w ayyilitni an)a l ala ma txu
ittawalt
A worthless squirt of a girl fooled me and let me carry a bagcan you
credit it?while she went to the loo SIG 102,5
(53) knu lbuada amln l tugg+ manga
Theyfar be they from youacted asunbelievable!mango traders
BAHN 82,6, = knu lbuada amln tugg+ manga
3.6 Comparative tiul

A rhetorical question formed by an introducing til You could say
followed by the interrogative suffix - as in tiul or provided with a
full negational ma- . . . - as in ma-tul, and a subordinated sentence as
reported speech, follows a proposition which describes an unexpected
and surprising fact as in (54, 55).18 Reported speech here communicates
a reason for this fact, but this reason is not the true one and in fact
contradicts reality. You could only compare it to the true reason. For
tiul ~ ma-tul, therefore, a pathway from Couldnt you say . . .?! =
You could say . . .! to As if it were > like can reasonably be assumed.19
On the discourse level, sentences with comparative tiul convey the
astonishment and excitement of the speaker.
(54) sabni wala saal anni, tiul illa kunti bawakkilu )b mi tasiyya w lamit
+s?
He left me and did not inform about me any longer, could you say that
[= as if] I had fed him with bricks, not with broth and meat of the head?
DAR 90,15.
(55) badi wayya asma dawa w kalm tiul il/a malyna ns
After a while I heard noise and talking, as if the room was full of people
GIZ 56,8
(56) labsa am ama+ min r ikmm wi riglha iryna ma-tuli ladaitni a+aba
She was wearing a red sleeveless shirt and her legs were naked, [I was
shocked] as if a scorpion had stung me MRR 80,8
(57) nazln al iddni iddni ma-tul innu huwwa nir ilwaf
They showered him with Give me, give me!, as if he were the supervisor
of the endowment NAZ 62,9
18
See Woidich 1989, 124 and 2006, 281.
19
For similar pathways of grammaticalization of say to like, as if in other langua-
ges, see Heine et al. 2003, 268.
tiul may be followed by inn introducing indirect speech (57). (56)

differs in some way from (54, 55) since the comparison expressed by
means of reported speech does not refer directly to the fact, but refers to
its consequence, which is not expressed itself [hence put into brackets
in (56)]: the unexpected seductive appearance of the lady causes a shock
to the speaker, which may be compared to the sting of a scorpion. We
notice here a shift of reference from a fact to the inferences thereof
similar to what occurred with l (cf. 3.5), which suggests that tiul
got more independence in its use and that it is on its way to become a
function word with the notion as if.
The question is how far tiul has developed in this way. It still may vary
and be inflected according to the person spoken to, that is, we can find it
as fem. tiul or pl. tiul followed by a complete sentence:
(58) mah dilwati aklitha wi labri bar/ak, ana mi fhim anhi banna aw fil
illi xtara ikkaki bi akli da? tiul illa byimilu xal)it mna
But nowadays, it takes you to the grave when you eat it. I do not
understand: which mason or bricklayer invented this sort of cake? You
could say they mix mortar! SMM 55,5
(59) wi lmg )li nzil tiul milya arr zara
And the waves went up and down, you could say like a ladies dark-blue
silk cloak DAR 141,6
(60) min sit ma simit ilxaba+ im wi hiyya nazla la)mi w buka tul ibnaha
Since she heard this awful news, she has been constantly slapping her
face and crying, as if it were her (own) son RUH 152,7
On the other hand, there are examples in the singular when we would
expect the plural, as in (61), a sentence that addresses a general public
(cf. ma-taxzun pl. Do not blame me! three lines further in the same
opening speech):
(61) w ah layym ammla tigri, tiul alla zalna minna!
And the days fly past, as ifalasthey were angry with us MAL 1,3
Whereas the facts with regard to inflection remain inconclusive,
other features suggest that tiul is on its way to be grammaticalized.
In (54, 58) above, both parts apparently still have separate intonational
contours as indicated by the authors by means of a question mark. But,
on the other hand, we see in (62, 63) below that tiul can be followed by
one argument only, not a complete sentence, which suggests that tiul
has been incorporated into the sentence and that it is used just as the
preposition zayy like. Indeed, the whole sequence has to be read here
with a single intonational contour.
690 manfred woidich
(62) amml yilsa fi llisn tiuli a+aba

It stings the tongue, as if it were a scorpion TAY 169,8
(63) w inta mkaa+ wi manfx a lxir tiuli ras wiz+a
And what are you, grave looking and utterly pompous, as if you were a
prime minister RAS 82,920
Moreover, tiul can leave its original head position and be moved within
the sentence, as in:
(64) lat makn azla zayyi ma huwwa, faritu tuli lissa mafra dilwati
I found azlas place unchanged, his bed was as if it had been made just
now LAB 49,2
(65) wala assi b-ayyuha ga r inn ila+abiyya wafa w humma tuli nizil
alhum sahm a&&h guwwa lutumbl
Nor did he feel anything but that the car had stopped and that they were
as if Gods arrow had fallen on them in the car (i.e. they had suddenly
fallen still) LAB 91,7
The original position in (64) would be tiuli faritu lissa mafra . . . and
in (65) tiul nizil alhum . . . . This strongly suggests that tiul in these
cases is grammaticalized and serves as a function word to express an
unreal comparison.
til you could say without a interrogative suffix - is rarely found in

Egypt, for example in (66) )amaha lazz tili hiyya malban It tastes
nice, youd say it is Turkish delight ARA 58,-1 and in texts from
Dakhla-Oasis (alMiyya): atgardit til di agr It rolled down as if
it were a stone. These last two examples coincide with the widespread
use of til in other Arabic dialects in the form of tegel and apparently
grammaticalized as a preposition like or something similar; see for the
Da%na tigel, tegel tu dirais Landberg (1942) 2542; for the Rwala tesma
eb nejrahom te4el dammm It is as if one could hear the angry voice
of their mortars, like to the sound of a great drum Musil (1982, 84,3)
and passim; for Souchne bitfu tik&innu cabal er sieht aus wie ein Berg
Behnstedt (1994, 353), originally When you see him, youd say it is a
mountain; for Syrian nomads see Cantineau (1937, 196); for Tunisia
see Marais (1959, 3310f.); for Morocco Colin (1993, 1624).
20
An example from Bri dialect in Upper Egypt: ilfurusa +a00a tgli ikle The
mare gallops like a young stallion.
3.7 Turn-taking device:21 baullak

Another phrase commonly used as a discourse marker by a speaker
in order take the lead in the conversation and to start a new topic, is
baullak Ill tell you something! Listen! see (66, 67) or simply bal . . .
I say . . . as in (68) without addressing the addressee directly. It attracts
the attention of the hearer by inviting the inference that he is telling him
something important. It is often used to prevent a longer discussion and
to give directions and instructions.
(66) : aywa ana ib ilfa/li alkum, adi nnu lolya ana ma-kani f addi
minkum gih hina, wi fi lxir absinni ba++a mi ayzn tidaxxalni ana
aywa baullak ! ana mi f/i li lhartala btatak di! lissa yiz abu) i/ati
w au)) kamirti . . .
: Yes, it is due to me, I mean, without me, none of you would have come
here, and, in the end, you shut me out and do not let me come in here any
longer!Yes, listen! I am not free for this nonsense of yours now! I want
to fix my lighting and set down my cameras . . . MUW 40,2.
(67) N: la ya **a, la! Fatiyya lzim tikammil talimha. kifya lli ga+li ana w
5a uxti min adit ilbt wi lirmn min ittalm.B: baullik ya Nagba,
ya binti. kalm kitr mi yiz asma w ana xal qa++a+t . . .
No, Daddy, no! Fat<iyya must finish her education. Enough what
happened to me and my sister @a from sitting at home and being
excluded from education!B: Ill tell you something, Nagba, my
daughter! I do not want to hear much fuss and I have made my final
decision . . .! HUD 41,8
(68) S: la, wala ga hti lahwa.H: ma-h f-di. itfa/dal ilahwa ya B!M:
mi yiz ya sitt, a&&h ilani an ahwitik, kida ya sitti Szi, inti ayzni
ar fhaS: bal ma-tsibha hina, u00l fn ilu)a Sayyid?
S: No, nothing, bring the coffee!H: I have it in my hand. Here is the
coffee, please!M: I do not want, Mam, I can do without your coffee!
Is it that, Mrs Suzy, you want me to die?S: Id say, leave it here, where
is Master Sayyid? ABM 193,27
At the same time the use of baullak creates kind of intimacy between
the speaker and the addressee, preventing the directions from coming
over too rude.
21
As to the exclamation of strong assertion bitl fha no question about it! you bet!
you said! adduced in Hinds et al. 1986, 722b, I could not find any example in the litera-
ture except the ones given in this entry: idirti truddi ala lmudr?bitl fha w atamtu
kamn Did you dare to answer the boss back?You bet I did! and I cursed him too!
692 manfred woidich
3.8 Reference to hearsay information and general knowledge

biylu they say introduces what is known from hearsay or is general
knowledge and from a non-specific source functioning as an evidential
marker, see Heine et al. (2003, 265). As usual, the 3rd pl is used in this
case, see Woidich (2006, 357):
(69) simiti agt kitr biylu kulli ga hatiba ala lbi)a
I have heard a lot. They say everything will be on a personal identity
card RUH 112,1
(70) inta lzim )allat illi biylu nnak miggawwizha
You must have divorced the one which, they say, you had married
FWQ 120,10
(71) biylu kamn kn skin fha wada w bintaha
They say, too, that a woman and her daughter were living in it RUH
18,10
Often a proverb is quoted in this way:
(72) wi l baa listigl da, da atta biylu lagala min ii)n
And why this haste, they even say, haste is from the Devil WAZ 377,15
4. l, yil introducing pseudo-complements22
In his article on serial verbs in EALL, Kees Versteegh mentions a use

of l, yil to say in Egyptian Arabic, which he considers as a com-
plementizer derived from a serial verb construction [SVC], which in his
view provides a syntactic slot for an object. Quoting an example taken from
Woidich (2002, 183) he compares this use of l, yil to similar develop-
ments in Creole languages such as Krio Creole English and Ki-Nubi.
(73) huwwa raddi alk allak ?
What did he answer you? LAB 118,8
Earlier, Catherine Miller brought the development of gl as a
complementizer in Juba Arabic to our attention, Miller (1998, 2000, 2001)
where gl is grammaticalized and used as a complementizer after verbs
of saying and cognition.23 Let me first cite some additional examples:
22
See Woidich 2002, 181ff, Woidich 2006, 401. Phrases of the same type, but used
to express the direction or a goal, are formed with +, as in kuntu btuxrugu tru fn?
where did you go out to? FWQ 62,6, see Woidich, loc.cit.
23
Hopper et al. (2003, 194196) discuss similar developments of say in Akkadian,
see as well Hopper et al. (2003, 1315) for West African languages. For a similar use
(74) ilmaxzin )aban atruddi tl ianfi r mawgd

The stores of course will answer and say, this kind is not available
RAQ 38,8
(75) ana kutti baanni bal ?
What, was I busy singing? [film]
(76) inta skit ya (asann, akkallim ana!atikkallimi tli ? waffari kkalm!
You are silent, asann? Then I shall speak!What are you going to
say? Keep your words to yourself ! ABM 144,6
(77) an kida dayman ana zabayni aulluhum iwa ya-bni tu))i ga f-gbak
This is why I always advise my customers and say, take care, my boy,
not to put anything in your pocket YUN 27,3
(78) rafa/u allu ina malna
They refused and said, not our business! NUS 7,6
(79) xayya+ni ammti alli taxdi bni ammik
My aunts made me choose, they said to me: you should take your
cousin RUH 59,3;
The differences between the use of l, yil in our case and gle in Juba
or se in Krio Creole English are indeed significant: the first mentioned
introduces quotations and noun phrasesmostly the question pronoun
what?and is fully inflected, whereas the other two serve as a
complementizer for subordinated sentences of various kinds and are
no longer inflected. Egyptian Arabic l, yil remains connected with
locutive and speech-related verbs, and its use is not extended to other
types of verbs as happens in many West African languages (Lord 1993,
176 ff) and in Juba-Arabic (Miller 2001). There is no trace of l in
Egyptian Arabic functioning as complementizer that, a fact earlier
stated for Arabic dialects in general in Versteegh (1984, 101).
Can we analyse l, yil constructions as an SVC, then? For a verbal

sequence to be recognized as an SVC it has to fulfil some conditions
formulated in Newmeyer (2004, 2f) and Kroeger (2004, 226256). Some
of these are met by our l, yil constructions:24
they are two verbs within the same clause, neither of which is an
auxiliary;
they belong to a single intonation contour25 and refer to a single event;
of the Turkish gerund diye saying see Kissling (1960, 190). For Bedja, see Vanhove
2004.
24
See in particular Kroeger (2004, 229).
25
For a case with no single intonation contour but two contours, see gaadit turgu
wu tanni, glit abbi a+ balad ilLayy, ina&&a m ni yiriddi alayy She started to
694 manfred woidich
they share at least one semantic argument, which is the agent in their
case;
they contain only one grammatical subject. 26
At first glance, this makes the l, yil constructions good candidates for
SVC, were it not for some other diagnostic features that exclude them
from the SVC. SVC do not contain any overt markers of subordination
or coordination. This is not true for l, yil constructions where wi
and may separate the two verbs, suggesting that what we have here is a
coordinate construction:
(80) raddi azla w allu mafhm ya ammi
azla answered and said to him, understood, Uncle! LAB 138,6 (cf.
74)
(81) wi anna w l mawawl
And he was singing Mawwls MAL 45,8 (cf. 75)
(82) af/al ana abar)am w al atta ladwiya kamn biyaddarha
I shall keep muttering: even the medicine they export as well GIL 58,8
(cf. 84 below)
(83) ultilu kulli da, raddi w alli inta-zzayy awlak? mi kuwayyis?
I told him all this, he answered and said to me: how are you? Not
good? ULA 66,2 (cf. 75)
Tense, aspect, modality, negation, etc. are normally expressed only once in
true SVC, and only occasionally are both verbs marked for these. l,
yil constructions, however, usually mark both verbs in the same way,
that is, both verbs appear in the same tense or aspect (see the examples
above), with the exception of bi- and a-prefixes: bi- may be and a- is
always omitted on yil; see (74, 76) above. For bi-, see (84):27
(84) ilbitti bitbar)am til ?
What is the girl mumbling? SIG 126,11
dance and to sing: my darling went to the country ilLayy, he will never return to me, I
hope (Ba<ariyya: MandBa).
26
This excludes, by the way, causative xalla clauses from the serial verb construc-
tions: xalltu ma/a I made him sign contains two agents/subjects.
27
Exceptional is katabu ylu ? What do they write and say? MRR 101,4 which
does not follow the above rules. It contains a sequence of a perfect and an imperfect
instead of perfect in both verbs, as in katabu lu ? What did they write and say? In
both cases, the question word what? asks for the opinion of the writers. Since the
act of writing happened in the past and the opinion expressed by this writing stays pre-
sent till the moment of asking, an imperfect yilu seems justified for this moment. The
semantic notion here overrules the syntactic one; see Woidich (2003, 131).
This is usual when two imperfects of this type are coordinated, see
Woidich (2006, 282). Even different verbal modals with more or less the
same semantic content are possible, as (85) shows, when the verbs are
coordinated by wi and:
(85) umm& h illi inta id tixa++af wi amml tilu da?
What is it then what you are raving about all the time? ABM 39,1
SVC verbs should not contain two overt NPs that refer to the same
argument; see Kroeger (2004, 230). In l, yil constructions this is
possible and both verbs may keep their original semantic and syntactical
structure, cf. (73) and (77) above and:
(86) inta bitwawi ilbinti bitullaha ?
What are you whispering to the girl? MRR 198,11
(87) amml tikallim nafsak til ?
What are you talking to yourself ? LIB 64,12
Finally, as Newmeyer (2004, 3) reports, in a coordinate structure moving
an NP of one of the verbs from its original position should be impossible
(coordinate structure constraint), but it is possible in an SVC. In the
l, yil construction, noun extraction by forming a cleft sentence is
not possible: * lli raddi alk alhlak? or * lli akkallim alu? Strangely
enough, (85) with its coordinated verbs offers such a cleft sentence
containing a moved NP, though it should be not possible in this case
according to the coordinate structure constraint adduced by Newmeyer
loc. cit.
In my view, these are enough reasons not to consider l, yil construc-
tions as SVC. Rather, I think that they developed from paratactic
entailments to intransitive verbs with a semantic component of saying
such as28 6anna to sing, yir to blame, ayya) to cry, bar)am to grumble,
daa to curse, dalla to give a nick-name, ikkallim to talk, kidib to lie,
na)a to pronounce, naa to give advice, radd to answer, rafa/ to refuse,
arrax to shout, wawi to whisper, xa++af to talk nonsense, which do
not have a syntactic slot to express the content of the performative act
and cannot quote directly. By addingbe it asyndetically or by means of
wia sentence with the locutionary verb l, yil that disposes of that
slot, the problem is solved. This coordinative structure describes one
28
The list of verbs with pseudo-complement as registered in my database. 13 of the
28 registered cases include the question l .
696 manfred woidich
fact and shows cohesion as to intonation, but not the syntactic cohesion
normally displayed in SVC. Following Seuren (1991, 196), I prefer to
interpret these l, yil constructions as pseudo-complements, especially
since l, yil is not the only type of these complements. Many verbs with
the semantic content of moving into a direction do not have a syntactic
slot for the direction of the movement. Here a pseudo-complement
with +, yir to go to fills the gap by introducing this direction, as in
mii + fn? Where did he go to?; see Woidich (2003, 181ff ) for more
details. Similar behavior is exhibited by xad, yxud to take and l, yil
to take away, which may be combined with wadda, yiwaddi to bring to
to form utterances such as anlu nwaddh fn AWL 44,6 Where shall
we bring him to?, imbri bi lll alha waddha lbt Yesterday evening
he took it home WAZ 369,-3f.
5. ql in delocutive derivations29
l, yil is combined with set phrases such as vocatives or idiomatic

expressions, thus enlarging the lexicon of the language. This is common
in Cairo Arabic. Here, it will suffice to give some examples:
(88) a))i dlu bn wi+ku w l ya fakk
It hung its tail between its legs and cleared off FAG 119,5
(89) aadt a/rab f l-yit lamma l ya bass
I kept beating him till he could take no more Hinds et al.
(1986, 74b)
(90) ittit dn ala txallha tl ai b raabti
An awful beating, which makes her give in RUH 63,2
(91) kunti bal ya ar/ inai w iblani
I said, O Earth split open and swallow me! = I wanted to vanish into
thin air LAB 218,4
Of a different type is l with sound-related interjections or ideophones
(or expressives, as they are sometimes called) to form descriptive
compounds. Only two examples have come to my attention, but there
are bound to be more:
(92) l wi He vomited Hinds et al. (1986, 947b)
(93) l g2y2 He gave up, resigned30 Azz 129
29
Plank (2002, 465 and 478).
30
For gy ~ gy help! see Hinds et al. 1986, 146a; for the Arabic etymology, see
Behnstedt 1997, 36.
We quite often find this type of gl, yigl with sound-related interjections
as a direct object in Upper Egyptian Arabic, for example in Bri.31 They
mostly describe a sudden event connected with a sound, and occur
commonly in narrative style, not necessarily as intransitive expressions
as described in Plank (2002, 468).
(94 + gyil ka+a+w bha He rushed at her
(95) itgl dibb fi lar/ She toppled over
(96) ma++a wida gl dradib All of a sudden, it went knack
(97) gl daradyy He crashed down
(98) gult ijlibb I jumped up and ran away
(99) gl, igl )) to break wind
(100) glat )irri minnh It buzzed away
(101) glat inn It bubbled up (boiling water)
(102) t ittit ul)a kida w gl hub, a+ glib ilgull wu lba wu adda min
r iml
A flat piece came, he made hub and threw down the jars and the bowl
and went off without a burden (Ba<ariyya: Manda)
(103) wu badn ilga)ir gl )), gallih ma tistanna lamma allabiyya tinif
And then the train made >>, he said to it, wait untill the gallabiyya
has become dry (Ba<ariyya: Manda)
6. References
6.1 General References

Azz, Araf. 2005. al-Kinyt al-mmiyya al-miriyya. al-Qhira: al-Hadara
Publishing.
Behnstedt, Peter. 1994. Der arabische Dialekt von Soukhne (Syrien). Teil 2 und 3.
Wiesbaden: Harrassowitz Verlag.
1997. Koptisch oder Arabisch? Wiener Zeitschrift fr die Kunde des Morgenlandes
87, 3139.
Cantineau, Jean. 1937. Etudes sur quelques parlers de nomades arabes dOrient.
Annales de lInstitut dtudes Orientales III,119237.
Chisarik, Erika and Wim van der Wurff. 2003. From Say to Because: Grammaticalisation
and Reanalysis. Paper read at the Conference on Comparative Diachronic Syntax,
University of Leiden Centre for Linguistics (ULCL), 2930 August 2003. http://www.
let.leidenuniv.nl/ulcl/events/compdiachr/programme.htm
Cohen, David, Marie-Claude Simeone-Senelle, Martine Vanhove. 2002. The
grammaticalization of say and do: An areal phenomenon in East Africa. In
Gldemann, Tom and Manfred von Roncador, eds. Reported discourse: A meeting
ground for different linguistic domains. Amsterdam: J. Benjamins, 227251.
31
For the very elaborate use of these expressions in Ethio-Semitic, which goes much
further than what we know from these Upper Egyptian Arabic dialects, see Cohen et al.
(2002, 227 and 238 ff). Cf. for the Da7na Landberg (1909, 1268) gl hubs se mit
courir.
698 manfred woidich
Colin, Georges S. 1993. Le dictionnaire COLIN darabe Dialectal Marocain. Sous la

direction de Zakia Iraqui Sinaceur. Volume 6. Rabat: Editions al Manahil.
Fischer, Wolfdietrich und Otto Jastrow. 1980. Handbuch der arabischen Dialekte. Porta
Linguarum Orientalium, Neue Serie 16. Wiesbaden: O. Harrassowitz.
Georgakopoulu, Alexandra and Dionysis Goutsos. 2004. Discourse Analysis. An
Introduction. 2nd print. Edinburgh: Edinburgh University Press.
Golato, Andrea. 2002. Self-quotation in German: Reporting on past decisions. In
Gldemann, Tom and Manfred von Roncador, eds.. Reported discourse: A meeting
Gldemann, Tom and Manfred von Roncador, eds. 2002. Reported discourse: A meeting
ground for different linguistic domains. Amsterdam: J. Benjamins.
Heine, Bernd and Tania Kuteva. 2002. World Lexicon of Grammaticalization. Cambridge:
Hinds, Martin and El Said Badawi. 1986. A Dictionary of Egyptian Arabic. Beirut:
Librairie du Liban.
Hopper, Paul J. and Elizabeth Closs Traugott. 2003. Grammaticalization. Second edition.
Cambridge: Cambridge University Press.
Kissling, Hans-Joachim. 1960. Osmanisch-Trkische Grammatik. Wiesbaden: Harrassowitz.
Klamer, Marian A.F. 2002. Report constructions in Kembera (Austronesian). In
Gldemann, Tom and Manfred von Roncador, eds.. Reported discourse: A meeting
Kroeger, Paul R. 2004. Analyzing Syntax. A Lexical-Functional Approach. Cambridge:
Landberg, Carlo de. 1909. tudes sur les dialectes de larabie mridionale. Deuxime
volume. Da%nah. Deuxime partie. Leiden: E. J. Brill.
1942. Glossaire Datnois. Troisime volume ( w) publi par K.V. Zettersten.
Leiden: E.J. Brill.
Lord, Carol. 1993. Historical Change in Serial Verb Constructions. Amsterdam: J.
Benjamins.
Marais, William and Abderrahmn Guga.1959. Textes Arabes de Takrona. II Glossaire.
Sixime Tome. Paris: P. Geuthner.
Miller, Catherine. 2000. The Grammaticalization of the Verb to say in Juba Arabic.
Proceedings of the Third International Conference AIDA, Malta 29 March2 April 1998,
ed. by Manwel Mifsud, 213218. Malta: Association Internationale de Dialectologie
Arabe.
2001. Grammaticalisation du verbe gale dire et subordination en Juba Arabic.
Leons dAfrique: Filiations, ruptures et reconstitution de langues: Un hommage
Gabriel Manessy, ed. Robert Nicolai, 455482. Louvain: Peeters.
Musil, Alois. 1928. The Manners and Customs of the Rwala Bedouins. New York:
American Geographical Society.
Newmeyer, Frederick J. 2004. Some thoughts on the serial verb construction. Paper
read at: Atelier du 9 dcembre 2004. CRLAO, EHESS, Paris. La notion de construction
verbale en srie est-elle opratoire? Fdration TUL (internet: http://www.typologie.
cnrs.fr/fr/ gabarits/TUL%20Newmeyer.pdf).
Plank, Frans. 2005. Delocutive verbs, crosslinguistically. Linguistic Typology 9,
459491.
Schwenter, Scott and Richard Waltereit. 2005. Presupposition accommodation
and language change: From additivity to speech-act marking. Paper read at From
Ideational to Interpersonal: Perspectives from Grammaticalization. Leuven, 1012
February 2005. Abstract: http://wwwling.arts.kuleuven.ac.be/fitigra/schwenter.pdf
Seuren, Pieter. 1991. The definition of serial verbs. Development and Structure of Creole
Languages: Essays in honor of Derek Bickerton, eds. Francis Byrne and Thom Huebner.
Amsterdam: J. Benjamins, 193205.
Vanhove, Martine. 2004. Dire et finalit en bedja: Un cas de grammaticalisation.

Journal of African Languages and Linguistics 25,2, 133153.
Versteegh, Kees. 1984. Pidginization and Creolization: The Case of Arabic. (Current
Issues in Linguistic Theory, Volume 33). Amsterdam: J. Benjamins.
2007. Serial verbs. In Kees Versteegh et al., eds., Encyclopedia of Arabic Language
and Linguistics, vol. 3. Leiden: Brill.
Woidich, Manfred. 1980. XIV. Text aus il-Birt. In W. Fischer and O. Jastrow, Handbuch
der arabischen Dialekte, 235242.
1989. illi da, illi weil und zayy illi als ob: Zur Reinterpretation von
Relativsatzgefgen im Kairenischen. Mediterranean Language Review 4/5, 109128.
1995. Some cases of Grammaticalization in the Egyptian Arabic Dialects.
Proceedings of the 2nd International Conference of Lassociation internationale pour
la dialectologie Arabe, Trinity Hall, University of Cambridge, 1014 September 1995,
259268. Cambridge: Association Internationale de Dialectologie Arabe.
2002. Verbalphrasen mit asyndetischem Perfekt im gyptisch-Arabischen.
Estudios de dialectologa norteafricana y andalus 6, 121192.
2006. Das Kairenisch-Arabische. Eine Grammatik. Porta Linguarum Orientalium,
Neue Serie 22. Wiesbaden: Harrassowitz Verlag.
6.2 Literary Sources

ABM = Amn Bakr, Xamas masraiyyt kumidya. al-Qhira 1986.
ARA = Lu>f al-Xl 1988. il-arnib. Kairo.
AWL Al Slim. 1986. awladna f Landan. al-Qhira.
BAHN = Smi< Farag. 1999. Bnhftrsa. Riwya bi l-mmiyya l-miriyya. al-Qhira.
BTR = Al Slim. 1991. ilBitrl )ili fi betna. al-Qhira.
DAR = usayn afq al-MiDr. 1929. il(aggi Darw w Umm Ismal. al-Qhira.
FAG = A<mad Fud Nigm. 1993. Muzakkirt ilFagmi. al-Qhira.
FWQ = Numn r. 1958. inNs illi f. al-Qhira: Dr an-Nadm.
GIL = Numn r. 1972. igGl iggidd. al-Qhira.
HAM = Bahg Isml. 1994. iln ilamra. al-Qhira.
HUD = Fat<iyya lAssl. 2002. (u/n al-umr. as-sra -tiyya. al-Qhira: al-haya
al-miDriyya al-mma li l-kitb.
LAB = Ysuf al-Qad. 1994. Laban ilafr. Riwyt al-Hill 545. al-Qhira.
LAY = Mu<ammad Inn. 1993. Laylat a-ahab. al-Qhira.
LIB = Mu<ammad Kaml Mu<ammad. 1994. Libit itttalib. al-Qhira.
MAL = Motie Ibrahim Hassan. 1971. in-Ns wil-malik. Copenhagen.
MAX = Ma<md Taymr. 1979. ilMaxba raqam tala))ar. An annotated phonemic
transcription by Stig. T. Rasmussen. Kopenhagen: Akademisk Forlag.
MRR = Rad Rud. 1975. Masra Rad Rud. al-Qhira.
MUW = Wald Ysif. 2000. ilMuw)in mahri. 2000. al-masra< al-arab 142. al-Qhira.
NAZ = Fat<i RaJwn. 1973. Nir Waqf. al-Qhira.
NUS = al-mmiyya al-miriyya f nu wa muwart. Nederlands-Vlaams Instituut
te Caro w.d.
RAQ = Ysuf f. 1997. Rqiat qi) mm. al-Qhira.
RAS = Nihd Gd. 1989. a rRaf. al-Qhira.
RUH = Kaf Abd al-Munim. 2005. Min (alwit irr. Riwya bi l-mmiyya. al-Qhira:
Madnat 6 Uktbar: Sanbil li n-nar wa t-tawz (a>->aba a%-%niya).
SIB = Sad ad-Dn Wahba. 1966. isSibinsa. Masra<iyyt Sad ad-Dn Wahba Nr. 149.
al-Qhira.
SIG = Fat<iyya al-Assl. 1993. Sign innis. al-Qhira.
SIK = Sad ad-Dn Wahba. 1967. Sikkit issalma. Masra<iyyt Sad ad-Dn Wahba
Nr. 184. al-Qhira.
700 manfred woidich
SMB = Bayram at-Tnis. 1925. isSayyid wi m+tu f Bars. al-Qhira.

SMM = Bayram at-Tnis. 1925. isSayyid wi m+tu f Mar. al-Qhira.
TAY = Mu<ammad Taymr. 1922. al-Masra al-mir. Muallaft Muammad Taymr,
al-juz a7-7li7. al-Qhira 1341 (1922).
ULA = Mu<ammad NDir. 2000. la awwil. al-Qhira: al-Majlis al-al li %-%aqfa.
WAZ = Sad ad-Dn Wahba. 1980. ilwazr l ittallga. al-Qhira.
XAMA = A<mad ams ad-Dn al-ajjj. 1988. ilXamasn. al-Qhira: al-masra<
al-arab.
YUN = Mu<ammad Inni. 1993. (alwit Ynis wi masraiyyt uxr. al-Qhira.
INDEX
, 571, 571 n. 24 Ab R Academic Hospital, 533

ending, 544 Ab Ubayda, 64
Abbeche Arabic, 648 Ab Xir, 90
Abbs, 315, 363 Ab Ysuf, 43
Abd al-l, 665 n. 8, 671 Abu Zinima, 566 n. 6
Abd al-Fatth Salm, 187 Abu-Haidar, 668669, 671
Abd al-H usayn al-Fatl, 22, 187 Abu-Lughod, Lila, 544, 563
Abd al-Qhir al-Jurjn, 232 Abuw lHl, 569 n. 16
Abd as-Salm Muhammad Hrn, 23, Abuw Znmah, 566, 566 n. 6, 578
43, 64 academy(ies) of, 655, 661
Abda**a lGum, 678 Baghdad, 661
Abdallh, 56, 61 Cairo, 661
Abdalmalik b. H abb, 195, 195 n. 10 Damascus, 661
Abdeen, 662 Jordan, 661
Abdel-Malek, Kamal, 563 Acc-ing, 478, 483, n. 10
Abderrahmn Guga, 698 construction, 476
Abdh, Ms, 548, 563 accompli, 315 n. 4
Abdul-Malek, 544 du prsent, 331
Abdul-Raof, 393395, 395 n. 85, 400 accusatif, 118, 130
abitud, 215 accusative, 78, 10, 1819, 19 n. 15, 95,
abitudines, 214215, 226 135, 135 n. 2, 138, 138 n. 19, 139140,
ablatiuo, 214, 216, 227 142, 144147, 216217, 227228, 230
ablative, 216217 ending, 138, 144
ablativus, 221 +genitive, 135
Ablaut, 256 accusativus, 221
abn al-Ajam, 115 acento, 215 n. 11
abn al-Arab, 115 achvement, 334, 341, 341 n. 24, 347
Abney, 475477, 478 n. 3, 483, 485, 498 du processus, 333334, 339
abniya, 316 Ackema and Neeleman, 475476, 478
Abraham, 324 n. 14, 363 479, 480 n. 5, 481, 481 n. 6, 482, 482
Ab Abdallh M. b. A. al-Kinn, 208 n. 8, 483, 483 n. 10, 484485, 495, 497
Ab Al al-Fris, 10, 165 n. 30, 498
Ab l-Aswad ad-Dual, 116, 120, n. 4, acquis, 326, 334335, 339341, 350
133 caractristique, 326327, 329331,
Abu Athera, Said Salman, 543 n. 1, 563 334, 337341, 344, 348, 348 n. 29, 351
Ab Bakr M. b. Abdallh b. Muhammad non, 327, 330331
b. Uta al-Isbahn, 200 n. 30 constitutif, 338
Ab Dwd, 93, 110 dernier, 326, 331, 333, 339340, 349
Ab H anfa, 35 final, 326
Ab l-H asan [al-Axfa], 167 instable, 338
Ab H usayn, 43 passager, 327 n. 18
Ab l-H usayn al-Basr, 38, 4041 permanent, 320
Ab Jafar Quray b. Uqba b. Bar, 198 rsultatif, 325, 325 n. 15, 326328,
Ab l-Qsim Abd al-Rahmn b. Ishq, 330332, 339, 341, 343, 349350,
113 352
Ab Qr, 130 non-caractristique, 329
Ab t-Tayyib al-Lughaw, 206 n. 45 temporaire, 335, 338
702 index
transitoire, 327 n. 18 degree, 384

acquisition expression, 682
dtat, 326327, 330, 334, 338340, manner, 384
342343, 346, 358, 360 phrasal, 387
acteur, 335337 phrase, 150, 152, 153 n. 5, 159,
lactif, 115 162163, 163 n. 20, 164, 164 n. 23,
activities 167168, 171172, 177180
durative, 299 prepositional, 175176, 179180,
habitual, 299 183186
actors, 396 semantic component, 170
acusatiuo, 214, 216 sentence, 376378, 387
acusativo, 227 structures, 54
d, 195, 202203 time, 384
Adab literature, 199 af l
Adam, 39, 190, 192195, 197198, 200, 206 al-madh wa-d-damm, 378 n. 4
adaptation(s) an-nqisa, 378 n. 41
morphological, 308 al-muqraba, 378 n. 41
phonological, 306, 308 al-qulb, 142, 378 n. 41
add, 86 ar-raj, 378 n. 41
Aden, 8283, 83 n. 18, 84 a-ur, 378 n. 41
adjectif(s), 325329, 331332, 335, 342, at-tafdl, 378 n. 41
346347, 349350 affirmatif, 519
analogue(s), 327, 341 affirmation, 398
assimil, 327, 331333, 341342, affix, 370 n. 11, 477480, 481 n. 6,
345352 482483, 483 n. 10, 484, 484 n. 12,
dtat 485, 492, 497 n. 30
acquis, 334 internal, 483
caractristique, 328, 333335, masdar, 485, 492, 494497, 497 n. 28
339342, 344345, 351354, 358, nominal, 480
360 nominalizing, 479, 482483, 483
non-caractristique, 328, 331, 333, 335 n. 11, 493
adjectival agreement, 29 phonological, 481, 496
adjective(s), 17 n. 13, 49, 153 n. 5, 154, prosodic, 495
156, 156 n. 11, 157, 158 n. 13, 160, stem, 490
167, 176, 189, 248249, 250251, 308, syntactic, 481, 496
370 n. 14, 373, 376 n. 34, 382 n. 53, Affix Grammars over Finite Lattices,
477, 484485 369, 369 n. 10, 370, 372 n. 18, 373
arab, 190 n. 23, 377 n. 36, 383
feminine, 250 affixation, 367
indefinite, 251 affixes, 230, 479481
nominalized, 382 non-, 372
adjunct stem, 488
adverbial, 391 terminal, 372
attributive, 391 Africa, 621, 639
adjunction, 293 East, 639, 697
Adnn, 201, 203 African, 607, 609, 611, 621
adverb(s), 214, 229, 252, 374375, 384, East, 620
385, 397, 477, 577 West, 692 n. 23, 693
interrogative, 398 n. 90 affricate, 597
sentence, 685 affrication, 611
temporal-, 63 Afroasiatic, 595
adverbial, 150, 153 n. 5, 159160, Agapito Valle Flemmarum, 212, 226,
162163, 167, 169 n. 26, 170, 172, 231, 232, 233234, 234 n. 48, 237, 239
180, 183, 371 n. 17, 375, 384, 387 Agylah
constituent, 166 dialect(s), 567, 572 n. 25
index 703
agent, 321, 321 n. 11, 322323, 326, 326 Alep, 113, 123
n. 17, 343, 348349, 351 n. 32, aleph(s), 595, 596 n. 6, 597, 597 n. 8,
354357, 376 n. 31, 380, 392394, 599 nn. 13, 15
394 n. 83, 395, 396 n. 85 Aleppo, 232 n. 44
explicit, 391 Aleut, 294
humain, 348, 351352, 356 Alexandrie, 130
implied, 391 Al-Falay, Ibrahim S., 655, 670671
agentives, 394 al-Alf, 530
agentivit, 321, 321 n. 11, 323, 328 n. 19, Alfiyya commentaries, 3
336, 344, 350351, 359 alayta, 141 n. 32, 160
entire, 323, 349, 359 Alger, 113
neutralise, 321322, 335, 338, Algerian, 640
343344, 348 n. 29, 351, 359 Arabic, 596
non, 321, 343344, 359 Al-Hassan, 671
partielle, 321, 323, 344, 347, 349, 349 Al b. Ab Tlib, 134, 206
n. 31, 351, 359 Al H aydar, 23
pleine, 321, 344 Al Slim, 699
Aggadic material, 192 Al Tawfq al-H amad, 187
Agha & Khalidi, 190 n. 1 Al, 117
Agha, Saleh Said & Tarif Khalidi, 207 Ali, 2627, 34, 3637, 3942
agreement, 368, 370, 374, 376 n. 31, 368 Al, caliph, 31
n. 4, 377, 381, 383384, 389, 397 Ali, Mohamed M. Yunis, 44
Aguila, Antonio de, 212, 231232 alif cequin, 218
gyptisch-Arabisch, 699 alif, 123 n. 12, 127, 598
Ahaywt, 566 n. 9 mdda, 684 n. 13
dialects, 567 8skin, 218
ahl al-arabiyya, 55 all, 17
ahl al-lua, 39 al-la, 658
Ahlwardt, Wilhelm, 21, 86 n. 23, 87 a&&a, 658
n. 23, 110 allad, 6772, 7681, 81 n. 8, 82 n. 13,
Ahmad Fud Nigm, 544, 563, 699 83 n. 19, 8485, 8889, 91, 93, 103
Ahmad Muhammad Shkir, 205 n. 41 clause, 68, 7071, 82, 85, 8889,
Ahmad Muxtr Umar, 65 9597, 102
Ahmad ams ad-Dn al-H ajjj, 700 conjunction, 75, 100101
Ahmad Ysuf Najt, 64 conjunctional, 70, 75, 79
Ahmad, 364 infinitival, 79
al-Ahrm, 530 Allh, 69, 72, 89, 9192, 97, 109, 200
id, 59 allemande, 118119
air, 256, 256 n. 8, 257 nn. 1213, 261, allemands, 118, 123
262, 268, 270 allophone(s), 273
ajam, 115116 etymon, 272273, 568, 576, 657
Akkadian, 597604 alternance(s)
early, 598 vocalique(s), 315, 317, 337, 344, 345
pre-, 598 n. 28, 350, 353
lakkadien, 125 Altoma, Salah J., 656, 671
al llad, 79 am, 17, 21
al, 17, 21 amal, 1011, 124, 140 n. 30, 141142,
Al-Ani, Salman H., 247 142 n. 37, 143146, 149152,
Albany, 202 n. 33 152 n. 3, 153, 159 n. 14, 160, 164,
Alcal, Pedrode, 209 n. 1, 210214, 214 166167, 171172, 180181, 184186,
n. 9, 215217, 217 n. 15, 218220, 220 224, 228, 234235
n. 22, 221222, 224226, 231 n. 41, jarr, 10
239, 241 jazm, 10
Alegat, 565 n. 3, 566 n. 7 nasb, 10
Aleiqat, 565 n. 3, 567 n. 9 raf , 10
704 index
Amara and Spolsky, 662 apercion, 225

Amari, M., 84, 110 apocopate, 227, 230 n. 39, 248, 575576,
American 581 n. 56, 592 n. 91
English, 661 Apostolou Panara, A., 299, 301303,
British, 560 307, 309
Americans, 560 n. 40, 561, 684 apposition, 69, 71, 102, 169 n. 26, 182
Amerindian, 210 clauses, 72
al-mid, 37 approximant, 266, 266 n. 28, 267, 272,
mil, 10 n. 8, 124, 135147, 149, 280, 280 n. 41, 285286
151152, 160, 160 n. 15, 161162, aprs
162 n. 18, 163, 168, 170, 175, vnementiel, 325
178179, 181, 184, 212, 234, 237, 240 Aquila, Antonius ab (Antonio
theory, 62 dellAquila), 123, 212 n. 4, 223, 237,
Amn Bakr, 699 239, 241
amir, 217 Aquilina, Joseph, 100, 111, 296, 310
$mmiyya, 403, 405, 411, 529, 529 n. 1, Arab, 26, 41, 236, 527, 534, 540, 543,
531532, 532 nn. 78, 533537, 543 n. 1, 551, 557
539540 countries, 541
amm, 13 n. 9, 1517, 21 genealogists, 201202
Ammar, Sam, 315, 318 n. 7, 319, 323, genealogy, 202
343 n. 25, 344, 344 n. 26, 353 n. 33, grammarians, 70, 79, 206 207
363 grammatical theory, 3
amorite, 316 grammatical thinking, 45
amr, 6, 34, 210, 217 linguistic studies, 41
Amsterdam, 565, 565 n. 2 literary studies, 543
an(na) clause, 24, 95, 99, 171 medieval grammarians, 595 n. 3
an-, 7 n. 5, 15, 79 Muslims, 67
8-clauses, 78 n. 3 nationalists, 197
analogical creations, 661 native speakers, 9
analysis North, 201
automated, 390, 391 philosophers, 192
etymonial, 258, 261262, 265 poetry, 3, 4 n. 1
anaphora, 171 theologians, 192
anaptyctic(s), 572, 574, 582 n. 60 tribal groups, 201
anaptyxis, 571572 world, 527528, 541, 543 n. 1, 561,
Anawati & Jomier, 93, 110 563
Andalousie, 128 writers, 543
al-Andalus, 78, 435, 438441, 444, arabe, 35, 114, 116121, 121 n. 8
447449, 453 122123, 131, 133
Andersen, R., 308309 dAbbeche, 652
angels, 190 ancien, 118119, 122, 124125, 130,
Anghelescu, Nadia, 19 n. 16, 23 313, 315317, 319, 321, 321 n. 10,
animal, 345, 346, 396 323, 325, 327, 332, 339, 343 n. 25,
animate, 367, 370, 383, 394, 396397 349, 353 n. 33, 354355, 355 n. 36,
non-, 394, 394 n. 81 357, 362, 501, 505, 508, 509 n. 21,
animit, 520, 522 509512, 515 n. 33, 519, 520521, 621
AnBs, Ibrhm, 364 dialectes, 125
annam, 16 classique, 116 n. 2, 119, 121 n. 7,
Ansr, 202 124125, 126, 128 n. 18, 130, 335
antecedens, 81 n. 8 n. 20, 510
antecedent, 59, 9192, 151, 153 contemporain, 347, 353
Antilles, les, 129 crole, 621
antinomy, 367 n. 3 dialect, 128
Antiquit grecque, 114 dialectal, 119, 121 n. 7, 122, 124 n. 13
Anubhti Svarpcrya, 209 gyptien, 335
index 705
fash , 120 559, 565 n. 1, 567568, 578, 595,

histoire de, 118-119, 120 n. 6, 125 595 n. 3, 596, 596 nn. 56,
marocain, 698 597598, 598 n. 11, 599604, 604
moderne, 116 n. 2 n. 17, 604605, 607, 609610, 614,
moyen, 118119, 128 n. 18 618, 619621, 639, 640, 641, 640
no-, 118119, 122, 124125, 130 n. 1, 643, 648, 649, 650, 651, 652,
sdentaris, 117 675, 699
vulgaire, 124, 124 n. 13 colloquial, 211212, 214, 223224,
Arabes, 115117, 120, 128129, 133 226, 595596, 596 n. 6, 597, 605,
langue des, 117 621
populations, 114 Damascene, 224 n. 31
populations, non-, 114116, 133 of Granada, 222
arabes, no-, 124 Northern Sudanese, 608
arab, 189 Northern
Arabia, 201, 563 Creole, 621, 639
Central, 563 Juba, 611, 698
Northern, 563 Damascene, 238
Arabian, 201 corpus, 532 n. 5
Ancient West, 652 dialect(s), 189, 578, 595, 677, 687,
Old South, 597, 599603 690, 693
Arabic, 9, 9 n. 7, 26, 35, 38, 61, 6768, dialectology, 72
70, 72, 75, 79, 87, 87 n. 24, 88, 90, dialects, 104, 595 n. 3, 596597, 604,
9293, 95, 98, 101, 104, 106109, 114, 604 n. 17, 658
119, 157, 165166, 168169 n. 26 dictionary, 221
171172, 177, 181, 189211, 211 n. 2, diglossia, 654, 656, 671
212, 213 n. 7, 214, 216218, 221, 222 doxology, 107
n. 26, 223, 223 n. 28, 224, 224 n. 32, early, 201
225230, 232, 232 n. 44, 233, 233 educated Cairene, 7
n. 45, 234235, 235 n. 51, 236, Egyptian, 675, 692693, 698699
238240, 247248, 250, 252253, 255, grammar, 23, 25, 35, 42, 75, 200, 438
260 n. 21, 266 n. 28, 272, 283, 292, grammatical
292 n. 2, 295296, 296 n. 4, 297, 306, literature, 79
309 sources, 23
dAbbeche, 648 terminology, 213, 219 n. 19
Adams-, 192 terms, 210212, 224
Algerian, 596 theory, 216, 219
alphabet, 200201, 219220 tradition, 23, 227, 237
Cairene, 7375, 85, 9495, 100102, early, 23
104, 568, 586 n. 70 Granadan, 212
of Cairo, 696 graphemes, 540
Christian, 71, 89 history, 189
classical, 13, 68, 72, 72 n. 2, 7677, history of, 192 ,211 n. 2
95, 98, 100, 155, 189, 210, 216219, Jewish, 296, 299
222224, 227, 229230, 237, 247, literary, 370
368 n. 5, 369, 369 n. 10, 370, 370 Maghribian, 291, 296297
n. 12, 377 n. 37, 381, 393, 395, 396, Magribian, 309
399, 403, 404, 405, 405 n. 2, 410, Moroccan, 292293, 295297, 308
415, 418, 420, 421, 421 n. 15, 427, Modern Standard, 248249, 251,
431, 433, 435, 437439, 449, 253, 367369, 372 n. 19, 373,
455456, 458, 465468, 470, 376, 376 nn. 31, 35, 377 n. 39,
471472, 475, 486, 486 nn. 1617, 379380, 384, 387388, 390394,
487 n. 18, 489, 490 n. 22, 491492, 398, 568569, 586 n. 70, 598
492 n. 25, 493, 493 n. 25, 527, 529, urban colloquial, of Damascus, 211
529 n. 1, 530531, 535, 537, vernacular, 210
540541, 544, 549 n. 9, 555 n. 31, imperfect, 70
706 index
Ismls, 197 scholars, 194

Jewish, 101 script, 200201
Juba, 128, 128 n. 19, 607612, 616, sentences, 181
618621, 692693, 698 sources, 211 n. 2
Judaeo, 68, 7172, 72 n. 2, 100, 100 South, 201
n. 30 speaking countries, 656, 661
dialect of Tunis, 100 n. 30 speech, 67
Juna, 617 spoken, 652
Jurhum, 200 standard, 621
Khartoum, 608 Egyptian, 679
Koranic, 531 standardization of, 189
Kuwaiti, 563 studies, 528, 565 n. 1
language, 47, 89, 654655, 661, Sudan, 685 n. 14
669672 Sudanese, 608
history, 639, 649, 651 sudanic, 648
Lebanese, 99 syntax, 88
letters, 220 Syrian, 86
lexical items, 668 terminology, 224, 227, 231, 233234,
lexicography, 196 236, 236 n. 53, 237, 239240
linguistic history, 652 terms, 214, 216, 218, 226, 231, 233,
linguistics, 28 237
literature, 544 tradition, 211 n. 2, 216, 220, 227
literary document, 189 Tunisian, 100101, 104
literary, 84 Upper Egyptian, 697, 697 n. 31
Maltese, 100 Uzbekistan, 651 n. 13
masdar, 484 vocabulary, 189, 661
media, 530531, 532 n. 5, 536, 540, Western Sudanese, 614
543 Western Sudanic, 611, 640
medieval, 181182 writing, 200
middle, 72, 72 n. 2, 118 written, 528, 541
modern, 541 Yarubs-, 199, 201
modern Cairene, 72 arabicization, 235
modern literary, 95 n. 28 arabicizing, 661
Moroccan Jewish, 81 n. 10 Arabie, 120
mother tongue, 662 centrale, 120 n. 4
native speakers, 9, 9 n. 7 mridionale, 563, 698
North, 201 arabique
neo, 7576, 95, 103, 118, 639 est-, 125
new, 649, 651 ouest-, 125
newspapers, 654 arabisants, 113, 118, 121, 123, 125
Nigerian, 639641 arabisation, 130
Northern Sudanese, 609 Arabisch, 697
old, 118119, 639, 640 n. 1, 650, Alt-, 118, 119
651 Mittel-, 118
orthography, 189 Neu-, 118, 119
Ottoman, 619 Arabists, 67, 527
poetry, 204 Western, 104
popular, 544 al-arabiyya, 53, 119120, 121 n. 7
pre-diasporic, 640 n. 1, 641, 649 Arabized, 620
pre-modern, 79 Arabness, 189
pre-neo-, 77 Arabs, 26, 33, 40, 86, 106107, 191, 194,
Prophet Muhammads, 205 197199, 201, 203, 205, 207
of prophetic language, 189 North, 202, 217, 227, 230, 235236,
Quraishi, 200 240, 528, 534, 543, 620, 642
religious pre-history of, 192 non-, 620
index 707
arafa, 660 Assyrian, 601

Araic Astarbd, 162 n. 19, 163 n. 20, 164,
Classical, 602 164 nn. 2223, 169 n. 26, 170 n. 28,
Aram, 70 171172, 175177, 186
Aramaic, 72, 189, 197, 201, 595, 597, Ar, 681
600, 605 aswad, 659, 664
Biblical, 597, 599601 Ataba, 538
Jewish, 600601 tr, 30 n. 3
Old, 601 athnach, 219 n. 19
Arazi & Masalha, 110 atnach, 219 n. 19
Arberry, A.J., 87 Atif Efendi, 44
arbre attention, 398
conceptuel, 358, 359 attribute, 71
archaisms, 595 attributive
Archimde, 501 clauses, 6869
ard sr, 194 relative clauses, 69
Arfaxad b. Sm, 195 Audebert, Claude, 335, 363
rid, 169 audience, 249, 405, 424, 432433
al-Ar Auezova, Zifa, 437, 446, 453
dialect of, 570 n. 21 augment(s), 440444, 446, 448
Aristotelian internal, 444
dialectic, 34 augmentation, 441
logic, 25, 41 Austronesian, 210
Aristotle, 33, 41 auxiliaries, 293, 302 379
arrangement finite, 293
morphological, 437 Auz, 199
orthographical, 437 avant
article(s), 209 n. 1, 210, 211 n. 2, 217 vnementiel, 325
n. 15, 226, 571, 573574, 576, 591 n. 88 Avi Shivtiel, L.J., 595 n. 1, 600601
al-, 573574, 576 Avokaya L., 609
il-, 573574, 576 AwaJ, L., 403 n. 1, 434
definite, 214215, 217, 306, 373, 411 awil
indfini, 509, 515 n. 33 -collections, 198
indefinite, 305 -literature, 198
separate [= not suffixed], 229 awmil, 1011, 140, 141 n. 32, 142, 160,
an, 76 185, 232, 234
Ashtor, Eli, 82, 82 nn. 1314, 110 8operants, 10
Asia, 232, 639 Awld Sad, 566 n. 9
Asn, Palacios, 222 Awld Silmiy, 567 n. 3
asl, 438, 448, 507, 507 n. 18 Aws Ab d, 195
marf, 169 awwaliyya, 507 n. 18
asm al-af l, 178 awzn, 314
ASP, 72 n. 2 al-Axfa, 45, 46 n. 3, 47 n. 4, 62, 51 n. 8,
aspect, 294, 370372, 384 n. 55, 389, 64, 154 n. 10, 157, 167168, 175176,
395 180182
perfect, 378, 378 n. 42, 520, 522 ya, 29, 46, 46 n. 1, 5054, 56, 58, 60, 191
Assaf, Simha, 84, 110 ayn, 540
Assemani, Joseph Simon, 220, 220 n. 22 ayna, 16, 20
assimilate, 574 Ayoub, G., 378 n. 41, 400
assimilating, 574 el-Ayoubi, 374 n. 25, 375 n. 29, 381
assimilation, 127, 568, 579 nn. 43, 48, n. 51, 385 nn. 56, 59, 386, 386 nn.
581 n. 55, 586 n. 73, 604 6061, 398 n. 91
of + h, 568 ayy, 14 n. 10
of initial h of suffixes, 568 Ayyoub, Georgine, 654, 655 n. 1,
Assise, Saint Franois d , 121 n. 8 664665, 669, 671
708 index
al-A'am, Muhammad Musta f, 200, Ban Qaht n, 195

200 n. 31, 201, 207 Ban Rala bint Yajub b. Yarub b.
Azz, Araf, 696697 Ldan b. Jurhum b. mir b. Saba b.
al-Azraq, 195, 195 n. 12, 207 Yaq tn b. bir b. lix b. Arafxaad
b. Sm b. Nh, 199
Baalbaki, Ramzi, 3, 6, 7 n. 5, 8 n. 6, Ban Yarub, 204
1223, 230 n. 39, 235 n. 49, 242 baqa, 660
Bb Tma, 222 n. 26 al-Bqilln, 34, 38, 43
bba, 658 baqiya, 660
**ai, 658 Baram, Amatzia, 197 n. 17
Bbil, 196, 202 Barhebrus, 107
Babylon, 196, 203204 Bari, 609611, 618620, 635, 638
Babylonian Barrk Dish Ab Tyih, 562
Old Egyptian, 692 n. 23 base
Tower, 191, 196 bi-consonantal, 256
back consonant, 658 basilectal, 608, 611612, 614, 615, 617,
Backus, A., 294, 310 620
bada llad, 79 Basra, 30 n. 3, 64, 114, 117, 133
badal, 169 n. 26, 182 Basran(s), 160 n. 15, 175176
badam, 17, 19 Basrien, 117
Badrah, 566 n. 9 Bassiouney, R., 407, 434
Badawi, El Said, 158 n. 13, 187, 313, Bath-party, 197
314 n. 3, 363, 371 n. 17, 374 n. 25, Baumgartner, Walter, 605
379 n. 47, 381 n. 51, 385 n. 56, 386 Baumstark, Anton, 105107, 111
n. 64, 387388, 388 n. 67, 400, 542, Bay, 610, 614, 617, 620, 626, 631
655656, 661, 664665, 671, 698 bayn, 38
Badawi-Hinds, 532, 538 n. 23, 542 al-Bayhaq, 203
al-bdiya, 115 Bayram at-Tnis, 563, 685, 700
Badrn, 557 n. 35, 559 Bedja, 693 n. 23, 699
Bagdad, 113 bedouin
Baghdad, 667, 669, 671 Arabic, 599
Bahariyya, 680, 684, 694 n. 25, 697 dialects, 565 n. 2, 567, 573, 575, 586
Bahg Isml, 699 n. 70, 657658
Bahga, 683 Sinai, 568, 571
Bahloul, Maher, 368 n. 4, 371 n. 17, 400 speech, 658
Bahr al Ghazal, 610, 614 tribe, 565566
Bahri, A., 272, 287 usage (qawl al-arab), 47
Bailey, Clinton, 544, 549 n. 11, 563, Bedouin(s), 25, 30, 46, 543 n. 1,
566 n. 4, 566 n. 7, 566, 570 n. 18, 544547, 548 n. 6, 550 n. 22, 552,
577578, 585 n. 68, 586 n. 75 553 n. 27, 557 n. 35, 560 n. 40,
Bailly, A., 276 n. 37, 287 560563, 565, 565 n. 2, 566568,
Bakhtin, M.M., 408, 434 571, 573, 575, 577578, 586 n. 70,
bala, 23 586 n. 75, 657
balbala, 194, 196 badu, 545
Bangala, 609 contemporary poetic tradition, 544
Ban Ati yya, 561562 Egyptian, 578
ban firn, 545 n. 2 in Iraq, 546
Ban H asan, 546 Iraqi, 546
Ban Sad, 206 Jordanian, 562
Ban Sad b. Bakr b. Hawzin, 206 law, 578
Baniy Wsi, 567 n. 9 Nothern Arabia, 563
Ban H m, 203 of Jordan, 546
Ban Hshim, 204205 poet, 545 n. 4, 560
Ban Isml, 195 poetry, 544545, 560 n. 38, 563
index 709
poets, 544545, 560, 563 Blair, Tony, 561

popular poetry, 544, 559 Blanc, Haim, 121 n. 10, 131, 567 n. 13,
Rwala, 698 570 n. 20, 570 n. 22, 573 n. 31, 574 n.
Sinai, 545 n. 2 32, 575 n. 37, 578579, 579 n. 49, 586
society, 544, 563 n. 71, 654, 656, 658659, 664, 671
tribal poetry, 561 Blas Francisco de Salamanca, 210 n. 2
Bdouins, 114115 Blau, Joshua, 68, 68 n. 1, 6972, 74, 77,
Beersheva, 543 n. 1 80, 89, 9394, 99, 111, 119, 131, 654,
Behnstedt, Peter, 567578, 652, 690, 696 655, 671
n. 30, 697 bleaching process, 675, 687
Beja, 685 n. 14 blending, 256259, 259 n. 20, 265, 272,
Bekir, M., 43 277 n. 38, 283, 285, 661
Belot, J.B., 313, 314 n. 3, 363 Bobzin, H., 109, 111, 671, 673
Ben Cheneb, Mohammad, 113, 232 Bohas, George, 4, 23, 255, 255 nn. 2,
n. 43, 242 45, 256, 257 n. 14, 258, 259 n. 18,
Benares, 209 260 n. 22, 262, 264 n. 27, 265, 269
beneficiary, 394396 n. 30, 272, 279 n. 40, 283, 285 nn.
Bengal, 209 4849, 287288, 371 n. 15, 400, 448,
Bengali, 679 n. 7 453
Benjelloun, Fauzia, 674 Bonaventura da Molazzana, 211 n. 2
Benveniste, mile, 320, 335337, 350, Boormans, Maurice, 313, 314 n. 3, 327
356, 363 n. 18, 363
Berk, 105106, 108109 Booth, Marilyn, 544, 563
Bergman, 375 n. 27, 387 n. 66, 400 Borer, Hagit, 494, 498
Bergstrer, G., 33 n. 4, 44 Borg, Gert, 527
Bri, 677, 678 n. 6, 697, 690 n. 20 borrowing
Berrendonner, 506 n. 16 foreign lexemes, 662
Beyrouth, 113 terms, 661
Beziehungsnomen, 69 Bosworth, C.E., 200 n. 28
Bhatia, K., 671, 673 Boucherit, Aziza, 653, 670, 671
Bible, 189, 196, 611, 623 Boumans, Louis, 291, 292 n. 12, 293,
Arab, 223 n. 28 309310
Hebrew, 219 n. 19 Boussofara-Omar, Naima, 404, 404 n. 2,
Jewish, 189 407, 434, 654, 671
Biblical, 106, 198 Brague, Rmi, 502 n. 2, 522
Aramaic, 597, 599601 Braham, A., 315, 363
Hebrew, 595, 595 n. 4, 596, 599, 599 Braslavski, 83, 110
n. 13, 602603 Bravmann, Josef, 28 n. 1, 6870, 111
material, 192 Bravmann, Meir, 44
Bickerton, Derek, 698 Brazilian, 304305
Bihr, 209 British, 544, 546, 560
Biir, 618 Brockelmann, Carl, 86, 86 n. 23, 87
bilabial stop, 595 n. 3 n. 23, 313, 313 n. 3, 327 n. 18, 363,
bilingual(s), 291, 293, 295, 301, 308, 408, 599
433 Broselow, Ellen, 671
bilingualism, 291, 295, 297, 299, 301, Bubenik, Vit, 595 n. 1, 601602
307, 309, 655 n. 2 Buburuzan, Rodica, 28, 44
homeland, 308 al-Buxr, , 78, 90, 110, 199
immigrant, 308 bukara-Dialekte, 578
biqi, 660 bukarra, 578, 580 n. 52, 583 n. 65
Br anNasb, 566 n. 8 Blq, 161
Bir Nasb, 565 n. 3 bunya, 435
biradical verbs, 449 Burton, John, 65
Blachre, Rgis, 125, 131, 313, 363 Bush, George, 560
710 index
CA, 72 n. 2, 540, 659 nominative, 383, 248251, 253

Caballero, 210 n. 2, 224, 224 n. 31, role(s), 369 n. 7, 391, 393394, 394
225231, 233, 238, 240241 n. 83, 395, 395 n. 85
Caballero, Lucas, 209 n. 1, 210 n. 2, 211 system, 217
n. 2, 222, 224, 224 n. 31, 237 Caspari, 363
Cachia, P., 434, 403 n. 1 Caspari-Uricoech, 313, 314 n. 3, 327 n. 18
cadi, 117, 221 n. 26 Castilian
Cadora, Frederic, 655, 667, 671 alphabeth, 213 n. 7
CAI, 659 n. 4 letters, 220
Caire, 113, 131 Castilians, 227
Cairene, 75, 80, 9495, 98, 100, 568, 586 Castro, 532
n. 70, 591 n. 87, 650 casuistique, 503, 503 n. 6
Arabic, 76, 78, 86, 568, 586 n. 70 casus accusativus, 228 n. 35
modern, 72, 94 cataphora, 154 n. 8
Cairo, 83, 101, 528, 598 n. 12, 684, 686, catgorie
687 n. 17 intermdiaire, 321
Arabic, 696 lexicales, 359
Genizah, 101 moyenne, 321
Old, 79, 80 n. 6 noun, 391, 399
caliph Al, 31 particle, 399
calque, 655, 661, 668 smantiques, 344, 359
Cameroon, 640 syntactic, 370
Camilleri, A., 296297, 310 verb, 391, 399
Canadian, 302 categories
Caes, Francisco, 210, 210 n. 2, 211, mixed, 475, 478, 484, 498
212, 222, 224227, 229233, 238, nominal, 477, 485
240241 verbal, 477, 485
canonical, non-, 61 categorization
Cantarino, 374 n. 25, 381 n. 51, 386 verb, 399
cantillation marks, 219 n. 19 Cauallero, Lucas, 224, 228
Cantineau, Jean, 283, 288, 314, 316, 316 Caubet, D., 292 nn. 12, 310
n. 5, 335 n. 20, 363, 690, 697 causal
capacit clause, 88, 92, 108109
de parole, 502 non-coreferential, 92
caractre subordinate, 89, 91
abstrait, 519 conjunction, 76, 91
Carter, Michael G., 4, 4 n. 2, 6 n. 3, 23, connection, 76
25, 3031, 35 n. 8, 39 n. 9, 44, 65, 140 function, 87 n. 24
n. 30, 147, 155, 162 n. 19, 187, 209 non-, 8788
n. 1, 228, 228 n. 35, 234, 240, 243, relation, 8788, 104, 109
313, 313 n. 3, 363, 503 n. 6, 522 structure, 87, 55
case(s), 368 n. 4, 370, 372373, 382384 causative, 441
389, 394, 399 function, 68
accusative, 248251, 253, 280, xalla clauses, 694 n. 26
382383, 476, 484, 484 n. 13, 493 causativit, 313
definite, 375 causativity, 450
indefinite 252 cause(s)
ending(s), 8, 9, 11, 18, 30 n. 2, 141, consequence relation, 263, 267268,
216, 251, 253 278, 284
feature, 373 de transformation, 317318, 318 n. 7
genitive, 248251, 253, 375, 383, 484, Central Arabia, 563
494 Chad, 640641, 648
marking, 374, 384 Chadian, 605
index 711
chanes de raisons, 501 coda, 487

Chambers, Jack, 648, 652 position, 488 n. 19, 489
changement dtat, 326331, 333335, code
339340 mixing, 618, 659, 665, 655 n. 2, 670
changes mixture, 653
educational, 404 switching, 291 n. 1, 295, 297 n. 7,
political, 404 298 n. 8, 300301, 307, 406408,
social, 404 417, 433, 665, 655 n. 2, 670,
technological, 404 672
chercheur, 504 codes
chesmantes, 230 linguistic, 403, 410
China, 223 Codoer, Carmen, 243
Chisarik, Erika, 697 Coelho, Paulo, 535
Chomsky, 503 cognate(s), 597, 597, 602, 653, 655656,
Chrtiens, 123 659 n. 4, 660, 666, 667
Christian lexemes, 653
Baghdadi speakers, 669 cognitum, 235 n. 51
doxology, 106, 109 Cohen, David, 296, 310, 314, 321, 335
faith, 221, 221 n. 26 n. 20, 344 n. 26, 363, 370 n. 14, 400,
liturgy, 105106 523, 697, 697 n. 31
Maronites, 224 Colin, Georges S., 690, 698
scholars, 34 collectivity, 397
speakers, 657 collocation, 391, 393
Christian(s), 79, 93, 106, 109, 195, 209 embedded, 309
New, 212, 213 n. 7 foreign, 309
Old, 212, 213 n. 7 language
circumstantial(s), 391, 393, 395, 399 colloquial Arabic, 595596, 596 n. 6,
local, 395 597, 604, 621, 653655, 655 n. 2,
optional, 395 656670
clan, 545, 565, 565 n. 3, 570 n. 20, 571, Chadian, 605
586 n. 71, 589 dialect(s), 656658, 667
classement, 503504 dictionairies, 658
grammatical, 504 lexemes, 655, 668
classer, 503504 lexical items, 656
classes pairs, 656
nominales, 503 n. 5 Proto-, 597
classical, 213, 428, 640 n. 1 speakers, 659, 664, 665
Arabic, 549 nn. 9, 13, 555 n. 31, Sudanese, 605
567568, 595596, 596 nn. 56, text, 659
597604, 639641 vocabulary, 656
elements, 224 n. 31 words, 666
language, 98, 639 colloquial, 211, 213214, 222, 224 n. 31,
non-, 79 403, 403 n. 1, 422, 529, 531, 533, 536
registers, 223 n. 13, 538, 540
post-, 70 elements, 224 n. 31
clause, 69, 100, 152, 164, 177 n. 35, 180 Egyptian, 529, 540
complement, 369, 376, 376 nn. 30, features, 614
33, 377379, 379 n. 45, 380, 386, modern, 529
388389, 391 speech, 227
Clauson, Gerard, 443 nn. 79, 453 urban speech, 224 n. 31
cleft sentence, 76, 687 n. 17, 695 written, 540
clusters colloquialisms, 224 n. 31
prepositional, 386 Colossoi of Memnon, 678 n. 6
712 index
combinatoire confirmation, 398

binaire, 521522 conjugaison, 316 n. 5, 317, 343, 357,
comitative, 393395 505, 517
commands, 397 arabe, 332
comment, 368 n. 4, 376377, 393, 397 conjugated, 644
n. 89 conjugation, 486, 640
Q+ allad + topic, 96 verbal, 436
Qcompound, 5, 14, 15 nn. 1112, of verbs, 613
1720, 22 conjunction(s), 6769, 7173, 75, 7779,
Qconditional, 17, 2021 86, 87 n. 24, 8889, 95104, 107, 109,
Q-topic, 76 213, 229, 374375, 388, 533, 541
QQstructure, 96 causal, 88, 103, 108109
Qxabar, 216 of comparison, 77
communication, 520, 522 coordinating, 388389
quotidienne, 122 final, 80 n. 6
savante, 122 h ukm arab, 190
community lisn arab, 190
Dutch-Moroccan, 293 subordinating, 388389
Companion(s) of, 193, 206 conjunctional, 6769
the City (ash b al-qarya), 195 allad, 69
the Prophet, 206 function, 71
competence connaissance, 501
language, 294, 308309 conocimiento, 212, 212 n. 7, 214, 217,
complement, 376 nn. 30, 32, 378, 379, 217 n. 14
381, 381 n. 50, 382, 386389 conqute islamique, 114, 116, 120, 128
function, 385, 388 consonant(s), 236 n. 53, 247, 258, 286,
phrase 368, 368 n. 6, 436437, 441444,
interjectional, 387 448451, 567568, 571, 611
prepositional, 375, 385, 390391, augmented, 436, 441, 445
396 basic, 436, 444445, 449
structure, 391 diffusion, 256
complementizer(s), 406407, 419, 676, doubled, 443
692693 emphatic, 614
complments, 518 final, 437, 451
collocational, 309 geminate, 437, 451
completive(s), 380, 391, 393, 395, 399 initial, 437, 451, 572
composites medial, 437, 451
binary, 287 middle, 446
ternary, 287 non-basic, 444, 450
compound(s), 456 non-regular, 437
Dutch, 456 pharyngeal, 614
verbal, 293 root, 371
words, 456 sound, 437, 451
Comrie, Bernard, 324 nn. 1314, 326 uvular, 256 n. 10
n. 17, 363 velar, 256 n. 10
concept vocalized, 447
religious, 471 voiceless, 579 n. 48, 592 n. 91
technical, 470 weak, 437
concord, 212, 368, 368 n. 4, 374, 376 n. consonantal, 276
31, 377, 383384, 389, 397 bi-, 596, 598, 598 n. 10
concrete, 370, 383, 396397 mono-, 596, 596 n. 6, 598
conditional clause, 177 phonemes, 567
conditionnements roots, 644
phontiques, 350 tri-, 598
index 713
consonantalism phontique, 510

bi-, 598 smantiques, 361
mono-, 598 contrle, 321, 343, 348349, 351 n. 32
tri-, 598 imparfait, 321
consonante, 215 n. 11 convention
consonne(s), 319, 506508, 508 structure, 521
n. 21, 509, 509 n. 22, 510511, 511 conventionalization, 294
n. 26, 512514, 515 n. 31 conversational implicature, 36
darrire, 318 coordination, 388, 517518, 520
double, 509 n. 21 asyndetic, 388 n. 67
glottales, 319 coordinator(s), 388
implosives, 512 adversative, 388
longue, 508 n. 21 coordonnants, 519
proto-, 510 n. 24, 511 n. 25, 515 n. 31 Coran, 125127, 130
vocalique, 515 n. 31 du Caire, 127
constant langue du, 120, 125
phonetic, 271 du Maghreb, 127
phono-semantic, 272 coranique, 125, 128 n. 18
semantic, 271 coreferential, 24, 2627, 33, 43
constituency non-, 109, 159, 165, 180
immediate, 369 subject, 92
constituent(s), 367368, 376, 383384, 390 coreferentiality, 92
phrasal, 398399 Cornips, L., 297 n. 6, 310
order coronal, 255257, 259261, 265266,
higher, 291 266 n. 28, 267, 270, 272, 275, 279
lower, 291, 292 n. 1 n. 39, 280 n. 41, 281282, 284286
constitution correspondence
de la langue input, 480, 482484, 493, 495496,
gnrale, 502 498
construccin linear, 479480, 483484, 493496,
de concordancia, 212 498
de rgimen, 212 quantitative, 479 n. 4
construct state, 574 Corriente, Frederico, 213, 215, 215 n. 11,
construction(s) 217 nn. 14, 15, 243
auxiliary, 293 countable, 396
impersonelle, 343 couplets, 548 n. 5
periphrastic, 293, 295, 296 n. 4, Cowan, William, 210, 213 n. 7, 214, 215,
297, 297 n. 6, 299304, 304 n. 11, 215 n. 11, 217218, 218, n. 17, 243,
305309 365
possessive, 493 Cowley, 320 n. 9
contextual, 5, 13 crement
elements, 5 initial, 262, 262 n. 25
continuant(s), 256, 256 n. 8, 257, inset, 262 n. 26, 265
260261, 266270, 275276, 278, crementation
280282, 285286 initial, 264
non-, 256 n. 8 creole, 692, 698
continuity, 247 Arabic, 639
Contossopoulos, N.G., 298, 310 Juba-, 611
contour principle languages, 692
obligatory, 259 n. 19 crole(s), 128129, 607, 608, 620621
contractual language, 25 tudes, 128
contrainte(s) crolisation, 129
formelles, 361 crolistes, 129
morphologiques, 361 creolistic, 608
714 index
creolization, 607, 621, 652, 699 endings, 227

crosslinguistically, 698 non-, 227
Crystal, David, 27, 44 decreolization, 608
culture dcrochage, 361
Arab, 424 deductive application, 25
Islamic, 468 dfaut, 345346
curvature, 257, 264, 270, 279280 defective, 252253
Cuvalay, Martine, 378 n. 41, 400 deficiency (illa), 53
Cuvalay-Haak, Martine, 400 dfinir, 503504, 506
Czapkiewicz, Andrzej, 192 n. 2, 207 definite, 250251, 253, 172, 373
marifa, 47
D ahab, 569 n. 16 article, 17 n. 13, 611613
Dahbi, 674 definiteness, 175, 368 n. 4, 370, 382, 384,
colloquial Arabic, 659, 662 389
discourse, 665 dfinition(s), 455456 504, 504 nn. 89,
dim, 56 505, 505 n. 12, 506 n. 16
Dakhla-Oasis, 690 deixis, 27
dlika, 15 n. 11 Dell, F., 256 n. 7, 288
Damas, 113 demonstratives, 373, 406407, 419
Damascene, 650 denativization, 308
Franciscan tradition, 211 Deng, 612, 622, 624625
Damascus, 210, 211 n. 2, 221, 221 n. 26, dnombrements; entiers, 501
222, 239, 532 denominatives, 373
damir, 210, 217218, 239 dental, 262
damm, 227 dependencies, 368369, 376, 381382
damma, 14, 218, 218 n. 16, 230, 239 Derenbourg, Hartwig, 43, 143 n. 43, 187
damma, 224, 231 derivation, 370, 372, 374, 395, 486
dammatn, 218 feature, 372373
Dankoff, R. and K., 437, 453 droulement
Dannenfeldt, 210 n. 2, 219 du temps, 505, 512
Dante, Alighieri, 121 n. 8 Descartes, Ren, 501, 503, 522
darf, 216 Descls, Jean-Pierre, 313 n. 1, 321, 324,
Darfouf, 255 n. 5, 287 324 n. 14, 325, 325 n. 15, 326, 326 n.
ad-Drim, 89, 110 17, 330, 332333, 336 n. 21, 341, 347,
Darwin, Charles, 274, 274 n. 35 362363
Dat, 255 n. 4, 257 n. 14, 272, 288 description(s), 456458, 465467,
Dat na, 690, 697 n. 31 469472
datiuo, 214, 227 paraphrased, 455456
dative, 214 formal, 367369, 369 n. 10, 370, 372
dativus, 221 nn. 1819, 374, 376 n. 35, 377
Dawsir, 563 nn. 3637, 39, 387, 391
Day of Judgment, 190 linguistic, 367, 376, 383, 385
Dbr, 566 n. 9 phonemic, 368
de Jong, Rudolf, 565, 567 nn. 10, 1213, translations, 467, 469
569 n. 17, 570 nn. 21, 23, 572 nn. 25, descriptive
2728, 573 nn. 3031, 574 n. 32, 575 analysis, 48
nn. 34, 37, 577 n. 40, 578, 583 n. 63, compounds, 696
586 n. 70, 591 n. 88, 652, 675 n. * dterminants, 513, 516
de Sousa, 211 n. 2 determination, 373, 374
dclaration, 517 n. 44, 518520, 520 n. 47 Dvnyi, Kinga, 23, 46 n. 1, 44, 45, 56
du lien, 517 n. 12, 65
declarative sentence, 169 dhamma, 233
declension, 373, 436 Diab, 257 n. 11, 288
declensional, 227228 diachronic, 67, 74, 656
index 715
dialect(s), 10, 97101, 216, 223, 403, inventory, 667

403 n. 1, 404, 404, n. 2, 405408, 410, Marocain, 698
427, 543, 545, 567, 573, 570 n. 20, dialecte(s), 120, 563
571, 573, 577578, 586 n. 71, 591 nn. ancien, 126
8889, 595, 595 n. 3, 596597, 604, arabe, 100, 128 n. 19, 129
640 n. 1, 641 n. 4, 650, 650 n. 12, 642, moderne, 126
653, 657, 665, 669674, 687 de larabie mridionale, 698
of the Agylah, 567 dialectic(s), 22
of the Ahaywt, 567 Aristotelian, 34
Arabic, 640641, 677, 690 dialectical, 34, 42
of the Arabs, 543 dialectique, 114
of al-Ar, 570, 570 n. 21 dialectologie historique, 129
bedouin Sinai, 571, 575, 578, 586 dialectologists, 211
n. 70 dialectology, 72, 605
Bri, 690 n. 20 diaspora communities, 302, 306
Cairene, 540 diathse, 316 n. 5, 335, 336
contemporary, 640 externe, 336337, 352, 353 n. 33,
of Damascus, 222 354360, 362
dependent, 658 objective, 513, 519
differences, 657 n. 3 subjective, 513, 519
Eastern Arabian, 555 n. 30 diatopic, 74
Egyptian Arabic, 675, 697 n. 31, 699 Daz, Lourido, 210 n. 2, 211 n. 2,
English, 596 n. 6 243
features, 658 dichotomy, 639
of the Jarjrah, 567 Dichy, Joseph, 313, 315, 317, 318 n. 7,
of the H amdah, 565 319, 323, 327, 341 n. 24, 343 n. 25,
H amdiy, 567 344, 344 n. 26, 347, 353 n. 33,
H ijz, 16 362363, 394 n. 81, 396397, 400
Jewish, of Tripolis in Libya, 100, 104 dictionaries, 455459, 462, 465473
of the Lgt, 565 Arabic, 455
Lgiy, 567 Arabic-Dutch, 455, 472
local, 403 Arabic-English, 578, 605
modern, 70, 75, 94, 100, 102, 527, 639 bilingual, 455456, 471
of the Mznah, 567, 567 n. 12, 568 classical and MSA, 532 n. 7
Mzniy, 575 Dutch-Arabic, 472, 455
of Negev, 570 n. 22, 565, 567, 574 Egyptian Arabic, 604, 698
n. 32, 575, 575 n. 37, 8 English-Arabic, 457
Neo-Aramaic, 595 n. 3 etymological, 443, 443 n. 7, 446, 449,
North African, 659 455456, 472
rural, 676 French-Arabic, 457
of the Samnah, 572 n. 25, 574 n. 32, 567 German-Arabic, 457
of Sinai, 568, 573, 574 n. 32, 575, 575 Hamito-Semitic, 605
n. 34 Hebrew, 605
bedouin, 568 lexicographical, 437
of southern Sinai, 567, 568, 568 n. 14, Modern Arabic, 605
572 n. 25, 573, 573 n. 29, 575 n. 35 monolingual, 455
speakers, 657 Moroccan Arabic, 605
Tamm, 16 multilingual, 455
of the Tarbn, 567, 569 n. 17, 575 dictionnaire(s), 315, 317, 326 n. 16, 340, 347
of the Tayhah, 567 dArabe Dialectal Marocain, 698
of Tunis, 100, 104 informatis de larabe, 315
Yemeni, 640 n. 1 didactic, 220
dialectal, 80, 81 n. 10, 403 n. 1, 405, 405, Diem, Werner, 8182, 110, 130131,
n. 3, 414, 422, 608 654, 664, 665 n. 8, 666, 671
716 index
Dietrich, Albert, 93, 110 Doniach, 473

difl (oleander), 21 dorsal, 256 n. 10, 257, 264266, 270,
diglossia, 189, 403404, 415, 421, 527, 279, 279 n. 39, 280
653, 654, 655 n. 2, 656, 664, 671672 non-, 256, 275
diglossic, 212 Doss, MadBha, 529
differences, 653 doubt, 398
makeup of Arabic, 670 Doxologie, 106107
situation, 403, 528 doxology, 106109
diglossie, 121 Driver, G.R., 597 n. 8
diglossique, 125, 129 dropping (ilq), 53
DIINAR, 391, 391 n. 72 Drost, 218 n. 18
diminutive(s), 373, 444, 450 Druze speakers, 657
dimuqrtiyye, 657 \ l-Qarnayn, 190
ad-Dnawar, 202 dual, 82 n. 15, 91, 153, 227, 383, 397
aJ-Xindn, 563 n. 86
Dinka, 610, 612613, 616, 622, 624 dure, 324325, 327328, 338, 341,
Nilotic, 610 351352
diphthong(s), 406407, 419, 491 n. 23, Dutch, 293, 295297, 297 n. 6, 299
568569, 576, 650 n. 12, 657 n. 9, 309, 455458, 465482, 565 n. 2
diptote, 124, 127 non-standard, 297
drah, 544, 566 nn. 78 standard, 297
direct speech, 534, 537, 539, 676, dzhamma, 226
678679, 683
disambiguation, 367 n. 2, 369, 390, 398 Edman, Johan, 222 n. 27
discourse, 368, 656, 672 Educated Spoken Arabic, 665
functions, 300, 403, 417 education, 404
marker(s), 619, 680681, 683, 685, Egyptian, 403, n. 1, 405406, 407,
687, 691 n. 5, 407423, 425, 429, 433
political, 405, n. 3 higher, 404
record, 681, 681 n. 9 public, 404
manipulation, 681 n. 9 Egypt, 81, 83, 83 n. 18 197, 224 n. 31,
routine, 680682 528530, 536, 544, 545 n. 2, 563, 566
strategies, 410 n. 6, 578, 655656, 662, 666, 671673;
structure, 27 686, 690
dispensable part, 136138, 142, 146 Upper, 690 n. 20
dispersion, 350, 356357 lEgypte, 130
semantique, 313, 315, 321, 356 Egyptian(s), 105, 190, 537, 544, 545
dissent, 397 n. 4, 578, 590 n. 84, 598 n. 12,
Ditters, Everhard, 367, 368 n. 4, 369 nn. 600604, 675, 679, 687
8, 10, 374 n. 26, 376, 376 nn. 31, 35, Ambassador, 529
379 n. 43, 382 nn. 5253, 385 n. 58, Arabic, 675, 679, 692693, 698699
386 n. 62, 400401 Arabic dialects, 675
divinity, 404 Babylonian Old, 692 n. 23
diviser, 501 history, 528
Dwn al-Adab, 436438, 445, 446 n. 11, 447 egyptianisation, 539
Dwn al-HudalByBn, 110 Eid, Mushira, 402403, 403 n. 1, 405, n.
Dwn Lut at-Turk, 437, 441447, 449, 3, 407, 410, 434, 498, 671
450452 Eisele, John, 371 n. 17, 384 n. 55, 401
dwns, 545 El Aissati, 309310
do-construction(s), 301, 307 El Assal, 403, n. 1
Dombay, Franciscus de, 223, 241 El Escorial, 210 n. 2, 211 n. 2
Dominicus Germanus de Silesia, 124 elatives, 373
n. 13, 221, 223 lment(s), 504505
domma rafo, 225 embedded, 292
index 717
Elgibali, Alaa, 402, 671 Erpenius, Thomas, 211, 231234, 236,

El-Hassan, 665 236 n. 52, 238, 240241
elision, 23, 572 Escavy, R., 243
Elzeiny, Nagwa, 375 n. 28, 401 espace, 512, 512 n. 27, 517 n. 41
emphatic(s), 284, 569, 642 n. 6, 653, lespagnol, 115
656658, 571 Ess, Josef, van, 192 n. 3, 194 n. 6
consonants, 658 tat(s), 317, 321, 323324, 324 n. 13,
lateralized, 283 325326, 326 n. 17, 327333, 337,
sounds, 651 n. 13 339340, 342, 347359
enantiosemy, 272, 272 n. 33, 286 accompli, 334
Encarnacin, Juan de la, 210 n. 2, 211 prsent, 332, 325 n. 15, 341
n. 2 rsultatifs, 330331, 347
ending(s) non-caractristiques, 329
feminine, 492, 250 Ethio-Semitic, 697 n. 31
inflectional, 216218, 225, 231233, etymological, 13, 14, 17, 2021, 91
239 etymology, 5, 8, 13, 17 n. 13, 2022, 203,
English, 9294, 291, 296299, 299 n. 262, 436, 440, 444, 696 n. 30
9, 300304, 304 n. 11, 305309, 391, etymon(s), 256258, 258 n. 15, 259262,
395, 399, 456, 465, 483 n. 10, 493, 548 262 n. 26, 264268, 270275, 277, 277
n. 5, 569, 578, 583 n. 63, 595, 596 n. 38, 278281, 283285, 287
n. 6, 605 homonymic, 261
Old, 596 n. 6, 609611, 618, 626, lEuphrate, 128
661662, 668, 673, 681, 681 n. 9, 682, Euphrates, 196
686, 692693 Europe, 209210, 219220, 121 n. 8,
ensemble, 503506 223, 232
entertainment, 403, 425, 432 Northern, 231, 240
enthusiasm, 397 European(s), 221 n. 26, 608
entit(s), 503, 505, 508, 512 n. 27, language, 213
520 scholars, 222
abstraites, 512 n. 27 vnement(s), 321, 324, 324 n. 13,
communes, 516 325326, 326 n. 17, 327, 330334,
multidimensionnelles, 512 339340, 342343, 347, 349, 351352,
universelle(s), 516, 516 n. 41 359, 505
entity(ies) event, 476
animate, 396397 nominal, 475
non-, 396 vora, 219 n. 20
human, 396397 Evripidou, D., 300, 303, 310
non-, 396397 Evstatiev, Simeon, 44
perceptible, 396 exceptive, 19, 21
pseudo-animated, 397 sentences, 171
entry(ies), 435437, 449, 456, 465466 excitative (tah /:/), 16
derived exclamation(s), 375, 387, 397398
non-, 375 exegesis, 49 n. 5
noun-, 375 Quranic, 438
double, 446 exegetes (ash $b at-t;wl), 54, 193
Dutch, 455 exegetical
lexical, 369, 381, 383, 395 methods, 63
verbal, 369 n. 7, 377, 379380, 391, tradition, 63
395, 397 n. 86, 391392, 394396, grammatical, 64
399 works, 45
epenthetic, 489 Exodus, 105
poque prislamique, 120 exprience(s), 505, 508, 511510, 520
Equatoria, 609, 611, 621 non-physical, 396
Ermers, Robert, 435, 437, 441, 449, 453 experiencer, 394, 396 n. 85
718 index
explicit, 368 n. 4, 378380, 389, 392 Ftihah, 105

n. 77, 397 n. 86, 394 fatina, 668
expression(s), 455456, 466, 469 faula, 313314, 314 n. 3, 316317, 320,
adverbial, 484 322, 327329, 337338, 338 n. 23,
frozen, 375, 375 n. 27, 387 340341, 343344, 350, 353, 353
extending meanings, 661 n. 34, 357358, 360361, 435436
extensions, 518 al-Fayrz bd, 353, 353 n. 35
verbal, 379, 380 feature(s), 368370, 370 n. 12, 371, 371
extra-textual linguistic source, 46, 61 nn. 1517, 372 n. 18, 373, 376, 381,
al-Ezbekiyah, 538 383, 389390, 392 n. 75, 394, 396398
clusters, 369, 383, 389
fa- after amm, 60 ideal, 386
f, 7 n. 5, 21, 38 inherent, 367 n. 1
faala, 313, 314 n. 3, 314, 316323, 337, inherited, 367 n. 1
350351, 353358, 360361, 435, 440 linguistic
factitif, 115 universal, 368
factitivit, 313 manner, 386
fil, 151152, 154, 154 nn. 7, 9, morphological, 371, 373
155158, 158 n. 13, 161, 164, 167, names, 367 n. 1, 370, 372, 372 n. 18,
175184, 323, 329 373 nn. 20, 23, 385, 391
faila, 314, 314 n. 3, 314, 316318, noun, 396
320323, 327, 327 n. 18, 337338, place, 386
344347, 349350, 351 n. 32, 356, prepositional, 396
358, 360361 semantic, 368, 368 n. 5, 369, 369 n. 7,
fal pattern, 659 383, 386, 390, 397399
falsafa, 114 semantico-syntactic, 369 n. 9
familles morphologiques, 315, 324325 time, 386
n. 15 values, 367 n. 1, 369 n. 9, 370, 370
far structure, 151 n. 11, 371372, 372 n. 18, 373 nn.
al-Frb, 4243, 120, 131, 436438, 445, 20, 23, 383, 385386, 390, 391, 392
447, 453, 504 n. 8, 513 n. 29, 522 n. 77, 393, 398
Farj, Muhammad, 247, 249, 253254 local, 390
Farazdaq, 7, 8 n. 6 semantic, 399
al-Fris, 11, 143 n. 43, 144 nn. 4950, variables, 373, 389, 397398
147, 164 n. 23, 165, 165 nn. 2425, featuring, 367, 367 n. 3, 380, 393
166168, 171, 177, 180, 185186 morphological, 368
al-Farr, 4546, 46 n. 3, 4748, 49 n. 6, nominal, 369
50, 50 n. 7, 5156, 56 n. 12, 57, 57 phoneme, 371 n. 15
n. 13, 5862, 62 n. 15, 6364, 64 semantic, 368
n. 17, 125, 129, 175 n. 33 syntactic, 368
fasd, 114, 129, 194, 207 verbal, 369
al-lua, 118, 124, 128 Feghali, Michel, 99, 111
fash a, 120 feminine
Fasold, Ralph, 655 n. 2, 670671 ending, 231
Fassi Fehri, Abdelkader, 386 n. 63, 401, marker, 182
475, 475 n. 1, 484485, 498 Ferguson, Charles A., 121 n. 9, 131, 404,
fatah ahu, 659 654, 656, 664, 667, 671
fataho, 659 Fernandez, Mauro, 654, 672
fath , 227 Ferrando, Ignacio, 126, 131
fath a, 1415, 218, 218 n. 16, 224225, Fez, 232
226, 230231, 239 figement lexical, 362
fath atn, 218 fil, 151, 154, 154 n. 9, 156157, 163
Fathi RaJwn, 699 n. 21
Fathiyya l-Assl, 699 +fil, 177
index 719
Filip IV of Spain, 223 n. 28 fraccion, 225

filiyya, 149, 177, 158 n. 13, 165 n. 24 franais, 331, 339, 343355, 355 n. 36,
fillers, 377379, 382, 385389 511 n. 25
Fillmore, Charles, 369, 401 Franciscain, 123
finite Franciscan(s), 210, 211 n. 2, 212, 221,
forms, 488 221 n. 26, 222, 231232, 232 n. 44,
verb, 163 237, 239240
fiqh, 31 authors, 211
Firanescu, Daniela, 387 n. 65, 401 Spanish, 211
firaq islmiyya dlla, 530 College, 222
Fischer and Jastrow, 656, 664 tradition, 222
Fischer, A. & A.K. Irvine, 208 Frank, Armin Paul, 243
Fischer, Wolfdietrich, 672, 698699 Frayha, Ans, 666 n. 9, 672
fiten, 668 French, 119, 292, 292 n. 2, 295296, 297
Fleisch, Henri, 314, 316, 316 n. 5, 320 n. n. 7, 298, 299 n. 9, 300, 302, 308, 456,
9, 321, 321 n. 10, 328329, 335, 335 465, 661
n. 20, 344 n. 26, 364, 371 n. 15, 373 n. fronted, 169 n. 26
22, 378 n. 40, 401, 513 n. 29, 522 fricative(s), 260, 282, 407, 597, 657
Fleischer, Heinrich Leberecht, 118119, final glottal, 639
121, 121 n. 7, 131, 132 interdental, 407, 407 n. 5, 409410,
flexion 413, 418420
casuelle, 125126 voiced, 419
dsinentielle, 117118, 120123, 130 voiceless, 268, 419
triptote, 130 Fritz Paul, & Horst Turk, 243
language, 662, 670 Fck, Johann, 119, 124 n. 13, 128, 132,
lexemes, 655656, 661 210 n. 2, 219220, 233, 243, 654, 672
loanwords, 661662, 668 Fuentes, Pedro, 223 n. 30
terminology, 661 function, 380
words, 662 word(s), 675, 676 n. 4, 689, 690
form, 380 Funktionsschwche, 75, 94, 96
pausal, 248 Funktionsschwchung, 69, 86
prejunctural, 250252 fush $, 403404, 404, n. 2, 405, 405,
prepausal, 250252 nn. 23, 406407, 407 n. 5, 408410,
formal, 403405, 423 410, n. 8, 411416, 416, n. 12,
literally Arabic, 658 417423, 427433, 529
formalism, 369 n. 10, 370, 396 al-Fust ^t , 7984, 101, 130
formality, 405, 410
format, 369 Gabriel, John, 615, 618620, 626, 635
formation gahawa(h)
external, 373 complex, 641 n. 5
internal, 373 syndrome, 572, 588 n. 81
forme(s), 315 n. 4, 332, 357358, Galloway, George, 561
360362 gap(s)
augmentes, 314 lexical, 455456, 458, 465472
construites, 316 Garawiyya, 609612, 614, 619620,
drive(s), 313, 313 n. 3, 314 622
verbale, 332, 362 Gardner-Chloros, P., 303, 310
forms azla, 690
bound, 374, 384 Gedichte, 563
finite, 369, 488 Geez, 597, 599603
free, 374, 384 Geiger, Ludwig, 231 n. 41, 243
hispanicised, 229-230 el-Gemei, Dalal M., 391 n. 71, 401
non-finite, 488 geminates, 611
paradigmal, 449 gemination, 489 n. 21, 490, 491
720 index
Gen(esis), 191, 194, 196, 198, 202 Goitein, Dov, 80, 80 nn. 4, 6, 8182, 110111
gender, 154 n. 6, 249250, 367, 368 Golato, Andrea, 684, 698
n. 4, 370374, 378, 380, 382383, 389, Goldenberg, Gideon, 7679, 111,
395, 615 153154, 163 n. 21, 173 n. 31, 174,
affixes, 611 187, 595 n. 1, 596 n. 6, 597 n. 8, 598
genealogical traditions, 192 n. 12, 600
genealogists, 203, 205 Goldziher, Ignaz, 119, 132, 208
gnitif, 130 Golius, Jacob, 211, 238
genitiuo, 214, 227 Gonzles Palencia, A., 43
genitive, 7, 10, 13, 106, 135 n. 2, 144, Gonzlez, Bernardino, 210 n. 2, 211 n. 2,
214, 216, 227228, 230, 230 n. 39 221222, 226
genitivus, 221 Gonzalez, Francisco, 212, 231, 238
Genizah, 7980 Goral, Mira, 655 n. 2, 673
genre, 513 Gottheil & Worrell, 8184, 111
Georgakopoulu, Alexandra, 677 n. 5, Goutsos, Dionysis, 300301, 310, 698
679 n. 8, 698 grammaticalized, 78
German, 81 n. 8, 9294, 98, 104, 456, grammaire(s), 113114, 118, 121, 123
465, 679 n. 7, 698 n. 12, 502 n. 4, 507
German-Arabic, 458, 462, 466468 arabe(s), 116117, 126, 130, 314, 348
gerund(s), 299, 301, 305, 476, 476 n. 2, n. 30
477479, 482, 492 arabisantes, 313, 319, 327
English, 475476, 482, 493 compar, 118
nominal, 477 gnrale, 517 n. 44
Gesenius, 320 n. 9 lhistoire de la, 114
Geva-Kleinberger, Aharon, 658, 672 grammairien, 113, 118, 122, 126
Ghassn Surr a-baylt (Ab Surr), arabes, 124125, 129, 505506
546548, 551556, 559 non-arabes, 506
Ghawnmah, M., 548, 563 grammar(s), 3, 4 n. 1, 23, 2526, 28, 31,
Ghazeli, et Hassour, 315 33, 37, 4243, 45, 62, 77, 79, 171, 209,
ghost words, 472 209 n. 1, 210, 211 n. 2, 212, 213, 215,
Giargianius, 235 217 n. 15, 218 n. 18, 219, 219 n. 19,
idir, 660 220, 223, 224, 226227, 232, 234236,
Giesbers, H., 297 n. 6, 310 239240, 370, 527528, 617
Gil, Moshe, 80, 80 nn. 4, 6, 81, 81 n. 8, of de Alcal, 219, 219 n. 19, 226, 232
82, 84, 110 Arabic, 248, 252253, 435437, 439,
Gilliot, Claude & Larcher, Pierre, 195 442, 448449
n. 11, 208 Classical, 223, 249
glide(s), 368, 437, 441, 446450 Granadan, 212
inserted, 444 of Caballero, 209 n. 1, 210, 210 n. 2,
glissement(s) 221, 222, 224 n. 31, 238
de sens, 338, 342 of Caes, 211, 221, 222
smantique(s), 321, 326, 334335, case, 369, 381, 393, 399
338341, 354358, 360362 Castilian, of Antonio de Nebrija, 215
global, 405, 410, 417, 433 dependency, 369 n. 8
glottal, 657 early, 25, 49 n. 5
catch, 571 formal, 367, 388
stop, 595 n. 3, 596, 596597, 597 n. 8, of Greek
598, 598 n. 12, 639640 nn. 12 classical, 223
final, 642 modern, 223 n. 30
goal, 380, 394395 generative, 4
God, 26, 3537, 3942, 189, 190191, of Hebrew, 209
193194, 196200, 202204, 207 Indian, 209
Godefroy-Demombynes, Maurice, 313, of Japanese, 210
314 n. 3, 363 Latin, 216, 239
index 721
of Martelottus, 238 systems, 4

medieval, 213 term(s)
missionary, 213 Arabic, 210
North-European, 211 exo-, 218
Renaissance, 213 technical, 217
of Sanskrit, 209 terminology, 233
Semitic, 605 orientalising Western, 240
Spanish, 210, 211 n. 2, 212, 223 n. 30, endo-, 240
237 theory(ies), 8, 228, 240
of Tagalog, 219 n. 19 Arab, 3
Tamil works of, 210 medieval Arab, 168, 178
traditional, 1112 non-Western, 239
transformational, 396 Western, 239
of Turkish, 224 n. 30 thinking
grammarian(s), 3, 68, 10 n. 8, 1113, medieval Arab, 152, 168169
17, 17 n. 13, 20, 22, 26, 33, 3940, thought
4546, 60, 70, 79, 135, 135 n. 2, 138 medieval Arab, 180
n. 21, 142, 143 n. 43, 146, 149, 150, tradition, 42, 64, 211 n. 2, 210, 227
150 n. 2, 151, 152, 152 n. 3, 153 n. 5, Hebrew, 209
154, 154 n. 7, 155157, 158, 158 n. 13, Italian, 211 n. 2
160165, 167170, 172, 173, 173 n. 31, Indian, 209
174175, 177179, 181, 184186, 220, of the Franciscans, 211 n. 2
223, 225, 235, 240, 595 n. 3 medieval Arab, 149
Basran, 175 works
classical, 5 canonical, 220
Franciscan, 212, 238 writings, 160 n. 15
medieval, 150, 154, 160, 162, 176177 grammaticalisation, 294, 621, 697699
Arab, 184 grammaticalization, 675, 676 n. 4, 678,
missionary, 238 681, 681 n. 9, 682, 684, 685 n. 14, 687,
traditional, 11, 13 n. 9 687 n. 16, 688 n. 19, 697699
grammatical, 34, 4 n. 1, 6,8, 10, 12, endo-, 236
15, 1718, 19 nn. 15, 17, 21, 23, 25, exo-, 236
3334, 4546, 46 n. 3, 47, 47 n. 4, 48, grammaticalized, 581 n. 54, 616, 675,
5557, 6164, 64 n. 17, 7879 677, 683684, 689, 690, 692
analysis, 4, 6, 12, 2223, 42, 4648, 49 grapheme(s), 540, 598
n. 5, 62, 64, 142, 146 Arabic, 371
categories, 34, 38 grec(que), 124, 130
coherence, 655 n. 1 classiques, 124
description, 209 grco-arabe, 119
features, 33 Greco-Latin, 219, 226, 228
form, 25, 30 Greek, 34, 105106, 108, 189, 193, 213,
framework, 210 221, 661
literature, 79 American, 301302, 304
medieval, 186 Classical, 223
Arab, 158 Cypriot, 300301, 303
medieval Arabic, 181 doxologies, 107108
markers, 679 n. 7 ethics, 42
monographs, 239 expansions, 108
problems, 25 grammar, 228 n. 35
rule, 49, 5762 Modern, 223 n. 30
of jaz, 59 vernacular, 223, 233 n. 46, 291, 298,
structure, 48, 152 n. 4, 655 298 n. 8, 299301, 303304,
of sentences, 160 306307, 309
subject, 694 verbal system, 108
722 index
greetings, 387 H amdullh, M., 43

formulaic, 375 Hamilton, Charles, 31, 44
Gregory James & Emilio Ridruejo, 244 H ammm Farawn, 566
Grice, H. Paul, 2730, 42, 44 H amw, Subh, 364
Gricean, 28, 29 n. 1 hamza, 214 n. 10, 226, 437, 441,
maxims, 33, 42 444445, 598, 598 nn. 11 12
griechisch, 106 initial, 437, 445446, 451
Grimshaw, Jane, 475, 475 n. 1, 476, 476 al-wasl, 598
n. 2, 498 H anaf, H ., 43
Gruntfest, Y., 123, 132 H anafs, 31, 35
Guadagnoli, Philip, 211 n. 2, 232235, h arakat, h arakt, 211, 225, 231
235 n. 51, 238241 ayn, 513
Guadaoli, Felipe, 212, 231 h aram, 199
Guentchva, Zlatka, 324 n. 14, 330, h arf, 165 n. 24, 218, 507 n. 18
362363 8jarr, 135
Guerssel, M., 256, 288 8nid$, 41
Guillaume, Jean-Patrick, 318, 317 n. 7, 8tat niya, 182
318 n. 8, 364 8w$id, 14, 1617
Gldemann, Tom, 676 n. 3, 697698 Harmon, R.M., 305, 310
Gully, Adrian, 313 n. 3, 314, 363 Harnell, Richard S., 605
guttural(s), 256 n. 9, 262, 266 n. 28, 280 H arrn, 196, 197
n. 41 Harsusi, 604 n. 17
consonant, 658 Hartley, A.H., 298, 301, 303, 310
Hartmann, Regina, 396, 401
Haak, Martine, 379, 401, 578, 652 Hrn, Salm, 143 n. 43, 161, 368 n. 6,
h abbad, 17 370 n. 13, 378 n. 40
habitudo, 215, 215 n. 12 h asan, 26
h add structure, 151 Hashemite(s), 549, 552
h adf (deletion), 61, 163 H assn, Tamm, 247, 249, 250, 254
al-h dira, 115 Hasse, Johann Gottfried, 123
H adt , h adt , 25, 30 n. 3, 36, 120, 163 Hasselmo, N., 308, 310
n. 21, 173 n. 31, 192, 198, 206, 530 Hassoun, Mohamed, 315, 363
canonical, 199 h att, 21
scholars, 30 n. 3 Haugen, E., 308, 310
H adramawt, 14, 16, 199 Hauschild, Richard, 209210, 243
Haeri, Niloofar, 404, 434, 654, 657, Hava, J.G., 578, 586 n. 77
665666, 669, 672 Hawa, 614615, 620
H afs an sim, 127 h wiy, 583
Hagar, 197 h ayawn
al-H akm, 315, 362 ntiq, 502
al-Hakkak, Ghalib, 313, 364 h ayt um, 1417, 19
h l, 50, 50 n. 7, 57, 57 n. 13, 60, 136, Haywood, 437
138, 138 n. 19, 139142, 145146, head(s), 69, 7374, 76, 8082, 84, 8789,
159160 9091, 9699, 103104, 156, 163, 368
hall, 17 n. 4, 376, 376 nn. 33-34, 377 n. 38,
Halle, M., 256 n. 8, 288 378379, 38182, 382 n. 53, 383390,
halumma, 13 n. 9, 16, 17 393, 396, 482
H m, 196 function, 376, 381, 381 n. 50, 384,
H amdah, 565, 567 n. 9, 571 388
h amdiy, 566 n. 8, 567, 573, 575576, nominal, 89, 480, 485, 494 n. 26
576 n. 39, 577 phrasal, 380
dialect, 568, 568 n. 14, 569, 571573 phrase
h amd, 46, 47 interjectional, 387
H amdalah, 67, 70, 85, 9192, 96, 104, verbal, 378380, 393, 480, 482, 485,
106107, 109 494, 496
index 723
header(s), 386, 390 history

prepositional, 390 Arab world, 528
Heath, Jeffrey, 295296, 308, 310, 653, Hoberman, Bob, 595 n. 1, 3, 596 n. 6,
670, 672 598
hbreu, 320 n. 9, 321 n. 10, 354355, Holes, Clives, 128, 132, 248249, 254,
355 n. 36 543, 563, 654, 656, 664665, 672
Hebrew, 68 n. 1, 70, 80 n. 6, 108, 189, Holy
193, 195196, 198199, 209, 219 Congregation of the Propaganda Fide,
n. 19, 221, 264, 273, 493, 595 nn. 211 n. 2
34, 596, 596 nn. 56, 597, 598, Land, 211 n. 2
598 n. 10, 604605, 662, 668, hominisation, 522
672673 homme(s), 501502, 502 n. 4, 503504,
Biblical, 595596, 599, 599 n. 13, 506, 512, 520521
602603 homo
classical, 198 loquens, 502, 522
dialectical, 198 sapiens, 521
grammar(s), 231, 231 n. 41 homonymy, 255 n. 1, 255, 257259,
grammatical tradition, 209 261262, 269, 272, 278, 286287,
Israeli, 596 n. 6 367 n. 3
Middle, 600, 602 homophonie, 515 n. 32
posodical-grammatical terms, 219 hongrois, 119
n. 19 Hoogland, Jan, 391 n. 71, 401, 455, 473,
pre-, 599 n. 14 532 n. 5, 542
proficiency, 662 Hopkins, Simon, 77, 111
Proto-, 596 n. 6, 598599 Hopper, Paul J., 692 n. 23, 698
Qumranic, 600 host, 297
terminology, 231 matrix, 291, 291 n. 1, 292, 295,
Hedjaz, 120, 120 n. 4 307309
hedjazismes, 125 minority, 297
Heine, Bernd, 607, 621, 676 n. 4, 688 superimposed, 291292, 308309
n. 19, 692, 698 Hd, 195
Heinrichs, Wolfhart, 595, 602 humain, 345346, 351 n. 32,
Hell, Joseph, 208 non-, 351 n. 32
hemistich(es), 548 n. 5 human, 370, 383, 389, 392, 394397, 397
Hernndez Terrs, M., 243 nn. 8687
Herrero Muoz-Cobo, B., 296, 310 humanist learning, 219
Hetzron, Robert, 672 Hurst, Nicholas, 305306
heuristics, 255, 286, 368 h urf, 38, 228, 237
high vowel, 572, 574 al-h alq, 318
H ijz, 16 jarr, 228
h ikya, 1315, 15 nn. 1112, 1617, Hurwitz, S., 262, 271 n. 32, 275, 288
1921, 22, 58 H uwt t, 561562
particles, 15, 15 n. 11, 2022 H uwt , 562
particles, non-, 15, 19, 2022 hybrid forms, 224 n. 31, 655 n. 1
h ikma, 910 hybridity, 403, 405407, 408, n. 7,
H imyar 409410, 413, 413, n. 10, 417418,
kings of, 203 421, 421, n. 14, 422, 432, 433434
Hindw, H asan, 187 hypallage, 512 n. 27, 517 n. 42
Hinds and Badawi, 596 hypercorrection, 80 n. 5, 596 n. 6
Hinds, Martin, 542, 604, 683, 691 n. 21, hyperonym(s), 465466, 469470
696, 696 n. 30, 698 translations, 469
Hindustn, 209
Hinnenkamp, Volker, 408, 433434 iarab, 218, 218 n. 17
hispanicised forms, 229230 Ibdu, 686
historians, 203 ibil, 47
724 index
Ibn Abbs, 193, 200 Icelandic, Old, 213

Ibn Ab d-Dam a-fi, 87, 111 Id Abuw Silm (al-Atra at-Turbniy),
Ibn Ab r-Rab, 150 n. 2, 153 nn. 565
56, 156, 168, 170171, 173174, id, 14 n. 10,
186 idea, 396
Ibn jurrm, 232, 232 n. 43, 233, 236, ideal, 396
238240 identit, 503504
Ibn al-Anbr, 124, 151, 152 n. 3, 153, identity(ies), 405406, 408, 423424,
160 n. 15, 162 n. 19, 163164, 175, 428, 430, 433
176 n. 34, 186187, 513 n. 29, 522 cultural, 403
Ibn Aql, 17 n. 13, 22, 155, 187 personal, 430
Ibn Askir, 194, 198 n. 21 ideophones, 696
Ibn Durayd, 203 idm, 127
Ibn Fris, 125, 131, 435, 505 n. 12, 507 idiomatic expressions, 611, 618, 696
n. 19, 513 n. 28, 522 idm, 1617
Ibn H anbal, 8687, 111 idmr, 45, 5161, 63
Ibn H azm, 201, 201 n. 32, 208 Idrs, 195
Ibn Him al-Ansr, 157, 175177, 177 Idriss, Socheil, 473
n. 35, 178187 Igla, B., 294, 310
Ibn Hishm, 203 ignoratum, 235 n. 51
Ibn Jinn, 89, 9 n. 7, 1012, 22, 164, ijtim al-kalm, 56
164 n. 22, 169, 169 n. 27, 170, 170 ilBawti, 684
n. 29, 187, 193, 317, 318 nn. 7-8 il, 141, 141 n. 32, 142, 142 n. 37, 143,
Ibn Kathr, 205 n. 40 145146, 147, 160
Ibn MaJ al-Qurt ub, 11, 22 ilhm, 192, 192 n. 3
Ibn Man'r, 115, 131 illa, 114, 317
Ibn Masd, 49 n. 6, 56, 56 n. 12, 61 ill, 7, 10 n. 8, 12, 15, 17, 1921
Ibn Mujhid, Ab Bakr, 640 n. 1, 652 illi, 7276
Ibn Mutarrif at-Taraf, 196 n. 13, 208 Ilys, Jsf, 326 n. 16, 331, 345, 364
Ibn Mut, 35, 43 iml, 63 n. 16, 141, 141 n. 32
Ibn Qutayba, 197 n. 19, 203 imla, 650 n. 12, 658
Ibn Sad, 196 n. 13, 198 nn. 2021, 23, Imlq, 203
199, 199 n. 25, 206 n. 44 imm, 16, 17
Ibn as-Sarrj, 4, 22, 152154, 162163, impratif, 343, 343 n. 25, 519
163 n. 21, 165 n. 24, 187 imperatiuo, 217
Ibn Taymiyya, 39 imperative(s), 34, 59, 217218, 295, 569,
Ibn Usfr, 154 n. 10, 163, 163 n. 20, 575, 575 n. 36, 681
164, 169, 171, 176 n. 34, 187 negative, 597
Ibn Uta, 200 imperfect(s), 70, 224 n. 31, 570, 574, 575
Ibn Walld, 3334, 43 n. 36, 576577, 581 n. 56, 586 n. 78,
Ibn Xlawayh, 318 n. 8, 353 n. 34, 355 603, 616, 640643, 643 n. 9, 645, 646,
n. 35, 362 681, 694 n. 27, 695
Ibn Xaldn, 3 n. 1, 23, 119120, 120 y-, 679
n. 6, 128129, 131, 543, 544, 563 imperfectif, 315 n. 4
Ibn Ya, 154, 154 n. 9, 157, 157 n. 12, imperfective, 296, 613
169, 172173, 178, 180, 187, 254, 316 implicatures, 27
n. 5, 316 n. 5, 319, 336 n. 22, 342, 362 implicit, 378379
Ibrhm, 190, 195200 implied, 392 n. 77
ibrn, 196 Imra al-Qays, 68, 86 n. 23
ibtid, 137, 139, 144 n. 46, 145, 149, Imru al-Qays, 324
153154, 156, 158159, 161, 162 in, 14 n. 10, 16, 21
n. 18, 167, 173 n. 32, 175176, 178, inaccompli, 315 n. 4, 332
185186 inaccuracies, 472
index 725
inceptive, 21 intention (qasd ), 9, 396397

incipient shift, 308 interaction, 405, 405, n. 3, 410, 424, 427, 432
incompatibilities interdental(s), 567568, 586 n. 70, 591
semantic, 286 n. 88, 657
incongruities interjection(s), 229, 374375, 387, 398
semantic, 286 interne, 321, 335337, 344, 346347,
incrementation, 256 349350, 352, 353 n. 33, 354360,
initial, 275, 277, 281 362
indefinite 26, 162, 248, 250251, 253, 373 interrogatif, 519
indefiniteness, 217 interrogation, 398
India, 210 interrogative(s), 1719, 161, 170, 577
Indian, 209 8a-, 156, 170, 184
indicative, 10, 228, 230 n. 39, 378 component, 170
indices sentence, 169 n. 27
contextuels, 362 intertextuality, 543
indirect speech, 676, 676 n. 3, 678, 687 intertwining vocabulary, 670
n. 16, 679, 683, 689 interview(s), 405, 405 n. 3, 409412,
indispensable 415, 423433
part, 135136, 142, 146 format, 425, 427
predicate, 135137, 40141, 141 n. 33, multi-textual, 426, 428
142143, 146147 set-up, 424, 427431
indo-europennes, 118 intransitifs, 337
Indo-Iranian, 295 intransitive 247
inductive derivation of rules, 25 intransitivit, 319, 320, 320 n. 9, 336
infinitive(s), 292293, 295296, 298, 301, invariant(s)
305306, 373, 383, 386 cognitifs, 324 n. 14
nominal notional, 255, 255 n. 5, 256257, 257
Dutch, 483 n. 10 n. 11, 259260, 260 n. 21, 261,
infixes, 436 263266, 268269, 275, 279281,
inflections 283285
verbal, 294 inversion, 151, 156
informal, 403, 405, 423, 432 irb, 3 n. 1, 8, 46, 5051, 5860, 117,
information, 403, 425 118, 121123, 123 n. 12, 124126,
Ingham, Bruce, 544, 563 156, 180, 211, 218 n. 17, 228, 230
Ing-of construction, 476477, 483 n. 10, n. 39, 232, 234235, 237
484, 492 endings, 46, 5051, 5860, 62 n. 15,
inherent, 373 64 n. 17
inherited, 373 ram b. Sm, 195, 202
inn-, 74 iranienne, 113
inna, 21, 135 Iraq, 197, 529530, 545546, 560, 560
Qsentences, 165 n. 40, 561, 563, 656
innam, 11, 1516, 19, 19 n. 17 Iraqi, 546, 561
in, 28 iref, 660
ind, 126 n. 15 Irish, Old, 213
instabilit formelle, 317 irrgularit, 507
instrument, 380, 393, 395 s, 190
insults, 387 arm a-ayh, 530
integrating terms, 661 ib, 448
integration al-Isfahn, 203
morphological, 292, 295, 297 n. 7, Ishq, 195
298300, 304, 306309 Ishmael, 198, 202, 578
morpho-phonological, 308 Ishml, 197
intensivit, 313 islh al-lafz, 170
726 index
Islam, 27, 39, 42, 92, 193, 196, 202, 205, izhr, 57
207, 404 zm mudf, 215
heilige Sprache des, 119 izzayy, 86
medieval, 42
religious language of, 104 Jabal Says, 127 n. 18
Islamic, 109 Jackendoff, Ray, 479, 498
conquest, 207 Jdis, 195, 202203
data al-Jaft, 562
historical, 42 Jhiliyya, 190
East, 109 al-Jhi', 13, 45, 122123, 131, 328
fundamentalists, 561 n. 19
law, 42 jhi'ien, 123
scholars, 41 Jahn, G., 136 n. 8, 147, 238
sciences, 28 Jir Ab _amd, 195
society, 42 jiz, 33 n. 4
Islamic times James, Gregory, 210, 243
pre-, 105, 109 Janah, A., 283, 287
history, 192 Japanese, 210
past, 207 scholars, 210
times, 189, 206 Jarajira, 567 n. 9
world civilization, 207 jarr, 29, 216, 225, 227, 228, 230
ism, 58, 155 Jsim, 203
al-fil, 341 Jastrow, Otto, 658, 671672, 698
fil, 177 jawb, 53
kna, 179 al-Jawhar, 3
al-maf l, 341 jaz, 59
wh id, 14, 18 al-Jazra, 194
Isml b. Ibrhm, 190, 195, 197200, 202 jazm, 59, 226228, 230231, 234
ismiyya, 149150, 157, 158 n. 13, 165, Jeremiah, 595 n. 2
165 n. 24, 177 Jerusalem, 8182, 84, 221
isoglosses, 608 Jerusalemites, 80 n. 6
Isolated Natural Subject, 28 n. 1 Jesuit, 209210
Israel, 546, 653660, 662, 667669, 673 Jesus, 190
sons of, 195 Jewish, 82 n. 14, 101, 106, 109, 296
Israeli(s), 561, 595 n. 3 Aramaic, 600601
Hebrew, 596 n. 6 Jews, 195
Israelite settlers, 197 jezme, 215 n. 11
istaqarra, 159160, 162164, 167, 180, Jibrl, 200
183185 Jhn Mahmd, 528
hypothesis, 162, 164, 167, 180, 185 jinn, 190
istidll, 34 Jordan, 543, 543 n. 1, 545548, 551,
istifhm, 4, 117 561562, 601, 661
isti lh , 192 Jordanian(s), 546, 548, 552, 554, 557,
istinf, 49 n. 5, 50 557 n. 35, 559, 560 n. 40, 560562
istiqrr, 141, 163 n. 20, 179180 society, 547
istish b al-h l, 36 tribe, 546
istit n, 31 Jos de Len, 221
iswad, 659, 664 Joseph, 190
Italian, 296, 298299, 661 Joon, Paul, 264, 288, 314, 320 n. 9, 321,
Italie, 124 n. 13 321 n. 10, 335 n. 20, 344 n. 26, 354,
itb, 48, 52, 63 364
itration, 313 Juba, 607 612, 614, 618, 620622,
itimd, 152, 152 n. 4, 153, 153 n. 5, 154, 637638
157 n. 12, 172, 179180, 184 Arabic, 607-621, 638, 692693, 698
index 727
courts, 607609, 620, 693 Kaufman, Terrence, 294, 311, 665

Rejaf, 609 Kautzsch, 320 n. 9
Judaeo-Arabic, 7980 Kaye, Alan S., 595, 597, 605, 654, 657,
Jdisch, 105 664, 665, 655 n. 1, 672
jumla, jumal, 113, 177 kayfa, 20
fi liyya, 149150, 157, 165, 165 n. 25, Kayne, Richard, 494, 499
166167, 177 n. 35, 178186 Kzim Bahr al-Marjn, 187
ismiyya, 149150, 157, 164167, 169, Kazimirski, A. de Biberstein, 115, 132,
177186 257, 277, 284, 288
inverted, 150, 150 n. 2, 162, 182 Kelly, James, 437, 453
arziyya, 177 n. 35 Kembera (Austronesian), 698
zarfiyya, 158, 167168, 179181, Kennedy, J.H., 305, 311
183185 Kenstowicz, M., 256 n. 9, 257 nn. 1213,
wasf, 158 n. 13 288
juncture 249 Kenya, 608, 621
Jurhum, 195, 197200, 202 Kerkhof, Maxim, 243
jurisprudence, 42 kesra, 224226, 231
al-Jurjn, 8, 1012, 2223, 154 n. 7, 163, kesre, 233
164, 174, 187, 232, 235, 238240 Khadduri, Majid, 26, 44
jussive, 10, 16, 56, 228, 230 n. 39, Khalidi, Tarif, 202 nn. 3334, 208
Juynboll, W.M.C., 243 Kharbush, A., 661, 672
Kharga, 536
ka, 78 Khartoum, 608, 620
kaanna, 16 Khatef, L., 272, 288
kaannam, 16, 19 Khim, Alain, 401
kaayyin, 16 Khubchandani, L.M., 308, 311
Kaba, 196, 197, 206 Kifya movement, 529
kabs(a), 668 Kihm, 368 n. 4
kad, 1617, 31, 3435, 35 n. 7 Killean, Cavolin G., 210 n. 2, 243
Kahle, Paul, 125, 132 Kinna, 116
Kairenisch, 72, 74, 699 kinya, 34
Kakwa, 609 Kinberg, Naphtali, 50 n. 7, 54 n. 11, 65,
kalm, 27, 47, 4950, 5255, 5759, 384 n. 55, 401
177 Ki-Nubi, 607608, 621, 692
al-arab, 49 Kirsten, Peter, 232, 234, 238
al-arab/badw, 61 al-Kis, 52 n. 9, 54
Kalb, 202 Kissling, Hans-Joachim, 679 n. 7, 692
kalima wh ida, 14, 1617 n. 23, 698
ka-llad, 7778 Kister, M.J., 205 n. 40
kallad, 78 kitb, 141, 316, 318, 320 n. 9, 329, 338,
kam al-xabariyya phrase, 171 338 n. 23, 353 n. 34, 504, 504 nn. 8,
kam, 16, 78 10, 507 n. 18
Kamel Al-Kabir French-Arabic, 458, 462, Klamer, Marian A.F., 698
466, 468 Kleiber, Georges, 506 n. 16, 522
kna Koehler & Baumgartner, 595 n. 2, 596,
an-nqisa, 135 597 nn. 89, 598, 598 n. 10, 599, 599
at-tmma, 178 n. 36 n. 13, 600602, 602 n. 16, 603605
Kaplony, Andreas, 113 n. 1 Koehler, Ludwig, 605
al-Kar, Mahmd, 435, 437, 441447, koinsation, 126
449450, 453 Konjunktion, 69, 75, 87 n. 24
kasr, 227 Koplewitz, Emanuel, 662, 672
kasra, 218, 218 n. 16, 239 Koptisch, 697
kasratn, 218 Koran, 105, 223 n. 28, 410, n. 8, 411
Kator, 609, 610611, 617620, 634 530531
728 index
Koranic reading, 640 n. 1 community, 291292, 308

Koster, Cornelis, 369 nn. 8, 10, 370, 389 contact(s), 291, 293298, 306309,
n. 69, 401 653, 670
Kouloughli, D.E., 23, 161, 164, 168, 175, dominant, 610, 662
185 n. 38, 187 embedded, 291, 291 n. 1, 295, 297,
Kremers, Joost, 475, 475 n. 1, 484485, 307308
485 n. 14, 493, 499 exposure, 309
Krio Creole English, 692693 faculty, 479
Kroeger, Paul R., 693, 693 n. 24, 695, 698 foreign, 308
Kropp, Manfred, 202 n. 35, 204 nn. institutes, 661
3839, 208 interference, 653, 670
Kfa, 117 matrix, 292, 307
Kfan(s), 35, 6364, 151, 154 n. 8, 157, natural, 367
160, 160 n. 15, 162 n. 19, 175176, 181 shift , 307
Kundlzza Rys, 543 n. 1, 563 socially, 291292, 307
Kurpershoek, P. Marcel, 544, 560 nn. specific pharyngeals, 657
3839, 563 structure, 398
Kuwaiti Arabic, 563 superimposed, 295
usage, 27
l, 14 metaphorical, 369
LA, 659 n. 4, 665 n. 7, 666 n. 9 ordinary, 369
laalla, 16 varieties, 408, 410, 418, 420, 433
laallam, 1617 langue(s), 502, 502 n. 4, 503, 505 n. 12,
labels, 367, 373 n. 20 506, 506 n. 16, 508, 511512,
labial(s), 255256, 256 n. 6, 257, 259, 517521, 521 n. 49, 522
264265, 267268, 270, 279282, arabe, 124, 121, 316, 317, 319, 349,
284286 355, 501, 505, 508 n. 21, 509-511,
Labd, 18, 86, 87 n. 23, 111 515 nn. 31, 32, 519-521
Labov, William, 129, 405, 410, 410, n. 9 commune, 521
ladu, 546 historique(s), 515 n. 31, 519 n. 46
lafz, 151, 169 n. 26, 173, 176 humaines, 521
structure, 151, 171 littraires, 121 n. 8
lafz, 10, 10 n. 8, 11 maternelle, 128
lafziyya, 235, 235 n. 49 naturelles, 521
Lgiy clan, 565 n. 3 ngro-africaines, 503 n. 5
law, 14, 14 n. 10, 1516, 136, 138, 146, 160 smitiques, 506, 508, 508 n. 20, 511, 514
lah n, 117118, 123, 206 vulgaire, 121, 121 n. 8, 124 n. 13
lam, 20 Larcher, Pierre, 28, 44, 113, 120 n. 5, n.
lm, 21 6, 121 n. 9, 126 nn. 14, 17, 128 n. 21,
al-lm al-friqa, 21 130, 132, 314, 316 n. 5, 321 n. 10, 335,
lamh al-asl, 17, 17 n. 13, 20 341 n. 24, 364
lamm, 16 laryngeal(s), 598, 642, 644, 646, 646
Lammens, 200 n. 28 n. 10, 647651
lan, 15 n. 12, 123 n. 12, 124 final, 639, 641645, 647650,
Landberg, Carlo le Comte de, 544, 563, 650 n. 12, 651, 658
690, 697 n. 31, 698 lta, 20
Landsberger, Benno, 601 lateral, 257, 257 n. 13, 268270,
Lane, E.W., 147, 581 n. 57, 600, 605 275276, 278, 280282, 286, 657
langage naturel, 521 Latin(s), 115, 119, 121, 121 n. 8, 124,
Langhade, Jacques, 120 n. 5, 132 211 n. 2, 211, 213217, 221, 229233,
Landsberger, Benno, 601 235236, 239240, 661
language(s) classiques, 124
academies, 661 letters, 220
as behavior, 4 sources, 211 n. 2
body-, 29 terminology, 211, 218, 236237
index 729
latine, 124, 502 domains, 662

no-, 119 cultural areas, 662
latinized form, 231, 233, 236 n. 53 politics, 662
Lavie, Smadar, 578 religion, 662
Lavorel, Pierre-Marie, 506 n. 15, 522 figement, 362
law, 2527, 33, 3840, 42, 209, 213, 243 forms, 641
Islamic, 25 items, 653, 655, 661, 667, 670
religious, 205 literary Arabic pairs, 656
Law, Vivian, 213 meaning, 3839, 675
lawl, 16 pairs, 656
lawm, 16 lexicographer(s), 207
laysa, 20 lexicographical, 39
layta, 20 work, 211, 219
Leeman-Bouix, Danielle, 332, 364 lexicography, 435, 448, 450, 471
Left-to-Right Association, 489 Arabic, 455, 471
legal, 2526, 33, 33 n. 4, 40, 43 lexicon, 255, 273274, 287, 370, 393,
argumentation, 27 395, 406413, 415416, 418, 418
contexts n. 13, 420422, 449, 608
medieval, 42 Arabic, 255, 286, 399
definitions, 37 semantics, 653
implications, 25 verbal
methodology, 25 lexique arabe, 326
obligation, 32 li + a relative clause, 140
principle, 37 li + genitive, 140
sciences, 35, 39, 41 li-ann, 76
semantics, 38 Libya, 656
system, 25, 43 Libyan
terminology, 31 Eastern, 640
theory, 25 Lichtenstdter, Ilse, 239, 243
valid speech, 37 lien, 506
works, 33, 41 gnral, 515
legalistic, 38, 42 limite, 504 n. 8
Lgt, 565566, 566 nn. 7, 9, 567 n. 13, linear
574575, 578 descriptive text analysis, 63
Lgiy, 565 n. 3, 566 nn. 78, 567, 570, interpretation, 63
573, 575578 linearization, 494495
dialect, 567568, 568 n. 14, 569, Lingala, 609
571574, 574 n. 32 lingua franca, 661
Lehmann, Christian, 87 n. 24, 111 linguist(s), 150, 609
lengthening, 446449 Egyptian, 529
consonants, 447 linguiste(s), 115, 118, 130
prosodic, 446, 448449 linguistic(s), 2530, 33 nn. 4, 6, 36,
Lentin, Jerome, 211, 243 4042, 4446, 52, 68, 190, 192, 295,
LeTourneau, Mark, 393, 401 307308, 369, 403404, 408, 410411,
Levant, 547 420, 422, 424, 433, 527528, 533, 543,
Levin, Aryeh, 135, 137 nn. 1314, 138 559, 565 n. 2, 595 n. 1, 596 n. 6, 605,
n. 19, 140 n. 30, 141 n. 33, 143 n. 41, 607, 609, 619621
147149, 153, 153 n. 6, 158, 162, 163 Arabic, 250, 435, 437, 441, 448449
n. 20, 187, 391, 391 n. 73, 395, 399, analysis, 34, 10, 12, 20, 22, 64 n. 17
401, 672 aspects, 26
lexemes, 568, 586 n. 70, 614, 646, awareness, 5, 89
653655, 655 n. 2, 657, 659 n. 4, 660, codes, 25
661, 663, 665 n. 8, 666, 670, 672 community(ies), 193, 201
lexical, 38, 56, 611612, 616, 619 corpus, 25
aspect, 654655, 676 developments, 541
730 index
exegesis, 63 verbs, 299, 303, 652

extra-, 33, 36 word(s), 298 n. 8, 299 n. 9, 301,
forms, 189 307308, 472, 540, 602, 655
history, 191, 195, 200, 203 English, 299300, 304, 307
history of, 209 n. 1, 210 n. 2, 211 local, 404405, 417, 433
ideas dialects, 659
history of, 4 localisation, 505
methods, 45 locus, 417
panorama, 201 London, 529530
phenomena, 4, 910 long
process, 25 monophthongs, 657
sciences, 35 vowel phonemes, 568
situation, 527 Lontos, S.S., 301, 303, 311
systems, 4 Lord, Carol, 679, 693, 698, 698 n. 7
theory, 213 Loucel, Henri, 39, 44, 208
typology, 80 Lowenstamm, J., 256, 288
works, 222 Luffin, Xavier, 607, 621
linguistique, 113, 118, 133, 502 lua(t)
arabe, 121 n. 9, 129, 317, 317 n. 7, 8mmiyya, 121 n. 7, 122
323 8drija, 121 n. 7
lhistoire de la, 113 8fush , 120, 120 n. 3, 121 n. 7, 125
mdivale 8Mudar, 120
arabe, 314315, 318320, 329 QQuray, 120, 125
historique, 118119, 121, 123, 130 Lt , 195
no-hellnique, 121 n. 9 Luxor, West Bank, n. 6, 678
Lipinski, E., 273 n. 34, 288 Lyons, John, 321, 324 nn. 1314, 326
liquids, 262 n. 17, 364
lisn, 196 n. 15, 257, 288
mm, 119, 121 m, 14 n. 10, 1819, 21, 79
mm, 121 interrogative, 13 n. 9
asl, 120 ma-...-, 688
mudar, 120 Maan, 562
arab, 190 Maadd, 201
Literary ma, 667
Arab, 658 Maa, Kurt-Jrgen, 243
Arabic, 653655, 655 n. 2, 656670, Mceiaru, Adrian, 595 n. 1
672673 Machrek, 128
articulation errors, 659 marf, 235 n. 51
and Colloquial, 653657, 659, md, 1718, 18 n. 14, 19, 22
662663, 665, 668670 maddo, 226
diphthongs, 657 madh , 555, 558, 561
lexemes, 655, 668 md, 59, 315, 315 n. 4, 316317, 323,
lexical items, 656 331332, 334, 339
pronunciation, 658 Madina, 601, 605
text, 659 Maeda, S., 266 n. 28, 289
vocabulary, 661, 666 al-Mafraq, 546
written and spoken, 654 maf l, 151, 179, 214, 217
language, 80, 85 maful, 214
literature, 403404 maful, 216
Arabic, 403, n. 1 Maghreb, 101, 128, 223
Littmann, Enno, 126 n. 16, 132 Maghrebine, 101
loan Maghrebis, 80 n. 6
translation, 655 mah all, 169 n. 26, 249
verbal, 302 al-waqf, 249
index 731
Mahdi, 120 n. 5, 131 rule, 479, 497 n. 29, 481 n. 6, 482, 492
al-Mahdiyyah, 81, 83, 110 idiosyncratic, 481, 497, 497 n. 30,
Mahmd Bayram at-Tnis, 544 498
Mahmd Taymr, 666, 699 Maqil b. Xuwaylid, 89
Mahmud, Ushari, 607608, 621 al-maqsr 252253
Maiduguri, 640, 642, 647648, 652 Marais, William, 121 n. 9, 132, 690,
Maingueneau, Dominique, 324 n. 14, 698
326 n. 17, 331, 364 Mardin, 640
majz, 507 n. 18 marf, 59
majrr, 177, 214, 216 marker
Makkah, 200 adjective, 249
Makram & Umar, 46 n. 2, 49 n. 6, 52 n. 9 adverbial, 249
Makram, Abd al-l Slim, 65 case, 411412, 476
Malak al-Araf, 221 n. 26 idiosyncratic, 490
malaka, 3 n. 1 infinitive, 298, 299 n. 9
malapropism, 613, 615, 617 mood, 409
Malbad, 566 n. 9 non-finite, 488
Malekiyya, 620 object 249
Malkiel, Y., 294, 311 present tense, 409
Malta, 698 relative clause, 406407, 419, 412
Maltese, 296, 299, 650, 650 n. 12, 651652 stem
Maluk, 612, 622625 verb, 496
Mamluk, 89 stem VII, 488489
maml, 124, 149, 151, 168, 170, 235 verb
man, 38, 151, 157, 176 foreign, 299
l-fi l, 176 marking
l-hadt , 38 nominal, 476
l-hudt , 342 Maronite, 211, 220, 222 n. 26
l-kalm, 38 Christians, 211
n-nasb, 38 martaba structures, 171
t -t ubt, 342 Martelotti, Francesco, 226
manaw(iyya), 10, 10 n. 8, 11, 235, Martelottus, Franciscus, 226 n. 33, 234,
235 n. 49 236, 238241
mand, 1819, 22 Maruyama, Toru, 210, 243
Mandarin, 223 Marwn Qadr, 535
Manda, 680, 684, 693 n. 24, 697 Marx, Emanuel, 578
Al Manhal French-Arabic, 458, 462, Masd, 566567, 569 n. 9
466468 masculin, 123 n. 12
maniabilit, 508 masdar(s), 47, 118, 210, 355, 381 n. 50,
manire, 504 n. 8 383, 436, 475, 484487, 487 n. 18,
manner, 392, 394395, 397 488494, 494 n. 27, 495497, 497
al-manqs, 252253, 449 n. 28, 498, 513514, 532 n. 7
mana, 507 n. 18 +acc, 484485, 493
Mansouri, W., 256 n. 6, 288 Arabic, 475, 484, 487488, 498
mansb, 124 formation, 475, 486, 492, 495496
Manwel Mifsud, 698 Arabic, 493
mapping, 482 idiosyncratic, 492
idiosyncratic, 481, 481 n. 6 +li, 484, 485 n. 14, 492493
one-to-one, 481 n. 6, 497 n. 30 mmy, 492
phonological, 498 stem II, 490, 492
principle, 479, 479 n. 4, 480, 484, 493, stem III, 492, 497
495 stem III, 497 n. 30
general, 480 stem IX, 489 n. 21
lexical, 480 stem V, 490 n. 22
732 index
mashl, 235 n. 51 Mecque, 117, 125

Masoretic times, 219 n. 19 mdde, 215 n. 11
mass media, 669 media, 403404, 404, n. 2, 405, 432
al-Masd, 196, 196 n. 16, 202 broadcasting, 403404, 404, n. 2, 405
matching, 377, 380, 390, 390 n. 70, 393, written, 404, 405 n. 2
396397 Medina, 202
agent, 395 mdio-passif, 335
feature, 393 Mehri, 604 n. 17
gender, 397 nn. 8687 Meijer, Hans, 370, 401
number, 397 n. 86 Meiseles, Gustav, 654, 656, 664665,
matres lectionis, 540 672
matrix(ces), 255256, 256 n. 6, 257, 257 Meissner, Bruno, 544, 563
nn. 11, 14, 258260, 260 nn. 21, 22, Mejdell, Gunvor, 434, 654, 655 n. 1, 672
261266, 266 n. 28, 267269, 269 mesolectal, 608, 610614, 616617,
n. 30, 270271, 273275, 277280, 619620
280 n. 41, 281285, 287, 291 Mesopotamian, 658
binary, 273 Messiah, 190
language, 670 metalanguage, 217 n. 15, 227
segment, 262 non-Western, 231
source, 263 Spanish, 240
Mauritania, 223 meteg, 219 n. 19
mawdi, 40 metheg, 219 n. 19
mawd methodological armory, 33
al-lua, 507 n. 18 mthodologies, 501503, 504 n. 8
Al Mawrid English-Arabic, 458459, 468 linguistique, 501
Mawsat al-hadt , 111 methodology, 236
mawsf, 151, 153 Michaelis, Johann Davis, 123
Mawwls, 694 Middle
maxim Ages, 221 n. 25
of manner, 28 Arabic, 655, 655 n. 1, 662, 665, 669
of quality, 27 Hebrew, 600, 602
of quantity, 27 mifal, 249
of relation, 27 mif al, 30, 31 n. 3
maxrr, 214 Mifsud, Manwel, 296, 311, 650 n. 12,
maxrr, 215 652, 672, 698
al-Maydn, 502 n. 3, 523 milehal, 219 n. 19
mazd, 313 Miller, Catherine, 128, n. 19, 607608,
Mzin al-Mubrak, 187 619, 621, 692693, 698
Mzin Mubrak, 113, 131 minb, 218
Mazraani, Nathalie, 655657, 659, 666, minenbn, 218
672 minib, 218
McAuliffe, Jane, 208 minib, 218
McCarthy, John, 259 n. 19, 288, 421 n. minib, 218
15, 476, 486, 486 n. 17, 488, 490, 492, mininbn, 218
494495, 497 n. 29, 498499, 671 minunbn, 218
McKenny, J., 304, 311 mirbad, 30
Mdaxxal Slmn, 565 n. 3 missionary(ies), 210, 211 n. 2, 212, 219
meaning(s) n. 19, 231
general, 466, 469 linguistics, 209 n. 1, 219, 219 n. 20,
specific, 465467, 469 240, 244
Mecca, 197, 202203, 206, 561 sources, 223
Meccan period Spanish, 211 n. 2, 219
late, 190 Mitchell, Terence F., 248, 254, 664665,
middle, 190 672
index 733
mix(ed) Mohamed Sami, 536

form(s), 611615, 617 Mokilko (East Chadic), 604 n. 17
language, 655 Mol, Mark van, 458, 462, 466, 468, 473
utterance, 658 Mombasa, 621
m-masdar, 492, 492 n. 25, 497 monocausal, 6970
mnemonic, 217218 monolingualism, 293, 297, 304
terms, 239 monophthongs and , 569
words, 220 monotheist, 197
Moabite, 602 monotheistic religions, 196
mobility Monroe, James T., 210 n. 2, 220222, 244
social Monteil, Vincent, 654, 672
upward, 291 mood, 248, 251252, 372, 384 n. 55
modalit, 516517, 517 n. 41, 519 n. 45 aspect, 371
danimit, 517 imperative, 378
ditration, 508 n. 21 perfect, 371 n. 17
de temps, 517 non-, 371 n. 17
de vie, 517 subjunctive, 386, 388
mode(s), 510 mora, 486 n. 16, 489, 491, 491 n. 23
of action, 263 Moravcsik, Julius, 396, 402
irrel, 519 Mrner, A.R. & Magnus Mrner, 222,
potentiel, 519 244
prepausal, 247, 252 Moro, 609
rel, 505 n. 13, 507, 507 n. 18, 519 Moroccan Arabic, 596, 605
model Morocco, 232, 690
descriptive, 368 morpheme(s) 491, 496
prosodic, 476 autosegmental
modle italien, 121 content, 421
modern fem., 570, 574
literary Arabic, 654, 667, 658 grammatical, 421; 496
Moroccan, 596, 605 masdar, 492 n. 24, 495, 498
proto-colloquial, 597 morphological, 435, 435 n. 2,
scholarly terminology, 665 436442, 442 n. 6, 443450, 452
Standard Arabic, 527, 545, 568569, nominal, 438
598, 598 n. 12, 654, 669, 671 nominalizing, 493
standard language, 560, 598 n. 12 non-finite, 490
technology, 661 non-finiteness, 489
version of Middle Arabic, 665 paradigmatic, 436
written Arabic, 605 prosodic, 486
modification root, 489, 495
smantique, 334 radical, 435
modifier(s), 367 n. 1, 368 n. 4, 376 n. 34, verbal, 438
adverbial, 378379 vowel, 438, 439
aspectual, 378 morphmes
modal, 378 dfinisseurs, 519
peripheral, 378379 spcificateurs, 519
temporal, 378 morphologic, 9
modus, 505 n. 11, 512, 512 n. 27, 513, morphological, 63, 87 n. 24, 91, 568
515, 517 nn. 4243, 44, 519, 519 n. n. 15, 570, 592 n. 91, 608, 611612,
46, 521 n. 49, 522 615616, 642643, 645, 648650
communs, 513 differences, 659
dexclamation, 516 reduction, 687
dassertion, 515 morphologie, 316
gnral, 515 arabe, 508 n. 21, 509 n. 22
infinitif, 514 mdivale, 344
734 index
morphology, 256, 292, 296297, 299, muh l, 29

307309, 368, 370, 370 n. 12, Muhammad al-Fanti l al-H ajy, 560
373375, 384, 388390, 396, 398, Muhammad b. Al, 198
406407, 412, 418420, 431, 436, 443, Muhammad Salm, 199
449, 475, 485, 573, 653654, 659 Muhammad Taymr, 700
Arabic, 475476, 486, 494 Muhammad, 36, 39, 86, 106, 189, 191,
concatenative, 498 193, 195, 198, 200, 202206
non-concatenative, 475, 498 muxtab, 4, 36
nominal, 573 Muhazzim Qaht n, 203
verbal, 574, 651 Muhtaseb, 396
morphophonological, 648649, 653 al-Mujam al-wast, 331, 340
morphosyntax, 212, 479 n. 4, 480 mujarrad, 313, 314 n. 3
Moscati, Sabatino, 315 n. 4, 316, 364, al-Muji, 162 n. 19, 164, 164 n. 23, 187
597, 605 mulan, 136, 138, 146
Moses, 190 multicausal, 70
mother tongue, 201 multilingual, 609
motion, 396, 397 Muluk, 616, 632
Mounin, Georges, 506 n. 16, 523 mund, 214, 217
Mount Jd, 195 al-mund, 221
Moutaouakil, Ahmad, 28, 44 munde, 214
mouvement smantique, 116 Munra, 683
Moyen Age, 121 n. 7 al-Munjid, 331, 340, 347
MSA, 527, 529, 529 n. 1, 531533, 535, muqaddar, 163
537538, 540, 586 n. 70 murd, 37
Muallaqah, 86, 8687 n. 23 Murray, George W., 565 n. 3, 567 n. 9,
(al-)Mubrak, Mzin, 113, 131, 187 567 n. 13, 575 n. 37, 578, 589
al-Mubarrad, 4, 23, 33, 175, 176 n. 34 Ms, 190, 548, 551552
al-Mubarrid, 329, 335, 343, 362 Musaylima, 204
mubtada, 48, 137, 137 n. 13, 138139, mush af, 36
144145, 149, 151153, 154 n. 9, Musharrafa, Musta fa, 434
55157, 157 n. 12, 158, 158 n. 13, Musil, Alois, 544, 549 n. 9, 563, 690, 698
159164, 168170, 170 nn. 2829, Muslim, 25, 29, 37, 42, 79, 92, 104, 189,
171173, 173 n. 32, 174176, 191192, 193 n. 4, 196, 530
178179, 181186, 214, 216 Arab
clause, 172 tradition, 199
definite sentence-initial, 174 scholarship, 192
indefinite, 169, 169 nn. 2627, 170, Baghdadi speakers, 669
170 n. 29, 172, 174 dogma and consensus, 189
al-kalm, 55 faith, 207
position narratives, 197
pre-, 149 revelation, 191
post-, 174 scholars, 34
xabar, 151152, 169, 170- 173, 177 Spain, 221 n. 25
mubted, 214 speakers, 657
mubtede, 215 speech, 92
mudf, 214 musnad, 177
mudf, 214, 216 al-mutak, 221
Mudar Badrn, 554, 557 n. 35 mustaqarr, 136139, 141142, 145146
Mudar, 120, 129 ayr, 136, 146
mudri, 230, 315 n. 4, 316319, 323, 338 mustaqm, 26, 37
mudmar, 151 al-mustaqti, 221
mufrad, 164 al-mustat, 221
muayyira, 1516 musulman, 128
muh addat anhu, 167, 173, 173 n. 31, mutakallim, 4
181 al-Muta rriz, 239240
index 735
al-mutasamm, 221 naw, 7 n. 4

al-mutawallid, 221 nazm, 12
mutawwalt, 3 Nebajoth/Nabat, 201
Mutazil, 39 Nederhoff, Mark-Jan, 370, 402
mutazilites, 11, 193 Neeleman, Ad, 475476, 478479, 480
muwallad, 115 n. 5, 481, 481 n. 6, 482, 482 n. 8, 483,
muxbar, 174 483 n. 10, 484, 486, 495, 497 n. 30,
Muysken, P., 293, 295, 302, 302 n. 10, 311 498
Muzeina, 566, n. 3 ngatif, 519
Myers-Scotton, C., 291 n. 1, 311, 670, 672 negation, 14, 378 n. 42, 398, 505 n. 13,
Mznah, 566 nn. 7, 9 578 n. 42, 694
dialect, 567 n. 12 value, 378379, 381, 381 n. 51,
Mznih, 567568 382
Mzniy, 575 negative(s), 14, 21, 406407, 413, 419
participle, 154
N projection, 477 sentence, 169 n. 27, 170 n. 29
Nabat, 201 Negev, 567, 574 n. 32, 575, 575 n. 37,
Nabatean(s), 197, 200201 577578
inscriptions, 197 dialect, 570 n. 22
kingdom, 201 type of dialect, 567
language, 201 neo-
script, 201 Arabic, 75, 77, 639, 649, 651
Nabawiyya, 682 Aramaic dialects, 595 n. 3
an-Nbiah, 86, 87 n. 23 neologism(s), 465467, 469472
Nad Tmn Ab Tyih, 562 Arabic, 471
Nagba, 691 translations, 467
nahw, 12, 23, 113114, 118, 121, 123 Nepos, Ferdinandus, 215, 215 n. 12,
n. 12, 133 242
nahwa, 118 van Ness, S., 308, 311
nahy, 34 Nestorian, 106
najr, 9 n. 7 nestorianische Kirche, 106
Nairobi, 620 The Netherlands, 565 n. 2
Najd, 544, 560 n. 38, 563 Netherlands-Flemish Institute, 529
Namrd, 196 neutestamentlich, 105
nqis, 178 new
narration(s), 415, 417, 428, 430432 terminology, 655, 661
an-Nas, 87, 111 Testament, 106, 189
nasal, 257, 259261 World, 212
nasb, 30, 30 n. 2, 32, 34, 41, 4749, 49 n. Newmeyer, Frederick J., 693, 695, 698
6, 5051, 5556, 58, 60, 124, 158, 160, news, 405, 412
162 n. 18, 169 n. 26, 175, 227228, Neyreneuf, Michel, 313, 364
230, 234, 238 Niederehe, Josef, 210 n. 2, 242, 244
nasbo, 225 Nigeria, 640641, 648, 652
nsib, 58, 124 Nigerian, 640
Nsf al-Yzij, 35 Arabic, 639641
Nsf, Jirjis, 326 n. 16, 331, 345, 364 Nijmegen, Arabic corpus, 532 n. 5
nassuantes, 230 Nile valley, 545 n. 2
nat, 47, 49 Nilotic
native speaker(s), 72, 94, 101, 527 Dinka, 610
of Arabic in Israel, 662 speakers, 613
Hebrew, 596 n. 6 nisba, 194
natural language niyya, 7, 7 n. 4, 151
processing, 367, 370, 389 Nizr, 201
Arabic, 367 Nizri(s), 202
Understanding System, 396 nodes, 369 n. 8
736 index
Nldeke & Schwally, 105, 107 nominativus, 221

Nldeke, Theodor, 111, 200 n. 30 non-
nom, 315 n. 4, 362, 504, 504 n. 8, 505, Bedouin Colloquial Arabic, 658
505 nn. 12, 41, 506 compound, 15, 19
abstrait, 512 n. 27 concatenative morphology, 498
commun, 508, 512 n. 27 contrastive compatibility, 667
nomad(s), 546 coreferential, 92
Syrian, 690 emphatic
nomade(s), 117, 120 articulation, 658
dOrient, 697 phonemes, 658
nomadism, 545, 563 finiteness, 488
nombre, 513 ikya, 15, 1922
nominal, 92, 109, 153, 157, 181, 183, 219 past verbal patterns, 660
complex, 476 n. 2 phonetic dictionaries, 657 n. 3
constituent, 153, 184185 restrictive, 26, 33, 103
declension, 216 clause, 91
element(s), 157, 165, 165 n. 25, 178, Nortier, Jacomine, 297 n. 5, 308, 311
181 Norwegian, 484 n. 13
ending, 228 Research Council, 209 n. 1
event, 476 nose, 257, 260261
simplex, 475 n. 1 notificacin, 226
indefinite, 172, 173 noun(s), 78, 13 n. 9, 14, 17 n. 13, 18,
non-referential, 155 41, 43, 50, 59, 6869, 84, 88, 92, 95,
paradigm, 214 102, 108, 135 n. 2, 137139, 141142,
phrase, 162 144145, 151152, 155156, 156
referential, 155 n. 11, 158 n. 13, 160, 165167,
sentence(s), 6, 18, 77, 135138, 177, 180, 213214, 218, 218 n. 17,
144146, 149, 166, 178, 216 225230, 230 n. 39, 231, 235, 235
inverted, 154 n. 8 n. 49, 239, 248252, 299, 302303,
result, 475 n. 1, 476 n. 2 305, 308, 367, 370, 370 nn. 1214,
simplex, 476 n. 2 373, 376 n. 34, 381, 383, 390,
system(s), 659 396397, 436437, 439, 441447,
type, 80, 86, 8990, 9293, 98 451452; 482 n. 8, 573
nominality, 476 abstract, 16, 20, 386, 436
nominalization, 293, 483 accusative, 8
nominalized agentive, 481, 481 n. 6, 482
clause, 171 biradical, 438
phrase, 58 clause(s), 6869, 172
nominalizer, 488492, 492 n. 24, collective, 392 n. 78
495497 colloquial Arabic, 226
idiosyncratic, 490492 common, 373, 382383
nominatif, 124, 130 corroborative, 18
nomination, 507508, 508 n. 20, 509 definite, 251
n. 22, 520, 522 definite subject, 150
des entits, 508 deverbal, 373374, 382, 484 n. 13, 488
des expriences, 508 ends, 227
smitique genitive, 7
nominatiuo, 214 indefinite, 169, 173, 175, 251, 381
nominatival individuality, 373
constituent, 156, 181, 183 multitude, 373, 381383
indefinite phrase, 173 nominatival, 149, 173, 173 nn. 3031, 181
nominative, 10, 13, 18, 19 n. 15, +noun, 165
137139, 142, 144145, 160, 178 phrase, 153, 169, 174175, 368 n. 4,
n. 37, 215, 227228, 230, 238, 374 376, 386, 693
nominativo, 227 definite, 150
index 737
nominatival, 159 predicate-, 369, 377, 380381,

plural (non-human), 155 383384, 389390, 397, 399
singular, 155 preposition-complement, 397399
proper, 213 prepositional, 379, 391
quality, 373 pronoun, 73
relation, 373 reason (lahu), 217
tetraradical, 442, 446 topic-comment, 377, 383384,
triradical, 439 389390, 397, 399
uniradical, 448 objet, 319, 327, 327 n. 18, 335, 337338,
verbal, 60, 247, 382, 436 354, 356357
vessel, 373 double, 357
vocative, 41 externe, 354355
NOW, 565 n. 2 obligatory
noyau, 517518 constituent, 160
de la phrase, 518 fronting, 169
NP, 695 occlusion, 515 n. 31
NSA, 609618, 620 official, 404, 414, 423, 427429
N-to-D movement, 494 language policy, 661
Nuayma, Mxl, 673, 666 old
Nubi, 607608, 621 Arabic, 639, 640 n. 1, 641,
Nh, 190, 192, 194196 650651
Nuijtens, E.T.G., 297 n. 6, 311 South Arabian, 597, 599603
Numn r, 699 Testament, 105, 108, 595 n. 2,
number, 368 n. 4, 370, 371374, 378, 605
382383, 389, 395 Syriac, 108
numerals, 35, 373, 619 World, 212
nunated forms, 218 Omayyade(s), 117
nunation, 225, 248, 251253 one-to-one
Nr ad-Dn, 313, 314 n. 3, 316 n. 5, 364 equivalent(s), 456, 465
Nuweiba, 565 n. 3 translation(s), 455
Nyckees, V., 255 n. 1, 288 onomatope(s), 517 n. 43, 520521
onset, 486 n. 16, 487, 491
Obicini, Thomas, 232, 232 n. 44, 233, operants, 8, 10 11
233 n. 48, 234, 238, 242 opration
object(s), 30, 37, 40 n. 10, 56, 6869, humaine, 503
7374, 9192, 109, 214, 217, 380, operator(s), 151, 160161, 175, 180181,
392395, 395 n. 85, 476, 484, 484 183, 237
n. 13, 493, 494 n. 27, 544, 611, 692 abstract, 178
accompaniment (maahu), 217 Oppenheim, Max Freiherr von, 565
accompanying, 393 n. 3, 578
clause, 6869, 102 oppositions
cognate, 393 binaires, 515 n. 34, 520
direct, 179, 217, 230 n. 39, 238, 293, optional constituent, 160
306, 309, 378 n. 40, 379380, oral
391392, 682, 697 articulation, 656
direct (bihi), 217 communication, 665
from it (minhu), 217 Order of St. Jerome, 210
fronted direct, 182 ordonner, 511
human, 393 n. 80 ordre, 501, 521 n. 49
non-, 393 n. 80 Orel and Stolbova, 604
indirect, 293, 380 Orel, Vladimir, 605
locative, 217 organisation
locative ( fhi), 217 binaire, 522
noun des langues
verbal, 293 gnrale, 501, 521522
738 index
primitive, 520 conditional, 375

primordiale, 522 coordinating, 375
Orient, 128, 527 formation, 492
Oriental languages, 221, 222 n. 27 Greek, 108
orientalising terminology, 211 n. 2 Greek attributive, 108
Orientalist(s), 220, 527 passive, 157, 488, 492, 497, 497 n. 30
orientalistes, 130 present, 482
orthographic, 595, 596 n. 5, 598 n. 11 selective, 375
sign, 217 singular, 155
orthographical, 540 singular feminine, 155
orthography, 189, 368, 448, 540, 598 particle(s), 5, 8, 13, 13 n. 9, 1415, 15
n. 11, 684 n. 13 n. 12, 17, 17 n. 13, 1822, 38, 41, 54,
Ottoman, 612 63, 165, 165 n. 24, 167, 170, 172,
Arabic, 619 177, 210, 214, 227229, 229 n. 37,
ougaritique, 316 230231, 233, 235, 235 n. 49,
Owens, Jonathan, 127, 132, 158, 165, 237239, 367, 370, 370 n. 12,
187, 188, 216217, 228, 228 n. 36, 374375, 376 n. 32, 378, 384, 388,
244, 453, 595 n. 1, 598 n. 11, 607, 397, 579 n. 49, 612, 681682
621, 639640, 642, 647, 650 n. 12, auxiliary, 292 n. 3, 384 n. 55
652653, 665, 670, 673 conditional, 375
Oxford Doniach conjunctive
English-Arabic, 458459, 466, 468, 473 conditional, 388
Oyanguren de Santa Ins, 219 n. 19, 242 connective, 388
of comparison, 78
palatalization, 611 coordinating, 374
palatalized variant, 657 cumulative, 374
Palestine, 113, 197, 201 dual-marking, 182
Palestinian exceptive, ill, 171
Ancient South, 72, 72 n. 2 interrogative, 152153, 155156, 170,
pan-Arab, 404405 179180, 184
Pap, L., 306, 311 m, 238
Papapavlou, A., 300, 303, 311 min, 135 n. 2
paradigm(s), 639, 642, 647648, 650, negative, 156
650 n. 12 relative, 7072, 76, 86, 95, 108
weak final, 642, 646, 651 selective, 374
stability, 651 type, 374
paradigmatic, 641, 649650, 650 n. 12, verbal, 614
651 vocative, 30
paragraph, 367 particularit, 346
paraphrases, 455 particulas charrantes, 230, 237
parser(s), 369370 particule(s), 504505, 505 n. 13, 506
part/whole relationship, 269 Partikel, 69
participants, 405, 425, 427, 429431 ara, 546
participe(s), 341, 341 n. 24, 342 pass
actif(s), 341, 341 n. 24, 342, 344, compos, 331, 332
347348, 348 n. 30, 349350 simple, 332
passif, 341342, 115 passif, 115, 343 n. 25
participial form, 151, 614 passive(s), 38, 63, 64 n. 17, 231, 235
participle(s), 82, 107108, 150156, n. 51, 281, 282 n. 43, 316, 323,
156 n. 11, 158 n. 13, 160, 163164, 406407, 419, 441
178179, 181, 183186, 295, 373, 498, Pastrana, Juan de, 215
588 n. 80, 602 patient, 323, 349, 393, 393 n. 80,
active, 108, 151152, 152 n. 3, 394395, 395 n. 85
153154, 154 n. 7, 157158, 167, 646 pattern, 370
index 739
Arabic, 444 philological methods, 219

biradical, 440, 447 philology (ilm al-lua), 12
heptaradical, 441 philosophers, 205
hexaradical, 441 philosophie, 502
morphological, 435437, 439450, 452 Phoenician, 597, 599602
nominal, 439 phoneme(s), 10, 256, 273, 284, 368, 370
pentaradical, 440 n. 14, 508, 510, 520, 522, 540,
tetraradical, 440, 446 567568, 576, 586 n. 70, 613, 640
triradical, 439440, 442444, 446 n. 1, 657658
nominal, 439 Arabic, 368, 371, 371 n. 16
uniradical, 440 auxiliary, 371
verbal, 436 voyelles, 510
Paul V, 221 consonnes, 510
pausal form, 30 n. 2 consonantal, 567
pause(s) 219 n. 19, 247249, 570 phonemic, 568, 576 n. 38, 598 n. 12, 641 n. 4
rules 248 phonetic(s), 9, 258, 260261, 273274,
syntaxique, 510 281, 286, 571, 650 n. 12
Payne Smith, S. and R., 107, 111112 differences of cognates, 660
PCCL, 377, 379 evolution, 273
Peled, Yishai, 149, 160, 169, 188 variants, 658
Pellat, Charles, 30 n. 3, 123 n. 11 vowel system, 658
Pellitteri, Antonino, 44 phonetical, 63
People of the Cave, 190 phonetically-conditioned urban/rural
Perez de Soto, Antonio, 222, 241 Colloquial Arabic, 658
perfect, 96, 371, 572, 577, 586 n. 78, 616, phontique, 319
640643, 645646, 680681, 694 n. 27 phonological, 602, 608, 611613, 616,
stems, 645 515 n. 34 618619, 641642, 649651
perfectif, 315 n. 4 analysis, 656
perfective, 613 categories, 655
Perfekt differences, 656567
asyndetisch, 699 features, 653, 655, 656, 668
performative, 28, 41 systems, 657
aspects, 41 phonology, 295, 297, 368, 406408, 416,
Persans, 115, 128 418420, 442443, 479480, 492496,
Persian, 193, 661 497 n. 30, 498, 567, 576, 597, 620,
person, 370374, 378, 380, 382, 389, 395396 653654, 656
verbal affixes, 611 Arabic, 490
persona(s), 423424, 428429, 431432 morpho-, 3
hybrid, 431 phonotactics, 571
private, 428, 430, 432 phrase(s)
professional, 432 adjective, 376, 376 n. 34, 377,
public, 428, 431432 379380, 382, 388, 397 n. 89, 504,
personal, 24, 417, 426432 506, 506 n. 16
personne indefinite, 379 n. 46
deuxime, 514515 adverb, 376, 376 n. 30, 377379,
premire, 509, 515 384386, 388, 390, 393, 395
Peter, William, 616 adverbial, 485 n. 14
le Petit Robert, 288 Arabic, 494
Pfaff, 295 complement, 388
Pharaoh, 105, 190 conjunctive, 376, 376 n. 30, 384, 388389
pharyngeal(s), 256, 256 n. 9, 257, 266 inflectional, 386 n. 63
n. 28, 275276, 279 n. 39, 284, 286, interjectional, 376 n. 30, 384, 387
598, 650 n. 12, 658 nominales, 341
final, 650 noun, 368 n. 4, 369, 370 n. 12, 376,
740 index
376 n. 33, 377381, 381 n. 50, 382, poetic(s), 43, 544545, 547548, 550
382 n. 53, 383385, 385 n. 57, n. 18, 559562, 578
386388, 391395, 397, 477, 485, 494 genres, 204
particle, 376, 376 n. 30, 384 poetical, 45, 561
prepositional, 369, 376, 376 n. 30, potique, 125, 126 n. 15,
377380, 382, 384387, 390395 poetry, 3 n. 1, 436, 448, 450, 543,
structure, 517 543 n. 1, 544, 545 nn. 34, 547,
non-, 517 n. 43 562563
verb, 376380, 386, 388, 393, 397 nn. Arabic popular, 543544
86, 88 Bedouin, 543 n. 1, 544, 560 nn. 38,
verbal, 386, 477, 482, 482 n. 7, 483, 40, 563
484 n. 13, 485, 493, 495, 497 n. 29 contemporary, 545
phraseology, 548 popular, 544, 559
Piamenta, Moshe, 6768, 112 tribal, 561
Pidgins, 128, 607, 621 Jahiliyya, 190
pidgin-crole, 128 love, 204
pidginisation-crolisation, 128 popular, 543544
Pidginization, 607, 621, 652, 699 praise, 204
pilgrimage, 199 pre-Islamic, 86, 559
Pisa, 84 political satire, 543, 559
place, 371, 373, 378 n. 40, 380, 394397 polysmie, 355, 359
of pause 249 polysemy, 255 n. 1, 367 n. 3,
plaidoirie, 504 n. 7 Pope John Paul II, 197
Plank, Frans, 696 n. 29, 697698 Portius, Simon, 223 n. 30
Plato, 547 Portugal, 219 n. 20
Platonic, 39 Portuguese, 210, 211 n. 2, 291, 294, 298,
plosives, 657 304307, 309
plural(s), 227, 374, 383, 397 n. 86, 436 American, 306
affixes, 611 Poss-ing, 478, 483, 483 n. 10
feminine, 224 n. 31 construction, 476, 483484,
form(s), 153, 660 495
nominal, 373 postdeterminer, 381, 381 nn. 5051,
patterns, 659 382
poem(s), 61, 204, 543 n. 1, 543, postmodifier, 381382, 384385
545548, 548 n. 5, 552, 554, 556, 559, pragmata, 33 n. 6
560 n. 40, 562 pragmatic, 2728, 29 n. 1, 33, 33 n. 6,
couplets, 548 n. 5 34, 42
hemistich, 548 n. 5 analysis, 36
meter, 548, 559 approach, 33
metrically, 548 n. 5 Gricean, 30
rhyme, 486 n. 16, 548 n. 5, 550 n. 18, modern, 41
559 Pragmatist, 40
scheme, 548 prceptes, 501
rhyming, 548 n. 5 predeterminer, 381382
syllables, 548 n. 5 prdicat(s), 339, 341
verse, 548 n. 5 verbaux, 339
posie, 123 n. 12, 125126 predicate(s), 7, 2829, 29 n. 1, 68, 74,
arabe ancienne, 323 135137, 141, 145, 149150, 155, 158
poet, 189, 543 n. 1, 544545, 545 n. 3, n. 13, 159, 162163, 163 n. 21,
547, 553 n. 27, 559560, 560 n. 38, 165166, 168, 171173, 173 n. 31,
561563 216, 230, 368 n 4, 376, 378, 389, 397
Bedouin, 544545, 545 nn. 2, 4, 560, 563 n. 86
Jordanian, 546 adjectival, 155156
pote(s), 117 adverbial, 150
modernes, 116 clauses, 68
index 741
indispensable, 138, 140141, 145146 pre-vocalic position, 647

non-verbal, 152 Prince, Alan, 421, n. 15, 476, 486, 486
prepositional, 150 n. 17, 488, 490, 495, 497 n. 29, 499
pre-subject verbal, 156 Priscian, 209
subject, 76, 369, 377, 380381, private, 403, 405, 423, 427430
383384, 389390, 397, 399 procs, 321324, 324 n. 13, 325326,
order, 162, 172 328 n. 19, 329330, 332, 335339,
predicatival 343, 347349, 351 n. 32, 354357
constituent(s), 152, 167,179180 accompli
non-verbal, 152 antrieurement, 331
prepositional phrase, 168 processing, 369, 377 n. 36
relationship, 149, 159160, 162, automated language, 367, 367 n. 2,
165166, 180 370 n. 14, 396, 398
sentences, 170 n. 29 processus, 321, 324, 324 n. 13, 325326,
predicative constituent, 149, 177, 326 n. 17, 327, 329331, 333, 340,
179181, 184 342343, 347, 349, 351352, 357, 359,
adverbial/prepositional, 185 361362
prefix(es), 15, 224 n. 31, 262 n. 25, 262, accompli, 332
264, 275, 277, 292, 407, 409, 412, 436, cognitif, 349
486, 488, 488 n. 20, 490, 490 n. 22, 492, professional, 426432
492 n. 24, 494, 597, 604, 612, 640 progressif, 341, 341 n. 24, 347
detransitivizing, 488 n. 19 projection
incrementation, 271 nominal, 485
prefixation, 17 n. 13 verbal, 485
prfixe, 509 pronominal, 7374, 97, 571 n. 24, 573
prfix, 315 n. 4, 316, 317319, 338 concatenation, 7374
prsent, 331332, 342 element, 154155
prefixed, 17 n. 13 system, 659
preformative(s), 262, 597 prononciation pausale, 126, 127 n. 18
prejunctural, 249 pronoun(s), 15 n. 12, 73, 88, 91, 106,
premodifier, 378379, 384385, 387389 151, 154 nn. 7-8, 164, 164 n. 23,
prepausal mode 247 173, 176, 180, 182, 578 n. 41, 613,
prepause 249 615616, 687 n. 16
preposicion, 215, 230 affix, 612613, 616
preposition(s), 7, 63, 79, 89, 93, 97, affixed, 615
101102, 163, 165, 214218, 229230, anticipatory, 171, 176
374375, 385386, 389, 391, 484, 497 conjunct, 217218
n. 28, 541, 576577, 579 n. 47, 581 free and bound, 659
n. 54, 598, 598 n. 10, 689690 implicit, 151
bi-, 7, 102 indefinite, 373, 381383
dummy, 476 interrogative, 373, 381383
Latin, 216 negated, 576
prepositional, 379 n. 48, 380, 384, 389 object, 615
n. 68, 391 suffix, 611
conjunctive, 389 personal, 89, 154, 154 n. 7, 164, 373,
linker, 385386 381383, 612613, 617
phrase, 5455, 152, 153 n. 5, 159160, possessive, 613, 615
162163, 163 n. 20, 164, 164 n. 23, prefixed, 379
167, 171172, 177, 179, 184 relative, 18, 67, 71, 78, 87 n. 24, 91,
definite, 150 94 n. 27, 9899, 101, 103104,
present tense, 224 n. 31 107108, 152, 576, 373
presentations, 397 resumptive, 73, 87 n. 24
prestige, 404 Spanish, 218
prestigious variety of Arabic, 664 suffixed, 239, 379, 381 n. 50
presuppositions, 27 Syriac relative, 108
742 index
pronunciation(s), 249, 253, 409410, qiri, 660

410, n. 8, 411413, 419 qisas al-anbiy, 192
proper names 250 qiys lafz, 11
properties qualit, 320, 327 n. 18, 328, 345346
phonological, 479 constante, 337
semantic, 479 constitutif, 338
syntactic, 479, 492 durable, 327328, 337
prophet(s), 26, 37, 190192, 195, 198, inhrente
200, 204206 naturelle, 320, 338
Muhammad, 189, 191, 193, 195, 198, permanente, 327328
200, 202, 204 temporaire, 327
prophetic language(s), 195, 204 transitoire, 327
prose contemporaine, 343 quantifiers, 373
prosody, 486, 490 quantitatives, 397
protasis, 377, 486 n. 16 quantity, 398
Proto- Quechua, 484 n. 13
Ethio-Semitic, 600 Queiroz, A., 304305, 311
Hebrew, 596 n. 6, 598 Quiniou, Y., 274 n. 35, 288
Semitic, 595, 596 nn. 56, 597599, Qumranic Hebrew, 600
601, 604 Qurn, 2526, 29, 3637, 4549, 5154,
proverb(s), 3 n. 1, 375, 387, 578, 692 57, 61, 64, 64 n. 17, 77, 79, 8687, 92,
proximity (itb), 48, 52 105106, 109, 189192, 196, 205207,
theory, 56 247-248, 611, 657
public, 403, 405, 415, 423424, 427432 ajam, 190
purposes (ard), 9 Arabic, 198, 205
exegesis of, 192193
Q. (Qurn), 155 language of, 189190
Qabbwa, 318, 364 Q 3/180, 36
qabh , 29 Q 67:1, 105
qad, 49, 50, 52, 57 Q 7:54, 105
qadara, 660 grammatical commentary, 45
qa/iyye, 657 vers, 193, 198
qfiya Qurnic, 4547, 4950, 5253, 5556,
muqayyada, 126 60, 8788, 105106, 108, 194
mutlaqa, 126, 126 n. 15 excerpts, 45
Qaht n, 202, 204 exegesis, 6263
al-Qahtani, 661, 673 linguistic outlook, 191
qalla, 19 message, 200
qallam, 17, 19 readings, 47
Qms, 257 revelation, 190, 205206
al-Qms al-Muht , 288 sciences, 49 n. 5
qnn, 42 text, 46, 58, 63, 155
qara, 660 traditions, extra-, 191
qaraa, 660 verses, 19 n. 17
qat, 49 n. 5, 50, 50 n. 7 Quray, 120, 125, 202, 206
Qatar, 532 Quayr, 117
qawl, 47, 5859 Qutrub, 126127
qawmiyye, 657
al-Qayrawn, 84 Rabbi David (i.e. Qimhi), 209
Qays, 201 Rabin, Chaim, 120 n. 4, 133, 640 n. 1, 652
q'ltu and g'l't dialects, 658 racine(s), 317318, 345, 507508, 508
qillat al-ijtim, 52 nn. 2021, 509, 509 nn. 2223, 510,
qira, 61 510 n. 24, 511516, 520, 521 n. 49, 522
qirt, 46, 48, 51, 652 consonnes, 513, 516
index 743
de consonnes, 508 n. 20, 509 n. 22 reading

de syllabes, 508 n. 20, 509 nn. 2223 oral 247
de voyelles, 508 n. 20 style, 411412, 414, 425
monoconsonantiques, 515, 515 n. 32, reasons (ilal), 9
516 recitation 247249
primitives, 507 Reckendorf, Hermann, 77, 112
proto-, 515516 rection, 1011, 1517, 19
triconsonantique, 516 Red Sea, 561
radical(s), 253, 256258, 262, 269270, Reese, 371 n. 17
274, 277, 277 n. 38, 282, 285, 287, reflexive, 271, 277, 441
371372, 435, 440441, 443 n. 8, 444, rflexivit, 313
446, 450 regular
augmented, 436, 440442 verb patterns, 660
base, 435 verbal sentence, 166
basic, 436, 441, 447 rgularit(s), 356, 507, 511
final 252 smantique(s), 313, 356
middle, 445446 regularities, 368, 389
non-ambiguous, 258 nn. 1617, relation(s)
primary, 441 biunivoque, 514, 517, 517 n. 44,
second, 446 518520, 520 n. 47, 521 n. 49
original, 441 causal, 276
primary, 440 univoques, 517520
three-, 435 relationships, 368369, 376, 381382
Turkic, 441 relative, 2, 6, 13, 26, 3334, 36, 3839,
radicale(s), 314315, 317, 353 n. 34, 511 4243
n. 26 clause, 6869, 7172, 7677, 87,
consonne(s), 313, 318319, 338, 344, 87 n. 24, 8889, 9193, 9798,
345 n. 28, 353 103105, 107109
deuxime, 317319, 345, 345 n. 28 asyndetic, 151152
premire, 318 causal, 104
troisime, 318 coreferential, 98
radio broadcast 251 non-restrictive, 98
Syrian 248 Relativpronomen, 85, 87n. 24
raf , 29, 33, 46, 48, 52, 56, 5860, 149, Relativsatz, 69, 75, 106
152154, 156160, 160 n. 15, 161, 162 syndetischer, 75
n. 18, 167, 170 n. 27, 171, 175176, Relativsatzgefge, 70, 74
180181, 183185, 225, 227228, 230, religious
234, 238 arengas (xutbas), 89
case, 161, 181, 183, 186 language, 104
ending, 46, 46 n. 3, 49 n. 6, 50, 58, orders, 221
158 scholars, 43
rfi, 55, 5859, 161, 167, 175 Renaissance, 209, 213
Rib, Ysuf, 93, 111 Rendsbourg, G.A., 595 nn. 12, 600601,
rajaz, 204 602 n. 16
Ramadan, 609, 616, 619 reported speech, 676, 676 n. 3, 678, 683,
ar-Ramlah, 566, 566 n. 7, 578 685, 687689
range, 393, 395 representation
Rappaport Hovav, Malka, 391 n. 73, 401 morphophonological, 479480
Rs adr, 566 morphosyntactic, 479480
Rs Sadr, 566 n. 7, 575, 575 n. 37 phonological, 479480
Rashd al-Kln, 548, 556 semantic, 479
rasm, 125 syntactic, 479
rasl, 191 res, 505 n. 11, 512513, 517, 519, 519
rteb, 667 n. 46, 520, 522
744 index
anime, 514 407, 435436, 446, 448, 450, 480,

banales, 514 486489, 491, 496, 498, 568, 580 nn.
communes, 513 50, 53, 581 n. 54, 586 n. 77, 596 nn.
dostension, 515 67, 597, 598 n. 10, 599 nn. 13,
de reprsentation, 515 600601, 604, 604 n. 17; 650 n. 12
gnrales, 515 bilateral, 486 n. 15
montres, 515 consonant(s), 489, 489 n. 21, 491
reprsentes, 515 consonantal, 370 n. 14, 395
vagues, 515 nominal, 438
restriction, 398 mono-, 596
rsultat, 330331 quadri-, 287
rsultatif, 329 tri-, 286, 287
resumptive, 8, 87 n. 24 paradigmatical, 435
retrieval quadriliteral, 486 n. 15, 497 n. 29
information, 369 n. 8, 389 n. 69 strong, 251
Reuchlin, Johannes, 209, 218, 231 n. 14, triradical, 442, 488
242 verbal, 438, 494
Reuland, Eric, 483 n. 10, 499 Rosenbaum, Gabriel M., 662, 665666,
de Reuse W., 308, 310 673
revelation(s), 190194, 198, 204205 Rosenhouse, Judith, 595 n. 1, 596 n. 6,
pre-Arabic, 198 653654, 655 n. 2, 656658, 660663,
revelationist 668, 672673
theory, 193 Rosenthal, Franz, 544, 563
tradition, 194 Roth, Heinrich, 209, 242
Rhema-Thema, 7576 Roth-Laly, Arlette, 648, 652
rhetoricians, 46 rubba, 19
rhyme, 486 n. 1, 548 n. 5 rubbam, 17
scheme(s), 436, 450 Rubin, Aaron D., 70, 112, 595 n. 1, 601
Ridruejo, Emilio, 210 n. 2 Rubin, Uri, 208
rime rules
absolue, 126 agreement, 367 n. 1
interne, 127 concord, 367 n. 1
lie, 126 ar-Rummn, 3435, 43, 504 n. 8
Ritchie, C., 671, 673 rural, 657
Ritchie and Bhatia, 653 dialects, 657
Ritter, Elizabeth, 493, 499 Russian, 294, 546, 561
Rizzi y Franceschi, Mariano, 229 n. 37 Rwala, 563, 690, 698
Ritter, Helmut, 23 ar-Rynih, 566 n. 7
Rmlt, 569 n. 17
Rodrigues, J., 304, 311 sa, 658
Rodrigues, Joo, 210, 242 Saad, George Nehmeh, 369, 371 n. 15,
Roeder, Larry, 565 n. 3 381, 393394, 394 n. 81, 396 n. 85,
Roldn, A., 243 399, 402
Roman(s), 121 n. 8, 228 n. 35, 314, 316, Saba, 199, 202
316 n. 5, 321 n. 11, 341 n. 24, 347, sabab, 114
349 n. 31, 353, 364, 501, 505 n. 13, sabat ah ruf, 64 n. 17
508 n. 21, 509 n. 22, 523 Sacy, Sylvestre de, 314, 364
Romance, 212 n. 7 Sad ad-Dn Wahba, 699700
romanes, langues, 119 sadaqa, 31
Romani, 294 Sadat, Anwar, 544
Rome, 223 n. 28, 232, 232 n. 44, 233, Saddam Hussein, 561
235, 238, 240, 243 d arf al-Jazr, 528
Romero, Juan Antonio, 223 n. 30 f, 667
root(s), 8, 9 n. 7, 13, 56, 203, 216, 231 Saf Abd al-Munim, 699
n. 41, 239, 286, 370, 370 n. 14, 371, as-Safad, 200 n. 30
index 745
a-fi, 26, 205, 205 n. 41 Schwenter, Scott, 681 n. 9, 698

al-Saffr, 369 n. 9, 402 science, 661
Saguer, Abderrahim, 255, 255 n. 2, 257 Seaman, P.D., 302304, 307, 311
n. 14, 258, 262, 262 n. 24, 288289, 400 Searle and Austin, 28
as-Shib, 125, 131, 505 n. 12, 507 nn. Seattle, 595 n. 1
1819, 513 n. 28 Second World War, 544
Sahh, 78, 90 Segeral, 256, 289
Saint-Barthlmy, 129 Sellheim, Rudolf, 239
as-Sakkk, 28 Selkirk, Elizabeth, 486 n. 16, 499
akl, 217 semantic(s), 3, 8, 1415, 22, 4849,
sakrn, 659 7071, 76, 86, 87 n. 24, 91, 9698,
sakrne, 659 101, 103104, 255 nn. 1, 5, 258, 261,
Salamanca, 221 262 n. 25, 263265, 269270,
Slih, Subh, 195, 206 n. 49 272275, 277278, 281, 284287,
salqiyya, 9 n. 7 367368, 369 nn. 78, 370, 370 n. 12,
Saloniki, 82 371, 387, 389391, 393, 397399, 421,
Sm, 196 479, 496, 656, 684687, 694695
Samnah analysis, 48, 49 n. 5
dialect(s), 567, 572 n. 25 aspects, 3, 22
Smih Farag, 699 content, 684, 694 n. 27, 695696
San Diego, 222 differences, 656, 666
San Pietro di Montorio, 211 n. 2 elements, 5
sand, 658 fields, 70, 661, 665, 667, 670
sane, sini, 658 finite, 370, 396
Sanskrit generalization, 93
grammars of, 209 level, 8
Santino, J., 618, 637 static, 389, 389 n. 69, 390
So Paulo, 219 n. 20, status of commands, 33
al-Sq, FJil, 370 n. 13, 402 semantically, 15, 54, 70, 96, 103, 103 n. 31
Sarbt al-Xdim, 565 n. 3, 566 n. 6, 578 deficient verb, 178
Sarah, 197 smantique, 314, 316, 319
arf al-Alf, 528 arabe, 324
art (cause), 54 semicolon, 219 n. 19
a-artn, Rad, 314, 318, 364 semiconsonants, 448
arw, 21 Semitic, 197, 201, 595, 597598, 598
satirical, 528 n. 10, 599, 599 n. 13, 604605
Saudi Arabia, 545, 561562, 661, 673 ancient, 597
Saudi(s), 545, 561563 East, 598
de Saussure, Ferdinand, 502, 523 languages, 108
Sawlhah, 567 n. 9 Proto-, 595, 596 nn. 56, 597599,
awq D ayf, 22 599 n. 13, 600601, 604, 657, 672
Saxon genitive, 476 Proto-Ethio-, 600
ay wh id, 16 West, 108
a-aybn, 40, 43 smitiques, langues, 125
Schabert, Peter, 100, 112 Semitist, 595
Schatz, H., 308, 311 seal, 217, 217 n. 15, 226
Schepper, Hugo de, 243 de demostracin, 217 n. 15, 226 n. 34
Schmidt, Richard, 418 n. 13, 420, 434 sens, 313, 315, 319, 322 n. 12, 324, 326,
Schnurrer, 210 n. 2, 212 nn. 35, 220, 328, 330, 333, 339344, 349, 352,
220 nn. 2122, 222223 354358, 360362
scholars of religious Law, 205 abstract, 508
Schregle, Gtz, 473 grammatical, 324, 358, 360362
Schuler, Bernard, 504 n. 7, 523 grammaticaux, 359, 362
Schulz, David, 408 n. 7, 434 lexical, 330, 334, 340
schwa(s), 219 n. 19, 296 lexicaux, 362
746 index
sense(s), 258259, 260 n. 21, 261, 227, 232, 238240, 316, 318, 320 n. 9,
263265, 267270, 272, 274279, 329, 336 n. 22, 338, 338 n. 23, 345
281282 n. 42, 283286 n. 27, 353, 353 n. 34, 362, 368 n. 6,
abstract, 275 370 n. 13, 378 n. 40, 402, 504, 504
contradictory, 272 n. 8, 505, 505 n. 12, 523
homonymic, 263, 266, 270 sibilant(s), 262, 407, 407, n. 5, 409, 413,
static, 264 418419
sentence, 367 pronunciation, 409410, 410, n. 8,
-initial position, 154, 156, 161, 170 420
n. 29, 172, 174, 185 sifa, 54, 140, 151, 157, 169, 169 n. 26,
nominal, 368 n. 4, 376, 376 n. 31, 377, 173 n. 30, 174
384, 388389, 393, 397, 399 muabbaha, 333, 341342
medieval theory of, 150, 156 signifiant(s), 507508, 508 n. 21, 510,
meaning, 28, 38 513, 520, 521 n. 49, 522
type(s), 149150, 156158, 158 n. 13, de cas, 510
162, 164168, 173 n. 31, 176177, de lanimit, 513
179181, 183185, 185 n. 38, 376, de mode, 509
398399 du temps, 513
Arabic, 157, 185 signifi(s), 509510, 520, 521 n. 49, 522
verbal, 368 n. 4, 376, 376 n. 31, sila, 59
377378, 384, 387389, 393, 397, Silesia, Dominicus Germanus of/de, 211
399 n. 2, 221, 223, 241
separator, 249 Simon, Udo, 28, 44
Serabit el-Khadim, 565 n. 3 sina, 3, 4 n. 1
Serhane, R., 257 n. 14, 288, 289 Sinai, 543 n. 1, 545, 545 n. 4, 560 n. 38,
serial verb construction(s), 675, 692, 694 563, 565, 566 n. 7, 9, 567 n. 9, 568,
n. 26, 698 572574, 574 n. 32, 575, 575 n. 34,
SVC, 692696 577578, 586 n. 70, 587, 587 n. 79
serial verbs, 692, 698699 bedouin dialect(s), 568, 571
serialization, 675 Sinaitic, 567568
Seuren, Pieter, 696, 698 Singer, Hans-Rudolf, 100, 100 n. 29, 112
Seville, 211 n. 2, 221222 single
x Ima, 680 phrase, 164
Sezgin, Fuat, 10, 23 word, 18
Shawqi Daif, 404, 404, n. 2, 405 singular, 157, 227, 380, 383, 397 n. 86,
Sheba, 603 659 n. 4, 660
Shibl, 198 n. 22 feminine, 224 n. 31
Shiites, 193, 206 form, 153
Shth, 195 as-Srf, 44, 117, 136 n. 8, 143 n. 43,
Shraybom-Shivtiel, 655, 661, 668, 673 144 n. 48, 145, 147, 164 n. 23, 336
Shuayb, 195 n. 22
Sbawayhi, 3, 3 n. 1, 4, 4 n. 2, 56, 6 a-irbn, 37, 44
n. 3, 7, 7 n. 5, 8, 10, 1215, 15 nn. lwh, 563
1112, 1718, 18 n. 14, 19, 19 n. 15, slots, 377380, 382, 385389
20, 2123, 2528, 28 n. 1, 29, 29 n. 1, Smith-Stark, Thomas, 219, 244
30, 30 n. 3, 3133, 33 nn. 45, 3443, Sobelman and Harrell, 596
45, 52, 62, 62 n. 15, 64, 64 n. 17, 129, Sobelman, Harvey, 605
135, 135 n. 2, 136138, 138 n. 19, social differentiation, 408
139, 139 n. 25, 140141, 141 n. 3233, sociality, 398
142, 142 n. 39, 143, 143 n. 43, 144 Socin, Albert, 544, 563
nn. 4951, 145147, 149, 151152, sociolinguistics, 291, 297, 307308, 652, 648
152 n. 3, 153, 156, 156 n. 11, 158159, sociolinguistique(s), 121, 123124
159 n. 14, 160162, 162 n. 18, 167, amricaine, 129
175, 178, 180, 182, 184, 187, 216217, Soden, Wolfram von, 605
index 747
Somekh, Sasson, 665, 673 n. 6, 444, 481, 486, 486 n. 17, 487,
sonorant(s), 256 n. 6, 257, 257 n. 12, 496, 640641, 641 n. 4, 643, 643 n. 9,
269, 282 644647, 649, 650 n. 12
non-, 257 n. 12 I, 488, 492
non-voiced, 256 n. 7 II, 490492, 497
voiced, 257 n. 12 III, 490492, 492 n. 25
Soqotri, 604 n. 17 V, 491492, 497
sorrow, 397 VI, 492, 497, 497 n. 29
Souchne, 690 IV, 489
Soudan, Sud, 128 V, 488, 490492, 492 n. 24, 497 n. 29
Soukhne, 697 VI, 488, 490492, 492 n. 24, 497 n. 29
sound-related interjections, 696697 VII, 488 n. 19, 496
source, 380, 392395 VIII, 488 nn. 1920
language, 455457, 466467 X, 488
sourdit, 515 n. 31 Arabic, 489
southern Sinai, 565 n. 2, 566567, 567 base, 372373, 488, 491
n. 12, 573, 573 n. 29, 575 n. 35 causative, 262
dialects, 567, 568 n. 14 denominative, 262
southern Sinaitic dialects, 568 derived, 372373
Southern Sudan, 609610 extension, 296297, 299
Sowayan, Saad Abdallah, 544, 563 final weak, 647
Spain, 210, 211 n. 2, 220, 223 n. 28, 639 laryngeal, 646
Spanish, 210, 210 n. 2, 212, 218, perfect, 645
222223, 223 n. 30, 225, 231, 294, perfective, 486487
296, 661 quadriliteral, 497 n. 29
tradition, 212 reflexive, 262
Sparvenfeldius, Johannes Gabriel, 220, verb, 478, 488
220 n. 23 Stephen, 610, 618620, 634, 638
speakers awareness, 5, 8, 12, 17, 20, 22 Stetkevych, Jaroslav, 654, 661, 673
speech(es) Stewart, Frank Henderson, 578, 580
acts, 27 n. 53, 584 n. 67, 586 n. 70, 591 n. 89,
community(ies) 404 592 n. 90
Arabic, 403 Stockholm, 222
context(s), 403405, 410, 410, Stoic, 33, 33 n. 6
n. 8, 413, 417, 423, 432433 Stolbova, Olga V., 605
event, 249250 stop(s) 407, 409, 418419
sounds, 657 glottal, 407
spontaneous, 427 uvular, 407
style(s), 406, 410, 414415, 419420, Strngns, 222
422, 426427, 427 n. 17, 429, strategy
431432 conversational, 417
religious, 251 stylistic, 417
Spitaler, Anton, 21, 2729, 3235, 47, stress, 18 n. 14, 21, 569572, 572 n. 25,
6872, 7475, 78, 78 n. 3, 86, 9294, 576, 591 n. 87, 617
97100, 105 n. 33, 112, 596 n. 6, 605 on the penultimate syllable, 219 n. 19
Sprenger, Alos, 125 stressed, 540, 569571, 573574
stabilit strike a blow to, 255257, 259260,
formelle, 317 263266, 270, 281282, 285286
standard, 404 structuration
variety, 403 binaire, 520521
state construct, 375 structure(s)
statif, 328 complement, 369, 369 n. 7, 379380,
status (manzila), 7 393 n. 79, 395396, 398399
stem(s), 371, 395, 436, 441442, 442 verbal, 369, 378380, 393 n. 79
748 index
exclamative(s), 117, 124 Sudanic, 640

interrogative(s), 117, 124 Arabic, 640, 648
morphological, 475, 495 WSA, 649650, 650 n. 12, 651, 651 n. 13
phonological, 480 n. 5, 481, 483, 488 suffix(es), 15, 292, 292 n. 2, 293294,
radicale, 519 296, 298, 436437, 441442, 444445,
syntactic, 480 n. 5, 481483, 492, 450, 478, 480481, 481 n. 6, 482483,
494495, 497498 486, 494, 494 n. 27, 568 n. 14, 571
syntaxiques, 506 n. 15 n. 24, 572573, 581 n. 56, 592
style(s) n. 91, 611, 640 n. 2, 643 n. 9,
casual, 405, 410411, 427 645646, 648649, 650 n. 12
conversational, 418, 427 agentive
differentiation, 429 -at, 644, 646
formal, 410411 C-initial, 644
shift, 415, 417 C-initial subject, 644
stylistics (bala), 1112, 368 diminutive, 482 n. 8
subject, 18, 2829, 29 n. 1, 29, 38, 68, English, 480481
7374, 89, 9192, 9698, 102, 104, final V-initial, 643
136138, 145146, 149, 152154, 154 finite, 488
n. 6, 155156, 158, 158 n. 13, first person, 541
159160, 162, 165168, 171, 173 future, b-, 224 n. 31
n. 31, 175, 181, 184, 210, 214, 230, inflectional, 249, 294
230 n. 39, 238, 397 n. 86, 476, 484, infinitive, 296
544, 559, 694, 694 n. 26 -kiy, 568 n. 14
clause, 68, 102, 154 n. 7 nominalizing, 484 n. 13
definite, 150, 168 non-finite, 482, 488, 490 n. 22,
explicit, 394 491492, 495, 497
human, 96 non-finiteness, 489, 496
implicit, 394 n. 84 object, 645646
indefinite, 150 paradigmal, 444
nominal, 55, 58 past tense, 442
position, pre-, 155 plural, 645
predicate relation, 68 pronoun, 576
Subjekt, 7375 -, 688, 690
Subjektsstze, 10 subject, 643 n. 8
subjonctif, 124 person, 643
subjunctive, 10, 16, 23, 228, 230 syntactic, 482
n. 39 template, 488
subordinate clause(s), 67, 9, 87 n. 24, verbal, 298
89, 9192, 104, 109, 687 V-initial, 644645
causal, 91 suffixation, 19
substantival clauses, 74 suffixe(s), 509, 612
substantive clause, 7172 C-initial, 644645, 648
subordination, 388, 517518, 520 C-initial subject, 647
subordinators, 388 object, 646, 646 n. 10
subordonnants, 519 relateur, 126, 128 n. 18
succession dtats, 326, 331 subject, 643, 646
Sud Soudan, 621 V-initial, 644, 647649
Sudan, 619, 640641 V-initial subject, 647
Arabic, 685 n. 14 verbal, 612, 640, 643
southern, 639 suffix, 315, 315 n. 4, 316317, 331332,
Sudanese, 614 334
Northern, 608609, 611, 620 suffixed, 19, 568, 572573, 575, 615, 646
South, 607 n. 10
Southern, 608609 object bound pronouns, 659
Southerners, 608 preposition(s), 576577, 579 nn. 4647
index 749
sujet, 123 n. 12, 316, 321 n. 11, 329, positions, 214

335337, 347, 521 n. 49 relations, 12
extrieure, 336 shift, 91
grammatical, 321323, 329, 338, 351 structure, 50, 62 n. 15
n. 32, 354357 subject, 96
humain, 337, 345346, 351, 351 syntactical, 8, 1114, 19 n. 17, 30,
n. 32, 352 42, 72, 76, 9394, 106107, 676,
non-humain, 346, 351352, 356 695
intrieur, 335336, 356 analysis, 11, 13
sukn, 218, 230, 253, 371 constructions, 79
Suleiman, Yasir, 9 n. 7, 23, 661, 673 ramifications, 18
Sumerian, 197 rules, 12
summarizing structure, 687
text-, 369 n. 8 syntactico-semantic, 3
sundq, 658, 664 syntaktisch, 69, 75
sunletters, 574 syntax, 12, 17, 21 28 n. 1, 29, 33, 35,
sunna, 42 62 n. 15, 72, 104, 212, 229, 232, 235,
support, 398 367369, 369 n. 7, 370, 371, 373376,
Sura, 190191, 193194, 198 378, 383384, 387391, 393, 396, 398,
srat Ysn, 195 406407, 412, 416, 418420, 431, 475,
surprise, 397 477, 479, 492, 495496, 497 n. 30,
Suryna, 194 527, 654
suryniyya, 194 morpho-, 3
as-Suyt , 10 n. 8, 17, 17 n. 13, 23, syntaxe, 120, 133, 316, 510
120 n. 5, 131, 192, 194 n. 89, de position, 120
195 n. 10, 198 n. 24, 199 n. 26, syntaxique, 126127
200 nn. 2930, 203 n. 37, 205 n. 42, Syria, 532, 546, 557 n. 35
208 Syriac, 4144 105, 105 n. 32, 106109,
Swadesh, Morris, 667, 673 193196, 201203, 600601
Swanson, D., 298, 311 doxology(ies), 107108
Sweden, 222 New Testament, 107
Sheri, 604 n. 17 Nhs, 192
syllabe(s), 508, 509 n. 23, 510, 511 n. 26, perfect, 103
520, 522, 573 relative, 43
open, 57, 570, 570 n. 21, 571, 574, script, 200
598 nomads, 690
closed, 570 nn. 2223, 574 Syrian, 220
syllable(s), 18 n. 14, 219 n. 19, 548 n. 5 Syrie, 113
closed, 598 Syrien, 697
CVC, 490 syrisch, 105
CVV, 490 Syro-Lebanese
extrametrical, 489, 491 Colloquial Arabic, 667
final, 617 dialects, 667
penultimate, 645 region, 667
synonyms, 38 varieties, 667
syntactic, 9, 16, 33, 50, 62 n. 15, 68, 73, system(s)
77, 79, 9093, 96, 101104, 135, 141, generative, 479
144 n. 44, 611, 675, 677, 687, 692, 694 natural language, 368
n. 27, 695696 nominal, 487
analysis of sentences, 174 verbal, 487
constructions, 135, 141, 144 systme(s)
context, 86, 91 de communication, 510511, 518520,
feature, 92 520 n. 47
non-human, 96 de nomination, 508 n. 21, 510512,
object, 92 514, 516, 519, 520 n. 47
750 index
matriels, 506 n. 14 tat niya, 4

syllabique, 509 n. 23, 510, 513 tawdu, 39, 192
universel, 518 tawahhum, 8 n. 6
Systemzwang, 569, 596 n. 6 Tawfiq al-Hakim, 666
tawld, 115
t marbta, 128 n. 18, 248250, 571, tawqf, 39, 192, 192 n. 3
658659 Tayhah, 566 n. 9
t mabsta, 250 dialects, 567
taajjub, 117 tayn, 170
at -_alib, 199 at-Tayyib, Muhammad Sulaymn, 565 n.
at-Tabar, 45, 5357, 6364, 193, 196, 3, 566 nn. 4, 7, 577
196 n. 14 technical and scientific vocabulary, 656
taba, 39 Teixidor, J., 287 n. 50, 289
tches, 510511 television
smantiques, 505 broadcast 251
Tckholm, Vivi, 578 news 248
tadiya, 158, 178 n. 37, 185 Tell Amarna, 316
tadmn, 116 tem, 217 n. 14
tafsr, 37, 51, 156, 193 temiz, 210, 217
tah qq, 507 n. 18 template(s), 486, 486 n. 17, 490491
tah rf, 194195 CV-, 486
Taine-Cheich, Catherine, 653, 670, 673 prosodic, 486487, 494
tajwd, 126 stem, 488490, 497 n. 29
talabtak, 659 stem I, 488
talabtuka, 659 stem II, 488, 491, 497
tlam, 19 n. 16 stem V, 497
tall, 9 n. 7 verbal, 490
Talmon, Rafael, 62 n. 15, 65, 152 n. 4, temple of Hathor, 566 n. 6
159, 161 n. 17, 173 n. 30, 175 n. 33, temporal markers, 619
188, 444, 453 temps, 504507, 511, 512 n. 27, 513,
_amnn, 195 517 n. 41, 517, 517 n. 41, 519520,
Tamil language, 210 522
Tamm, 120, 126 n. 15 gnral, 515
Tamm, 9, 16 tenifa, 210
Tamis, A., 302, 311 tenivofa, 210
tamyz, 138, 138 n. 21, 140, 146, 217 tense, 248, 251252, 294, 370371, 371
_amd, 202203 n. 17, 372, 384 n. 55, 389, 395
tanwn, 126 n. 15, 126127, 217, 225, mood, 384
509, 515 n. 33 past, 437
taqdm, 120, 149, 161, 168169 present, 294
wa-taxr, 149, 161 Terabin, 567 n. 13
wa-tarx, 120 terminals
taqdr, 56, 6 n. 3, 7, 7 n. 5, 8, 11, non-, 367
142143, 143 n. 41, 146147, 151, terminological, 62
156, 164 terminology, 47 n. 4, 62
_aqf, 199 Texas, 560, 560 n. 41
Tarbn, 566 n. 9, 575, 575 n. 37 text data, 376, 390 n. 70
dialects, 567, 569 n. 17 Arabic, 367, 395, 398399
target, 393, 395 textes arabes
language, 456, 466 mdivaux, 315
Trx, 186 n. 14, 198 n. 21, 199 n. 22 texts
Tasm, 202, 203 religious, 411
Tassa, 277 n. 38 textuality, 425, 432
index 751
Theodory, Constantin, 473 traditionists, 199

theologians, 205 trait commun, 503
thorie transitif(s), 115, 337, 343
du levier, 501 transitive, 247
Theory of Matrices and Etymons, 255, transitivit, 313314, 314 n. 3, 319320,
272 336, 336 n. 22, 343, 344, 349350,
Thom, 506 n. 15 521 n. 49
Thomason, S.G., 294, 311, 665, 673 directe, 320, 337
Thomason and Kaufman, 665 indirecte, 320, 337
Tibriade, 113 transitivity, 379, 379 n. 44, 380382, 391,
tier 435, 441
autosegmental, 488, 498 translation(s), 456458, 465467,
prosodic, 494495 469470, 473
segmental, 488489, 495496 Arabic, 456459, 462, 471
syllabic, 489 English, 457, 467468
tightening, 256, 284, 286 equivalent(s), 457458, 466, 471472
Tigre, 128 of foreign words, 661
Th, 566 French, 457
Tihama, 650 German, 457
time, 373, 380, 394396, 398, 522 profile(s), 467470
Tingstadius, Johan Adam, 210 n. 2, 222, unidirectional, 456
222 n. 27 tree
TMA, 611, 614616 conceptual, 396
token, 29 lexicogenetic, 269270, 276, 278,
Toledano, Moses, 81, 111 280281, 286
Toledo, 82 semantic, 396
tongue, 256 n. 10, 257, 257 n. 13, structures, 477
269270, 275278, 280283, 286, 371 syntactic, 494
topic(s), 368 n. 4, 376, 376 n. 33, 377, Treffers-Daller, 299 n. 9, 311
377 nn. 3738, 405, n. 3, 414417, tribal, 544545, 555, 558559, 561563
421, 423429, 431433 community, 201
agent, 376 n. 31 history, 201, 204
personal, 414 tribales, 129
shift , 417, 427 tribe(s), 189, 195, 199200, 559,
Torre, Patricio Jos de, 220 561, 563, 565, 566 nn. 34, 569,
Tosco, Mauro, 607, 621 571, 586 n. 71, 589 n. 82, 601,
Tottoli, Roberto, 208 626
Toulouse, 221 Arab, 192, 201
traction, 257, 260261 Arab Canaanite, 197
tradition(s), 7778, 87, 90, 9293 Arabic-speaking, 198
arabe, 114, 118, 507, 509 bedouin Sinai, 577
Arab-Islamic 247 Canaanite, 197
grammatical Jordanian, 546
Arabic, 373, 378, 378 n. 40, 380 Jurhum, 197200
grammaticale South Arabic, 202
arabe, 503 n. 6, 504 n. 10, 506507, sub-, 203
507 n. 19, 513, 519 n. 45 Yamani, 202
non-, 509 tribu
orientaliste, 507 de la rgion de la Mecque, 117
sunan, 205 triptote, 124125, 127
traditional Literary Arabic Troupeau, Grard, 6 n. 3, 7 n. 4, 23, 38,
articulation, 657 44, 141 n. 32, 148, 234, 244, 504
phonological system, 659 n. 8, 523
752 index
Truckenbrodt, Hubert, 486 n. 16, 499 u<l:(s), 27, 3438, 4043

Tubayq, 561562 Qsemantics, 40
Tubba Qsources, 37, 41
kings of, 203 `Utayba, 563
Tunis, 84 utterance, 28, 38
Tunisia, 101, 690 uvular(s), 266 n. 28, 657658
Turbniy, 575 n. 37, 576577, 580 n. 53 al-Uwayst, 546
Turcs, 128 Uygur, 449
Turkey, 640 Uzbekistan, 651
Turkic, 435438, 442443, 446, 446 Arabicism, 651 n. 13
n. 10, 448450
Turkish, 294295, 568, 661, 690, 693 V projection, 477
n. 23, 679 n. 7 valeur(s)
Turku, 607, 621 aspectuo-temporelles, 315 n. 4, 325
Twi, 679 n. 7 n. 15, 327, 341, 362
attributive, 340341, 343
Ubayl, 195 causative-factitive, 357
Ubayy b. Kab, 49 n. 6, 56, 56 n. 12, 61 dacquisition dtat, 333
Uganda, 608, 621 dimmutabilit smantique, 342
Ugaritic, 597 n. 8, 597, 599603 doccurrence smantique, 342
uira, 667 morale, 338
UK, 561 physique, 338
ulqiyat, 61 smantique(s), 313314, 319320,
Umayyim, 203 322, 324, 325 n. 15, 327, 329, 334,
Umm al-Jiml, 126 n. 16 337, 339, 341 n. 24, 343344,
Umm Anwar, 683 346347, 350, 354, 358359
umma, 404, 404, n. 2 morpho-lexicale, 327
Unz Ab Slim at-Turbn, 545 n. 4 stative, 330, 333334
unification-principle, 370, 383 Valladolid, 219 n. 20
unit(s), 508, 511512 van Dijk-Wittop Koning, A.M., 298, 311
amorphes, 519 Vanhove, Martine, 100, 112, 685 n. 14,
de nomination, 506, 509, 509 n. 23, 693 n. 23, 697, 699
510, 512, 515 n. 33, 516518 van Staden, M., 292, 311
banale, 514 variety(ies)
commune, 514 Arabic, 418, n. 13
gnrale, 514 dominant, 408, 419422
de nomination des entits, 508 literary, 403
de nomination des expriences, 508 mixed
flchies, 519 Arabic, 420
units, 367 non-dominant, 422
lexical, 456 superposed, 404
Upper Egyptian Arabic, 697 velarization, 371, 571, 576, 578 n. 41,
Uppsala, 222 n. 27 591 n. 88
Ur, 197 verb(s), 68, 13 n. 9, 16, 18, 30, 41,
urban 47, 4951, 5354, 5657, 59, 70,
Colloquial Arabic, 657 82, 83 n. 18, 84, 86, 89, 93, 9799,
dialects, 657, 658 102103, 105, 108109, 139, 142, 142
speakers, 657 n. 37 144, 144 n. 51, 149, 151152,
Ursprache, 118 152 n. 3, 153, 154 n. 6, 154, 154 n. 7,
US, 531532, 536 155157, 158 n. 13, 159, 163, 165
uskitat, 61 167, 177178, 178 n. 36, 179180,
al-Umn, 17 n. 13, 23 183186, 196, 213, 218 n. 17, 221 n.
u<l al-=qh, 2628, 33, 4243 24, 225, 228, 230, 230 n. 39, 231, 231
index 753
n. 41, 233 n. 47, 235, 235 n. 49, 239, gl, yigl, 675, 677678, 697
258260, 263, 270271, 282, 284, 292, of the heart, 378 n. 41
292 n. 2, 293297, 297 n. 7, 298304, Hebrew, 596 n. 6
304 n. 11, 305, 307309, 367, 369, 369 of hope, 378 n. 41
n. 7, 370, 370 nn. 12, 14, 371, 379, imperative, 34
391, 393394, 397, 407, 435437, incomplete, 378 n. 41
439442, 447, 449, 451452, 475 n. 1, incorporation, 294
475476, 481, 482 n. 7, 484 n. 13, 486, integration, 292293, 295, 296 n. 4,
487 n. 18, 490, 497 n. 29, 498, 532 297298, 306309
n. 5, 572, 575 n. 36, 576577, 577 intransitive, 391, 435, 482, 695
n. 40, 578 n. 42, 580 nn. 50, 53, 581 lexical, 675
n. 56, 589 n. 83, 592 n. 91, 597 n. 8, li- +, 54
612613, 616, 619, 639641, 641 n. 4, -like constituent, 152
642644, 646, 646 n. 10, 647, 650 -like element, 160
n. 12, 649, 651, 676 n. 4, 679 loan, 297, 652
-acting constituent, 156 locutive and speech-related, 693
of action, 394 modal, 371 n. 17
l, yil, 675678, 681, 692696 monotransitive, 380
of approximation, 378 n. 41 +noun dichotomy, 165
banish, 391, 395, 399 +noun type, 165
base, 371 patterns, 660
of beginning, 378 n. 41 of praise and blame, 378 n. 41
bitransitive, 380 prepositional, 63
C- final, 649 prohibitive, 34
of change, 394 pseudo-, 73
classes, 391, 396, 399 remove, 392, 395
cognitive, 178 serial, 675
conjugation, 617 stem(s), 292, 294, 297, 299, 301, 306,
construction 641, 645, 649
bilingual periphrastic, 291, 297 foreign, 292, 306
coreferential, 92 I, 407
delocutive, 698 IV, 406407, 419
embedded, 307, 309 strong, 640, 644, 646 n. 10, 649
English, 305306 Syriac, 109
of esteem, 378 n. 41 system, 660
of experience, 394 temporal, 371 n. 17
final of transformation, 394
//, 650 transitive, 391
/h, 648649 bi-, 391
// or /h/, 642, 647 mono-, 391
laryngeal-, 646 n. 10, 648649, 650 tri-, 391, 380
n. 12 triliteral, 487
pharyngeal-, 650 n. 12 Turkic, 437, 440, 442 n. 6
weak, 641 type, 369, 372, 395
finite, 97, 155, 163, 178, 183, 294, 296, weak, 369, 372, 644, 649
371372, 372 n. 19, 378, 378 n. 42, final, 252253, 640, 644, 647650,
379380, 384, 394 n. 84, 395, 491 650 n. 12, 648649
foreign, 292293, 295, 306309 verbal, 5455, 57, 72, 86 n. 23, 89,
imperfect, 228 108109, 608, 611614, 616, 640
embedded, 307 analysis, 151, 153154, 154 n. 8,
morphologically, 303304, 306 155157
transitive, 293 clause, 163
form(s), 294, 407, 409, 640 construction, 155
754 index
effect, 156 mdio-passif(s), 335336, 344, 346,

element, 165 n. 25, 181 352, 354356, 358360
ending(s), 210, 228 moyen(s), 320321, 335, 335 n. 20,
force, 155156, 168, 176 336337, 344, 346347, 349350,
form(s), 531, 679, 613, 615616 352, 354, 359360
function, 151, 153156, 166 intransitif(s), 355, 337
measures, 572 transitif(s), 328 n. 19, 338 n. 23
modals, 695 direct, 337
paradigm(s), 224 n. 31, 230 n. 39 indirect, 337
phrase, 55 passif, 343
prepositional phrases, 54 de perception, 323
sentence(s), 6, 77, 139, 149, 151, primitif, 314
154 n. 8, 157 pronominaux, 354
status, 178 de qualit, 321 n. 11, 328329
structures, 54 redoubls, 353, 353 n. 34, 354, 356
system, 230 n. 39 regular, 575
type(s), 10, 81, 86, 8990, 92-93, de sentiment, 323
98, 109 simple(s), 313, 315316, 316 n. 5,
Verbalphrasen, 699 317322, 322 n. 12, 323, 324,
verbe(s), 115, 118, 134, 314315, 326327, 337338, 344, 350, 353,
318320, 323, 326330, 332335, 357360, 362
339343, 345352, 353 n. 35, 354356, statifs, 320
358362, 504505, 506, 506 n. 15, 509, transitifs, 320, 337, 338 n. 23, 341,
513, 515 n. 35, 573, 575, 621 349, 352, 354, 356357
dacquisition dtat, 335, 344, 351, directs, 348351
358360, 362 indirects, 347351
caractristique, 352 vernaculaire(s), 125
non-caractristique, 334 vernacular(s), 223, 403, 527, 559560,
actif(s), 320, 336337, 350 607, 609610, 613
daction, 314, 319321, 323, 336, 359 African, 609, 611
agentif(s), 354 Non-Arabic, 610
entirement, 359 speech, 227
non-, 322-323 336, 344, 346, 351, vernacularization, 608, 618
351 n. 32, 355, 359 vernacularized, 611, 620
partiellement, 322, 359 non, 612
pleinement, 322, 328 n. 19, Arabic, 607
350351 verse(s), 46 n. 2, 189, 204, 548 n. 5
dagentivit neutralise, 328 n. 19 pre-Islamic, 89
arabe, 313 Versteegh, Kees (C.H.M.), 33, 4445, 65,
augment(s), 313, 324 114, 117, 120 n. 6, 121 n. 10, 124, 126,
non, 315 128, 128 n. 20, 133, 209 n. 1, 227, 364,
de changement dtat, 328329, 331, 371 n. 15, 402, 434, 436, 441, 453, 455,
333 473, 503 n. 6, 504 n. 9, 523, 565 n. 1, 578,
commun, 508 607, 621, 639, 652, 654, 656, 664665,
concaves, 353354, 356 671, 673, 675, 675 n. 1, 692, 693, 699
caractristique, 322, 330, 334, vocabulary, 654, 656, 662, 665666,
341343, 352, 357, 359, 361 668670
dtat, 314, 319, 320, 320 n. 9 321, vocalisation, 233 n. 45, 445
323, 326, 328, 334336, 342344, vocalism, 604
354, 359, 362 vocatiuo, 214
externe(s), 336337, 350 vocative(s), 41, 214, 217, 375, 387,
intransitifs, 320, 338, 344, 346348, 397398, 676, 696
351, 356 noun, 41
index 755
vocativus, 221 short, 247, 251, 449, 568, 569

voice(s), 370372, 395, 426432 final, 248249, 251
active, 379 short high front and back, 641 n. 4
passive, 379380, 397 n. 88 system, 224, 224 n. 32, 233, 657658
voiced, 256 n. 7 terms, mnemonic, 217
non-, 256, 256 n. 7, 267, 275 omission of, 224 n. 31
voicing, 371 zero-, 227
Voigt, Rainer, 595 n. 1, 599 voyelle(s), 314315, 316 n. 5, 317,
volition, 370, 383, 396397 319, 338, 345 n. 28, 510, 510 n. 24,
Vollers, Karl, 125, 130 511, 511 nn. 2526, 512513, 515
Vopadeva, 209 n. 31
vowel(s), 9, 15, 47, 214 n. 10, 217219, dsinentielle, 509
219 n. 19, 224 nn. 3132, 225228, premire, 513
230, 233, 239240, 247, 250, 256, 368, mdiane(s), 313, 314 n. 3, 315,
435, 445, 446 n. 10, 447450, 540, 317319
543, 568570, 570 n. 23, 572, 572
n. 27, 573575, 575 n. 34, 577, 579 wa-, 38
n. 44, 582 n. 60, 595, 597598, 601, wa/, 39
612613 w/i (creator), 10
back, 296 Wdiy
Canaanite, 598 sSa haw, 566
declensional, 230 n. 39 Frn, 566
deletion 251 arandal, 566
empty, 371 Isla, 566
ending(s), 4647, 63 n. 16, 252253, Lihyn, 566
543 Wahhabi scholars, 200
short-, 63, 64 n. 17 Wahd, 685
final, 16, 63, 253, 295, 646 n. 10 wahy, 192, 193
stem, 643644, 649 wajh, 48, 249
front, 296 wjib, 169
high stem, 641, 643 n. 9 Walid Irfa, 535
infixed, 371 Wald Ysif, 699
inflectional, 216, 224227 wallada, 115
-initial, 572, 572 n. 27 Wallin, Georg August, 544, 563
length, 449, 488 waqf, wuqf, 85, 227, 247, 247253
lengthened, 448 War an Nfi, 127
long, 252253, 407, 407 n. 4, 440, 449, wasfiyya, 158
488489, 491, 491 n. 23, 611, 614, wasl, 127
640, 641 n. 4, 649 Watson, Richard, 607, 619, 621
low stem, 643 n. 9, 641 wafa, 249
marks, 251 wazn, 435, 444, 509
mutation, 9 Wehr, Hans, 344, 365, 599, 600, 601,
omission, 253 602, 603, 604, 605, 668, 673
pattern(s), 370, 411412, 447, 488, Wehr-Cowan, 315, 331, 339, 340, 343,
490, 492 353
prefix(ed), 371, 641 n. 4 Weinreich, Uriel, 665, 674
preformative, 262, 597 Weiss, Bernard G., 193, 208
quantity, 488 Wei, Josef, 228, 234, 244
rear, 256 n. 10 Wellens, Ineke, 607, 621
semantic, 224 Werner, John Henry, 220
semi-, 641 Wernitz, C.J., 292, 311
sequence, 436, 439 West Germany, 546
stem, 641 n. 4, 644645 Western
756 index
languages, 104 Xawrij, 530

Sudanese Arabic, 614 al-Xidr, 195
Sudanic, 611 xiftu, 660
tradition, 219 xilf (or muxlafa), 162
Wetzstein, Johann Gottfried, 126, 133 XP, 478
wi (and), 677, 679, 694, 695 xucla, 210, 217
wi, 163 xuft, 660
Wild, Stefan, 435, 436, 437, 438, 453 al-Xwrazm, 33
Winograd, 393, 402
wish, 398 y, 41
Woidich, Manfred, 7278, 80, 85, 86, Yahwe, 105
9498, 100-102, 112, 224, 224 n. 31, Yaman, 202, 203
402, 455, 473, 565 n. 2, 675, 675 Yamani(s), 202
n. 2, 678, 683, 683 n. 12, 687, 687 yaqdiru, 660
n. 17, 688, 688 n. 18, 692, 692, n. 22, al-Yaqb, 199
694696, 699 Yqt, 202, 203
word yarifu, 660
classes, 213, 236 Yarub, 202, 203, 204
-final position, 657 Yarub b. Qaht n, 196, 199, 201204
formation, 475, 479, 498 Yajub b. Qaht n, 195, 199, 202
order, 12 Yajub/Yaman, 202
Wright, William, 314, 320, 327, 338, 365, Yassin, M.A., 555, 563
373, 374, 378, 385, 386, 388, 435, yxud, 660
436, 453, 487, 499 yaxudu, 660
written tradition, 210 Yemen, 195
WSA, 610, 616 Yemeni dialects, 640
Wurff, Wim van der, 613, 614, 697 Yeou, M., 266, 289
Wurzelbegriff, 219 yieraf, 660
Wstenfeld, Ferdinand, 195, 203, 207 yidar, 660
Yishml, 198
xabar, 48, 49, 138, 144 n. 19, 149, 150, Yngve, Victor, 389, 402
151, 152, 153, 154, 155, 156, 157, Yoda, Sumikazu, 100, 112
158, 159, 160, 161, 162, 163, 164, Yokwe, Elisai M., 607, 621
168, 169, 170, 171, 172, 173, 174, Yoon, James, 483, 499
176, 178, 179, 180, 182, 183, 184, 216 Youssi, Abderrahim, 674
accusatival, 159 yxed, 660
adverbial, 169 yxud, 660
adverbial/prepositional, 164, 168 Yucatn, 219
clausal, 164, 180 Yumn Bassiouni, 533, 536
fronted, 153, 182 Ynus, 195
kna, 179 Ysuf al-Qad, 699
non-adverbial, 169 Ysuf f, 699
muqaddam, 153 Zaborski, Andrezej, 402
mubtada, 180 hir, 151
obligatorily fronted, 168, 173, 174 |hirite(s), 11
obligatory fronting, 170, 171 az-Zajjj, 57, 58, 64, 113
preposed obligatorily, 173 az-Zajjj, 113, 114, 121124, 126, 131,
as-sifa, 173 153, 154, 157, 186, 187, 318, 362
xaf/, 60, 234 zakh, 31
xfi/, 58, 60 Zakariyy, Michel, 3, 23
Xlid b. Sinn al-Abs, 195 Zakia Iraqui Sinaceur, 698
al-Xall b. Ahmad, 28, 29 n. 1, 33 az-Zamaxar, 89, 111, 157, 177, 180,
151, 152, 155156, 238, 444, 453 287, 320, 329, 338, 342, 343, 345, 362
index 757
Zande, 609 arfiyya, 177, 184

arf, 135, 136, 137, 138, 139, 140, 141, Zarpetea, P., 303, 311
142, 143, 144, 145, 146, 147, 162, az-Zarq, 546
177, 181, 214, 217. zawid, 509
dispensable part, 135, 136 Zayd, 13
indispensable part, 135 zayy, 78, 689
indispensable predicate, 136 illi, 74, 77, 78
maa, 144 zero endings, 230
mulan, 142, 147
STUDIES IN SEMITIC
LANGUAGES AND LINGUISTICS
3. Corr, A.D. The Daughter of My People. Arabic and Hebrew Paraphrases of
Jeremiah 8.13-9.23. 1971. ISBN 90 04 02552 9
5. GrandHenry, J. Les parlers arabes de la rgion du Mza b (Sahara algrien). 1976.
ISBN 90 04 04533 3
6. Bravmann, M.M. Studies in Semitic Philology. 1977. ISBN 90 04 04743 3
8. Fenech, E. Contemporary Journalistic Maltese. An Analytical and Com-
parative Study. 1978. ISBN 90 04 05756 0
9. Hospers, J.H. (ed.). General Linguistics and the Teaching of Dead Hamito-Semi-
tic Languages. Proceedings of the Symposium held in Groningen, 7th-8th
November 1975, on the occasion of the 50th Anniversary of the Institute
of Semitic Studies and Near Eastern Archaeology of the State University at
Groningen. 1978. ISBN 90 04 05806 0
12. Hoftijzer, J. A Search for Method. A Study in the Syntactic Use of the H-
locale in Classical Hebrew. With the collaboration of H.R. van der Laan
and N.P. de Koo. 1981. ISBN 90 04 06257 2
13. Murtonen, A. Hebrew in its West Semitic Setting. A Comparative Survey of
Non-Masoretic Hebrew Dialects and Traditions. Part I. A Comparative
Lexicon.
Section A. Proper Names. 1986. ISBN 90 04 07245 4
Section Ba. Root System: Hebrew Material. 1988. ISBN 90 04 08064 3
Section Bb. Root System: Comparative Material and Discussion. Sections C,
D and E: Numerals under 100, Pronouns, Particles. 1989.
ISBN 90 04 08899 7
14. Rets, J. Diathesis in the Semitic Languages. A Comparative Morphological
Study. 1989. ISBN 90 04 08818 0
15. Rouchdy, A. Nubians and the Nubian Language in Contemporary Egypt. A Case of
Cultural and Linguistic Contact. 1991. ISBN 90 04 09197 1
16. Murtonen, A. Hebrew in its West Semitic Setting. A Comparative Survey of
Non-Masoretic Hebrew Dialects and Traditions. Part 2. Phonetics and
Phonology. Part 3. Morphosyntactics. 1990. ISBN 90 04 09309 5
17. Jongeling K., H.L. Murre-van den Berg & L. van Rompay (eds.). Studies in
Hebrew and Aramaic Syntax. Presented to Professor J. Hoftijzer on the Occa-
sion of his Sixty-Fifth Birthday. 1991. ISBN 90 04 09520 9
18. Cadora, F.J. Bedouin, Village, and Urban Arabic. An Ecolinguistic Study.
1992. ISBN 90 04 09627 2
19. Versteegh, C.H.M. Arabic Grammar and Qur"a nic Exegesis in Early Islam.
1993. ISBN 90 04 09845 3
20. Humbert, G. Les voies de la transmission du Kita b de Sbawayhi. 1995. ISBN
90 04 09918 2
21. Mifsud, M. Loan Verbs in Maltese. A Descriptive and Comparative Study.
1995. ISBN 90 04 10091 1
22. Joosten, J. The Syriac Language of the Peshitta and Old Syriac Versions of Matthew.
Syntactic Structure, Inner-Syriac Developments and Translation Tech-
nique. 1996. ISBN 90 04 10036 9
23. Bernards, M. Changing Traditions. Al-Mubarrads Refutation of Sbawayh
and the Subsequent Reception of the Kita b. 1997. ISBN 90 04 10595 6
24. Belnap, R.K. and N. Haeri. Structuralist Studies in Arabic Linguistics. Charles A.
Fergusons Papers, 1954-1994. 1997. ISBN 90 04 10511 5
25. Talmon R. Arabic Grammar in its Formative Age. Kita b al-"Ayn and its Attribution
to ]all b. Ah.mad. 1997. ISBN 90 04 10812 2
26. Testen, D.D. Parallels in Semitic Linguistics. The Development of Arabic la- and
Related Semitic Particles. 1998. ISBN 90 04 10973 0
27. Bolozky, S. Measuring Productivity in Word Formation. The Case of Israeli
Hebrew. 1999. ISBN 90 04 11252 9
28. Ermers, R. Arabic Grammars of Turkic. The Arabic Linguistic Model Applied to
Foreign Languages & Translation of #Abu- ayya-n al-#Andaluss Kita-b al-"Idra-k li-
Lisa-n al-"Atra-k. 1999. ISBN 90 04 113061
29. Rabin, Ch. The Development of the Syntax of Post-Biblical Hebrew. 1999.
ISBN 90 04 11433 5
30. Piamenta, M. Jewish Life in Arabic Language and Jerusalem Arabic in Communal Per-
spective. A Lexical-Semantic Study. 2000. ISBN 90 04 11762 8
31. Kinberg, N. ; Versteegh, K. (ed.). Studies in the Linguistic Structure of Classical
Arabic. 2001. ISBN 90 04 11765 2
32. Khan, G. The Early Karaite Tradition of Hebrew Grammatical Thought. Including a
Critical Edition, Translation and Analysis of the Diqduq of "Ab Ya#qb
Ysuf ibn N on the Hagiographa. 2000. ISBN 90 04 11933 7
33. Zammit, M.R. A Comparative Lexical Study of Qur"nic Arabic.
ISBN 90 04 11801 2 (in preparation)
34. Bachra, B.N. The Phonological Structure of the Verbal Roots in Arabic and Hebrew.
2001. ISBN 90 04 12008 4
35. kesson, J. Arabic Morphology and Phonology. Based on the Mar al-arw by
Amad b. #Al b. Mas#d. Presented with an Introduction, Arabic Edition,
English Translation and Commentary. 2001. ISBN 90 04 12028 9
36. Khan, G. The Neo-Aramaic Dialect of Qaraqosh. 2002. ISBN 90 04 12863 8
37. Khan, G., ngeles Gallego, M. and Olszowy-Schlanger, J. The Karaite Tradi-
tion of Hebrew Grammatical Thought in its Classical Form. A Critical Edition and
English Translation of al-Kitb al-Kf f al-Lua al-#Ibrniyya by "Ab al-Faraj
Hrn ibn al-Faraj. 2 Vols. 2003. ISBN 90 04 13272 4 (Set), ISBN 90 04
13311 9 (Vol. 1), ISBN 90 04 13312 7 (Vol. 2)
38. Haak, M., De Jong, R., Versteegh, K. (eds.). Approaches to Arabic Dialects.
A Collection of Articles presented to Manfred Woidich on the Occasion of
his Sixtieth Birthday. 2004. ISBN 90 04 13206 6
39. Takcs, G. (ed.). Egyptian and Semito-Hamitic (Afro-Asiatic) Studies in Memoriam
W. Vycichl. 2004. ISBN 90 04 13245 7
40. Maman, A. Comparative Semitic Philology in the Middle Ages. From Sa#adiah
Gaon to Ibn Barn (10th-12th C.). 2004. ISBN 90 04 13620 7
41. Van Peursen, W.Th. The Verbal System in the Hebrew Text of Ben Sira. 2004.
ISBN 90 04 13667 3
42. Elgibali, A. Investigating Arabic. Current Parameters in Analysis and Learning.
2004. ISBN 90 04 13792 0
43. Florentin, M. Late Samaritan Hebrew. A Linguistic Analysis of Its Different
Types. 2004. ISBN 90 04 13841 2
44. Khan, G. The Jewish Neo-Aramaic Dialect of Sulemaniyya and \alabja. 2004.
ISBN 90 04 13869 2
45. Wellens, I. The Nubi Language of Uganda. An Arabic Creole in Africa. 2005.
ISBN 90 04 14518 4
46. Bassiouney, R. Functions of Code Switching in Egypt. Evidence from Mono-
logues. 2006. ISBN 90 04 14760 8
47. Khan, G. Semitic Studies in Honour of Edward Ullendorff. 2005.
ISBN 90 04 14834 5
48. Mejdell, G. Mixed Styles in Spoken Arabic in Egypt. Somewhere between Order
and Chaos. 2006. ISBN-10: 90 04 14986 4, ISBN-13: 978 90 04 14986 1
49. Ditters, W.E. and Motzki, H. (eds.). Approaches to Arabic Linguistics. Presented to
Kees Versteegh on the Occasion of his Sixtieth Birthday. 2007.
ISBN 978 90 04 16015 6

Approach To Arabic Linguistics Diglosie 403 PDF

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Approach To Arabic Linguistics Diglosie 403 PDF

Hochgeladen von

Copyright:

Verfügbare Formate

Approaches to Arabic Linguistics

Studies in Semitic Languages

Library of Congress Cataloging-in-Publication Data

Copyright 2007 by Koninklijke Brill NV, Leiden, The Netherlands.

All rights reserved. No part of this publication may be reproduced, translated,

Authorization to photocopy items for internal or personal use is granted by

printed in the netherlands

Bibliography Kees Versteegh ............................................................ xv

Inside the Speakers Mind: Speakers Awareness as Arbiter of

Pragmatics and Contractual Language in Early Arabic

Idmr in the Man of al-Farr: A Grammatical Approach

Arabic allad as a Conjunction: An old Problem and a New

Les origines de la grammaire arabe, selon la tradition:

Sbawayhis View of the zarf as an mil ......................................... 135

Problems in the Medieval Arabic Theory of Sentence Types ......... 149

Inflection and Government in Arabic According to Spanish

The Linguistic Analysis and Rules of Pause in Arabic ................... 247

The Explanation of Homonymy in the Lexicon of Arabic ............. 255

The Periphrastic Bilingual Verb Construction as a Marker of

Faula, faila, faala: dispersion et rgularits smantiques dans

Featuring as a Disambiguation Tool in Arabic Natural

Arabic on the Media: Hybridity and Styles ..................................... 403

The Use of Morphological Patterns in Arabic Grammars of

Lexical Gaps in Arabic: Evidence from Dictionaries ...................... 455

Masdar Formation .............................................................................. 475

Mthodologie linguistique: organisation de la langue arabe.

How to be KOOL in Arabic Writing: Linguistic Observations

Hello, I say, and welcome! Where from, these riding men?

Notes on the Dialects of the Lgt and H amda h of Southern

Classical and Colloquial Arabic Archaisms ..................................... 595

Do They Speak the Same Language? Language Use in Juba

Paradigmatic Stability and Final Laryngeals in Nigerian Arabic:

Some Aspects of Diglossia as Reflected in the Vocabulary of

Everything you Always Wanted to Know about l, yil to say

Index .................................................................................................... 701

From when we began to compile a Festschrift for Kees Versteegh on the

apparent in his search for Hellenistic elements in Arabic linguistics.

On the tunes of birds that speak true Arabic,

who use metres that are neglected by al-Khall and unknown

Everhard Ditters and Harald Motzki

Latinitas, Hellnisms, Arabiyya. The History of Linguistics in the Clas-

Geschiedschrijving in de klassieke Arabisch-Islamitische samenleving.

New approaches towards the history of Arabic. Taqaddum al-lisniyyt

Robins, Robert H. 1993. The Byzantine Grammarians: Their place

honor of El-Said Badawi. Edited by Alaa Elgibali. Cairo: American

Bekkum, Jan Houben, Ineke Sluiter and Kees Versteegh. Amsterdam:

C. E. Bosworth, E. van Donzel and W. P. Heinrichs. Leiden: E. J.

by Geert Boo, Christian Lehmann, Joachim Mugdan and Stavros

The Arab grammatical theory as represented by the later grammar-

One of the most significant aspects of Sbawayhis method of gram-

of operations performed by the speaker in order to achieve a specific

2. Meaning and speakers awareness

The relationship between meaning and the speakers awareness in

(suppletive insertion of elided elements), mainly because, as a gram-

yasruxu) before the accusative noun (Kitb I:355f.). The intention of

Jinn stands out as a unique scholar whose innovative approach in his

over others (II:33), or to the onomatopoeic nature of phonemes within

a reflection of the meaning he intends. Based on this and on the simi-

4. The importance of the particles

The importance which Sbawayhi attaches to the speakers competence

h ikya, as in the use of ayyayni/ayyna or manayni/manna in response

h aytu- is added to the conditional sense which h aytum expresses. Sim-

1. innam: h ikya (III:331); h arf wh id (II:418; III:57); laysat -m bi-

h aytu- is added to the conditional sense which h aytum expresses. Sim-

regarded as single words) while kad wa-kd stands for compound

2.3 Man I:1112 regarding Q 2:2 [dlika l-kitbu l rayba fhi]

2.5 Man I:16 regarding Q 2:1718 wa-tarakahum f ulumtin . . .

2.6 Man I:17 regarding Q 2:19 yajalna asbiahum . . . h adara

case, e.g. in Q 2:19 (Man I:17, 1ff.): yajalna asbiahum . . . h adara

3.1.2 Man I:35 regarding Q 2:72: wa-id qataltum nafsan

3.2.3 Man I:271 regarding Q 4:46 mina lladna hd yuh arrifna