Bo Isaksson - Maria Persson - Clause Combining in Semitic - The Circumstantial Clause and Beyond-Harrassowitz (2015)

Abhandlungen für die Kunde
des Morgenlandes
Band 96
Clause Combining in Semitic:

The Circumstantial Clause and Beyond
Edited by
Bo Isaksson and Maria Persson
Deutsche Morgenländische Gesellschaft

Harrassowitz Verlag
© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden
ISBN Print: 978-3-447-10405-0 ISBN E-Book: 978-3-447-19049-7
Abhandlungen für die Kunde
des Morgenlandes
Im Auftrag der Deutschen Morgenländischen Gesellschaft
herausgegeben von Florian C. Reiter
Band 96
Board of Advisers:
Christian Bauer (Berlin)
Desmond Durkin-Meisterernst (Berlin)
Lutz Edzard (Oslo/Erlangen)
Jürgen Hanneder (Marburg)
Herrmann Jungraithmayr (Marburg)
Karénina Kollmar-Paulenz (Bern)
Jens Peter Laut (Göttingen)
Joachim Friedrich Quack (Heidelberg)
Michael Streck (Leipzig)
2015
Harrassowitz Verlag . Wiesbaden

Clause Combining in Semitic:
The Circumstantial Clause and Beyond
Edited by
2015
Harrassowitz Verlag . Wiesbaden

Bibliografische Information der Deutschen Nationalbibliothek
Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen
Nationalbibliografie; detaillierte bibliografische Daten sind im Internet
über http://dnb.dnb.de abrufbar.
Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publication in the Deutsche
Nationalbibliografie; detailed bibliographic data are available on the internet
at http://dnb.dnb.de.
For further information about our publishing program consult our

website http://www.harrassowitz-verlag.de
© Deutsche Morgenländische Gesellschaft 2015
This work, including all of its parts, is protected by copyright.
Any use beyond the limits of copyright law without the permission
of the publisher is forbidden and subject to penalty. This applies
particularly to reproductions, translations, microfilms and storage
and processing in electronic systems.
Printed on permanent/durable paper.
Printing and binding: Hubert & Co., Göttingen
Printed in Germany
ISSN 0567-4980
ISBN 978-3-447-10405-0
e-ISBN 978-3-447-19049-7

Contents
Editors’ preface ............................................................................................ 7
Clause Combining in Arabic dialects
Heléne Kammensjö
Circumstantial Clause Linking in Egyptian Arabic Narration ..................... 15
Maria Persson
Non-main Clause Combining in Damascene Arabic:
A scale of markedness.................................................................................. 55
Clause Combining in Written Arabic
Michal Marmorstein
The Domain of Verbal Circumstantial Clauses in Classical Arabic............. 125
Clause Combining in Biblical Hebrew
Bo Isaksson
The Verbal System of Biblical Hebrew. A Clause Combining Approach ... 169
Clause Combining in Modern Spoken Aramaic
Eran Cohen
Circumstantial Clause Combining in the Jewish Neo-Aramaic Dialect
of Zakho ....................................................................................................... 271
Clause Combining in Epigraphic South Arabian
Jan Retsö
The Problem of Circumstantial Clause Combining (CCC) in Sabaean........ 297
Clause Combining in East Semitic
Eran Cohen
Circumstantial Clause Combining in Old Babylonian Akkadian ................. 365
Index of terms ............................................................................................. 407

Editors’ preface
We became aware of the role played by ‘clause combining’ while reading the
ground-breaking book Clause combining in grammar and discourse (1988).1
In the chapter “The structure of discourse and ‘subordination’” Matthiessen
and Thompson discuss how “clause combining is a grammaticization of the
rhetorical discourse”. They state that clause combining reflects the rhetorical
intentions of the author or narrator, and that “the interesting cross-linguistic
issue is how and to what extent the grammar of clause-combining in a given
language reflects the rhetorical organization of discourse in that language”.2
Discovering the importance of ‘clause combining’ in Semitic texts was one of
the major achievements of the research project Circumstantial Qualifiers in
Semitic: The Case of Arabic and Hebrew,3 which preceded and paved the way
for the research presented in this volume.
Clause Combining in Semitic: The Circumstantial Clause and Beyond ex-
amines how different kinds of clauses combine to a text in a number of Semit-
ic languages. Specifically, many of its chapters examine how circumstantial
clauses are coded in individual Semitic languages.4
The book comprises the results of a research project, Circumstantial
Clause Combining in Semitic, funded by the Swedish Research Council. As is
nearly commonplace in research projects, considerations regarding the most
‘useful’ or ‘fruitful’ or ‘productive’ terminology resulted, for some of us, in a
more general conceptual approach. The term ‘circumstantial’ in the project
title was originally a reflex of the phenomenon of ḥāl (‘circumstance’) clauses
1 Haiman, John, and Sandra A. Thompson, eds., Clause combining in grammar and
discourse (Typological studies in language 18. Amsterdam; Philadelphia: John Benja-
mins, 1988).
2 Matthiessen, Christian, and Sandra A. Thompson, “The structure of discourse and ‘sub-
ordination’”, in Clause combining in grammar and discourse, 299, 317.
3 Abhandlungen für die Kunde des Morgenlandes 70 (Wiesbaden: Harrassowitz, 2009).
4 We could also have chosen ‘clause linking’ for the title of the book as the terms are used
interchangeably. ‘Clause linking’ was used by R. M. W. Dixon in his chapter “The se-
mantics of clause linking in typological perspective”, in The semantics of clause linking:
A cross-linguistic typology (edited by R. M. W. Dixon and Alexandra Y. Aikhenvald, 1-
55, Oxford: Oxford University Press, 2009). However, ‘clause combining’ inspires us to
imagine how a creator of a text (oral or written) actively combines clauses to achieve a
specific communicative efficiency.

8 Editors’ preface
in Arabic. But it was soon felt that ‘circumstantial’ was too narrow a concept
to cover the complexity of the phenomena we wanted to investigate. As an
example, it was questioned whether a clause that expresses an elaboration (of
the action in a previous clause) could justly be called ‘circumstantial’.5 Thus,
the general scope of the project had to be widened to an investigation of non-
main clause linking in Semitic. This was done with some limitations, though.
We did not primarily study relative clauses, nor subject and object clauses
(occupying the subject or object ‘slot’ in a main clause).6 Clauses introduced
by subordinating conjunctions were left out as well, since the linking of them
is made semantically explicit by the conjunction.
The questions put forward in the project were: How is hypotaxis marked in
Semitic, other than by conjunctions? How does this affect the organization of
texts? More specifically, what constitutes a circumstantial clause? To find an
answer to these questions, all the major Semitic language families and some
modern spoken Semitic dialects were covered within the project.7
The Semitic varieties were chosen with the aim to cover a wide range of
the Semitic linguistic spectrum. We were also aiming, as far as possible, for
varieties that are, or have been at some point in time, the native tongues of
their users, and to capture them in that time and function. In other words, our
goal was, wherever possible, to analyse texts which represent a native compe-
tence of the language users. Thus, for the classical Hebrew texts, for example,
the ambition was to choose texts that were so early that they may be assumed
to represent a living language (though possibly belonging to a higher register).
Kammensjö describes types of circumstantial clause linking in a corpus of
spoken Egyptian Arabic narratives. Asyndetic hypotactic linking was found to
be more than twice as common as syndetic hypotactic linking in Kammensjö’s
5 Thus M. Waltisberg, Satzkomplex und Funktion: Syndese und Asyndese im Althochara-

bischen (Wiesbaden: Harrassowitz, 2009), who calls such a clause ‘Umstandssatz’
(213). Cf. Bo Isaksson’s review of Waltisberg’s book in Zeitschrift der Deutschen mor-
genländischen Gesellschaft 164 no. 1 (2014): 247-251.
6 Thus also the limitation in Dixon’s “The semantics of clause linking in typological
perspective”, 1, cf. note above.
7 For Ethio-Semitic we refer to Lutz Edzard, “Complex predicates and Circumstantial
Clause Combining (CCC): Serial verbs and converbs in a comparative Semitic perspec-
tive”, in Strategies of Clause Linking in Semitic Languages: Proceedings of the Interna-
tional Symposium on Clause Linking in Semitic Languages 5-7 August 2012 in Kivik,
Sweden, edited by Bo Isaksson and Maria Persson (Wiesbaden: Harrassowitz, 2014,
207-230).

Editors’ preface 9
data. She concludes that EA circumstantial clauses are usually subtly marked,
most often through interlacing (sharing semantic elements) and gram switch-
ing (verb form contrasts). This means that EA circumstantial clauses are, in
general, ‘unmarked’ for semantic relationships, such as temporality or cau-
sality, which have to be inferred from the context. Kammensjö also points to
the pivotal role of the active participle in the EA verbal system in general and
the area of circumstantial clause combining in particular.
Persson concludes, from her analysis of Damascene Arabic, that there are
no syntactic grounds for the establishment of a class of “circumstantial claus-
es” on a par with, for example, “conditional clauses” and “relative clauses”.
She finds only one clause type which combines form and function to such an
extent that it can be said to be used specifically to encode a circumstantial
meaning. Rather, she points to the discovery of gram switching which consti-
tutes an overarching system of minimally marked non-main clause linking
within which the clause combinations subsumed under headings such as “cir-
cumstantial” or “ḥāl” constitute an integrated part.
Marmorstein uses a large body of Classical Arabic prose, composed or
compiled by the end of the 10th century A.D to discuss the whole semanto-
syntactic domain of event integration and complex predications to which
verbal circumstantial clauses belong. She discusses the entire range of com-
plex predications: from closely integrated (and grammaticalized) verbal com-
plexes, via syndetic circumstantial clauses, to textual units consisting of mu-
tually-dependent, setting and presentative clauses. She concludes that these
constructions, despite the apparent heterogeneity of the group, are intrinsically
related by the presence of the same set of predicative verbal forms: yafʿalu,
the participle and qad faʿala, marking an ongoing situation, a state and an
outcome, respectively. All three are co-temporal, either simultaneous or coin-
cidental with the time frame set in the matrix clause. Her survey also sheds
new light on the use of the Arabic verb forms as such in the classical texts she
surveys.
Isaksson examines the use of the three basic finite verbal grammatical
morphemes (grams) in Standard Biblical Hebrew (SBH) and archaic Hebrew
poetry: the suffix verb (Vsuff), the short prefix verb (VprefS) and the long
prefix verb (VprefL), and how they behave in clause combining. The starting
point of his survey is that the verbal system can only fully be understood
when seen in the light of how clauses are linked together. Isaksson demon-
strates, with numerous examples, how the discourse function of a digression

10 Editors’ preface
from the main line of a text is often encoded in the syntax by a shift of grams.
He also establishes that the conjunction we/wa is used in both hypotactic and
paratactic clause combining to mark a clause as an addition (accompanying
action) in relation to a preceding clause. The main part of Isakson's study is
devoted to a systematic demonstration of the use of each gram under study,
the types of clause combining for which each of them is used, and the dia-
chronic development that each of them has undergone. Based on this, Isaksson
is able to establish that the wa-yiqtol syntagm is not a conjugation of its own.
There is, he concludes, only one VprefS gram, and this gram may occur in
three types of affirmative clauses: Ø-VprefS (modal or indicative), we-VprefS
(modal/purposive), and wa-VprefS (indicative).
Cohen examines circumstantial expressions at text and sentence level in
the Jewish Neo-Aramaic Dialect of Zakho spoken in northern Iraq using a
corpus of folk literature. He finds the circumstantial expressions at sentence
level to be rather uncomplicated in that they all take the same syntactic slot
and may interchange with each other (see table). At text level, different expo-
nents are found expressing circumstantiality. The examples discussed are all
opposed to the preterites which stand for narrative events and make up the
main line of the story whereas the circumstantial expressions constitute off-
line information.
Retsö puts Sabaean circumstantial clause combining (CCC), and indeed the
whole Sabaean verbal system, in a typological Semitic perspective. Starting
with an overview and discussion on previous research on the Sabaean verbal
system, he proceeds to discuss the relevant semantic relations involved when
considering CCC, relating these to the types presented by Dixon (2009) and
presenting the morphosyntactic realizations of each of this in the Arabiyya.
Retsö’s discussion of the Sabaean verbal system in the light of other Semitic
verbal systems, and not the least a comparison with Ugaritic on one hand, and
some varieties of modern spoken Arabic on the other, leads him to the conclu-
sion that Sabaean stands out as a fourth type of organization of the verbal
system alongside the Akkadian, the Northwest-Central Semitic, and the Ethio-
semitic systems. Retsö then proceeds to exemplify the use of the Sabaean verb
forms and, specifically, the switch between Vsuff and Vpref and the possible
semantic significance of these switches. His examples are taken from three
texts representing Old Sabaean, Middle Sabaean, and Late Sabaean respec-
tively and he is able to discern, in these texts, a possible diachronic develop-

Editors’ preface 11
ment from a lack of formal distinction between main-line and off-line clauses
to the use, at least to some degree, of special marking of background clauses.
Cohen concludes the volume by a survey of the strategies which are used
to express circumstantiality in Old Babylonian Akkadian. Within the clause he
finds three strategies: nominals marked by -iššī- and -ūt-, but also an infinitive
construction with ina. Circumstantial clause combining seems, however, to be
more common than circumstantial expressions within the clause. Here Cohen
is able to add to the traditionally recognized stative form also the non-verbal
clause in circumstantial use as well as a number of predicative forms (UL
IPRUS, ADĪNI UL IPARRAS and LIPRUS). The CCC paradigm is further compared
to the background forms in the epic narrative and with the group of forms
which occur in the paratactic conditional pattern.
This book, as stated above, comprises the results of a research project. Bo
Isaksson, in his capacity of initiator and project leader, has directed the pro-
cess leading to the research results that are presented here. As for the practical
work with editing this volume, it has been a joint effort.
Uppsala and Lund in February 2015.

Clause Combining in Arabic dialects

Circumstantial Clause Linking in Egyptian Arabic
Narration
Heléne Kammensjö, Uppsala University
1. Introduction
1.1 Aim and frame of reference
This paper will examine and describe types of circumstantial clause linking in
a corpus of spoken Egyptian Arabic narratives. It is the continuation of an
earlier study of circumstantial clauses in modern Arabic fiction prose which
formed part of a research project involving more Semitic varieties, see Cir-
cumstantial qualifiers in Semitic (Kammensjö 2009; Isaksson et al. 2009). In
order to identify circumstantial clauses in the Egyptian Arabic (EA) corpus,
the same semantic approach was used as in the above-mentioned project
(Isaksson et al. 2009). That is, “instead of examining a predefined set of syn-
tactic structures, we have approached our databases in search of any part of
speech that functions as a circumstantial qualifier” (Persson 2009, 207). Such
an approach is beneficial because it allows a more holistic view on hypotactic
clause linking in Arabic and avoids the trap of simply reiterating details of the
so called ḥāl category in prescriptive grammar.
The circumstantial clauses will be treated in their capacity to combine with
other clauses to form larger units in the flow of discourse. The fruitfulness of
such an approach – especially for spoken language – has been pointed out by
Halliday (1994; 2004), Matthiessen and Thompson (1988), Fleischman (1985;
1990) and others. As Halliday (1994, 224) puts it:
The clause complex is of particular interest in spoken language, be-

cause it represents the dynamic potential of the system – the ability to
‘choreograph’ very long and intricate patterns of semantic movement
while maintaining a continuous flow of discourse that is coherent
without being constructional.

16 Heléne Kammensjö
In their seminal article, dedicated to the issue of hypotaxis as a discourse

phenomenon, Matthiessen and Thompson (1988, 303) make the claim that
“clause combination represents rhetorical organization”, meaning that a major
function of a clause combination is “to reflect the scoping and structuring of a
rhetorical unit in a text”. Rhetorical units consist of a nucleus and a satellite
and since the nucleus and the satellite are rhetorically distinct from one anoth-
er, they can be expected to be grammatically distinct as well, and their asym-
metrical relationship to be grammatically marked. Semantically, the hypotac-
tic enhancing clauses are related “circumstantially” to their nucleus, typically
fulfilling one of the following functions: temporal, conditional, reason, con-
cessive, purpose, means or manner (Matthiessen and Thompson 1988, 277,
283-84).1
EA circumstantial clauses are usually subtly marked, most often through
interlacing (sharing semantic elements) and gram switching (verb form con-
trasts), sometimes with the addition of a general additive connective, wa
‘and’, followed by an anaphoric pronoun referring to a participant in the main
clause. Interlacing and gram switching, however, are general features of hypo-
tactic linking, not markers of semantic relationships, such as temporality or
causality. This means that EA circumstantial clauses are ‘unmarked’ for such
relationships, which have to be inferred from the context. EA circumstantial
clauses, thus, represent the ‘unmarked’ subset of this category which also
includes ‘marked’ enhancing clauses (Isaksson, 2009, 3f; Persson 2011).
Marked enhancing clauses are typically introduced by a subordinative con-
junction to signal their precise relationship to the head clause, i.e. clauses
introduced by lamma ‘when’ or ʿašān ‘because; in order to’ (EA).
Fleischman’s work (1985) brings verb morphology into the study of clause
linking, as a marker of head-satellite relationships, stating:
The function of tense forms in narrative is frequently not the basic

tense function of temporal reference, which in most narrative forms is
1 Enhancing, as a concept, was first used by Halliday as part of his taxonomy for the
expansion of the clause, where it is defined in the following way: “In enhancement
one clause (or sub-complex) enhances the meaning of another by qualifying it in
one of a number of possible ways: by reference to time, place, manner, cause or
condition” (Halldiay 2004, 410). In traditional grammar such clauses are termed
adverbial clauses, a term avoided here since it suggests constituency in the main
clause.

Circumstantial Clause Linking in Egyptian Arabic Narration 17
established a priori as past. Rather, tense contrasts may be pressed into

pragmatic service in the organization of narrative discourse.
She shows how tense switching, in particular the use of present tense in past
narration, functions in Old French as a strategy for ‘narrative subordination’,
or ‘grounding’. It can be shown that Fleischman’s observations of Old French
have parallels in Semitic languages (Isaksson 2009, 39-46). So also in Egyp-
tian Arabic narrative, where shifts of verb form across clause junctures are
legion and become particularly conspicuous in asyndetic clause linking. In
order to allow the whole of predicate morphology into the analysis of the form
switch between a circumstantial clause and its main clause, Fleischman’s term
“tense switch” has been replaced by gram switch in the present volume. The
word gram, borrowed from Bybee et al. (1994, 2) is used in this study for any
Arabic verb form, as well as for the zero morpheme of a ‘verbless’ nominal or
participle clause.
1.2 Data and transcription

The corpus comprises transcribed oral data from three sources:
(i) ÄAD = All the narratives and a few procedural texts in Behnstedt and
Woidich (1985-1994). These texts are transcriptions of audio-recordings
of Egyptian Arabic made in 1976-1985 and represent the language spoken
by the rural population in different areas of Egypt.
(ii) Gha = Ca. 100 pages of Ghada Abdel Aal’s blog book, ana ayza
atgawwaz (I want to get married) (Abdel Aal 2007).
(iii) TAK = A number of short, unpublished, Cairene anecdotes, audio-
recorded and transcribed by Professor Manfred Woidich who generously
lent me both recordings and transcriptions.
The bulk of the data, around 700 clause units, have been elicited from the
published transcriptions of Behnstedt and Woidich’s Sprachatlas, which,
however, only represent rural Egyptian usage, some of it recorded already in
the 1970s. In order to provide a more contemporary and urban frame of refer-
ence, data sources (ii) and (iii) have been added, comprising around 470
clause units.
The transcription follows that of the original although Behnstedt and
Woidich use a more allophonic transcription than needed for the present

study. Even punctuation marks have been represented as in the original, since
it can be assumed that they mirror real prosodic features in the narration. The
original audio-recordings are no longer available. The transliteration of Gha-
da’s blog book, on the other hand, follows standard phonemic conventions for
transcribing spoken Arabic.
1.3 Relevant features of spoken Egyptian Arabic

Spoken language obviously differs from written language as a result of a dif-
ference in production conditions. Even if spoken language may have its own
‘literary’ conventions, may be planned beforehand for a performance, its main
characteristic is that it is produced in the moment, with little time for planning
ahead or editing. On the other hand it can afford repetitions and self-
amendment as speakers strive to adapt to the situation and their conversation
partner. If discourse is both a flow and a hierarchy, the flow aspect becomes
so much more conspicuous in spoken language. Spoken language needs expe-
dient strategies in order to make itself clear with as little effort as possible.
This may imply leaving implicit whatever can be inferred from context and
making use of the principle of adjacency, as a clue to the relationship between
clauses. In spoken Arabic there are few connectives as markers of discourse
hierarchy and semantic relationships between parts of utterances, with the
exception of wa, which is additive but not necessarily coordinating, as can be
seen in example (1) below, where the switch from the suffix form in the first
clause to participles and a prefix form in the clauses that follow suggests a
non-main clause reading of subsequent clauses.
(1) gaʿadt aʿayyaṭ yāma baga, wu miš ʿārif amašši wu ggamūsa tintišni wu
xayfa ʿawza tmašši. (ÄAD 23:14)
sit/continueVs1s SCreamVp1ps so much (exclam particle) then, wa NEG know-
PAms walkVp1s wa DEF-buffalow (fem) snapVp3fs-PRON1s wa fearPAfs
wantPAfs walk.awayVp3fs
‘I continued to scream a lot then, not being able to walk away while the
buffalo cow snapped at me, she was afraid and wanted to get away’
In particular, there are four features in EA narration that render the work of
identifying circumstantial clauses particularly difficult, (i) that asyndesis is
such a ‘normal’ and widespread type of clause linking; (ii) that stories are
often told, using prefix conjugation to describe habits or procedure, or simply
to bring narration closer to the moment of speech (‘dramatic present’); (iii)

that the present participle plays such an active role in the EA verb system
where it functions as the ‘unmarked’ alternative to the tensed forms of the
verb; and (iv) that some main clause verbs frequently involved in asyndetic
clause linking, are subject to an auxiliation process where they become dis-
course markers or void tense ‘holders’ as a result of grammaticalization and
semantic ‘bleaching’ or loss of semantic content (Hopper and Traugott 1992;
2003). These are predominantly general verbs of motion and posture, such as
ʾām, ‘rise; set out’; rāḥ ‘go’; ‘get going’, gih ‘come; set out’ and ʾaʿad ‘sit’;
‘keep doing’.
The obvious consequence of features (i) and (ii) is that two of the most
common traits of circumstantial clause linking, i.e. asyndesis and verbal gram
contrast, become neutralized. One of the most conspicuous consequences of
feature (iii) is an increased number of verbal gram distinctions in the EA verb
system as compared to Modern Written Arabic, cf. participles in example (1).
Feature (iv) results in many borderline cases, where it becomes a question of
interpretation whether the main verb has in fact become ‘bleached’ of its se-
mantic content or not. When the grammaticalization process is complete, we
are no longer talking of two clauses but one, which contains a composite verb
phrase, sometimes containing more than two verbs, as in example (2). The
effect of ʾām ‘get up; get going’ is similar to that of the connective phrase
(and) then in English.
(2) ʾumtı̊ gayyı̊ mṛawwaḥ (ÄAD 8:82)
get.upVs1s comePAms go.homePAms
‘and then I went home’
1.4 Notes on the EA participle and the prefix forms

The participle, time and aspect
Reflecting over the fact that the Arabic active participle (henceforth just ‘par-
ticiple’) is so common in dialectal Arabic narratives but relatively rare in
genres of written Arabic discourse, Brustad concludes that it must be “more
strongly associated with the speech act than other verb forms” and this is
because “narratives are often present in a way that emphasizes the relevance
to the moment of speaking” (Brustad 2000, 182).2
2 The EA passive participle has similar functions as the active participle but is far less
common. Besides, most derived verb stems in EA have only one participle form which
is formed through addition of mi- to the stem of the prefix conjugation, e.g. mbayyaḍ

The participle clause has mainly a verbal function in spoken Arabic, much
like present participle clauses in English, cf. she is coming or I saw her com-
ing. In EA, the participle clause appears mostly in its ‘non-past’ form, that is
without the copula kān ‘to be’, e.g. (hu) gāy ‘he’s coming’. In this form, (i.e.
when not preceded by a tensed copula verb) it behaves rather ‘chameleon-
like’ when it comes to time reference; it appears to be able to replace a main
verb of any tense (past, present, or future), cf. ǧāy ‘come’ as a main verb with
past time reference in example (3) below. This means that the EA predicative
participle forms part of the verb system, without being a tense in its own right,
i.e. it does not possess inherent time reference (Brustad 2000, 162-64).
(3) fa baʿad ma bāʿu ǧǧiḥš, ǧāy igabbḍ ilfilūs gallhum: (ÄAD 89:8)
And.then after sellVs3mp DEF-donkey, comePAms grabVp3ms DEF-money
sayVs3ms-PREP-PRON3mp
‘and then after they had sold the donkey, he came to grab the money (and)
said to them:’3
The high frequency of the predicative active participle in modern Arabic dia-
lects suggests a process of verbicization at work. Speakers keep choosing this
nominal derivative, which (along with the unmarked prefix form) is felt to be
the ‘neutral’, more convenient alternative once the time frame has been set by
the context. It is not difficult to imagine that the lack of copula in participle
clauses may have triggered the reanalysis in spoken Arabic as part of a gram-
maticalization process. Even when it is preceded by the past tense copula, kān
‘to be’, it looks no different than other compound tense constructions with
kān, cf. examples (4) and (5) below.4
(4) il-makana kānat šaġġāla billēl (EAD 102:3b)
DEF-engine (f) beVs3fs workPAfs PREP-DEF-night
(stem II), cf. the following examples from Cairene: iḥna mbayyaḍīn il-ʾōḍa ʾrayyib ‘we
have whitewashed the room recently’; il-ʾōḍa mbayyaḍa ‘the room is whitewashed/has
been whitewashed’ (Woidich 2006, 83-86).
3 In its grammaticalized meaning, Cairene gih may signify ‘to begin to do’ or to ‘want to
do’, of which the latter meaning has been chosen in the German translation “… da
wollte er das Geld in Empfang nehmen” (Behnstedt and Woidich, vol. 3II, 1988, 181.
See also Woidich (2006, 331f).
4 Brustad (2000, 163) provides an example from a Syrian dialect where the participle has
developed a personal conjugation for the second-person feminine which resembles the
perfect ti-suffix of the perfect, e.g., šāyəftī la-ʾaḥmad? ‘have you (sg. f.) seen Ahmed?’.

‘the engine used to be running at night’

(5) il-makana kānat taštaġilu bi-l-layli (MWA or higher style EA)
DEF-engine (f) beVs3fs workVp3fs PREP-DEF-night
‘the engine used to be running at night’
Schlonsky (1997, 25-56) makes a similar point concerning the verbicization of

the participle in Modern Hebrew, where the participle (Benoni) frequently
functions as the main verb encoding present tense, e.g. dani kotev sipurim
‘Dani writes/is writing stories’. Schlonsky interprets this construction syntac-
tically as a compound tense construction consisting of a ‘null’ auxiliary plus
participle, saying that “the null auxiliary is an incorporation trigger” which
raises the participle to the highest clausal level, where it may function as a
fully tensed verb (Schlonsky 1997, 25-57).5
To sum up, the participle in EA (and other Arabic dialects) frequently
functions as the main verb of a sentence with past, present, or future time
reference. This makes it part of the EA verb system, but precludes it from
being a tense form in its own right, i.e. it does not have inherent time refer-
ence (Brustad 2000, 162-64).
The aspectual value of the participle in spoken Arabic has been discussed
by many (Wild 1964; Woidich 1975; Mitchell 1978; Eisele 1990b; 1999;
2006; Brustad 2000, 182ff; Eades and Persson 2013). It will not be possible
within the scope of this article to make a definite stand for or against the pro-
posed aspectual nature of the Arabic verb system. In her research review of
these matters, covering Arabic dialects from Morocco, Egypt, Syria and Ku-
5 See also Gordon (1982) for a description of the process of verbicization of the Hebrew
participle. In Late Biblical Hebrew it is already possible to discern a reanalysis of the
participle towards being an inherently time-referenced verb; in Mishnaic Hebrew the
participle functions (with very few exceptions) as the present tense, while it continues
to exist in its nominal capacity. In Modern Hebrew the verbal uses of the participle are
clearly distinct from its nominal uses, and its present time reference is mutually exclu-
sive to other time references. In Neo-Aramaic the development has advanced even fur-
ther, to the point where the active and passive participles have completely replaced the
old Semitic prefix and suffix conjugations as tense forms and become fully inflected for
person. The inflectional suffixes are shortened forms of the independent personal pro-
nouns. Note also that there are peripheral modern Arabic dialects that have developed a
personal conjugation for the verbal participle, for instance the Arabic vernacular of
Boukhara in Uzbekistan, possibly under influence of Aramaic (Cohen 1984, 270-334;
Jastrow 1997, 360-67).

wait, Brustad concludes that “the trend in more recent studies has been to
view the verb system of spoken Arabic as combining aspect and time refer-
ence” (Brustad 2000, 203; Eisele 1990a; 1990b; 1999; Ingham 1994).6
It is obvious that the time reference of the EA participle is relative to the
moment of speaking, or to the ‘now’ or ‘then’ of the story. However, the in-
terpretation of its temporal/aspectual function does not solely depend on the
context but also on the lexical aspect (Aktionsart) of the verb itself. 7 Partici-
ples of some verbs, predominantly action verbs, express resultant states and
have anterior time reference while others, predominantly motion verbs and
stative verbs are given a co-temporal progressive or habitual reading.
To come to grips with the elusive functioning of the EA participle it will
thus be necessary to define the interplay between the formal and lexical as-
pects (or Aktionsart) in the variety, as suggested by Eisele and others. Eisele
proposes a rather complex taxonomy of lexical aspect for Cairene Arabic
(Eisele 1999, 214-252, 254; 2007, 198). 8 Brustad’s division, adapted from
Ingham, is much simpler, using a main division into: state/motion verbs as
opposed to action verbs further divided into telic and atelic processes (pro-
cesses with punctual ending vs open-ended processes) (Brustad 2000, 170-
172, 202; Ingham 1994, 89-90).
Lexical aspect and the participle
Telic Atelic
State/Motion Resultant state Progressive
Action Resultant state Progressive
Chart 1: Brustad’s simplified model (2000, 171)
6 The same view has been put forward concerning Koranic Arabic by Kinberg (1992) and
for Classical Arabic by Comrie (1976, 78-81). Comrie concludes, on the basis of Clas-
sical Arabic examples taken from Wright (1898) that the difference between the two
conjugations is one of aspect and relative tense intertwined. The perfective (suffix con-
jugation) indicates perfective meaning and relative past time reference, while the imper-
fective indicates everything else, even a relative past tense in subordinate clauses, as in
wa-ttabaʿu mā tatlū l-šayāṭīnu ʿala mulki sulaymāna ‘and they followed what the de-
mons used to recite in Solomons reign’.
7 Cf. Comrie (1976, 6, n. 4).
8 Eisele’s taxonomy of lexical aspect was based on that of Dowty and Vendler from
1979, which in turn was inspired by Aristotelian categories (Eisele 2007, 197).

According to Brustad’s model (Chart 1), telic verbs in participle form express
a resultant state (perfect formal aspect), while the participle of atelic verbs has
a progressive reading. Below is an example from the EA data containing a
telic action participle, fātiḥ ‘open’ having perfect aspect and anterior time
reference followed by two participles, qāʿid mistannīni having atelic lexical
aspect, thus generating a progressive reading. The second participle may or
may not be taken as an auxilized aspectualizer, in which case it has lost its
original meaning ‘sit’ and only signals progressivity, ‘keep doing something’.
(6) wu ṛawwaḥtı̊ laqát axūya fātiḥ iddukkān wu qāʿid mistannīni (ÄAD
36:28)
wa go.homeVs1s findVs1s brother-PRON1s openPAms DEF-shop wa sitPAms
waitPAms-PRON1s
‘I went home and found my brother having opened the shop and (sitting)
waiting for me’
Ingham recognizes that some verbs are multivalent, in that they may be both
telic (punctual end) and atelic (open-ended action), or inceptive (punctual
start) and non-inceptive. The verb read in read a book is telic while read in a
book becomes atelic because of the preposition. In Arabic the verb nām
‘sleep’ may be either inceptive, i.e., ‘to fall asleep’ or non-inceptive (and atel-
ic), i.e., ‘sleep’. The same is valid for ʾaʿad ‘sit down’ or ‘be sitting’ and ʿirif
‘get to know’ or ‘know’. Participle aspects may also vary across dialects.
While ḏākar ‘study, review a lesson’ has an atelic meaning in Najdi (Ingham
1994, 91), it may be either telic or atelic in Egyptian. When telic, it has a
perfect (resultative) interpretation: ʾana mizākir id-dars ‘I have studied the
lesson’ (Brustad 2000, 169).
Unmarked and marked prefix forms

The ‘unmarked’ prefix form (without bi- or ḥa-) is used for the general con-
tinuative or habitual situations in the past and the future (Eisele 1999;
Woidich 2006, 273-77). Woidich further reports that the unmarked prefix
form can replace the marked prefix forms introduced by bi- or ḥa-, which
means that both bi- form and the unmarked prefix form function well in narra-
tive discourse. Apart from its temporal uses, the unmarked prefix form ex-
presses modal meanings, such as directive, optative, potential, prospective,
obligative or desiderative meaning (Woidich 2006, 275-77).

Woidich describes the meaning of the bi-form of the prefix conjugation

(bi-imperfect in his terminology) in non-dependent clauses as the function of
the lexical meaning of the verb (Woidich 2006, 280-82). For verbs with non-
stative and non-repetitive meaning, the bi-form expresses the present of hic et
nunc, e.g., inti lmaṛṛādi btikallimi ṣaḥḥ ‘this time you (f) speak the truth’. For
stative or ingressive verbs, including verbs of motion and perception, the
active participle is used in this function instead, e.g., ana šāyif ‘I see’. For all
lexical verb types the bi-form is used to express repeated, habitual or lasting
actions or states, e.g. ikkahṛabaʾiyya btikkallim kida ‘electricians (usually)
talk this way’.
2. Types of circumstantial clause linking

2.1 Syndetic and asyndetic linking as division
This section lists and comments on the structural types of circumstantial
clauses (CC) in the EA corpus, the prototypical together with the less clear-
cut. The data has been sub-divided into two main categories, syndetic clauses
(syndetically joined clauses), appended by means of the general conjunction
wa, on the one hand, and asyndetic clauses (asyndetically joined clauses), i.e.
conjunction-less clauses, on the other. The latter category also includes a sub-
category of desententialized clauses, i.e., participles, adjectives and other
substantive nouns that in descriptions of written Arabic are usually catego-
rized as nominal. In EA, however, where participles with verbal function may
replace finite verbs in most positions, it is not meaningful to categorize them
as nominal phrases. The distinction syndesis/asyndesis is relevant since it also
represents an important division between two structural types with differing
degrees of dependency on and integration into the main clause. Woidich,
acknowledges this tighter clause connection of asyndetic clauses, saying that
they function as predicates of the subject/object in the main clause (Woidich
1991; 2006, 397f).
Asyndetic hypotactic linking was found to be more than twice as common
as syndetic hypotactic linking in the data (800 to 380 instances of circumstan-
tial clause linking). Neither the syndetic or asyndetic circumstantial clauses
are unique form classes. Non-definite relative clauses are linked asyndetically.
Not even the more clearly marked syndetic circumstantial clause lacks com-
petitors. Clauses beginning with a connective wa plus a thematic noun or

pronoun are generally rather frequent in EA, e.g. in non-narrative passages

where topicalized clauses form the second half of binary complementary
statements. In example (7), the two bracketed clauses could be interpreted as
of a lower rank, i.e. as circumstantial hypotactic clauses. With another read-
ing, however, they may be thought of as complementary paratactic coordina-
tions, which is the choice of the German translation.9
(7) iwhib nafsak liyya [w-ana niwhib lik], wu tibʾa inte gōzi, [wu ana
miṛātak] (ÄAD 26:34)10
giveIMPms-PRON2ms PREP-PRON1s [wa PRON1s giveVp1s PREP-PRON2ms]
wa becomeVp2ms PRON2ms husband-PRON1s, [wa PRON1s wife-PRON2ms]
‘give yourself to me, [as/and I give myself to you], and you become my
husband, [as/and I (will be) your wife]’
2.2 Types of syndetic clauses

Below, the syndetically linked circumstantial clauses will be examined more
closely according to syntactic type, using the predicate as a point of departure.
2.2.1 The prefix form in the circumstantial clause

Around one third of the examples of syndetic clause linking in the corpus
have prefix verb forms in the circumstantial clause, i.e., either the basic un-
marked form, or the marked form, bi-, ḥa- and their variants, as exemplified in
(8) to (10) below. Woidich also observes that Cairene circumstantial clause
verbs may appear in the marked form of prefix conjugation (1991; 2006, 397).
(8) wi onkil disku mitʿalliʾ fi bāba bibūs rās-u [wi bāba yiʾūl-l-u: “iš-šayṭā
šāṭir yā ḥāǧǧ Disco”] (GhA 73:5b, Cairo)
wa Uncle Disco clingPAms PREP-Dad ASP-kissVp3ms 11 head-PRON3ms [wa
Dad sayVpms]-PREP-PRON3ms: DEF-Satan clever VOC-Hajj Disco]
‘and Uncle Disco would be clinging to Dad kissing him on the head [while
Dad tells him: “The Devil is a shrewd one, Ḥaǧǧ Disco”]’
9 “Sie sagte zu ihm: Schenk dich mir, und ich schenk mich dir, und dann bist du
mein Mann und ich bin deine Frau” (Behnstedt and Woidich 1987, 189).
10 This example comes from Alexandria, where the verbal prefixes in first person both
singular and plural is ni- as in niktib ‘I write’; niktibu ‘we write’. Note the suffix –u to
distinguish the plural.
11 ASP = aspectualizer.

(9) laGenāhum [wu huṃṃa biyliffu sagāyir ē? wu byišṛábu ḥašīš] (ÄAD

28:39)
findVS1P-PRON3mp [wa PRON3mp ASP-roll-Vp3mp cigarettes INT wa ASP-
smoke-Vp3mp hashish]12
‘we met them [as they were rolling cigarettes and smoking hashish]’
(10) ya ʿAbd ilʿAl, inta nāyim [w-iḥna hanitfíḍiḥ filbilād] (ÄAD 68:69)
VOC-ʿAbd ilʿAl, PRON2ms sleepPAms [wa PRON1p be.shamedFUT-Vp1p
PREP-DEF-land]
‘ʿAbd ilʿAl, you sleep [while we are going to be shamed in the land]’
The bi-marked prefix form appears to be the preferred choice for marked
simultaneous time reference, cf. example (9).
Belonging to a ‘grey zone’ are some clauses showing gram contrast but
lacking the normally preposed anaphoric pronoun, cf. example (11).
(11) ḥaṭṭaha fkitfu, [w ygūl ēā?] (ÄAD 87:121, Upper Egypt)
putVs3ms-PRON3fs PREP-shoulder-PRON3ms, [wa sayVp3ms INT]
‘he put it (the sack) on his shoulder, [saying what?]’
2.2.2 Participle or nominal clause as circumstantial clause

Taken together these two related clause types represent around 60 per cent of
the total number of syndetic clause linking, the vast majority of them having a
suffix form verb in the main clause, cf. example (12).
(12) istannū fiṭṭarīg [wa huwwa māši wu ǧārir il-ḥumār] (ÄAD 75:4)
waitVs3mp PREP-DEF-way [wa PRON3ms walkPAms wa pullPAms DEF-
donkey]
‘they waited for him along the way [as he was walking and pulling the
donkey]’
The order of clauses may be reversed, so that the circumstantial clause comes
first in the sentence, cf. example (13).13 Or it may be embedded in the main
12 Note that the interrogative particle ēh ‘what’ is used rhetorically and as an ‘empty filler’
in EA oral discourse.
13 Rosenhouse (1978, 229) suggests that the high frequency of CQs (circumstantial quali-
fiers) preceding their heads has led to that “the division of emphasis/semantic weight
between heads and CQs has shifted in favour of the CQs in modern dialects as com-
pared to classical Arabic” (cf. Persson 2009, 275-77). Note that the reversed word order
has also been reported by Woidich (2006, 395).

clause, as in (14), in which case it must be interpreted as operating at the

phrase level, as relative clauses do when embedded in the noun phrase.
(13) [wu huṃṃa mašyīn] ʾāmu liʾyu gneyh marmi filʾarḍ (ÄAD 53:38)
[wa PRON3mp walkPAmp] do.suddenlyVs3mp findVs3mp Pound throwPPms
PREP-DEF-ground
‘[as they were walking] they suddenly found a Pound thrown on the
ground’
(14) onkil disku da [w-ism-u l-ʾaṣli disūʾi] ṣāḥib bāba min zamān (GhA
72:11)
Uncle Disco-DEMms [wa name-PRON3ms DEF-originalADJms Disū’i] friend-
Dad PREP-time
‘this Uncle Disco [his real name is Disouqiyy] has been Dad's friend for
ages’
In example (15), word order has been reversed in a way that the subject pro-
noun precedes the connective, a feature belonging to the dialect area of Upper
Egyptian, according to Behnstedt and Woidich (1988, vol. 3II, 320, n. 4).
Note also that the head clause is a circumstantial clause on a higher rank.
Circumstantial clauses can thus be nested so that they serve both as non-main
and main clause in the same clause complex.
(15) ṛāḥu yirmaḥu, hūwa yṣawwir fīhum [huṃṃa w ǵaryīn] (ÄAD 111:10)
set.offVs3mp runVp3mp, PRON3ms photographVp3ms PREP-PRON2mp
[PRON2mp wa runPAmp]
‘they set off running, while he was taking photos of them [as they were
running]’
There are some cases lacking a preposed noun or pronoun, cf. example
(16).
(16) ana baʾēt māši filṃayye [wu xāyif niṭlaḥ fōG]14 (ÄAD 27.2:31)
PRON1s beginAUXs1s walkPAms PREP-DEF-water [wa fearPAms as-
cendVp1p15above]
‘I began walking in the water [fearing to ascend]’
14 [G] is sometimes used by Behnstedt and Woidich for the phoneme /q/ to transcribe
speech from Alexandria and Rosetta where both /q/ and /ǧ/ are sometimes realized as
[g], the latter being a Cairene influence (Behnstedt and Woidich 1987, vol. 3I, 177-78).
15 1p used for 1s here.

The circumstantial clause in (16) has a causal interpretation, supported by the

context. It could of course also be thought of as coordinated to the participle
in the main clause, māši, in which case it too refers back to the auxiliary baʾa
with a less likely translation as a result: *‘I began walking in the water and
was afraid to ascend’.
2.2.3 The suffix form in the circumstantial clause

A few circumstantial clauses have been found with their verb in the suffix
conjugation, such as example (17) below. Note that it is not entirely clear
whether the general action verb ʿamal ‘do’, in this example, is inceptive in
meaning or a simultaneous act.
(17) atābih Bibbi indāṛa lilġarb, maʿa lfalāyik, [wu rrīḥ ʿamal ʿalī šaṛGi,
yiḥallílu ʿalʿAlamēn] (ÄAD 27.3:10)16
oh no Bibbi turnVs3ms PREP-DEF-west, PREP-DEF-feluccas, [wa DEF-wind
work/moveVsms PREP-PRON3ms easternADJ, driveVp3ms-PRON3ms PREP-
ʿAlamēn]
‘oh no, Beppi moved to the west, with the feluccas, [since the wind pushed
him from the east, driving him towards ʿAlamēn]’
2.2.4 Unusual head types in syndetic circumstantial clauses

Note that the head of the circumstantial clause may be a noun phrase, as in
(18), or a prepositional phrase, as in (19). The circumstantial clause may also
serve as a kind of conclusion after a complex of clauses, as in (20), or it may
lack a head clause altogether and instead refer back to the discourse context as
a whole, as in (21).
(18) wu šwayya [wu Bibbi da gayy] (ÄAD 27.3:16)
wa a little [wa Bibbi-DEMms comePAms]
‘after a little while [Beppi came]’
(19) … w-ana sāyabā-hu wāʾif fātiḥ buʾʾ-u zayy ʿamm ʿabd rabbu l-bawwāb
[wi huwwa bitfarrag ʿa l-musalsil il-kūri] (GhA 40:13)
wa-PRON1s leavePAfs-PRON3ms standPAms openPAms mouth-PRON3ms like
Amm Abderrabbu DEF-door keeper, [wa PRON3ms ASP-watchVp3ms PREP-
DEF-series DEF-CoreanADJms]
16 The expression atābī- with a suffix pronoun is a demonstrative “indicating the discov-
ery of a reason or a fact” (Badawi and Hinds 1986, 5). Note that there is another em-
bedded circumstantial clause in the example: [yiḥallílu] ‘driving…’.

‘…as I left him standing (there) with his mouth open like Amm Abder-
rabbu, the door keeper, [when he watches the Korean soap]’
(20) wu lina naṣīb lissa, wu ʿandína ʿiyāl ḥaṛṛabbūhum [wu ṛabbína fbitna]
(ÄAD 31:86)
Wa PREP-PRON1p share still, wa PREP-PRON1p children FUT-raiseV1p-
PRON3mp [wa Lord-PRON1p PREP-hous-PRON1p]
‘we still have a share in this life and we have children that we will bring
up [the Lord is in our house (is with us)]’
(21) (new paragraph) [w-iḥna kullı̊ -na lissa wāʾfīn ʿa-l-bāb binbuṣṣ li baʿḍ]
(GhA 33:5)
[wa PRON1p all-PRON1p still standPAmp PREP-DEF-door ASP-lookVp1p
PREP-each other]
‘…[and we were all still standing at the door looking at each other]’
2.3 Types of asyndetic clauses

In the following, types of asyndetically linked circumstantial clauses will be
presented and analysed using the predicate of the circumstantial clause as the
point of departure.
2.3.1 The unmarked prefix form in the circumstantial clause

This is the most frequent type representing more than half of all examples of
asyndetic clause linking found in the corpus. The prototypical and clear case
involves a gram switch from the suffix form to the prefix form in the non-
main clause, as in (22).17
(22) umr-aha mā nizilit is-sūʾ [tištari l-xuḍār] (GhA 29:16)
Life-PRON3fs NEG go.downVs3fs DEF-market [buyVp3fs DEF-vegetables]
‘In her entire life she has never gone down to the market [to buy vegeta-
bles]’
2.3.2 The marked prefix form in the circumstantial clause

In the corpus the bi- form (and ḥa-) is common after the verb find or percep-
tion verbs, as in example (23).
17 The terms used here, main and non-main clause have been applied by Dixon for the
syntactic classification of clauses in his book The semantics of clause linking (Dixon
2009, 4).

(23) wi simiʿnā-hu [biyitaḥrig daḥraga razīna giddan] (GhA 36:15)

wa hearVs1p-PRON3ms [ASP-rollVp3fs18 rollingVN heavy very]
‘we heard him [tumbling down in a very heavy tumble]’
Thus, the bi- (and ḥa-) modifiers serve to complement the ‘bare’ prefix form
in circumstantial clauses, as the marked alternative. The bi-prefix in (23)
signals simultaneous progressivity or iterativity, and the ḥa-prefix prospectivi-
ty in purpose clauses.
2.3.3 Desententialized clauses as circumstantial clauses

The group of desententialized circumstantial clauses in the data are made up
of active participles mostly, but some passive participles, adjectives and sub-
stantive nouns are also found, as exemplified in (24) to (27) below. They
constitute around one fourth of the asyndetic data. The majority of them have
the suffix form as main clause verb, often liʾa ‘find’, as in example (24).
(24) liʾa xwātu [naymīn], kúlluhum (ÄAD 11:54)
findVs3ms brothers-PRON3ms [sleepPAmp], all-PRON3mp
‘he found his brothers [sleeping], all of them’
(25) šāfu ssulūk bitūʿ ilkahṛába [maḥruqīn] (ÄAD 31:87)
seeVs3mp DEF-cables of DEF-electricity [burnPPmp]
‘they saw that the electric cables [had been burnt].’
(26) wu ragadit ḥawāli min asbūʿ [ʿayyāna] (ÄAD 73:49)
wa lieVs3fs about PREP week [illADJfs]
‘and she lay [ill] around a week’
(27) būki ʿaṭʾāki liyya [hadiyye] (ÄAD 87:164)
father-PRON2fs giveVsms-PRON2fs PREP-PRON1s [gift]
‘your father gave you to me [as a gift]’
As a desententialized constituent of the main clause, the adjunct participle
belongs to a lower clause rank in the hierarchy of the text than a full clause.
The problem of distinguishing it from the full participle clause is that the latter
may consist of only one word too, as in gāy? ‘are you coming?’ (Woidich
2006, 248). This is why gāy ʿalēni ‘coming in on us’ in (28) below, may be
18 ASP stands for aspectualizer, meaning the verb modifiying prefix bi- or its equivalent
signaling simultaneous progressive or iterative aspectual value.

interpreted either as a desententialized circumstantial clause, or as an inde-

pendent/main clause in a series of paratactically arranged participle clauses.
(28) ḥassēna bilbaḥr, [gāy ʿalēni], ilbaḥṛı̊ gāy, samʾīnu [gāy] (ÄAD 10:96)
noticeVs1p PREP-DEF-sea, [comePAms PREP-PRON1p], DEF-sea comePAms,
hearVs1p-PRON3ms [comePAms]
‘we felt/noticed the sea [coming in on us], the sea was coming, we heard it
[coming].’
Furthermore, as pointed out in section 1.2, participles mix freely with tensed
verb forms in EA narration. The latter interpretation has been the choice in the
German translation: “wir spürten die See, sie kam über uns, die Wellen ka-
men, wir hörten sie kommen” (Behnstedt and Woidich 1987, vol. 3I, 97).
When both the main clause and the circumstantial clause have participles,
some ambiguous cases arise as a result. Then clause rank will have to be in-
ferred from context, cf. example (29).
(29) ilMaḥrūs gayyı̊ [ṣāyid wu mṛawwaḥ] (ÄAD 32:29)
DEF-Maḥrūs comePAsm [fishPAms wa go.homePAms]
‘The Maḥrūs (name of the boat - author’s note) came [having been fishing
and being on his way home]’
Note that in (29), due to the differing lexical aspects of the verbs, the clauses
have different time references; ṣāyid ‘fishing’ has an anterior temporal refer-
ence (to the time of the story), resultative (perfect) aspect, whereas murawwaḥ
‘going home’ as a motion verb has simultaneous time reference and is pro-
gressive in aspect.
2.3.4 Participle and other nominal clauses as circumstantial clause

This is a minor group among the asyndetically linked clauses. Asyndetic par-
ticiple and clauses with explicit subjects are very similar to syndetic circum-
stantial clauses, giving the impression that they belong to that category but
lost their connective, cf. (30).
(30) gēna ʾaṣad Būr Saʿēd, [iḥna naymēn] (ÄAD 10:15)
comeVs1p PREP-Port Said, [PRON1p sleepPAmp]
‘we came opposite Port Said, [while we were sleeping]’
Such circumstantial clauses may precede their heads, as in (31).

(31) [huwwa dāxil], ligi iǧǧanāyni nāyim (ÄAD 73:82)

[PRON3ms enterPAms], findVs3mp DEF.garderer sleepPAms
‘[when he entered/had entered] he found the gardener sleeping’
2.3.5 The suffix form in the circumstantial clause

A suffix conjugation verb in the main clause may occasionally combine with a
suffix conjugation verb in the asyndetic non-main clause, as in (32). This
interpretation is analogous to the construction in examples (24) above, con-
taining the liʾa ‘find’ plus a participle as the circumstantial clauses. In the
example, the suffix form has anterior time reference and resultative (perfect)
aspect, referring to what happened before the ‘now’ of the story, while it still
has a bearing on it. This hypotactic clause may also be interpreted as an asyn-
detic object complement clause, not unusual in EA (Woidich 2006, 392).
(32) laʾēt-u [nizil maʿāya fi nafs il-maḥaṭṭa] (GhA 24:7)
findVs1m-PRON3ms [get.downVs3ms PREP-PRON1s PREP-same DEF-station]
‘I found him [having got off with me at the same stop]’
3. Significant gram switching

In Chart 2 below, the 800 combinations of asyndetic clause linking included
as data have been presented in order to visualize what combinations are tech-
nically possible and which ones are in fact exploited. No exact figures for
different categories have been given, since there are far too many ambiguous
and intermediate examples in the database to justify such a treatment. Only
rough proportions can be provided, as indicated by the signs (+), (++) and
minus (-), see key of symbols below. The potential for different predicate
combinations, as a result of clause linking, is striking. The number of choices
and possible combinations and gram-switches, appears to be greater in EA
than in written Arabic, which is only natural in view of the fact that the parti-
ciple in written Arabic plays a less prominent role and that its prefix form has
no marked alternative for non-past time reference. Note that main clause verbs
of the suffix form attract desentential clauses (usually participles) more often
than do main clause verbs of the prefix form. Switches from a suffix form to a
prefix form or to a nominal participle, are all in all the most frequent gram
switch.

Non-main
clause
Main Vsuff Vpref pVpref Imp PCl NCl Pa et al.
clause
Vsuff + ++ + - + + ++
Vpref (+) ++ (+) - + (+) +
pVpref - + + - (+) - +
Imp - + (+) (+) (+) - (+)
PCl (+) + (+) - (+) (+) +
NCl (+) + + - (+) + +
Pa et al. - + (+) - (+) (+) +
Chart 2: Clause type combinations in asyndetic circumstantial clause linking

in EA corpus
Abbreviations and symbols

Vsuff Suffix conjugation
Vpref Prefix conjugation
pVpref Marked prefix conjugation19
Imp Imperative
PCl Participle clause20
NCl Nominal clause
19 This category is composed of what is known as the present or progressive bi-imperfect

with its isoglottic variants in rural Egyptian Arabic. It also contains verbs of the prefix
conjugation introduced by ḥa- for the future. For the dialectal distribution of the prefix
verb modifiers, see Behnstedt and Woidich (1985-1994, vol. 2, maps 219-225); see also
Fischer and Jastrow (1980, 227). Woidich (2006, 282): “ʿammāl, meist mit folgenden y-
Imperfekt, druckt Dauer und Intensität individueller Sachverhalte aus. Die Zeitlage ist
aktuelle Gegenwart. Es kongruiert… Auch mit bi-Imperfekt oder Partizip kommt es
vor, ʿammāla biybuṣṣilh, ʿammāl wāʾif.
20 Participle clauses have been distinguished from nominal clauses although syntactically
they represent the same construction. This is because the participle, as described in this
study, forms part of the Egyptian verb system in a very marked way.

Pa et al. Participles, adjectives and substantive

nouns.21
- No occurrences
(+) 1-4 occurrences
+ 5-44 occurrences
++ 45 or more occurrences
It is interesting to note that switching from the marked prefix form (pVpref) to
the unmarked (Vpref) may indeed function as a rank-shifter, as in example
(33) below, in which the narrator describes how he and his newly wedded
wife spent their ten days of honeymoon in Alexandria. The past auxiliary kān
‘be’ introduces a series of bi-forms with habitual meaning. There are also
three circumstantial clauses marked by gram switches to the unmarked prefix
form, which have a relationship of purpose with their heads.
(33) da kān ḥaẓẓina nn huwwa ggawwı̊ ḥilw, fa kunna bnuxrug iṣṣubḥı̊ [ninzil
ʿala lbaḥṛı̊ šwayya], binitġadda ʿa lbaḥṛı̊ ʾaw binirgaʿ [nitġadda fi lbēt],
wi billēl binuxrug [nirūḥ hina fi Skandariyya] ṭabʿan fi Maʾmūra fi malāhi
fī maḥaṭṭit iṛṛaml, fī ḥagāt kitīra ʾawi lwāḥid mumkin yitfassaḥ fīha (TAK
12:48-49, Cairo)
DEMms beVs3ms luck-PRON1p that PRON3ms DEF-weather nice, so beVs1p
ASP-go.outVp1p DEF-morning [go.downVp1p PREP-DEF-sea a little], ASP-
eat.lunchVp1p PREP-DEF-sea or ASPreturnVp1p [Vp1p PREP-DEF-house], wa
PREP-DEF-evening ASP-go.outVp1p [goVp1p LOC PREP-Alexandria] of course
PREP-Mamoura in-it amusement parks in it Raml station, in it things manyADJfs
very DEF-one possiblePAms go.visitVp3ms PREP-PRON3fs
‘we were lucky that the weather was nice, so we used to leave in the morn-
ing [to go down to the sea for a while], we would have lunch on the beach
or return [to have lunch at home], and in the evening we would leave [to
walk in Alexandria (to…)] of course there is the Mamoura (beach), there
are amusement parks, there is the Raml station, there are very many things
that you can visit in it (Alexandria)’
21 Includes primarily active participles, but also passive participles, adjectives and sub-
stantive nouns, a category here defined as desententialized circumstantial clauses by
Lehmann (1988, 193; Isaksson 2009, 52-53). In main clauses, this category is mainly
comprised of modal modifiers, frozen in their participlee form, such as lazim andʿāwiz
followed by the prefix form, meaning ‘must do’ and ‘wants to do’.

Switching to a marked prefix form, however, is another story. The distinction

between the ‘bare’ prefix form and the bi-form is here, as elsewhere in EA, a
distinction of markedness, where the bi-form gives a clearer sense of co-
temporality and present progressivity, see 1.4 (Unmarked and marked prefix
forms).
It has been assumed in this study that the reasons behind gram switching
between a main clause and a circumstantial clause - or unmarked hypotactic
clause for that matter – are that it functions as marking of dependency and
hierarchy in discourse, see 1.1. It is also sensible to think that gram switching
plays a less significant role in syndetic circumstantials, since these clauses are
already marked through the connective phrase, wa plus anaphoric pronoun
initially, e.g. wa huwwa. Since there are also other factors influencing form
choices, analysis must combine formal and semantic criteria. It should also be
remembered that in one fourth of the collected data there is no gram switching
between clauses at all! The general conclusion, however, is that contrasting
gram forms between clauses is indeed a sign of clausal rank shift. That is, as
long as the verb (the first verb if many) of the circumstantial clause is in its
prefix form (unmarked or marked), or the circumstantial clause as a whole is
nominal: a desententialized clause (adjunct participle), a participle clause or a
nominal clause, cf. examples (22), (23), (24), (12) and (13) above. All forms
of the verb system have been documented as main clause grams.
4. Semantic relationships
It is the semantic functions of circumstantial clauses in context which makes
sense of the bewildering variation in form. Adding the semantic properties to
the analysis, including the choice of main clause verb, actually reduces the
number of types to a few highly frequent ones, as will be seen below. For
instance, the syndetic circumstantial clauses mainly function as temporal sub-
clauses, indicating simultaneity and progressivity. In the data this type repre-
sents more than 60 percent of the cases. It is often possible to use the English
conjunction ‘while’ in translation. But it may also express logical functions
such as consecutive (34), adversative/concessive (35), causal or consecutive
(36), as in the following examples.

(34) ʾāl: Gūmu ṭʿumu lġazl, [w-ḥna Gaymīn], mahu rrayis bitaʿna baGa
(ÄAD 28:20)
sayVs3ms: get.upIMPp baitIMPp DEF-net, [wa-PRON1p get.upPAmp], EMPH-
PRON3ms22 DEF-captain GEN-PRON1p PARTICLE
‘he said: get up and put bait (on the net), [so we stood up], because he was
our captain!’
(35) fa ṛṛāǧil dih, masak fiyye, [w ana ma naʿrafūš gablı̊ kidi], fimaḥaṭṭit
maṣr (ÄAD 16:18)
and.then DEF-man-DEMms, grabVs3ms PREP-1s [wa PRON1p NEG-knowVp1p-
PRON3ms-NEG before this], PREP-station-Cairo
‘and then this man took hold of me [although we did not previously know
each other] at the station in Cairo’
(36) šūf dilgēti ʿala kdih, min ilmiġrib ma-yaṭlaḥš milbēt, [wa hūwa xāyif]
(ÄAD 109:23)23
seeIMPms now like this, PREP-DEF-evening.prayer NEG-go.outVp3ms-NEG
PREP-DEF-house, [wa PRON3ms fearPAms]
‘see now (the case is) like this, from the evening prayer he has not left the
house, [because he fears]’
In the EA corpus are also found what can be described as a classical type of
syndetic ǧumla ḥāliyya ‘circumstantial clause’ in traditional Arabic grammar.
That is the type of co-temporal clause which describes the state of a partici-
pant. Almost all of the syndetic ones were found in the Cairene blog book, cf.
example (37). This may be an influence from the written language or simply
more common in Cairene than in the rural dialects.
(37) ʾaʿad ʾayman ʾuddām-i [wi huwwa xaglān wi bāṣiṣ fi l-ʾarḍ] (GhA
63:17)
sitV3ms Ayman PREP-PRON1s [wa PRON3ms shy wa lookPAms PREP-DEF-
floor]
‘Ayman sat down in front of me [being shy and looking at the floor]’
22 Emphatic particle
23 The word dilgēti is a variant of of dilwakti ‘now’, used in Upper Egypt (Behnstedt and
Woidich 1985, vol. 2, map 181).

The relationships signaled by asyndetically linked circumstantial clauses are

most frequently logical (purpose, prospect), as in example (38), or temporal
(simultaneity), as in example (39).24
(38) bigīb il-ʿiyāl ṣiḥāb-u kullı̊ xamīs [yitfarragu ʿa l-xināʾa] (GhA 73:11,
Cairo)
ASP-bringVp3ms DEF-children friends-PRON3ms every Thursday [watchVp3mp
PREP-DEF-fight]
‘he brings the children, his friends every Thursday [to watch the fight]’
(39) wu yugʿud iššāʿir [yiġanni wu ygūl ē?] yifawwit illēli, ʿala ē? ʿala
mistáwa kwayyis (ÄAD 15:31)
wa sitVp3ms DEF-poet [singVp3ms wa sayVp3ms INT] make.passVp3ms DEF-
night, PREP-INT PREP-level goodADJms
‘and the poet sits there [singing and says…] making the night pass at a
good level’
A smaller group has a specifying relationship to the main clause, as in exam-
ple (40). Specifying circumstantial clauses serve to transpose content from a
more general to a more specific level in that they define or explain something
further. The term has been used by Halliday (2004, 396f).
(40) wu dawwaṛ ilgiyād taḥtī. [yigīd bilḥṭab giṭn wu xašab.] (ÄAD 69:19-
20)25
wa get-goingVs3ms DEF-lightVN underneath. [lightVp3ms PREP-fire wood cot-
ton and wood.]
‘he started to light the fire underneath [lighting with cotton wood and or-
dinary wood].’
The asyndetic circumstantial clause unit in example (41) makes a predication
about one aspect/part of the subject participant in the main clause. This type
has been described and exemplified by Woidich who states that these clauses
may be either asyndetically or syndetically linked (Woidich 2006, 395-96).26
24 Woidich points out that the purpose function is modal rather than logical, as it involves
the intention of the speaker (2006, 397).
25 The punctuation here follows Behnstedt and Woidich (1985-1994, vol. 3II, 92). Note
that they have interpreted the two clauses as coordinated.
26 Woidich gives the following example of syndetic linking, gayyı̊ w ʾalbu maḥrūʾ ‘he
came with a burnt heart’ (with great sadness).

Since there is no gram switch in this example it is a most ambiguous case of

circumstantial clause combining.
(41) da ṛṛāgil da kwayyis ʾawi, [ṣiḥḥítu kwayyisa] (ÄAD 4:4)
DEMsm DEF-man-DEM goodADJms very, [health-PRON3ms goodADJfs]
‘this man is very well, [his health is good]’
A borderline case is also (42) which could also be interpreted as a case of
coordinate clause linking, where the specifying circumstantial clause (if it is
circumstantial) is negative. This type is obviously very similar to asyndetic
relative clauses.
(42) idrāʿu yxīb, [ma-yigdaṛš yištaġil] (ÄAD 85:15)
arm-PRON3ms be.uselessVp3ms, [NEG-be.able.toVp3ms-NEG workVp3ms]
‘his arm will be useless, [he will not be able to work]’
Note that there are quite a few asyndetic conditional clauses in the corpus, cf.
example (43). These, however, are in need of further attention, something that
is beyond the scope of this study.
(43) taxd ibnu wi wlādu… li-ḥadd ilbīr, maṣarifhum ʿalēk (ÄAD 1:30)
takeVp2ms son-PRON3ms and sons/children-PRON3ms.. up.to DEF-well, costs-
PRON3mp PREP-PRON2ms
‘(if) you take his son and his children… the costs will be on you’
5. Conclusion about Woidich as a point of reference

As a point of reference Woidich’s work on circumstantial clauses may be
used. It is the most thorough treatment of circumstantial clauses in EA
(Woidich 1991; 2006, 394-401). Woidich sees the need for a combined se-
mantic-syntactic approach but does not attempt to answer the question about
what really constitutes a circumstantial clause. Neither does he attempt to
settle the matter with the ḥāl category of traditional Arabic grammar, but
follows its syntactic taxonomy rather faithfully when identifying circumstan-
tial clauses. Woidich presents three structural types adapted from Bloch
(1965, 68ff):
Type 1: (wa) + subject + predicate
Example: ʾallaha [wi huwwa nāzil] ‘he told her [as he went down]’.

Type 1 is, according to Woidich, the most frequent and “vielseitig” (versatile)
type found, representing more than 80% of his data. The predicate of the main
clause may be verbal or non-verbal and that of the non-main clause may be
verbal or nominal (noun or participle as predicate). The relationship between
them may be temporal or modal.
Type 2: (wa) + predicate + subject
Example: rigiʿ [wi maʿā šanṭa blastik] ‘he came back [with a plastic bag]’
Type 2 is a rather marginal group that could perhaps have been grouped to-
gether with type 1, consisting of nominal clauses (copula-less clauses) where
the predicate is an adverbial or prepositional phrase and the subject is usually
non-definite.
Type 3: (main clause) + predicate
Example: biyimši [yzukk] ’he walked [limping]’
Type 3 where the circumstantial clause stands in a much tighter relationship

with the main clause than that of type 1, both semantically and syntactically
because of asyndesis and shared elements. The circumstantial clause may be
thought of as a (second) predicate of the subject or object in the main clause.
Semantically, the circumstantial clause expands or specifies the meaning of
the main clause verb modally. The main clause verb of this type belongs to a
restricted semantic class, containing verbs of motion (intransitive or transi-
tive), posture (or situation) or perception (inclusive of verbs for ‘find’ and
‘leave’), e.g. rāḥ ‘go’, miši ‘go; walk’, gih ‘come’; gāb ‘bring’, šāl ‘take’;
gaʿad ‘sit’, nām ‘sleep’; šāf ‘see’, lāʾa ‘find’.
The question remains as to what motivates the choice of one structural type
over another. According to Woidich this question primarily concerns modal
circumstantial clauses, since in his view temporal circumstantial clauses
would always be syndetic (Woidich 1991, 95). An example of a meaning shift
resulting from a choice of structural type are circumstantial clauses preceded
by the verb laʾa ‘find’, which acquires a more abstract meaning when fol-
lowed by an asyndetic circumstantial clause (Woidich 1991, 88), as in laʾētu
biyʿayyat ‘I noiced that he cried’ as opposed to laʾētu w huwwa biyʿayyat ‘I
found him crying’.

It is my view that the reality of circumstantial clauses in modern spoken

Arabic is less clear-cut than Woidich’s classification suggests. Such a view is
also congruent with the one presented for Gulf Arabic by Persson (2009). The
rich variation in form and meaning suggests that circumstantial clauses in
Arabic may be treated as a more general category of hypotactic clause linking
than is usual in traditional treatments (Isaksson 2009, 3f; Persson 2011). A
striking difference between the results of this study and that of Woidich is that
his Type 1, roughly the syndetic clauses, amounts to about 80 percent of the
data, whereas in this study, asyndetic combinations are twice as frequent as
the syndetic ones. This is the result of ‘throwing the net out wide’ and making
more use of the semantic context when identifying circumstantial material.
With such an approach, it is natural to find more asyndetically linked clauses,
since these have less marking to guide the analyst. In addition, asyndesis is a
general characteristic of expedient EA story-telling.
To summarize the findings for syndetic combinations, the participle cir-
cumstantial clauses (and other nominal ones) represent nearly two thirds of
the total number, while the rest are clauses containing the prefix form of the
verb. The syndetically linked clauses function most often as temporal com-
plement clauses. In the Cairene blog data the syndetic circumstantial clause is
a common expression for the state of a participant (subject/object) of the main
clause. In addition, logical relationships occur too, such as concessive, causal
and consecutive. Also the group of asyndetic clauses is dominated by a few
highly frequent patterns, which can be identified through choice of lexical
main verb, typically a general transitive or intransitive motion verb, through
the semantic relationship of purpose, and through the prefix form of the verb,
as in example (22) above. Furthermore, one fourth of the asyndetically linked
clauses are desententialized, containing participles mostly and usually preced-
ed by a main clause verb in the suffix form having a meaning of perception or
finding, as in example (24) above. Other semantic relationships are temporal,
specifying or logical. The asyndetic circumstantial clause combinations are
the most intriguing field of study because they open the door to a continuum
of gradual clause merger, leading eventually to grammaticalization in ways
that will be described below.

6. Auxiliation and clause merger

This and the next section will give a brief description of the processes of aux-
iliation and clause merger that are relevant to the argument of this study. The
line of argument will be based on Lehmann’s theory of clause rank and clause
merger, assuming that “the lower the level, the more tightly the subordinate
clause is integrated into the main clause” along a continuum of clause integra-
tion (Lehmann 1988, 189).
6.1 Auxiliation processes

The study of circumstantial clauses inescapably enters an adjacent field of
study, namely periphrastic verb constructions containing auxiliaries and main
verbs. The route is via the ambiguous or intermediate forms within the area of
asyndetic circumstantial clause linking, where mere juxtaposition and verb
form contrasts are the only marks of asymmetry in clause rank. Most conspic-
uously prone to grammaticalization are constructions involving verbs of mo-
tion or posture and with a semantic relationship of simultaneity or finality.
Grammaticalization processes resulting in new auxiliary verbs, so called aux-
iliation are well documented in languages generally (Hopper and Traugott
2003, 55-58, 206-207). There are also other asyndetic clause combinations
which give rise to ambiguous cases, where the context determines whether
there are two clauses or just one, but these will have to be treated in another
study. The ambiguous examples typically involve a gram switch so that, in
form, they become identical to a circumstantial clause preceded by its head,
the main clause. Does rāḥ ‘go’ in example (44) really indicate motion away
from a deictic centre, or is it a void discourse marking verb used to express
sudden action? Likewise, in example (45) it has to be understood from the
context whether the boy actually sat down or whether gaʿad ‘sit’; ‘proceed
doing’ is the aspectualizer expressing progressivity. Typical for grammatical-
ized items in the language is that the old and the new ‘bleached’ meanings
continue to exist side by side.
(44) ṛāḥ [ḍarbu] bilkaff. (ÄAD 73:82)
goVs3ms [hitPAms] PREP-DEF-flat.of.hand
‘he went (over) and slapped him with the flat of his hand’ or ‘and suddenly
he slapped him’

(45) ilwalad gaʿad [yifakkar] (ÄAD 24:19)

DEF-boy sitVs3ms [thinkVp3ms]
‘the boy sat down to think (it over)’ or ‘the boy was thinking/sat thinking’.
The result of auxiliation in EA can perhaps be compared to that in the Scandi-
navian languages, where independent clauses with the verbs go, come, sit, lie
and stand may integrate semantically with the following clause to form one
single process and one single clause (Hopper and Traugott, 2003, 206), e.g.,
Vad står du och tänker på då? ‘What are you (standing there) thinking about
then?’ Other frequent EA auxiliary verbs of this kind are fiḍil ‘remain’ > ‘con-
tinue doing’; baʿa ‘to remain’ > ‘begin to do’; ‘continue’ > ‘then’; ʾaʿad ‘to
sit’ > ‘to keep doing’.
Once the function of a lexeme has changed, as its semantic content has
been reduced, there is usually also a gradual reduction in form in grammati-
calization processes generally. In EA we can see such tendencies, for instance
in the undeclined ʾām. Even the verb baʾa ’remain; be, become; begin; then’
has been grammaticalized into a general discourse marker or “filler” with the
undeclined form baʾa or yibʾa, as in muš hatiʾdar timši yibʾa ‘you will not be
able to go then’. Note that the different functions of baʾa coexist in the lan-
guage (Woidich 2006, 325).27
In their ‘bleached’ form, the above mentioned motion verbs, rāḥ, ʾām
(qām), and gih, function more like discourse markers, something like ‘pace-
makers’, leading the story on in a rhythmic way. In this function they are
perhaps not translatable at all, or perhaps a compound connective, like ‘and
then’ or ‘and suddenly’ may be used in translation. Woidich, who has done
the most thorough analysis of asyndetic verb phrases in Cairene, says that
periphrases with ʾām, rāḥ and gih serve to give the text structure, as they bind
together dynamic situations (Woidich 2002; 2006, 329). Brustad calls the
auxiliary motion verbs ‘narrative contour verbs’, “since, as a group, they seem
to mark twists and turns of narrative events” (Brustad 2000, 192-202). As
regards the participles which so often succeed ām and rāḥ, their use has been
documented in Syrian and Kuwaiti Arabic as well. Brustad compares them to
27 Note also the development in EA of rāḥ and ʿimil ‘do’ having taken a further step
and become clitic tense/aspect markers, the former (as rāyiḥ or indeclinable rāḥ)
for future tense and the latter (as ʿammāl- or ʿa-) for present progressive as a vari-
ant of the bi-prefix, e.g. laʾētu […] ʿammāl yiliff ‘I saw him turning’ (TAK 4:64,
Cairo).

the Mandarin perfect particle which is used to signal what has just happened
or what is about to happen, as described by Li, Thompson, and Thompson
(1982, 36). Thus the participle of verbs – yielding perfect aspect – is very
much suitable “to highlight sudden or important plot events” (Brustad 2000,
199).
6.2 A general framework for clause linking
Elaboration Compression
Parataxis (Downgrading of subord. cl.) Embedding

Sentence (Syntactic level) Word
Clause (Desententialization) Noun
Lexical verb (Grammaticization of predicate) Grammatical affix
Clauses disjunct (Interlacing) Clauses overlap
Syndesis (Explicitness of linking) Asyndesis
Lehmann 1988, 217
Chart 3: Lehmann’s framework for a typology of clause linkage

One of the conclusions of this study is that the bewildering diversity of form
in EA circumstantial clause linking ought to be handled with some sort of
multi-variable tool. As I see it, Lehmann (1988) provides a framework that
could satisfy this need. Its aim is to survey the most important aspects of
complex sentence formation in languages around the world. Its clause concept
is defined as containing one predication and the linking between clauses as “a
relation of dependency or sociation”. As shown in Chart 3, the framework
consists of six variable parameters, all subsumed under the headings elabora-
tion at one pole of the continuum and compression at the other. The variables
are interdependent, but not in an absolute way, e.g., there are asyndetically
linked clauses that show no trace of reduction.
On the left of the continuum, a clause complex is made up of two clauses
that are maximally elaborated, i.e., independent full clauses that are syntacti-
cally equal. They are linked by a connective giving explicit information of the
semantic relationship. To the right of the continuum the complex likewise
consists of two clauses, but here one is a reduced predication, governed by

and embedded in the main clause. It is on a low syntactic level, strongly nom-
inalized and interlaced with the main predication, lacking specific information
on the semantic relationship. The six parameters are described as follows:
1. Hierarchical downgrading
Degrees on an axis running from parataxis (independence) to embedding
(dependence, constituency). Lehmann (1988, 183-84) exemplifies six degrees
using different languages, here represented by English, Latin and glosses (for
Hittite Australian Walbiri and Kobon from PNG).
2. Syntactic level
Pertaining to the main clause: levels between the morpheme and the para-
graph. Advanced hierarchical downgrading of a subordinate clause implies a
low syntactic level, but the opposite may not be true.
3. Desententialization
Nominalization of the main verb and reduction process where the clause loses
the properties of a clause. These properties are typically: illocutionary force,
mood, tense, aspect, actants and circumstants. The subject tends to be put in
the form of the possessive, see Lehmann’s chart, Lehmann 1988, 200). Chart
4 exemplifies two degrees of desentialization in English, note the possessive
form of the actant in (ii).
Sententiality
(i) She objected to [his constantly (ii) She objected to [his constant
reading magazines] reading of magazines]
Lehmann 1988, 200
Chart 4: Degrees of desententialization

4. Grammaticalization
Main clause verb grammaticalized into auxiliary to verb in dependent
clause.28 The whole sentence ceases to be syntactically complex as two claus-
es merge into one, sometimes with a new word order as a result. The former
subordinate verb becomes the main verb (perhaps only semantically to start
with) and the former main verb becomes its modifying operator, at an ad-
vanced stage to an undeclined clitic or a mere affix, as fecit in Miles picem
fervefecit ‘the soldier boiled the pitch’ (fervere ‘to glow’; facere ‘to do’)
(Lehmann 1988, 202).
5. Interlacing
Semantically, two clauses share elements of meaning, such as actants, predi-
cates, tense and aspect. Syntactically, these elements are not specified, e.g.,
tense is only marked in the main clause from which the non-main borrows its
tense value. The interlacing of actants, may be marked by anaphora pronouns
(cf. Arabic syndetic circumstantial clauses).
6. Explicitness of linking (syndesis vs asyndesis)

In natural unpretentious text (spoken or written), Lehmann believes that “the
explicitness of the linking device is adjusted to the size of the entities linked”:
large units of text needs explicit linking whereas for smaller units the implicit
relationship becomes “sufficiently clear from the mere adjacency” (Lehmann
1988, 211).
The most interesting point concerning Lehmann’s framework is that it in-
cludes desententialization and grammaticalization (auxiliation) into the sys-
tem. These phenomena are presented as interdependent ways of reducing a
complex sentence to a simple one: (1) to desententialize the subordinate
clause; (2) grammaticalize the governing verb, turning it into an affix which
modifies the meaning of the semantically subordinate verb. “In both cases, the
subordinate verb becomes a constituent of the main clause: in the first case, a
dependent one, in the second case, its main verb” (Lehmann 1988, 204). The
spectrum of clause linking in EA could be presented according to Chart 5
below.
28 “Grammaticalization is a diachronic process and a synchronic continuum which lead

from lexical to grammatical items” (Lehmann 1988, 201). This particular type of
grammaticalization is termed auxiliation in this study, see 4.

Elaboration Compression
Syndetic CC Asyndetic CC Participle CC Auxiliated

(anaphoric main clause
pronoun) verb
Chart 5: Arabic types of clause linking along a cline of elaboration-

compression.
6.3 Conclusions
In an earlier study on Modern Written Arabic, I presented a theory along simi-
lar lines as that presented here, making use of some examples from Halliday
(Kammensjö 2009, 195-98); Halliday 1994, 241). The idea then too, was that
gradual integration of the non-main clause (notably circumstantial clauses)
into the main clause correlates with a gradual loss of independence and
grammatical information. The drive towards integration and merger comes
from the speaker’s need to control the packaging of information to give the
story a desired ‘pace’, since “clause combination represents rhetorical organi-
zation”, that is “the scoping and structuring of a rhetorical unit in a text” (Mat-
thiessen and Thompson 1988, 303), see 1.1. In written language, clause com-
pression may involve a drive towards economy. In oral narration, which of
course is audible, the factors of pace or rhythm naturally play a greater role.
Lehmann describes clause linking as being affected by two opposing forces,
the first acting towards “the elaboration of a phrase into a more fully devel-
oped construction which contains its own predication with all the accessories
(…)”. The opposing force acts towards ”the compression of a full-fledged
clause to a nominal or adverbial constituent of a matrix clause” (Lehmann
1988, 216-17).
The phenomenon of gram switching offers a clause internal, morphology-
based marker of clause rank, which makes asyndetic constructions highly
economical in terms of effort. Thus, asyndetic clause linking becomes a fa-
voured device of construction. And with such widespread asyndesis, auxilia-
tion processes can easily get started. The typical case is represented by clause
combinations where the main verb belongs to the semantic class of general
motion and posture, such asʾām ‘rise; set out’, rāḥ ‘go’ and gih ‘come’ in
addition to ʾaʿad ‘sit’. From the EA data it is now possible to conclude that

asyndetic clause linking is a primary zone for the kind of grammaticalization

that works across clause boundaries and results in desententialization and
clauseal integration (Persson 2010; 2013).
Gram switching is also exploited as a meaning-making tool in the verb
phrase resulting from auxiliation, as illustrated by Woidich in Chart 6 below.
V1 V2 V2 V2 V2
Perfekt Akt. Partizip y-Imperfekt Imperativ
Perfekt + + + –
Imperfekt – + + –
Akt. Partizip – + + –
Imperative – – – +
Chart 6: Gram combinations in periphrases with ʾām, rāḥ and gih (Woidich
2006, 329)29
Chart 6 illustrates Woidich’s findings regarding gram combinations in asyn-

detic verb phrases in EA. It shows that whenever the first verb is a perfect (the
suffix form), the second verb may take anyone of three forms, the suffix form,
the participle or the prefix form. If, however the first verb is an imperfect or
an active participle the second verb may be either one of these two grams. If
the first verb is an imperative, the second one has to be an imperative too. The
distribution of combinations coincides with those documented in the present
research. They also roughly coincide with the gram switching patterns docu-
mented in hypotactic asyndetic clause combinations, with the exception of
imperative main clause verbs being followed by a prefixed verb in the circum-
stantial clause, rather than another imperative.
In the present corpus, the most frequent verbal gram contrast involves a
general verb of motion, such as rāḥ plus a participle. However, in circumstan-
tial clause linking the desententialized adjunct participle is mostly used after
verbs of perception or finding, as in (24) above. Only rarely would they be
preceded by an intransitive verb of motion. This means that there is no one-to-
one correspondence in patterns of verb form contrast between circumstantial
clause boundaries and the verb combinations in grammaticalized periphrastic
structures. The participle is used as the main verb in these verb combinations
and as such borrows its time reference from the auxiliary. It functions rather
29 V1 = first verb in verb phrase; V2 = second verb in verb phrase.

as the English infinitives do in similar positions. The verbicization of the EA

participle where it is frequently used as a main verb may have triggered a
development where the syntactic ‘main verb’ becomes superfluous and thus
available for other functions, such as auxuilary and discourse markers of se-
quence in the storyline.
As suggested above, there must, of course, be a drive in the opposite direc-
tion too, towards increased informativity and independence, in this context
represented by the syndetic circumstantial clause linking which is marked off
from the main clause through the connective wa plus an anaphoric pronoun.
Summary
This contribution reports on the types of circumstantial clause linking found in
a corpus of spoken Egyptian. It is similar to an earlier study of circumstantial
clauses in modern Arabic fiction prose which formed part of a research pro-
ject involving more Semitic varieties, see Circumstantial qualifiers in Semitic
(Kammensjö 2009; Isaksson et al. 2009). The circumstantial clauses are treat-
ed in their capacity to combine with other clauses, to make up larger units in
the flow of discourse. The fruitfulness of such an approach – especially for
spoken language – has been pointed out by Halliday (1994; 2004), Matthies-
sen and Thompson (1988); Fleischman (1985; 1990), and others. EA circum-
stantial clauses are usually subtly marked, most often through interlacing
(sharing semantic elements) and gram switching (verb form contrasts), some-
times with the addition of a general additive conjunction, wa ‘and’, followed
by an anaphoric pronoun referring to a participant in the main clause. Interlac-
ing and gram switching, however, are general features of hypotactic linking,
not markers of semantic relationships, such as temporality or causality. This
means that EA circumstantial clauses are ‘unmarked’ for such relationships,
which have to be inferred from context. EA circumstantial clauses, thus, rep-
resent the ‘unmarked’ subset of enhancing hypotactic clauses, also including
‘marked’ enhancing clauses (Isaksson, 2009, 3f; Persson 2011). Marked en-
hancing clauses are typically introduced by a subordinating conjunction to
signal their precise relationship to the head clause, i.e. clauses introduced by
lamma ‘when’ or ʿašān ‘because; in order to’ (EA).
The corpus is composed of transcribed oral material from Behnstedt and
Woidich (1985-1994); Ghada Abdel Aal (2007) (blog book), and Woidich
(2010). Altogether around 1170 clause units were elicited from these sources

for the purpose of this study. The data has been divided into two main catego-
ries, syndetic clauses (syndetically joined clauses), appended by means of
general conjunction wa, on the one hand, and asyndetic clauses (asyndetically
joined clauses), i.e. conjunction-less clauses, on the other. The latter category
also includes a sub-category of desententialized clauses, i.e. participles, adjec-
tives and other substantive nouns that in descriptions of written Arabic are
usually categorized as nominal. Asyndetic hypotactic linking was found to be
more than twice as common as syndetic hypotactic linking in the data (800 to
380 instances). Neither the syndetic nor asyndetic circumstantial clauses are
unique form classes. Among the syndetic combinations, the participle circum-
stantial clauses (and other nominal clauses) represent nearly two thirds of the
total number, while the rest are clauses containing the prefix form of the verb.
The syndetically linked clauses function most often as temporal complement
clauses. In addition, logical relationships occur to, such as concessive, causal
and consecutive. The group of asyndetic clauses is also dominated by a few
highly frequent patterns, which can be identified through the choice of lexical
main verb, typically a general transitive or intransitive motion verb, through
the semantic relationship of purpose, and through the prefix form of the verb.
One fourth of the asyndetically linked clauses are desententialized, usually
preceded by a main clause verb in the suffix form having a meaning of per-
ception or finding. Other semantic relationships are temporal, specifying or
logical. The asyndetic circumstantial clause combinations are the most intri-
guing field of study because it opens onto a continuum of gradual clause mer-
ger, leading eventually to grammaticalization.
Following Fleischman (1985; 1990) who showed that tense switching in
Old French functions as a strategy for ‘narrative subordination’ it has been
assumed in this study that gram switching between a main clause and a cir-
cumstantial functions as a marker of dependency and hierarchy in discourse.
However, since there are also other factors influencing form choices, analysis
must combine formal and semantic criteria. It should also be remembered that
for around one fourth of the collected data there is no gram switching between
clauses at all! The general conclusion, however, is that contrasting gram forms
between clauses is indeed a sign of clausal rank shift. That is, as long as the
verb (the first verb if many) of the circumstantial clause is in its prefix form
(unmarked or marked), or the circumstantial clause as a whole is nominal: a
desententialized clause (adjunct participle), a participle clause or a nominal
clause. The phenomenon of gram switching offers a clause internal, morphol-

ogy-based marker of clause rank, which makes asyndetic constructions highly

economical in terms of effort. Thus, asyndetic clause linking becomes a fa-
voured and device of construction. And with such widespread asyndesis, aux-
iliation processes can easily get started.
From the EA data it is now possible to conclude that one zone of change
towards further desententialization and integration is hypotactic asyndetic
linking, where certain main clause verbs are moving towards auxiliation. The
typical case is represented by clause combinations where the main verb be-
longs to the verb classes of motion and posture verbs, such asʾām ‘rise; set
out’, rāḥ ‘go’ and gih ‘come’ in addition to ʾaʿad ‘sit’ and the circumstantial
clause containing a prefix verb or a participle. The predilection of participles
as main verb may have triggered a development where the ‘main verb’ be-
comes superfluous and thus be available for other functions, such as discourse
markers of turns in the storyline. There is of course a drive in the opposite
direction, towards increased informativity and independence, resulting in the
syndetic circumstantial clause linking which is marked off from the main
clause through the connective wa plus a anaphoric pronoun.
One of the conclusions of this study is that the bewildering diversity of
form in EA circumstantial clause linking ought to be handled with some sort
of multi-variable tool. Lehmann (1988) provides a framework that could satis-
fy this need. His framework consists of six variable parameters, all subsumed
under the headings elaboration at one pole of the continuum and compression
at the other. The variables are interdependent, but not in an absolute way, e.g.
there are asyndetically linked clauses that show no trace of reduction. The
most interesting point concerning Lehmann’s framework is that it includes
desententialization and grammaticalization (auxiliation) in the system. These
phenomena are presented as interdependent ways of reducing a complex sen-
tence to a simple one: (1) to desententialize the subordinate clause; (2) gram-
maticalize the governing verb, turning it into an affix which modifies the
meaning of the semantically subordinate verb. “In both cases, the subordinate
verb becomes a constituent of the main clause: in the first case, a dependent
one, in the second case, its main verb” (Lehmann 1988, 204).
References
Abdel Aal, Ghada. 2008. ʿayza atgawwaz (I want to get married). Cairo: Dar Al
Shourouk.

———. 2010. I want to get married. Transl. from Arabic by Nora Eltahawy. Center
for Middle Eastern Studies, U. of Texas, Austin.
Badawi, Elsaid, and Martin Hinds. 1986. A dictionary of Egyptian Arabic. Beirut:
Librairie du Liban.
Badawi, Elsaid, Michael G. Carter, and Adrian Gully. 2004. Modern written Arabic: A
comprehensive grammar. London: Routledge.
Behnstedt, Peter, and Manfred Woidich. 1985-1994. Die ägyptisch-arabischen Dialek-
te. 4 vols. Beihefte zum Tübinger Atlas des vorderen Orients, series B, 50/1-4.
Wiesbaden: Ludwig Reichert.
Bloch, Ariel. 1965. Die Hypotaxe im Damaszenisch-Arabischen mit Vergleichen zur
Hypotaxe im Klassich-Arabisch. Mit Vergleichen zur Hypotaxe im Klassisch-
Arabischen. AKM 35:4. Wiesbaden.
Brustad, Kristen E. 2000. The syntax of spoken Arabic: A comparative study of Moroc-
can, Egyptian, Syrian, and Kuwaiti dialects. Washington D.C.: Georgetown Uni-
versity Press.
Bybee, Joan, Revere Perkins, and William Pagliuca. 1994. The evolution of grammar:
Tense, aspect, and modality in the languages of the world. Chicago: The Universi-
ty of Chicago Press.
Cohen, D. 1984. La phrase nominale et l'évolution du système verbal en sémitique.
Études de syntaxe historique Paris. Société de Linguistique de Paris.
Comrie, B. 1976. Aspect: An introduction to the study of verbal aspect and related
problems. Cambridge, Cambridge University Press.
———. 1985. Tense. Cambridge, Cambridge University Press.
Dixon, R. M. W. 2009. “The semantics of clause linking in typological perspective”. In
The semantics of clause linking: A cross-linguistic typology, edited by R. M. W.
Dixon and A. Y. Aikhenvald, 1-55. Oxford University Press.
Eades, Domenyk, and Persson, Maria. 2013. “Aktionsart, word form and context: On
the use of the active participle in Gulf Arabic dialects.” Journal of Semitic Studies
58 no. 2: 343-67.
Eisele, John C. 1990a. “Time reference, tense, and formal aspect in Cairene Arabic.”
In Perspectives on Arabic linguistics I, edited by Mushira Eid, 173-212. Current
issues in linguistic theory 72. Amsterdam: John Benjamins.
———. 1990b. “Aspectual classification of verbs in Cairene Arabic.” In Perspectives
on Arabic linguistics II, edited by Mushira Eid and John McCarthy, 192-233. Cur-
rent issues in linguistic theory 72. Amsterdam: John Benjamins.
———. 1999. Arabic Verbs in Time: Tense and aspect in Cairene Arabic. Wiesbaden:
Harrassowitz.

Fleischman, Suzanne. 1985. “Discourse functions of tense-aspect oppositions in narra-

tive: Toward a theory of grounding.” Linguistics 23 no. 6: 851-882.
———. 1990. Tense and narrativity: From medieval performance to modern fiction.
Croom Helm romance linguistics series. London: Routledge. Reprint, 2002.
Gordon, A. 1982. “The development of the participle in Biblical, Mishnaic, and Mod-
ern Hebrew.” Afroasiatic Linguistics 8, no. 3: 121-179.
Halliday, Michael A. K. 1985, 1994. An introduction to functional grammar. 2nd ed.
London: Arnold.
———. 2004. An introduction to functional grammar. Edited by Christian M. I. M.
Matthiessen. 3rd rev. ed. London: Arnold.
Hopper, Paul J. and Traugott, Elizabeth. 1993, 2003. Grammaticalization. Cambridge:
Cambridge University Press.
Ingham, Bruce. 1994. Najdi Arabic: Central Arabian. London Oriental and African
Language Library. Amsterdam: John Benjamins.
Isaksson, Bo. 2009. “Introduction” and “An outline of comparative Arabic and Hebrew
textlinguistics.” In Circumstantial qualifiers in Semitic: The case of Arabic and
Hebrew, edited by B. Isaksson, 1-35 and 36-150. Wiesbaden: Harrassowitz Verlag.
Jastrow, Otto. 1997. “The Neo-Aramaic languages.” In The Semitic languages, edited
by Robert Hetzron, 334-377. London: Routledge.
Kammensjö, Heléne. 2009. “Circumstantial Qualifiers in contemporary Arabic prose -
with a view to understanding variation.” In Circumstantial qualifiers in Semitic:
The case of Arabic and Hebrew, edited by B. Isaksson, 151-205. Wiesbaden: Har-
rassowitz Verlag.
Kinberg, Naphtali. 1992. “Semi-imperfectives and imperfectives: A case study of
aspect and tense in Arabic participal clauses.” Lingua 86: 301-330.
Lehmann, Christian. 1988. “Towards a typology of clause linkage.” In Clause combin-
ing in grammar and discourse, edited by J. Haiman and S. Thompson, 181-225.
Amsterdam: John Benjamins.
Li, Charles N., Sandra A. Thompson, and R. McMillan Thompson. 1982. “The dis-
course motivation for the perfect aspect: The Mandarin particle le.” In Tense-
Aspect: Between semantics and pragmatics, edited by Paul Hopper, 19-44. Am-
sterdam: John Benjamins.
Matthiessen, Christian, and Sandra A. Thompson. 1988. “The structure of discourse.
and ‘subordination’”. In Clause combining in grammar and discourse, edited by
John Haiman and Sandra A. Thompson, 275-329. Amsterdam; Philadelphia: John
Benjamins.

Mitchell, T. F. 1978. “Educated spoken Arabic in Egypt and the Levant, with special
reference to participle and tense.” Journal of Linguistics 14: 227-258.
Persson, Maria. 2009. “Circumstantial qualifiers in Gulf Arabic dialects.” In Circum-
stantial qualifiers in Semitic: The case of Arabic and Hebrew, edited by Bo Isaks-
son, 206-289. Wiesbaden: Harrassowitz.
———. 2010. Asyndetisk satsfogning i gränslandet mellan omständighetssatser,
hjälpverb och parallellism. (Asyndetic clause combining at the borders between
circumstantial clauses, auxiliaries, and parallellism.). Paper presented at Semitiska
seminariet, Uppsala University, Department of Linguistics and Philosophy, 7/12
2010.
———. 2011. “Circumstantial clause.” In Encyclopedia of Arabic Language and
Linguistics Online, ed. by L. Edzard and R. de Jong. Leiden, Boston: Brill.
———. 2013. “Asyndetic clause combining in Gulf Arabic dialects. Auxiliary, adver-
bial and discourse. functions.” Zeitschrift für Arabische Linguistik 57: 5-39.
Rosenhouse, Judith. 1978. “Circumstantial clauses in some Arabic dialects.” Zeitschrift
der Deutschen morgenländischen Gesellschaft 128: 227-36.
Schlonsky, Ur. 1997. Clause structure and word order in Hebrew and Arabic: An
Essay in Comparative Semitic syntax. New York, Oxford: Oxford University
Press.
Wild, Stefan. 1964. “Die resultative Funktion des aktiven Partizips in den syrisch-
palästinischen Dialekten des Arabischen.” Zeitschrift der Deutschen morgenländi-
schen Gesellschaft 114: 239-254.
Woidich, Manfred. 1975. “Zur Funktion des aktiven Particips im Kairenisch-
Arabischen.” Zeitschrift der Deutschen morgenländischen Gesellschaft 125: 273-
93.
———. 1991. “Die Formtypen des Zustandssatzes im Kairenischen.” Zeitschrift für
Arabische Linguistik 23: 66-98.
———. 2002. “Verbalphrasen mit asyndetischem Perfekt im Ägyptisch-Arabischen.”
Estudios de Dialectología Norteafricana y Andalusí 6: 121-92.
———. 2006. Kairenisch Arabisch: Eine Grammatik. Wiesbaden: O. Harrassowitz.
———. 2010. “Texte aus Kairo.” (unpublished CDs and transcriptions of 17 texts).

Non-main Clause Combining in Damascene Arabic:
A scale of markedness
Maria Persson, Lund University
1. Introduction
In the proceedings of the International Symposium on Clause Linking in Se-
mitic Languages that took place in Kivik, Sweden, 5-7 August 2012, I pre-
sented data that puts the concept of circumstantial clauses in Arabic into ques-
tion (Persson, 2014a).1 The analysis led to the conclusion that there is, in the
hypotaxis of spoken Syrian Arabic of the 1950s, a scale of markedness rang-
ing from no other marking than context and/or intonation, i.e. no hypotaxis in
syntactic terms, 2 to the use of specific subordinating conjunctions. On this
scale, ‘gram switching’ plays an important role as a general, non-specific
marker of a digression from the main line of a discourse or the main clause of
a clause combination (cf. Fleischman 1990; Givón 2001, 299). The term
‘gram switching’ has been developed from the observation, made by Joan
Bybee and Östen Dahl (1989), that grammatical morphemes, for which they
coined the term ‘gram(s)’, play a crucial role in signalling grammatical and
discourse structure as well as temporal and aspectual relations. Inspired by
Fleischman’s observation, Isaksson (2009b, 121-122) noted a high frequency
of switching from one verb form to another in hierarchical clause combining
in Biblical Hebrew. He discerned that the use of different verb forms in two
combined clauses constitutes a general way of marking clausal relations. This
device he labelled ‘gram switching’. Subsequent studies in Arabic have shown
that gram switching as a syntactic marker is not unique to the Biblical Hebrew
1 The study supported similar findings in modern urban Gulf Arabic data (Persson,
forthcoming).
2 This may sound like a contradiction in terms. However, the possibility of marking
clause hierarchy in other ways than syntactic must be taken into consideration when
analysing spoken language.

56 Maria Persson
texts in which it was first recognized. That a switch between verb forms exists
in circumstantial clause combining, i.e. the coincidence of a verb form switch
and circumstantial clause combining, has been noted before (e.g. Addeweesh
1985, 86-87; Fischer 2002; Premper 2002, 17-27). However, its pivotal role as
an independent grammatical marker has, to my knowledge, not previously
been recognized. This is somewhat surprising considering that gram switching
in Arabic and Hebrew, specifically a switch of verb forms, may not only sup-
plement marking through other devices, such as conjunctions, but often func-
tions as the sole marker of hypotaxis (Isaksson 2011; 2013; Isaksson, Kam-
mensjö, and Persson 2009; Persson 2014a; forthcoming).
The recognition of a scale of markedness, and of general strategies for
nonspecific marking of clause hierarchy, leads to a deeper understanding of
the function and structure of non-main clauses in Arabic dialects and sheds
new light on the widespread use of asyndetic clause linking in colloquial Ara-
bic. Analyses of such linking have been presented elsewhere (Persson 2013;
2014a). Furthermore, the results of the study presented in Persson (2014a)
have, as mentioned, led to a questioning of the existence of circumstantial
clauses as a relevant syntactic category within Arabic non-main clauses. The
different types of clauses which, based on formal criteria, have been thought
by various scholars to be circumstantial were shown to be part of a larger
system of encoding non-main clause linking. Thereby, the problem of finding
a coherent syntactic definition of an assumed class of circumstantial clauses
has been resolved, at least in the data under consideration. The results of the
study, together with previous research on modern Gulf Arabic and Biblical
Hebrew, have disclosed a need for a re-analysis of non-main clause linking in
Semitic in general. This volume is one endeavour towards that end.
Nonetheless, the term “circumstantial clauses” will be used here as there is
no reason to question the relevance of talking of circumstantial clauses in
semantic terms. Clauses that qualify other clauses circumstantially do exist as
a semantic category. What is put into question is the circumstantial clause as a
valid syntactic category. Through a discussion, based on examples from the
database, of clauses with the semantic function of circumstantially qualifying
other propositions, I will assess the validity of the concept of ‘circumstantial
clause’ in Damascene Arabic of the 1950s. Furthermore, I will demonstrate
the systematic use of gram switching as a syntactic marker of hypotaxis in
Arabic. Similar studies of modern urban Gulf Arabic presented in Persson

Non-main Clause Combining in Damascene Arabic 57
(forthcoming) are thus supplemented by older data from a different Arabic

dialect.
In the first part of this presentation, (3. Formal coincidence in Arabic non-
main clause linking), I survey ways in which non-main clauses that circum-
stantially qualify other clauses coincide, in terms of syntactic form, with
clauses that serve other functions in relation to their main clauses. This sec-
tion, in other words, gives examples of cases of ambiguity: cases where the
same syntactic structure may be interpreted as either a circumstantial qualifi-
cation or something rather different. The discussion in this section discloses
general strategies of non-main clause linking which are, subsequently, dis-
cussed in the next section, (4. Gram switching), where a possible disambiguat-
ing function of gram switching in the previously discussed clauses is evaluat-
ed. Finally, the formal types of circumstantial clauses suggested by Bloch
(presented in 2.3 Bloch 1965) are discussed in the light of the role played by
gram switching.
The survey further supports the theory, presented in Persson (2014a), of a
scale of markedness where gram switching plays a pivotal role. Depending on
the speech situation or the speaker’s preferences, a non-main clause may be
left unmarked at the syntactic and morphological level, marked with a gram
switch, or doubly marked by both a gram switch and a conjunction. Gram
switching, hence, is found to be a marker of non-main clause linking in gen-
eral. What have been labelled circumstantial clauses in various descriptions of
Arabic grammar are, to a large extent, clauses where non-main clause linking
is marked in this vague way. The (semantic) function of the clause; whether it
is expressing the condition, cause, consequence, interpretation, time, place,
manner, and so forth, of the head, is not made explicit in the clause structure.
Circumstantial clauses, thus, do not constitute a distinct grammatical class of
clauses. Rather, they belong to the said overarching system of hypotaxis with
non-specific marking, i.e. where the existence of non-main clause is signalled,
but the type of non-main clause is not specified. The same clause types, the
same syntactic forms, are used for more than one function. Within this system
of hypotaxis, gram switching emerges as significant marker of the relationship
between clauses.3
3 For a more extensive discussion, see Isaksson et al. (2009).

58 Maria Persson
2. Definitions, scope and previous research

2.1 General definitions and scope
Neither Arabic grammars, nor general works on clause combining contain a
unified, established, definition of the term “circumstantial”. Thompson, Lon-
gacre, and Hwang (2007) count “circumstantial clauses” as one specific type
among the “adverbial clauses” they describe. These are broadly described as
“clauses expressing the circumstances by which a given state of affairs comes
to be” (2007, 253). They are, however, in the next sentence, restricted to such
clauses as those that, in English, are introduced by by and without followed by
the participial form of the verb. Lichtenberk (2009, 259), in his description of
clause linking in Toqabaqita, speaks of “attendant circumstances” and vaguely
states that in such clauses, “the Supporting clause expresses a characteristic or
a circumstance of the state of affairs expressed in the Focal clause, as relating
to one (set) of the participants”.4 Premper (2002, 20-21) notes that some of the
definitions found in grammars of Arabic are rather “complementary” and cites
Beeston who includes most of the “adverbial clauses” surveyed by Thompson,
Longacre, and Hwang except the conditionals:
The logical relationship between the main proposition and the circum-
stantial one is highly variable: the circumstantial clause may be purely
temporal, or adversative (…), or explanatory (…). But there is an un-
sophisticated lack of overt marks of the logical intention. (Beeston
1970, 89)
Thompson, Longacre, and Hwang (2007, 237-243) also mention that clause
linking in some cases or languages may be a matter of juxtaposition of claus-
es. The semantic relationship between the clauses in such linking is not ex-
plicitly marked but inferred from the context. The “unsophisticated lack of
overt marks of the logical intention” described by Beeston is thus found cross-
linguistically. Like the present study, Addeweesh concludes that only a se-
mantic definition of the circumstantial clause is feasible:
The primary importance we give to the semantic function of ḥāl is

based on the fact that this function is the only feature that is inclusive
of all types functioning as such, yet is exclusive of other related syntac-
4 Cf. also Matthiessen (2002, 263-267).

tic patterns. Including the syntactic types of ḥāl in the definition is not
practical; there is a variety of types. (Addeweesh 1985, 185)
Similarly, Premper (2002, 29) concludes from his survey of standard works on
Arabic grammar, that there are no specific morphologic or syntactic exponents
that single out ‘the circumstantial clause’ from other types of clauses. Gram-
marians have listed the structures they have seen in the texts they have sur-
veyed, but when compared side by side it is clear that they do not agree. This
conclusion is further supported by the findings of the project that has led to
this publication and its forerunner.
Nevertheless, in nearly all descriptions of Arabic syntax, be they modern
or classical, standard or colloquial, there is a section or paragraph on ‘circum-
stantial clauses’.5 In other words, ‘circumstantial clauses’ are described as a
syntactic category on a par with conditional clauses, relative clauses and simi-
lar, perpetuating the illusion. That this ‘class’ of clauses in spoken Arabic
should take on some specific forms, or display some peculiarities, is presup-
posed, just as, for example, the relative pronoun has different forms in collo-
quial Arabic, or as differences in lexicon and morphology have consequences
for the realization of conditional clauses. Hence, also descriptions of colloqui-
al Arabic grammar include sections on the ‘circumstantial clause’. ‘Circum-
stantial clauses’ (Zustandssätze) in the data used for this study have been
previously described by Bloch (1965). His definition of what constitutes a
circumstantial clause in Damascene Arabic (see 2.3) is purely formal. He
recognizes three syntactic structures as the only forms in which circumstantial
clauses appear in the texts. While the clauses mentioned by Bloch and others
do function to qualify other clauses circumstantially, no list is comprehensive.
The function of a ‘circumstantial clause’ evades being caught within the
boundaries of a syntactic category.
Thus, the starting point for this study is that any definition of what can be
said to constitute a circumstantial clause must be based on function rather than
form. For the purposes of this study, a clause will be considered to function as
a circumstantial qualification of another clause if
a) it carries either a descriptive or a logical relationship to its head

clause, and
5 The actual term chosen to signify the category may, of course, vary.

60 Maria Persson
b) this relation is not made explicit by a conjunction, but has to be in-

ferred from context
Circumstantial clauses, as defined here, serve as unmarked alternatives to

clauses that are explicitly marked by conjunctions.6 When the relationship to
the head clause is descriptive, the non-main clause provides additional or
background information about time, place, manner, or other details concern-
ing the head clause; when the relationship is logical, the non-main clause
conveys a condition, a cause, a result, an interpretation, and so forth, of the
head clause.7
Gram switching mainly involves a switch between morphological forms of
the verb such as a switch from a participle or suffix form to a prefix form. It
may also involve mood marking. Arabic also makes use of verbless clauses,
i.e. clauses in which no verbal gram is found. For the purposes of this study, a
switch between the use of a verbal gram and a clause without a verbal gram
will also be counted as a gram switch.8
2.2 Data
The data for this survey has been collected from the Syrian Arabic texts rec-
orded on audio tape in Damascus by Hans Wehr in 1956. The texts were tran-
scribed, translated into German, and published together with a commentary
and glossary by Bloch and Grotzfeld (1964). Heinz Grotzfeld subsequently
published works on phonology, morphology and syntax in Syrian Arabic
(Grotzfeld 1964; 1965) based mainly on these texts, and Ariel Bloch pub-
lished a study on hypotaxis in Syrian Arabic (1965).9
6 Cf. also Beeston (1970, 89), quoted above; Isaksson (2009a); Persson (2009).
7 Cf. Persson (2009, 238-249)
8 An alternative would be to speak of switches between clause types (cf. Isaksson 2013).
This, however, would remove focus from the verbal grams which, in all switches, con-
stitute the pivotal part. In the context of this paper it may also cause terminological con-
fusion since “clause type” may also refer to the semantic types of clauses discussed here
such as relative clauses, conditional clauses, final clauses and so forth.
9 The original intention of collecting new data from informants in situ had to be aban-
doned as civil war broke out shortly after the onset of this research project. Collecting
data from informants outside of the country was considered but risk of lost authenticity
was deemed too high and, hence, the choice was made to concentrate on the older mate-
rial and, by that, create a basis for later comparison.

Two texts, numbers 14 and 19, were excluded from the survey. Text 14
was excluded since it is based on written material and is described by Bloch
and Grotzfeld (1964, 88) as partly unnatural and ‘classicizing’. Text 19 was
omitted because it has been edited and because the language is said not to be
representative of the Damascus dialect (1964, 146). Set expressions copied
from the standard language, such as the general congratulation “kəll ʿām w-
ʾənte bḫēr”, ‘every year and/while you are well’, have also been excluded.10
The functional definition given in 2.1 constituted the basis for the selection
of data. A phrase or a clause selected for scrutiny may have more than one
possible interpretation. However, at least one possible interpretation must
imply that it functions as a clause that qualifies another clause circumstantial-
ly. These criteria led to a collection of 210 clauses that were found to clearly
have a circumstantial reading as one of their possible interpretations. Another
45 were initially studied but discarded as they were found to be too ambigu-
ous, i.e. a circumstantial reading was unlikely. These are, thus, not among the
210. Finally, an additional 73 cases of larger units of text, i.e. not separate
clauses, with a circumstantial function, were added to the database and con-
sidered in Persson (2014b). The total database, then, consists of 328 instances,
whereof 210 are under consideration in this study.
As the data consists of spoken language, an attempt has been made to rec-
ognize false starts and anacoluthon and to exclude such anomalies from the
study. With only the transcribed text available and no comments of this kind
made by the editors – and the text being edited – there is a risk, however, that
some of these have not been recognized. With these difficulties in mind and
with the aim of finding structures that have not previously been recognized,
the strategy has been to discuss, rather than discard, cases of ambiguity and
doubt. By making such cases available to the reader’s judgment, it is hoped
that their inclusion in the study will broaden the perspective on hypotaxis in
Arabic (and Semitic) and advance the linguistic analysis in this area. In this
sense, the study can be said to be pre-theoretical and an invitation to further
discussion. Although a few numbers will be mentioned, the main approach in
the study is qualitative, not quantitative.
10 Bloch, however, includes examples from these texts, as well as set greetings as the one
described above, for his survey of circumstantial clauses. So, for example, in §70a and b
and in §71b (1965, 69-71).

62 Maria Persson
2.3 Bloch 1965

Bloch’s study of hypotaxis in Damascene Arabic is mainly based on the texts
that also constitute the basis for this study.11 He (1965, 68) divides circum-
stantial clauses in Arabic into three structural types which, by default, consti-
tute his definition of circumstantial clauses:
1. (w-) + subject + predicate (verbal or not verbal)

2. (w-) + predicate (prepositional phrase) + subject
3. finite verbal predicate
Examples of these are:

(1) ʾaḍḍēna mədde mn əzzamān [w-nəḥna ʿam nətšawwa w-nətʾalla]
(1:67:26)12
spendSF1cp period PREP DEF.time [w-1cp PROGR be.grilledPF1cp w-be.friedPF1cp
‘I spent some time [sizzling and frying (=frustrated)]’13
(2) māreʾ wāḥed [w-maʿo marto] (20:4:166)
passAPms one [w-PREP3ms wife3ms]
‘there was one passing by [(having) with him his wife]’
(3) tammēt ʾāʿed [ʿam ʾətfarraž ʿalēhon] (7:8:68)
remainSF1cs sitAPms [PROGR watchPF1cs PREP3cp]
‘I remained seated [watching them]’
11 In addition to the texts collected by Wehr, Bloch has used Bergsträsser (1924) and
Malinjoud (1924) and, to some extent, Dietrich (1956), Oestrup (1897) and Cantineau
and Heibaoui (1953).
12 Examples cited from Bloch and Grotzfeld (1964) are referred to by text
number:section:page. This way of referring differs from the one used by Bloch and
Grotzfeld themselves (Bloch 1965; Grotzfeld 1964; 1965). I believe that it is important,
however, for the reader to be able to easily identify examples that derive from the same
text.
13 Plural verb forms are used to denote the actions of the (singular) narrator. Note that
English translations of Arabic examples in this article are given as a help to understand-
ing the structure of the Arabic clauses. Hence, the use of English in the examples is not
always idiomatic. In this example, the Arabic expression “nətšawwa w-nətʾall” is a fig-
urative way of saying that the speaker was frustrated. I have, however, used a literal
translation to show that the predicate contains verbs.

Bloch’s analysis of circumstantial clauses in Damascene Arabic, thus, builds

on his assumption that they can be recognized by their structure. For this rea-
son, he differentiates between circumstantial clauses and circumstantial ex-
pressions (1965, 79-81). He states that the latter often consist of two juxta-
posed parallel clauses of equal status such as:
(4) rəḥt ʿalbēt - [ḥēṭ ʿam biṣəddni w-ḥēṭ ʿam yrəddni] (1:88:34)
goSF1cs PREP.DEM.house – [wall PROGR b.push.awayPF3ms.1cs w-wall PROGR
pull.backPF3ms.1cs]
‘I went home [one wall pushes me away one wall pulls me back
(=staggering from wall to wall)]’
Bloch excludes these ‘circumstantial expressions’ from what he defines as
circumstantial clauses on a formal basis since they do not always conform to
any of the three structural forms described above. He further states that, as
compared to what he defines as circumstantial clauses, they also display a
difference in their function in that they describe the details of a (whole) situa-
tion.
Furthermore, Bloch (1965, 81) refutes the existence of the so called ‘ḥāl
muqaddar’, the ‘implied circumstantial clause’, in Damascene Arabic. The
reason for this is that the clauses concerned have syndetic parallels in clauses
introduced by la-, ‘in order to’. From this, Bloch concludes that these are final
clauses, not circumstantial clauses. Moreover, he states that the circumstantial
clauses on one hand, and this type of final clause on the other, are formally
distinct in Damascene Arabic: the final clauses have verbs in the bare prefix
form, whereas the asyndetic circumstantial clauses contain prefix verbs pre-
ceded by b- or ʿam, but never by the bare prefix form.14 By the invention of b-
and ʿam, Damascene Arabic, he states, has created a way to differentiate for-
mally between the final clauses and the circumstantial clauses.15
As to function, Bloch (1965, 77-78) concludes that the main function of
circumstantial clauses in Damascene Arabic is to encode simultaneity. Within
the borders of this simultaneity, other types of relationships between the two
clauses may also exist. Such relations may be causal relationships or instances
of logical opposition.
14 Cf. also Bloch (1965, 75).

15 Cf. 5.6 below.

64 Maria Persson
Bloch’s types, as well as his exclusion of ‘circumstantial expressions’ and

‘ḥāl muqaddar’, will be discussed in section 5 in the light of the results pre-
sented in sections 3 and 4.
3. Formal coincidence in Arabic non-main clause linking
The relevant point is that hypotactic or paratactic relations among

clauses are logical relations, not necessarily syntactic ones. Counting
up things that look like subordinate clauses is not the way to arrive at
judgements about relative hypotaxis or parataxis. (Johnstone 1990,
222)
3.1 A concept that evades formal definitions

Bloch, as described previously, exclusively bases his analysis on formal crite-
ria, declaring clauses of three formal types to be the only circumstantial claus-
es in the database:
Charakteristisch für den Zu sta nds sa tz ist der Aufbau nach bestimm-
ten Grundschemata, die hier als ‘Formtypen’ bezeichnet werden. In ei-
nem solchen Schema liegen Wesen, Reihenfolge und Anzahl der ein-
zelnen Glieder fest. Die drei Formtypen: …16 (Bloch 1965, 68)
I agree with Bloch that the features that his three types display are all possible
ingredients of a circumstantial clause and, conversely, clauses of these types
may very well express an “attendant circumstance”.17 However, as declared
already in my introduction, I do not believe that such a narrow syntactic defi-
nition of circumstantial clauses, nor Bloch’s even more limited view on their
function, serve to capture an existing function in the language.
As Premper concluded, there are no specific morphologic or syntactic ex-
ponents that single out the circumstantial clause from other types of clauses
16 I.e. “Characteristic of the circumstantial clause is the[ir] construction according to

certain basic patterns, that are here referred to as 'form types'. In such a pattern, the na-
ture, order and number of the individual members are fixed. The three types of shapes:
…”
17 Cf. Isaksson (2009a, 5).

(2002, 29).18 Since there is no designated morphological marker that is used

solely for marking circumstance, the structures used for circumstantial clauses
coincide with a number of other clause functions. Often, the function can only
be discerned through the context; sometimes ambiguity remains even when
the context is taken into consideration:
Die Grenze zwischen Zustandsätzen und asyndetischen/syndetischen

beigeordneten Sätzen ist fliessend. Eigentlich mag jeder asyndeti-
sche/syndetische beigeordnete Satz als Zustandsatz dienen; Bloss der
Zusammenhang, und wohl auch der Ton (einschliesslich Pause) ma-
chen den beigeordneten Satz zu einer Art Nebensatz, eben zum Zu-
standsatz. [---] Da aber die Konstruktion der Zustandsätze so frei ist,
mag nicht nur manches, das wir als Zustandsatz aufgefasst haben, ein-
fache Parataxe sein, sondern auch manches, das wir als beigeordnete
Sätze angeführt haben, hierher gehören, da die zwischen dem Haupt-
satz und dem Nebensatz bestehenden Begriffsverhältnisse mannigfach
sind,…19 (Blau 1960, 223; cf. also Persson 2009, 256-263)
Meanwhile, many scholars have noted that asyndetic clause combining, in

general, is common in Arabic dialects and that this often causes ambiguity
between (semantic) coordination and subordination (Blau 1960, 214; Fischer
2002, 153; Isaksson et al. 2009)20. Thus, Bloch describes what he considers to
be asyndetically joined paratactic clauses with a temporal or conditional, i.e.
hypotactic, semantic relationship. These, he states, are numerous and often
ambiguous (1965, 92-94; cf. Johnstone 1990, 222-223, 227).
18 Cf. Addeweesh who writes that: ”Even a preliminary review of medieval presentations
of the ḥāl reveals that certain vague generalizations and strong ambiguities do exist in
the description” (1985, 4).
19 I.e.: “The boundary between circumstantial clauses and asyndetic/syndetic juxtaposed
clauses is fluid. Actually, every asyndetic/syndetic juxtaposed clause may serve as a cir-
cumstantial clause; only the context, and arguably also the intonation (including pause)
make the juxtaposed clause into to a kind of subordinate clause, even to a circumstantial
clause. [---] However, since the construction of the circumstantial clauses is so free, not
only may many things that we have considered as circumstantial clauses be simple para-
taxis; some things that we have called coordinated clauses may belong here, since the
notional relationships that exist between the main clause and the subordinate clause are
manifold ...”
20 Cf. also Eksell (1995); Isaksson (2008, 252, 256-257) and Premper (2002, 329-346). See
Woidich (2002, 121) for a comprehensive list of references.

66 Maria Persson
Some main areas of coincidence in form between circumstantial clauses

and other types of clauses will be exemplified in this study with data from
Bloch and Grotzfeld’s texts. In the following section I will show how a hither-
to rather unnoticed grammatical marker is actually at work in these clauses
and how the recognition of this marker can resolve some of the ambiguity.
The most obvious cause of ambiguity is the fact that the only conjunction
that, at times, is found to introduce circumstantial clauses is the general junc-
ture marker w(ə)- which is used to indicate juncture also in simple coordina-
tion of clauses. The most well-known coincidence, on the other hand, is the
one between the asyndetic circumstantial clause and the asyndetic relative
clause with indefinite correlate. However, the most common coincidence in
the present database is between a category of circumstantial clauses which
Bloch decides not to recognize as such, namely the so called ḥāl muqaddar on
one hand, and a reading of the verb in the main clause as an auxiliary on the
other. The last major area of overlap, that I will discuss, concerns a more
pronounced digression from the main line of the narrative, where a single
clause can be interpreted either as a comment on the temporal/situational
circumstance of the main event, or as an explanatory side-comment with a less
direct link to the main clause/main verb of the head clause.
3.2 Coincidence between circumstantials and ordinary

coordination or sequence
That circumstantial clauses may be introduced by the non-specific conjunction
w(ə)- results in ambiguity between circumstantial and coordinated clauses.21
However, ambiguity between coordination and circumstance; between para-
taxis and hypotaxis, occurs also when there is no conjunction. The syntactic
point at issue is that circumstantial clauses are often not syntactically embed-
ded in their head clauses. A moderately subordinated relationship between
two clauses may be realized only at the semantic level, i.e. without any syn-
tactic expression. In the following example, the clause that starts with “w-
hadāk” is identical in form to Bloch’s Type 1 circumstantial clause:
(5) ṣār hūwe yərkod [w-hadāk yəlḥaʾo] (20:2:166)
22
AUXSF3ms 3ms runPF3ms [w-DEM followPF3ms.3ms]
21 Cf. Kammensjö (2008, 2009), Persson (2009, 257-258).

22 Lit: he became. Cf. 5.2.4 below.

‘He started running [and that one followed after him]’

Depending on how this second clause is interpreted, the example means either
“he started running and that one followed him” or “he started running while
that one followed him”. The ambiguity in the Arabic clause combination is
found also in the English translation but is (irrespective of language, of
course) resolved by the context: there would be no sense in following after
someone who does not, simultaneously, keep on running. Thus, the sense of
“simultaneity”, often described as a hallmark of circumstantial clauses, here
follows from our knowledge of the world, not from any syntactic marker. A
similar example is (6) where w(ə)- is followed immediately by a prefix verb:
(6) kəll yōm təṭbḫīlo sabʿ tmən ʾaškāl [w-tʾūli mū ẓarīfe mənšān wāžeb əḍḍēf]
(17:53:140)
every day cookPF2fsPREP3ms sev’ eight kinds [w-sayPF2fs NEG nice for.sake.of
duty DEF.guest]
‘Every day you cook sev’ eight courses for him [(and/while) you
say/saying it isn’t proper (not to do it) for the sake of hospitality]’23
In (6) it is the temporal expression “every day”, telling us that the event in the
first clause is repetitive, combined with the lack of separately stated time in
the second clause, which leads to an interpretation where the second clause is
dependent on the first. If there had been a temporal expression also in the
second clause such as “and now you tell me…” the two activities would have
been seen as separate. As it is, the iterative of the first clause carries over to
the second clause creating a (semantically) hypotactic relationship. Sometimes
the ambiguity is even more extensive. Consider the following:
(7) ərrəžžāl byəʾʿod [wəlmara btəštəġəl] (11:2:76)
DEF.man b.sitPF3ms [w-DEF.woman b.workPF3fs]
The context allows for at least three possible interpretations of the second
proposition in (7):24
23 Literally: ‘for the duty of the guest’ but the implied meaning is ‘for the duty one has in
relation to the guest’. Bloch and Grotzfeld (1964, 141) translate it into German as “Es ist
nicht schön (wenn ich es nicht tue), wegen der Gastfreundschaft”.
24 The context is as follows: For the first time in twelve years, a judge is coming to town
and the town is going to prepare for the visit. However,…

68 Maria Persson
a) it can be temporal: ‘The man sits while the woman works’

b) it can be adversative: ‘The man sits (idle) but the woman works’
c) it can be simple coordination: ‘The man sits and the woman works’
That all three meanings can also be implied in the English translation with
“while” in (a) above suggests that this kind of ambiguity is not specific for
Arabic. As stated above, ambiguity between coordination and subordination
exists also when there is no conjunction:
(8) biḥəṭṭu məšwāye ʿala halfaḥm – w-byəʾʿdu baʾa [byəšwūha] (16:12:120)
b.putPF3cp gridiron PREP DEM.DEF.charcoal – w-b.sitPF3cp PTCL [b.fryPF3cp.3fs]
‘They put a gridiron on the charcoal – and then, you know, they sit down
[(and) fry it]’
A literal translation of the last two predicates would be “and they sit down,
they fry it”25. This combination may express either a coordinate construction,
‘they sit down and fry it’, or the circumstances of their sitting down, i.e. ‘they
sit down frying it’. As may be noted, the English sentence with ‘and’ is also
ambiguous allowing a choice between a circumstantial reading and a literal,
coordinated reading.
Yet another area of overlap between coordination and circumstance ap-
pears when suffix verbs are asyndetically joined. With these, there is usually
an implication of sequence since the suffix form implies that each of the
events or activities involved is completed. There are, however ambiguous
cases:
wa-ʾiz žarāʾim ktīr bəlbalad – w-ḫəṣṣa hənne wənnəswān - ərrəžžāl byəʾʿod [wəlmara
btəštəġəl] – ma fī ʿandhon hadīk əlʾēš? – hadīk əlḥanān hadīk əlʾənsānīye – bišaġġlu
lmara wərrəžžāl byəbrok bəlʾarḍ - əlḥāsel - ʾaktarīyəthon mətḫānʾīn hənne w-
nəswānhon.
‘Now, there are a lot of crimes in town – and especially (between) them and the women
– the man sits while/and/but the woman works – they do not have that what (do you say)
– that compassion, that humanity – they put the woman to work and/while/but the man
would sit idle on the ground – the result (of this was that) – most of them were in a fight
(they) with their womenfolk’.
In other words, an adversative interpretation is adduced by the following explanation.
The very fact that the speaker feels a need to add this clarification emphasizes the ambi-
guity of the statement quoted in (7).
25 The word baʾa, functions as a discourse particle and may often, as here, be left untrans-
lated. For a comment on its use, see Persson (forthcoming b).

(9) ʾaʿadna [trawwaʾnālna šwayye] (1:44:18)

sitSF1cp [have.breakfastSF1cp some]
‘we sat [and had/having breakfast for a while]’
The verb is a locational: ʾaʿad, ‘to sit’. Locational verbs often develop into
auxiliaries (cf. 3.4). The core activity in the narrative is not that someone was
sitting but what this someone was doing while ‘sitting’. The less the focus is
on the act of sitting, the more likely it is that the verb ‘to sit’ becomes
bleached and, eventually, transforms into an aspectual marker for the progres-
sive (‘be doing’).
Finally, there are instances introduced by w(ə)- that Bloch and Grotzfeldt
have translated as paratactic, but that, as far as I can see, describe circum-
stances:
(10) ṣār ʿanna ṣaby w-bəntēn [w-nəḥna ʿam nəštəġel ləssāʿətna bəlmaṭbaʿa
lhāšmīye] (1:106:40)
becomeSF3ms with.1cp boy w-two.girls [w-1cp PROGR workPF1cp still
PREP.DEF.printshop DEF.Hashemite]
‘(now) I had a boy and two girls [and (?) I was still working at the Hashe-
mite Press]’26
The first clause describes a development and change that has occurred over
time. When a stable situation such as ‘I was still working at the Hashemite
Press’ is juxtaposed to this expression of development, this last part is, by
default, perceived as a comment on the circumstances. It is, however more of
an adversative “and/but, yet, I was still working” than a simultaneous reality
“while still working”. Shortly after this there is a similar clause combination:
(11) la kānet tzīd ʾəžrətna mətl ənnās w-la kānet tənʾaṣ – [w-ləssaʿətna ʿam
nətḥammal mənno halmaʾāsi] (1:108:40)
NEG beSF3fs increasePF3s salary1cp like DEF.people w-NEG beSF3fs de-
creasePF3fs - [w-still PROGR endurePF1cp PREP3ms DEM.DEF.severity]
‘My salary did not increase as it should and/but it (also) did not decrease –
[and I was still enduring this harshness on his part]’27
26 Bloch and Grotzfeldt’s translation into German is as follows: „Ich hatte nun einen Jun-
gen und zwei Mädchen und arbeitete immer noch in der Haschimiyya-Druckerei”.

70 Maria Persson
While clearly circumstantial, both of these are side-comments rather than the
kind of “attendant circumstances” described in the literature.28 There are more
instances in the database of comments like these that digress from the main
line of the narrative. They will be the topic of section 3.5.
3.3 Coincidence between circumstantials and relative clause

The coincidence in form between relative clauses with indefinite correlate and
circumstantial clauses has been frequently observed. Judith Rosenhouse, writ-
ing on classical Arabic, even called this “the main problem of ambiguity
which cannot be solved (possibly) by using only linguistic markers” (1978,
235). Premper, in his analysis of this phenomenon, states that, in these cases,
the relationship to the head is not specified (2002, 319). The ambivalence can
be double in that, firstly, it is not clear whether the non-main clause refers to
the head as a whole or only to the head clause predicate, and secondly, wheth-
er it refers to the action (the predicate) as such or to one of the actors. Hence,
it is not clear if the appended clause functions as an adverbial (circumstantial)
or an attributive (relative) modification.29 Bloch (1965, 75) mentions the fol-
lowing example:
(12) hūwe māši bhaṭṭarīʾ - māši wāḥed [maʿo ḥmār] w-ʿabāyto ṭawīle
(20:1:166)
3ms walkAPms PREP.DEM.DEF.road – walkAPms one [PREP.3ms donkey] w-
cape3ms long
‘(as) he was walking on the road – there was one walking [who
had/(having) with him a donkey] and his cloak was long’
The same ambiguity may occur when the non-main clause has a verbal predi-
cate:
27 Bloch and Grotzfeldt’s translation: „Kurz und gut, (…) wurde mein Lohn nicht höher
wie es sich gehört, aber auch nicht geringer. Und ich ertrage jetzt immer noch von ihm
diese Strenge“.
28 Cf. Lichtenberk (2009, 259).
29 Moreover, as Premper points out (2002, 92), so called circumstantial clauses are not
strictly “ad-verbial”. In providing details about a situation, they modify not only the
event/action (the verb) but also the participants of this event/action. “Circumstantial
clauses”, in other words, have a broader scope. Cf. Matthiessen and Thompson (1988,
276-286) for a discussion on the difference between enhancing hypotaxis and adverbial
clauses.

(13) yōm byəṭlaʿ ʿalīyi kalb [biʿawwi ʿalīyi] w-yōm byəṭlaʿ ʿalīyi sakrān
[biḫawwəfni] (1:10:4)
day b.overtakePF3ms PREP1cs dog [b.barkPF3ms PREP1cs] w-day
b.overtakePF3ms PREP1cs drunk [b.scarePF3ms1cs]
‘One day a dog overtakes me [barking at me/that barks at me] and one day
someone drunk overtakes me [scaring me/who scares me]’
(14) šāf wāḥed [ʿam yǝṭmor msaddas bǝlʾarḍ] yaʿni fard (5:5:60)
seeSF3ms one [PROGR buryPF3ms gun PREP.DEM.earth] meanPF3ms revolver
‘He saw someone [burying/who burried a gun in the ground] a revolver,
that is’
Finally, an example where the reading may be relative or final:
(15) kəllma … yəḍṭarru yṭālʿu wāḥed mətl əḥkāyti [yʾammənlon əlmuḫābara]
(1:97:38)
every.time … be.forcedPF3cp sendPF3cp one like kind301cs [en-
surePF3ms.PREP3cp DEF.telephone.connection]
‘Every time that… they had to send out someone like me [who would/in
order to ensure the telephone connection for them]’
A locational verb such as ṭālʿa, ‘send’, in this example, followed by a verb in
the bare prefix form, and with no conjunction to separate the two, is typical of
the final clauses that have been called ḥāl muqaddar. This leads to the discus-
sion in the following section on circumstantial clauses with final meaning.
3.4 Coincidence between ḥāl muqaddar and auxiliary

The coincidence between the so called ḥāl muqaddar, and a reading of the
first of two juxtaposed verbs as an auxiliary, derives from the fact that a ḥāl
muqaddar is made up of two asyndetically combined verbs. The head clause
of a ḥāl muqaddar is most often a verb of motion. Verbs denoting posture or
change of posture or location, i.e. either stable position or movement of some
kind and in some direction; what Eisele (1999, 142) calls locationals, are
well-known from typological studies as a nurturing ground for grammaticali-
zations with temporal or aspectual value (Bybee, Perkins, and Pagliuca 1994,
11; Lord 1993, 215-217; Payne 1997, 311-312). Ambiguity appears when the
30 The word literally means ‘story’ but is used in this way, with an attached pronoun, in the
sense of ‘kind’ or ‘type’.

72 Maria Persson
locational which is the first of two combined verbs is also found in auxiliary
use in the language. The relationship between the two verbs becomes skewed
through the said processes of grammaticalization, or auxiliarization, where the
first verb is bleached and the emphasis is placed on the second verb. The first
verb turns into an auxiliary and, if the process continues, may be reduced into
no more than a modal or aspectual particle. This has, at varying degrees, hap-
pened in some dialects to the verb qām for example, but also to verbs such as
rāḥ, ‘go’, and žāʾ, ‘come’ (Firanescu 2003; 2008, 188; Persson 2013).
Asyndetic verb combinations where the first verb is a locational verb are
very common in the data. The following example contains no less than three
coordinated pairs that, theoretically, could function as either auxiliary+main
verb or main verb+“ḥāl muqaddar” clauses.31
(16) ʾana ʾāyme [ʾətġaṭṭa] w-rūḥ [ʾəštəki] w-yəžu [yṭālʿū] bəlʾūwe žžabrīye
(17:49:138)
1cs riseAPfs [coverPF1cs] w-goPF1cs [complainPF1cs] w-
comePF3cp[oustPF3cp.3ms] PREP.DEF.force DEF.compulsory
‘I’m going to 32 [cover 33 ] and/to go [complain] and/so that they come
[and/to oust him] by force’
The interpretation of the first combination is rather unambiguous. The verb
qām, here, does not literally mean to get up; the woman is most likely already
standing. Rather, it implies to “get going”, i.e. an auxiliary reading. The two
other locationals both retain their sense of movement. The lady will
go/translocate to present her complaint and they will, or so she hopes, literally
come to the house to oust the unwelcome guest. Still, there is a degree of
auxiliarization here too, at least in the first of the pairs. Focus is on the second
verb, the act of complaining, rather than on the translocation. A similar exam-
ple, with the same matrix verb, is (17):
(17) ʿam byəži bbāli ʾənni fūt [fayyʾo] (17:5:126)
PROGR b.comePF3ms PREP.mind1cs COMP1cs enterPF1cs [wakePF1cs.3ms]
31 Note that it is the second verb in the combination that is a candidate for being a non-
main clause (what has been called ḥāl muqaddar). Hence, although the first verb is in
focus in the discussion, the second verb has been placed within square brackets to be
concistent with the marking in the rest of the article.
32 Lit: getting up.
33 I.e. get dressed for going out.

‘I’m getting the idea that I (should) go in [and (= in order to) wake him
up]’
The German translation by Bloch and Grotzfeldt (1964, 127), “Es kommt mir
gerade der Gedanke, zu ihm hineinzugehen und ihn aufzuwecken”,34 suggests
that ‘entering’ and ‘waking’ are two activities that are coordinated. Conceptu-
ally, however, they may also be perceived as one combined activity. With this
reading, the verb fāt, ‘enter’, has taken a step towards auxiliarization, towards
creating a compound verb phrase together with the verb ‘wake up’. I will
discuss the combinations in (16) and (17) further in 4.3.3.
3.5 Coincidence between circumstantials at clause and text level

Circumstantial clauses constitute a digression from the main line. Previous
studies have shown that the same strategies as those used by a narrator for a
short digression from the main line of the narration at the clause level, may
also be used to signal major digressions consisting of larger units of discourse
(Persson 2009, 229-238). When the digression is made up of a whole unit of
discourse, there is no risk of confusion between this and a circumstantial
clause. Also, in these cases, there is no main clause that functions as a head
clause to the digression. Such examples from the present database are dis-
cussed in Persson (2014b).
However, the data also contains a number of clauses for which a more tra-
ditional circumstantial interpretation overlaps with the function of a comment
to a larger unit of discourse. A single clause can, in other words, function
either as a circumstantial qualification of the predication in an adjacent main
clause; or as a comment at the text level having as its head, not a main clause
but a whole unit of text. Thus, the clause in (18) starts with a w(ə)-followed by
a pronoun which means that it, structurally, conforms to a pattern that has
often been identified with circumstantial clauses (Bloch’s Type1, cf. example
5 above):
(18) twaffa ʾabi - [w-ʾana kān ʿəmri ʿašr ṭnaʿšar səne bələktīr] (1:1:2)
dieSF3ms father1cs – [w-1cs beSF3ms age1cs ten twelve years PREP.DEF.much]
‘My father died – [(and) my age was ten, twelve years at the most]’
34 I.e. ‘The idea comes to me now to go in and wake him up’.

74 Maria Persson
With a circumstantial interpretation, the translation would be “my father died

when I was ten, or twelve at the most”. The context, however, makes it more
sensible to interpret the clause as an explanatory comment, added outside of
the actual narrative:
(18b) ʾana lammen kənt əẓġīr - twaffa ʾabi – [w-ʾana kān ʿəmri ʿašr ṭnaʿšar
səne bələktīr] – kənt ʾāʿed bəlmadrase w-mfakker (1:1:2)
1cs CONJ beSF1cs small – dieSF3ms father1cs – [w-1cs beSF3ms age1cs eight
twelve years PREP.DEF.much] - beSF1cs sitAPms PREP.DEF.school w-thinkAPms
‘Me, when I was small - my father died - [I was ten or twelve at the most.]
- I was at school, thinking…’
The clarification about age is added to emphasize that the narrator was, in-
deed, young. It does not have the adverbial flavour of a circumstantial clause:
it does not qualify a specific predication.35 In (19) there is a longer digression
from the main line of the narration where the aim is to provide a comment and
where the clauses conform (in form) to the types mentioned by Bloch:
(19) tḥammalət - (…) - ṣərt ʾəštəġel maʿ haššaġġīle - [w-hənne ytaʿʿbūni
laḥatta yharrbūni] – [w-ʾana ʾətḥammal] – w-kəll yōm nəntəʾəl mən balad
labalad - (…) - w-hēk laḥatta yəntəhi ʾāḫer əlḫaṭṭ… (1:93:36)
persevereSF1cs – (…) - AUXSF1cp36 workPF1cs PREP DEM.DEF.workmen – [w-3cp
make.tiredPF3cp.1cs PREP.CONJ make.fleePF3cp.1cs] – [w-1cs perseverePF1cs] –
w-every day movePF1cp PREP town PREP.town – (…) – w-thus PREP.CONJ
endPF3ms last DEF.line…
‘I persevered – (…) - I started working with these workmen – [(and) they
were working me hard to make me leave] – [(and) I/while I was persever-
ing] – every day we were moving from town to town – (…) - thus until the
end of the line was reached’37
In spite of the formal correspondence of each of these clauses to the oft-noted
structure of circumstantial clauses with w(ə)+pronoun, they are not subordi-
nate to main clauses. They do not clearly refer back to a main clause predi-
cate. Rather, they describe the social conditions (unkind colleagues) that
called for perseverance, followed by a description of the general work condi-
35 Cf. Isaksson (2009b, 103-105).

36 Lit: I became. Cf. 5.2.4 below.
37 Lit: ended.

tions (constant travelling) that, also, created a need for perseverance. Yet an-
other example is found in (20):
(20) ʾəltəllo ma biṣīr ġēr ma žəbli38 šaʾfet ḥašīše. 3. [msāwīn baʾa nəḥna šāy]
– rəḥna ḥaṭṭēnālo bkāset əššāy šaʾfet ḥašīše –(9:2f:70)
saySF1cs.PREP3ms NEG b.happenPF3ms other REL bringPF2ms?PREP1cs piece
hashish. 3. [makeAPmp PTCL 1cp tea] – goSF1cp placeSF1cp.PREP3ms PREP.glass
DEF.tea piece hashish
‘I said to him, no you must bring me39 a piece of hashish. 3. [We had made
some tea, you see] – we went and put a piece of hashish for him in the
glass of tea’
Bloch (1965, 74) mentions these, too, and calls them independent circumstan-
tial clauses (versebständigter/unabhängiger Zustandssatz). As is further elabo-
rated on in Persson (2014b), even larger units of text may constitute digres-
sions from the main line of a story that carry what may be called a ‘circum-
stantial flavour’.
3.6 Summary
Form, then, is not a sufficient criterion for deciding whether a specific clause
functions as a circumstantial qualification of another clause. The syntactic
forms that have been associated with circumstantial clauses are used by other
classes of clauses too, with resulting ambiguity. Furthermore, the “circum-
stance” may be a comment that functions at the text level rather than at the
clause level.
With such ambiguity, one may wonder if there is still a reason to talk of
circumstantial clauses. Although no formal criteria identify the circumstantial
clause, there are, however, clauses that, from a semantic perspective, function
as circumstantial qualifications to other clauses. This qualification can be
temporal or adversative, final or concessive, and so forth. The semantic type
of the qualification is not specified; the clauses are not marked as adverbial,
final or interpretive, causative or adversative. This lack of specific marking
leads to ambiguity between circumstantial meanings and other semantic cate-
gories of non-main clauses. In other words, and as has been stated already,
38 Sic! The quoted form žəbli appears to be an imperative. One would have expected 2ms
prefix form: tžəbli. Thus, it may be a case of assimilation of the t. Cf. Bloch and
Grotzfeld (1964, 70, footnote 78).
39 Lit: it will not happen without that you bring me…

76 Maria Persson
circumstantial clauses do not form a specific grammatical category on a par

with, for example, conditional or relative clauses.
Moreover, although the type of non-main clause is not specified, hypotaxis
itself is largely marked in these clauses. The marking device used is one that
has received little attention and that has rarely, if ever, been recognized in this
capacity. However, when taken into consideration, it may indicate which of
the alternative readings is most plausible in ambiguous cases, such as the ones
discussed here. This marking device will constitute the focus of the next sec-
tion.
4. Gram switching
4.1 A scale of markedness
When Premper (2002, 17-18, 23), as a summary of his findings in the main
works of Arabic grammar, lists four characteristic traits of the circumstantial
(ḥāl) clause, a switch from a suffix form verb to a prefix form verb is one of
the traits that he mentions. Gram switching has also been mentioned in con-
nection with other non-main clause linking structures. Thus, Lewin (1966,
*39*), observed that complement clauses (that-clauses) in his data were, op-
tionally, marked by a gram switch construction. Similarly, Addeweesh gives
clear examples of where a gram switch is necessary for a clause combination
to receive a hypotactical interpretation.40 Moreover, of the examples already
cited in this article, two thirds feature a switch between two different verbal
grams, or a switch between a clause with a verbal gram and one without. In
clause combinations such as the following, gram switching is the only marker
of hypotaxis:
Participle → [progressive particle + prefix form verb]:
(21) ʾāʿed laḥālo [ʿam yəšrab ʾargīle] hāda. (4:7:52)
sitAPms alone [PROG drinkPF3ms waterpipe] DEM
‘This guy was sitting there alone, [drinking a waterpipe]’
40 Addeweesh (1985, 107-109). Cf. also his discussion on how the gram switch between a
suffix form with qad and one without is necessary for a circumstantial reading (1985,
95-104). In both instances, Addeweesh has neither recognized the gram switch per se
nor seems to have understood its dynamics. Instead, he refers to stylistic reasons.

Suffix form + participle → [progressive particle + prefix form verb]:

(22) tammēt ʾāʿed hēk ləssāʿa təsʿa [ʿamma ʾətṣaffan] (7:4:66)
staySF1cs sitAPms thus PREP.hour nine [PROG ponderPF1cs]
‘I remained sitting like that until nine o’clock, [pondering]’
Suffix form verb → [progressive particle + prefix form verb]:
(23) šāfo [ʿam yəġsəlha] (32:6:192)
seeSF3ms.3ms [PROG washPF3ms.3fs]
‘He saw him [washing it]’
Neither Western nor Arab grammarians have, it seems, noted the pivotal
syntactic role played by gram switching. It has, merely, been observed as a
feature commonly occurring in connection with circumstantial clauses. Yet, as
recent studies have shown, it is not primarily gram switching that coincides
with circumstantial clause combining. Rather, gram switching is a basic syn-
tactic marker of hypotaxis. It may be the only marker of a certain instance of
clause linking, but it may also be reinforced by a conjunction. By the recogni-
tion of how gram switching is used to mark non-main clause linking, some
cases of ambiguity between a paratactic and a hypotactic reading of asyndeti-
cally linked clauses can be resolved. At the same time, it becomes evident that
the concept of “circumstantial clauses” is a semantic taxonomy that bears
little, if any, relevance to the syntactic classification of clauses.41
When noting the frequent occurrence of gram switching to mark hypotaxis,
both alone and together with other markers, it is equally important to observe
that such marking is not necessary. A clause that, semantically, functions to
circumstantially qualify another clause may be totally unmarked. For the data
presented here, the following scale has been found to apply to the marking of
hypotaxis (Persson 2014a):
− No marking. Only context and/or intonation

− Gram switching only
− Gram switching + neutral conjunction (“and”) indicating the location, and
thus the existence, of the linking
41 Hence, the concept of ḥāl muqaddar also bears little relevance for the analysis of claus-
es (cf. Persson, forthcoming).

78 Maria Persson
− Gram switching and conjunction - or conjunction alone - specifying se-

mantic type of non-main clause
As has already been demonstrated, a proposition may be semantically subor-

dinate to another proposition without being syntactically marked as such.
Hence, some of the ambiguity mentioned by Bloch and Blau cannot be elimi-
nated.42 As was seen from the English translations of examples in the previous
chapter, such ambiguity is not unique to Arabic. Quite naturally, less marking
coincides with an increase in ambiguity. Thus, it may be observed that all
examples in the present survey, in which a gram switch does not occur, are
instances where the clause is ambiguous between a paratactic and a hypotactic
reading.43 This total lack of marking, such as in examples (24) and (25) which
are also discussed by Bloch (1965, 92-95, 98), means that the tem-
poral/conditional meaning has to be derived from the context alone:
(24) [btəsʾal] bidəllūk (1:64:24)
[b.askPF2ms] b.directPF3cp.2ms
‘[(if/when) you ask] they will show you (lit. you ask they show you)’
(25) [baʾa bifūt wāḥed ʿassəžn] byəšlaḥ ʾawāʿī (21:8:168)
[PTCL b.enterPF3ms one PREP.DEF.prison] b.take.offPF3ms clothes3ms
‘[now, (if/when) someone enters prison] he takes of his clothes’
Bloch mentions how the logical link between the events in the clauses
makes us perceive a connection between the clauses instead of interpreting
them as independent. That the clauses are asyndetically juxtaposed strength-
ens this impression, as do, possibly, intonation patterns (Bloch 1965, 94).44
The semantic relationship between the clauses follows from logic and our
knowledge of the world and, hence, no linguistic marker of this is needed. In
addition to temporal/conditional clauses, the database also contains a number
of final clauses where the first verb is a locational:
42 Cf. 3.2 above.

43 Cf. the already quoted examples 5-9, 13, 15, 17 as well as two of the three verb combi-
nations in 16.
44 Bloch discusses this asyndetic clause combining in terms of parallelism as it, to him, is
the fact that the verb forms are the same, rather than the lack of a coordinating or subor-
dinating conjunction, that creates the link between the clauses.

(26) ḫāf ʾənno yšəmm əssabʿ rīḥet ʾəns w-yəži [yāklo] (28:3:186)
fearSF3ms COMP smellPF3ms DEF.lion smell human w-comePF3ms
[eatPF3ms.3ms]
‘He feared that the lion would smell the scent of a human being and come
[and/to eat him]’
The man in (26) is afraid that the lion would come with the intention of eating
him. Judging from what we know of lions, there is justification behind his
fear. There is little reason to believe that the lion would come over just for a
visit. Our knowledge of the world, thus, is enough to make us detect the inten-
tion and, hence, the finality of the clause. Similarly:
(27) byəstanna mišān hūwe lʿaskari yəṭlaʿ [ynām] (26:2:182)
b.waitPF3ms for.sake 3ms DEF.soldier ascendPF3ms [sleepPF3ms]
‘He was waiting for him, this soldier, to go up [and/to sleep]’
Here, too, it is the context and our knowledge of the world that tell us that the
young soldier would go up the stairs with the intention of going to bed. A
change of location in particular, such as in these examples, often comes with a
purpose. The motion involved, itself, implies directionality. Therefore, finality
does not have to be overtly marked in these clauses.45 It is in this way that the
first category above (“No marking. Only context and/or intonation”) should be
understood. When there is no syntactic marking of hypotaxis, the clause is
syntactically not a non-main clause. Semantically, however, there is a hierar-
chy between the propositions.
One step further down the scale, gram switching functions as a marker of
non-main clause linking in general. It often constitutes a switch from one verb
form to another, such as a switch from a participle or suffix form to a prefix
form; or from a verbal predicate to a verb-less clause.46 Gram switching that
signals hypotaxis may also involve morphological entities on lower levels
45 One notes, however, that examples of locationals + non main clause with final meaning
where a gram switch does occur also abound, namely when the locational is in the suffix
form. They are so common that Arab grammarians have awarded them a name: ḥāl
muqaddar, ‘implied circumstantial clause’. Cf. 3.4, 4.4 and 5.6. Here, as well, it may be
assumed that it is the motion encoded by the locational that brings about the final inter-
pretation of the clause; the gram switch only signals hypotaxis.
46 Such as in examples 21-23 above. Cf. also the examples and discussion in 4.2 below.

80 Maria Persson
than the verb form, such as mood marking (Lewin 1966, *39*; Persson, forth-
coming). 47
The location of juncture between the two clauses may be marked by a se-
mantically vague conjunction such as w(ə)-. The specific classification of
clauses is, however, still dependent on the context:
Genau wie man sich in BZ in manchen Situationen, in denen wir im

Deutschen genauere Konjunktionen gebrauchen, mit Asyndese begnügt
[…], so verwendet man oft die ”Allerwelts”-Konjunktion u und über-
lässt es der Phantasie des Zuhörers, die genauere Beziehung zwischen
den Sätzen zu ergänzen.48 (Blau 1960, 218)
Thus, while not defining the semantic relationship between the clauses or the
type of clause combining, it does add saliency to the juncture. The use of such
a juncture marker will therefore be regarded as another step on the scale from
no marking to full marking of clause hierarchy.
For maximum transparency as to the semantics of the non-main clause
linking, the speaker can choose to use a semantically specific subordinating
conjunction which, thus, constitutes the last step of the marking scale.
4.2 Types of clauses marked by gram switching

Gram switching is used to mark such varying types of non-main clauses as
adverbial clauses, final clauses, and conditional clauses. As mentioned above,
the use of gram switching alone seems to constitute the least specific marking
of non-main clause linking that the language offers. As seen by the examples
below, drawn from the present data base, it is in no way limited to the kind of
hypotaxis traditionally labelled as circumstantial.
4.2.1 Conditional clauses

In the previously surveyed modern Gulf Arabic data, conditionals were fre-
quently marked by a gram switch only. In the present older database of Dama-
scene Arabic, however, only less explicit conditionals seem to be marked by
gram switching as demonstrated by the examples below. More research is
47 Cf. (30) below.

48 I.e. “Just as one in B[ir]Z[eit] in some situations is satisfied with asyndesis where we
use more precise conjunctions in German [...], one often uses the "all round" conjunc-
tion u and leaves to the imagination of the listener to supply the more precise relation-
ship between the clauses ”.

needed to discern if the difference between the two databases in the frequency
of conditionals marked by gram switching is incidental for these specific texts
or if the age of the data, or differences between the two dialects, lie behind it.
4.2.2 Temporal/conditional clauses

There is often a semantic overlap between temporal clauses and conditionals
(Bloch 1965, 92-95; Persson 2009, 260-263). Such temporal/conditional
clause combining without any gram switching was discussed above (24, 25).
A few examples with gram switching are also found in the present database.
(28) ṣār wazn əṣṣabāṭ fōʾ əttlāte kīlo - [bəddo yəmši] byəsḥab ḥālo ṣaḥb
bhaṣṣabāṭ (21:2:168)
becomeSF3ms weight DEF.boots over DEF.three kilo – [wish3ms walkPF3ms]
b.dragPF3ms himself dragVN PREP.DEM.DEF.boots
‘The weight of the boots had become over three kilo – [(if/when) he want-
ed to walk] he would literally drag himself wearing49 these boots’
The use of no other marker than a change of construction type to encode tem-
poral/conditionals is noted cross-linguistically (Dixon 2009, 16). In (28),
where the temporal/conditional non-main clause precedes the main clause, the
switch is from a combination of bədd50+a bare prefix form verb to a prefix
form verb with b-prefix. On a similar note, result clauses may be marked by
gram switching:
(29) ġassel wəššak [btəṣḥa w-btətnaššaṭ] (17:28:132)
washIMPms face2ms [b.wake.upPF2ms w-b.become.energeticPF2ms]
‘wash your face [(and) you will wake up and get energy]’
In this example there is a gram switch from the imperative to prefix form with
b-prefix. The prefix form verbs describe the expected consequence of obeying
the order expressed by the imperative. Bloch discusses this example, interpret-
ing the imperative as (semantically) a conditional (if you wash…). However,
as he is not aware of how gram switching functions to signal hypotaxis, he
considers the clause linking to be paratactic.
49 Lit: with these boots.

50 For a description of clauses with bədd, see 5.4 below.

82 Maria Persson
4.2.3 Final clauses

Clauses without gram switching, but with a sense of finality were also dis-
cussed above (26, 27). In the following examples, however, the hypotactic
clause linking is evident through the gram switches. Nevertheless, the sense of
finality, here as well, is only an implication and other interpretations are
sometimes possible:
(30) ʾən kān ʿažabak haššarṭ hāda - btəži tāni yōm [təštəġel] (1:15:6)
CONJ beSF3ms pleaseSF3ms.2ms DEM.DEF.condition DEM – b.comePF2ms second
day [workPF2ms]
‘If you are pleased with this condition, you come tomorrow51 [(in order to)
work]’
(31) ʾəmmi bədda maṣāri [təṣref] (1:28:10)
mother1cs wish3fs money [spendPF3fs]
‘My mother needed money [to spend]’
(32) fāt laʿandi ʿalʾōḍa [yṣabbeḥ ʿalīyi] (17:34:136)
enterSF3ms PREP.LOC.1cs PREP.DEF.room [greetPF3ms PREP1cs]
‘he came into the room to me [to greet/greeting me good morning]’
(33) ʾana žāye [ʾətfarraž] ʿala ʿadlkon (20:8:168)
1cs comeAPcs [watchPF1cs] PREP justice2cp
‘I have come [to watch] your jurisprudence’
The gram switch in (30) is one between modal forms of the verb: a prefix
form with the b-prefix alternates with one without. In (31) the switch is from a
non-verbal clause with bədd- to a prefix form verb. Example (32) is taken
from a context with suffix verb forms. Here, hypotaxis is marked on the last
verb by the switch to a prefix form verb. In (33), finally, the switch is from a
participle to a bare prefix form verb. Before uttering these words, the person
speaking has, for a while, been watching the judge in action. The judge has,
here, turned to him to ask why he has come.
4.2.4 Adversative clauses

Other non-main clauses that may be marked by gram switching are those that
express the adversative. These clauses encode limitations on the validity of
51 Lit: the second day.

the proposition in the head, or present circumstances that are contradictory to

the statement in the head:
(34) ḍallēt sabʿa tməntəšhor bmaṭbaʿt əlḥukūme [w-ma kān yərḍa halmudīr
halmanẓūm yəḥsəbli ʾəžra] (1:26:10)
staySF1cs seven eight.months PREP.print-shop DEF.goverment [w-NEG beSF3ms
consentPF3ms DEM.DEF.director DEM.DEF.dignified creditPF3ms.PREP1cs salary]
‘I stayed seven, eight months in the national print shop [without that the
dignified director agreed to pay me a salary]’
The first clause here contains a suffix form verb, whereas, in the second
clause, there is a combination of the copula verb kān in suffix form and a
prefix form verb. In addition to this gram switch, the clause juncture is sig-
nalled by w(ə)-.
4.2.5 Relative clauses

The ambiguity inherent in asyndetic relative clauses was discussed above
(3.3). Here are a few more examples:
(35) barki ʾənno ʾalla byəžmaʿak bḥada [təṭlaʿ ʾənte wīyā] (1:68:26)
maybe COMP God b.bring.togetherPF3ms.2ms PREP.someone [set.outPF2ms 2ms
with3ms]
‘Maybe God brings you together with someone [with whom you can go]’52
(36) žamaʿna bzalame [ydabbərna bšaġle ʿattalifōnāt] (1:91:34)
bring.togetherSF3ms.1cp PREPchap [arrange.forPF3ms.1cp PREP.business
PREP.DEF.telephones]
‘brought me together with someone [who (would) get me53 a job in the tel-
ephone business]’
The contexts of these examples clearly give us relative clause interpretations
of the clauses. Specifically, a relative clause reading is supported by the tense
that does not support a reading where the action or event in the non-main
clause is simultaneous with the head clause action or event. Hypotaxis is
marked in (35) by a switch from a prefix form with b-prefix to one without;
and in (36) by a switch from a suffix form to a prefix form. The syntactic and
morphologic marking, however, does not disclose the clause type.
52 Lit: you go with him.

53 Lit: he makes arrangements for me.

84 Maria Persson
4.2.6 Causal clauses

A non-specific non-main clause marked by gram switching may also express
the cause for the action or event expressed in the head clause:
(37) yḍallu raġbānīn [ʿam ymaššu ḥālon] (1:25:10)
remainPF3cp wantAPmp [PROGR make.walkPF3cp themselves]
‘(they) remain keen [(as) they are making it through]’
Since the interpretation of clause type, in lack of conjunctions or other specif-
ic markers, is context dependent other interpretations may be possible. Similar
to my suggested translation above, Bloch and Grotzfeldt (1964, 11) have
opted for a translation that allows for both a simultaneous (“circumstantial”)
and a causal reading: “und willig blieben, indem sie sich einigermaßen über
die Runden brachten”.54 The gram switch here is from a bare prefix form verb
to one marked for the progressive by ʿam-.
4.2.7 Manner clauses

One category that, traditionally, has been associated with circumstantial
clauses is descriptive clauses with adverbial meanings such as where, when or
how the head clause event or activity has taken place:
(38) waʾʾafūna bʾǝrne [ʾafāna lǝlḥēṭ] (5:10:62)
stopSF3cp.1cp PREP.corner [backs1cp PREP.DEF.wall]
‘they placed us in a corner [with our backs to the wall]’
In (38) a non-verbal clause effectively breaks the succession of activities - the
flow of the narrative related in suffix form verbs – to give us a description of
the activity: namely that the placement of the men was done in such a way
that their backs were against the wall.
4.2.8 Complement clauses

Complement clauses do not belong to the types of clauses that have been
counted as circumstantial. Nevertheless, the syntactic marking may very well
be the same:
(39) ʾūlīlo [yəži lahōn] lašūf… (17:31:134)
sayIMP2fs.PREP3ms [comePF3ms PREP.here] PREP.seePF1cs…
54 I.e. “… and remaining keen, while/as fairly making the ends meet”.

‘Tell him [to come here] that I may see…’

The gram switch here is between an imperative and a bare prefix form verb.
Examples of a gram switch that marks complement clauses were also men-
tioned by Ambros (1977, 88; cf. also Lewin 1966, *39*). Note that a switch
between an imperative and a prefix form was also used in (29) above, where
it, instead, marked a result clause. The type of switch, thus, has no implication
for the (semantic) type of non-main clause linking.
4.2.9 Clausal circumstantials at the text level

Finally, an example where, in my view, the translation provided by Bloch and
Grotzfeldt is not entirely felicitous, and where I believe the reason for their
choice of translation may be a lack of awareness of how gram switching is
used to mark clause linking:
(40) btəži mən halfərn - bətkūn ṣəḫne - mənṣaḫḫen nətfet mayy mənbəḫḫha
(16:21:122)
b.comePF3fs PREP DEM.DEF.oven – b.bePF3fs hot – b.heat.upPF1cp bit water
b.vaporizePF1cp.3fs
“Wenn sie vom Backhaus zurückkommt, ist sie warm. Wir wärmen ein
bißchen Wasser und besprühen sie.”55
Bloch and Grotzfeldt (1964, 123) interpret the first line in this example as
temporal: “Wenn sie vom Backhaus zur̈ uckkommt, ist sie warm”, i.e. “when it
comes from the bakery it is warm”. However, the main line of the story con-
sists of a number of coordinated verb clauses expressing a temporal sequence
“we cover it, we send it, we heat … etc”. All these verbs are in the prefix form
with the b-prefix (realized as m- for 1 person singular).
This line of simple prefix form verbs is broken by a clause without a full
verb. Instead the copula verb kān ‘to be’ is used: bətkūn ṣəḫne ‘it will be
warm’. Even though the b-prefix is attached to the copula verb, this use of a
different gram than the surrounding also breaks the story line. It does not
move the story forward but is an aside comment, a digression from the main
line. It is marked as such by this gram switch. With all due respect to the
translation work done by Bloch and Grotzfeldt and their Syrian consultants, I
would therefore suggest this slight adjustment to the translation: “(then) it
55 I.e. When it comes from the bakery it is warm. We heat some water and sprinkle it.

86 Maria Persson
comes from the bakery – warm – and (then) we heat up some water and spray
it.
4.3 Cases of ambiguity revisited

The awareness of how gram switching works to signal hypotaxis lessens the
ambiguity of asyndetically linked clauses. The clauses described as ambigu-
ous in section 3 above ought, therefore, to be revisited. In 3.5 there is no ques-
tion of ambiguity between different types of non-main clauses. In all exam-
ples, gram switching marks a general, non-specific digression from the main
line. The ambiguity, instead, concerns the scope of the digression. This kind
of ambiguity will not receive more attention here.
4.3.1 Coincidence between circumstantials and ordinary coordination

or sequence
A survey of the ambiguous examples in section 3 where one possible reading
of the clause linking is as coordination or sequence shows that, in most of
these cases there is no gram switch. In (5) and (6), both clauses feature a bare
prefix verb; in (7) and (8) both verbs are prefix forms with the b-prefix; and,
finally, in (9) both clauses have suffix form verbs. However, in four instances
gram switching does occur:
In (10) the switch is between a verb-less clause56 with the copula verb ṣār, ‘to
become’, on the one hand, and a prefix verb preceded by ʿam on the other; in
(11) the switch is between a verb combination with the copula verb kān +
prefix forms of the verb and, again, a prefix verb preceded by ʿam:
(10) ṣār ʿanna ṣaby w-bəntēn [w-nəḥna ʿam nəštəġel ləssāʿətna bəlmaṭbaʿa
lhāšmīye] (1:106:40)
becomeSF3ms with.1cp boy w-two.girls [w-1cp PROGR workPF1cp still
PREP.DEF.printshop DEF.Hashemite]
‘(now) I had a boy and two girls57 [and (?) I was still working at the Hash-
emite Press]’
(11) la kānet tzīd ʾəžrətna mətl ənnās w-la kānet tənʾaṣ – [w-ləssaʿətna ʿam
nətḥammal mənno halmaʾāsi] (1:108:40)
NEG beSF3fs increasePF3fs salary1cp like DEF.people w-NEG beSF3fs de-
creasePF3fs - [w-still PROGR endurePF1cp PREP3ms DEM.DEF.severity]
56 More specifically, a clause without a full verb.

57 Lit: with me had become a boy and two girls.

‘My salary did not increase like it should and/but it (also) did not decrease
– [and I was still enduring this harshness on his part]’
Both of these are, as described in 3.2, comments on the situation that the nar-
rator provides alongside the main narrative. This digression from the main
line of the story is marked by gram switching.
The corpus also contains examples with gram switching where an adversa-
tive interpretation of the clause is reinforced by the use of independent, pre-
posed pronouns:
(41) ṣār hūwe yākol mən hazzawāde [w-ʾana yḫallīni ʿalfaḍle]. (1:71:28)
58
AUXSF3ms 3ms eatPF3ms PREP DEM.DEF.provisions [w-1cs leavePF3ms.1cs
PREP.DEF.rest]
‘He started eating from these provisions [leaving the left-overs to me]’59
(42) yəlḥašli rġīf w-šaʾfe žəbne w-ḫyāra [w-hūwe lʾaklāt ʿando ʿala kēfo ʿam
yākol] (1:72:28)
throwPF3msPREP1cs loaf w-piece cheese w-cucumber [w-3ms DEF.food LOC3ms
PREP discretion3ms PROGR eatPF3ms]
‘He threw over a loaf and a piece of cheese and cucumber to me [(while)
he, having the food, was eating at his discretion]’
The added observation, that the clause linking in these clauses is also marked
by gram switching, means that a plain coordinative reading can, probably, be
ruled out.
The rest of the examples in 3.2 are, as said, unmarked. As was discussed in
that section, it is implications from the context, from logic and from the basic
meanings of the verb form (suffix form), that, in those examples, lead to a
possible hypotactic reading of clause linkings that, syntactically, are paratac-
tic.
4.3.2 Coincidence between circumstantials and relative clause

Of the four relative clauses mentioned, two feature gram switching. In (12)
the switch is between a participle and a verb-less clause; in (14) it is between
a suffix form verb and a prefix form verb preceded by ʿam. It is interesting to
note that it is the two clauses not marked by gram switching that are clearly
58 Lit: he became. Cf. 5.2.4 below.

59 Lit: and me, he left to me the left-overs.

88 Maria Persson
translated as relative clauses by Bloch and Grotzfeldt; the ones marked by

gram switching are ambiguous. This does not mean that relative clauses are
not marked by gram switching. 60 It may, however, be seen as supporting
Johnstone’s theory that little or no subordination/embedding is involved in
relative clauses in Arabic (1990, 222-223, 227).
4.3.3 Coincidence between “ḥāl muqaddar” and auxiliary

In (16), with three pairs of verbs, there is a gram switch between the first two
verbs but not between the verbs in the following two pairs:
(16) ʾana ʾāyme [ʾətġaṭṭa] w-rūḥ [ʾəštəki] w-yəžu [yṭālʿū] bəlʾūwe žžabrīye
(17:49:138)
1cs riseAPfs [coverPF1cs] w-goPF1cs [complainPF1cs] w-comePF3cp
[oustPF3cp.3ms] PREP.DEF.force DEF.compulsory
‘I’m going to61 cover62 and/to go complain and/so that they come and/to
oust him by force’
The gram switch between the first two verbs in (16) is one between an active
participle and a bare prefix form verb. In example (17) there is no gram switch
between fūt, ‘I go in’, and fayyʾo, ‘I wake him up’:
(17) ʿam byəži bbāli ʾənni fūt [fayyʾo] – (17:5:126)
PROGR b.comePF3ms PREP.mind1cs COMP1cs enterPF1cs [wakePF1cs.3ms]
‘I’m getting the idea that I (should) go in and (= in order to) wake him up’
All “head” verbs in these examples are locationals: ʾām, ‘get up’, rāḥ, ‘go’,
žāʾ, ‘come, and, fāt, ‘enter’, and all but the last of these are commonly found
as auxiliaries. That the first verb in (16) is in the form of a participial, which is
what causes a gram switch, is, most likely, to add liveliness and a sense of
immanency to the action; i.e. “I’m going to get going (now!) to …”. The lack
of gram switch in the other verb pairs promotes a non-final, close to auxiliary
reading of these, i.e. something like “I’ll go get dressed and (go) complain and
they’ll come oust them”. It is interesting to compare (17) with (32) as this
contains the same head verb:
60 Cf. examples (12) and (14) above.

61 Lit: getting up.
62 I.e. get dressed for going out.

(32) fāt laʿandi ʿalʾōḍa [yṣabbeḥ] ʿalīyi (17:34:136)

enterSF3ms PREP.LOC.1cs PREP.DEF.room [greetPF3ms] PREP1cs
‘he came into the room to me to greet/greeting me good morning’
In (17) there is, as said, no gram switch. Both verbs are in the bare prefix
form; in (32), on the other hand, there is a switch between a suffix form and a
bare prefix form. In both examples there is a sense of intentionality. However,
the close relationship between the verbs in (17) suggests a combined activity
where the entering into the room is part of the process of waking the man
from his sleep. In (32) the separation of the two clauses into one main clause
and one non-main clause is more pronounced and signalled by a gram
switch.63
4.4 Summary
Gram switching may, thus, be used as the only marker of such diverse types
of clause combining as the conditional and the relative, and it covers the full
spectrum of what has been labelled circumstantial, i.e. logical relationships
such as a condition, a cause, a result, or an interpretation of the head clause as
well as descriptive relationships such as providing additional or background
information about time, place, or manner of the head.64 Since final clauses are
also covered, there is no need to posit a specific class of ḥāl muqaddar to
cater for “circumstantial clauses” that fall outside of the pattern; they all be-
long to the same general system of marking of hypotaxis through a switch of
grams.
The discovery of gram switching is no magic wand that will resolve all
ambiguity in Semitic clause combining. Ambiguity is part of normal language
use. It does, however, resolve some ambiguity and, more importantly, enables
us to look at non-main clause combining from a new and more comprehensive
perspective.
63 See Persson (2013) for a discussion on auxiliation and the possible occurrence of verb
serialization in Arabic.
64 Cf. Persson (2009, 238-240).

90 Maria Persson
5. The relevance of form. Bloch’s three types revisited

5.1 Introduction
Bloch’s formal approach mirrors the way in which “[m]ost Arabic Grammars
(…) [have] been content with giving a very general characterization (‘at-
tendant circumstance’ or the like) together with an enumeration of examples
of ḥāl constructions...” (Isaksson 2009a, 5). The data presented above demon-
strate that the idea of the circumstantial clause as a distinct syntactic category
in Arabic cannot be upheld.
Nevertheless, a classification based on formal criteria does not appear un-
less there is a considerable quantity of clauses that conform to a set number of
patterns and that seem to have the same function. Arab grammarians, and
grammarians of Arabic, have noticed a large number of clauses with more or
less adverbial function and found structures commonly used for these. Though
it is now obvious that they have been a bit too quick to place a label on them
and assume a distinct class of clauses, the reason for their discovery is still
valid. Although a coherent syntactic class of “circumstantial clauses” cannot
be established, there are specific structures that, more often than not, are
found with clauses enhancing other clauses circumstantially. Therefore, when
the clauses for this survey had been collected, I investigated how many of the
clauses in the database conformed to the three structural types described by
Bloch. All in all, I found eighty instances that conform to Bloch’s Type 1
pattern65 in which he also includes clauses with participial predicates as well
as what he calls “verselbstständigter Zs”66, the latter largely corresponding to
what I here have labelled ‘clausal circumstantials at the text level’. There are
also five instances of his Type 2 pattern67 and twenty-four instances of his
Type 3 pattern68 adding up to a total of 109 instances in the database of claus-
es that meet the formal requirements for a circumstantial clause suggested by
Bloch. This means that just over half of the collected database corresponds to
Bloch’s definition of a circumstantial clause.
Basic examples of Types 1-3, drawn from the present database, were given
in the introduction to this article (examples 1-3). Below, I will discuss addi-
65 Type 1: (w-) + subject + predicate (verbal or not verbal).

66 I.e. “Circumstantial clauses that have become ‘independent’” (Bloch 1965, 74).
67 Type 2: (w-) + predicate (prepositional phrase) + subject.
68 Type 3: finite verbal predicate.

tional examples showing possible variation, mainly within Types 1 and 3,

which Bloch recognizes, but also problems posed by the formal criteria he
posits.
5.2 Type 1. (w-) + subject + predicate (verbal or not verbal)

5.2.1 General remarks
Bloch’s Type 1 contains the most syntactically marked of the types of circum-
stantial clauses that have traditionally been recognized: the clause introduced
by w(ə)- followed by a pronoun:
(43) tšaḥḥaret ktīr [w-hīye mablūle bəlmayy] (32:4:190)
become.sootySF3fs much [w-3fs soakPPfs PREP.DEF.water]
‘It (the cat) became very sooty [(as) it was soaked by the (rain-) water]’
(44) ḍallēna mədde mn əzzaman [w-nəḥna ʿala haššaġle] (1:97:38)
staySF1cs period PREP DEF.time [w-1cp PREP DEM.DEF.work]
‘I stayed some time [(being) in this business]’
The use of the pronoun here specifically indicates that the non-main clause
describes one of the participants of the main clause. This fact will receive
further attention below (5.7). However, not many clauses are so clearly
marked. This means that this marking, though typically marking what has
been perceived as a circumstantial clause, cannot, in any sense, be perceived
as a marker of a class of circumstantial clauses. It marks a fragment of what
has, traditionally, been labelled circumstantial.
Nineteen of the instances in the database are “verselbstständigter Zs” or
what I have chosen to label ‘clausal circumstantials at the text level’, i.e they
do not refer back to a specific head clause. Because of this rather significant
structural difference, they will be treated separately. Of the sixty-one remain-
ing Type 1 examples that are recorded in the present database, twenty-five are
introduced by w(ə)- and thirty-six are not. In eighteen of the twenty-five that
are introduced by w(ə)-, this w(ə)- is immediately followed by a noun or a
pronoun. This means that only eighteen of the sixty-one clauses feature the
double marking of w(ə)- and a preposed subject.69
69 Six of the nineteen circumstantials at the text level have this double marking, i.e. in both
groups these constitute about 30% of the cases. The number of clauses is, however, too
small for any reliable conclusions to be drawn.

92 Maria Persson
Looking instead at the types of predicates included, it turns out that in all
but four of the thirty-six clauses that are not introduced by w(ə)- the predicate
is a participle. When preceded by a noun or pronoun, participial predicates
are, however, among the clauses introduced by w(ə)-.70 This indicates that the
use of w(ə)- may not be optional after all, but conditioned by the form of the
predicate. This will be further discussed below. Other types of predicates
found after w(ə)-, with or without a preposed subject, are verb-less (noun)
clauses, finite verb clauses, and clauses following an expression with ṣār l(i)-.
Bloch also mentions preposed circumstantial clauses as belonging to Type 1.
However, as will be seen below, the data base contains several preposed cir-
cumstantial clauses that do not conform to Bloch’s criteria for Type 1.
5.2.2 Finite verbs

Only eleven of the sixty-one instances of Type 1 are verbal clauses with finite
verbs. Four of these are preposed and will be discussed further on. Of the
remaining six, all but one are preceded by w(ə)-. The one example without
w(ə)- is (45):
(45) ʾāmet ʾallaʿto [nəḥne ʿam nərʾoṣ] (8:6:70)
71
AUXSF3fs oustSF3fs.3ms [1cp PROGR dancePF1cs]
‘She threw him out [(as) we were dancing]’
Likewise, of the six, all but one contain the pre-verbal particle ʿam and, thus,
are similar to (45) in all but the use of w(ə)-:
(46) štaġalna mədde mn əzzaman [w-nəḥna ʿam nəštəġel ʿala ḥsābna
bəṣṣafḥa] (1:102:38)
workSF1cp period PREP DEF.time [w-1cp PROGR workPF1cp PREP invoice1cp
PREP.DEF.page
‘I worked for a while, [working as a self-employed freelance (paid) per

page]’72
The one example where the verb is not preceded by ʿam is (47):
(47) laḥatta ʾarīb əlʿaṣər fənyet rəžlayyi [w-ʾana ʾəftel mən ʾərne laʾərne] -
laḥatta ʾalla baʿatli ṣāḥeb maṭbaʿa … (1:63:24)
70 See for example (55) below.

71 The verb used as an auxiliary literally means “she stood up”. Cf. 4.3.3 above.
72 Lit: working on my own invoice, per page.

PREP.CONJ close DEF.afternoon be.spentSF3fs legs1cs [w-1cs spinPF1cs PREP cor-

ner PREP corner] – PREP.CONJ God sendSF3ms.PREP1cs owner print-shop …
‘Until, in the late afternoon 73 , my feet were worn out [(from) whirling
from one place to the other]74 – until God sent me a print-shop owner…’
All the examples with finite verbs feature a gram switch. In four instances,
such as the examples quoted here, the switch is one between a suffix form and
a prefix form; in the remaining two it is a switch from either a participle or a
bare prefix form to a prefix form preceded by ʿam.
The clauses are temporal, expressing a simultaneous activity or event
(while/as) such as (45), or they describe the manner in which something is
done or experienced such as (46) as well as (1) above. They may also have a
meaning that is between the two such as (47).
5.2.3 Non-verbal predicates

The texts include six examples of Type 1 where the circumstantial clause is
entirely non-verbal. One of these is preposed (see below) and all are intro-
duced by w(ə)- such as (48):
(48) sakkar əṭṭāwle [wəddaʾʾ bʾīdi] (18:5:142)
closeSF3ms DEF.board [w-victory PREP.hand1cs]
‘He folded the board [(when) victory was in my hand]’
While the non-main clause in (48) is temporal, most of the other examples in
the database are descriptive such as (49):
(49) ṭəleʿ əssabʿ mn əlʾafaṣ [w-ʿalāyem əlġaḍab ʿalē] (28:4:186)
exitSF3ms DEF.lion PREP DEF.cage [w-signs DEF.anger PREP.3ms]
‘The lion left the cage [with (showing) signs of anger75]’
The meaning of the Type 1 non-verbal circumstantial may also be a combina-
tion of both:
(50) wṣəlna [wəddənye ʿətme] (1:35:14)
arriveSF1cs [w-DEF.world dark]
‘We arrived [and/when it was dark]’
73 Lit: close to (time of) the afternoon prayer.

74 Lit: and I spin from corner to corner.
75 Lit: and signs of anger (were) on him

94 Maria Persson
In (50), the statement that it was dark works both to locate the event in time
and to describe the scene. The narrator goes on to comment on the lack of
electricity in the village. In five of the examples there is a gram switch from a
suffix form verb to the non-verbal clause, in the sixth example the main clause
contains a combination the copula verb kān in suffix form and a participle.
5.2.4 Circumstantials following an expression with ṣār l(i)-

Within the Type 1 group, Bloch recognizes that an expression with ṣār l(i)- +
noun/pronoun may be the head of a circumstance:76
(51) ṣār lak šahr ramaḍān ṭūlo [ġāyeb ʿanhon] (17:19:130)77
happenSF3ms PREP2ms month Ramadan lenght3ms [be.absentAPms PREP3cp]
‘all of Ramadan has passed for you [(being) away from them]’
(52) ʾana ṣār li ḫamsīn səne [mṣawwer] (27:3:184)78
1cs happenSF3ms PREP1cs fifty years [photographer]
‘Fifty years have passed for me [(being) a photographer]’
Only five examples of this were found in the texts. As in (51, 52) the predicate
in all but one is a participle. In one, the participle is preceded by w(ə)- + pro-
noun:
(53) ṣār lo tmānā-w-ʿəšrīn yōm [w-hūwe lāzeʾ bəlbēt] (17:4:126)
happenSF3ms PREP3ms eight-w-twenty days [w-3ms be.gluedAPms
PREP.DEF.house]
‘Twenty-eight days have passed for him [(while) he has refused to leave
the house]’79
The one example with a finite verb instead of a participle consists of a negated
progressive:
(54) w-ʾana ṣār li tlət əsnīn – [māli ʿamma bəštəġel bhaṣṣanʿa] (1:101:38)80
76 Cf. also Grotzfeld (1965, 102). The expression consists of a combination of the verb ṣār,
‘happen’, and the preposition l(i)-, ‘for’, ‘to’, plus a pronoun or noun expressing the per-
son to whom the period of time had elapsed. Whether this time span is long or short, the
activity or event expressed in the clause must be of such a kind that it can have been on-
going for all that time; such as is expressed by a stative or a progressive.
77 Cf. Bloch (1965, 71).
78 Cf. Bloch (1965, 71).
79 A more literal translation would be “and he is glued to the house”.

w-1cs happenSF3ms PREP1cs three years – [NEG1cs PROGR b.workPF1cs

PREP.DEM.DEF.profession]
‘Three years had passed for me [(during which) I had not worked/while I
was not working in this profession]’81
Thus, in all the examples there is a gram switch. The combination of the nega-
tion and the progressive in (54) creates the sense of stativity that is also im-
plied in the participles and, I would assume, a necessary collocation for an
expression such as ṣār l(i)-.82
5.2.5 Participial predicates

With twenty-six instances of participle alone (whereof one introduced by
w(ə)-), and with seventeen examples of pronoun/noun+participle (whereof ten
introduced by w(ə)-), participial predicates constitute the most common varie-
ty (2/3) of what has been counted as Bloch’s Type 1 in my collected data. An
example of the (w-) + pronoun/noun + participle variety is (55):
(55) w-ʾəmm rəšdi raḥa tṭəʾʾ əlməskīne [w-hīye sākte] … (17:4:126)
w-mother Rəšdi FUT burstPF3fs DEF.poor [w-3fs be.quietAPfs]
‘and Umm Rushdie was on the verge of exploding, poor one, [and/but she
was quiet]’
The corpus, as said, contains thirty-six clauses that conform to Type 1 that are
not introduced by w(ə)- and, of these, thirty-two are participial. The reason for
the lack of conjunction here is structural. Consider (56-57):
(56) bšūf ənnās [rāyḥa žāye] (1:6:2)
b.seePF1cs DEF.people [goAPfs comeAPfs]
‘I see people [coming and going]’
(57) ʾərkod83 ʿala halbēt ʿala ḍaww ʿalbēt [mabṣūṭ] (1:17:6)
runPF1cs PREP DEM.DEF.house PREP light PREP.DEF.house [happy]
‘I would run [happy] to the house in daylight’
80 Cf. Bloch (1965, 70, 72).

81 As in all examples here the aim is not for idiomatic English translations but translations
that are fairly close to the Arabic original. An even more literal translation would be
“me not being at work in this profession”.
82 Cf. footnote 74 above.
83 Sic!

96 Maria Persson
To insert w(ə)- here is hardly conceivable.84 Since w(ə)-, then, is not an op-
tion, there is no overt marking in most of these clauses. It is the gram switch
alone, the switch from a finite verb to a participle, which signals the non-main
clause.
5.2.6 “Preposed circumstantial clause”

Bloch also mentions the possibility of preposing the circumstantial clause, a
construction that Kaye and Rosenhouse (1997, 308) suggested is especially
common in colloquial Arabic.85 Brustad explains the phenomenon as a case of
“frame-setting” or topicality:
When circumstantial clauses occur in sentence-initial position, they

function to give a temporal and descriptive frame within which the ac-
tion of the main verb takes place. (Brustad 2000, 340)
Bloch only recognizes the existence of Type 1 preposed circumstantial claus-

es. However, there are several examples in my database of preposed circum-
stantial clauses that do not conform to Bloch’s criteria. Of the twenty-nine
preposed circumstantial clauses I have collected, sixteen conform both to
Bloch’s description of Type 1 and the added comment that he gives that only
pronouns, no nouns, are found to introduce these. These will be discussed in
5.2.6.1 followed by a survey of those that do not conform to Bloch’s descrip-
tion.
According to Bloch, the use of w(ə)- is optional. It is possible that Bloch
did not do a statistical survey of this; he found that both options existed and
84 See Bloch (1965, 71). I have found one possible exception to the rule that a participial
circumstantial cannot be introduced by w(ə)-:
(i) lakān hēk bəddna nḍall ḥāṭṭīno ġēme sōda bəlbēt [w-ʾāʿdīn]? (17:52:140)
then thus wish1cp stayPF1cp putAPcp3ms cloud black PREP.DEF.house [w-
sitAPcp]
‘should we then stay like this, having put him in the house as a black cloud and
[(remain) sitting (doing nothing)]?’
As this is the only exception, and as the subject is mentioned shortly before both as a
verbal inflection (nḍall) and, implicit, in the plural of the previous participle, I believe it
is reasonable to posit a deleted pronoun here (w-nǝḥna ʾāʿdīn [w-1pc sitAPpc] ‘us sit-
ting’).
85 Cf. also Blau (1960, 225) and Rosenhouse (1978, 229).

concluded that the w(ə)- is optional.86 In my survey of the texts, however, I

find that the cases without an introductory w(ə)- are much more common than
those that are introduced by w(ə)-. The whole database contains only five
examples of preposed circumstantial clauses introduced by w(ə)-. Conspicu-
ously, all of these are found among the sixteen clauses conforming to Bloch’s
Type 1 pattern. This still leaves the majority even of this limited group (elev-
en examples) without an introducing w(ə)-.
Brustad suggested, in the above quote, that these clauses set a temporal and
descriptive frame to the main activity or event. It seems, however, that the
tendency for temporal frame setting is the strongest. In the modern urban Gulf
Arabic data most of the preposed circumstantial clauses express time (Persson
2009, 276-277). In the present corpus, all the preposed circumstantial clauses
are temporal. This indicates that this particular construction functions, at least
in the two surveyed types of Arabic, as a specific type of temporal clause.
After this introduction I will proceed to a survey of the clauses found in the
database. I will first describe the sixteen clauses that do conform to Bloch’s
criteria and then proceed to discuss the thirteen clauses that do not conform to
this type.
5.2.6.1 “Preposed circumstantial clauses” that do conform to Bloch’s

criteria
Examples (58-60) are among those that conform to Bloch’s Type 1:
(58) [w-nəḥna ballašna nəštəġel] tətḥassan ʾəžrətna (1:104:40)
w-1cs startSF1cp workPF1cp] improvePF3fs salary1cs
‘[(As) I started to work] my salary was (increasingly) better’
86 Another possible reason would be that Bloch’s definition of circumstantial clauses has
led to a different selection than that of the present study. While this may be true, it is
worth noting that all but one of the clauses included in the present study have been
translated by Bloch and Grotzfeld (1964) as temporal (als/wenn…) or simultaneous
(während). The one exception is an example translated as a conditional:
[bikūn hassammān māsek halʾannīne ʿamma ykǝtt bǝlʾannīne hadīke zēt] byəmroʾ hāda
ššēḫ (25:2:180), translated as “Hatte der Krämer eine Flasche in der hand gehalten, in-
dem er in diese Flasche Öl goß, so kam dieser Scheich vorbei ”. In other words the
translation is “If … had a bottle in his hand, [then!] this sheikh would pass by”. As seen
from my insertion of “then” the clause may be compared to other clauses included in the
discussion in 4.2.2 on the ambiguity between the temporal and the conditional.

98 Maria Persson
(59) [w-hū rāked] ma laʾa ʾəlla ʾārme maktūbe lḥākem qaraqāš (20:5:166)
[w-3fs runAPms] NEG findSF3ms except sign writePPfs DEF.judge Qaraqash
‘[(As) he was running] he suddenly found a sign that said “judge Qara-
qash”’
(60) [w-hūwe ʿam yfakker] šāf bāb əssabʿ nfataḥ (28:4:186)
[w-3ms PROGR thinkPF3ms] seeSF3ms door DEF.lion be.openedSF3ms
‘[(As) he was thinking] he saw the door of the lion being opened’
As is seen from these examples, the double marking, where the clause is in-
troduced by w(ə)- followed by a pronoun, is also found in the preposed claus-
es. However, as said above, most of the preposed clauses do not feature a
w(ə)-. Example (61) has a participial predicate and, thereby, represents the
most common type (ten of the sixteen examples):
(61) [hūwe māši bǝššāreʿ] šāf sərk (28:1:184)
[3ms walkAPms PREP.DEF.road] seeSF3ms circus
‘[(As) he walked in the street] he saw a circus’
A less common variety (four examples) has a finite verb:
(62) [hūwe ʿam yġanni] - ma kənna nəfham ʿalē nəḥna (4:5:50)
[w-3ms PROGR singPF3ms] – NEG beSF1cp understandPF1cp PREP3ms 1cp
‘[(While) he was singing] we did not understand him (i.e. what he was
singing)’
There is also one instance of a non-verbal predicate:
(63) [w-hūwe binǝṣṣ ǝlḥlāʾa] fāt kalb (30:3:188)
[w-3ms PREPhalf DEF.shaving] enterSF3ms dog
‘[(As) he was halfway through shaving] a dog entered’
In addition to these, there is one example with bədd-:
(64) [ʾana bǝddi rūḥ laʾand 87 žamāʿti lahnīke] - yəṣraḫu ʿalīyi hadōle
(4:13:52)
[1cs wish1cs goPF1cs PREP.LOC group1cs PREP.there] – shoutPF3cp PREP1cs DEM
87 Sic!

‘[(As) I want to go to my group over there] those (people) are shouting at

me’
Although clauses with bədd- are generally counted among Type 3 by Bloch
(see 5.4 below), I deem this to be closer to Type 1 as it is introduced by a
pronoun.88 Bloch (1965, 73-74) also mentions cases such as (65) where the
preposed circumstantial is followed by a more or less frozen non-verbal
phrase with a nomen vicis of a locational verb:
(65) [w-nǝḥna ʿam nǝštǝġel] - bžayyet wāḥed taḥṣaldār (4:16:54)
[w-1cs PROGR workPF1cs] – PREP.comeVN one tax.collector
‘[(While) we were working] there came a tax collector’
As may be seen from the glosses, in all these examples there is a gram switch.
Another peculiarity found within this group of clauses, but one that Bloch
does not mention, is the placement of w(ə)- in the following examples:
(66) fī wāḥed doktōr ʾāʿed bməstašfa lmažanīn - [māreʾ fi sāḥet əlməstašfa]
w-ʾām ṣadaf wāḥed mažnūn. (43:1:198)
there.is one doctor sitAPms PREP.hospital DEF.lunatics – [walkAPms PREP square
89
DEF.hospital] w-AUXSF3ms stumble.uponSF3ms one lunatic
‘There was a doctor (located) at the lunatic asylum – [(as) he was walking
in the hospital square] he happened to meet someone mad.’
(67) [hūwe ʿamma yǝrfaʿlo ʿam ysarreḥ ǝššaʿrāt] - (…) - w-ʾāmet nʾarṭet
šaʿra rāḥet (37:2:194)
[3ms PROGR liftPF3msPREP3ms PROGR combPF3ms DEF.hairs] - (…) – w-
90
AUXSF3fs be.cut.offSF3fs hair goSF3fs
‘[(As) he was lifting his hair up combing it] – (…) – one hair broke and
was gone’
The use of the w(ə)- here is interesting. Obviously, the circumstance in (66) is
that the doctor was walking in the main area of the hospital and it was “as he
was walking…” that he bumped into one of the patients “a crazy one”. This
88 Theoretically, this could also be an example of Type 3 where the pronoun (ʾana, ‘I’) is
topicalized and, hence, the ‘real’ circumstantial, anyway, starts with bədd-. Such an in-
terpretation is, however, not supported by the context.
89 Lit: he stood up. Cf. 4.3.3 above.
90 Lit: she stood up. Cf. 4.3.3 above.

100 Maria Persson
also has the form of an ordinary preposed circumstantial. Yet, it is the second
clause that has received a w(ə)-.
Similarly in (67), it was while the barber was trying to create some kind of
hairdo with the customers remaining three hairs that the (one) hair broke. The
w(ə)- added to the second (main) clause is certainly not a coordinative con-
junction but expresses the simultaneity inherent in the circumstantial. Yet, the
placement of this w(ə)- seems to have become skewed as the speaker chose to
use a “preposed circumstantial clause” - where use of w(ə)- is not very com-
mon in the surveyed texts.
5.2.6.2 “Preposed circumstantial clauses” that do not conform to

Bloch’s criteria
As said, of the twenty-nine preposed circumstantials I have collected, sixteen
conform to Bloch’s description of Type 1, to which he adds the comment that
only pronouns, no nouns, are found to introduce the preposed clauses. How-
ever, in the remaining thirteen instances found in the corpus, both w(ə)- and
the pronoun are missing. For some of these, alternative interpretations are
available, but some clearly have the same enhancing function as the sixteen
cases discussed above.
a) Noun instead of pronoun

The smallest deviation from Bloch’s description of what a ‘preposed circum-
stantial clause’ in Damascene Arabic should look like is found in (68):
(68) [ʾabu ḥamdo rāžeʿ mn əlʾahwe] ləssa ma fāt mən bēto wəlla ʿabʾet ʿalē
harrawāyeḥ (21:15:170)
[Abu Hamdo returnAPms PREP DEF.café] yet NEG enterSF3ms PREP house3ms ex-
cept pressSF3fs PREP3ms DEM.DEF.fragrances
‘[(As) Abu Hamdu was returning from the café] even before he entered his
house the fragrances overwhelmed him’
Bloch states that only pronouns may introduce a preposed clause. Here, how-
ever, and in one other example, we find (the name) Abu Hamdo at the head of
the clause. Similarly in (69) the clause is introduced by a noun:
(69) [ǝlmfatteš ʿamma yāḫod ǝlbīlēt mn ǝrrǝkkāb] - žtamaʿ bṣāḥəbna
(29:2:186)
[DEF.inspector PROGR takePF3ms DEF.ticket PREP DEF.passengers] – meetSF3ms
PREP.friend1cs

‘[(As) the inspector was collecting the ticket(s) from the passengers] – he
came across our friend’
It is possible to regard this circumstantial expression, not as preposed, but as
following the previous clause, which is how Bloch and Grotzfeld have inter-
preted this paragraph.91 The full context, with a rather literal translation, is as
follows:
(69b) wəṣel laryāʾ - [ǝlmfatteš ʿamma yāḫod ǝlbīlēt mn ǝrrǝkkāb] - žtamaʿ
bṣāḥəbna (29:2:186)
arriveSF3ms PREP.Ryāʾ - [DEFinspector PROGR takePF3ms DEF.ticket PREP
DEF.passengers] – meetSF3ms PREP.friend1cs
‘he arrived in Ryāʾ - [the inspector is collecting the ticket(s) from the pas-
sengers] – he came across our friend’
To my understanding, it is equally, if not more, plausible to see the two latter
events as a unit, i.e. that the conductor reached the protagonist’s seat while
collecting the tickets. No matter the reading, the prefix form verb in the in-
serted clause creates a gram switch in the story otherwise told with suffix
forms. By this, the non-main clause is marked as such.
b) Participial clause with no preposed subject

In (70) there is neither a pronoun nor a noun subject to introduce the preposed
circumstantial:
(70) [šāleḥ əlfarde ttānye] tzakkar ʾənno … (26:5:182)
[take.offAPms DEF.unit DEF.second] rememberSF3ms COMP…
‘[(as) he had taken off the other one (of his pair of boots)] he remembered
that …’
The preposed clause starts with the predicate in the form of a participle. A
gram switch occurs as this participial clause is followed by a suffix form verb.
c) Prefix form verb with ʿam or b-

There are also instances with prefix form verbs (Bloch’s Type 3, see 5.4 be-
low) in preposed position:
91 “Er kam in Ryāʾ an, da nahm der Schaffner die Billets von den Passagieren. Er traf
unsern Freund und…”. I.e. “He arrived in Ryāʾ; then the conductor took the tickets from
the passengers. He met our friend and…” (1964, 187).

102 Maria Persson
(71) w-mnənzel bšwayyet ṭarḫūn - [btəġli lkəbbe] mənḥəṭṭəlha šwayyet naša

(16:25:124)
w-b.take.downPF1cp PREP.some tarragon – [b.boilPF3fs DEF.kebbeh]
b.putPF1cpPREP.3fs some starch
‘and we put in some tarragon - [(when) the kebbeh boils] we add some
starch to it’
Interestingly, it is only among the preposed circumstantials that I have found
examples in the texts of Type 3 where the verb has a b-prefix. In all other
examples of Type 3, the verb is preceded by the progressive particle ʿam- (cf.
5.4). Even more interesting, perhaps, is the fact that (71) is one of three exam-
ples in the database, only, of a preposed circumstantial without a gram
switch.92 Context alone gives the temporal meaning here. One of the other two
examples without a gram switch is also interesting since it contains a main
clause that is surrounded by circumstantials. The second of these, the Type 2
[maʿo ḥmār], was mentioned in 3.3 above as example (12). The first circum-
stantial, however, is preposed. The example is repeated here as (12b) with
both circumstantials marked by square brackets:
(12b) [hūwe māši bhaṭṭarīʾ] - māši wāḥed [maʿo ḥmār] w-ʿabāyto ṭawīle
(20:1:166)
[3ms walkAPms PREP.DEM.DEF.road] – walkAPms one [PREP.3ms donkey] w-
cape3ms long
‘[(as) he was walking on the road] – there was one walking [(having) with
him a donkey] and his cloak was long’
Not only is there no gram switch as both predicates are participles; the predi-
cates are also identical (māši). Thus, what creates the sense of a preposed
circumstantial here, in addition to context, is the pronoun that introduces the
clause.
Returning to examples with gram switching, there are some where a com-
bined tense (auxiliary + main verb) causes the gram switch:
(72) [bǝtkūn hannaṣbe ġalyāne bmayyet ǝlḥamḍ bmayyet ǝlbanadōra]
mənsaʾʾeṭ halkūsa hēk layəstəwi (16:3:116)
b.bePF3ms DEM.DEF.broth boiling PREP.water DEF.lemon PREP.water DEF.tomato]
b.drop.downPF1cp DEM.DEF.zucchini thus CONJ.maturePF3ms
92 The other two feature participles in both clauses.

‘[(When) the broth with lemon juice and tomato juice is boiling] we drop
down the zucchini like this so that it will become done’
Here, again, there is a b-prefix, this time on the copula verb kān.
d) Suffix form
There are two examples in the texts of preposed circumstantials with a suffix
form verb. In one of the examples the suffix verb is a main verb:
(73) [stawu baʾa] - nazzəlha lahaṭṭanžara baʾa … (16:11:118)
[become.doneSF3cp PTCL] – putIMP2ms PREP3fs PREP.DEM.DEF.pot PTCL
‘[Then, (when) they are done], put into the pot…’
The main clause predicate nazzəlha, ‘put into it’, is an imperative which caus-
es a gram switch. In the second example there is a combined tense where the
suffix verb is an auxiliary combined with a main verb in the bare prefix form:
(74) [ṣāru yǝġlu baʾa] ġaṭṭēnāhon (16:11:118)
[AUXSF3cp93 boilPF3cp PTCL] coverSF1cp3cp
‘[then, (when) they boil] we cover them’
The main clause verb ġaṭṭēna, ‘we cover(ed)’ is a simple suffix form verb,
which means that there is a gram switch in this example, too.
e) Type 3 with bədd-

Finally, the database also includes one example of a preposed circumstantial
introduced by bədd- that conforms to the kind that Bloch counts as Type 3 (cf.
5.4):
(75) [bəddo yəmši] ʾallo taʿa lahōn ya ʾaḫūna (22:9:174)
[wish3ms walkPF3ms] saySF3ms.PREP3ms comeIPM2ms PREP.here VOC broth-
er1cp
‘[(when) he wanted to go] he said to him, “Come here my friend!”’
As mentioned above in 5.2.6.1, there is yet another example of a circumstan-
tial with bədd- that does not fit the patterns, namely (64) above.
93 Lit: They became. Cf. 5.2.4 above.

104 Maria Persson
5.2.7 Clausal circumstantials at the text level. Bloch’s

“verselbstständigter Zs”
When Isaksson (2009b, 103-105) coined the term “circumstantial qualifiers on
the text level”, and pointed out the hitherto unnoticed use of specific marking
for these, the discovery itself of clauses that coincide in form with clauses that
appear in various lists of ‘circumstantial clauses’, but that qualify a whole
piece of text instead of (only) a particular clause, was not new. Bloch (1965,
74) described their occurrence in the Damascene data and called them
“verselbstständigter Zs” i.e. circumstantial clauses that have become inde-
pendent. Bloch also remarks that these clauses, such as (76), tend to function
as general background remarks set off from the main line of the discourse:
(76) ṣār yəġsəlha - [wəddənye šəte bard] - ṣār yəġsəl halʾəṭṭa (32:5:192)
94
AUXSF3ms washPF3ms.3fs – [w-DEF.world winter cold] - AUXSF3ms
washPF3ms DEM.DEF.cat
‘(and) started to wash her – [and it was winter and cold] – he started to
wash the cat’
Note that the expression ṣār yəġsəlha, ‘he started washing it’, in (76) is re-
peated on both sides of the clause to mark the beginning and the end of the
digression from the main line of the story. Bloch comments on this use of
what he perceived as the circumstantial clause type, and even mentions the
discourse-marking function of the repeated words. Nevertheless, he does not
seem to have noticed the gram switch that marks the clause as a digression
from the main line even when it is not introduced by w(ə)-:
(77) kəll yōm yəṣrəfni – [əddənye ṣēf wənnhār ṭawīl] - yəṣrəfni ʾabl əlməġreb
(1:17:6)
every day dismissPF3ms.1cs – [DEF.world summer w-DEF.day long] – dis-
missPF3ms.1cs before DEF.sunset
‘every day he would dismiss me – [it was summer and the day was long] -
he would dismiss me before sunset’
(78) ʾəžu ḍayyafūni bʾadaḥ ʿaraʾ zġīr - [ʾana ma bəšrab ʿaraʾ] - ʾalabton
(7:9:68)
95
AUXSF3cp entertain.withSF3cp.1cs PREP.glass arak small – [1cs NEG
b.drinkPF1cs arak] – tiltSF1cs.3cp
94 Lit: he became (both instances in this example). Cf. 5.2.4 above.

‘they gave me a small glass of arak – [I don’t drink arak] – I tossed it

down’
In all the examples quoted here there is a gram switch. In (76-77) that are also
mentioned by Bloch, the switch is from verbal clauses too verb-less clauses;
in (78) the story line is told with suffix form verbs moving the story forward
and the digression is marked by the use of a prefix form verb.
5.3 Type 2: (w-) + predicate (prepositional phrase) + subject

Type 2 is, in Bloch’s description, rather cohesive and does not receive much
attention. Moreover, with only five instances, it is almost non-existent in the
texts under survey. My collection of Type 2 clauses contains two examples
that entirely coincide with the ones given by Bloch, one of them cited as (2)
above. Of the other three, one is without w(ə)- :
(79) marra – kənna sahrānīn bəllādʾīye [maʿna wāḥed darwīš šwayye]
(9:1:70)96
time – beSF1cp spend.the.eveningAPmp PREP.Latakia [PREP1cp one poor some-
what]
‘one time – we were hanging out in Latakia in the evening [(having) with
us someone who was a somewhat gullible (lit. poor)]’
The gram switch in this clause is one from a combined verb form with the
copula verb kān to the non-verbal clause of the circumstantial. The next Type
2 clause contains a slightly more complex clause:
(80) ənsaḥabna mən halḫān - [w-bwəššna nlāʾīlna (…) ši ʾərne nətʾāwa fīha
šwayye] - labēn ma ʾənno təmši ḥarket əlbalad (1:40:16)
withdrawSF1cp PREP DEM.DEF.inn – w-PREP.face1cs findPF1cp.PREP.1cp (…) some
corner retreatPF1cs PREP3fs somewhat] – until REL COMP walkPF3fs movement
DEF.city
‘We withdrew from this inn – [and we intended/intending97 to find (…)

some corner to hide away in for a bit] – until the city started to move’
95 Lit: they came. Cf. 4.3.3 above.

96 Bloch (1965, 75).
97 Lit: in our face.

106 Maria Persson
The prepositional expression, which makes up the Type 2 circumstantial,

consists of a w(ə)- plus a noun creating a metaphor: “in our face (was)” mean-
ing “we set before us to”, followed by a verbal phrase.
The fifth instance of Type 2, mentioned also by Bloch (1965, 75) does not
contain any new information:
(81) šažžaʿət ḥāli - halmarra ṭāleʿ laḥāli [ma maʿi rfīʾ] - šažžaʿət ḥāli w-fətt
(1:56:22)
encourageSF1cs myself – DEM.DEF.time set.outAPms alone [NEG PREP1cs friend]
- encourageSF1cs myself w-enterSF1cs
‘I took courage, this time I had gone on my own, no friend was with me;
I took courage and entered…’
The phrase ma maʿi rfīʿ, ‘my friend was not with me’, which corresponds to
Bloch’s Type 2, is an appositive to laḥāli, ‘on my own’. Both are added com-
ments that constitute remarks on the circumstances surrounding the expedi-
tion. The remark in the second, that no friend was with the narrator, is, how-
ever, not a circumstance of his going on his own but a restatement of the
same, in Dixon’s words a “same-event addition” (2009, 27). Consider the
following, similar example:
(82) waʾʾafūna žžēš – kəllīyātna musallaḥīn [maʿna msaddasāt] (5:3:60)
stopSF3cp.1cp DEF.army – all1cp armPPcp [PREP1cp guns]
‘The army stopped us, we were all armed [(having) guns with us]’
Bloch gives a temporal-simultaneous interpretation to the phrase waʾʾafūna
žžēš – kəllīyātna musallaḥīn: “das Militär hielt uns an, während wir alle be-
waffnet waren und Revolver mit uns führten.” (my emphasis). The preceding
clause, kəllīyātna musallaḥīn, ‘all of us were armed’, is according to Bloch
(1965, 70) a circumstantial of Type 1 (without w(ə)-). Bloch, later on (1965,
75) mentions the expression within brackets, maʿna msaddasāt, ‘having with
us revolvers’, as an example of Type 2 without w(ə)-. Hence, in his analysis,
this is an example of two (different) circumstantial clauses placed after one
another. Judging only from the German translation, however, Bloch and
Grotzfeldt consider these phrases to be of equal status and possibly, with me,
read the second as an appositive of the first. That the people had guns is not
the circumstances of their being armed; it is just another way of saying that
they were. Both clauses describe the circumstances surrounding the encounter
with the police.

More importantly, however, the circumstantial comment is far longer than

Bloch in his analysis, with his formal approach, has recognized:
(83) wṣəlna wəšš əṣṣəbḥ ʿand ʿala bəkra – waʾʾafūna žžēš – [kəllīyātna
musallaḥīn maʿna msaddasāt – w-kamān žamāʿa tḫān yaʿni w-zkərtīye
(…)98 w-kəllon tḫān yəlli maʿna] – waʾʾafūna žžēš ʿalʾasās yfattšūna –
(5:3:60)
arriveSF1cs face DEF.morning LOC PREP tomorrow - stopSF3cp.1cp DEF.army –
[all1cp armPPmp PREP1cp guns– w-also group thugs meanPF3ms w-
rowdies(…) w-all3cp thugs REL PREP1cp] - stopSF3cp.1cp DEF.army
PREP.grounds searchPF3cp.1cp
‘We arrived in the early morning the next day - the army stopped us, [we
were all armed (having) guns with us – and also a group of thugs, you
know, and rowdies (…) and all of them were thugs who were with us] –
the army stopped us in order to search us’
The whole passage, from the first instance of waʾʾafūna žžēš ‘the army
stopped us’ to the repetition of those same words, is a digression by the narra-
tor from the main line of the story in order to provide the listener with more
details, to describe the scene – the circumstances. The description of the cir-
cumstances, thus, contains much more than the fact that they were armed. The
way in which this digression from the main line is set off as such, by a repeti-
tion of the words waʾʾafūna žžēš ‘the army stopped us’, was recognized by
Bloch when he discussed “verselbstständigter Zs”(Bloch 1965, 74) and is also
described elsewhere (Persson, 2014b).
5.4 Type 3: finite verbal predicate

In his final category, Type 3, Bloch only includes finite verbs preceded by the
particles b- or ʿam such as (84) and (85):
(84) laʾa lʿālam [ʿamma tnādi] (31:1:188)
findSF3ms DEF.world [PROGR shoutPF3fs]
‘he found people [shouting]’
98 mən žamāʿet əššāġūr w-mən žamāʿet mādənt əššaḥm = from the group of Šāġūr and
from the group of Mādənt əššaḥm

108 Maria Persson
(85) laʾēna lʾarḍ [ʿam təlheb mətl ənnār] (1:73:28)

findSF1cp DEF.earth [PROGR burnPF3fs like DEF.fire]
‘we found the ground [burning like fire]’
Only three such examples were found in the database: these two and example
(3) above. Two of the three examples contain the doubly transitive verb laʾa
(classical Arabic: laqiya), ‘to find’, whereas, in the third (3), the head clause
features a combined tense with an auxiliary plus a participle.
The most common variety of Type 3 in the database, constituting twenty-
one of the twenty-four instances, is an alternative form of Type 3 which is
mentioned by Bloch: the finite verb introduced by bədd+pronoun. The noun
bədd means ‘wish’ and the pronoun added to a noun is possessive.99 Hence,
these clauses are final, expressing intention:
(86) w-ʾəz ətḥarrak ərrəžžāl [bəddo ysāwīlna kāst əššāy] (1:44:16)
w-PTCL moveSF3ms DEF.man [wish3ms makePF3ms.PREP1cs glass DEF.tea]
‘and then the man moved [(wanting) to make us a cup of tea]’
Bədd is a noun and constitutes the subject of a non-verbal clause which, in all
the examples mentioned in this section, creates a gram switch. If the attached
pronoun has a referent other than the subject of the main clause, the meaning
is only final, with a mild imperative sense instead of a wish:
(87) hāda lḥmār byəmši biwaʾʾef - žāybo [bəddak tsāwīli yā]. (36:3:194)
DEM DEF.donkey b.walkPF3ms b.stopPF3ms – bringAPms.3ms [wish2ms
doPF2ms.PREP1cs ACC3ms]
‘this donkey (repeatedly) walks and stops – I have brought it [so that you
can/for you to fix it for me]’100
In addition to the 21 examples mentioned here, the database contains three
examples with bədd+pronoun that do not conform to the pattern. In two cases,
bədd- is preceded by a w(ə)-:
99 Literally: ‘my/their/our etc. wish is to…’ used to express ‘I/they/we etc. want to…’.
A suggested etymology is that it derives from the Classical Arabic bi-widdi x ‘in the
desire of x’ (Ambros 1977, 92).
100 The donkey, that refuses to walk as soon as the owner wants to ride on it, has stopped
outside a watchmaker’s shop and the owner likens it to a watch that constantly stops
working.

(88) baʾa wāḥed fāyet [w-bəddo yəshar] (9:5:72)

PTCL one enterAPms [w-wish.3ms spend.eveningPF3ms]
‘now, here someone has entered [wanting/intending to spend the evening]’
(89) kəll wāḥed ṭamar yəlli maʿo bhalʾarḍ [w-bəddo yənfod bʾēš? brīšo bəddo
yənfod] (5:4:60)
every one burySF3ms REL PREP3ms PREP.DEM.DEF.earth [w-wish.3ms
get.throughPF3ms PREP.what? PREP.feathers.3ms wish.3ms get.throughPF3ms]
‘every one buried what he had with him in the ground [wanting to escape
with what? Wanting to escape saving his skin]’101
As is often the case, w(ə)- introduces ambiguity between parataxis and hypo-
taxis. The next example deviates even more from the Type 3 pattern as both
w(ə)- and a pronoun precedes bədd+pronoun:
(90) baʾāl ənnōme ʿando bḫamsā w-ʿəšrīn līra – [w-ʾana bəddi bāt ʿala šara-
fo hallēle] (7:2:66)
PTCL DEF.sleep LOC.3ms PREP.five w-twenty pound – [w-1cs wish1cs
spend.nightPF1cs PREP honour3ms DEM.DEF.night
‘Now, a night at his place was twenty five pound – [and I was intending to
sleep at his expense102 this night]’
This example is similar to Bloch’s Type 1. The clause with bədd can be inter-
preted either as paratactic: “and I want to…” or as hypotactic: “I, however,
want to…”.
5.5 Bloch’s ‘circumstantial expressions’

Bloch stated two main exceptions when formulating his definition of circum-
stantial clauses. One was the ‘ḥāl muqaddar’, to which I will return in the
next section, and the other was what he called ‘circumstantial expressions’.
The reason for their exclusion was twofold. One reason was formal: they do
not always conform to any of his three structural forms. The second reason
was functional: they describe the details of a (whole) situation. One example
of these was given in 2.3 (example 4). Another one has been discussed later
on (4.2.7, example 38). They are both repeated here for reference:
101 Lit: with his feathers.

102 Lit: at his honour.

110 Maria Persson
(38) waʾʾafūna bʾǝrne [ʾafāna lǝlḥēṭ] (5:10:62)

stopSF3cp.1cp PREP.corner [backs1cp PREP.DEF.wall]
‘they placed us in a corner [with our backs to the wall]’
(4) rəḥt ʿalbēt - [ḥēṭ ʿam biṣəddni w-ḥēṭ ʿam yrəddni] (1:88:34)
goSF1cs PREP.DEF.house – [wall PROGR b.push.awayPF3ms.1cs and-wall PROGR
pull.backPF3ms.1cs]
‘I went home [one wall pushes me away one wall pulls me back
(=staggering from wall to wall)]’
In the light of the present study, there is no formal reason to exclude these
clauses. In both the examples the hypotaxis is marked by a gram switch from
a suffix verb to a verb-less clause (38) or to a clause with prefix verbs preced-
ed by ʿam and, in one case, also by b- (4). Moreover, both are, in fact, qualify-
ing other clauses, not whole situations. Therefore, in these two particular
instances, I also see no reason to exclude them from the discussion on the
functional grounds mentioned by Bloch. The same construction may, howev-
er, also be used for larger digressions at the text level where Bloch’s statement
applies.
5.6 ḥāl muqaddar

Bloch (1965, 81) also refutes the existence of the so called ‘ḥāl muqaddar’,
the ‘implied circumstantial clause’, in Damascene Arabic. His main reason for
this is the available parallel with la-, ‘in order to’ which, to him, shows that
these clauses are final, not circumstantial.
Interestingly, Bloch (1965, 76), as seen above, accepts the finality in claus-
es with bədd+pronoun and counts them as an alternative of his Type 3 clauses.
He calls the combination a verb compound (Verbaleinheit) and likens
bədd+pronoun with ʿam, saying that it functions as ‘a modifying particle’
(vermodifikatorischen Partikel).103 Bloch (1965, 76) notes that these examples
often have a purposive meaning. The semantics of purpose, or more literally
‘wanting’, is implied in the bədd+pronoun phrase. At least in some instances,
one could say that bədd functions as a pseudo-conjunction introducing a final
clause. Bloch further observes that these clauses often occur with main clause
verbs of motion. This would mean that they are parallel to the so called ḥāl
103 Cf. Ambros (1977, 92) who suggests that bədd- be analysed as a preposition or a
“pseudo-verb”.

muqaddar. Bloch’s motive for excluding the ḥāl muqaddar was, as men-
tioned, the possibility of a paraphrase with the preposition l(i)-.
Such a paraphrase is, however also possible with clauses introduced by bədd.
Cf. (86), which is repeated here for reference, with (86b):
(86) w-ʾəz ətḥarrak ərrəžžāl [bəddo ysāwīlna kāst əššāy] (1:44:16)
w-PTCL moveSF3ms DEF.man [wish3ms makePF3ms. PREP1cs glass DEF.tea]
‘and then the man moved (wanting) to make us a cup of tea’
(86b) w-ʾəz ətḥarrak ərrəžžāl [li-ysāwīlna kāst əššāy]
w-PTCL moveSF3ms DEF.man [CONJ-makePF3ms. PREP1cs glass DEF.tea]
‘and then the man moved to make us a cup of tea’
The only difference between (86) and (86b) is that the volitive (wish) flavour
of the intention is not explicitly expressed in the latter.
As an argument to exclude the asyndetically juxtaposed clause combina-
tions that have been called ‘ḥāl muqaddar’, Bloch also states that these and
circumstantial clauses are formally distinct in Damascene Arabic. As men-
tioned above, Bloch only includes prefix forms preceded by b- and ʿam
among the Type 3 circumstantials. This is because, according to him, the bare
prefix form marks the final clause.104 Now, it is correct that all final clauses in
the database are in the bare prefix form and this also corresponds to similar
findings in modern Gulf Arabic (Persson 2009, 277-279). Bloch, thus, is right
in making the connection between the bare prefix form and a sense of finality.
The hypothesis that all (other) circumstantial clauses of this type are always
introduced by a particle, however, is not entirely supported by the present
study.105
Most importantly, however, with the recognition of gram switching, there
is a different argument to abandon the concept of ‘ḥāl muqaddar’. This cate-
gory was created in an effort to accommodate, within the perceived class of
circumstantial clauses, those clauses that formally would fit into the lists of
what had been labelled circumstantial clauses but that expressed finality in-
stead of ‘attendant circumstance’ (cf. Persson 2009, 253-256). These so called
‘ḥāl muqaddar’ clauses fit well into the new perspective of gram switching as
a general marker of hypotaxis. Together with other semantically non-specific
104 Cf. also Bloch (1965, 75).

105 Cf. Persson (2014).

112 Maria Persson
non-main clauses, they occupy various positions on a scale of hypotactic

marking, while the specific semantic interpretation of each clause is left to, as
Blau (1960, 218) puts it, “die Phantasie des Zuhörers”.106
5.7 The significance of the ‘doubly marked’ circumstantial

I noted above, (5.2.1), that Bloch’s Type 1 contains clauses introduced by
w(ə)- followed by a pronoun and I briefly mentioned that the use of the pro-
noun here specifically indicates that the non-main clause describes one of the
participants of the main clause. While many of the circumstantials dealt with
in this treatise have an adverbial meaning, this particular construction is de-
cidedly adjectival. Here, it is not the verb of the main clause that is qualified
by the non-main clause. Rather, the non-main clause is a description of the
state of the referent at the time of the state or event described in the main
clause. So, for example, Umm Rashdi’s being quiet in (55) describes her, at
the time of her being on the verge of exploding; it does not describe the way
in which she was about to explode.
(55) w-ʾəmm rəšdi raḥa tṭəʾʾ əlməskīne [w-hīye sākte] … (17:4:126)
w-mother Rəšdi FUT burstPF3fs DEF.poor [w-3fs be.quietAPfs]
‘and Umm Rushdie was on the verge of exploding, poor one, [and/but she
was quiet]’
Similarly, the non-main clause in (1) does not qualify the act of spending time
but is a description of the state of the narrator while he was spending the
time.
(1) ʾaḍḍēna mədde mn əzzamān [w-nəḥna ʿam nətšawwa w-nətʾalla]
(1:67:26)
spendSF1cp period PREP DEF.time [w-1cp PROGR be.grilledPF1cp w-be.friedPF1cp
‘I spent some time [sizzling and frying (=frustrated)]’
The non-main clause in (55) is a non-verbal clause – it consists of a subject
and a predicative attribute. 107 It is semantically closely related to what in
standard Arabic is realized as the so called “circumstantial accusative”, in the
dialects as a participial or adjective:
106 I.e. “the imagination of the listeners”.

107 It is also, particularly due to the explicit pronoun subject, semantically rather inde-
pendent of the main clause. Cf. also Addeweesh (1985, 47-49).

(57) ʾərkod ʿala halbēt ʿala ḍaww ʿalbēt [mabṣūṭ] (1:17:6)

runPF1cs PREP DEM.DEF.house PREP light PREP.DEF.house [happy]
‘I would run [happy] to the house in daylight’
In (57) the passive participle mabṣūṭ, ‘happy’, does not in any way describe
the act of running in the main clause. The boy is not running in a happy way.
He is happy and he is running. It is his state of being that is described by the
non-main clause.108 What made these clauses stand out to the Arab grammari-
ans was the fact that, in the grammar of standard Arabic, the participle here is,
as indicated above, in the accusative case.109
(91) yarkuḍu ḍāḥikan
runPF3ms laughPAmsACC
‘He runs laughing’
(92) yarkuḍu dāʾiman
runPF3ms constantPAmsACC
‘He runs always’
In (91) the verb is followed by a participle that indicates the state of the one
running. The verb “run” cannot “laugh” – it is the person running who is, at
the same time, laughing. In the second example the participle describes the
frequency of the act of running. The boy is not constant – it is the act of run-
ning that is constant. Language, however, does not differentiate morphologi-
cally or structurally between these.
Clauses such as (1) and (57) also fit into the system of gram switching.
There is, usually, a gram switch between the two clauses. However, the use of
the independent pronoun specifically signals that the non-main clause consti-
tutes a description of the state of the referent of this pronoun. In this sense,
these clauses may be said be markedly circumstantial, i.e. to have a syntactic
form that coincides with a circumstantial reading, namely in the sense of ex-
108 Cf. Addeweesh (1985, 186): ”The major function of the single-word type of ḥāl is to
describe the state or condition of the referent of the, ḥāl rather than the action of the
regent verb of the ḥāl ”. Similarly, on page 192, Addaweesh describes examples of
the type found in (1) above saying that “the basic function of the ḥāl is to describe the
state or condition of the referent of the ḥāl rather than the action”.
109 Examples 91-93 are construed for the purpose of demonstration. Cf. for example
Addeweesh (1985, 3).

114 Maria Persson
pressing a state or activity that is simultaneous with the state or activity in the
main clause.110
A clause featuring two verbs such as (93) also describes a state or activity
that is simultaneous with the state or activity in the main clause:
(93) rakaḍa yaḍḥaku
runSF3ms laughPF3ms
‘He ran laughing’
However, this does not constitute a specific clause type. The gram switch
indicates general hypotaxis and the same combination of one suffix form verb
followed by a prefix form verb may, as has been demonstrated above, receive
varying semantic interpretations depending on the semantic value and aktion-
sart of each specific verb and the general context in which the clause is ut-
tered. Or, as Beeston put it in the passage quoted above (2.1); “… there is an
unsophisticated lack of overt marks of the logical intention” (Beeston 1970,
89).
5.8 Summary
The above has shown that Arab grammarians and grammarians of Arabic have
discovered one part of circumstantial clause combining by recognizing a
number of clause types regularly used for qualifying other clauses circumstan-
tially. What they failed to see is that their lists did not capture a cohesive class
of clauses but, rather, general strategies of non-main clause linking. Their lists
were neither exhaustive nor cohesive and a large number of clauses with cir-
cumstantial meaning evaded the different nets that they had thrown.
Only in the case of one of the clause types discussed by Bloch do form and
function coincide to such an extent that it is possible to say that a specific
clause type is used specifically to encode a circumstantial meaning. Namely,
in the clauses featuring w(ə)- +pronoun discussed in the preceding section.
In yet another case, clause type coincides with a specific meaning: non-
main clauses that precede their main clauses and that are introduced by a
pronoun alone or w(ə)- + pronoun seem to be used, at least in the present
110 This corresponds well with the definition of ḥāl that Addeweesh (1985, 184-185)
concludes to be the only possible: “a ḥāl is a construction whose function is to depict
circumstances attendant on the referent of the ḥāl at the time the action or event of the
regent occurred, whether such circumstances are temporary or permanent”. Cf. also
Addeweesh (1985 191-192).

corpus and the Gulf Arabic corpus previously surveyed, as a specific type of
temporal clause.
On a different note, a gram switch was observed in all but three of the 110
examples in the database which correspond to Bloch’s types. This further
supports the hypothesis, stated above, that it is not gram switching that hap-
pens to coincide with circumstantials; instead it is the hypotaxis in circum-
stantials that is marked by gram switching.
6. Conclusion
A variety of clauses have been described, in previous studies, under headings
such as “ḥāl-clauses” or “circumstantial clauses”. As has already been pointed
out by several scholars (Abboud 1986, 191; Isaksson et al. 2009; Premper
2002, 29), this approach has not led to the formulation of a clear syntactic
category of “circumstantial clauses”. The analysis presented here, based on a
database of Damascene Arabic from the 1950s, together with previous re-
search, has confirmed that the reason for this confusion is that there are no
syntactic grounds for the establishment of a class of “circumstantial clauses”
on a par with, for example, “conditional clauses” and “relative clauses”. Of all
the surveyed clauses, only one clause type combines form and function to
such an extent that it can be said to be used specifically to encode a circum-
stantial meaning. Non-main clauses featuring w(ə)- +pronoun do generally
receive a circumstantial reading, namely in the sense of expressing a state or
activity pertaining to the referent of the pronoun that is simultaneous with the
state or activity expressed in the main clause. Yet another clause type was
distinguished by performing a specific function, namely non-main clauses that
precede their main clauses and that are introduced by a pronoun alone or w(ə)-
+ pronoun. These were found to function, at least in the surveyed types of
spoken Arabic, as a specific type of temporal clause.
The survey took its starting point in Bloch’s treatment of, among others,
the same database that has constituted the basis for the present study. Bloch
followed the tradition in listing syntactic types of ‘circumstantial clauses’. In
his final remarks he did, however, pinpoint the problem that constituted the
impetus for the present survey:
Nun ist es nicht leicht festzustellen, unter welchen Umständen für den
gleichen Sachverhalt einmal die parataktische, einmal die hypotakti-

116 Maria Persson
sche Ausdrucksweise gewählt wird, wann man also etwa sagt: btəsʾal
bidəllūk, und wann ʾəza btəsʾal bidəllūk.111 (Bloch 1965, 98)
Whereas Bloch did not find an answer to his question, the results of the pre-
sent project and its forerunner show that the choice between an asyndetic and
a syndetic juncture is often not as arbitrary as it seemed to him. Moreover,
many of the clause combinations that, to Bloch, seemed perfectly paratactic
are, in fact, marked for hypotaxis, namely by gram switching.
Gram switching, mainly in the form of a switch between verb forms but al-
so comprising variations in mood marking as well as a switch from a clause
containing a verbal gram to a verb-less clause, has been found to function as a
general marker of non-main clause linking. This general marker occupies the
second position in a scale of marking starting from no syntactic marking, via
gram switching that marks the existence of hypotaxis, to more specific mark-
ing where the semantic type of the non-main clause is also marked:
− No marking. Only context and/or intonation

− Gram switching only
− Gram switching + neutral conjunction (“and”) indicating the location, and
thus the existence, of the linking
− Gram switching and conjunction - or conjunction alone - specifying se-
mantic type of non-main clause
The discovery of gram switching implies the discovery of an overarching

system of minimally marked non-main clause linking. Gram switching alone
is used to mark such varying types of non-main clauses as adverbial clauses,
final clauses, and conditional clauses. Then, there is a possibility of using a
semantically non-specific conjunction such as w(ə)- (and) to indicate the loca-
tion of juncture albeit not the (semantic) type of the non-main clause. Finally,
the speaker can choose to use a semantically specific subordinating conjunc-
tion to clearly specify the semantics of the clause linking.
Pragmatic reasons will thus decide whether a non-main clause is left un-
marked at the syntactic and morphological level, marked with a gram switch,
or marked by both a gram switch and a conjunction. Gram switching functions
111 I.e. “It is, however, not easy to determine the circumstances under which, at times a
paratactic, at times a hypotactic way of expression is chosen for the [expression of]
the s a m e fact; when one says btəsʾal bidəllūk, and when ʾəza btəsʾal bidəllūk”.

as a marker of non-main clause linking in general. It is not limited to any of

those types of clauses that have been called circumstantial. The diverse claus-
es that appear on various lists of “circumstantial” or “ḥāl-” clauses are found,
together with other semantic clause types, at various levels of marking. Fur-
thermore, gram switching as a general marker of hypotaxis is neither isolated
to colloquial Arabic nor to spoken data, but is supported by Isaksson’s
(2009a) findings in both Classical Arabic and Biblical Hebrew.
References
Abboud, P. 1986. “The ḥāl construction and the main verb in the sentence.” In The
Fergusonian impact: in honor of Charles A. Ferguson on the occasion of his 65th
birthday. Vol. 1, From phonology to society, edited by J. A. Fishman, A. Tabouret-
Keller, M. M. Clyne, B. Krishnamurti and M. Abudlaziz, 191-196. Berlin: Mouton
de Gruyter.
Addeweesh, R. A. 1985. A syntactic and semantic study of ḥāl ‘circumstantial’ struc-
tures in modern literary Arabic prose literature. (PhD Diss.), University of Michi-
gan, Ann Arbor, University Microfilms International, Ann Arbor.
Ambros, A. A. 1977. Damascus Arabic. Malibu, California: Undena Publications.
Beeston, A. F. L. 1970. The Arabic language today. London: Hutchinson.
Bergsträsser, G. 1924. Zum arabischen Dialekt von Damaskus. Vol. 1. Phonetik -
Prosatexte. Beiträge zur semitischen Philologie und Linguistik 1. Hannover: Ver-
lag der Orient-Buchhandlung Heinz Lafaire.
Blau, J. 1960. Syntax des palästinensischen Bauerndialektes von Bīr-Zēt, auf Grund
der "Volkserzählungen aus Palästina von Hans Schmidt und Paul Kahle. Walldorf-
Hessen: Verlag für Orientkunde Dr. H. Vorndran.
Bloch, A. 1965. Die Hypotaxe im Damaszenisch-Arabischen: mit Vergleichen zur
Hypotaxe im Klassisch-Arabischen. Wiesbaden: Deutsche Morgenländische Ge-
sellschaft, Kommissionsvlg. F. Steiner.
Bloch, A., and Grotzfeld, H. 1964. Damaszenisch-arabische Texte. Wiesbaden: Deut-
sche Morgenländische Gesellschaft, Kommissionsvlg. F. Steiner.
Brustad, K. E. 2000. The syntax of spoken Arabic: a comparative study of Moroccan,
Egyptian, Syrian, and Kuwaiti dialects. Washington, DC: Georgetown University
Press.
Bybee, J., and Dahl, Ö. 1989. “The creation of tense and aspect systems in the lan-
guages of the world.” Studies in language, 13(1): 51-103.

118 Maria Persson
Bybee, J. L., Perkins, R. D., and Pagliuca, W. 1994. The evolution of grammar: tense,
aspect, and modality in the languages of the world. Chicago: Univ. of Chicago
Press.
Cantineau, J., and Heibaoui, Y. 1953. Manuel élémentaire d'arabe oriental (Parler de
Damas). Paris: Klincksieck.
Dietrich, A. 1956. “Damaszener Schwänke”. Zeitschrift der Deutschen morgenländi-
schen Gesellschaft 106: 317-344.
Dixon, R. M. W. 2009. “The semantics of clause linking in typological perspective.” In
The semantics of slause linking. A cross-linguistic typology, edited by R. M. W.
Dixon and A. Y. Aikhenvald, 1-55. Oxford: Oxford University Press.
Eisele, J. C. 1999. Arabic verbs in time: tense and aspect in Cairene Arabic. Wiesba-
den: Harrassowitz.
Eksell, K. 1995. “Some punctual and durative auxiliaries in Syro-Palestinian dialects.”
In Proceedings of the 2nd Internal Conference of l'Association internationale pour
la dialectologie Arabe: held at Trinity Hall in the University of Cambridge, 10-14
September 1995, edited by J. Cremona, C. Holes and G. Khan, 41-49. Cambridge:
University Publications Centre.
Firanescu, D. R. 2003. “Le modalisateur aspectuel-temporel qām dans le parler sy-
rien.” In AIDA, 5th Conference Proceedings, Cádiz september 2002, edited by I.
Ferrando and J. J. Sánchez Sandoval, 481-492. Association Internationale de dia-
lectologie Arabe. Cádiz: Servicio de Publicationes, Universidad de Cádiz.
———. 2008. “The moving sands of the modals raaH and ija in Syrian Arabic.” In
Between the Atlantic and Indian Oceans. Studies on contemporary Arabic dialects.
Proceedings of the 7th AIDA Conference, held in Vienna from 5-9 September
2006, edited by S. Prochàzka and V. Ritt-Benmimoun, 185-194. Wien: LIT Ver-
lags.
Fischer, W. 2002. “Unterordnende und nebenordnende Verbalkomposita in den neu-
arabischen Dialekten und im Schriftarabischen.” In "Sprich doch mit deinen
Knechten aramäisch, wir verstehen es!": 60 Beiträge zur Semitistik: Festschrift für
Otto Jastrow zum 60. Geburtstag, edited by W. Arnold and H. Bobzin, 147-163.
Wiesbaden: Harrassowitz.
Fleischman, Suzanne. 1990. Tense and narrativity: from medieval performance to
modern fiction. Croom Helm romance linguistics series. London: Routledge.
Givón, Talmy 2001. Syntax: an introduction. Vol. 1. Philadelphia: J. Benjamins.
Grotzfeld, H. 1964. Laut- und Formenlehre des Damaszenisch-Arabischen. Wiesba-
den: Deutsche Morgenländische Gesellschaft.

———. 1965. Syrisch-arabische Grammatik (Dialekt von Damaskus). Wiesbaden:

Harrassowitz.
Isaksson, B. 2008. “Circumstantial qualifiers in the Arabic dialect of Kinderib (East
Turkey).” In Between the Atlantic and Indian Oceans. Studies on contemporary
Arabic dialects. Proceedings of the 7th AIDA Conference, held in Vienna from 5-9
September 2006, edited by Veronika Ritt-Benmimoun and Stephan Procházka,
251-258. Wien: LIT Verlags.
———. 2009a. “Introduction.” In B. Isaksson, H. Kammensjö and M. Persson, 2009,
1-35.
———. 2009b. “An outline of comparative Arabic and Hebrew textlinguistics.” In B.
Isaksson, H. Kammensjö and M. Persson, 2009, 36-150.
———. 2011. “The textlinguistics of the Suffering Servant: Subordinate structures in
Isaiah 52,13-53,12.” In En pāsē grammatikē kai sophiā. Saggi di linguistica ebrai-
ca in onore di Alviero Niccacci, ofm, edited by G. Geiger and M. Pazzini, 173-212.
Analecta: Studium Biblicum Franciscanum 78. Jerusalem and Milano: Franciscan
Printing Press and Edizioni Terra Santa.
———. 2013. “Subordination: Biblical Hebrew.” In Encyclopedia of Hebrew Lan-
guage and Linguistics, edited by G. Khan, vol. 3, 657-664. Leiden: Brill.
Isaksson, B., Kammensjö, H., and Persson, M. 2009. Circumstantial qualifiers in
Semitic. The case of Arabic and Hebrew, edited by B. Isaksson. Abhandlungen für
die Kunde des Morgenlandes 70. Wiesbaden: Harrassowitz.
Johnstone, B. 1990. “'Orality' and discourse structure in Modern Standard Arabic.” In
Perspectives on Arabic Linguistics, edited by M. Eid, vol. I, 215-233. Amsterdam:
John Benjamins.
Kammensjö, H. 2008. “Omständighetsbestämningar i samtida arabisklitteratur. (Cir-
cumstantial qualifiers in contemporary Arabic literature).” Paper presented at the
Nordiska semitistsymposiet, Kivik, Sweden.
———. 2009. “Circumstantial qualifiers in contemporary Arabic prose – With a view
to understanding variation.” In B. Isaksson, H. Kammensjö and M. Persson, 2009,
151-205.
Kaye, A. S. and Rosenhouse, J. 1997. “Arabic dialects and Maltese.” In The Semitic
languages, edited by R. Hetzron, 263-311. London: Routledge.
Lewin, B. 1966. Arabische Texte im Dialekt von Hama: mit Einleitung und Glossar.
Wiesbaden: Steiner.
Lichtenberk, F. 2009. “The semantics of clause linking in Toqabaqita”. In The seman-
tics of clause linking. A cross-linguistic typology, edited by R. M. W. Dixon and A.
Y. Aikhenvald, 239-260. Oxford: Oxford University Press.

120 Maria Persson
Lord, C. 1993. Historical change in serial verb constructions. Amsterdam: J. Benja-

mins.
Malinjoud 1924. “Textes en dialecte de Damas.” Journal Asiatique, 204: 259-332.
Matthiessen, C. M. I. M. 2002. “Combining clauses into clause complexes. A multi-
faceted view.” In Complex sentences in grammar and discourse. Essays in honor
of Sandra A. Thompson, edited by J. Bybee and M. Noonan, 235-319. John Ben-
jamins.
Matthiessen, C. M. I. M., and Thompson, S. A. 1988. “The structure of discourse and
‘subordination’” In Clause combining in grammar and discourse. Typological
studies in language, edited by J. Haiman and S. A. Thompson, 275-329. Amster-
dam: J. Benjamins.
Oestrup, J. 1897. Contes de Damas. Leyde: Brill.
Payne, T. E. 1997. Describing morphosyntax: a guide for field linguists. Cambridge:
Cambridge Univ. Press.
Persson, M. 2009. “Circumstantial qualifiers in Gulf Arabic dialects.” In B. Isaksson,
H. Kammensjö and M. Persson, 2009, 206-289.
———. 2013. “Asyndetic clause combining in Gulf Arabic dialects. Auxiliary, adver-
bial and discourse functions.” Zeitschrift für Arabische Linguistik 57: 5-39.
———. 2014a. “Non-Main Clause Linking and Verb Form Switch in Syrian Arabic: Is
there a circumstantial clause?” In Strategies of Clause Linking in Semitic Lan-
guages: Proceedings of the International Symposium on Clause Linking in Semitic
Languages 5-7 August 2012 in Kivik, Sweden, edited by B. Isaksson and M.
Persson, 27-50. Abhandlungen für die Kunde des Morgenlandes 93, Wiesbaden:
Harrassowitz.
———.2014b. “Verb form switch as a marker of discourse hierarchy in Semitic: a
case study on Syrian Arabic.” In: From Tur Abdin to Hadramawt. Semitic studies.
Festschrift in honour of Bo Isaksson on the occasion of his retirement, edited by
Tal Davidovich, Ablahad Lahdo, and Torkel Lindquist, 117-128. Wiesbaden: Har-
rassowitz.
———. Forthcoming. “Verb form switch as a marker of clausal hierarchies in urban
Gulf Arabic.” In: Arabic and Semitic Linguistics Contextualized, edited by Lutz
Edzard. Wiesbaden: Harrassowitz.
Premper, W. 2002. Die „Zustandssätze“ des Arabischen in typologischer Perspektive.
Frankfurt am Main: P. Lang.
Rosenhouse, J. 1978. “Circumstancial Clauses in some Arabic Dialects.” Zeitschrift
der Deutschen Morgenländischen Gesellschaft 128(2): 226-237.

Thompson, S. A., Longacre, R. E., and Hwang, S. J. J. 2007. “Adverbial clauses.” In

Language typology and syntactic description, edited by T. Shopen. Vol. 2, Com-
plex constructions, 237-300. 2nd ed. Cambridge: Cambridge Univ. Press.
Woidich, M. 2002. “Verbalphrasen mit asyndetischem Perfekt im Ägyptisch-
Arabischen.” Estudios de Dialectología Norteafricana y Andalusí 6: 121-192.

Clause Combining in Written Arabic

The Domain of Verbal Circumstantial Clauses
in Classical Arabic
Michal Marmorstein, The Hebrew University, Jerusalem
1. Introduction
Circumstantial clauses (henceforth: CCs) constitute a main type of clauses in
Classical Arabic. In both the medieval and modern grammatical literature, the
subject of CCs is dedicated a special chapter, and in recent years it has become
the primary focus of three volume-length studies (Premper 2002; Isaksson et
al. 2009; Waltisberg 2009). CCs make a highly intriguing topic of study, not
just because of their particular syntactic and semantic character, but also be-
cause they challenge a basic ambition of any linguistic analysis, to reach a
neat classification and a sharp demarcation of structural categories. Indeed,
CCs in Arabic defy a categorical definition and thus compel us to think of the
entire domain to which they belong.
The present study is concerned with a variety of circumstantial clauses in
Classical Arabic whose predicate is a verbal form. The discussion will focus
on two issues: (a) the syntactic domain to which verbal CCs belong, i.e. the
domain of complex predications; and (b) the paradigm of verbal forms shared
by all complex predications. The aim is to deal with both the wide variety of
structures related to the notion of the ‘circumstantial’, and the common trait of
these structures, which justifies their analysis under a single heading. The
analysis of verbal CCs in their syntactic domain will also be shown to be high-
ly instructive in reaching a better understanding of the internal mechanism of
the Classical Arabic verbal system.
The analysis of verbal CCs is based on a large body of Classical Arabic
prose, composed or compiled by the end of the 10th century A.D. The corpus
comprises excerpts from the historical accounts of Ibn Hišām, al-Wāqidī and
al-Ṭabarī, the belles lettres of al-Ǧāḥiẓ, Ibn al-Muqaffaʿ, and al-ʾIṣbahānī, and
the traditions collected by al-Buxārī (see below the references list).

126 Michal Marmorstein
Before discussing the domain of verbal CCs in Classical Arabic, a brief ac-
count of the literature on CCs in Arabic is deemed necessary. The purpose of
this short review is not to cover the vast amount of writing on the topic, but
only to sketch the development of the traditional concept of the ḥāl ‘circum-
stantial expression’ in modern studies on Arabic CCs, and to point out the
principal ideas which characterize the discussion throughout, especially the
hesitation as to the syntactic status of this type of clauses and the problem of
capturing the semantic essence of the circumstantial relation.
2. A brief account of the circumstantial category in the literature

2.1 The ḥāl in Arabic grammatical tradition
The concept of the ḥāl, as formulated in the Arabic grammatical tradition,
deserves a separate study. Here I will only provide a general outline of the
analytic procedure by which the ḥāl category was defined, and call attention
to some important insights that are still relevant to a modern understanding of
CCs.
A simple clause such as (1) features the typical ḥāl ‘circumstantial expres-
sion’:1 an indefinite nominal (participle) which is governed by a verb (or by
whatever has a verbal meaning) and thereby assigned the accusative case:2
(1) ḏahabtu rākiban
goSC1csg PTCPmsg-acc
‘I went riding.’
The Arab grammarians define the ḥāl in relation to other constituents of the
clause. These constituents are the object, the adverbial specification, the at-
tribute and the predicate.
By contrast to the mafʿūl ‘object’ (e.g. ḍarabtu ʿabda llāhi ‘I hit
ʿAbdallāh’), the ḥāl is always indefinite, it cannot undergo transformation,
1 The term ḥāl is an abbreviation of ḥāl waqaʿa fī-hi al-fiʿl ‘the circumstance in which the
verbal action took place’ (Sībawayhi I, 15).
2 The examples are glossed according to the Leipzig glossing rules, with the following
additions: EMPH=emphasis particle, INTRO=introductory particle, JUSS=jussive,
MOD=modifying particle, PASS=passive, PC=prefix conjugation (the yafʿalu pattern),
PRON=pronoun, SC=suffix conjugation (the faʿala pattern), TOP=topicalizer. Proper
names of people and places which don’t have a conventionalized English form are ren-
dered in the original Arabic form in the translation.

The Domain of Verbal Circumstantial Clauses in Classical Arabic 127
from patient (of an active verb) to agent (of a passive verb), and it is compati-
ble with intransitive verbs (Sībawayhi I, 15-16). The subject and the object
have a different status (manzila) than the ḥāl, being the constituents by which
the kernel structure of the clause is made complete. The ḥāl is therefore con-
sidered a faḍla ‘redundancy’ or ziyāda ‘addition’, as it occurs baʿdu tamāmi l-
kalāmi ‘after a complete clause’ (Ibn al-Sarrāǧ I, 213; Ibn Yaʿīš II, 4).
The ḥāl and the tamyīz ‘(adverbial) specification’ (e.g. la-hū ʿišrūna dir-
haman ‘He has twenty Dirhams’) are similar in being both indefinite accusa-
tives whose function is to distinguish one situation from other possible ones
and thus restrict the (main) predication (Ibn Yaʿīš II, 36). Both the ḥāl and the
tamyīz are governed by the preceding predication, and hence marked as accu-
sative or ‘dependent’, yet they are external to the structure which they follow
(Sībawayhi I, 16).3
The ḥāl diverges from the general category of the ṣifa or waṣf ‘attribute’
(e.g. ḍarabtu Zaydan ʾabā-ka ‘I hit Zayd, your father’),4 as it can only indi-
cate a transitory state, an accidental attribute and not an inherent property (Ibn
al-Sarrāǧ I, 213). The ḥāl is not employed to identify or particularize an enti-
ty, but to convey an additional predicate. As such, the ḥāl (like other predi-
cates) is in principle indefinite. The attribute, by contrast, agrees with its ante-
cedent and hence may be either definite or indefinite.
The grammarians often designate the ḥāl constituent xabar ‘predicate’. In
his Kitāb, Sībawayhi interchanges between the terms ḥāl and xabar in cases
where the ḥāl constituent follows a definite noun. In analyzing a clause such
as hāḏā ʿabdu llāhi munṭaliqan ‘Here is ʿAbdallāh departing’, Sībawayhi
refers to the constituent munṭaliqan as both xabar lil-maʿrūf ‘a predicate of a
definite noun’ and ḥāl (I, 218). Later grammarians define as a rule the ḥāl as
xabar and even xabar ṯāni ‘secondary predicate’ (Ibn al-Sarrāǧ I, 214; Ibn
Yaʿīš II, 6). The only difference between a ḥāl and a genuine predicate is that
3 In the example ḏahabtu rākiban (lit. ‘went-I riding’), the ḥāl constituent rākiban is
made external to the governing verb ḏahab-tu by the ‘interposition’ of the agent -tu, the
same way as the tamyīz constituent dirhaman in the clause la-hū ʿišrū-na dirhaman (lit.
‘to-him twenty Dirhams’) is ‘separated’ from the governing noun ʿišrū-na by the tanwīn
(or its equivalent -na); cf. Carter (1972) for the general principle of tanwīn boundary in
Sībawayhi’s Kitāb.
4 Ibn Hišām (194-195) employs the term waṣf to refer to a super-category, under which
the ḥāl, the xabar and the ṣifa are subsumed. The ḥāl is distinct from the xabar in being
dispensable to the complete structure of the clause; it is distinct from the ṣifa in indicat-
ing a predicative rather than an attributive relation.

the first, in its typical form, is not considered an essential part of the grammat-
ical structure of the clause (Ibn Yaʿīš II, 17).5 However, the ḥāl may become
indispensable when the predicative information, the point for which the clause
is designed, is ‘gathered’ (munʿaqida) by it (Ibn Yaʿīš II, 9). Thus, in the
example cited above, the ḥāl constituent munṭaliqan is analyzed as an essen-
tial part of the clause, communicating the sight of ʿAbdallāh while departing.
The fact that the terms ḥāl and xabar sometimes overlap or are used inter-
changeably (the latter defining the first) is quite telling: it discloses the spe-
cial identity of this syntactic constituent, which at the same time is both
predicative (in its essence) and adverbial (in its position). In their strict
formal analysis of the clause, the grammarians regard the ḥāl as peripheral,
like any other adverbial; however, as far as the informativity (or ‘usefulness’)
of the clause is concerned, the ḥāl is viewed as part of the predicative core.
According to the Arab grammarians, the typical ḥāl is a plain noun-phrase,
or to be more precise, a participial form. This is due to a basic principle of
their grammatical theory, namely, that the prototype of a syntactic category, to
which other members are analogous, is the most basic (non-derived or aug-
mented) morphological constituent. When the ḥāl is realized as a full clause, it
is considered to occupy the syntactic position of a simple ḥāl constituent (Ibn
Yaʿīš II, 24ff.), and therefore be embedded.6 The dependency of the ǧumla
ḥāliyya ‘circumstantial clause’ upon the main clause is marked by the con-
junction wa- and/or by a resumptive pronoun, as illustrated in the examples
below; in (2) the linkage is implicitly marked by the pronoun embodied in the
verb, whereas in (3) it is explicitly marked, by the overt pronoun huwa and the
conjunction wa-:
5 Thus, Ibn Yaʿīš (II, 7) explains that in a clause such as marartu bi-l-farazdaqa qāʾiman
‘I passed by al-Farazdaq standing’ the predication (ʾixbār) of ‘passing by’ is added an-
other predicate (xabar ʾāxar), the only difference is that the first is obligatory (ʿalā
sabīli l-luzūmi) whereas the latter is additional and therefore can be renounced
(ziyādatan yaǧūzu l-istiġnāʾu ʿan-hā).
6 Cf. Goldenberg (1987-1988, 108): ‘In Arab grammatical tradition, subordinate (includ-
ed, embedded) clauses are consistently defined in terms of the simple forms whose sta-
tus they assume: the position where a noun, e.g., is required can be occupied either by a
“plain noun” (ism ṣarīḥ) or by what virtually equals its meaning as a noun (=al-
muʾawwal bihī or mā fī taʾwīlihī) that is its “periphrasis” or “paraphrase”’.

(2) ǧāʾa Zaydun yarkabu

comeSC3msg Zayd-nom ridePC3msg
‘Zayd came riding.’
(3) ǧāʾa Zaydun wa-huwa rākibun
comeSC3msg Zayd-nom and-PRON3msg ridePTCPmsg-nom
‘Zayd came while he was riding.’
2.2 The circumstantial/predicative in modern grammatical

descriptions of Arabic
Detailed accounts of the discussion on CCs in Western grammars of Classical
Arabic are found in Premper (2002, chapter 4) and Waltisberg (2009, 23-59).
In this section I will only refer to the points where the modern approach does
not follow the traditional one, but introduces some new observations, distinc-
tions and terminology regarding the matter of CCs in Arabic.
Western grammars do not conform with the Arabic grammatical tradition
on one significant point: the ḥāl, which is conceived by the Arab grammarians
as a unified category comprising simple and complex manifestations, is split
into forms (word-units) and full clauses. The first are treated as parts of the
simple clause, the latter as parts of the complex clause. It is interesting to
notice the different renderings of the terms ḥāl and ǧumla ḥāliyya in these
grammars. Wright (1898 II, 112-121) provides a literal translation of ḥāl as
‘state or condition’ whereas the clause initiated by wāw al-ḥāl is called ‘cir-
cumstantial clause’ (332). Reckendorf (1895 I, 101-102; 1921, 97-100) refers
to both a nominal and an asyndetic verbal form as Prädikativum. By contrast,
syndetic clauses are labeled Zustandssätze (549-564; 447-453). The same
division between ḥāl forms and clauses is reflected in Fischer’s terminology
(2002, 196, 220): both nominal and verbal circumstantial forms are defined as
‘predicative’, whereas syndetic clauses are treated separately and labeled
‘coordinate circumstantial clauses’ (209).
Another point in which Western grammars differ from the Arabic tradition
is in their applying a diachronic-comparative perspective to explain certain
issues of Arabic grammar. The relationship between the ḥāl and the xabar is
described by the Arab grammarians in structural terms: the first is analyzed as

an equivalent of the latter in a ‘transformed’ or ‘expanded’ clause.7 Western

grammars, by contrast, describe an historical development in the opposite
direction, whereby the predicative has originated from a circumstantial clause
(cf. Reckendorf 1921, 295).
Recent studies of CCs in Arabic (see above 1) depart to an even greater ex-
tent from the traditional model. The unified category of the Arab grammari-
ans, split in Western grammars into predicative forms and circumstantial
clauses, is further deconstructed. When considered within a modern typologi-
cal approach, the topic of CCs is re-contextualized: rather than stressing the
semantic homogeneity and the syntactic analogy between simple and complex
CCs, the heterogenic nature of this set of structures – which extend from sim-
ple adverbial constituents to textual units – is put into focus. The circumstan-
tial is no longer viewed as a micro-syntactic category of the simple clause, but
as part of the macro-syntactic phenomenon of clause-linking. The traditional
ḥāl-paradigm is replaced by a scale, accommodating varying degrees and
manifestations of clause combining. Some CCs are accordingly involved in
constructions exhibiting a high degree of ‘elaboration’, whereas others form
part of highly ‘compressed’ constructions (Lehmann 1988, 216).
As the paradigm is abandoned in favor of the scale, one is no longer con-
cerned with the formal division between forms and full clauses; rather, all
circumstantials are considered to be clauses, featuring a higher or a lower
degree of finiteness or sententialization. On the junction scale (Raible, 1992),
a participle such as rākiban (1) – which was treated by the Arab grammarians
as the prototype of the ḥāl-paradigm – is analyzed as syntactically ‘degraded’
with respect to a finite form such as yarkabu (2), and even more so, a nominal
clause such as wa-huwa rākibun (3). By the same token, the latter are not
necessarily regarded as embedded (due to their substitution with the participi-
7 Talmon (2003, 40, 188-189, 296-297) explains the notion of xabar manṣūb ‘an accusa-
tive predicate’ as a syntagm that, due to the creation of a new syntactic structure, is ‘de-
prived of its first-degree predicate status and transforms consequently to a second-
degree predicate position marked grammatically by naṣb’. The alternation between ḥāl
and xabar in Sībawayhi’s terminology may be better understood if put in a broader con-
text. According to Levin (1979, 193ff.), besides genuine predicates, the term xabar is
applied to those constituents in expanded verbal or nominal clauses – such as initiated
by the auxiliary verb kāna, by di-transitive mental verbs or by the presentative hāḏā –
which correspond to the predicate in the basic (underlying) nominal structure. Although
marked by the accusative, all these have the function of a predicate.

al form), but rather as dependent clauses, integrated to some extent with their
main clause.8
Not only from a syntactic point of view, but also from a semantic one, the
circumstantial category is notoriously hard to define (König 1995). One of the
solutions is to view the semantically unmarked or ambiguous character of CCs
as their distinct property. Thus, Isaksson et al. (2009, 4) define CCs as ‘en-
hancing clauses’ which – by contrast to ‘particle-marked’ adverbial clauses –
are not specific and can express a whole range of semantic relations: ‘general
circumstantial, condition, reason, purpose, cause, time, space, manner, or
means’. Waltisberg (2009, 358ff.) also regards the syndetically-linked variety
of CCs as polyfunctional; while syndetic CCs indicate a ‘catalogue’ of seman-
tic relations, their asyndetic counterparts serve the single ‘unspecific’ function
of Erläuterung. Premper (2002, 275, 369) stresses the temporal relation and
the textual function which correlate with the semantic ‘generality’ of CCs:
syndetic CCs are simultaneous and thus serve background descriptions, where-
as asyndetic CCs are sequential, thus marking a further progression or inten-
tion of the participants in the verbal event.
2.3 The problem of the circumstantial/predicative in Arabic:

recapitulation
Modern thinking on CCs in Arabic, influenced by the general study of clause-
linking and clause-typology, redefines the discussion of the topic by address-
ing such problems as the formal identity of CCs and the nature of the semantic
relation which they convey. CCs are found to be a heterogenic class of con-
structions, including verbal complexes, on the one hand, and with the text-
level structures, on the other (Premper, 369). Whether analyzed as semantical-
ly vague or polysemous, the meaning of CCs is considered to be implicit,
thereby interpretable as ‘temporal’, ‘causal’, ‘modal’, ‘contrastive’ etc. in
different contexts. Unlike other types of adverbial clauses, CCs appear to defy
a clear-cut formal as well as functional determination. Rather than forcing
8 Some scholars draw a clear distinction between ‘hypotaxis’ or ‘dependency’ and ‘em-
bedding’: the first indicate the combination of a main clause and a subordinate clause,
the latter the inclusion of a clause as a constituent in a ‘higher’ clause, cf. Van Valin
(1984), Halliday (1985, chapter 7). Circumstantial clauses, according to the same view,
are not commutable with simple adverbial constituents (but only by way of a ‘grammat-
ical metaphor’) and hence are regarded as essentially non-embedded, cf. Matthiessen
and Thompson (1988).

them into arbitrary boundaries, modern researchers resort to the large syntac-
tic domain within which CCs operate.
The present study follows the same general course and adopts the scalar
model. Verbal CCs are accordingly not viewed as a micro-syntactic category,
but as various realizations of the macro-syntactic phenomenon of complex
predications and its semantic-pragmatic correlate of event-integration (cf.
Givón 2001, II, 39ff.). Drawing on the traditional terminology we may say
that, dealing with verbal predicates, we are not so much concerned with ḥāl
constituents, but mainly with the fact of their being xabar.
While studying verbal CCs from a ‘horizontal’ perspective, the ‘vertical’
perspective is not abandoned either. A close examination of the circumstan-
tial-predicative variety of clauses in Classical Arabic discloses the strong
formal tie by which all are bound: although considerably different from each
other, in all these clauses the same set of verbal forms, marking distinct as-
pects of the complex predication, recurs. I shall apply the term ‘the predica-
tive paradigm’ to this set of forms. The following discussion will focus on this
paradigmatic regularity, which can be shown to cross-cut diverse syntactic
levels, and the semantic oppositions marked by the predicative forms in each
clause type.
3. The predicative paradigm

The term ‘predicative’, as used here, refers to the syntactic position assumed
by a predicate whose function is to complete the content expressed by another
predicate, so as to form a complex predication.9 Complex predications consist
of (at least) two predicates: the main or primary predicate, which is the gram-
matical nucleus of the complex, and the secondary predicate, which is often
the semantically salient constituent. The fact that both predicates are fused
into one complex entails that not only the predicative, but also the main verb
does not constitute a self-contained predication; rather, both predicates depict
one common occasion.
9 The term ‘converb’ would have been quite proper for the description of the predicative
forms in Arabic, if not typically associated with non-finite verbs, cf. Haspelmath’s
(1995, 3) definition of a converb as ‘a nonfinite verb form whose main function is to
mark adverbial subordination’. I prefer the term ‘predicative’ for being general enough,
i.e., for not being necessarily associated with a specific word-class (e.g. verb or noun) or
a grammatical realization thereof (e.g. non-finite or accusative).

In Classical Arabic, the predicative paradigm consists of three forms: the

prefix conjugation form yafʿalu, the participle fāʿilVn/mafʿūlVn and the modi-
fied suffix conjugation form qad faʿala. In this syntactic slot one may speak
of a basic aspectual meaning of the forms: yafʿalu marking a dynamic-
progressive situation, the participle marking a static state and qad faʿala
marking a state resulting from a previous process. As for their temporal val-
ue, the predicative forms are essentially co-temporal, indicating either simul-
taneity or coincidence with the time frame established in the main clause:
Table (1) – The predicative paradigm

PREDICATIVE FORM ASPECT TEMPORAL VALUE
yafʿalu dynamic-progressive simultaneous, coincidental

(terminal)
fāʿilVn/mafʿūlVn static simultaneous
qad faʿala resultative coincidental (initial)
The triad of yafʿalu, fāʿilVn/mafʿūlVn and qad faʿala constitutes the core of
the predicative paradigm. It is not the normal case for the suffix conjugation
form faʿala to function as a predicative. This may be explained by the fact that
faʿala is used to indicate self-contained events (whence its use as the narrative
form), not coinciding with other events. The same goes for the future form sa-
yafʿalu, which rarely participates in complex predications (see below 4.1.4).
Quite often, when faʿala and sa-yafʿalu are used, the predication involves a
certain abstraction at the semantic level, allowing for some extension of the
notion of common occasion.
4. The domain of complex predications

A wide definition of complex predications, as such realizing a certain degree
of syntactic and semantic integration between (at least) two predicates, covers
a large and quite heterogenic group of structures. However, what is clearly not
included in this group is the coordination of main predicates marking self-
contained events. One may sometimes encounter parallel versions of narra-
tives, where the same set of events is either ‘compressed’ into one predication
or presented in sequence. The difference between these two forms of packag-

ing is significant: in the first, an internally-viewed detailed scene is depicted,

whereas in the latter the events are plainly reported by the external voice of
the narrator. This distinction is clearly evident in (4) and (5), featuring two
versions of the same story:
(4) ṯumma nazala qābīlu min-a l-ǧabali ʾāxiḏan bi-yadi ʾuxti-hī (Taʾrīx 1,
144)
then descendSC3msg Qābīl-nom from DEF-mountain-gen takePTCPmsg-acc in-
hand-gen sister-gen-PRON3msg
‘Then Cain descended from the mountain holding his sister by the hand.’
(5) lammā qatala qābīlu ʾaxā-hu hābīla ʾaxaḏa bi-yadi ʾuxti-hī ṯumma habaṭa
bi-hā min ǧabali būḏa (Taʾrīx 1, 144)
when killSC3msg Qābīl-nom brother-acc-PRON3msg Hābīl-acc takeSC3msg in-
hand-gen sister-gen-PRON3msg then go_downSC3msg with-PRON3fsg from moun-
tain-gen Būḏ-gen
‘After Cain killed his brother Abel he took his sister by the hand and then
went down with her from mountain Būḏ.’
In the following sections, I will start by discussing closely integrated com-
plexes, in which the predicative form is embedded; then, one type of depend-
ent clauses – the syndetic circumstantial clause – will be discussed, and final-
ly, mutually-dependent, setting and presentative clauses, will be shortly exam-
ined.10 These various constructions are all tied by the presence of the predica-
tive paradigm.
4.1 Verbal complexes

The term ‘verbal complex’ covers various manifestations of [main verb +
embedded verb]. These range from closely integrated structures, involving
auxiliaries and modifying verbs, to lexically and grammatically looser ones,
where the main verb retains its full semantics.11 The main verb is the gram-
matical nucleus of the complex, in the sense that it marks the syntactic status
of the entire complex, thereby ‘adjusting’ it into the text; the embedded verb
10 I do not discard the notion of embedding when discussing clause-combining (see above
n. 8); rather, I maintain the distinction between ‘embedded’ and ‘dependent’ clauses: the
first commute with plain nominal forms and are juxtaposed to the main clause, the latter
are appended to the main clause by means of a conjunction.
11 The same verb can be realized either as a lexically ‘full’ verb (tāmm) or as a lexically
‘deficient’ one (nāqiṣ), which is then followed by a predicative form.

is usually the lexical pivot of the complex. Nonetheless, it should be stressed

that both verbs convey some amount of grammatical and lexical information:
the main verb is never entirely depleted (even the auxiliary kāna may be said
to convey the notion of ‘being’), while the predicative form marks such cate-
gories as number, gender, diathesis and aspect.
Verbal complexes present the same syntactic structure, regardless of the
lexeme of the main verbs: the predicative – a verb(-derived) form in adverbial
position – is juxtaposed to the main verb;12 both verbs exhibit subject-identity
(including inalienable entities, see below (16)). The adverbial status of the
predicative obtains a formal mark when the predicative is realized as a nomi-
nal (verb-derived) form, i.e., as the accusative participle. The finite forms
yafʿalu and qad faʿala occupy the same syntactic position as the participle and
thus acquire – by virtue of their paradigmatic interrelation – an adverbial
status.
The following discussion will focus on verbal complexes initiated by verbs
of motion and state, perception and permission verbs, and speech verbs. Com-
plexes initiated by auxiliary and modifying verbs, featuring a further semantic
condensation (from a common to a single occasion), will be mentioned in
brief, in order to provide a full picture of the phenomenon of verbal complex-
es. All the examples exhibit verbal complexes in main clauses.
4.1.1 kāna-compounds
Compound kāna forms exhibit the highest degree of integration within a ver-
bal complex. The auxiliary verb kāna indicates either a temporal or a modal
meaning, whereas the predicative form expresses the content of the verbal
situation as well as its internal unfolding. The opposition between the predica-
tive forms resides thus in two domains: (a) the lexical domain, to which the
issues of lexical compatibility and the valence of the verb-phrase belong; and
(b) the grammatical domain, in which the aspect marked by the verbal form
comes into play.
The compound form kāna yafʿalu is the least lexeme-sensitive of all verbal
forms:13 it may convey all types of verbal situations, both static and dynamic
12 On the internal constitution of compound verb forms and the adverbial status of the
predicative complement, see Goldenberg (2000).
13 The compound form kāna yafʿalu was thoroughly studied by Nebes (1982). Nebes
concludes that kāna yafʿalu is an imperfect whose marked time reference is past. The
present discussion is not concerned with the general function of kāna yafʿalu; rather, the

(telic and a-telic), and be realized in both intransitive (active and passive) and
transitive verb-phrases. 14 As for its grammatical aspect, yafʿalu depicts an
unbounded situation: either one that continues throughout the period of time
indicated by kāna, or one that constantly repeats itself. The repetition is
frequentative by nature, i.e., it is not a mere iteration of the verbal situation,
but a regular and predictable recurring which is valid through the entire period
of time indicated by kāna.15 These distinctions are illustrated in (6)-(7):
(6) wa-kāna rasūlu llāhi yuḥibbu l-faʾla wa-yakrahu l-ṭīrata (Maġāzī 1, 218)
and-beSC3msg messenger-nom God-gen lovePC3msg DEF-good_omen-acc and-
hatePC3msg DEF-evil_omen-acc
‘The Messenger of God used to love the good omen and to hate the evil
omen.’
(7) wa-kāna mraʾan tanaṣṣara fī l-ǧāhiliyyati wa-kāna yaktubu l-kitāba l-
ʿibrāniyya fa-yaktubu min-a l-ʾinǧīli bi-l-ʿibrāniyyati mā šāʾa llāhu ʾan
yaktuba (Ṣaḥīḥ 1, 5)
and-beSC3msg man-acc become_ChristianSC3msg in DEF-pre_Islam-gen and-
beSC3msg writePC3msg DEF-writing-acc DEF-Hebrew-acc and-writePC3msg of DEF-
Gospel-gen in DEF-Hebrew-gen what wantSC3msg God-nom that writeSBJV3msg
‘And he was a man [who] became Christian in the pre-Islamic era; he used
to write in the Hebrew script and would write in Hebrew whatever God
wished him to write from the Gospel.’
The compound form kāna fāʿilan/mafʿūlan shows a clear contrast to kāna
yafʿalu, at both the lexical and the grammatical levels. The predicative parti-
ciple is encountered almost exclusively in intransitive configurations. It is
used to indicate a static situation which is viewed in its entirety, i.e., which
does not break down into internal phases or recurrent instances.
contrast between yafʿalu and the other verbal forms which co-occur with kāna is in fo-
cus.
14 ‘Intransitive’ and ‘transitive’ are used here in a strict sense, to refer to the grammatical
relation between a verbal situation and its accusative complement(s). As a semantic
concept, transitivity is obviously multi-faceted and scalar, see Hopper and Thompson
(1980).
15 Cf. Kleiber (1987, 115) on the contrast between iterative and frequentative repetition:
‘Une phrase simplement itérative est une phrase qui présente une situation comme étant
vérifiée à deux, trois… plusieurs reprises à l’intérieur d’un intervalle temporel […] Une
phrase fréquentative, au contraire, présente l’itération comme s’étendant sur tout
l’intervalle temporel’.

The predicative participle is used with stative lexemes, as illustrated in (8);

notice that muḥibban is connected with its object by means of the preposition
l-,16 whereas yuḥibbu in (6) has a direct object:
(8) wa-kullu-hum kāna lī muḥibban wa-ʾilay-ya māʾilan wa-lī muṭīʿan
(Riwāyāt 1, 35)
and-all-nom-PRON3mpl beSC3msg to-PRON1csg lovePTCPmsg-acc and-to-PRON1csg
inclinePTCPmsg-acc and-to-PRON1csg obedientPTCPmsg-acc
‘And everyone loved me and was favorably inclined to me and was obedi-
ent to me.’
The participle is also compatible with potentially transitive lexemes. In such
cases, however, transitivity is often not exercised, as the verb occurs without
an explicit object complement. Comparing the participle in (9) and yafʿalu in
(7), we observe two points of contrast: (a) the participle kātiban has no object
complement; (b) while yaktubu indicates recurring instances of writing, kāti-
ban expresses not only a state, but one so stable that has achieved the status of
a skill. This stands to reason, as the participle is an adjectival form, which by
itself solely indicates an attributive relation. Whether that be accidental (tem-
porary) or inherent is not specified by the participial form, but only by the
context:17
(9) wa-qāla li-ʾuxti-hī ʾaʿṭī-nī hāḏihi l-ṣaḥīfata […] ʾanẓuru mā hāḏā llaḏī
ǧāʾa bi-hī muḥammadun wa-kāna ʿumaru kātiban (Sīra 1, 226)
and-saySC3msg to-sister-PRON3msg giveIMP2fsg-PRON1csg DEMfsg DEF-
leaf_of_book-acc […] look_atPC1csg what DEMmsg RELmsg comeSC3msg with-
PRON3msg Muḥammad-nom and-beSC3msg ʿUmar-nom writePTCPmsg-acc
‘And he said to his sister: Give me this leaf [of book…] so I will take a
look at that which Muḥammad has brought; and ʿUmar was literate (lit.
‘writer’)’.
16 The Arab grammarians consider this l- as al-lām li-taqwīyat al-ʿāmil ‘the lām which
strengthens the regent’, see Wright (1898 II, §29). Being a nominal form, the participle
has less ‘power’ to govern an object complement; the lām thus serves as an explicit ex-
ponent of this grammatical relation.
17 Cf. Reckendorf (1906, 256). There are languages in which the distinction between an
accidental and an inherent attribute is marked on the (inflected) adjectival form, see
Goldenberg (1991).

The compound form kāna qad faʿala (or qad kāna faʿala) embodies the mean-
ings of anteriority and completion. Anteriority is doubly marked, by both
the modifier qad and the auxiliary kāna. kāna qad faʿala functions in a similar
way to kāna faʿala: it is mostly used to indicate background in the narrative.
kāna qad faʿala has, however, a higher frequency and a wider distribution
than kāna faʿala (in main clauses). The only lexical class which does not
occur with kāna qad faʿala is that of verbs of state such as ʾaqāma ‘to stay’.
This might be explained by the fact that kāna qad faʿala – due to the effect of
qad – indicates a temporally framed situation, which is incompatible with
stative background descriptions. Indeed, we often encounter kāna qad faʿala
in contexts where temporality, or to be more precise, the successive order of
the events, is salient to the narrative:
(10) fa-lammā raǧaʿa baʿda ḥīnin ṭalaba ḥadīda-hū wa-kāna l-raǧulu qad
bāʿa-hū (Kalīla wa-Dimna, 119)
and-when returnSC3msg after time-gen ask_forSC3msg iron-acc-PRON3msg and-
beSC3msg DEF-man-nom MOD sellSC3msg-PRON3msg
‘And after a while when he came back he asked for his iron, [but] the man
had already sold it.’
The compound form kāna faʿala is less frequently used than kāna qad faʿala.
In narratives, kāna faʿala is used when the relative order of the events is not
deemed as important as the assertion of their actual occurrence. Notice the
difference between (10) and (11), extracted from the same story: in the first
case the temporal sequence is crucial to the point of the narrative (the man
claims back his iron after the iron has already been sold); in the latter case, it
is the events themselves (reported in the direct speech) that are given the most
focus:
(11) kuntu waḍaʿtu ḥadīda-ka fī nāḥiyatin min-a l-bayti fa-ʾakala-hū l-
ǧurḏānu (Kalīla wa-Dimna, 119)
beSC1csg placeSC1csg iron-acc-PRON2msg at side-gen of DEF-house-gen and-
eatSC3msg-PRON3msg DEF-rats-nom
‘I had placed your iron at a corner of the house and the rats ate it.’
4.1.2 Modifying verbs

Modifying verbs, ʾaxawāt kāna ‘kāna’s sisters’ in the grammatical tradition,
serve to describe a certain phase or aspect of the verbal situation, which is
expressed by the predicative form. In Classical Arabic, modifying verbs com-

prise several lexical classes;18 the main semantic distinction, however, can be
drawn between two groups of verbs: those which indicate the initial phase of
the verbal situation and those indicating its continuance or duration. There
are no modifying verbs referring to the terminal phase of the verbal situation
or to its accomplishment, hence this group of verbs is incompatible with the
resultative meaning of qad faʿala.
The predicative form yafʿalu is compatible with both groups of modifying
verbs. The most common representative of the first is ǧaʿala ‘to start’:
(12) fa-ǧaʿalū yaḍribūna-hum ḥattā nuqiḍat ṣufūfu-hum (Maġāzī 1, 226)
and-startSC3mpl hitPC3mpl-PRON3mpl until destroySC.PASS3fsg lines-nom-
PRON3mpl
‘And they started to hit them until their lines were destroyed.’
To the same group of inchoative verbs belong also verbs indicating immi-
nence or intention: these verbs do not refer to the actual outset of the verbal
situation but to the phase immediately preceding it, either in the physical or in
the mental world of the agent:
(13) fa-lammā ntahā ʾilā l-nahri lam yaǧid ʿalay-hi qanṭaratan li-yaqṭaʿa-hū
wa-l-ḏiʾbu kāda yudriku-hū (Kalīla wa-Dimna, 63)
and-when getSC3msg to DEF-river-gen NEG findJUSS3msg on-PRON3msg bridge-acc
to-crossSBJV3msg-PRON3msg and-DEF-wolf-nom be_aboutSC3msg reachPC3msg-
PRON3msg
‘And when he got to the river he did not find a bridge to cross it over,
while the wolf was about to reach him.’
yafʿalu often co-occurs with the second group of modifying verbs, indicating
the continuance or duration of the verbal situation. The predicative participle
is also quite common with the second group of modifying verbs. Example
(14) illustrates the contrast between the predicative participle and yafʿalu with
the modifying verb mā zāla ‘to not cease’: the first depicts a static situation
while the latter depicts a dynamic one:
18 Modifying verbs, like kāna, can also be used as full verbs. See Waltisberg (2009 198-
199) for some ambiguous examples, where the verbal form may be interpreted either as
a modifying or a full verb.

(14) wa-qīla ʾanna-hū lam yazal muqīman bi-makkata yaḥuǧǧu wa-

yaʿtamiru (Taʾrīx 1, 164)
and-saySC.PASS3msg that-PRON3msg NEG ceaseJUSS3msg stayPTCPmsg-acc in-
Makka-gen performing_the_ḤaǧǧPC3msg and-performing_the_ʿUmraPC3msg
‘It was reported that he continued to stay in Mecca, performing the pil-
grimages of the Ḥaǧǧ and the ʿUmra.’
As mentioned above, modifying verbs – expressing inchoative or durative
meaning – are incompatible with qad faʿala. The form faʿala also rarely co-
occurs with modifying verbs; the few existing attestations stem mostly from
poetry.19 In my corpus I have encountered only one example in which a faʿala
form co-occurred with a modifying verb, as illustrated in (15); the semantic
compatibility of ʾarāda ‘to want’ and ʿasā ‘it might be’ may be explained by
the fact that, as opposed to qad faʿala, faʿala does not imply the complete and
concrete realization of the verbal situation:
(15) fa-ʿasā-hu ʾarāda l-tafḍīla fī l-qismati (Buxalāʾ, 91-92)
and-it_might_beSC3msg-PRON3msg wantSC3msg DEF-preference-acc in DEF-
division-gen
‘Perhaps he wanted [to be given] preference in the allotment [of the
gifts]?’
4.1.3 Motion and state verbs

Verbs expressing a movement towards a destination or a certain position or
location in space are very common in complex predications. Such verbs indi-
cate the outset or the setting of the verbal situation which is specified by the
following predicative form. With this group of verbs we find the predicative
triad yafʿalu, fāʿilan/mafʿūlan and qad faʿala, marking the opposition between
a progressive, a static and a resultative aspect, respectively. As both the main
verb and predicative verb refer to the same situation and are co-temporal,
faʿala – being neither simultaneous nor coincidental – is excluded from the
predicative paradigm.
In both the traditional and modern grammatical literature, verbal complex-
es initiated by motion and state verbs provide the most typical example of
circumstantial constructions (see above example 1). According to the Arab
grammarians, the predicative form yafʿalu has two manifestations: (a) as ḥāl
19 See Reckendorf (1921, 297) for poetry quotations such as ʾaṣbaḥat ʿaḏalatnī.

muqārin ‘simultaneous circumstantial’, or as (b) ḥāl muqaddar ‘intended

circumstantial’ (cf. Wright 1898 II, §8, 19-20). Modern grammars maintain
the same distinction between yafʿalu forms which are ‘simultaneous with or
following the action expressed by the governing verb’ (Fischer 2002, 220). In
a short article, Abboud (1986) diverts attention from the predicative yafʿalu to
the semantics of the main verb: since yafʿalu co-occurs with ‘event-
completion’ verbs, it may refer either to the event or to its ‘eventual comple-
tion’. According to Abboud, such an explanation ‘obviates the need for a ḥāl
muqaddar’ (195). A somewhat different view is presented by Waltisberg
(2009, §5.2 and §5.6) in his work on circumstantial clauses. Although he, too,
ascribes the interpretation of yafʿalu to the semantics of the main verb,
Waltisberg suggests a neat separation between a ‘modal’ (simultaneous) and a
‘final’ (posterior) function of yafʿalu, which are then paired off with syndetic
circumstantial clauses and final clauses respectively.20
A detailed examination of all the possible combinations reveals that the
temporal value of yafʿalu is not solely determined by the content of the main
verb, nor by that of the predicative verb, but by the interaction of both. We
observe a general rule: if (at least) one of the verbal lexemes is potentially
unbounded (a-telic), then yafʿalu is interpreted as simultaneous (totally over-
lapping); if neither is unbounded, then yafʿalu is interpreted as coincidental
(partially overlapping), i.e., the situation expressed by yafʿalu proceeds from
the one expressed by the main verb. This semantic relation should not be
analyzed as mere succession, since both the main verb and yafʿalu refer to the
same occasion, the first depicting its outset or setting, the latter its destination.
The rule outlined above is demonstrated in the following set of examples,
where the main verb indicates: (a) movement towards a destination, i.e., mo-
tion and goal; (b) movement in space with no goal; or (c) static position in
space. In (16)-(17) the main verb belongs to the first group of motion verbs
while the predicative verb indicates an unbounded situation, yafʿalu is thus
interpreted as simultaneous; in (18) the main verb belongs to the first group of
motion verbs while the predicative verb indicates a bounded situation, yafʿalu
is thus interpreted as coincidental; in (19) the main verb belongs to the second
group of motion verbs, yafʿalu is thus interpreted as simultaneous; also in
(20), where the main verb belongs to the third group of state verbs, yafʿalu is
interpreted as simultaneous:
20 See also the review of Waltisberg in Marmorstein (2011, 381-382).

(16) fa-raǧaʿa bi-hā rasūlu llāhi yarǧufu fuʾādu-hū (Ṣaḥīḥ 1, 5)

and-returnSC3msg with-PRON3fsg messenger-nom God-gen shiverPC3msg heart-
nom-PRON3msg
‘And the Messenger of God returned with them (i.e. the verses) his heart
shivering.’
(17) fa-labisa dirʿa-hū wa-ʾaxaḏa sayfa-hū fa-xaraǧa yaʿdū (Maġāzī 1, 370)
and-wearSC3msg armor-acc-PRON3msg and-takeSC3msg sword-acc-PRON3msg
and-go_outSC3msg runPC3msg
‘He put on his armor, took his sword and went out running.’
(18) fa-xaraǧū yaṭlubūna-humā fī kulli waǧhin (Riwāyāt 2, 24)
and-go_outSC3mpl look_forPC3mpl-PRON3mdu in all-gen direction-gen
‘They went out looking for both of them in all directions.’
(19) wa-marrū yaḍribūna bi-l-dufūfi wa-yazmirūna bi-l-mazāmīri (Maġāzī 1,
375)
and-goSC3mpl strikePC3mpl in-DEF-tambourines-gen and-playPC3mpl in-DEF-
pipes-gen
‘They marched striking tambourines and playing the pipes.’
(20) fa-bātū yaṭlubūna-hū ḥattā yaʾisū min-hu (Riwāyāt 2, 178)
and-spend_the_nightSC3mpl look_forPC3mpl-PRON3msg until give_hopeSBJV3mpl
of-PRON3msg
‘They spent the night looking for him until they gave up all hope of [find-
ing] him.’
Not only with verbs of (vectorial) motion, but also with verbs of caused mo-
tion, yafʿalu depicts the goal of the event launched by the main verb:
(21) fa-ʾarsala ʾilay-hi yadʿū-hu (Riwāyāt 2, 29)
and-sendSC3msg to-PRON3msg callPC3msg-PRON3msg
‘And he send to him [a message/messenger] inviting him [to come].’
yafʿalu in such cases is not strictly successive, neither does it indicate finality;
rather, it indicates the terminal stage of the verbal situation. That the two
notions, i.e., ‘final’ and ‘terminal’, are not simply overlapping can be demon-
strated, inter alia, by the fact that proper final clauses, such as are introduced
by an explicit operator (e.g. li- ‘for’), are external to the verbal situation and

hence can be negated, whereas the terminal yafʿalu – being an internal and
inseparable part of the verbal situation – is never negated.
While coincidental or terminal yafʿalu forms cannot be negated, yafʿalu
indicating simultaneity does exhibit negation with lā. Quite often, lā yafʿalu
paraphrases the content expressed by the preceding (affirmative) predicative
form:
(22) wa-bātat ṣāhilatan xaylu-hum lā tahdaʾu (Maġāzī 1, 217)
and-spend_the_nightSC3fsg neighPTCPfsg-acc horses-nom-PRON3mpl NEG
calm_downPC3fsg
‘Their horses stayed up the night neighing, they would not calm down.’
The predicative participle depicts a static situation. It may indicate: (a) the
outcome of a previous process, in the passive form; (b) the persistence in a
certain state, with dynamic lexemes; or (c) the endurance of a state, with
stative lexemes. These three options are illustrated in the examples below. In
(24), as is often the case, the participle is used to describe physical appear-
ance, dressing and, especially, the girding of a sword (other parallel examples
are Riwāyāt 2, 185: mutaqallidan; Sīra 1, 225: mutawaššiḥan); example (25)
demonstrates the functional contrast between the participle and yafʿalu with
regard to transitivity: the first figures in intransitive verb-phrases whereas the
latter takes an object complement:
(23) fa-makaṯa mumallakan ʿalay-hā ʾašhuran (Riwāyāt 2, 184)
and-remainSC3msg enthronePTCP.PASSmsg-acc on-PRON3fsg months-acc
‘He remained its king (lit. ‘enthroned’) for several months.’
(24) fa-ʾaqbala muṣlitan sayfa-hū fī nafarin min-a l-yahūdi (Maġāzī 1, 372)
and-approachSC3msg drawPTCPmsg-acc sword-acc-PRON3msg in group-gen of
DEF-Jews-gen
‘He approached unsheathing his sword amid a group of Jews.’

(25) fa-xaraǧnā xāʾifīna naxāfu l-raṣada (Maġāzī 1, 28)
and-go_outSC1cpl fearPTCPmpl-obl fearPC1cpl DEF-ambush-acc
‘We set out afraid, we were fearing an ambush.’
The last example illustrates a general principle of complex predications. One
may refer to it as the principle of ‘increasing specificity’: each predicative is
‘piled up’ upon the previous one, thereby depicting in greater detail the given

situation. The increased specificity is obtained by the mere accumulation of

predicates, and not by their internal order. Consider, for instance, (26)-(27), in
which rakiba ‘to ride’ functions as either the main (specified) verb or the
predicative (specifying) form:
(26) lam ʾarkab xaṭwatan ḏāhiban wa-lā rāǧiʿan (Maġāzī 1, 26)
NEG rideJUSS1csg step-acc goPTCPmsg-acc and-NEG returnPTCPmsg-acc
‘I did not ride a single step either going or coming.’
(27) ʾaqbaltu rākiban ʿalā ḥimārin ʾatānin (Ṣaḥīḥ 1, 31)
approachSC1csg ridePTCPmsg-acc on she_ass-gen
‘I came close riding on a she-ass.’
Quite often, the lexemes of both the main verb and the predicative form per-
tain to the same class of motion verbs. According to Waltisberg (2009,
286ff.), in such cases the participle serves to mark ‘situation-identity’ between
the two verbal forms. It appears, however, that the notion of ‘situation-
identity’ fails to capture the specifying function of the participle and its se-
mantic contribution to the verbal complex. To be sure, there are cases where
the content of both verbs is pretty similar. However, even in these, one is not
concerned with mere tautology, but with the elaboration of the content of the
main verb, often by indicating the point of departure or the direction of the
motion:
(28) wa-ʾaqbala ʾabū ǧubaylata sāʾiran min-a l-šāmi (Riwāyāt 2, 11)
and-approachSC3msg ʾAbū_Ǧubayla-nom goPTCPmsg-acc from al-Šām-gen
‘And ʾAbū Ǧubayla came proceeding from al-Šām.’
(29) ḥattā qumtu fī qiblati-hī mustaqbila-hū (Sīra 1, 228)
until get_upSC1csg in direction_of_praying-gen-PRON3msg facePTCPmsg-acc-
PRON3msg
‘Until I stood in his direction of praying facing him.’

Unlike yafʿalu, the participle is rarely interpreted as indicating the terminal
stage or destination of the verbal situation. Example (30) is one case that may
be interpreted as such:
(30) ǧiʾtu-ka ʿāʾiḏan bi-ka (Riwāyāt 1, 55)
comeSC1csg-PRON2msg seek_protectionPTCPmsg-acc in-PRON2msg

‘I came to you asking for your protection.’

In other cases where the main verb indicates vectorial motion, what one usual-
ly finds is the participle indicating an appointment or assignment, rather than
the destination or goal. As an adjectival form, the participle is most suited for
the expression of such attributes, whether these have a temporary or a perma-
nent validity. Example (31) provides a good illustration of the distinction
between the participle, indicating an appointment, i.e., an (intended) attribute
(‘being sent as’), and yafʿalu, which breaks down this attribute into its actual
instances (‘being sent to’):
(31) fa-baʿaṯa llāhu ʾiblīsa qāḍiyan yaqḍī bayna-hum (Taʾrīx 1, 85)
and-sendSC3msg God-nom ʾIblīs-acc judgePTCPmsg-acc judgePC3msg between-
PRON3mpl
‘God sent ʾIblīs as a judge to judge among them.’

As already mentioned above, a predicative faʿala is incompatible with motion
verbs, due to its self-contained temporal framing (this is not to be confused
with the notion of telicity: faʿala, with stative lexemes, may well indicate
unbounded persisting situations). A predicative qad faʿala, on the other hand,
does co-occur with motion verbs, indicating a process whose result is coinci-
dental with the situation expressed by the main verb. As opposed to the coin-
cidental yafʿalu, the tangent point of qad faʿala and the main verb is not the
terminal but the initial stage of the latter. Like the other predicative forms, qad
faʿala may refer to the motion itself, indicating the point of departure, or to
the physical state of the agent:
(32) fa-ǧāʾa ʾaʿrābiyyun qad ʾaqbala min tihāmata (Maġāzī 1, 46)
and-comeSC3msg Bedouin-nom MOD approachSC3msg from Tihāma-gen
‘A Bedouin arrived [after] approaching from Tihāma.’
(33) wa-ʾaqbala l-mušrikūna qad ṣaffū ṣufūfa-hum (Maġāzī 1, 220)
and-approachSC3msg DEF-polytheists-nom MOD alignSC3mpl lines-acc-PRON3mpl
‘The polytheists approached [being] already arranged in lines.’
4.1.4 Perception and permission verbs

Perception verbs and (manipulation) verbs indicating permission form togeth-
er a sub-group of complement-taking verbs in Classical Arabic. Both these
classes of verbs head raising constructions. The term ‘raising’ refers to the

syntactic fusion of two clauses, a complement-taking verb and its proposition-

al complement, whereby the subject of the second clause is fronted to the
object position of the first. The raised element is in fact shared by both claus-
es, and cannot be analyzed as an exclusive member of either.21
The mechanism of raising allows for two interpretations of perception
verbs: either as indicators of concrete perception, of an object and its condi-
tion (e.g. ‘I saw him doing’), or as indicators of notional perception, of a
whole situation or fact (e.g. ‘I saw that he was doing’). The latter use may
trigger a further shift or abstraction of the meaning of the verb, from physical
perception to mental comprehension (e.g. ‘I realized that he was doing’).22
The distinction between the patterns of concrete and notional perception is
not only semantic, but it also has syntactic correlates: (a) complement clauses
introduced by the operator ʾanna are not free variants of raising constructions,
but mostly interchange with the pattern of notional perception (Waltisberg
2009, 340); and (b) the paradigm of predicative forms which is compatible
with the expression of notional perception is wider and includes also verbs not
indicating co-temporality with the main verb. Thus, while the pattern of con-
crete perception involves only the predicative triad, yafʿalu, fāʿilan/mafʿūlan
and qad faʿala, the pattern of notional perception also includes faʿala and sa-
yafʿalu, both referring to events which are not envisaged as co-occurring with
the moment of perception.
The predicative form yafʿalu occurs with both patterns of concrete and no-
tional perception. With the first pattern, yafʿalu depicts an ongoing situation,
simultaneous with the moment of perception:
(34) fa-raʾā-nī ʾatawaḍḍaʾu min kūzin xazafin (Buxalāʾ, 37)
21 In complex clauses such as ‘I found her gone’, Jespersen (1924, 122) suggests regarding
the entire combination (‘nexus’) her gone as the object of the main verb. According to
Givón (2001 II, 272), such cases exhibit the process of raising, whereby an argument of
the subordinate clause is converted to an argument of the main clause. Waltisberg (2009,
322-323), on the other hand, views the raised element as still belonging to the embedded
clause. As a matter of fact, this question cannot be decided, for the raised element is
formally marked (through its case and agreement) as relating to both clauses at the same
time.
22 According to the Arabic grammatical tradition, when raʾā and waǧada are not intended
in their physical denotation (i.e., in the sense of ruʾyat al-ʿayn ‘the seeing of the eye’ or
wiǧdān al-ḍālla ‘the finding of the lost beast’), but rather in their mental denotation,
their ‘second object’ is indispensable to the structure, serving as the predicate of the first
object (Sībawayhi I, 13).

and-seeSC3msg-PRON1csg perform_ablutionPC1csg from jug-gen pottery-gen

‘He saw me performing the ablution using [water from] a pottery jug.’
yafʿalu is also compatible with raʾā, when this indicates – rather than concrete
ocular perception – the seeing of a certain scene in a dream or the envisioning
of a scene:
(35) ʾinn-ī qad raʾaytu ruʾyan […] wa-ʾarā bn-ī yaṭlubu-nī ṭalaban ḥaṯīṯan
(Sīra 1, 254)
INTRO-PRON1csg MOD seeSC1csg dream-acc […] and-seePC1csg son-PRON1csg
searchPC3msg-PRON1csg search-acc hastily-acc
‘I dreamt (lit. ‘I saw a dream’) […] and I see my son looking for me anx-
iously.’
With the second pattern of notional perception or comprehension, yafʿalu
expresses a prediction, a situation that is not concomitant but posterior to that
indicated by the main verb. As the conceived situation does not coincide with
the moment of conceiving, the future form sa-yafʿalu may also be used:
(36) fa-tarā muḥammadan yaḥṣiru-nā sanatan (Maġāzī 1, 368)
and-seePC2msg Muḥammad-acc besiegePC3msg-PRON1cpl year-acc
‘Do you think that Muḥammad will besiege us for a year?’
(37) wa-ʾammā ṭalabu bn-ī ʾiyyā-ya ṯumma ḥabsu-hū ʿann-ī fa-ʾinn-ī ʾarā-hu
sa-yaǧhadu ʾan yuṣība-hū mā ʾaṣāba-nī (Sīra 1, 254)
and-TOP search-nom son-PRON1csg PRON1csg-acc then confinement-nom-
PRON3msg from-PRON1csg and-INTRO-PRON1csg seePC1csg-PRON3msg MOD-
strivePC3msg to befallSBJV3msg-PRON3msg REL befallSC3msg-PRON1csg
‘And as for my son’s looking for me and being withheld from me, I see it
[as if] he will strive so that what happened to me will happen to him [too].’
The same as with verbs of motion and caused motion, when yafʿalu co-occurs
with permission verbs it may be coincidental, referring to the terminal stage of
the complex situation:
(38) fa-qad ʾamara-nā ʾan lā nadaʿa-ka tastaqirru ʿalā l-ʾarḍi (Riwāyāt 1,
248)
and-MOD orderSC3msg-PRON1cpl that NEG letSBJV1cpl-PRON2msg
hold_firmPC2msg on DEF-ground-gen
‘He has instructed us to not let you stick to the ground.’

The predicative participle is compatible with the expression of concrete as

well as notional perception. In both cases it depicts a static situation, one that
exists or persists at the moment of mental or physical perception:
(39) ṯumma ǧāʾat-i mraʾatu l-ḥaǧǧāmi baʿda sāʿatin li-musāmarati ṣadīqati-
hā mraʾati l-ʾiskāfi fa-waǧadat-hā marbūṭatan (Kalīla wa-Dimna, 79)
then comeSC3fsg wife-nom DEF-cupper-gen after hour-gen to-evening_chat-gen
friend-gen-PRON3fsg wife-gen DEF-shoemaker-gen and-findSC3fsg-PRON3fsg
tiePTCP.PASSfsg-acc
‘Then the wife of the cupper came after one hour to have an evening chat
with her friend, the wife of the shoemaker, and she found her tied up.’
(40) fa-lammā waǧada-hū qāʿidan fī ʾaṣḥābi-hī ʾakabba ʿalay-hi wa-ʿānaqa-
hū (Buxalāʾ, 43)
and-when findSC3msg-PRON3msg sitPTCPmsg-acc among friends-gen-PRON3msg
bend_downSC3msg on-PRON3msg and-embraceSC3msg-PRON3msg
‘When he found him sitting among his friends he bent down over him and
embraced him.’
(41) fa-qultu yā rasūla llāhi mā la-ka ʿan fulānin fa-wallāhi ʾinn-ī la-ʾarā-hu
muʾminan (Ṣaḥīḥ 1, 15)
and-saySC1csg VOC messenger-nom God-gen what to-PRON2msg on so-and-so-gen
and-by_God INTRO-PRON1csg EMPH-seePC1csg-PRON3msg believePTCPmsg-acc
‘So I said: O Messenger of God, what do you have [in mind] about so-and-
so, for by God, I think he is a believer.’
Also with verbs indicating permission, the predicative participle depicts a
static situation; notice that in (42) the participle, as elsewhere, is intransitive,
whereas the yafʿalu that follows has an object complement:
(42) fa-ʾaxbara-nā ʾanna muḥammadan kāna ʿaraḍa li-ʿīri-nā fī badʾati-nā
wa-ʾanna-hū taraka-hū muqīman yantaẓiru raǧʿata-nā (Maġāzī 1, 28)
and-informSC3msg-PRON1cpl that Muḥammad-acc beSC3msg inspectSC3msg to-
caravan-gen-PRON1cpl in start-gen-PRON1cpl and-that-PRON3msg leaveSC3msg-
PRON3msg stayPTCPmsg-acc waitPC3msg return-acc-PRON1cpl
‘And he informed us that Muḥammad was observing our caravan since we

started our [journey] and that he had left him to stay [there] and watch for
our return.’

The predicative form qad faʿala in both patterns of concrete and notional
perception, is used to indicate a process whose resultant state coincides with
the time of perceiving or conceiving:
(43) ʾasmaʿu l-ṣawta qad-i rtafaʿa fī ʾaʿlā qawrā (Riwāyāt 2, 47)
hearPC1csg DEF-voice-acc MOD riseSC3msg from TOP-gen Qawrā-gen
‘I hear the voice already risen at the top of Qawrā.’
(44) ʾinn-ī ʾarā qurayšan qad ʾazmaʿat ʿalā l-xurūǧi (Maġāzī 1, 36)
INTRO-PRON1csg seePC1csg Qurayš-acc MOD decideSC3fsg on DEF-going_out-gen
‘I truly think that Qurayš have already decided to go out.’
As is usually the case, qad faʿala is preferred to faʿala when the chronological
order of the events is deemed salient to the narrative; thus in (45), the fact that
ʿAdī was already dead when the messenger found him, and not just the mere
fact of his death, has great bearing on the later development of the narrative:
(45) ʾinn-ī waǧadtu ʿadiyyan qad māta qabla ʾan ʾadxula ʿalay-hi (Riwāyāt 2,
191)
INTRO-PRON1csg findSC1csg ʿAdiyy-acc MOD dieSC3msg before that enterSBJV1csg
upon-PRON3msg
‘I had found ʿAdī already dead before I entered upon him.’
In some cases, the perceived situation consists of a number of ‘slides’. Ex-
ample (46) is a good illustration of the predicative triad; we observe that the
order of the forms – first the participle, then yafʿalu and qad faʿala – is a fixed
one, regardless of the nature of the matrix clause. This order may be regarded
as iconic, reflecting the decreasing degree of integration of the predicative
form with the main verb:
(46) wa-la-ka-ʾann-ī ʾanẓuru ʾilay-kum ẓāʿinīna yataḍāġā ṣibyānu-kum qad
taraktum dūra-kum xulūfan wa-ʾamwāla-kum (Maġāzī 1, 365)
and-EMPH-as_if-PRON1csg lookPC1csg at-PRON2mpl departPTCPmpl-obl
cry_outPC3msg children-nom-PRON2mpl MOD leaveSC2mpl homes-acc-PRON2mpl
behind-acc and-possessions-acc-PRON2mpl
‘It is as if I look at you departing, your children crying out, [after] you
have left your homes and possessions neglected.’
In my corpus, a predicative faʿala was not too often found in raising construc-
tions. Unlike the temporally-bounded, coincidental qad faʿala, faʿala refers to

a self-contained period of time which is not referential to another. With con-

crete perception, faʿala is found with lexemes indicating an enduring state;
notice in (47) the indefinite time frame indicated by the adverb zamānan ‘for a
while’:
(47) ʾinn-ī qad raʾaytu l-malika ʾaqāma bi-makāni-hī hāḏā zamānan lā
yabraḥu min-hu (Kalīla wa-Dimna, 73)
INTRO-PRON1csg MOD seeSC1csg DEF-king-acc staySC3msg in-place-gen-PRON3msg
DEMmsg time-acc NEG movePC3msg from-PRON3msg
‘Indeed I have seen [that] the king remained in this place of his for a
while, not moving from it.’
faʿala is more likely to occur when perception is not intended in its physical
sense, but in its mental one. Thus, faʿala is used in visions and dreams, or in
the expression of realizations and conclusions:
(48) raʾaytu rākiban ʾaqbala ʿalā baʿīri-hī […] fa-ʾarā l-nāsa ǧtamaʿū ʾilay-
hi (Maġāzī 1, 29)
seeSC1csg rider-acc approachSC3msg on camel-gen-PRON3msg […] and-seePC1csg
DEF-people-acc gatherSC3mpl to-PRON3msg
‘I saw [in a dream] a rider [that] approached on his camel […] I saw [that]
the people gathered to him.’
(49) fa-ʾinn-ī ʾarā rīḥan qad hāǧat min ʾaʿlā l-wādī wa-ʾinn-ī ʾarā-hā buʿiṯat
bi-naṣri-ka (Maġāzī 1, 29)
and-INTRO-PRON1csg seePC1csg wind-acc MOD riseSC3fsg from TOP-gen DEF-
valley-gen and-INTRO-PRON1csg seePC1csg-PRON3fsg sendSC.PASS3fsg with-help-
PRON2msg
‘I see a wind has risen from above the valley and I think it has been sent to
help you.’
As is the case elsewhere, the predicative faʿala is not encountered in the nega-
tive form; instead, lam yafʿal is used:
(50) inhaḍ-i l-sāʿata ʾilā l-faḍli bni yaḥyā fa-ʾinna-ka taǧidu-hū lam yaʾḏan
li-ʾaḥadin baʿdu (Riwāyāt 1, 30)
get_upIMP2msg DEF-hour-acc to al-Faḍl_ibn_Yaḥyā-gen and-INTRO-PRON2msg
findPC2msg-PRON3msg NEG allowJUSS3msg to-anyone-gen yet
‘Get up [and go] now to al-Faḍl b. Yaḥyā; you will find him not allowing
anyone [in] yet.’

4.1.5 Speech verbs

Another defined class of verbs which are realized in verbal complexes are
speech verbs. The predicative forms which co-occur with speech verbs are
yafʿalu and the participle. These forms increase the specificity of the verbal
situation by referring either to the content expressed or to the vocal quality of
speech itself; notice that (52) exhibits an unusual case where the participle is
followed by an object complement:
(51) qāla l-ʾaʿšā yamdaḥu l-samawʾala (Riwāyāt 2, 27)
saySC3msg al-ʾAʿšā-nom praisePC3msg al-Samawʾal-acc
‘Al-ʾAʿšā said praising al-Samawʾal.’
(52) qāla l-walīdu rāfiʿan ṣawta-hū (Buxalāʾ, 65)
saySC3msg al-Walīd-nom raisePTCPmsg-acc voice-acc-PRON3msg
‘Al-Walīd said raising his voice.’
4.2 Syndetic circumstantial clauses

The predicative paradigm is found not only in verbal complexes, i.e. in em-
bedded clauses, but also at higher syntactic levels. In this section I will discuss
one type of dependent clause, the circumstantial clause, in which the predica-
tive triad is found. The aspectual oppositions between the forms are the same
as those stated above: yafʿalu indicates an ongoing situation or process, the
participle indicates a static situation and qad faʿala indicates a result.
As mentioned above (2.1), the Arab grammarians described the ǧumla
ḥāliyya ‘circumstantial clause’ as a complex (‘periphrastic’) manifestation of
the ḥāl category. However, modern researchers, such as Premper (2002) and
Waltisberg (2009), have demonstrated that the asyndetic ḥāl constituent and
the syndetic ǧumla ḥāliyya do not interchange freely with each other. In fact,
there are some significant formal and functional distinctions between them.
Firstly, the subject of the syndetic CC is not necessarily co-referential with
the subject of the main clause, thus both clauses do not necessarily refer to the
same situation. It should be noted, however, that the subject of the CC is not
entirely new, but can be retrieved from the previous context.23
23 It is rather unusual that the subject of the CC is newly introduced into the text. As the
following example shows, although the subject is indefinite, it is strongly associated
with other topics and hence may be regarded as presupposed (like inalienable entities):
kuntu ʿinda šayxin min ʾahli marwa wa-ṣabiyyun la-hū ṣaġīrun yalʿabu bayna yaday-hi

Secondly, the syndetic CC is not lexeme-sensitive: it does not co-occur

with specific classes of verbs.
Thirdly, syndetic CCs are backgrounded whereas their asyndetic counter-
parts are foregrounded. The latter distinction is often correlated with the tem-
poral relation marked by the syndetic and asyndetic clause, to wit, simultanei-
ty vs. sequentiality or chaining (Premper 2002, 275). Although generally
correct, this correlation is too sweeping: as has already been discussed, the
predicative forms are either concomitant to or coincident with the time frame
set in the main clause, thus they may precede, overlap or proceed from the
situation indicated by the main verb.
The formal and functional distinctions outlined above reflect different de-
grees of integration between the asyndetic and syndetic CCs and their matrix
clause. The asyndetic predicative forms, occupying the position of an adverbi-
al (accusative) complement, show a higher degree of integration than syndetic
CCs, which are connected to their matrix clause as autonomous units (note that
the participle in CCs takes the nominative case!). Also from a functional point
of view, asyndetic predicative forms serve to single out a certain aspect, fea-
ture or stage of the complex situation; syndetic CCs, on the other hand, are
comments elaborating on a certain entity, depicting in greater detail the situa-
tion indicated in the main clause or describing the setting in which the latter
takes place.
Except for some minor cases in which fa- is used, the CC is connected as a
rule with -wa.24 wa- is a general connective particle, indicating the adjoining
of two or more elements or clauses. The particular semantic relation between
the adjoined clauses, whether it be chronological, causal, contrastive or other,
is not indicated by wa- but determined by the given context.25
wāw al-ḥāl ‘the circumstantial wa-’ may introduce either a ‘nominal
clause’ or a ‘verbal clause’: the first is headed by a topic entity with which the
predicate is in full concord; the latter is initiated by a third-person verb whose
subject shows only a partial agreement with it. Both patterns signal the func-
tional distinction between ‘entity-oriented’ and ‘event-oriented’ clauses
(Buxalāʾ 38) — ‘I was at [a place of] a sheikh from the people of Marv, and a young
boy of his was playing in front of him’.
24 For CCs introduced by fa-, see Nebes (1999).
25 For a different view of syndetic CCs, as such indicating a ‘catalogue’ of semantic rela-
tions, see Waltisberg (2009, 358).

(Holes 2004, 251-253).26 When the predicate of the circumstantial is yafʿalu

or the participle, the nominal pattern is mandatory; with qad faʿala, the pat-
tern of the verbal clause is mostly used, although one may encounter a few
cases where the subject is fronted.
Circumstantial clauses whose predicate is yafʿalu may refer to a situation
parallel to the one indicated in the matrix clause. Being more general, the term
‘parallel’ is preferred to ‘simultaneous’, which captures only one possible
interpretation of the circumstantial relation, namely the temporal. The parallel
relation is sometimes interpreted as contrastive, especially when the subject of
the CC is not only different from, but in fact confronted with the subject of the
main clause:
(53) fa-ǧaʿaltu ʾamšī ruwaydan wa-rasūlu llāhi qāʾimun yuṣallī yaqraʾu l-
qurʾāna (Sīra 1, 228)
and-startSC1csg walkPC1csg slowly-acc and-messenger-nom God-gen
get_upPTCPmsg-nom prayPC3msg readPC3msg DEF-Qurʾān-acc
‘I started to walk slowly while the Messenger of God was standing, pray-
ing [and] reciting the Qurʾān.’
(54) fa-daxala ʿalay-hi raǧulun kāna la-hū ǧāran wa-kāna lī ṣadīqan fa-lam
yaʿriḍ ʿalay-hi l-ṭaʿāma wa-naḥnu naʾkulu (Buxalāʾ, 38)
and-enterSC3msg upon-PRON3msg man-nom beSC3msg to-PRON3msg neighbor-acc
and-beSC3msg to-PRON1csg friend-acc and-NEG offerJUSS3msg to-PRON3msg DEF-
food-acc and-PRON1cpl eatPC1cpl
‘When in came a man, a neighbor of his and a friend of mine, and he did
not offer him food, though we were eating.’
Like the asyndetic yafʿalu, the circumstantial yafʿalu – when co-referential
with the subject of the main verb – refers to the same occasion as the latter.
Thus, in (55)-(57), the same verb, viz. qāla ‘to say’, is followed each time by
a circumstantial clause, specifying the manner of speech, its location, or the
content expressed:
26 See also Goldenberg (2006) for the functional distinction between verb-initial sentences
and topicalizations in Arabic.

(55) fa-lammā qāla surāqatu mā qāla wa-huwa yanṭuqu bi-lisāni ʾiblīsa

šaǧuʿa l-qawmu (Maġāzī 1, 39)
and-when saySC3msg Surāqa-nom what saySC3msg and-PRON3msg speakPC3msg
with-tongue-gen ʾIblīs-gen be_courageousSC3msg DEF-people-nom
‘And when Surāqa said what he said, and he was speaking with the tongue
of ʾIblīs, the people were encouraged.’
(56) samiʿtu ʿaliyyan yaqūlu wa-huwa yaxṭubu bi-l-kūfati (Maġāzī 1, 57)
hearSC1csg ʿAliyy-acc sayPC3msg and-PRON3msg deliver_a_speechPC3msg in-al-
Kūfa
‘I heard ʿAlī saying while he was delivering a sermon in Kufa.’
(57) […] qāla wa-huwa yuḥaddiṯu ʿan fatrati l-waḥyi (Ṣaḥīḥ 1, 6)
[…] saySC3msg and-PRON3msg talkPC3msg about intermediate_break-gen DEF-
revelation-gen
‘He said, while delivering a ḥadīṯ about the period of pause in revelation’
In syndetic CCs, yafʿalu is negated by lā. Quite often, lā yafʿalu occurs with
verbs of knowledge, depicting a situation where one subject is ignorant about
the activity of another:
(58) fa-marrat-i l-ḥayyatu ʿalā l-xazanati wa-hum lā yaʿlamūna (Taʾrīx 1,
104)
and-passSC3msg DEF-snake-nom by DEF-keepers-gen and-PRON3mpl NEG
knowPC3mpl
‘The snake passed by the keepers [and entered] while they did not know.’
Circumstantial clauses whose predicate is the participle are either co-
referential with the situation indicated by the main verb, or refer to a parallel
situation, as illustrated in the examples below. We observe that the participle
in (59) follows a non-derived adjectival pattern (viz. faʿīl), depicting the men-
tal state of the subject; in (60) the participle, as elsewhere (see above example
24), is used to refer to the physical appearance of the subject, his dress or
girding (another example is Maġāzī 1, 39: wa-huwa mutawaššiḥun bi-sayfi-
hī):
(59) fa-ǧāʾa-nī wa-huwa ḥazīnun munkasirun (Buxalāʾ, 90)
and-comeSC3msg-PRON1csg and-PRON3msg sad-nom breakPTCPmsg-nom
‘He came to me sad and [heart] broken.’

(60) wa-ʾaxaḏa l-nabiyyu l-qawsa wa-ʾaxaḏa qanātan bi-yadi-hī […] wa-l-

muslimūna mutalabbisūna l-silāḥa (Maġāzī 1, 215)
and-takeSC3msg DEF-prophet-nom DEF-bow-acc and-takeSC3msg spear-acc in-
hand-gen-PRON3msg […] and-DEF-Muslims-nom dressPTCPmpl-nom DEF-weapons-
acc
‘The Prophet took the bow and the spear in his hand […] while the Mus-
lims were putting on their weapons.’
As with the asyndetic participle, the circumstantial participle occurs as a rule
in intransitive verb-phrases, either with intransitive lexemes or, with transitive
lexemes, in the passive form:
(61) laqiya-hū safīhun min sufahāʾi qurayšin wa-huwa ʿāmidun ʾilā l-kaʿbati
(Sīra 1, 246)
meetSC3msg-PRON3msg fool-nom of fools-gen Qurayš-gen and-PRON3msg pro-
ceedPTCPmsg-nom toward al-Kaʿba-gen
‘One of Qurayš’ fools came across him while he was proceeding to the
Kaʿba.’
(62) li-ʾanna-hū lā šayʾa yatawahhamu-hū mutawahhimun fī qawli qāʾili
ḏālika ʾillā wa-huwa mawǧūdun fī qawli qāʾilin (Taʾrīx 1, 58)
because-PRON3msg NEG thing-acc imaginePC3msg-PRON3msg one_imagining-nom
in saying-gen one_saying-gen DEMmsg but and-PRON3msg findPTCP.PASSmsg-nom
in saying-gen one_saying-gen
‘Because there is nothing which one may presume [to be implied] in this
statement without existing in a statement such as […].’
On rare occasions the participle is encountered with an object complement.
We observe, however, that the object in such cases is not a prototypical one,
i.e. an individualized affected entity, but in fact, forms a collocation with the
verbal form:
(63) wa-kayfa yastaṭīʿu ḏālika wa-huwa ʾākilun ʿušban (Kalīla wa-Dimna,
92)
and-how be_ablePC3msg DEMmsg and-PRON3msg eatPTCPmsg-nom grass-acc
‘How is he able to do that while being a grass-eater?’
Circumstantial clauses whose predicate is qad faʿala exhibit in the main
the pattern of the verbal clause, although one may encounter a few cases in
which the subject is fronted, as in (65):

(64) wa-raǧaʿa l-naǧāšiyyu wa-qad ʾahlaka llāhu ʿaduwwa-hū (Sīra 1, 221)

and-returnSC3msg al-Naǧāšiyy-nom and-MOD destroySC3msg God-nom enemy-
acc-PRON3msg
‘The Negus came back [after] God had already destroyed his enemy.’
(65) wa-xaraǧa l-xabaru ʾilā l-nāsi wa-riǧālu banī l-muṣṭaliqi qad-i qtusimū
wa-mulikū (Maġāzī 1, 411)
and-go_outSC3msg DEF-news to DEF-people-gen and-men-nom Banū_al-Muṣṭaliq-
gen MOD divideSC.PASS3mpl and-possessSC.PASS3mpl
‘The news went out to the people, while the men of the Banū Muṣṭaliq had
already been divided [among their captors] and become [their] property
[…].’
A plausible explanation to the different order of wa-qad faʿala vis-à-vis wa-
huwa yafʿalu/wa-huwa fāʿilun may be that the latter, which take the form of
the nominal clause, are indeed plot-external descriptions, sometimes even
generic or encyclopedic comments, centered on a certain entity, while wa-qad
faʿala, though deviating from the main faʿala-plot-line and depicting an ante-
rior event is not purely descriptive, but rather incorporated in the stream of
events.
Circumstantial clauses can be realized in the ʾinna la- clausal pattern. The
particle ʾinna introduces the entire clause whereas la- precedes the predicate.
When the predicate is verbal, la- is prefixed to either yafʿalu or the participle.
The structure wa-ʾinna la- has an emphasizing function: it indicates that the
content expressed in the clause stands against a certain expectation, explicit or
implicit in the surrounding context, and that it is therefore remarkable. There
is an important functional distinction between ordinary circumstantial clauses
and wa-ʾinna la- CCs: unlike the first, the pattern wa-ʾinna la- is not merely
descriptive or orientational (if at all), but it presents the personal evaluation
of the narrator regarding the narrated situation. Thus, in (66) ʿĀʾiša says that it
was an extremely cold day when the revelation came upon the Prophet, but
nevertheless she saw that he was sweating; in (67), Maymūna is reported to be
given medicine, despite the fact that she was fasting:
(66) wa-la-qad raʾaytu-hū yanzilu ʿalay-hi l-waḥyu fī l-yawmi l-šadīdi l-bardi
fa-yufṣimu ʿan-hu wa-ʾinna ǧabīna-hū la-yatafaṣṣadu ʿaraqan (Ṣaḥīḥ 1, 6)
and-EMPH-MOD seeSC1csg-PRON3msg come_downPC3msg upon-PRON3msg DEF-
revelation-nom in DEF-day-gen DEF-strong-gen DEF-cold-gen and-abatePC3msg
from-PRON3msg and-INTRO forehead-acc-PRON3msg EMPH-dripPC3msg sweat-acc

‘And I saw him [when] the revelation came upon him, in a very cold day,
then it departed from him, while his forehead was dripping with sweat.’
(67) fa-la-qad luddat maymūnatu wa-ʾinna-hā la-ṣāʾimatun (Sīra 2, 1007;
Waltisberg 2009, 159)
and-EMPH-MOD administer_medicineSC.PASS3fsg Maymūna-nom and-INTRO-
PRON3fsg EMPH-fastPTCPfsg-nom
‘And Maymūna was given the medicine while she was fasting.’
qad faʿala, which is rarely found in the nominal clause pattern, is also not
found in the ʾinna la- pattern. Nevertheless, qad faʿala is compatible with the
emphasizing particle la- which precedes the modified form. In the same man-
ner as the emphasizing pattern wa-ʾinna-hū la-yafʿalu/wa-ʾinna-hū la-fāʿilun,
wa-la-qad faʿala also has an evaluative function: besides the plain mention of
an anterior event (which would have taken the form of wa-qad faʿala), la-qad
faʿala imparts the impression of the narrator regarding this event:
(68) la-qad ḥaddaṯa-nī ʿabdu llāhi bni ʿabbāsin ʾanna ʾādama nazala ḥīna
nazala bi-l-hindi wa-la-qad ḥaǧǧa min-hā ʾarbaʿīna ḥiǧǧatan ʿalā riǧlay-
hi (Taʾrīx 1, 124)
EMPH-MOD tellSC3msg-PRON1csg ʿAbdallāh_ibn_ʿAbbās-nom that ʾĀdam-acc de-
scendSC3msg when descendSC3msg in-al-Hind-gen and-EMPH-MOD perform-
ing_the_ḤaǧǧSC3msg from-PRON3fsg forty-obl Ḥaǧǧ-acc on two_feet-PRON3msg
‘ʿAbdallāh b. ʿAbbās told me that when Adam came down it was in India;
from there he had performed the pilgrimage to Mecca on foot forty times.’
4.3 Mutually-dependent clauses

Mutually-dependent constructions are exocentric, that is, neither their first nor
their second part may be said to function as the main constituent or nucleus to
which the other is subordinate. In Classical Arabic, mutual dependency is
marked as distinct from other types of interdependency by inverting the usual
order of the conjoined dependent clause, from subsequent position to initial.
The global meaning of a mutually-dependent construction is gathered from
the sum of both its parts, so that neither one can be omitted without giving up
much of the sense of the entire construction.
Conditional sentences are perhaps the best known example of mutually-
dependent constructions. The conditional meaning is only obtained by the
juxtaposition of a protasis and an apodosis. In contrast to other types of de-
pendent clauses, the conditional clause heads the entire construction. Moreo-

ver, it exhibits a different verbal paradigm than the one found in adverbial
clauses which follow their main clause.27
Apart from conditional clauses, there are other types of mutually-
dependent constructions. The present discussion focuses on those exhibiting
the predicative paradigm. These constructions can be divided into two interre-
lated kinds: (a) setting clauses and (b) presentative clauses. There is an im-
portant difference between the verbal complexes and the syndetic CCs dis-
cussed above and the setting-presentative constructions to be discussed below:
the first operate at the syntactic level of the complex-clause, and thus may be
found in any type of discourse, e.g., dialogues, narratives, expositions, etc.;
the latter operate at the text level and can only be found in narratives. They
are, in fact, marked patterns of narration.28
4.3.1 Setting clauses

Setting clauses are introduced by the conjunction (fa-)baynā/baynamā ‘while’.
They take the first position in the complex construction (like conditional
clauses), followed by a presentative clause (see below 4.3.2). baynā/baynamā-
clauses exhibit the pattern of the nominal clause (see above 4.2); in case the
predicate is verbal, it is realized as either yafʿalu or the participle, always in
the affirmative. Here, as well, we observe the opposition between the dynam-
ic-progressive-transitive yafʿalu and the static-intransitive participle:
(69) baynā ʾanā ʾamšī ʾiḏ samiʿtu ṣawtan min-a l-samāʾi (Ṣaḥīḥ 1, 6)
while PRON1csg walkPC1csg when hearSC1csg voice-acc from DEF-heaven-gen
‘As I was walking I suddenly heard a voice from heaven.’
(70) fa-baynamā humā wāqifāni bayna yaday-hi ʾiḏ saqaṭa ṭāʾirāni ʿalā l-sūri
(Riwāyāt 2, 180)
and-while PRON3mdu standPTCPmdu-nom between two_hands-obl-PRON3msg
when landSC3msg two_birds-nom on DEF-wall-gen
27 Prototypical conditional sentences exhibit a limited range of possibilities, the verbal

form in the protasis - faʿala or yafʿal - triggers off the verbal form in the apodosis -
faʿala or yafʿal, both forms assuming a hypothetical meaning. By contrast, modifying
adverbial clauses are free to follow a wide variety of clause patterns (Peled 1992,
140ff.), and the temporal or modal meaning of their verb, as is generally the case in de-
pendent clauses, is relative to the referential point established in the main clause.
28 A fuller discussion of mutually-dependent constructions and their function in the narra-
tive is found in Marmorstein (2014).

‘And while they were standing in front of him suddenly two birds landed
on the wall.’
The modified form qad faʿala, incompatible with the durative (unbounded)
meaning of baynā/baynamā, is not attested in this clause type. However, qad
faʿala may be incorporated into the setting in the form of a circumstantial
clause; notice that in (71) the subject of the circumstantial is fronted, so as to
match the order of the baynā/baynamā clause:
(71) fa-baynā ʾanā fī maǧlis-ī wa-l-xadamu qad ḥaffū bī wa-ǧawāriy-ya yata-
raddadna bayna yaday-ya ʾiḏā ʾanā bi-šayxin (Riwāyāt 1, 45)
and-while PRON1csg in session_room-PRON1csg and-DEF-servants-nom MOD sur-
roundSC3mpl with-PRON1csg and-maids-nom-PRON1csg frequentPC3fpl between
two_ hands-obl-PRON1csg when PRON1csg with-old_man-gen
‘And while I was in my living room, the servants had already surrounded
me and my maids were coming and going in front of me, suddenly there
was an old man with me.’
Setting clauses may also take the form of the ʾinna la- pattern. The same as
in baynā/baynamā-clauses, yafʿalu and the participle function as verbal predi-
cates. The distinction between setting clauses introduced by baynā/baynamā
and those introduced by ʾinna is not a syntactic one: both types of clauses
exhibit a mutually-dependent construction with the same verbal paradigm.
Rather, the distinction resides in the domain of expressivity: ʾinna, in this case
(and in a different manner than its emphasizing function in CCs, see above
4.2), signals the presence of an internally involved, ‘homodiegetic’ narrator
(Genette 1980, 245), telling the story from his own first-hand experience:
(72) fa-wallāhi ʾinn-ī la-ʾamšī naḥwa-hū [...] ʾiḏ xaraǧa naḥwa bābi banī
sahmin (Maġāzī 1, 31)
and-by_God INTRO-PRON1csg EMPH-walkPC1csg toward-PRON3msg […] when
go_outSC3msg toward gate-gen Banū Sahm-gen
‘By God, I was walking towards him [...] when suddenly he went out to-
wards the gate of Banū Sahm.’

(73) fa-wallāhi ʾinn-ī la-qāʿidun fī ʾahl-ī ʾiḏ naẓartu ʾilā ẓaʿīnatin (Sīra, 2,
948; Nebes 2001, 120)
and-by_God INTRO-PRON1csg EMPH-sitPTCPmsg-nom among family-gen-PRON1csg
when lookSC1csg at woman_in_camel_borne_sedan-gen
‘By God, I was sitting among my people when suddenly I noticed a wom-
an in a camel-borne sedan.’
4.3.2 Presentative clauses

Presentative clauses take the second position in mutually-dependent construc-
tions. There are two kinds of such presentatives: the first are headed by the
particle ʾiḏ, followed by the verbal form faʿala; in the second, the particle ʾiḏā
introduces the clause which exhibits the pattern [noun-phrase + the predica-
tive paradigm]. 29 The predicative paradigm comprises the triad of yafʿalu,
fāʿilun/mafʿūlun and qad faʿala. Interestingly, the participle in ʾiḏā-clauses –
rather than taking the accusative case (as in verbal complexes) – assumes as a
rule the nominative case:30
(74) wa-qāla qum bi-ʾiḏni llāhi fa-ʾiḏā huwa qāʾimun yanfuḍu l-turāba ʿan
raʾsi-hī wa-qad šāba (Taʾrīx 1, 187)
and-saySC3msg get_upIMP2msg with-permission-gen God-gen and-behold
PRON3msg get_upPTCPmsg-nom shakePC3msg DEF-dust-acc from head-gen-
PRON3msg and-MOD become_gray_hairedSC3msg
‘And he said: Rise, with God’s permission! And there he was standing,
shaking the earth from his head, already gray-haired.’
29 The nominal presentee after ʾiḏā may be definite or indefinite. It either takes the nomi-
native case or is realized as the genitive complement of the preposition bi- ‘with’. How-
ever, when followed by a predicative form, the nominal presentee is nearly always at-
tested in the nominative.
30 In my corpus, as well as in the major grammars of Classical Arabic, there are no exam-
ples of ʾiḏā-presentatives in which the participle is attested in the accusative case. On
the other hand, there are quite a few examples of presentatives introduced by hāḏā in di-
rect speech, in which the participle takes the accusative. This double manifestation of
the participle is explained by Bloch (1986) as a semanto-grammatical development of
presentatives in Arabic, from ‘amplified’ constructions, in which the participle (or some
other form) is adverbial (i.e. accusative), to ‘proclitic’ constructions, in which it is pre-
dicative (nominative). In a synchronic view, however, the fact that a fluctuation between
both manifestations exists is by itself instructive: it reflects the adverbial-yet-kernel sta-
tus of this ‘amplifying’ term, which, unlike other adverbials (e.g., temporal or locative),
forms part of the predicative core of the clause.

(75) fa-ǧiʾtu ʾilā ʾibrāhīma l-mawṣiliyyi fa-ʾiḏā l-bābu maftūḥun wa-l-dihlīzu

qad kunisa wa-l-bawwābu qāʿidun (Riwāyāt 1, 28)
and-comeSC1csg to ʾIbrāhīm_al-Mawṣiliyy-gen and-behold DEF-door-nom open-
PTCP.PASSmsg-nom and-DEF-hall-nom MOD sweepSC.PASS3msg and DEF-
door_keeper-nom sitPTCPmsg-nom
‘I came to ʾIbrāhīm al-Mawṣilī, and behold, the door was opened, the hall
was already swept and the door-keeper was sitting.’
(76) fa-fataḥa-hā la-hū fa-ʾiḏā fī-hā ṣūratu ʾādama wa-ḏurriyyati-hī kulli-him
fa-ʾiḏā kullu raǧulin maktūbun ʿinda-hū ʾaǧalu-hū wa-ʾiḏā ʾādamu qad
kutiba la-hū ʿumru ʾalfi sanatin (Taʾrīx 1, 156)
and-openSC3msg-PRON3fsg to-PRON3msg and-behold in-PRON3fsg picture-nom
ʾĀdam-gen and-progeny-gen-PRON3msg all-gen-PRON3mpl and-behold all-nom
man-gen writePTCP.PASSmsg-nom at-PRON3msg term-nom-PRON3msg and-behold
ʾĀdam-nom MOD writeSC.PASS3msg to-PRON3msg term-nom thousand-gen year-
gen
‘He opened it (i.e. His hand) for him, and behold, in it there was the pic-
ture of Adam and all his progeny, and there was the [life] term of each
man written down with him, and there was Adam, a term of thousand
years already written down for him.’
Presentative clauses introduced by ʾiḏ and ʾiḏā express something unexpected,
mufāǧaʾa ‘surprise’ in traditional terms, a sudden development or realization,
perceived or conceived by a certain character. Unlike ʾiḏ-clauses, which pre-
sent a further dynamic progression in the plot, ʾiḏā-clauses present an unfold-
ing scene, a static tableau. In both cases, the overall construction exhibits
what may be described as an aspectual asymmetry: in ʾiḏ-initiated presenta-
tives, a static situation (baynā/baynamā-clause) is interrupted by a dynamic
peak in the story; in ʾiḏā-initiated presentatives, a dynamic step forward in the
plot (faʿala) is concluded in a static situation. It is this aspectual asymmetry
that creates the dramatic moment of surprise in the narrative.
5. Conclusions
Circumstantial clauses are one of the most intriguing topics of Arabic syntax.
In this study I approached the problem of defining the circumstantial category
by giving up a clear-cut demarcation of verbal CCs and resorting to the seman-
to-syntactic domain of event integration and complex predications to which

they belong. I discussed the entire range of complex predications: from close-
ly integrated (and even fully grammaticalized) verbal complexes toward more
loosely integrated structures, such as syndetic circumstantial clauses and tex-
tual units consisting of mutually-dependent, setting and presentatives clauses.
Despite its heterogeneity, this variety of constructions was shown to be intrin-
sically related by the presence of the same set of predicative verbal forms:
yafʿalu, the participle and qad faʿala, marking an ongoing situation, a state
and an outcome, respectively. All three are co-temporal, either simultaneous
or coincidental with the time frame set in the matrix clause. Table (2) summa-
rizes the discussion on the domain of complex predications:
Table (2) - The domain of complex predications in Classical Arabic

SYNTACTIC LEVEL THE PREDICATIVE TRIAD OTHER VERBAL FORMS
faʿala
yafʿalu
(auxiliary, perception)
verbal complexes fāʿilan/mafʿūlan
sa-yafʿalu
qad faʿala
(perception)
wa-huwa yafʿalu
circumstantial clauses wa-huwa fāʿilun/mafʿūlun
wa-qad faʿala
baynā/baynamā huwa
yafʿalu
baynā/baynamā huwa
setting clauses
fāʿilun
*qad faʿala not adjacent to
baynā/baynamā
ʾiḏā huwa yafʿalu
presentative clauses ʾiḏā huwa fāʿilun/mafʿūlun
ʾiḏā huwa qad faʿala

The study of verbal CCs has also led to some interesting observations with
regard to the Classical Arabic verbal system in general. Perhaps the most
remarkable observation is that, in contrast to the general opinion, the simple
finite forms faʿala and yafʿalu do not appear to function as a binary pair: as a
predicative form, faʿala is quite marginal vis-à-vis the dominant role played
by yafʿalu and its counterparts, the participle and qad faʿala. Moreover, it was
shown that despite their close syntactic ‘resemblance’, yafʿalu and the partici-
ple are distinct at several semantic levels: 31 yafʿalu indicates a dynamic-
progressive aspect and is used in transitive, more informative verb-phrases,
whereas the participle indicates a static aspect and even permanent attributes
(including appointments). Lastly, a clear functional distinction between qad
faʿala and faʿala was noticed: the first is a resultative form which indicates a
coincidental situation, while the latter is an eventive form which is used to
indicate self-contained situations and thus is less frequently found in the pre-
dicative position.
References
Primary sources
Riwāyāt = Rannāt al-maṯāliṯ wal-maṯānī fī riwāyāt al-ʾaġānī. 2 vols. 1923 (vol I),
1946 (vol II). (Extracts from Kitāb al-ʾAġānī, ʾAbū al-Faraǧ al-ʾIṣbahānī). Edited
by ʾAnṭūn Ṣāliḥānī. Beirut: al-Maṭbaʿa al-Kāṯūlīkiyya.
Buxalāʾ=ʾAbū ʿUṯmān ʿAmr b. Baḥr b. Maḥbūb al-Ǧāḥiẓ, Al-Buxalāʾ. 1993. Edited by
ʿAbbās ʿAbd al-Sātir. Beirut: Dār wa-Maktabat al-Hilāl.
Kalīla wa-Dimna=ʿAbdallāh ʾIbn al-Muqaffaʿ, Kalīla wa-Dimna. 1926. Edited by
Louis Cheikho. Beirut: Maṭbaʿat al-ʾĀbāʾ al-Yasūʿiyyīn.
Maġāzī=ʾAbū ʿAbdallāh Muḥammad b. ʿUmar al-Wāqidī, Kitāb al-Maġāzī. vol I.
1966. Edited by Marsden Jones. London: Oxford University Press.
Ṣaḥīḥ=ʾAbū ʿAbdallāh Muḥammad b. ʾIsmaʿīl al-Buxārī, al-Ǧāmiʿ al-Ṣaḥīḥ. vol I.
1862. Edited by Ludolf Krehl. Leiden: Brill.
Sīra=ʾAbū Muḥammad ʿAbd al-Malik b. Hišām, Kitāb Sīrat Rasūl Allāh. 2 vols. 1858-
1860. Edited by Ferdinand Wüstenfeld. Göttingen: Dieterichsche Universitäts-
Buchhandlung.
31 The finite form yafʿalu is termed by the Arab grammarians al-muḍāriʿ l-ism al-fāʿil ‘the
[form] resembling the participle’. The muḍāraʿa ‘resemblance’ stems from the paradig-
matic interchangeability of both forms, see Sībawayhi (I, 2).

Taʾrīx=ʾAbū Muḥammad b. Ǧarīr al-Ṭabarī, Taʾrīx al-rusul wa-l-mulūk. 15 vols. 1879-

1901 (reprint 1964). Edited by Michael J. de Goeje et al. Leiden: Brill.
Secondary sources
Abboud, Peter. 1986. “The Ḥāl Construction and the Main Verb in the Sentence.” In
The Fergusonian Impact: In Honor of Charles A. Ferguson on the Occasion of His
65th Birthday, edited by Joshua A. Fishman et al., vol. 1, 191-196. Berlin: Mouton
de Gruyter.
Bloch, Ariel A. 1986. “Presentative Structures and Their Syntactic and Semantic De-
velopment.” In Studies in Arabic Syntax and Semantics, 54-101. Wiesbaden: Har-
rassowitz.
Carter, Michael G. 1972. “‘Twenty Dirhams’ in the Kitāb of Sībawayhi.” Bulletin of
the School of Oriental and African Studies 35: 485-496.
Fischer, Wolfdietrich. 2002. A Grammar of Classical Arabic. 3rd ed. Translated by
Jonathan Rodgers. New Haven/London: Yale University Press.
Genette, Gérard. 1980. Narrative Discourse. Translated by Jane E. Lewin. Ithaca:
Cornell University Press.
Givón, Talmy. 2001. Syntax: An Introduction. 2 vols. Amsterdam/Philadelphia: Ben-
jamins
Goldenberg, Gideon. 1987-1988. “The Contribution of Semitic Languages to Linguis-
tic Thinking.” Jaarbericht van het Vooraziatisch-Egyptisch Genootschap Ex Ori-
ente Lux 30: 107-115 [=Studies in Semitic Linguistics, 1-9. Jerusalem: The Magnes
Press].
———. 1991. “On Predicative Adjectives and Syriac Syntax.” Bibliotheca Orientalis
48: 716-726 [=Studies in Semitic Linguistics, 579-590. Jerusalem: The Magnes
Press].
———. 2000. “The Structure of Compound Tense-Forms.” Igeret 18: 15-17 [In He-
brew].
———. 2006. “On Grammatical Agreement and Verb-Initial Sentences”. In Loquentes
linguis: Studi linguistici e orientali in onore di Fabrizio A. Pennacchietti, edited by
Giorgio Borbone, Alessandro Mengozzi and Mauro Tosco, 333-339. Wiesbaden:
Harrasowitz.
Halliday, M.A.K. 1985. An Introduction to Functional Grammar. London: Edward
Arnold.
Haspelmath, Martin. 1995. “The Converb as a Cross-Linguistically Valid Category.”
In Converbs in Cross-Linguistic Perspective, edited by Martin Haspelmath and
Ekkehard König, 1-55. Berlin: Mouton de Gruyter.

Holes, Clive. 2004. Modern Arabic: Structures, Functions, and Varieties. 2nd ed.
Washington: Georgetown University Press.
Hopper, Paul J. and Thompson, Sandra A. 1980. “Transitivity in Grammar and Dis-
course.” Language 56/2: 251-299.
Ibn Hišām, Ǧamāl al-Dīn al-ʾAnṣārī. 2003. Šuḏur al-ḏahab. Edited by Muḥammad al-
Saʿdī Farhūd. Cairo/Beirut: Dār al-Kitāb al-Maṣrī wal-Lubnānī.
Ibn Sarrāǧ, ʾAbū Bakr Muḥammad b. Al-Sarī. 1985-1988. Kitāb al-ʾUṣūl fī al-naḥw. 3
vols. Edited by ʿAbd al-Ḥusayn al-Fatlī. Beirut: Muʾassasat al-Risāla.
Ibn Yaʿīš, Muwaffaq al-Dīn Yaʿīš b. ʿAlī. 2001. Šarḥ al-Mufaṣṣal lil-Zamaxšarī. 6
vols. Edited by Emīl Badīʿ Yaʿqūb. Beirut: Dār al-Kutub al-ʿilmiyya.
Isaksson, Bo, Kammensjö, Heléne and Persson, Maria. 2009. Circumstantial Qualifi-
ers in Semitic: The case of Arabic and Hebrew. Wiesbaden: Harrassowitz.
Jespersen, Otto. 1924. The Philosophy of Grammar. London: G. Allen and Unwin.
Kleiber, Georges. 1987. Du Côte de la référence verbale: Les phrases habituelles.
Bern/Frankfurt am Main: Peter Lang.
König, Ekkehard. 1995. “The Meaning of Converb Constructions.” In Converbs in
Cross-Linguistic Perspective, edited by Martin Haspelmath and Ekkehard König,
57-95. Berlin: Mouton de Gruyter.
Lehmann, Christian. 1988. “Towards a Typology of Clause Linkage.” In Clause Com-
bining in Grammar and Discourse, edited by John Haiman and Sandra A. Thomp-
son, 181-225. Amsterdam/Philadelphia: Benjamins.
Levin, Aryeh. 1979. “Sībawayhi’s View of the Syntactic Structure of kāna
waʾaxawātuhā.” Jerusalem Studies in Arabic and Islam 1:185-213.
Marmorstein, Michal. 2011. “A Review Article: Michael Waltisberg Satzkomplex und
Funktion: Syndese und Asyndese im Althocharabischen.” Jerusalem Studies in Ar-
abic and Islam 38: 361-390.
———. 2014. “Verbal Syntax and Textual Structure in Classical Arabic Prose.” In
Strategies of Clause Linking in Semitic Languages: Proceedings of the Interna-
tional Symposium on Clause Linking in Semitic Languages 5-7 August 2012 in Ki-
vik, Sweden, edited by B. Isaksson and M. Persson, 63-87. Abhandlungen für die
Kunde des Morgenlandes 93,Wiesbaden: Harrassowitz.
Matthiessen, Christian, and Thompson, Sandra A. 1988. “The Structure of Discourse
and ‘Subordination’.” In Clause Combining in Grammar and Discourse, edited by
John Haiman and Sandra A. Thompson, 275-329. Amsterdam/Philadelphia: Ben-
jamins.
Nebes, Norbert. 1982. Funktionsanalyse von kāna yafʿalu. Hildesheim: Georg Olms.

———. 1999. “Das Satzschema fa-huwa yafʿalu/fāʿilun/ Prädikativ für Vergangenheit

in frühklassischer arabischer Erzählliteratur.” In Tempus und Aspekt in den semiti-
schen Sprachen. Jenaer Kolloquium zur semitischen Sprachwissenschaft, edited by
Norbert Nebes, 77-100. Wiesbaden: Harrassowitz.
———. 2001. “Das Inzidenzschema im klassischen Arabischen: Ein Vorbericht.” In
Sachverhalt und Zeitbezug: Semitistische und alttestamentliche Studien, Adolf
Denz zum 65. Geburtstag, edited by Rüdiger Bartelmus and Norbert Nebes, 113-
128. Wiesbaden: Harrassowitz.
Peled, Yishai. 1992. Conditional Structures in Classical Arabic. Wiesbaden: Har-
rassowitz.
Premper, Waldfried. 2002. Die ‘Zustandssätze’ des Arabischen in typologischer Per-
spektive. Frankfurt am Main: Peter Lang.
Raible, Wolfgang. 1992. Junktion: Eine Dimension der Sprache und ihre Realisie-
rungsformen zwischen Aggregation und Integration. Heidelberg: Carl Winter.
Reckendorf, Hermann. 1895-1898. Die Syntaktischen Verhältnisse des Arabischen. 2
vols. Leiden: Brill.
———. 1906. “Zum Gebrauch des Partizips im Altarabischen.” In Orientalistische
Studien, Theodor Nöldeke zum siebzigsten Geburtstag gewidmet, edited by Carl
Bezold, vol. 1, 255-265. Gieszen: Alfred Töpelmann.
———. 1921. Arabische Syntax. Heidelberg: Carl Winter.
Sībawayhi, ʾAbū Bišr ʿAmr b. ʿUṯmān. 1881-1889. Al-Kitāb. 2 vols. Edited by Hart-
wig Derenbourg. Paris: L’imprimerie nationale.
Talmon, Rafael. 2003. Eighth-Century Iraqi Grammar: A Critical Exploration of Pre-
Ḫalīlian Arabic Linguistics. Winona Lake: Eisenbrauns.
Van Valin, Robert. D. 1984. “A Typology of Syntactic Relations in Clause Linkage.”
In Proceedings of the Tenth Annual Meeting of the Berkeley Linguistics Society,
542-559. Berkeley: BLS.
Waltisberg, Michael. 2009. Satzkomplex und Funktion: Syndese und Asyndese im
Althocharabischen. Wiesbaden: Harrassowitz.
Wright, William. 1896-1898. A Grammar of the Arabic Language. 2 vols. 3rd ed.
Cambridge: Cambridge University Press.

Clause Combining in Biblical Hebrew

The Verbal System of Biblical Hebrew.
A Clause Combining Approach
Bo Isaksson, Uppsala University
Introduction
The classical Hebrew verbal system is one of the greatest problems of Semitic
linguistics. It has been a topic of debates and academic discussions in more
than a millennium. The sad fact is that the grammatical understanding of the
Hebrew tradition texts had been forgotten already in the Middle Ages (McFall
1982).
Since then, the Biblical Hebrew verbal forms (tense- and/or aspect forms)
have remained a mystery, and theories about them have, since the Jewish
grammarian Jafeth ha-Levi (10th century), been based on inductive study of
translations into other languages. “In the explanation of the tenses no appeal is
made to a body of tradition, such as the Masoretes”, and this because the tra-
dition was no more available (McFall 1982, 16).
The theory that emerged among Jewish scholars was that the conjunction
we put before a finite verb had a ‘conversive’ function and thereby could
transform this verbal form into another tense form. The verb for past tense,
qatal, with a prefixed we often acquired a futural or modal sense, which re-
minded of the meanings of the other finite tense form (yiqtol). By analogy, the
conjunction wa (allomorph of we) was considered ‘converting’ yiqtol into a
narrative past tense (wa-yiqtol).1
This “conversive theory” made overwhelming impact on the subsequent
scholarly discussion on the Hebrew verbal system and still dominates the text
books on Biblical Hebrew. It is a system of four different finite tense forms
1 In the present article we refers to the Hebrew morpheme wə with allomorphs wə, ū, wī,
wå, wɛ, wa, and wā. In a similar way, wa refers to the morpheme wa + gemination with
allomorphs way, wat, wan, wå̄ , and wa.

170 Bo Isaksson
(or aspect forms), of which two lack a prefixed we and two have a prefixed
conjunction. Though the system has four tenses, those tenses are distributed
semantically into two main classes: 1) qatal and wa-yiqtol means past tense
and anterior/perfective aspect, and 2) yiqtol and we-qatal means present time,
future, modal meaning and imperfective aspect.
This is the starting point for nearly all scholarly studies even today. On this
four-part verbal system, various theories of tense-aspect-mood are applied to
the Hebrew texts in a never-ending stream of dissertations and monographs,
not to talk about papers and articles. And still, as with the Jewish grammari-
ans, in focus of the discussion are the verbal forms as such and their relations;
clauses are usually not considered.2
It is conspicuous that a comparative Semitic perspective has had so little
impact on the question of the four-part Hebrew verbal system. No other Se-
mitic language exhibits this strange constellation of four verb forms semanti-
cally ordered in pairs, and in no other Semitic language are there a ‘conver-
sive’ or ‘consecutive’ wa that can transform verb forms from one conjugation
to another.
Seen in the perspective of the close relatives of Hebrew in Central Semitic,
and taking into account the second millennium Amarna Canaanite (Rainey
1996), we would rather expect Biblical Hebrew to behave as one of several
first millennium daughter languages of early Canaanite. We would expect that
Biblical Hebrew had three basic finite verb forms (not four), two with pre-
fixed inflection (reflexes of yaqtul and yaqtulu);3 and one with suffixed inflec-
tion (qatal). Utterly few scholars have dared to maintain this concerning Bib-
lical Hebrew (among the few are Tropper 1988; Van de Sandhe 2008).
We have reminiscences of the Canaanite system in Ugaritic and in Classi-
cal Arabic. It is significant that both Ugaritic and Classical Arabic have re-
tained short final vowels, both in the nominal and the verbal inflections. This
is not the case with the Northwest Semitic languages attested in the first mil-
lennium, B.C. Huehnergard writes:
As one studies the tense-mood-aspect systems of Arabic, Ugaritic,

Amarna Canaanite, and early Hebrew (and perhaps also early Aramaic,
in view of the Tel Dan inscription with its preterite yqtl forms), one is
struck by their overall formal and semantic similarity: besides the im-
2 Two exception are Andersen (1974) and Heller (2004), cf. Isaksson (2009, 22).
3 Bibliography in Bloch (2007, 142 note 3).

The Verbal System of Biblical Hebrew. A Clause Combining Approach 171
perfective yaqtulu, we may mention the volitive or subjunctive yaqtula

and the emphatic energic forms. Although the details remain unclear, it
seems increasingly likely that the entire tense-mood-aspect system of
these languages, both formally and semantically, can be reconstructed
to a common ancestor. (Huehnergard 2005, 165).4
The most simple and straightforward and reasonable expectation of the

Biblical Hebrew verbal system is that it should behave like an ordinary de-
scendant of early Canaanite (Isaksson 2011, 173 note 1). This is the position
taken in the present study. Biblical Hebrew is Central Semitic and it belongs
to the Canaanite subgroup. Biblical Hebrew possessed the grams short yiqtol,
long yiqtol, and the qatal, the imperative (IMP) and the cohortative.
Hypothesis: A word order revolution in Proto-Hebrew

Somewhere in the period between the Amarna age (14th century), and before
the Old testament texts were created, there occurred in Proto-Hebrew and its
Canaanite sister languages no less than a grammatical revolution: a phonolog-
ical change resulted in the loss of all short final vowels, and this in turn occa-
sioned a subsequent morphological merger of the short and long prefix grams.
The types yaqtul and yaqtulu both became yaqtul. This change occurred in all
adjacent Northwest Semitic languages, Aramaic, Phoenician, Hebrew; but not,
as far as we can understand, in the earlier Ugaritic (extinguished 1180 B.C.).
We can observe a similar process of change in the modern Arabic vernaculars,
which, in contrast to Classical Arabic, have lost short final vowels.5
4 The cohortative is discussed only occasionally in the present article. The cohortative
ending was added only to first person forms of the prefix conjugation and the ending
was facultative (Notarius 2013, 305), as it also seems to have been when added to the
imperative (many first person prefix conjugation forms are clearly volitive without hav-
ing a cohortative ending). When appended, the ending results in an explicitly signaled
volitive gram. The cohortative ending seems to have been facultative also in Amarna
Canaanite (Tropper and Vita 2010, 78). The cohortative was not subject to word order
constriction in Biblical Hebrew (as were the VprefS and VprefL grams). I am hesitant to
regard the ‘energic’ (which could be labeled ‘Vpref-N’) a separate gram. The functional
meaning of the seemingly facultative n suffixes, except for some sense of emphasis, is
still an open question (J-M § 61f.).
5 The whole Neo-Arabic language type is characterized by the loss of the ancient Central
Semitic marking of the moods in the prefix conjugation. Corresponding functions are
upheld with other syntactical means “zum Ausdruck temporaler, aspektueller oder
modaler Sonderfunktionen” (Fischer and Jastrow 1980, 41-43).

172 Bo Isaksson
Hebrew academic research has never seriously discussed the Biblical He-
brew verbal syntax in the light of the syntactical strategies that must have
developed in Proto-Hebrew in order to handle the new situation with a partial
morphological coalescence of two conjugations. It is the more remarkable,
since the morphological difference between short form and long form – when
this difference of phonetic reasons remained discrete (as in the Hifil and in
many forms of the weak verb) – has been retained in Biblical Hebrew. Bibli-
cal Hebrew still upholds the distinction between a jussive yāqém ‘may he
raise up’ and an indicative yāqīm ‘he will raise up’. It seems obvious that the
linguistic instinct among the speakers of Biblical Hebrew retained a sensibil-
ity for the distinction between the two prefix conjugations, also after the par-
tial merging of the forms.
The individual Northwest Semitic languages all developed their own ways
to handle this merger. Some discarded the short yaqtul type completely (thus
Aramaic except in the most ancient inscriptions). The strategy that the Proto-
Hebrew speakers developed, to retain the distinction between short form and
long form, was a limitation of word order. While word order in Amarna Ca-
naanite was relatively free for both the short and the long prefix conjugations,
Biblical Hebrew (even in the archaic poetry) has relegated the affirmative
short yiqtol to first position in the clause, whereas the long yiqtol is used in
other positions (such as after a negation, or after a topicalized first element).6
The word order constraints were part of a strategy to compensate for the
partial formal mergers of the grams, whereby the short prefix form (the old
Semitic yaqtul) “was assigned the clause-initial position” (Gzella 2011, 442),
while the long prefix verb (the old Central Semitic yaqtulu) was placed in
non-initial position. “[S]o word-order constraints to some extent restore the
functional differentiation” (Gzella 2012, 101).
Few Hebrew scholars have understood this restriction of word order in the
light of a morphological merger (one of the few is Gzella 2011, 442; 2012,
6 Joosten (2012, 12) recognizes that “they may still be distinguished, at least in classical
Hebrew prose, by paying attention to the syntax. The jussive usually takes the first posi-
tion, while YIQTOL tends to occupy non-initial position in the clause” (also Joosten
1999, 15 note 3; Niccacci 1987). When Joosten talks about YIQTOL he means the reflex
of Central Semitic yaqtulu. This usage of the term ‘yiqtol’ is misleading, since yiqtol in
Biblical Hebrew may also code indicative narrative short prefix conjugation (in poetry,
especially archaic poetry, Bloch 2009), and the jussive, both being reflexes of Central
Semitic yaqtul.

101). No one has tried to explain how the three basic conjugations of the He-
brew verbal system work in the light of this insight.7 Neither has an explana-
tion been suggested for how clauses with these basic grams combine to com-
municate meaning.
‘Clause combining’ is a concept taken from Clause combining in grammar
and discourse (Haiman and Thompson, eds., 1988). With this approach the
scholar examines how different kinds of clauses combine in a specific lan-
guage.8 The pattern of clause combining used in a text reflects the rhetorical
intentions of the author or narrator (Matthiessen and Thompson 1988, 290,
299, 275). I assume that this is true also for Biblical Hebrew. In analysing
texts, the ‘proof’ of my assumption will be that the texts communicate mean-
ing in a more successful way than before; that the textual structure will be-
come more understandable, more transparent.
The details will be discussed below.
The clause in Biblical Hebrew

A clause is a piece of linguistic code, a syntagm (spoken or written) that con-
tains one predication (Lehmann 1988, 182; Haspelmath 1995, 11; Isaksson
2009, 7-23).9 The predication could be just a noun phrase (NP) as in (1).
(1) Pattern: Spron-NP;

zɛ šaʿar haš-šå̄ mayim
‘this is the gate of heaven’ (Gen. 28:17).
7 It is a pleasant duty to express thanks to my Autumn 2013 and Spring 2014 master and
PhD student classes on Old Aramaic texts and Biblical Hebrew prose. Many of the top-
ics discussed in the present article were opened up or refined in the discussions. I am es-
pecially indebted to PhD student Ambjörn Sjörs (forthcoming) who insisted on the im-
portance of word order in Biblical Hebrew and pointed out negative Vsuff clauses as
possible candidates for a renewed coding of negative clauses in the narrative storyline.
8 For an introduction to the notion of ‘clause combining’ in Arabic and Hebrew, see
Isaksson (2009, 1-35). In the chapter “The structure of discourse and ‘subordination’”,
Matthiessen and Thompson discuss how clause combining can be seen as “a grammati-
calization of a very general property of the hierarchical structure of the discourse itself”
(1988).
9 There is no need of the English term ‘sentence’ in biblical Hebrew. It is usually under-
stood that a sentence includes several clauses, but the concept of ‘sentence’ has never
been defined in a Semitic setting and it is unclear which clauses to include and which to
exclude. In this article I prefer to use the more informative concept ‘clause combining’
in a discussion of relations between clauses. ‘Sentence’ is a concept that hides more than
it clarifies.

174 Bo Isaksson
The predicate of a verbless clause might be practically any non-verbal element

of a clause, also a prepositional phrase (PrP) as in (2), where the utterance
receives an implied modal tone.
(2) Pattern: PrP-Snoun, VOC;

ʿå̄ lay qillå̄ ṯəḵå̄ bənī
‘Your curse be on me, my son’ (Gen. 27:13).
In the present article, clauses without a verbal element are called ‛noun claus-
es’ (NCl).
It is not necessary that a clause contains a subject. If it does not, the subject
is understood, as in Moses’ answer to YHWH at the burning bush, (3).
(3) Pattern: NP (= NCl);

maṭṭɛ
‘(It is) a rod’ (Exod. 4:2).
This quotation constitutes a complete utterance with a predicate consisting of

only one word. It is a main clause.10
If not a noun clause, the clause contains a predication encoded by a finite
or infinite verb. A clause with a finite verb is probably the most common type
of clause in the Hebrew Bible, as in (4).
(4) Pattern: wa-VprefS-ADV;

way-yå̄ måṯ šå̄ m
‘He died there’ (Judg. 1:7).
Example (4) shows a clause consisting of a conjunction, a finite verb and an

adverb.
The verbal predicate in a clause may of course also be infinite. It is unusual
that a verbal noun (VN) constitutes the predicate in a main clause.11 The active
participle (PA), however, is rather often predicate in main clauses, and then
usually in direct speech, to indicate the immediate future or an ongoing action
(progressive aspect). An example of immediate future is shown in (5).
10 This is not to say that all utterances consist of clauses (that is, ‘predications’). Some
utterances are just interjections (ʾăhå̄ h ‘Ah!’) or vocatives (ham-mɛlɛḵ ‘O king!’), (J-M
§§ 105, 137g).
11 The prime examples of VN are the infinitive construct and the infinitive absolute (J-M §
49a), but other verbal nouns are sometimes used in similar functions.

(5) Pattern: Spron-PA;

ʾå̄ nōḵī mēṯ
‘I am about to die’ (Gen. 50:24).
In the example, the subject is explicitly stated.

A progressive aspect (with past reference) is found in example (6) in a
main line clause.
(6) Pattern: we-hinnē-PA-PrP;

wə-hinnē ʿōmēḏ ʿal ha-yʾōr
‘He was standing by the Nile’ (Gen. 41:1).
In (6) the subject is understood.

It is justified to claim that the main characteristic of a clause is its predi-
cate.12
The concept of ‘gram’ in Biblical Hebrew

It is taken for granted that Biblical Hebrew was a normal human language
which once possessed native speakers, and it is also assumed that Hebrew
under its life span as a native spoken medium underwent changing processes
according to the same rules as other languages. In particular, it is supposed
that the verbal grammatical morphemes (‘grams’) in Biblical Hebrew had a
history involving new formation, reanalysis, and interaction with the other
grams. 13 Linguists who have investigated the grammaticalization of verbal
grams in a cross-linguistic perspective have found that a gram develops ac-
cording to the patterns of a few characteristic trajectories. This fact enables
12 In the present article clauses are given names by their predicates. A clause with a short
prefix verb predicate is called ‘short prefix verb clause’ or simply ‘VprefS clause’, a
clause with a suffix verb predicate is called ‘suffix verb clause’ (‘Vsuff clause’), etc.
Clauses with infinite verbal predicates are called ‘verbal noun clauses’ (VN) and ‘parti-
ciple clauses’ (‘PA’ or ‘PP’) respectively. Clauses with non-verbal predicates such as
adjectives (ADJ), noun phrases (NP), prepositional phrases (PrP), adverbs (ADV), etc.,
are all subsumed under the common designation ‘noun clause’ (‘NCl’).
13 The structuralist approach is limited to just that, not more, not less. The grams have
meanings of their own, and those meanings interact with and influence other grams in
the verbal system. Since grams are regarded the basic units in a verbal system, TAM
categories are not. “The B&D approach differs from most other treatments of tense and
aspect in that the basic units are not ‘the category of tense’ and ‘the category of aspect’
but rather what we call grams” (Dahl 2000, 7; “B&D” here stands for Joan Bybee and
Östen Dahl).

176 Bo Isaksson
the Hebrew scholar to discuss the meanings of existing verbal grams in Bibli-
cal Hebrew, and with reasonable accuracy identify some plausible trajectories
for each of the finite verbal grams (Isaksson 2011, 173-176).14
In the present article neither ‘aspect’ nor ‘tense’ 15 – not even ‘relative
tense’16 – are taken as forming the basic constituents of the verbal system in
Biblical Hebrew, but instead grams; “tense-aspect grams can crosslinguistical-
ly be classified into a relatively small set of types”. Gram types should be
thought of “as relatively stable points along the paths of development that
grams take in the course of grammaticalization processes” (both quotations
Dahl 2000, 7).
The general meanings of grams

Grams have meanings that may be aspectual or temporal or modal or a com-
bination of such meanings. A gram has usually acquired several meanings
during its development. Early ‘archaic’ meanings are often synchronically
attested side-by-side with later meanings. Because of this, grams cannot be
said to possess a ‘basic’ meaning from which the others in some way can be
derived.
Working with grams makes the long debated question whether SBH is an
‘aspect language’ or ‘tense language’ a less crucial issue. However, the mean-
ings displayed in the grammaticalization paths discussed below mainly con-
cern aspectual concepts. In the present article the aspect meanings are defined
according to the ground-breaking study The Evolution of Grammar, by Bybee,
Perkins and Pagliuca (1994):17
− “Resultatives signal that a state exists as a result of a past action. The
resultative is often similar to the passive in that it usually makes the pa-
tient the subject of the clause but differs in that a resultative may apply to
an intransitive verb, as He is gone, without a change of subject” (Bybee et
al. 1994, 54).
− “Anteriors (or ‘perfects,’ as they are often called) usually develop from
resultatives. Anteriors differ from completives in being relational: an an-
terior signals that the situation occurs prior to reference time and is rele-
14 This approach to the Hebrew verbal system is taken also by T. D. Andersen (2000),
Cook (2006; 2012), and Andrason (2010; 2011a; 2011b; 2012).
15 For an up-to-date survey of research, see Cook (2012, chapter 1).
16 Advocated by some Hebrew scholars, such as Hatav (1997).
17 Boldface within quotation is my emphasis.

vant to the situation at reference time. Anteriors are typically translated

with the English perfect and often accompanied by the relational adverbs
‘already’ and ‘just’” (Bybee et al. 1994, 54). Unlike resultatives, anteriors
can be used with numerical adverbs like ‘twice’. Anterior grams general-
ly tend to develop additional perfective meanings.
− “Perfectives signal that the situation is viewed as bounded temporally.
Perfective is the aspect used for narrating sequences of distinctive events
in which the situation is reported for its own sake, independent of its rele-
vance to other situations (Hopper 1982). It is thus often used to refer to
situations in the past” (Bybee et al. 1994, 54-55).
− Immediate future is a future meaning that is typically expressed by a
perfective gram (‘bounded future’; Bybee et al. 1994, 83; Isaksson 2009,
134).
− An imperfective gram “views the situation not as a bounded whole, but
rather from within, with explicit reference to its internal structure”; “an
imperfective situation may be one viewed as in progress at a particular
reference point, either in the past or present, or one viewed as characteris-
tic of a period of time that includes the reference point, that is, a habitual
situation. Imperfective forms are typically used in discourse for setting up
background situations, in contrast with perfective forms, which are used
for narrating sequences of events”. “Imperfectives may be applicable to
either past, present, or future time” (Bybee et al. 1994, 125-126).
− “Progressive views an action as ongoing at reference time” (Bybee et al.
1994, 126).
− Gnomic presents “apply to generic subjects and basically hold for all
time” (Bybee et al. 1994, 126).
− Zero present: “some languages have two grams that we have classified
as presents. In all these cases, it is clear that one is an older, more gram-
maticized (or zero) gram and the other is a younger developing gram, and
in three out of four cases, it appears that this younger gram has developed
from a progressive” (Bybee et al. 1994, 144).

178 Bo Isaksson
The grams in Biblical Hebrew18

The short prefix gram (VprefS)
The most archaic finite gram in Semitic except the imperative is the short
prefix verb (Arab. yaqtul, Akk. iprus, SBH short yiqtol, Rainey 2003a, 400).
We can trace its formation back in Afro-asiatic times, but it is presently not
possible to state how it was formed or according to which trajectory it was
grammaticized. We do not know whether it, in its first beginning, followed a
resultative gram trajectory or an imperfective gram trajectory. Or, was another
type of verbal formation.
We do know, however, the meanings displayed by the VprefS in the attest-
ed Semitic languages. The most frequent meanings, also in Biblical Hebrew,
were the jussive and the indicative perfective (Bloch 2009, 38; Rainey 1986,
5). In the earliest stages of Semitic, VprefS was the prime narrative form in
the storyline.19 But VprefS could be used also for generally valid facts, a gen-
eral present (Tropper 1998, 161, 172f; Andrason 2011a, 48). Whatever the
VprefS was in early Semitic, it was not a ‘preterite’, it was not a past tense.
The enigma of the VprefS gram in Semitic consists of the seemingly con-
tradictory meanings by which the gram is used (Joosten 2012, 15). There
seems to be no common grammaticization path by which a narrative ‘tense’
may develop into a ‘jussive’, nor one by which a ‘jussive’ develops into a
narrative gram (Isaksson 2009, 126).
The most plausible explanation is that VprefS was inherited in Semitic as a
so-called ‘zero-gram’ (Huehnergard 1988, 22; Stempel 2012) or ‘zero present’
(Bybee et al. 1994, 144). In many respects it behaves in SBH like the ‘old
present’ described by Haspelmath (1998; cf. Isaksson 2009, 126) or the so-
called ‘injunctive’ in Indo-European, “the zero or unmarked tense and mood”
(Kiparsky 1968, 34). A zero gram can be used as a general gnomic present. A
zero gram may also develop ‘jussive’ connotations. And it can usually be used
as a storyline gram in narratives.
18 ‘Biblical Hebrew’ is here taken as the variety that is commonly called Standard Biblical
Hebrew (SBH) with the exclusion of Late Biblical Hebrew. Occasionally, also archaic
Biblical Hebrew will be considered. The linguistic difference between SBH poetry and
archaic poetry has often been overestimated. The most prominent feature adduced for
the archaic poetry – the narrative perfective (wa-)VprefS clauses – are found in ordinary
pre-exilic psalms as well (Bloch 2009, 37).
19 The term ‘storyline’ refers to the main line in a specific discourse type: narrative prose.

More common than marking narrative contexts, however, is not mark-

ing them – quite a considerable number of languages use unmarked
verb forms in narrative contexts (Dahl 1985, 113).
The long prefix gram (VprefL)

The long prefix gram, as found in Central Semitic, is probably an innovation
in Semitic. The Central Semitic formation of the VprefL, yaqtulu, starts with
the short prefix gram and adds some suffixes to mark the imperfective aspect
(Huehnergard 2005, 164-165; Rainey 1986, 5; Meyer 1966 § 63). This point
of departure is significant also for an understanding of the nature of the short
prefix gram. In such a process of renewal of the verbal system, a fresh encod-
ing of the imperfective aspect was needed. The starting point for such a re-
newal cannot have been something completely different. It could not have
been a decidedly ‘jussive’ gram, nor a dedicated ‘preterite’ formation. What-
ever it was, its meaning was used to encode the imperfective aspect with the
help of the aspect markers -u and -na: 3ms yaqtul-u, 3mp yaqtulū-na. In the
imperfective trajectory, this long prefix gram acquired the meanings of actual
present, habitual action, future, and a command (Tropper 1998, 178-181).
The suffix verb gram (Vsuff)

The suffix verb was inherited in Semitic as a stative/resultative formation
(Rainey 2003b). The West-Semitic innovation is a development according to
the resultative trajectory and has acquired all the expected meanings of such a
gram: stative/resultative with sometimes a general present, but also anterior,
perfective, immediate future, and optative (Tropper 1998, 182-184). The opta-
tive meaning of the Vsuff is archaic and close to the use of the passive partici-
ple in SBH noun clauses, as in (7).20
(7) Pattern: PP-Spron-PrP;

bå̄ rūḵ ʾaḇrām lə-ʾēl ʿɛlyōn
‘Blessed be Abram by God Most High’ (Gen. 14:19; ESV).
The same construction, but with a divine subject, may be perceived as an

eternal truth without optative connotations, which is seen in (8).
20 The examples of NCl clauses with PP predicate are adduced to illustrate how the proto-
typical archaic meanings of the Vsuff gram occurred. For a treatment of Vsuff clauses,
see section C.

180 Bo Isaksson
(8) Pattern: PP-Snoun;

bå̄ rūḵ ʾēl ʿɛlyōn
‘Worthy of praise is the Most High God’ (Gen. 14:20; NET).21
Prose and poetry in Biblical Hebrew

It is reasonable to suppose that most poetic texts in the Hebrew Bible were
composed by poets who were contemporary with – and shared the same cul-
tural community as – the original composers of the prose texts, otherwise they
would not have been understood (Joosten 2012, 414). There are poetic pieces
which are more archaic and lack some of the features that are found in SBH
prose, and such differences should be treated accordingly as representing a
somewhat earlier state of the language. However, most poetic texts must be
assumed to belong to the same state of Hebrew as the prose texts.
This conclusion has far-reaching consequences. A trustworthy linguistic
approach to the prose texts in SBH should be expected to stand the test of
poetic texts as well. The poetic texts were produced in the same language as
the prose texts (Isaksson 2009, 38 with note 7). In the present article poetic
text samples will sometimes be adduced as SBH usage, side-by-side with
prose examples. It is assumed that “verbal meanings in poetry are basically
the same as in prose” (Joosten 2012, 413).22
A cross-linguistic typology of semantic clausal relations

It is a great advantage, while working with a dead language, to refer to a
cross-linguistic typology of the semantics of clausal relations such as that in
Dixon (2009). Such a typology may suggest what kind of semantic relations
between clauses should be expected, also in a Biblical Hebrew text.23
21 Some optative meanings of Vsuff may be connected with the immediate future meaning,
since ‘imminent activity’ (be about to) easily receives a modal nuance (Dixon 2012, 26),
cf. Isaksson (2009, 131-132).
22 This also implies that a verbal form like VprefS that may have general present or ‘zero’
meaning in poetry, cannot, just on the basis of prose usage, be deemed to be inherently
‘preterite’ (against the assumption of Joosten 2012, 413).
23 In the present article ‘main clause’ and ‘non-main clause’ refer to the syntactic marking
of a clausal relation, while ‘focal clause’ (FC) and ‘supporting clause’ (SC) pertain to
the semantic relation. It is often the case that the syntactically marked ‘non-main clause’
must be analysed as ‘focal clause’, Dixon (2009, 1-5).

The following semantic clausal relations are common in Hebrew:24

− Temporal succession: second clause in temporal sequence. Indicates that
the actions or states happened in that iconic order (Dixon 2009, 9).
− Relative time: “The supporting clause serves to place the event or state
of the Focal clause in temporal perspective” (in English: before, when,
while, etc.), (Dixon 2009, 10).25
− Comment/background: Although Dixon’s relative time clause accounts
for many of the typical Central Semitic attendant circumstance clauses, it
does not seem to include ‘comment clauses’ and more general ‘back-
ground’ clauses, which are frequent in SBH narrative prose.
− Consequence relations are basically of three types: 1) clauses showing
cause or reason; 2) clauses showing result; 3) clauses showing purpose.
To these basic types Dixon adds clauses showing ‘cause with no neces-
sary result’ and clauses showing ‘natural result’ (Dixon 2009, 17-23, 44-
45), which are pertinent in some SBH contexts.
− Additions are exceptionally common in SBH. In such a clause linking
there are “two pieces of information (one in each clause) which are not in
a Temporal relation, or in a relation of Condition, Consequence, Possible
consequence, Alternatives, or Manner. We refer to this as Addition”
(Dixon 2009, 26). Dixon recognizes four basic types of additions, and all
of them can be attested in SBH: 1) Unordered addition: “two distinct
events which are semantically or pragmatically related but for which no
temporal sequence is assumed”. Dixon’s example is: Mary peeled the po-
tatoes and John shelled the peas. “The temporal information is not con-
sidered relevant and is not stated”. 2) Same-event addition: “two clauses
describe different aspects of a single event” (Dixon 2009, 27). The sup-
porting clause here “describes an aspect of the event which follows from
that described by FC”, and the focal clause “details main aspect of event”.
One of Dixon’s examples is: Mary came first in her race [and won the
24 The term ‘clause combining’ (taken from Haiman and Thompson 1988) is here used as
synonymous with ‘clause linking’ in Dixon (2009). Both terms presuppose some kind of
syntactic marking of the linking. If marking is only semantic (‘inference’) it is more
fruitful to utilize Dixon’s ‘focal clause’ in relation to ‘supporting clause’.
25 ‘Focal clause’ (FC) is a clause that “refers to the central activity or state of the biclausal
linking”; the opposite concept is ‘supporting clause’ (SC) “which may set out the tem-
poral milieu for the Focal clause, or specify a condition or presupposition for it or a pre-
liminary statement of it” (Dixon 2009, 3).

182 Bo Isaksson
prize]SC. 3) Elaboration: “the second clause echoes the first, adding addi-
tional information about the event or state described” by the focal clause.
An English example given by Dixon is: [John telephoned]SC he invited us
to dinner. 4) Contrast: “In this kind of linking, the information conveyed
by the Focal clause contrasts with that provided in the Supporting clause,
and may be surprising in view of it” (Dixon 2009, 28). One of Dixon’s
examples is, [John is rich]SC but he is not happy.
Dixon’s definition of ‘addition’ is basically negative (what it is not). A posi-

tive definition of ‘addition’ will be used here: ‘accompanying action’, which
involves clauses that in some respects describe events occurring in a certain
(general or logical or temporal) connection with the first clause. From this
definition follows that a clause describing an accompanying action always
comes after another clause (which is termed ‘pre-clause’ in the present arti-
cle). Accompanying action clauses are not necessarily simultaneous with the
action described by the pre-clause, nor are accompanying actions necessarily
consequences of the action in the pre-clause. As described by Dixon (2009,
28), addition clauses may – by inference from the context – describe ‘tem-
poral linking’ and ‘result’.26
The concept of ‘main line’ and non-main clause linking

It is assumed that practically all texts in the Hebrew Bible contain syntactical-
ly marked main lines (Niccacci 2014; Isaksson 2013).27 Significant features in
the syntactic coding of main lines are, 1) the conjunction wa or we, 2) absence
of conjunction (asyndesis), 3) a specific verbal gram (or no verbal gram in a
NCl), and 4) a specific word order. The coding of the main line may be stable
as in narrative prose (wa-VprefS) or instructional discourse texts (we-Vsuff
clauses), but it may also vary, and usually does so in direct speech and poetry.
A poem may begin with three jubilating Ø-Vsuff clauses, as in (9).
(9) Pattern: Ø-Vsuff+Ø-Vsuff+Ø-Vsuff;
26 Dixon illustrates by [Mary left John]SC and he went into a monastery, an example of
inference of both ‘temporal succession’ and ‘consequence: result’ (Dixon 2009, 9, 17,
28)
27 The concept of ‘main line’ in my terminology replaces that of ‘coordination’. As I
abstain from using the cumbersome term ‘subordination’ it is probably wise not to use
‘coordination’ (Isaksson, forthcoming c). The term ‘main line’ has the advantage of
stressing the textual nature of our primary linguistic source.

ʿå̄ laṣ libbī be-YHWH rå̄ må̄ qarnī bə-YHWH rå̄ ḥaḇ pī ʿal ʾōyəḇay
‘My heart rejoices in the LORD; in the LORD my horn is lifted high. My
mouth boasts over my enemies’ (1 Sam. 2:1; NIV).
Then the same poem may go on with a new main line of NCl clauses, (10):
(10) Pattern: Ø-NCl+kī-NCl+Ø-NCl;

ʾēn qå̄ ḏōš kə-YHWH kī-ʾēn biltɛḵå̄ , wə-ʾēn ṣūr k-ēlōhēnū
‘There is no one holy like the LORD; there is no one besides you; there is
no Rock like our God.’ (1 Sam. 2:2 NIV).28
The same poem continues by changing the main line once again, and now to a
negated jussive VprefS clause (11):
(11) Pattern: jussive ʾal-VprefS;

ʾal-tarbū ṯəḏabbərū gəḇōhå̄ ḡəḇōhå̄ ...
‘Do not keep talking so proudly ...’ (1 Sam. 2:3; NIV).
Besides the discourse types in which the main line coding is more or less a
convention (such as in narrative prose and instructional discourse), it is not
possible to define syntactically what constitutes a main line. “There are no
devices in the grammar specific to coordination” (Cormack and Smith 2005,
395). This must be left to the pragmatics of the text itself. In archaic poetry
the narrative short prefix verb is unrestrained by the (oral) literary style of a
narrative storyline (in which the clause must begin with the conjunction wa)
and quite often exhibits a main line of indicative perfective Ø-VprefS clauses
with past time reference, as in (12).
(12) Pattern: [VN]+[VN]+Ø-VprefS!

[bə-hanḥēl ʿɛlyōn gōyīm] [bə-hap̄ rīḏō bənē ʾå̄ ḏå̄ m] yaṣṣeḇ gəḇūlōṯ
ʿammīm lə-mispar bənē yiśrå̄ ʾēl
‘[When the Most High gave the nations their inheritance], [when he divid-
ed all mankind], he set up boundaries for the peoples according to the
number of the sons of Israel’ (Deut. 32:8).
28 The kī-clause is here taken with NIV as a main clause, the kī being interpreted as an
emphatic adverb, ‘surely’. The kī-clause could also be taken as a reason clause with kī
as a subordinating conjunction (in which case it should not be treated as part of the main
line).

184 Bo Isaksson
In direct speech the coding of the main line likewise shifts freely, as the
shift to a VprefL main line shows in (13).
(13) Pattern: Spron-Vsuff; Spron-VprefL!;

mī śå̄ m pɛ lå̄ -ʾå̄ ḏå̄ m; ʾō mī yå̄ śūm ʾillēm ʾō ḥērēš ʾō p̄ iqqēaḥ ʾō ʿiwwēr
‘Who has made man’s mouth? Who makes him mute, or deaf, or seeing, or
blind?’ (Exod. 4:11; ESV)
The main line in (13) begins with a Vsuff clause with anterior meaning. And
then it proceeds with a main line VprefL clause with continuous or habitual
meaning. None of the clauses is dependent on the other. They constitute inde-
pendent rhetorical questions.
Within a main line, clauses are linked with ‘equal status’ (Halliday 2004,
374). Since “the grammar is only capable of providing asymmetric structures”
a shift of status is encoded by a ‘digression’ from a main line in which
“[c]oordination appears to be symmetric”.29 The shift of status may be marked
by several syntactic signals. It may be signaled by a ‘switch’ in the clausal
pattern, that is, a switch of ‘clause type’ (Isaksson 2013); by the presence (or
absence) of the conjunction we/wa; and/or by an explicitly subordinating
conjunction (which usually also involves a ‘gram-switch’). It may be argued
that a shift of status may also be signaled by the semantic context alone. In the
latter case the linking is syntactically unmarked (it is ‘inferred’).
Since signaling a status shift by subordinating conjunctions is a relatively
commonplace syntactic feature it will be treated just accidentally in this arti-
cle. Only the conjunction we/wa will receive a fuller attention.
The conjunction we/wa and the absence of it (asyndesis)

In accordance with the comparative Semitic evidence, it is assumed that He-
brew inherited only one conjunction wa from early Canaanite, and that the
morphological differentiation we/wa is an internal or secondary innovation.30
29 Isaksson (2013); the quotations are from Cormack and Smith (2005, 395). Since a non-
main clause may precede its main clause, it could be argued that the term ‘digression’ is
misleading. But ‘main line’ is a textual (‘macro-syntactic’) concept that usually involves
several clauses, and so the term is justified at least on a textual level. A ‘digression’
must be distinguished from the case when there is a shift from one main line to another
main line.
30 Thus Müller (1983; 1991); Pardee (2012, 287 note 12). The Masoretic pointing of the
conjunction before the short prefix verb indicates two distinct morphemes, wa (the main
allomorph of which shows a gemination of the following consonant wayyiḵtoḇ < *wa-

The we/wa is one conjunction with complementary allomorphic distribution.

The wa allomorph serves basically to distinguish the syndetic indicative
VprefS (wa-VprefS) from modal (“jussive”) VprefS (we-VprefS).31
I propose that the basic function of the conjunction we/wa is to signal a
clause as an addition in the sense of being an accompanying action in relation
to the action or state in a preceding clause (Isaksson, forthcoming a).32 Thus,
an accompanying action clause practically always occurs in a certain connec-
tion with a preceding ‘pre-clause’.33 This is illustrated in (14):
(14) Pattern: Ø-Vsuff+Ø-Vsuff+{wa-VprefS}+{wa-VprefS}+{wa-VprefS};

så̄ rū mahēr min had-dɛrɛḵ ʾăšɛr ṣiwwīṯīm, ʿå̄ śū lå̄ -hɛm ʿeḡɛl massēḵå̄
{way-yištaḥăwū lō} {way-yizbəḥū lō} {way-yōmərū ...}
‘So quickly they have turned aside from the way I ordered them to follow!
They have cast a metal statue of a calf, {worshipped it}, {sacrificed to it}
{and said, ...}’ (Exod. 32:8; CJB).34
yaktub, Gzella 2012, 101) and wə (with its allomorphs). There are two distinct reflexes
of one original morpheme, wa (with allomorphs) before an indicative VprefS, and we
(with allomorphs) before a non-indicative VprefS and all other instances of the conjunc-
tion. Several explanations of this fact are possible: 1) The Masoretes, 1000 years later in
their texts with vowel signs for synagogal recitation, handed over a syntactic signal that
did not belong to the original Hebrew language (thus Van de Sande 2008; and Revell
1984, 444, advocates the end of the biblical period). 2) The distinction might have been
introduced already in Proto-Hebrew in order to explicitly signal the indicative (‘narra-
tive’) short yiqtol (wa-VprefS), in distinction to the jussive and purposive short yiqtol
(we-VprefS, J-M § 116).
31 Rainey (1986, 6). Joosten (2012, 14-15): “an ancient prefixed preterite was preserved in
a well-defined syntactic environment. The term ‘waw conservative’ has been proposed”.
Revell (1984, 443): “This could have arisen naturally before 3ms waw consecutive im-
perfect forms from roots III h, which had initial stress. This form of the conjunction
could then have been transferred to other forms as a distinguishing mark of the other-
wise generally unmarked semantic category ‘waw consecutive imperfect’. Since other
forms did not have initial stress, the consonant following the conjunction had to be dou-
bled to maintain the length of the syllable.”
32 And contrary to my conclusion in Isaksson (2009, 117).
33 When some types of main line coding (like wa-VprefS) develop to a standard feature of
a certain type of discourse, some exceptions may occur, as when a new narrative begins
with a wayyiqtol clause. An example is wayhī ʾīš ʾɛḥå̄ ḏ ‘Once there was a certain man
...’ (1 Sam. 1:1). In a prose text a yəhī (Ø-VprefS) would have been perceived as a dedi-
cated modal verb.
34 Curly brackets are used in this article to mark out indicative uses of VprefS that do not
take part in a narrative storyline.

186 Bo Isaksson
In this example of direct speech in the mouth of YHWH, the main line is
coded by two (asyndetic) Vsuff clauses. The three wa-VprefS clauses are
additions related to the immediately preceding Vsuff clause (ʿå̄ śū lå̄ -hɛm ʿeḡɛl
massēḵå̄ ), which constitutes the pre-clause. The wa-VprefS clauses code ac-
tions that accompany the ‘They have cast a metal statue of a calf’. It would be
a mistake to translate the wa-VprefS clauses with anterior tense, as if they
were on par with the two Vsuff clauses. An interpretation should be able to
account for the gram switch from Vsuff to VprefS clause. A translation of all
five verbs as being part of the same main line and with the same temporal or
aspectual meaning fails to explain the syntax. 35 The accompanying actions
could be rendered by a “then” or “on that occasion” (if meaning is temporal)
or with a “at that” if maximal clarity would be needed in the translation: ‘in
that occasion they worshipped it, sacrificed to it, and said, ...’. An accompany-
ing action clause marked by an initial we/wa is not an independent clause. The
event time of the wa-VprefS clauses depends on that of the pre-clause.
In a similar English example,
(15a) I have seen John and he smoked a cigar,
there are two clauses, one in the perfect tense, and one in the past tense. The
perfect tense clause I have seen John has a specific inferred event time: the
time when the seeing of John occurred. The second clause is an accompanying
action. It could be translated and on that occasion he smoked a cigar, but this
‘on that occasion’ is not necessary. What signals the special and possibly non-
main accompanying action is the shift to the English past tense. With this
construction the past tense clause and he smoked a cigar ‘takes over’ the
event time of the perfect tense clause. We could say that an accompanying
action has a ‘local event time’ that depends on the event time of the pre-
clause. This dependency on the pre-clause indicates that and he smoked a
cigar has not an illocutionary force of its own.
In this dependency of the event-time in the pre-clause (and often also of
other properties in the pre-clause) lies the special character of ‘non-main
clause’, which a clause with an initial conjunction we often displays in Bibli-
cal Hebrew, and this also pertains to the so-called ‘consecutive tenses’, which
sometimes, in specific text types have stiffened in a certain function (such as
35 Thus NIV, CSB, ESV, etc.

in procedural discourse). It is in the light of this dependency upon a pre-clause

we can understand the perceived ‘conversion’ of tense, mode and aspect,
which occurs in clauses of the type we-qatal and wayyiqtol.36
How an addition easily receives notions of result or consequence is illus-
trated by (15b).
(15b) Do that again and I’ll break your neck.37
It is easy to see that relations between the clauses in (14) and (15a-b) are not
‘commutative’: the order of the clauses cannot be changed without drastically
changing the meaning of the utterances. The clauses after we and and respec-
tively presuppose the fore-going pre-clause. This is usually a distinguishing
feature of additions in Biblical Hebrew (except for some rare cases of unor-
dered addition in main clauses, Dixon 2009, 26; 1 Sam. 2:7).
An accompanying action is not the same as an attendant circumstantial
clause (Isaksson 2009, 19-21). Attendant circumstantial clauses are regularly
signaled by asyndesis in Biblical Hebrew (absence of we). In (16) the asyndet-
ic participle clause is somewhat ambiguous since it could also refer to the
plural subject of the main clause, but the pragmatic setting settles the case and
we infer that the participle niṣṣå̄ ḇīm refers to the two objects Moses and Aa-
ron. In the translation below the circumstantial clause is rendered by a relative
‘who were waiting’ in order to avoid referential ambiguity (thus ESV).
(16) Pattern: wa-VprefS-Onoun+[Ø-PA];

way-yip̄ gəʿū ʾɛṯ Mōšɛ wə-ʾɛṯ ʾAhărōn [niṣṣå̄ ḇīm liqrå̄ ṯå̄ m bə-ṣēṯå̄ m mē-ʾēṯ
parʿō]
‘They met Moses and Aaron, [(who were) waiting for them, as they came
out from Pharaoh]’ (Exod. 5:20).
An example of an explicitly coded attendant circumstantial clause in Eng-

lish is (17).
(17) I have seen John smoking a cigar.
36 The details of the two syntagms will be discussed in C.2 and A.1.2 respectively.
37 The example is taken from Verstraete (2005, 618).

188 Bo Isaksson
In example (17) the circumstantial clause smoking a cigar is much more spe-
cific about the simultaneousness of the action than the accompanying action in
(15a).38
When a we clause describes a state its predicate is usually non-verbal (i.e.
it is a NCl clause) or a participle. Such clauses are better characterized as
signifying an accompanying state. On a scale from dependence to independ-
ence (in relation to the main clause) accompanying state clauses in Biblical
Hebrew are relatively less dependent, while asyndetic attendant circumstance
clauses, as in (16), are more tightly linked to the main clause. The relation is
often hard to render in English. General but clumsy phrases that usually do the
job are ‘at that’ or ‘on that occasion’ (the latter temporal), an example of
which is (18).
(18) Pattern: wa-VprefS+[we-Snoun-PA];

wat-tērɛḏ baṯ parʿō li-rḥōṣ ʿal ha-yʾōr [wə-naʿărōṯɛhå̄ hōləḵōṯ ʿal yad ha-
yʾōr]
‘Then Pharaoh's daughter went down to the Nile to bathe, [and on that oc-
casion her attendants were walking along the riverbank].’ (Exod. 2:5).
The translators hesitate on how to translate the participle clause in (18). Some
(like ESV) has a circumstantial clause, ‘while her young women walked be-
side the river’, while others (like NIV) prefer a less dependent rendering, ‘and
her attendants were walking along the riverbank’.39
In SBH prose, which is characterized by the preferences of an oral narra-
tive tradition (Isaksson 2009, 36), a characteristic feature is the concatenation
of clauses with the conjunction we/wa, a feature that resembles the structure
of modern spontaneous speech in English (Givón 2001, I:299; Isaksson
2014b). This characteristic of narrated prose texts has often been interpreted
as a general preference for ‘coordination’ in Biblical Hebrew. This is not at all
38 Polak (2014, 192-193) points out that the asyndetic circumstantial clause is more inte-
grated in the main clause, while a we clause (with preposed subject) that breaks the nar-
rative sequence is more marked and salient. Its status is different, and it may act “as a
trigger for the ensuing narrative”, often used in literary composition. As Lehmann
(1988) has shown, integration in another (main) clause is a scalar category, not a matter
of either-or.
39 An example with NCl is 1 Sam. 1:24 (wa-VprefS+[we-NCl]).

evident.40 The pre-clause and the accompanying action clause are sometimes
of equal status, but in many other cases they are of unequal status (Bloch
2009, 40 note 26).
Since addition clauses (signaled by we/wa) may describe a wide variety of
clausal relations, such as background information, comment, elaboration,
contrast, temporal linking (by inference), result of a previous action (by infer-
ence), temporal succession (by inference) and even (depending on the predi-
cate) something close to attendant circumstances, a definition of ‘coordina-
tion’ that includes the whole gamut of such clauses is not productive.41 In a
narrative context, specifically, such clauses constitute digressions from the
storyline (except for the storyline of ‘wayyiqtols’ itself).42
An asyndetic clause is often a non-main clause.43 If non-main, it is practi-
cally always circumstantial (Isaksson 2014d, 129, 132, 134).44
Fronting as a signal of topicalization in Biblical Hebrew

It is nowadays recognized in the Hebrew grammatical discussion that fronting
of a non-verbal element serves to signal that element as in some sense topical-
ized (Van der Merwe, et al. 1999, 346-347).45
It is widely agreed among those Hebraists and linguists who have ad-
dressed the problem that preverbal clause components are marked ei-
ther in order to highlight an element that is in focus, or to show a new
topic that is entering the discourse. (Lunn 2006, 29).
40 Also the English conjunction and may connect clauses with unequal status, cf. Ver-
straete (2005, 618). The concepts used here, ‘main line’ and ‘non-main clause’ have the
advantage of supporting a textlinguistic perspective (Isaksson 2013; forthcoming c).
41 The explanatory force of the term would be drastically reduced.
42 In the ‘normal’ case, a main line is characterized by the consistent use of the same gram.
There are several exceptions to this rule of thumb, however, many of which will be dis-
cussed below. An exception that will not be discussed in this article is the negation of
the imperative, which is coded by ʾal-VprefS (jussive, prohibitive meaning), as in the
equal status clauses in Ps. 10:12.
43 For a discussion of non-main clauses, see Isaksson (2014d, 115; 2013; 2014e; 2015).
44 The remarkable exceptions are the VprefS clauses, which may be non-main, but which
never seem to express attendant circumstantial action.
45 In accordance with the terminology introduced by Niccacci, such a non-verbal element
is called an “X” element. Such an element can also “be adverbial, i.e. an adverb proper,
or a preposition + noun / pronominal suffix” (Niccacci 1987, 8). It is understood that the
‘X’ element is not a negation, since a preverbal negation is not topicalized.

190 Bo Isaksson
In (19) the first VprefL clause has a topicalized object noun. In the second
clause (also VprefL) a prepositional phrase is topicalized. Very few translators
have managed to topicalize both corresponding elements in English. In the
ESV rendering below, emphasis of the object in the first clause is achieved by
the powerful adverb ‘nothing’. In the second clause ESV has topicalized the
prepositional phrase by giving it clause-initial position.
(19) Pattern: Ø-Onoun-lō-VprefL + Ø-PrP-VprefL;

kål maḥmɛṣɛṯ lō ṯōḵēlū bə-ḵōl mōšəḇōṯēḵɛm tōḵəlū maṣṣōṯ
‘You shall eat nothing leavened; in all your dwelling places you shall eat
unleavened bread.’ (Exod. 12:20; ESV).
Summary of the introduction

Biblical Hebrew is an ordinary descendant of early Canaanite, with the five
basic finite grams short prefix verb (VprefS), long prefix verb (VprefL), suffix
verb (Vsuff), imperative (IMP) and cohortative (VprefA). It is hypothesized
that the word order restrictions of VprefS and VprefL are caused by the partial
morphological merger of the respective grams.
A cross-linguistic typology of semantic clausal relations has been present-
ed (Dixon 2009). It will constitute the basic semantic terminology when ana-
lysing clause combining in the biblical texts.
The concept of ‘main line’ has been introduced as the basic textlinguistic
concept that defines a main clause: a main clause takes part in the coding of a
main line. A non-main clause represents a ‘digression’ from the main line.
The conjunction we/wa has been discussed as basically signalling a clause
as an ‘addition’ in relation to a previous ‘pre-clause’. Additions can be main
clauses or non-main clauses.
A. Clause combining and the VprefS gram in Biblical Hebrew

Due to the restricted word order in Biblical Hebrew practically all VprefS
verbs in affirmative clauses are clause-initial in our extant Biblical Hebrew
texts. There are some exceptions to this rule, and most of them are due to the
specific character of the Biblical Hebrew clause. Vocatives and pre-
dislocations were not perceived to take part in the clause, for example. The
exceptions will be discussed below.

Negated clauses pose no problem when it comes to identify a prefix verb

as short or long. The negation ʾal is specialized for the VprefS clause with
jussive meaning. Since there is no risk of confusion with an ʾal-VprefS clause,
word order is free in such clauses.
The negation lō poses no problem either, because it does not negate indica-
tive VprefS clauses. When a prefix verb is negated by lō, we know that the
verb is long.
So indicative VprefS clauses are not negated. This means that the familiar
storyline in prose, the narrative core of wa-VprefS clauses is not negated, at
least not by a negation lō and a VprefS verb. The wa-VprefS syntagm is not
‘split up’, and this in order to avoid confusion. Indicative lō was used before
the VprefL, not before VprefS.
But that is not the end of the story. Since indicative perfective lō-VprefS
did exist in the Canaanite of the Amarna tablets,46 the question arises as to
what replaced it as negated clause in the storyline. The reasonable answer is
the we-lō-Vsuff clauses, discussed in C.3.3.
Masoretic pointing indicates a secondary distinction in the morphology of
the original Semitic conjunction wa: the allomorph wa (usually with gemina-
tion of following consonant) was read before an indicative VprefS, and the
allomorph we before a modal VprefS (and all other cases). Thus affirmative
VprefS clauses exhibit three variant patterns, Ø-VprefS, we-VprefS, and wa-
VprefS, all three having a clause-initial VprefS.
A.0.1 Exceptions in word order and morphology

There are some exceptions to the rule of a clause-initial VprefS. One concerns
the vocative. An initial vocative was not perceived to belong to the clause, and
so does not affect the fronted position of the VprefS, as can be seen in (20).
(20) Pattern VOC, Ø-VprefS+Ø-VprefS;

ʾattå̄ YHWH tišmərēm, tiṣṣərɛnnū min had-dōr zū lə-ʿōlå̄ m
‘Do thou, O LORD, protect them! Guard us forever from this generation’
(Ps. 12:8).47
46 Even without word order constraint: a-na-ku / la-a iš-me a-na ša-šu-nu (pattern: Spron-
lā-VprefS) ‘I did not listen to them’ (EA 136:14-15 quoted from Rainey 1996, 212).
47 Another example is Yəhūḏå̄ ʾattå̄ yōḏūḵå̄ ʾaḥɛḵå̄ ‘Judah, may your brothers praise you!’
(Gen. 49:8; VOC, PRE-DIS, Ø-VprefS with resumptive pronoun).

192 Bo Isaksson
Likewise, a pre-dislocation was felt not to belong to the clause, and so does
not violate the rule of initial verb position (Niccacci 1987, 12):48
(21) Pattern: PRE-DIS, Ø-VprefS!;

wə-han-naʿar, yaʿal ʿim ʾɛḥå̄ w
‘As for the boy, let him go back with his brothers.’ (Gen. 44:33)
The Masoretic accent at han-naʿar is distinctive (Ṭip̄ ḥå̄ ) and indicates that the
han-naʿar is not the direct subject of the clause.49
With the negation ʾal clarity was achieved as to which prefix gram was in-
tended. For this reason it was unnecessary to place the ʾal-VprefS syntagm in
clause-initial position. This is illustrated in example (22).
(22) Pattern: ADV-ʾal-VprefS!;

raq ʾal yōsep̄ parʿō hāṯēl
‘Only let not Pharaoh cheat again’ (Exod. 8:25; ESV).50
Mostly, the morphological distinction between VprefS and VprefL is upheld

when possible (Rainey 1986, 7).51 When a discrete VprefL! is encountered in
a text, it is normally intended to have the meaning of a long prefix verb.52 This
is of great value in the interpretation of Hebrew texts. But there are excep-
tions, and they are not few. Examples of morphologically long forms intended
to function as VprefS are found here and there and the signal that the short
form meaning was intended is most often a clause-initial position.
48 Same observation by Niccacci (1987, 2.1.2). ‘Pre-dislocation’ is here used in the sense
of ‘left-dislocation’, or ‘casus pendens’ in older literature. The terms ‘right’ and ‘left’
presuppose a written medium and are less pertinent when large parts of texts may have
been orally transmitted at a first stage after their creation.
49 Another example is wə-hå̄ -ʿōp̄ yirɛḇ bå̄ -ʾå̄ rɛṣ ‘and the birds, may they multiply on the
earth’ (Gen. 1:22; pattern we-PRE-DIS, Ø-VprefS!), with accent ṭip̄ ḥå̄ .
50 Another example is Gen. 37:22, wə-yāḏ ʾal tišləḥū ḇō ‘but don't lay a hand on him’
(NIV), with pattern we-Onoun-NEG-VprefS. Also Exod. 16:19.
51 Rainey’s scheme of three indicative verb forms and three injunctive verb forms (1986,
8) is, however, contradictory. He speaks of “The injunctive use of the imperfect”, alt-
hough the yaqtulu in his scheme is only indicative.
52 When a verb form is morphologically discrete, as the yōsep̄ in (22), I add an exclama-
tion mark to the abbreviation. The exclamation mark is not added to a wa-VprefS syn-
tagm, since the wa allomorph of the conjunction exclusively marks the gram as short
(and indicative). When way-yiqtol syntagms occur as formal wa-VprefL, it is a later de-
velopment in Biblical Hebrew, and has no significance for the meaning and function of
the verb (Bloch 2007).

(23) Pattern: VprefSᴸ+[we-VprefS];

yå̄ qūm ʾå̄ ḇī [wə-yōḵal miṣ-ṣēḏ bənō]
‘Let my father sit up [and eat of his son’s game]!’ (Gen. 27:31; TNK)
The yāqūm in (23) has the meaning of a jussive and it is fronted. We would
have expected a short yå̄ qŏm, of course, but the context forces us to analyse
the verb as a jussive VprefS, although the formal reading is a long prefix verb
(‘VprefSᴸ’). It seems that a clause-initial position in some instances, also in
archaic poetry,53 was felt to be a sufficient signal of short prefix verb mean-
ing, even when the morphology of the verb contradicted its position.54
A.1 VprefS clauses encoding a main line

Main clauses with a VprefS verb as predicate are most frequently either narra-
tive perfectives with past time reference, or ‘jussives’. Less frequently the
VprefS gram may also have a general present meaning. The meanings will be
discussed below.
A.1.1 The VprefS clause with jussive meaning

(24) Pattern: Ø-VprefS+[we-VprefS]+Ø-VprefS!;
yå̄ qūmū [wə-yaʿzərūḵɛm] yəhī ʿălēḵɛm siṯrå̄
‘Let them rise up [to help you]! Let them give you shelter!’ (Deut. 32:38;
NIV).
In (24) two asyndetic VprefS clauses form a main line with jussive meaning.
The two asyndetic clauses are seemingly of equal status. Of the two, the sec-
ond form is discretely short (yəhī), while the first (yå̄ qūmū) could formally be
long. What settles the analysis is 1) that also the first VprefS verb is fronted in
its clause, 2) that the main line also contains a discretely short form (yəhī),
and 3) that the semantic context favours a jussive meaning in both clauses.
53 An example is yå̄ qūm ʾɛ̆lōhīm yå̄ p̄ ūṣū ʾōyəḇå̄ w ‘May God arise, may his enemies be
scattered!’ (Ps. 68:2; NIV; pattern Ø-VprefSᴸ+Ø-VprefS). Examples with indicative
(‘perfective’) meaning but long form are taḵrīăʿ in Ps. 18:40 (Ø-VprefSᴸ), and tå̄ ḵīn in
Ps. 18:2 (elaboration). The phenomenon is further discussed in the introduction to sec-
tion C.2.
54 The [we-VprefS] in (23) is a purpose clause. Such clauses will be treated under A.2.3
below.

194 Bo Isaksson
The main line pattern in the example is coded by asyndetic VprefS clauses
with jussive meaning.55
The negation ʾal is the exclusive negation of the jussive VprefS gram. If
ʾal is prefixed to a Vpref verb, every receiver of the text knows that the verb is
short and jussive in meaning, as in (25).56
(25) Pattern: Ø-ʾal-VprefS;

ʾal taʿămoḏ
‘don't delay!’ (Gen. 45:9; NIV).
The verb in (25) is not discretely short, but because of the negation ʾal it is
safe to analyse the verb as a jussive.
A non-main clause may of course precede its jussive VprefS main clause
and does not violate the rule of clause-initial affirmative VprefS. This is
shown in the following example.
(26) Pattern: [PREP-VN]+Ø-VprefS!+Ø-VprefS;

[bə-šūḇ YHWH šəḇūṯ ʿammō] yāḡel yaʿăqōḇ yiśmaḥ yiśrå̄ ʾēl
‘[When the LORD restores his people], let Jacob rejoice and Israel be
glad!’ (Ps. 14:7; NIV)
Also in (26) the main line is encoded by asyndetic VprefS clauses with jussive
meaning. The ‘digression’ in the form of a VN clause is placed before the two
main line clauses. The meaning of the VN clause in relation to the main line is
‘relative time’, rendered by NIV with the conjunction ‘when’.57
A main line of affirmative jussive VprefS clauses is characterized by asyn-
desis and clause-initial position of the verb.
55 The we-VprefS clause constitutes a non-main clause with consequence-purpose mean-

ing (Dixon 2009, 22; J-M § 116). See A.2.3 below.
56 If indicative VprefS ever was negated in pre-Biblical (‘proto-’) Hebrew, the expected
negation would have been lō, as can be seen in Amarna Canaanite (Rainey 1996, 212).
Possible exceptions (the existence of lō-VprefS clauses) are discussed by Ges-K § 109
d,k; König 1897 III § 191 c,g (I thank Ambjörn Sjörs for these references). The fact that
lō-yiqtol was associated with the long prefix verb (VprefL) and indicative VprefS was
not (no longer?) negated, raises the question which type of negative clause replaced the
indicative lō-VprefS in Biblical Hebrew. A reasonable alternative is that (we-)lō-Vsuff
did the job, see C.3.3.
57 VN clauses as non-main clauses are only incidentally discussed in the present article.
They are treated in Isaksson (2007).

A.1.2 The VprefS clause with indicative perfective meaning

In Biblical Hebrew prose texts, the literary convention requires the affirmative
clauses in a storyline to be syndetic with the ‘indicative’ allomorph wa of the
conjunction.
(27) Pattern: wa-VprefS+wa-VprefS;

wayḇīʾūhū yərūšå̄ layim wayyå̄ måṯ šå̄ m
‘They brought him to Jerusalem and he died there.’ (Judg. 1:7).
Both clauses in (27) are syndetic and have an initial conjunction wa. This
allomorph of the conjunction is a signal in the Masoretic reading tradition that
the prefix verb is short and indicative.58 The basic meaning of the conjunction
(‘accompanying action’) is usually neutralized in the chain of a narrative main
line (‘storyline’). In the typical case, as in (27), the chain is iconic, and a sec-
ond clause is perceived to express a temporal succession (the default interpre-
tation).
The wa allomorph in a VprefS chain is sometimes absent, especially in ar-
chaic poetry, although the meaning is unmistakably indicative and the aspect
perfective with past reference:
(28) Pattern: Ø-VprefS+Ø-VprefS+Ø-VprefS+Ø-VprefS;

yimṣå̄ ʾēhū bə-ʾɛrɛṣ miḏbå̄ r ū-ḇə-ṯōhū yəlēl yəšīmōn yəsōḇəḇɛnhū
yəḇōnənēhū yiṣṣərɛnhū kə-ʾīšōn ʿēnō
‘He found him in a desert land, and in the howling waste of the wilder-
ness; he encircled him, he cared for him, he kept him as the apple of his
eye.’ (Deut. 32:10; ESV).
All interpreters take this passage as a narrative with perfective verbs. None of
the verb forms is discretely short, but the context and the fact that all four
verbs are put in fronted position in their clauses, signal that the prefix verbs
are short in (28).59
In a West-Semitic perspective, VprefS is the older formation while Vsuff
is an innovation. As can be seen in (28), the indicative perfective VprefS
58 This is a valuable information to a late reader when confronted with difficult contexts,
where the form of the verb is not discrete.
59 This does not mean that the four Ø-VprefS have the same textual function as wa-VprefS
clauses. In (28) the VprefS clauses have equal status, but they are not additions, and
temporal succession is not implied.

196 Bo Isaksson
could be used independently of Vsuff in archaic poetry (as it could in Amarna

Canaanite). This is a feature that is less common in Biblical Hebrew prose,
where the Vsuff tends to introduce a series of events (and gradually takes over
more and more indicative functions from the old VprefS).60
When negative storyline clauses of the type *we-lō-VprefS turned out un-
acceptable in pre-Biblical Hebrew because of the risk of confusion with the
we-lō-VprefL clause type, it was replaced by the we-lō-Vsuff clause (see
C.3.3).
A.1.3 Main line VprefS clauses that are neither jussive nor perfective
As was indicated in the introduction, the VprefS gram is by itself not just a
jussive, nor by its nature just a perfective. The jussive and perfective mean-
ings are triggered by specific contexts, and there are meanings of this old
‘zero’ gram that cannot be classified as jussive or perfective.
(29) Pattern: wa-VprefS;

way-yilqəṭū ʾōṯō bab-boqær bab-boqɛr
‘Morning after morning they gathered it.’ (Exod. 16:21)
The VprefS clause in (29) has frequentative or habitual meaning, but it would
be premature to conclude that the VprefS gram itself has this meaning. Habit-
uality is signaled by the two adverbs bab-boqɛr bab-boqɛr ‘morning after
morning’. It is more to the point to state that the VprefS gram allows for a
frequentative meaning in certain contexts. As the example shows, the VperfS
gram is not ‘punctual’ as is sometimes maintained, and it is not just perfective.
That the VprefS has a prototypical meaning that is far from ‘punctual’ is
shown by its ability to express also a general or continuous actions:
(30) Pattern: wa-VprefS;

wa-yḥī yaʿăqōḇ bə-ʾɛrɛṣ miṣrayim šəḇaʿ ʿɛśrē šå̄ nå̄
‘Jacob lived in Egypt seventeen years.’ (Gen. 47:28)
60 I know of only a few possible prose examples of indicative perfective Ø-VprefS clauses
in SBH: Judg. 2:1 (J) (Ø-VprefS+wa-VprefS in the beginning of direct speech), 6:5 ([kī-
Spron-Snoun-VprefL]+Ø-VprefS ‘[When they invaded with their cattle and tents], they
came as thick as locusts’). Clause-initial in relative clause, 2 Kgs 8:29 (J), 9:15 (J), Hab.
3:14, Ps. 7:16. In poetry numerous examples: Deut. 32:8,9,10,11, 33:8,10, Exod. 15:7,
17, Ps. 8:7, 18:17,19,41,44,45, 22:19. (J) refers to Joosten (1999, 24) who quotes the ex-
ample as in some sense a problematic “yiqtol”.

What is described in (30) is a situation, a state, rather than a single event.

Even as a state it is characterized by a certain activity.61
The ‘zero’ meaning of the VprefS gram makes it suitable for gnomic con-
texts as well, which can be seen in (31):
(31) Pattern: Ø-VprefS!;

yaḏreḵ ʿănå̄ wīm bam-mišpå̄ ṭ
‘He guides the humble in what is right.’ (Ps. 25:9; NIV)
The most natural translation of this poetic sentence is a general present. It

characterizes the way the Lord deals with humble persons. The VprefS is
certainly dynamic but it expresses a general (or gnomic) present.62
A discrete short form is found in (32) where a general state of affairs with
past reference is expressed:63
(32) Pattern: wa-VprefS+[kī-NCl]+[we-Snoun-PA];

way-yɛʾɛ̆haḇ yiṣḥå̄ q ʾɛṯ ʿēśå̄ w [kī ṣayid bə-p̄ īw] [wə-riḇqå̄ ʾōhɛḇɛṯ ʾɛṯ
yaʿăqōḇ]
‘Isaac loved Esau because he had a taste for fresh game, but Rebekah
loved Jacob.’ (Gen. 25:28; NET)
The VprefS in (32) describes a state in a main line, and the corresponding
state of Rebekah as an accompanying concurrent circumstance is coded by a
participle clause signifying a continued state of attention.
61 Bo-Krister Ljungberg has rightly remarked at a seminar in Uppsala that the 17 years can
be viewed ‘as a single whole’ and thus with a perfective aspect.
62 Other examples are Ps. 25:12 and Prov. 1:5 (Ø-VprefS+[we-VprefS!] ‘The wise man
listens [to increase in learning]’). The form of the initial verb in the latter examples is
not discrete and could formally be a long prefix verb as well. But it is clause-initial,
which, as we have seen, is an indication of a short prefix verb. The second clause is a
purpose clause and does not belong to the main line (J-M § 116).
63 I regard the form to be discrete because of the wa allomorph, which signals that the
following Vpref is short.

198 Bo Isaksson
A.2 The VprefS clause encoding a non-main clause

A.2.1 The asyndetic VprefS clause
A.2.1.1 Ø-VprefS with a modal nuance (purpose)
Non-main asyndetic VprefS clauses with a discrete (‘short’) morphology are
rare. A purpose clause with a VprefS gram usually accords with the pattern
we-VprefS (J-M § 116). An asyndetic example is (33).
(33) Pattern: IMP+we-IMP+[Ø-VprefS!];

qaḥ ʾɛṯ maṭṭəḵå̄ wə-hašleḵ lip̄ nē parʿō [yəhī lə-ṯannīn]
‘Take thy rod, and cast it down before Pharaoh, [that it become a serpent].’
(Exod. 7:9; JPS)
Asyndetic purpose clauses with the VprefS gram are unusual, but the example
shows that in a clear context (in this case a preceding imperative) it was per-
missible to discard the conjunction we.64
A.2.1.2 The non-main Ø-VprefS with indicative meaning

A syndetic wa-VprefS clause in non-main position may often describe an
elaboration of the preceding main line clause, and this function is sometimes
found also with an asyndetic clause. For the modern interpreter the distinction
between an elaboration and a temporal linking (and result) is sometimes diffi-
cult to discern, especially in poetry, in that both can be coded by the same
syntactical construction. A reasonably clear example is (34).
(34) Pattern: Ø-Vsuff+{Ø-VprefS};65

nå̄ ṭīṯå̄ yəmīnəḵå̄ {tiḇlå̄ ʿēmō ʾå̄ rɛṣ}
‘You stretched out your right hand {(and) the earth swallowed them}.’
(Exod. 15:12)
64 Another asyndetic VprefS! clause with purpose nuance is hiššå̄ mɛr ləḵå̄ [ʾɛl tōsɛp̄ rəʾōṯ
på̄ nay] ‘Take care [that you do not see my face again]’ (Exod. 10:28; NRS; pattern
IMP+[Ø-NEG-VprefS!]), where the negation ʾal is read with an unusual vowel ɛ. Fur-
ther example: Ps. 13:6, with a Ø-VprefS in a third person purpose clause, and then a Ø-
VprefA purpose clause in the first person. This is in accord with J-M § 116, which also
discusses the syndetic we-IMP clause as having a sense of purpose in second person
clauses, as in example (54): wə-hašleḵ which could be interpreted as ‘to cast it down
(before Pharaoh)’. Other examples: Gen. 47:29 and Exod. 7:9.
65 Indicative non-main VprefS clauses are enclosed within curly brackets.

In (34) there is a temporal succession between the stretching out the hand and
the swallowing of the earth. But the receiver of the text also perceives that
there is a causal relationship between the two actions: it is implied that the
swallowing in some sense is a consequence (result) of the Lord’s stretching
out his hand.
A Ø-VprefS clause may also code an elaboration of a state in a noun
clause. Also the following example, (35), is from a poetic text and the VprefS
may be interpreted as a result clause.66
(35) Pattern: NCl + {Ø-VprefS + [we-Onoun-VprefL] + [we-Onoun-

VprefL]};
ʾēl mōṣīʾō mim-miṣrayim kə-ṯōʿăp̄ ōt rəʾēm lō {yōḵal gōyīm ṣå̄ rå̄ w [wə-
ʿaṣmōṯēhɛm yəḡå̄ rēm] [wə-ḥiṣṣå̄ w yimḥå̄ ṣ]}
‘The God who brought him out of Egypt is for him like the horns of the
wild ox, {that he devours the nations who are his adversaries [and at that
he crushes their bones in pieces], [and shatters them with his arrows]}.’
(Num. 24:8)
The noun clause is best interpreted as a main line clause characterizing the
God who brought Israel out of Egypt. He defends them. When an enemy ap-
proaches them, this enemy will find that Israel’s God is like the horns of the
wild ox. The switch to a VprefS clause has the effect of a result of this charac-
teristic of God: he devours the nations who are his adversaries.67
A non-main asyndetic VprefS clause is not an expression of concomitant
(circumstantial) action, as is usually the case with non-main asyndetic VprefL
and Vsuff clauses (B.2.1 and C.3.1).
A.2.2 The non-main syndetic wa-VprefS clause

The wa-VprefS clause is indicative, and usually describes a dynamic action.
The conjunction (wa) indicates an action or state that occurs in connection
with another action or state. Thus, as a digression from another type of main
line, a wa-VprefS clause may describe for example an elaboration, a same-
event addition, an additional action with an implied temporal linking, an im-
66 The example could also be interpreted as a purpose clause (thus modal), ‘to devour the
nations ...’, but no translator seems to take this path of interpretation.
67 The remaining Vpref grams are not clause-initial, and should be analysed as long prefix
grams in accompanying action clauses: crushing, shattering. See B.2.2, below.

200 Bo Isaksson
plied temporal succession, or an implied result (Dixon 2009, 26-28). The

details are elaborated below.
A.2.2.1 The non-main wa-VprefS as a general accompanying action

The non-main wa-VprefS as a general accompanying action is rare (but see
(39)). The more frequent meanings of non-main wa-VprefS clauses are elabo-
ration and temporal succession. They are implied meanings derived from the
wa-VprefS clause as an accompanying action clause (signaled by wa) with a
VprefS gram predicate. The general semantic meaning of addition, without
implied nuances are most easily perceived when the pre-clause describes a
habitual action. In (36) this habitual action in the pre-clause is coded by an
active participle (PA).
(36) Pattern: we-Spron-PA+{wa-VprefS};

wǝ-hī yōšɛḇɛṯ taḥaṯ-tomær dǝḇōrå̄ bēn hā-rå̄ må̄ ū-ḇēn bēṯ-ʾēl bǝ-har
ʾɛp̄ rāyim {way-yaʿălū ʾēlɛhå̄ bǝnē yiśrāʾēl lam-mišpå̄ ṭ}
‘She used to sit under the palm of Deborah between Ramah and Bethel in
the hill country of Ephraim; {and on that occasion the Israelites came up
to her for judgment}.’ (Judg. 4:5)
Within a context of narrative prose, in the story of Deborah, the participle

clause introduces a piece of background information. The listener is informed
that the prophetess Deborah used to sit under a certain palm, and when she sat
there, the Israelites came to her. She did not sit there all the time, but from
time to time she did, and then the Israelites came to her for judgment.
The wa-VprefS clause depends on this local time reference, i.e. the event
time of the PA clause which brings about the interpretation of punctual but
habitual events (each time she sat there). Thus, the VprefS clause can be in-
terpreted as having a temporal connotation ‘at those occasions’ = ‘then...’ (cf.
C.2.1.2 below). Since the implied meaning is habitual, it is perhaps natural to
analyse the VprefS clause as a general accompanying action that in the con-
text takes on a frequentative meaning.
A habitual meaning is found also in a direct quotation of Jacob in his ar-
gument with Laban.
(37) Pattern: NCl+[Ø-Vsuff+{wa-VprefS}];

zɛ lī ʿɛśrīm šå̄ nå̄ bə-ḇēṯɛḵå̄ [ʿăḇaḏtīḵå̄ ʾarbaʿ ʿɛśrē šå̄ nå̄ bi-štē ḇənōṯɛḵå̄
wə-šēš šå̄ nīm bə-ṣōnɛḵå̄ {wat-taḥălep̄ ʾɛṯ maśkurtī ʿăśɛrɛṯ mōnīm}]

‘It is now twenty years for me in your household: [I have served you four-
teen years for your two daughters, and six years for your flock; {and you
changed my wages ten times}].’ (Gen. 31:41)
In this direct quotation, the circumstantial asyndetic Vsuff clause works as an

elaboration of the initial noun clause: ‘in that I have served you fourteen years
for your two daughters...’. The Vsuff clause adds additional information about
the twenty years described in the NCl. Then, there is a switch to a wa-VprefS
clause to add further information with an accompanying action clause: during
these twenty years, Laban changed Jacob’s wages ten times. The wa-VprefS
clause is not an elaboration of Jacob’s serving, and it is not a temporal succes-
sion. It is a separate accompanying action (with a different actant) that occurs
during Jacob’s 20 years of hard work. The most natural aspectual meaning of
the Vsuff clause is anterior. Jacob stands in front of Laban and summarizes
his life in Laban’s household: I have served fourteen years, I have served six
years. The wa-VprefS clause is dependent on the Vsuff clause and takes over
its event time. It has a ‘local’ temporal reference, and should not be interpret-
ed as an anterior, because it is not anterior in relation to the Vsuff clause. This
is the effect and meaning of the switch: in relation to the Vsuff clause the wa-
VprefS clause describes an action that accompanies it, and it is semantically
dependent on the action described by the Vsuff clause.
A general indicative accompanying action is found in wa-VprefS2 of (38).
(38) Pattern: kī-ADV-Vsuff1 + {wa-VprefS1} + Ø-ADV-Vsuff2 + {wa-

VprefS2};
kī šešɛṯ yå̄ mīm ʿå̄ śå̄ 1 YHWH ʾɛṯ haš-šå̄ mayim wə- ʾɛṯ hå̄ -ʾå̄ rɛṣ ʾɛṯ hay-yå̄ m
wə-ʾɛṯ kål ʾăšɛr bå̄ m {way-yå̄ naḥ1 bay-yōm haš-šəḇīʿī}
ʿal kēn bēraḵ2 YHWH ʾɛṯ yōm haš-šabbå̄ ṯ {wa-yqaddəšēhū2}
‘For in six days the LORD made1 heaven and earth, the sea, and all that is
in them, {and rested1 on the seventh day}. Therefore the LORD blessed2
the Sabbath day {and made it holy2}.’ (Exod. 20:11; ESV)
Example (38) illustrates that the interpretation of a switch of clause type (here
perfective Vsuff / perfective VprefS) depends on the text type. In narrative
prose, we would interpret the Vsuff clauses as coding a perfective event with
topicalized preposed adverbs, kī šešɛṯ yå̄ mīm ‘for in six days’ and ʿal kēn
‘Therefore’. In that case both Vsuff clauses and wa-VprefS clauses would
have been clauses within a narrative storyline (the wa-VprefS clauses without

202 Bo Isaksson
topicalized elements). But the passage is quoted from the ten commandments.
The two Vsuff clauses are uttered as a reason for the Sabbath commandment.
In such a context, the Vsuff clauses are perceived as main clauses. The two
wa-VprefS clauses are accompanying actions, and in the first clause (wa-
VprefS1), the relation, by implication, is one of temporal succession. First,
God created heaven and earth and sea in six days, then, on the seventh day, he
rested. The second wa-VprefS (wa-yqaddəšēhū) expresses a general ‘and (in
that connection) he made it holy’. The most natural and less strained interpre-
tation of wa-VprefS2 is that the consecration (‘made it holy’) was an act that
accompanied the blessing. Not necessarily as a result, however, nor as some-
thing that temporally succeeded it.
A.2.2.2 The wa-VprefS when the pre-clause has anterior meaning

When the pre-clause has anterior meaning, the wa-VprefS clause has a time
reference that relates to the time reference of the pre-clause, i.e. the wa-
VprefS has a ‘localized’ time reference, cf. example (14) above. In other
words, the wa-VprefS does not have an independent time reference in such
cases. The translators may be pressed to render the wa-VprefS clause with the
anterior or pluperfect, but this is not what the switch to wa-VprefS expresses.
The accompanying action of the wa-VprefS clause ‘takes over’ the event time
of the pre-clause, and even when the Vsuff of the pre-clause has anterior
meaning, the following wa-VprefS still expresses the meanings that accord
with the properties of an indicative non-main VprefS addition clause: perfec-
tive action, temporal succession, elaboration, etc. A nuance of temporal suc-
cession is found in (39).
(39) Pattern: we-Snoun-Vsuff+{wa-VprefS};

wə-ḥɛḇɛr haq-qēnī nip̄ rå̄ ḏ miq-qayin mib-bənē ḥōḇå̄ ḇ ḥōṯēn mōšɛ {way-yeṭ
ʾåhå̆ lō ʿaḏ ʾēlōn ba-ṣʿannīm}
‘Now Heber the Kenite had left the other Kenites, the descendants of
Hobab, Moses' brother-in-law, {and on that occasion he pitched his tent by
the great tree in Zaanannim near Kedesh}.’ (Judg. 4:11)
The whole example (40) is a piece of background (‘comment’) information.

The receiver of the text needs the information of an event that had happened
before the now of the storyline about Deborah. This is one of the passages that
have been taken to prove that VprefS grams may also have resultative mean-
ing. But this conclusion does not take into account the reason for the gram

switch. The wa-VprefS clause depends on the Vsuff clause for its event time.
In relation to the Vsuff clause it is not anterior, it is just a clause describing an
accompanying action that supplies more information about Heber’s moving to
a place near Kedesh. When we analyse the wa-VprefS clause as an accompa-
nying action, it is not necessary to interpret the clause as resultative or anteri-
or. In the example, the wa-VprefS has a meaning of ‘pure’ accompanying
action, that is, it lacks implied temporal or logical connotations.68
In other instances of wa-VprefS with anterior pre-clause, the meaning is
often one of temporal succession, as in (40).
(40) Pattern: kī-Vsuff+{wa-VprefS};

kī-śå̄ rīṯå̄ ʿim ʾɛ̆lōhīm wə-ʿim ʾănå̄ šīm {wat-tūḵå̄ l}
‘for you have striven with God and with men, {and in that connection you
prevailed}.’ (Gen. 32:29)
As usual, the temporal value of the wa-VprefS clause depends on the (point
in) time of the event in the preceding anterior Vsuff, and it is therefore not
itself anterior. The reason for the switch (to a wa-VprefS clause) is to express
a temporal succession, nearly a result. Jacob’s prevailing can be conceived of
as a result of his striving with God and men.69
It is not necessary that the temporal reference is past time. With a present
or present anterior pre-clause, a switch to a wa-VprefS clause may describe a
generally valid action in the present, as in (41).
(41) Pattern: Onoun-Vsuff+{wa-VprefS};

qaštō ḏå̄ raḵ {wa-yḵōnənɛhå̄ }
‘he has bent his bow {and (on that occasion he) aims it}.’ (Ps. 7:13)
In (41) the interpretation of a temporal succession presupposes that the aiming

follows after having bent the bow (with the foot).70
68 This is the case also in Num. 23:4 with the switch Onoun-Vsuff+{wa-VprefS}, ‘I have
prepared the seven altars {and on that occasion I offered a burnt offering of one bull and
one ram on each altar}.
69 A nuance of temporal succession after an anterior pre-clause is found also in Exod. 18:8
(Vsuff+{wa-VprefS} within a relative construction), Gen. 31:34 (we-Snoun-Vsuff +
{wa-VprefS}+{wa-VprefS}), 39:13 (with kī-Vsuff+{wa-VprefS}), Judg. 4:11 (Snoun-
Vsuff+{wa-VprefS}).
70 A case with general present tense in the pre-clause is Ps. 3:5 with Onoun-VprefL+{wa-
VprefS}, and a clear temporal linking with a nuance of result: ‘I call out to the LORD,

204 Bo Isaksson
An addition that expresses a temporal linking may often receive result

connotations (Dixon 2009, 28). This is the case in (42).
(42) Pattern: we-Onoun-Vsuff+{wa-VprefS};

wǝ-ʾɛṯ ha-yḇūsī yōšēḇ yǝrūšå̄ layim lō hōrīšū bǝnē ḇinyå̄ mīn {way-yēšɛḇ
ha-yḇūsī ʾɛṯ bǝnē ḇinyå̄ mīn bīrūšå̄ layim ʿad hay-yōm haz-zɛ}
‘The Benjamites, however, did not drive out the Jebusites, who were living
in Jerusalem; {to this day the Jebusites live there with the Benjamites}.’
(Judg. 1:21; NIV)
This passage in the account of the conquest of Canaan is a comment on a

shortcoming of the Benjamites: they did not drive out the Jebusites. The func-
tion of the switch to a VprefS clause is to describe the result of this imperfec-
tion, a result that lasted until the present time of the storyteller. Considering
the function of this switch, it is justified to translate, ‘therefore the Jebusites
live there with the Benjamites to this day’.71
Sometimes, an addition with a connotation of temporal linking does not
describe a result clause. It may receive other nuances according to the context,
and one example is (43) which displays a temporal linking with a nuance of
contrast.
(43) Pattern: kī-Vsuff+{wa-VprefS};

kī rå̄ ʾīṯī ʾɛ̆lōhīm på̄ nīm ʾɛl på̄ nīm {wat-tinnå̄ ṣel nap̄ šī}
‘I have really seen God face to face, {and yet my life was spared}.’ (Gen.
32:31)
In the example, the Vsuff clause is a resultant present state of facts. The main
point here is that, although the wa-VprefS clause certainly describes a tem-
poral linking, this action cannot in any way be called a result of the preceding
action. The addition coded by the wa-VprefS clause in this context describes a
contrast (Dixon 2009, 28).
An anterior pre-clause may also. in rare cases. express an irreal action,
something that could have happened, but for some reason did not. In such
cases a following wa-VprefS clause is also irreal, as in (44).
{and he answers me from his holy mountain}’ (NIV). Another present tense example is
1 Sam. 2:6 (with a PA pre-clause).
71 Similar result connotations of wa-VprefS clauses (after anterior pre-clauses) are found
in Gen. 30:27, 44:20 (result of divination).

(44) Pattern: ADV-Vsuff+{wa-VprefS}+{wa-VprefS};

kī ʿattå̄ šå̄ laḥtī ʾɛṯ yå̄ ḏī {wå̄ -ʾaḵ ʾōṯəḵå̄ wə-ʾɛt ʿamməḵå̄ bad-då̄ ḇɛr} {wat-
tikkå̄ ḥēḏ min hå̄ -ʾå̄ rɛṣ}
‘For by now I could have stretched out my hand {and (in that connection)
I struck you and your people with a plague} {and you were wiped off the
earth}.’ (Exod. 9:15)
The irreality of the action is signaled by the ʿattå̄ ‘now’ and the Vsuff gram.
The Vsuff clause, in this context, receives a nuance of ability or possibility.
There is nothing in the two wa-VprefS clauses that signals an irreal action.
Instead, the notion of irrealis is taken over from the Vsuff clause on which the
wa-VprefS clauses depend. The additions, coded by the wa-VprefS clauses,
are implied temporal successions, the last (‘and you were wiped off the earth’)
with a nuance of result.
A.2.2.3 The wa-VprefS expressing temporal succession or result

A temporal succession is one of the most common implied meanings of a wa-
VprefS clause, as in (45).
(45) Pattern: Snoun-Vsuff+we-ADV-VNabs-Vsuff+{wa-VprefS};

ʾīš miṣrī hiṣṣīlå̄ nū miy-yad hå̄ -rōʿīm wə-ḡam då̄ lō ḏå̄ lå̄ lå̄ nū {way-yašq ʾɛṯ
haṣ-ṣōn}
‘An Egyptian rescued us from the shepherds. He even drew water for us
{and watered the flock}.’ (Exod. 2:19)
In this direct quotation, the main line is coded by the two Vsuff clauses that
are not temporally related. The wa-VprefS clause codes an action that specifi-
cally accompanies the action in the second Vsuff clause, and the implied
meaning of the wa-VprefS is a temporal succession.72
A result is a frequent implied meaning of a wa-VprefS clause (Dixon 2009,
28). Only the context decides on the correct meaning for a translation. A
straightforward example is (46).
72 Similar cases of wa-VprefS clauses with an implied meaning of temporal succession.

but without the nuance of result. are: Gen. 24:16 (PA+{wa-VprefS}+{wa-VprefS} +
{wa-VprefS}), 25:8 (NCl/wa-VprefS in genealogy), 27:33 and 30:30 (both NCl/wa-
VprefS in direct speech), 37:7 and 41:3-4 and 41:18 (all three in dream account), 44:28
(Vsuff/wa-VprefS in direct speech), Exod. 6:25 (genealogy), 14:10 (Vsuff/wa-VprefS),
15:19 (Vsuff/wa-VprefS archaic poem), Judg. 18:18 (Vsuff/wa-VprefS), 2 Sam. 1:27
(Vsuff/wa-VprefS poem), Ps. 3:6 (Vsuff/wa-VprefS ‘I lie down {and sleep}’).

206 Bo Isaksson
(46) Pattern: Snoun-Vsuff+{wa-VprefS};

han-nå̄ ḥå̄ š hiššīʾanī {wå̄ -ʾōḵēl}
‘The serpent tricked me, so I ate.’ (Gen. 3:13; CJB)
It is of course acceptable to translate, ‘The serpent deceived me, and I ate’

(thus ESV), since also the English addition clause ‘and I ate’ has an implied
sense of result.
A wa-VprefS clause may express a result also from other types of pre-
clauses. In (47) a habitual meaning is expressed in a pre-clause with a long
prefix verb gram.
(47) Pattern: VN+ADV-VprefL+{wa-VprefS+we-lō-Vpref};

middē ʿălōṯå̄ h bə-ḇēṯ YHWH kēn taḵʿīsɛnnå̄ {wat-tiḇkɛ [wə-lō tōḵal]}
‘Whenever Hannah went up to the house of the LORD, her rival provoked
her {till she wept and did not eat}.’ (1 Sam. 1:7; NIV)
In the example, the habitual meaning of the first VprefL is triggered by the
initial adverbial VN phrase. The wa-VprefS clause is an accompanying action
that in the context receives a habitual meaning and a sense of result. The wa-
VprefS clause is dependent on the pre-clause for such meanings. It takes over
the past reference and the habituality, and adds an implied nuance of result.
In (48) the temporal reference is past time, but there is no notion of habitu-
ality.
(48) Pattern: Ø-PrP-Vsuff+{wa-VprefS};

ʾɛl han-naʿar haz-zɛ hiṯpallå̄ ltī {way-yittēn YHWH lī ʾɛṯ šəʾēlå̄ ṯī ʾăšɛr
šå̄ ʾaltī mē-ʿimmō]
‘This is the child I then prayed for, {and the LORD granted my request}.’
(1 Sam. 1:27)
This is taken from the direct speech of Hannah to the priest Eli, and before
this quotation she says ‘Pardon me, my lord. As surely as you live, I am the
woman who stood here beside you praying to the LORD’ (NIV). Now she has
the boy with her, and in (49) she says with a topicalized object constituent
(formally PrP): this is the child. The main clause in (48) is the Vsuff, and the
wa-VprefS is just an addition, ‘and the Lord granted my request’, with im-
plied temporal succession, which of course may be interpreted as a result
(Dixon 2009, 28).

A wa-VprefS clause may also express a temporal succession from the ac-
tion in a participle pre-clause, as in (49).
(49) Pattern: Snoun-PA+we-PA+Ø-PA+{wa-VprefS};

YHWH mēmīṯ ū-məḥayyɛ mōrīḏ šå̄ ʾōl {way-yå̄ ʿal}
‘The LORD deals death and gives life, casts down into Sheol {and raises
up}.’ (1 Sam. 2:6)
Example (49) displays four clauses, semantically divided into pairs, PA + we-
PA and PA+{wa-VprefS}. The first pair describes the continual activity of the
Lord dealing with life and death, but the second pair, with the switch to a
VprefS clause, encodes a temporal succession. First the Lord casts down into
Sheol, and then, after that, he raises up. There is no consequence implied in
this switch, and the wa-VprefS clause, which is marked as indicative by the
form of the conjunction, has an unmistakable general present meaning. Alt-
hough the example is from a piece of poetry which is regarded archaic, we can
observe that a generally valid present was compatible with the trajectory of
the VprefS gram.73
A.2.2.4 The wa-VprefS clause expressing elaboration

An addition may describe an elaboration, and this is true also of wa-VprefS
clauses. What characterizes an elaboration with the VprefS gram is a dynamic
action. An example is the comment on Joram son of Ahab in (50).
(50) Pattern: wa-VprefS+lō-PrP+{wa-VprefS};

way-yaʿăśɛ hå̄ -raʿ bə-ʿēnē YHWH, raq lō ḵə-ʾå̄ ḇīw ū-kə-ʾimmō {way-yå̄ sar
ʾɛṯ maṣṣəḇaṯ hab-baʿal ʾăšɛr ʿå̄ śå̄ ʾå̄ ḇīw}
73 Other wa-VprefS clauses with result meaning are Gen. 3:10 (Onoun-Vsuff+{wa-VprefS
+[kī-NCl]}+{wa-VprefS}), 3:17 (kī-Vsuff+{wa-VprefS}), 7:20 (Vsuff+{wa-VprefS}),
12:19 (direct speech: ADV-Vsuff+{wa-VprefS}), 15:6 (‘Abram believed the LORD,
{and he credited it to him as righteousness}’ NIV), 19:11, 19:30, 20:6, 20:12
(NCl+{wa-VprefS}: ‘the daughter of my father, but not the daughter of my mother;
{and so she became my wife}’ JPS), 21:1, 26:28, 27:36, 29:17, 31:42, 34:7, 36:8, 39:14,
42:6, 44:20, 45:7, 49:17, Judg. 6:34, 20:5, 1 Sam. 1:13 (‘Hannah was speaking in her
heart; only her lips moved, and her voice was not heard. {Therefore Eli took her to be a
drunken woman}’ ESV), 1:27, Hab. 3:6, Prov. 1:24 (Vsuff+{wa-VprefS}: ‘Because I
have called {and you refused to listen}’ ESV). The we-lō-Vpref clause is discussed in
section B.2.2.1.

208 Bo Isaksson
‘He did evil in the sight of the LORD, but not to the same degree as his fa-
ther and mother. {He did remove the sacred pillar of Baal that his father
had made}.’ (2 Kgs 3:2; NET)
The first wa-VprefS clause in (51) partakes in the storyline. It is followed by a

clause that could be taken as a noun clause (‘though not like his father and his
mother’) with implied subject (‘he’) or as a Vsuff clause with implied verb
(‘he did not act like his father ...’). Be that as it may, the last wa-VprefS clause
elaborates on this negative fact (not like his father and mother) and adds fur-
ther details about this shade of positive behaviour: he removed the pillar of
Baal.
Elaborations with wa-VprefS are fairly common in direct speech, as in
(51), where two wa-VprefS clauses add additional facts about the rhetoric
question by Laban in dialogue with Jacob.
(51) Pattern: Opron-Vsuff+{wa-VprefS}+{wa-VprefS};

mɛ ʿå̄ śīṯå̄ {wat-tiḡnoḇ ʾɛṯ ləḇå̄ ḇī} {wat-tənaheḡ ʾɛṯ bənōṯay ki-šḇūyōṯ
ḥå̄ rɛḇ}
‘What have you done? {You deceived me}, {and you carried off my
daughters like captives in war}.’ (Gen. 31:26)
The initial rhetorical Vsuff clause is a question that amounts to an accusation:

‘you have done something unacceptable’. The two wa-VprefS clauses add
further information about the action referred to by the Vsuff clause and speci-
fy the unacceptable actions taken by Jacob. As usual, the wa-VprefS clauses
depend on the event time of the Vsuff clause and do not in themselves signal
anterior action.
Elaboration is a frequent function also in poetry. In a poetic context, the
dynamic action of the VprefS often elaborates on a pre-clause that expresses
an eternal truth, as in (52).74
(52) Pattern: NCl+{wa-VprefS};

kī l-YHWH məṣūqē ʾɛrɛṣ {way-yå̄ šɛṯ ʿălēhɛm tēḇēl}
74 Another conspicuous case of a NCl expressing a state and wa-VprefS describing a

dynamic elaboration is 2 Sam. 14:5 ʾăḇå̄ l ʾiššå̄ ʾalmå̄ nå̄ ʾå̄ ni {way-yå̄ måṯ} ‘Alas, I am a
widow; {my husband died}’ (TNK). Also 2 Sam. 14:6 with NCl+{wa-VprefS}+{wa-
VprefS}+{wa-VprefS}.

‘For the foundations of the earth are the LORD’s; {on them he set the
world}.’ (1 Sam. 2:8)
The noun clause in this passage of archaic poetry describes an eternal truth
and no action is implied in it. The wa-VprefS adds additional details on this
truth with a finite dynamic transitive verb. It is not necessary, as most transla-
tions do, to render the VprefS gram by an anterior.75 A perfective past refer-
ence, as in a narrative storyline, is enough, ‘he set the world on them’. This
ability of the wa-VprefS clause to characterize a generally valid truth is em-
ployed in the self-presentation of the Lord by his name, as in (53).
(53) Pattern: NCl+{wa-VprefS+[we-ADV-lō-Vsuff]};

ʾănī YHWH {wå̄ -ʾērå̄ ʾɛl ʾaḇrå̄ hå̄ m ʾɛl yiṣḥå̄ q wə-ʾɛl yaʿăqōḇ bə-ʾēl šaddå̄ y
[ū-šəmī YHWH lō nōḏaʿtī lå̄ hɛm]}
‘I am the LORD. {I appeared to Abraham, to Isaac, and to Jacob, as God
Almighty, [but by my name the LORD I did not make myself known to
them]}.’ (Exod. 6:3; ESV)
The repeated self-presentation of the Lord by his name, ‘I am YHWH’ might

seem tautological, but is repeatedly accompanied by actions that are intended
to reveal who he is. His name, his identity, is step by step made known by
actions. In this example, what it means that his name is YHWH is elaborated
by information about his actions in the past, when he revealed himself to the
patriarchs. By implication, his name will be more fully understood by the
actions that led to the exodus out of Egypt. The wa-VprefS clause “echoes the
first” noun clause ‘I am YHWH’, ”adding additional information about the
event or state described” (Dixon 2009, 27).76
75 Thus NIV, ‘For the foundations of the earth are the LORD’s; on them he has set the
world’.
76 The Vsuff clause is an addition to the VprefS clause, with a meaning of contrast (Dixon
2009, 28). Further examples of elaborating wa-VprefS clauses are Gen. 10:19, 12:16,
13:12 (‘but Lot lived in the cities of the valley {and set up his tent near Sodom},
HCSB), 19:19, 19:25, 26:29, 32:24; 34:13 and 37:18 (both without switch and implied
elaboration), 40:23, 41:51, 46:12 (elaboration in genealogy), 46:18, 50:13 (no switch),
Exod. 1:7, 1:17, 1:18 (Vsuff/wa-VprefS: ‘Why have you done this thing, {letting the
boys live}?’ TNK), 9:23, 16:17, 16:20, 18:4, 24:11, Judg. 2:11, 2:17, 2 Sam. 22:39
(VprefL/wa-VprefS: ‘they cannot get up; {they fall at my feet}’ NET), 1 Kgs 8:24,
Amos 2:9, Job 7:6, Ps. 18:33 (PA/wa-VprefS: ‘It is God who arms me with strength
{and makes my way perfect}’ NIB), 29:9.

210 Bo Isaksson
Wa-VprefS clauses are often elaborations of resultative pre-clauses in po-

etry. Example (54) shows a wa-VprefS having a general present meaning in
(possibly) archaic poetry:
(54) Pattern: Snoun-Vsuff+{wa-VprefS};

YHWH lam-mabbūl yå̄ šå̄ ḇ {way-yēšɛḇ YHWH mɛlɛḵ lə-ʿōlå̄ m}
‘The LORD sits enthroned over the flood; {the LORD sits enthroned as
king forever}.’ (Ps. 29:10; ESV)
The context is eternal truths, so it is natural to follow most translators here and
interpret the Vsuff gram, not as an anterior, but with its prototypical resulta-
tive meaning ‘is enthroned’ (which ESV renders ‘sits enthroned’). The wa-
VprefS expresses an elaboration of this eternal fact (emphasized by the PrP lə-
ʿōlå̄ m ‘for ever’). The addition cannot, in this instance, be interpreted as a
temporal succession, and a nuance of result is not a natural reading.77
A.2.2.5 The wa-VprefS clause expressing same-event addition

The case of a same-event addition, that is, “two clauses describe different
aspects of a single event” is often hard to distinguish from elaboration (Dixon
2009, 27). A possible example is (55).
(55) Pattern: Vsuff+{wa-VprefS};

kå̄ lå̄ ʿå̄ nå̄ n {way-yēlaḵ}
‘As the cloud fades {and vanishes}.’ (Job 7:9; ESV)
The zero-gram VprefS is well equipped to express also generally valid ac-
tions, corresponding to an English general present. The wa allomorph of the
conjunction indicates that the VprefS is indicative, it describes a real action
(though not in a storyline and not with past time reference). There is no tem-
poral difference between the clauses, and what is expressed by the switch is
an addition, in which the pre-clause and the addition clause are just different
aspects of the same event.78
77 The mɛlɛḵ ‘as king’ is an adverbial expression, which can be interpreted as an infinite
predication of its own (and thus a clause), ‘being a king’, Isaksson (2009, 39).
78 Some other possible examples of same-event addition are Gen. 34:29 (Vsuff/wa-VprefS
‘All their wealth, all their little ones and their wives, all that was in the houses, they cap-
tured {and plundered}’ ESV), 35:3, 39:18.

A.2.3 The syndetic we-VprefS clause and the we-...ʾal-VprefS clause

The we-VprefS clause is distinguished in the Masoretic reading tradition as
being modal in contradistinction to wa-VprefS which is indicative. Both are
clause-initial. With the negation ʾal this type of clause is discrete (as to the
‘short’ nature of the Vpref) and there is no restriction of word order.
The syntactic construction we-...VprefS is an addition with a jussive mean-
ing, ‘and thereby he should ...’, but it seems that this type of syntagm has
received a more restricted meaning. Such clauses practically always describe a
purpose and relates to a pre-clause that may be an imperative, noun clause,
cohortative, long prefix clause, we-Vsuff clause, a jussive VprefS clause, and
even an indicative narrative VprefS clause.79 Even when there is no gram-
switch, as in a clause linking of the type Ø-VprefS/we-VprefS, the we-VprefS
clause usually has an unmistakable flavour of purpose. The purpose meaning
of we-VprefS clauses is recognized by standard grammars, and there is no
need of an extensive exposition of all cases (J-M § 116; Nyberg 1972, § 87d).
A simple example is (56).
(56) Pattern: IMP+[we-VprefS];

qirʾɛn lō [wə-yōḵal lå̄ ḵɛm]
‘Call him, [that he may eat bread].’ (Exod. 2:20; ESV)
Purpose clauses often conform to the symmetry that third person clauses have
a we-VprefS pattern, and second person clauses a we-IMP pattern (thus J-M §
116). This is illustrated in (57).
(57) Pattern: IMP+we-IMP+[we-VprefS!];

śå̄ nā ḥaṭṭå̄ ṯī ʾaḵ hap-paʿam wə-haʿtīrū l-YHWH ʾɛ̆lōhēḵɛm [wə-yå̄ ser mē-
ʿå̄ lay raq ʾɛṯ ham-må̄ wɛṯ haz-zɛ]
‘Now therefore, forgive my sin, please, only this once, and plead with the
LORD your God [only to remove this death from me].’ (Exod. 10:17;
ESV)
The example is instructive, in that the first addition (we-IMP) is not as dis-
tinctly purposive as is the we-VprefS clause. The addition of the imperative,
‘and plead thereby with the Lord’, can be interpreted as a purpose clause ‘for-
79 Examples of pre-clauses (with following we-VprefS): VprefL (as a positive command)

Gen. 42:20; VprefA Gen. 34:23, Exod. 8:4; we-Vsuff (we-qatal) Exod. 2:7; PA (partici-
ple clause) Exod. 14:17.

212 Bo Isaksson
give my sin [so that you plead with the Lord]’ (J-M § 116f), but also with a
looser connection to the first imperative as an accompanying imperative.80
The pre-clause may also be a VprefS clause. It is rare, though, that the pre-
clause is an indicative perfective. One of these rare cases is from archaic poet-
ry (58).
(58) Pattern: Ø-VprefS+[we-VprefS];

təḇīʾēmō [wə-ṯiṭṭå̄ ʿēmō bə-har naḥălå̄ ṯəḵå̄ ]
‘You brought them in [to plant them on the mountain that is your own.’
(Exod. 15:17)
The passage is an account of the exodus, and in such an archaic context a

fronted ‘narrative’ perfective VprefS without the conjunction wa is a fairly
common feature. The verbs are morphologically non-discrete, but they are
clause-initial, which is a signal of the short prefix verb, and the following we-
Vpref clause fits well in an interpretation as a purpose clause against many
translations.81
There are several examples of a jussive pre-clause with a succeeding pur-
posive we-VprefS clause, which means that the linking lacks a gram-switch,
only the added conjunction signals the semantic relation. This is shown in
(59).
(59) Pattern: Ø-VprefS+[we-VprefS]+[we-ʾal-VprefS];

tiḵbaḏ hā-ʿăḇōḏå̄ ʿal hå̄ -ʾănå̄ šīm [wə-yaʿăśū ḇå̄ h] [wə-ʾal-yišʿū bə-ḏiḇrē
šå̄ qɛr]
‘Make the work harder for the people [so that they keep working] [and pay
no attention to lies].’ (Exod. 5:9)
The example shows an initial main line jussive VprefS in the second person
(the ‘pre-clause’). It is followed by one positive purpose clause and another
80 Other examples of we-VprefS purpose clauses are, Gen. 27:29, 30:3 (‘Sleep with her [so
that she can bear children for me]’, NIV), 31:37, 38:24, 42:16, Exod. 4:23 (and many
similar), 9:22, 10:21. In this first person the syndetic cohortative (we-VprefA) is the
common alternative although in many cases the form of the verb is often not discrete
(could be a VprefS or VprefL). Only our acquaintance with the discrete cases make us
conclude that the cohortative is intended when first person is involved. Two morpholog-
ically discrete examples are Gen. 19:5 (IMP+[wə-nēḏəʿå̄ ]) and 23:4 (IMP +[wə-
ʾɛqbərå̄ ]).
81 Such as NAB, ‘You brought them in, you planted them on the mountain that is your
own’.

purpose clause negated by the modal negator ʾal. It is reasonable to conclude

that the two purpose clauses have equal status.82 As can be seen in (59), pur-
pose clauses may be negated by the modal negation ʾal, which is illustrated
also in (60).
(60) Pattern: IMP+[we-ʾal-VprefS];

riḡzū [wə-ʾal-tɛḥɛ̆ṭå̄ ʾū]
‘Tremble with fear [that you do not sin]!’ (Ps. 4:5)83
Since ʾal only negates jussive VprefS verbs, word order in such clauses is free
and a topicalized constituent may be placed before the negated verb, as in the
legal language of (61), where two object nouns are placed in topicalized posi-
tion before the verb and its negation:
(61) Pattern: Ø-PrP-VprefL+[we-Onoun-we-Onoun-ʾal-VprefS];

mid-dəḇar šɛqɛr tirḥå̄ q [wə-nå̄ qī wə-ṣaddīq ʾal tahăroḡ]
‘Have nothing to do with a false charge [not to put an innocent or honest
person to death].’ (Exod. 23:7)
A.3 Summary
In affirmative clauses the VprefS gram is restricted to clause-initial position in
order not to be confused with the VprefL gram. In negative clauses (negation
ʾal) there is no such restriction of word order for the VprefS gram.
The VprefS gram is used in both main line and in non-main clause linking.
VprefS is zero-marked as to tense and mood. Its most frequent uses in a main
line are as a perfective verb in past time contexts, and as a modal (jussive)
verb. But it is also used to express a general (timeless) present.
The wa allomorph signals that the VprefS is indicative (wa-VprefS),
whereas the we allomorph signals the verb to be modal (usually with a purpos-
ive nuance). In prose (with very few exceptions) an asyndetic VprefS signals
that the verb has jussive meaning (Ø-VprefS), whereas early poetry shows
examples of indicative (usually narrative perfective) Ø-VprefS clauses.
82 Other examples of VprefS/we-VprefS clause linkings with purpose nuance: Gen. 1:6
(‘Let there be a vault between the waters [to separate water from water]’ NIV), 1:9, Ex-
od. 5:21.
83 Or, as NJB, ‘Be careful not to sin’.

214 Bo Isaksson
As a non-main clause wa-VprefS often expresses elaboration, temporal

succession, and result. Sometimes, however, it is just an addition, an accom-
panying action in relation to the pre-clause.
B. Clause combining and the VprefL gram

As already mentioned, the partial morphological coalescence of the short
prefix gram and the long prefix gram led to radical restrictions in the word
order of VprefL clauses. While the yaqtulu gram in Old Arabic retained a
greater freedom, which allowed for a fronted position in the clause, Proto-
Hebrew (preceding the Biblical Hebrew texts) developed word order strate-
gies to avoid confusion between the original VprefS and the original VprefL.84
A clause-initial prefix verb came to be perceived as short (indicative or jus-
sive), and this position was consequently avoided when a long prefix verb was
to be used in affirmative clauses. In such clauses a VprefL form was generally
non-initial.85
In negative clauses on the other hand, there was generally no danger of
confusion since ʾal was exclusively used before short jussive VprefS, and lō
came to be the normal negation before VprefL (Tropper 1998, 178). A nega-
tion lō before a prefix verb was usually a signal of a VprefL verb in Biblical
Hebrew, in sharp contrast to the free use of lā in Amarna Canaanite before
both VprefS and VprefL verbs (Rainey 1996, 211-212). In poetry the VprefL
gram could also be negated by bal.
84 With the term ‘Proto-Hebrew’ I suggest a transition stage when short final vowels
(including the case vowels of the noun) were being dropped and resulting syntactical
changes took place. “These developments were essentially completed or at least in an
advanced stage when the Northwest Semitic languages of Iron Age Syria-Palestine ap-
peared on the stage of history after ca. 1000 BCE” (Gzella 2012, 6).
85 Relative clauses are unaffected by the word order rule, as in Ps. 34:9 ʾašrē hag-gɛḇɛr
<yɛḥɛ̆sɛ bō> ‘blessed is the one who takes refuge in him’ (...NP-REL<Ø-VprefL>;
NIV). In some cases the syntactical status of a VN phrase may cause bewilderment, as
in asyndetic Exod. 30:20 [Ø-be-VN]+Ø-VprefL, which could be argued to represent an
exception. But it should be observed that ‘embedding’ is a scalar concept. The gradual
integration of the desententialized (VN) clause in the VprefL clause (Lehmann 1988,
183) in such cases makes the linguistic instinct perceive the VprefL as non-initial (the
VN phrase being positioned before it as an element in the same clause). Similar exam-
ples are Exod. 33:8,9.

B.1 VprefL clauses encoding a main line

It is well known that the VprefL clause in the main line often codes future
time reference or habitual action. Sometimes, but not as often as could be
expected, it is also used with its prototypical imperfective meaning of a pro-
gressive present (Bybee et al. 1994, 126).
Future time reference is probably the most common meaning of a main
line VprefL clause. One example is (62).
(62) Pattern: kī-VprefL!+we-PrP-VprefL;

kī yå̄ ḏīn YHWH ʿammō wə-ʿal ʿăḇå̄ ḏå̄ w yiṯnɛḥå̄ m
‘The LORD will vindicate his people and relent concerning his servants.’
(Deut. 32:36; NIV)
In this passage the two clauses form a main line of VprefL clauses with future
reference. None of the verbs is clause-initial. It is reasonable to assume that
the addition indicated by the conjunction we, in this case (without a gram-
switch), means a concatenation of clauses of equal status. (63) is an illustra-
tion of the typical morphology of the VprefL gram: when a discretely long
form is available (in this case yå̄ ḏīn in contradistinction to VprefS yå̄ ḏen) then
this long form is also used.86
The rule of non-initial VprefL is sometimes virtually violated by ellipsis,
as in (63).
(63) Pattern: ʾāz-VprefL! + we-lō-VprefL + ()-VprefL! + we-lō-VprefL!;

ʾå̄ z yiqrå̄ ʾūn-nī wə-lō ʾɛʿɛ̆nɛ, yəšaḥărūn-nī wə-lō yimṣå̄ ʾūn-nī
‘Then they will call to me but I will not answer; they will look for me but
will not find me’ (Prov. 1:28; NIV)
The four clauses in (63) form two pairs of the type ʾāz-VprefL!+we-lō-
VprefL. In the second pair, however, the ʾāz is understood (here indicated by
parentheses). Three of the four verbs are discretely long and have future refer-
ence.87
86 The exceptions are usually in the opposite direction. Surprisingly often long forms are
used as short, and then always in initial position.
87 Three of the verbs in (64) exhibit a ‘nun paragogicum’, which is commonly supposed to
originate from the long prefix form in Central Semitic, cf. the Arabic ‘indicative’ plural
yaqtulūna (Rainey 1986, 7). As for the III infirmae verb ʾɛʿɛ̆nɛ, it should probably not be
regarded as discrete (Tropper 1998, 165; J-M §§ 79m, 114g note 3). Other examples
with future reference (and discretely long forms): Gen. 49:10 (lō yå̄ sūr šeḇɛṭ mīhūḏå̄

216 Bo Isaksson
An expression of future easily receives a nuance of obligation and indirect

command. This is shown in (64).
(64) Pattern: ADV-VprefL!;

kō tōmərūn l-aḏōnī lə-ʿēśå̄ w
‘Thus you shall say to my lord Esau.’ (Gen. 32:5; ESV)
The verb is not fronted and it has a ‘nun paragogicum’, both signals of a long
prefix verb. It is a command by Jacob directed at his servants.
A progressive present VprefL is perceivable in the two clauses of (65),
which displays two asyndetic clauses of equal status in direct speech.
(65) Pattern: Ø-ADV-VprefL!+Ø-ADV-VprefL!;

ma-t-tərīḇūn ʿimmå̄ dī, ma-t-tənassūn ʾɛṯ YHWH
‘Why do you quarrel with me? Why do you put the LORD to the test?’
(Exod. 17:2; NIV)
Both verbs show a ‘nun paragogicum’ and the progressive present meaning.
This meaning of the VprefL gram is sometimes encountered in SBH, although
the active participle increasingly takes over this function.88
As can be expected, instead of a progressive present meaning the VprefL is
often used as a general present (Bybee et al. 1994, 127, 153). This is shown in
(66).
(66) Pattern: VOC, Spron-VprefL!+Ø-Spron-VprefL;

YHWH, mī yå̄ ḡūr bə-ʾåhå̆ lɛḵå̄ , mī yiškon bə-har qåḏšɛḵå̄
‘LORD, who may dwell in your sacred tent? Who may live on your holy
mountain?’ (Ps. 15:1; NIV)
The two clauses in (66) have equal status, and in the first the VprefL is dis-
crete. The (timeless) general present receives a nuance of permission: ‘who
may ...’. The verb is non-initial in both clauses.89
‘The scepter will not depart from Judah’), Prov. 2:6, Ps. 16:4 (with purpose nuance in
main line).
88 A progressive present VprefL, in a usage that comes close to a performative function, is
found in Deut. 32:40, kī ʾɛśśå̄ ʾɛl šå̄ mayim yå̄ ḏī ‘I lift my hand to heaven and solemnly
swear’ (NIV).
89 A general present is perceivable also in Ps. 9:9 (wə-hū yišpoṭ tēḇēl bə-ṣɛḏɛq, yåḏīn
ləʾummīm bə-mēšå̄ rīm ‘and he judges the world with righteousness; he judges the peo-
ples with uprightness’, ESV), where the second verb is discretely long in an elliptic

The VprefL is not confined to one (or two) specific temporal reference(s),
which is shown in its use as past habitual (67).
(67) Pattern: ADV-VprefL;

kēn yaʿăśɛ šå̄ nå̄ ḇə-šå̄ nå̄
‘He did the same every year.’ (1 Sam. 1:7; CJB)
The verb is not discrete here, but it is non-initial and this together with the
meaning makes it reasonable to conclude that the long form is intended. Ha-
bituality is signaled by the adverbial phrase šå̄ nå̄ ḇə-šå̄ nå̄ ‘year by year’.90
B.2 The VprefL clause encoding a non-main clause

When the VprefL gram is used in a non-main clause its prototypical imperfec-
tivity often results in a dynamic description of an attendant or accompanying
action (“background situations”, Bybee et al. 1994, 126).
The function of attendant circumstantial clause is apparent in an asyndetic
VprefL clause, but also an addition signaled by we and a VprefL clause often
results in a meaning close to a circumstantial expression, although in a more
independent clause.
In narrative prose, the circumstantial VprefL clause is rather uncommon, a
fact that contrasts with classical Arabic (Isaksson 2009, 84-89, 91-92). Two
contrasting VprefS/VprefL are only rarely found when the VprefS is a story-
line clause. An attendant action is more often coded by other types of clauses,
such as an active participle.
B.2.1 The asyndetic VprefL clause
B.2.1.1 The Ø-NEG-VprefL clause

A Ø-NEG-VprefL clause in non-main position codes an attendant circumstan-
tial action without a topicalized element. The emphasis is on the absence of
the action itself. The negation is lō, in poetry sometimes bal. A negated non-
clause (understands an initial subject hū from the first clause). Another case of general
present is Gen. 2:24 ʿal kēn yaʿăzåḇ ʾīš ʾɛṯ ʾå̄ ḇīw wə ʾɛṯ ʾimmō ‘That is why a man
leaves his father and mother and is united to his wife’ (NIV) with a slight nuance of ha-
bituality. Also Gen. 43:32. The rare instances with ʾå̄ z and VprefL represent a special
case and should be discussed separately (possibly a case of diegetic present).
90 Other examples of habitual VprefL clauses are Gen. 32:33 (or general present), Exod.
4:11, 34:34, Job 1:5.

218 Bo Isaksson
main asyndetic VprefL clause with negation lō in prose is used to describe the
old aged patriarch Israel in (68).
(68) Pattern: we-Snoun-Vsuff+[Ø-lō-VprefL];

wə-ʿēnē yiśrå̄ ʾēl kāḇəḏū miz-zoqɛn [lōʾ yūḵal li-rʾōṯ]
‘Now the eyes of Israel were dim with age, [so that he could not see].’
(Gen. 48:10; ESV)
In (68) the circumstantial meaning of the VprefL clause (‘in that’) receives a
nuance of result (‘so that’). The general attendant circumstance meaning of
the clause, ‘not being able to see’, is understood as a resulting state (Isaksson
2009, 7-10).
An example with the negation bal in poetry is (69).
(69) Pattern: Ø-Vsuff+[Ø-bal-VprefL];

ṣå̄ rap̄ tanī [bal timṣå̄ ]
‘You have tried me [without finding]’ (Ps. 17:3)
Example (69) shows an asyndetic VprefL clause expressing an attendant cir-

cumstance. The poetic negation bal with a VprefL gram introduces a circum-
stantial clause ‘not finding’ which is better rendered ‘without finding (any-
thing)’.91
B.2.1.2 The Ø-X-VprefL clause

When an element ‘X’ (other than a negation) is positioned before the verb,
this element becomes topicalized. 92 Examples of X-VprefL clauses being
digressions from a narrative storyline of wa-VprefS in prose are unusual and
show quite specific constructions. One example is (70) with repetitive clauses.
91 The CSB (2009) renders the negated VprefL as if the clause was an addition with per-
fective aspect (though neutralized by the negation): ‘You have tried me [and found noth-
ing evil]’. But the syntax, an asyndetic VprefL clause, is that of an attendant circum-
stance. Other cases of an asyndetic switch Vsuff/lō-VprefL are: Exod. 10:29 (ADV-
Vsuff+[Ø-lō-VprefL!]), Amos 8:2 (Ø-Vsuff+[Ø-lō-VprefL!] with reason nuance), Ps.
5:5-7 (kī-lō-Vsuff+[Ø-lō-VprefL]+[Ø-lō-VprefL] with two reason clauses).
92 The clause is still an attendant circumstantial clause, but that does not mean that it
always codes an ongoing action concomitant with the preceding clause. It can also code
a rule that is valid at the same time, as in Exod. 35:2 Ø-ADV-VprefL+we-PrP-VprefL +
[Ø-Snoun-VprefL] ‘Six days work shall be done, but on the seventh day you shall have
a Sabbath of solemn rest, holy to the LORD. Whoever does any work on it shall be put
to death.’ (ESV).

(70) Pattern: wa-VprefS + [Ø-Snoun-VprefL] + [we-Snoun-VprefL];

wa-yhī qōl haš-šōp̄ å̄ r hōlēḵ wə-ḥå̄ zēq məʾōḏ [mōšɛ yəḏabber] [wə-hå̄ -
ʾɛ̆lōhīm yaʿănɛnnū ḇə-qōl]
‘The blast of the shofar grew louder and louder, [while Moses was speak-
ing and God was answering him with thunder].’ (Exod. 19:19; NAB)
The storyline wayhī is succeeded by two repetitive participles hōlēḵ wə-ḥå̄ zēq
(J-M § 123s), and then follow two VprefL clauses that describe what was
going on during the repeated blasts of the shofar. The VprefL clauses express
two dynamic actions that we infer interact with each other: God and Moses
speak to each other in a dialogue during a period of time. The first VprefL
clause is asyndetic and the second is added to the first with a we. The two
VprefL clauses have seemingly the same status. The signal of the linking is
the asyndesis and the switch to a long prefix verb gram. The two subject
nouns Moses and God are topicalized (placed before the verb).93
If a narrative storyline is interrupted by a VprefL clause, an explicit adverb
is usually needed in the construction, as with ṭɛrɛm ‘not yet’ in (71), and ʿal
kēn in (72).
(71) Pattern: wa-VprefS+[Ø-ṭɛrɛm-VprefL];

way-yiśśå̄ hå̄ -ʿå̄ m ʾɛṯ bəṣēqō [ṭɛrɛm yɛḥmå̄ ṣ]
‘So the people carried off their dough [still unleavened].’ (Exod. 12:34;
NJB)
The event time of the VprefL clause is that of the storyline. In this non-main
position, the imperfective VprefL takes over the event time of the main clause
and describes a concomitant state ‘not yet being leavened’. The adverb ṭɛrɛm
‘not yet’ emphasizes the simultaneity of the two actions.
When a comment with habitual meaning is inserted into the storyline a
VprefL clause may be utilized, with an explanatory adverb in topicalized
position.
93 A similar example is 1 Sam. 13:17-18 (wa-VprefS+[Ø-Snoun-VprefL+we-Snoun-

VprefL + we-Snoun-VprefL]) where the “difficult case” Ø-Snoun-VprefL is discussed
by Joosten (1999, 24: “prospective”) and Driver (1892 § 163) as a main clause (like
most other clauses). Also Exod. 19:19 (wa-VprefS+[Ø-Snoun-VprefL] + [we-Snoun-
VprefLN]), and 1 Kgs 3:4.

220 Bo Isaksson
(72) Pattern: (wa-VprefS +) [Ø-ADV-VprefL + [we-Vsuff] + [we-Vsuff]];

[ʿal kēn yaʿăzåḇ ʾīš ʾɛṯ ʾå̄ ḇīw wə ʾɛṯ ʾimmō [wə-ḏå̄ ḇaq bə-ʾištō] [wə-hå̄ yū
lə-ḇå̄ śå̄ r ʾɛḥå̄ ḏ]]
‘[That is why a man leaves his father and mother [and is united to his
wife] [and they become one flesh].]’ (Gen. 2:24; NIV)
The digression from the storyline in (72), coded by an asyndetic VprefL

clause (and subsequent accompanying actions with we-Vsuff clauses),94 has
the function of a narrator’s comment. The comment is beside the main line, it
is not part of the narrative scene, but at the same time it is directly relevant to
what has been related (the creation of the woman).95
Non-main asyndetic VprefL clauses in SBH are mostly found in direct
speech or poetry. In (73) a circumstantial clause describes a reason in Sara’s
direct speech.
(73) Pattern: Onoun-Vsuff+[Ø-Snoun-VprefL];

ṣəḥōq ʿå̄ śå̄ lī ʾɛ̆lōhīm [kål haš-šōmēăʿ yiṣḥaq lī]
‘God has brought laughter for me, [in that everyone who hears will laugh
at me].’ (Gen. 21:6)96
Abraham’s wife is pregnant at an old age and she feels embarrassed for what
people will think. She feels that God has brought laughter on her, and in a
circumstantial clause she gives the reason for this statement: since everyone
will laugh at me. Since the circumstantial clause expresses her expectation of
people’s laughter, a future time rendering is natural. Sara’s expectation is a
present state, but what she expects lies in the future.97
94 The we-Vsuff clauses will be discussed in section C.2. See also Isaksson (forthcoming
a).
95 In archaic poetry, where a ‘storyline’ is still sometimes coded by asyndetic VprefS
clauses, a circumstantial VprefL is sometimes found to code a digression, as in Ps. 18:21
Ø-VprefS + [Ø-PrP-VprefL!] ‘the LORD dealt with me according to my righteousness
[rewarding me according to the cleanness of my hands]’, where the circumstantial
clause amounts to a same-event addition (different aspect of the same action).
96 The LXX marks the reason nuance by a γὰρ: ὃς γὰρ ἂν ἀκούσῃ συγχαρεῖταί μοι.
97 A poetic example of a circumstantial switch from a Vsuff clause is found in Ps. 6:10
(Vsuff+[Ø-Snoun-Onoun-VprefL]). It is a same-event addition of the type Mary came
first in her race, [winning the prize] (Dixon 2009, 27).

An asyndetic VprefL clause may encode an attendant circumstance in rela-

tion to practically all types of main clauses. In (74) the main line is a PA
clause.
(74) Pattern: we-Spron-PA+[Ø-Snoun-VprefL];

wa-ʾănī hinnī mēḇī ʾɛṯ ham-mabbūl mayim ʿal hå̄ -ʾå̄ rɛṣ lə-šaḥeṯ kål bå̄ śå̄ r
ʾăšɛr bō rūaḥ ḥayyīm mit-taḥaṯ haš-šå̄ må̄ yim [kål ʾăšɛr bå̄ -ʾå̄ rɛṣ yigwāʿ]
‘I am going to bring floodwaters on the earth to destroy all life under the
heavens, every creature that has the breath of life in it [whereby everything
on earth will perish].’ (Gen. 6:17)
The VprefL clause in this example has the same event time as the preceding
PA clause, namely a point in the future.98
A VprefL clause often constitutes an attendant circumstantial clause in re-
lation to a preceding imperative, jussive (VprefS) or we-Vsuff clause. In (75)
the main line is a jussive clause.
(75) Pattern: ʾal-VprefS+[Ø-Snoun-VprefL+[we-Vsuff]];

ʾal yippol lēḇ ʾå̄ ḏå̄ m ʿå̄ lå̄ w [ʿaḇdəḵå̄ yēlēḵ [wə-nilḥam ʿim hap-pəlištī haz-
zɛ]]
‘Let no man’s heart fail because of him; [for your servant will go [and
fight with this Philistine]].’ (1 Sam. 17:32)
The asyndetic VprefL clause with fronted subject is a circumstantial clause

saying, ‘in that your servant is going’. In this case the VprefL does not take
over an event time from the main line (which, here, is a jussive), instead it
forms a reason for the appeal in the jussive (main line) clause.99
An imperative clause with a circumstantial VprefL clause is found in (76).
98 Other cases of a switch PA/VprefL: Exod. 13:22, 1 Sam. 2:8 (with two VprefL clauses
PA + [Ø-PrP-VprefL!] + [we-Onoun-VprefL]), Num. 24:16 (PA + [Ø-Onoun-VprefL! +
PA+PP]: ‘Word of him who hears God's speech, who obtains knowledge from the Most
High [in that he beholds visions from the Almighty [prostrate], [but with eyes un-
veiled]]’). Examples of a switch NCl/VprefL: Exod. 16:25 (with result nuance), Gen
49:27 (with two VprefL clauses: NCl+REL+[Ø-PrP-VprefL]+[we-PrP-VprefL]), Ps.
5:10 (NCl+[Ø-Onoun-VprefL]).
99 The we-Vsuff clause expresses an affirmative accompanying action (without topicalized
element) in relation to the VprefL clause, taking over its event time: ‘and (thereby) fight
with this Philistine’, see section C.2.

222 Bo Isaksson
(76) Pattern: IMP+[Ø-ADV-VprefL-Onoun];

ʿăśē ləḵå̄ tēḇat ʿăṣē gop̄ ɛr [qinnīm taʿăśɛ ʾɛṯ hat-tēḇå̄ ]
‘Make yourself an ark of gopher wood [making the ark with rooms].’
(Gen. 6:14)
The VprefL clause in (76) has the same event time as the imperative (that is,
future), but does not describe a separate action. The asyndesis signals con-
comitant action and codes one of the actions that should be taken while mak-
ing the whole ark: ‘making the ark with rooms’.100
As is well-known, we-Vsuff (‘we-qatal’) clauses are extremely frequent in
instructions, and in such contexts they must usually be analysed as forming
the instructional main line. It is less well-known that VprefL clauses often
form circumstantial clauses in relation to the backbone chain of we-Vsuff
clauses. An example is (77).
(77) Pattern: we-Vsuff+[Ø-PrP-VprefL];

wə-nå̄ ṯattī ʾɛṯ haṭ-ṭabbå̄ ʿōṯ ʿal ʾarbaʿ hap-pēʾōṯ ʾăšɛr lə-ʾarbaʿ raḡlå̄ w [lə-
ʿummaṯ ham-misgɛrɛṯ tihyɛnå̄ haṭ-ṭabbå̄ ʿōṯ lə-ḇātīm lə-ḇaddīm lā-śēṯ ʾɛṯ
haš-šulḥå̄ n]
‘you fasten the rings to the four corners at its four legs [in that the rings
that hold the poles used for carrying the table shall be close to the rim].’
(Exod. 25:27)
VprefL clauses such as in (77) are usually translated as if they were part of the
back bone of instruction, but switches of this kind create an information struc-
ture even in an instructional text (Isaksson 2014a, 35-37).
We-Vsuff clauses may, just as well, describe repetitive actions in a narra-
tive context. In such instances, a VprefL clause may also function as a cir-
cumstantial. In (78) two circumstantial clauses are involved in the clause
combining.
(78) Pattern: we-Vsuff+[Ø-Onoun-VprefL]+[we-Onoun-VprefL];

wə-šå̄ p̄ əṭū ʾɛṯ hå̄ -ʿå̄ m bə-ḵål ʿēṯ [ʾɛṯ had-då̄ ḇå̄ r haq-qå̄ šɛ yəḇīʾūn ʾɛl Mōšɛ]
[wə-ḵål had-då̄ ḇå̄ r haq-qå̄ ṭōn yišpūṭū hēm]
100 I analyse the qinnīm ‘rooms’ as adverbial, ‘with rooms’. Other examples of
IMP/VprefL: Exod. 16:16, 19:12, Job 7:7 (IMP+[Ø-lō-VprefL!] with a nuance of rea-
son); possibly also Ps. 17:8 (IMP + [Ø-PrP-VprefL]).

‘and they rendered decisions for the people at all times, [and hereby they
referred to Moses the more difficult cases], [while they themselves settled
all the minor cases].’ (Exod. 18:26)
In (78) a habitual action is described by a we-Vsuff clause. Two attendant

circumstantial actions, likewise habitual, are coded by two VprefL clauses of
which the first is asyndetic. It is a description of a procedure with two actions
going on simultaneously, the difficult cases and the minor ones.101 In this kind
of concatenation of two circumstantial clauses, the asyndesis of the first
VprefL clause is the signal of simultaneity (with the action in the we-Vsuff
clause) and the we in front of the second VprefL clause signals that the two
have equal status (both are circumstantial in relation to the main line we-Vsuff
clause).
B.2.2 The syndetic VprefL clause

Clauses of the types we-NEG-VprefL or we-X-VprefL connect as additions to
a previous clause. The simple *we-VprefL clause became unacceptable in
Proto-Hebrew because of confusion with the (purposive) clause-initial we-
VprefS clauses (A.2.3). Instead the emerging we-Vsuff clause (C.2), also
affirmative and with no topicalized element, replaced the old *we-VprefL
clause.
This is the reason why the innovative we-Vsuff clauses gradually ‘took
over’ meanings typical of the VprefL gram. This is also the reason why the
innovative we-Vsuff clauses with new meanings ‘could not be negated’ nor
‘could be split up’ by a topicalized element ‘X’ between the we and the Vsuff.
There was no need to. Such functions (negative meaning, topicalized constitu-
ent) were retained by the old VprefL clause types.102
101 Other instances of we-Vsuff/Ø-X-VprefL clause combining: Gen. 17:20 (we-Vsuff +

[Ø-Onoun-VprefL!]), Exod. 8:7 (NCl + [we-Vsuff + [Ø-ADV-VprefL]), 12:4 (we-
Vsuff + [Ø-PrP-VprefL]), 25:28 (we-Vsuff + [Ø-Onoun-VprefL!]), 25:29 (we-Vsuff +
[Ø-Onoun-VprefL!]), 25:31 (we-Vsuff+[Ø-Onoun-VprefL!]+[Ø-Onoun-VprefL]).
102 This is also the reason why the we-Vsuff syntagm with innovative meanings similar to
the VprefL gram has been perceived by the grammars to constitute an unbreakable
unit, a conjugation, a ‘gram’ of its own (Isaksson, forthcoming a).

224 Bo Isaksson
The verb in VprefL clauses is strictly non-initial.103 The various functions

of a syndetic VprefL clause derive from two basic facts. As an accompanying
action (we marks ‘addition’) it is connected to a previous action, state, wish or
command. And as an imperfective clause (because of its VprefL predicate) it
may describe many possible nuances, such as futurity, habituality, and contin-
uous action. And take on apodictic meanings (‘you shall’ > ‘you must’).
B.2.2.1 The syndetic NEG-VprefL clause

The dominating negation before a VprefL verb is lō. This negation is normally
a signal that the prefix verb is long. When the VprefS/VprefL distinction was
threatened by collapse and the affirmative we-Vsuff syntagm in many addi-
tion clauses superseded the VprefL gram, the negative addition clauses of the
type we-lō-VprefL were retained because they were unambiguous. In this
sense, we-lō-VprefL clauses, as also the we-X-VprefL clauses, represent a
retention, an ‘archaism’.
If the explanation above is right, and the we-Vsuff type of clause in af-
firmative clauses without topicalized element replaced the *we-VprefL clause
type, then we should expect to encounter cases where we-Vsuff clauses inter-
act with the we-NEG-VprefL clauses which are of equal status. This is, in
fact, a very frequent feature in Biblical Hebrew texts. One example is (79).
(79) Pattern: we-kī-VprefL!+[we-Vsuff]+[we-lō-VprefL!]+[we-Vsuff];

wə-ḵī yərīḇūn ʾănå̄ šīm [wə-hikkå̄ ʾɛṯ rēʿēhū bə-ʾɛḇɛn ʾō ḇə-ʾɛḡrōp̄ ] [wə-lō-
yå̄ mūṯ] [wə-nå̄ p̄ al lə-miškå̄ ḇ]
‘If men quarrel [and one hits the other with a stone or with his fist] [and he
does not die] [but is confined to bed],’ (Exod. 21:18; NIB)
The whole verse 18 in (79) is a protasis in legal discourse (the apodosis does
not concern us here). This protasis is built by four clauses of which the first is
the main clause introduced by the particle kī, which here functions as a tem-
poral conditional marker (conjunction).104 What concerns us in this legal prot-
asis is the series of accompanying actions (legal circumstances) that are coded
by one we-Vsuff clause, one we-lō-VprefL clause, and again one we-Vsuff
103 This syntactic feature is the decisive mark of a VprefL verb. Usually, the morphology
agrees with the syntactic mark, but in the rare cases when a morphologically long pre-
fix verb is clause-initial, it is intended as short, as in Ps. 18:40.
104 The we before the kī probably connects this legal case with previous legal cases stated
in Exod. 21.

clause. Legal cases have to be very specific and the main clause in the prota-
sis, ‘If men quarrel’ must be qualified by three further circumstances, of
which one is negative. These accompanying actions all have equal status. The
we-lō-VprefL represents the old accompanying action construction, the we-
Vsuff represent the innovation when *we-VprefL could no longer be used.
Another example, with a we-Vsuff clause and a we-lō-VprefL clause of
equal status, is (80).
(80) Pattern: Ø-Snoun-VprefL + ʾim-lō-Vsuff + [we-Vsuff] + [we-lō-VprefL];

šəḇūʿaṯ YHWH tihyɛ bēn šənēhɛm ʾim lō šå̄ laḥ yå̄ ḏō bi-mlɛḵɛṯ rēʿēhū [wə-
lå̄ qaḥ bəʿå̄ lå̄ w] [wə-lō yəšallēm]
‘(then) an oath before the LORD shall settle (the issue) between them that
the neighbor did not lay hands on the other person's property, [and at that
the owner shall accept (this)], [and (the neighbor) shall not make restitu-
tion].’ (Exod. 22:10)
Example (80) shows the apodosis of a legal case concerning a man who leaves
his animal to his neighbour for safety, “If a man gives a donkey, an ox, a
sheep or any other animal to his neighbour for safekeeping and it dies or is
injured or is taken away while no-one is looking” (NIB). Example (80) de-
scribes how such a case is settled legally. There shall be an oath before God,
and in two additive clauses the actions are stated that should accompany the
oath, from the side of the owner of the animal, and from the side of the neigh-
bor. The negative clause is coded by a we-lō-VprefL clause, the affirmative
accompanying action is coded by a we-Vsuff clause.105
The general meaning of a we-lō-VprefL clause is an accompanying negat-
ed action with imperfective aspect. When the pre-clause has a Vsuff verb
predicate, the Vsuff sometimes describes an anterior action which contrasts
with the imperfective aspect of the additive syndetic VprefL clause, which
may result in a habitual present or a future (81).
105 Other examples of equal status we-Vsuff and we-lō-VprefL clauses: Gen. 9:15 (we-
Vsuff+we-lō-VprefL), 17:5 (we-Vsuff+we-lō-VprefL+we-Vsuff), Exod. 9:4 (hinnē-
Spron-PA+[we-Vsuff]+[we-NEG-VprefL!]), 10:5, 12:13, 12:23, Exod. 21:22 (kī-
VprefL+[we-Vsuff] + [we-Vsuff] + [we-lō-VprefL]), 21:29 (ʾim-NCl+[we-Vsuff] +
[we-lō-VprefL]+[we-Vsuff], in protasis), 21:33 (we-kī-VprefL+ʾō+kī-VprefL+we-lō-
VprefL), 22:10 (we-Vsuff+we-lō-VprefL, within apodosis), 1 Sam. 1:11 (ʾim-VNabs-
VprefL+we-Vsuff+we-lō-VprefL).

226 Bo Isaksson
(81) Pattern: Spron-Vsuff + [we-lō-VprefL] + we-Vsuff + [we-lō-VprefL];

ha-hū ʾå̄ mar [wə-lō yaʿăśɛ] wə-ḏibbɛr [wə-lō yəqīmɛnnå̄ ]
‘Has he said, [and will not do it]? Or has he spoken, [and will not fulfill
it?]’ (Num. 23:19; ESV)
The proposed translations of the VprefL clauses in (81) vary, of course, con-
siderably. A present tense rendering ‘Has he spoken [and at that does not
fulfill it]?’ is quite as correct. The point in this clause combining is to express
an action that is closely connected with the Vsuff pre-clause, not as an at-
tendant circumstantial clause, but as an action that temporally succeeds the
Vsuff clause and is closely connected with it.
Also a modal (jussive) VprefS clause may be qualified by one or two syn-
detic VprefL clauses that express an accompanying (apodictic) action.
(82) Pattern: ʾal-VprefS+[we-lō-VprefL]+[we-Onoun-VprefL];

ʾal tōḵəlū mim-mɛnnū nå̄ ū-ḇå̄ šēl məḇuššå̄ l bam-må̄ yim kī ʾim ṣəlī ʾēš rōšō
ʿal kərå̄ ʿå̄ w wə-ʿal qirbō [wə-lō ṯōṯīrū mim-mɛnnū ʿaḏ bōqɛr] [wə-han-
nōṯå̄ r mim-mɛnnū ʿaḏ boqɛr bå̄ -ʾēš tiśrōp̄ ū]
‘Do not eat any of it raw or boiled in water, but roasted, its head with its
legs and its inner parts. [And you shall let none of it remain until the morn-
ing]; [anything that remains until the morning you shall burn].’ (Exod.
12:9-10; ESV)
The three clauses in (82) should not be rendered as having equal status, as
many translations do. The role of the switch from jussive to syndetic VprefL
clauses is to express accompanying (apodictic) actions in the future, ‘at that
you shall let none of it remain ...’. The second VprefL clause has a topicalized
object constituent han-nōṯå̄ r mim-mɛnnū ʿaḏ boqɛr ‘anything that remains
until the morning’ (see next section).106
Since the VprefL gram is not confined to a specific temporal reference, a
we-lō-VprefL may code an accompanying action to a narrative main line of
wa-VprefS clause, although this is rather infrequent in narrative prose. In such
a position the VprefL gram is “neutralized” by the negation (the event is not
realized) and may in some instances to function as a perfective Vsuff gram.
106 Other examples of modal pre-clauses are with we-lō-VprefL: Gen. 42:20 (we-Onoun-
VprefL+[we-VprefS+[we-lō-VprefL]), 47:19 (IMP+[VprefA+[we-lō-VprefL!] + [we-
Snoun-lō-VprefL]]), and Exod. 9:28 (IMP+[we-VprefA+[we-lō-VprefL!]]).

(83) Pattern: wa-VprefS+[we-lō-VprefL];

way-yuggaḏ lə-šå̄ ʾūl kī ḇå̄ raḥ då̄ wīḏ gaṯ [wə-lō-yōsēp̄ ʿōḏ lə-ḇaqšō]
‘And it was told Saul that David was fled to Gath; [and he sought no more
again for him].’ (1 Sam. 27:4; JPS)
The example gives the kethiv rendering with a verb ywsp that formally can be
also a Qal participle, but a prefix verb is more probable. 107 A we-lō-Vsuff
clause would have been more mainstream SBH, and this is also what the qere
offers here: wə-lō-yå̄ sap̄ (cf. C.3.3).108
B.2.2.2 The syndetic X-VprefL clause

The ‘X-VprefL’ type of clause has a topicalized element placed before the
verb, as was shown already in example (83). Since the verb in such a clause
was non-initial, this type of clause was not in danger of being confused with a
VprefS clause. In some contexts such a clause describes a background action.
In other contexts, and depending on the pre-clause, it describes a subsequent
action or state. The latter is true of the we-X-VprefL clauses in example (84).
(84) Pattern: NCl + [we-Snoun-VprefL] + [we-Snoun-VprefL] + [we-Snoun-

VprefL];
šənē ḡōyīm bə-ḇiṭnēḵ [ū-šənē ləʾummīm mim-mēʿayiḵ yippå̄ rēḏū] [ū-ləʾōm
mi-lʾōm yɛʾɛ̆må̄ ṣ] [wə-raḇ yaʿăḇoḏ ṣå̄ ʿīr]
‘There are two nations in your womb. [From birth they will be two rival
peoples]. [One of these peoples will be stronger than the other], [and the
older will serve the younger].’ (Gen. 25:23; CSB)
The pre-clause in (84) is a noun clause describing a present state in a direct

speech quotation. The three VprefL clauses that follow describe actions that
clearly succeed the state in the noun clause. The three are directly connected
107 Thus also Joosten (1999, 24) who regards it a “yiqtol”, but refers to it as the qere
version.
108 A case with we-lō-Vsuff and the same verb is found in 2 Sam. 2:28 which also exhibits
an example of we-lō-VprefL in a prose text (J: “past modal”). Instances of we-lō-
VprefL in similar positions with past time reference have often been regarded difficult
or problematic. Some more examples are: Gen. 2.25 (J: “past modal”) (wa-VprefS +
[we-lō-VprefL]), Judg. 6:4 (wa-VprefS+[we-lō-VprefL]), 12:6, 1 Sam. 1:7 (wa-VprefS
+ [we-lō-VprefL]), 2 Sam. 1:22 (PrP-Vsuff + [we-Snoun-lō-VprefL!]), Ps. 18:38 (Ø-
VprefS + [we-VprefS] + [we-lō-VprefL!]; with perfective past time Ø-VprefS), 18:39
(Ø-VprefS+[we-lō-VprefL]). (J) = Joosten (1999, 24).

228 Bo Isaksson
to the state coded by the noun clause, but with future event time. Semantical-
ly, they may also be analysed as elaborations of the NCl clause. The elabora-
tions constitute a prophecy on the meaning of the twins in Rebekah’s womb.
What marks the VprefL clauses as accompanying actions is the ū (allomorph
of we) before the first VprefL clause. The other two occurrences of we bind
the three VprefL clauses together as having equal status (cf. the discussion of
(78)).109
An accompanying action of the type we-X-VprefL sometimes acquires a
nuance of contrast, due to the topicalized X-element. An example is (85).
(85) Pattern: ADV-Spron-PA+[we-Onoun-VprefL];

ʿal kēn ʾănī zōḇēăḥ l-YHWH kål pɛṭɛr rɛḥɛm haz-zəḵå̄ rīm [wə-ḵål bəḵōr
bå̄ nay ʾɛp̄ dɛ]
‘Therefore I sacrifice to the LORD all the males that first open the womb,
[but all the firstborn of my sons I redeem].’ (Exod. 13:15; ESV)
In this example, the VprefL clause codes an action that is concomitant with
that in the preceding clause: ‘but at that I redeem all firstborn of my sons’.
The PA clause expresses a habitual action, and so does the VprefL clause,
which describes a contrasting exception in the general procedure of sacrificing
all that opens a womb.
A we-X-VprefL clause may, in specific contexts, receive a nuance of re-
sult, as in (86). The example shows an addition, and the context favours a
result interpretation (Dixon 2009, 28).110
(86) Pattern: Ø-Vsuff+[we-Snoun-VprefL];

på̄ ḏå̄ nap̄ šī mē-ʿăḇōr baš-šå̄ ḥaṯ [wə-ḥayyå̄ ṯī bå̄ -ʾōr tirʾɛ]
‘He redeemed my life from going down to the place of corruption, [and
my life sees the light]!’ (Job 33:28; NET)
In (86) Elihu son of Barakel states, with a Vsuff clause, that God redeemed
(perfective) or has redeemed (anterior) his life from going down in šaḥaṯ (here
109 Other instances of NCl + syndetic VprefL are Ps 1:2 (kī-NCl+[we-Onoun-VprefL!]

with result nuance), and 18:23 (kī-NCl+[we-Onoun-VprefL!]).
110 An explicit result is rendered by the German SCL, ‘er hat meine Seele erlöst, daß sie
nicht ins Verderben hinabgefahren ist, so daß mein Leben das Licht wieder sieht!’.
Other instances of Vsuff pre-clause with syndetic X-VprefL are: Gen. 4:14
(Vsuff+[we-PrP-VprefL] with result nuance), 2 Sam. 1:22 (PrP-Vsuff+[we-Snoun-
VprefL!]).

in pause: šå̄ ḥaṯ). The consequence is that his life can see light. The accompa-
nying action clause takes over the event time from the Vsuff clause and adds a
result clause with dynamic action: ‘from then my life sees light’.
In relation to a storyline of wa-VprefS clauses, the syndetic VprefL clause
is not as frequent as could be expected. The extant examples show the same
variation of meaning as we have seen above. A pure addition is found in (87).
(87) Pattern: wa-VprefS +[we-VNabs+VprefL];

wat-tiṯpallel ʿal YHWH [ū-bå̄ ḵō ṯiḇkɛ]
‘Hannah prayed to the LORD, [weeping bitterly].’ (1 Sam. 1:10; NIV)
The NIV translation has correctly perceived that there is a linking between the
wa-VprefS clause and the VprefL clause. The addition expresses ‘and at that
she wept bitterly’ with an imperfective aspect and with a concomitant event.111
An addition with a VprefL verb may also function as an elaboration (Dix-
on 2009, 27). This is shown in (88).
(88) Pattern: wa-VprefS+[we-Snoun-VprefL];

way-yarʿem baš-šå̄ mayim YHWH [wə-ʿɛlyōn yitten qōlō bå̄ rå̄ ḏ wə-ḡaḥălē
ʾēš]
111 The prototypical imperfective meaning of the VprefL gram is sometimes hard to per-
ceive in non-main addition clauses with VprefL predicate. In (87) the VprefL gram is
not more ‘iterative’ than the initial wa-VprefS clause (which however is compatible
with continuous or iterative actions, see (29) and (30)). But the main clause in (87) is
by implication continuous (‘prayed to the Lord’) and this is probably the reason for the
choice of a VprefL gram in the addition clause. It is to be expected that the properties
of the VprefL gram are not fully realized in non-main clauses. “Difficult cases” of we-
X-VprefL clauses with ‘neutral’ or close to ‘punctual action’ after a wa-VprefS clause
are: Gen. 37:7 (J: “past modal”) (we-hinnē-Vsuff + we-gam-Vsuff + we-hinnē-VprefL;
‘then your bundles came gathering around mine’), Exod. 8:20 (wa-VprefS+[we-PrP-
VprefL]), Deut. 2:12 (J) (we-PrP-Vsuff+ [we-Snoun-VprefL], ‘but the descendants of
Esau drove them out (successively)’), 2 Sam. 15:37 (wa-VprefS+[we-Snoun-VprefL];
‘So Hushai David's friend came into the city; and Absalom was at the point of coming
into Jerusalem’ JPS), 23:10 (J: “iterative”) (wa-VprefS+[we-Snoun-VprefL]; ‘The
Lord brought about a great victory that day [and the troops came strolling back to him
– only to strip the slain]’), 1 Kgs 7:15,23 (J: “iterative”) (a special usage specifying the
circumference of bronze pillars), 20:33 (wa-VprefS + "..." + [we-Snoun-VprefL]), 21:6
(J: “anomalous”) (kī-VprefL+wa-VprefS ‘Because I was talking to Naboth the Jezreel-
ite and said to him’), Ps. 8:6 (wa-VprefS+[we-Onoun-VprefL] ‘You made him a little
lower than the heavenly beings and crown him with glory and honour.’), 22:19 (Ø-
VprefS+[we-PrP-VprefL]), 78:72 (wa-VprefS+[we-PrP-VprefL]), Isa. 6:4 (wa-
VprefS+[we-Snoun-VprefL]). (J) refers to Joosten (1999, 24).

230 Bo Isaksson
‘Then the LORD thundered from heaven, [and at that the Most High gave
forth His voice – hail and fiery coals].’ (Ps. 18:14)
A reasonable interpretation of this piece of archaic poetry is that the two

clauses describe the same action, not two different actions. The description in
the VprefL clause, ‘gave forth his voice – hail and fiery coals’, supplies addi-
tional information about the thundering from heaven.112
A syndetic X-VprefL clause may of course also be added to an imperative.
A pure addition is shown in (89).
(89) Pattern: IMP+[we-Onoun-VprefL!];

wə-śīm kɛsɛp̄ ʾīš bə-p̄ ī ʾamtaḥtō [wə-ʾɛṯ gəḇīʿī gəḇīăʿ hak-kɛsɛp̄ tå̄ śīm bə-p̄ ī
ʾamtaḥaṯ haq-qāṭōn]
‘and put each man's money in the mouth of his bag. [In the mouth of the
youngest one’s bag put also my silver goblet].’ (Gen. 44:1-2; NAB)
The translators have had difficulty when rendering this clause combining. The
action performed by the VprefL clause is not an elaboration and not a result,
nor a reason. It is a separate accompanying action that is to be performed in
close connection with that in the imperative clause.113
When a we-X-VprefL clause interacts with a we-Vsuff clause, the two
clauses usually have equal status, the we-X-VprefL clause representing a
syntactic retention, while the we-Vsuff clause is the innovative expression
with no topicalized element (C.2). The meaning is often futural, habitual, or
apodictic. A future meaning is found in example (90).114
(90) Pattern: we-Vsuff+*we-Onoun-VprefL!;

ū-nəṯattīw lə-ḡōy gå̄ ḏōl *wə-ʾɛṯ bərīṯī ʾå̄ qīm ʾɛṯ yiṣḥå̄ q
‘and I will make him into a great nation, *but my covenant I will establish
with Isaac’ (Gen. 17:20b-21a)
112 Other instances of indicative VprefS + syndetic X-VprefL are: Exod. 8:20 (wa-VprefS
+[we-PrP-VprefL]), Isa. 6:4 (wa-VprefS+[we-Snoun-VprefL]), Ps. 8:6 (wa-VprefS +
[we-Onoun-VprefL]), 78:72 (wa-VprefS + [we-PrP-VprefL] same-event addition).
113 Other instances of IMP + syndetic VprefL are: Gen. 24:14 (IMP+[we-Onoun-VprefL]
pure addition), Exod. 5:18 (IMP + IMP + [we-Snoun-NEG-VprefL] + [we-Onoun-
VprefL];), and Job 6:24 (IMP+[we-Spron-VprefL!] result).
114 The equal status we-X-VprefL clauses are marked by *.

The two clauses are separated by a verse border in the written text, which for
a modern reader tends to conceal the continuity of the procedural main line in
which the clauses are of equal status, only that the second clause has a topical-
ized object noun, ‘my covenant’.
A similar addition of a clause with topicalized subject and equal status can
be perceived in (91).
(91) Pattern: we-Vsuff+*we-Snoun-lō-VprefL;

ū-nəṯattīw l-YHWH kål yəmē ḥayyå̄ w *ū-mōrå̄ lō yaʿălɛ ʿal rōšō
‘then I will give him to the LORD all the days of his life, *and no razor
shall touch his head.’ (1 Sam. 1:11; ESV)
In contexts of longer promises, the core of the promised actions is often coded
by the affirmative we-Vsuff clauses without topicalized element, expressing
only the actions themselves. In such a main line, a syndetic VprefL clause
expresses either a negative action or a clause with topicalized clause constitu-
ent, mōrå̄ ‘razor’. In (91), ‘I shall give him to the Lord ... [and razor shall not
touch his head]’, there is no result nuance involved in the VprefL clause, nor a
contrast. The VprefL has the same event time as the we-Vsuff clause (future).
A negation in such a clause is placed after the topicalized element (mōrå̄ ‘ra-
zor’), immediately before the verb.115
B.3 Summary
The VprefL gram is restricted to non-clause-initial position. Its negation is lō
(in poetry sometimes bal). As a main clause it expresses the meanings to be
expected from an imperfective gram trajectory: progressive (less frequent in
SBH), future, habitual, iterative, and apodictic meanings.
The VprefL gram is frequent in instructions and procedural discourse (al-
ternating with we-Vsuff clauses).
115 Other instances of equal status we-Vsuff and syndetic X-VprefL clauses: Gen. 6:19
(we-Vsuff+we-PrP-VprefL!), 17:6 (we-Vsuff+we-Snoun-VprefL), 30:42 (we-Vsuff +
we-PREP-VN+lō-VprefL!), 44:9 (REL-VprefL + [we-Vsuff + we-Spron-VprefL]),
47:24 (we-Vsuff+we-Snoun-VprefL), Exod. 4:21 (we-Vsuff+we-Spron-VprefL+we-lō-
VprefL), 7:15, 7:17-18, 12:12, 12:22 (we-Vsuff+we-Spron-lō-VprefL), 13:13, 18:23,
21:35, 23:11, Judg. 2:3, 1 Sam. 1:5 (wa-VprefS + [we-Vsuff + we-PrP-VprefL, with
habitual action and contrast), 1 Kgs 8:35, Amos 2:3 (we-Vsuff+we-Onoun-VprefL),
4:7 (we-Vsuff+we-PrP-VprefL!] habitual main line), Ps. 1:3 (we-Vsuff+we-Snoun-
VprefL!).

232 Bo Isaksson
When the clause is affirmative and there is no need of a topicalized clausal

element, the expected *we-VprefL clause is supplanted by a we-Vsuff clause;
a syntax that is characteristic of SBH (but not of the archaic poetry).
The non-main asyndetic VprefL clause usually codes an attendant circum-
stance (always of the type Ø-X-VprefL or Ø-lō-VprefL). This clause type is
mostly found in poetry.
The non-main syndetic VprefL clause is an addition carrying meanings
that are expected from an imperfective gram. It often has habitual or futural
meanings, but sometimes it also comes close to a circumstantial clause. As an
addition it may also acquire a nuance of result, depending on the context.
Also, as non-main clauses we-X-VprefL or we-lō-VprefL clauses alternate
with equal status we-Vsuff clauses (affirmative and without topicalized ele-
ment).
C. Clause combining and the Vsuff gram

The grammaticalization path of a resultative gram starts with a resultative
meaning (‘he is gone’) and then normally develops an ability to take on also
anterior meanings (‘he has gone away’). In a final stage the gram can express
a perfective aspect (‘he went away’) and eventually also a past tense. In addi-
tion to this gamut of meanings, with a perfective aspect that is not temporally
restricted to the past, future events may be viewed as bounded (‘prospective
future’) and express optative nuances. The Vsuff gram in Standard Biblical
Hebrew can express all these meanings. In such cases, the gram is still not a
past tense (Bybee et al. 1994, 63-87, 93 and the summary on 95; Isaksson
2009, 134).
C.1 Vsuff clauses encoding a main line

A resultative Vsuff in main line is mostly encountered in poetry. An example
is found in Hannah’s prayer (92).
(92) Pattern: Ø-Vsuff+Ø-Vsuff+Ø-Vsuff;

ʿå̄ laṣ libbī b-YHWH rå̄ må̄ qarnī be-YHWH rå̄ ḥaḇ pī ʿal ʾōyəḇay
‘My heart exults in the LORD; my horn is exalted in the LORD. My
mouth derides my enemies.’ (1 Sam. 2:1; ESV)

Hannah bursts out in praise after having born a son. In all three Vsuff clauses,
a present state of joy and praise is described.116
The anterior meaning of the Vsuff is found in all kinds of texts, not the
least in direct speech, as in (93).
(93) Pattern: Ø-Vsuff+we-gam-Vsuff+{wa-VprefS};

då̄ nannī ʾɛ̆lōhīm wə-ḡam šå̄ maʿ bə-qōlī {way-yittɛn lī bēn}
‘God has vindicated me; he has listened to my plea {and gave me a son}.’
(Gen. 30:6)
The reference is to an already completed action, the results of which are still
relevant to the present situation of the speaker. It is reasonable to hold that the
two Vsuff clauses have equal status.117
The perfective aspect views the action as a bounded whole and in narra-
tive this is, consequently, rendered by an English past tense. In a royal chroni-
cle, this usage of the Vsuff often codes a main line, as in (94).
(94) Pattern: Snoun-Vsuff;

w-īhōrå̄ m bɛn ʾaḥʾå̄ ḇ må̄ laḵ ʿal yiśrå̄ ʾēl bə-šōmərōn bi-šnaṯ šəmōnɛ ʿɛśrē
l-īhōšå̄ p̄ å̄ ṭ mɛlɛḵ yəhūḏå̄
‘Joram son of Ahab became king of Israel in Samaria in the eighteenth
year of Jehoshaphat king of Judah.’ (2 Kgs 3:1; NIV)
The action is viewed as a bounded whole, in this context in the past, in a spe-
cific year in the history of Israel.118
116 Other instances of resultative Vsuff are: Num. 23:21 (‘Er schaut kein Unrecht in Jakob,
und er sieht kein Unheil in Israel’, SCL), 2 Sam. 1:26 (Ø-Vsuff+ Ø-Vsuff+ Ø-Vsuff),
Ps. 6:8 (Ø-Vsuff+Ø-Vsuff), 14:2. A significant relative clause with Vsuff is found in 2
Kgs 3:14 (‘As the LORD of hosts lives, before whom I stand’, ESV).
117 The main point here is not the wa-VprefS clause, which is treated under another head-
ing above (A.1.2), and which codes an elaboration of the preceding Vsuff (or, alterna-
tively, a temporal succession). Nearly all translators render the VprefS as an anterior.
But when linked to the Vsuff clauses as an addition it takes over the temporal reference
from them. It is not a new main clause. Other instances of anterior Vsuff in main line:
Exod. 5:3 (Snoun-Vsuff), 5:10 (kō-Vsuff ‘This is what Pharaoh has said’, but many
translators render this with present tense), 5:22 (ADV-Vsuff+ADV-Vsuff).
118 Other perfectives in main line: Gen. 27:35, 31:31 (kī-Vsuff+kī-Vsuff, with adverbial
kī), Exod. 15:13 (Ø-Vsuff+Ø-Vsuff), 2 Sam. 23:2 (Snoun-Vsuff+[NCl]+Ø-Vsuff+PrP-
Vsuff).

234 Bo Isaksson
A proof that Vsuff is still not a past tense in SBH would be that it can be
used for the immediate future (Bybee et al. 1994, 95). An example is (95).
(95) Pattern: IMP, Vsuff;

rəʾē nəṯattīḵå̄ ʾɛ̆lōhīm lə-p̄ arʿō
‘See, I make you as God to Pharaoh.’ (Exod. 7:1; NAS)
The clause is often rendered by English perfect tense, but the event is still not
fulfilled at the time of speech. Immediately before God’s utterance, Moses has
exclaimed, ‘Behold, I am unskilled in speech; how then will Pharaoh listen to
me?’ (NAS).119
The immediate future may also have nuances of potentiality, or an action
that nearly could have happened, as in (96).
(96) Pattern: ADV-Vsuff;

kī ʿattå̄ šå̄ laḥtī ʾɛṯ yå̄ ḏī
‘For by now I could have stretched out my hand.’ (Exod. 9:15; NIV)
God’s utterance is a threat to Pharaoh and an expression of the ability of God,

and, yet, it is still not performed.
The optative meaning of Vsuff is rare also in Arabic (Bybee et al. 1994,
93; Isaksson 2009, 132), and in Hebrew it usually needs an adverbial particle
to be realized, as in (97).
(97) Pattern: kī-ʾim-Vsuff+[CONJ-VprefL];

kī ʾim zəḵartanī ʾittəḵå̄ [kaʾăšɛr yīṭaḇ lå̄ ḵ]
‘Only remember me, [when it is well with you].’ (Gen. 40:14; ESV)
Neither kī nor ʾim are conjunctions here, it is not a protasis. The ʾim particle
expresses a polite ‘if only ...’.
C.2 The we-Vsuff (‘we-qatal’) clause

The we-Vsuff clause in SBH exhibits an astonishing variation in meaning and
temporal reference. The meanings are partly those otherwise found with the
VprefL gram. This has led to the supposition that the ‘we-qatal’ syntagm
119 The immediate future is close in meaning to the bounded whole of a performative
action. It is often hard to distinguish the one from the other: Gen. 1:29 (hinnē-Vsuff),
9:13 (Onoun-Vsuff, NET: ‘I will place my rainbow in the clouds’, NAS: ‘I set My bow
in the cloud’), 14:22, 23:11.13, 41:41, 47:23 , 48:22.

represents a separate gram in Biblical Hebrew (the term commonly used is

‘conjugation’). But there is no comparative evidence of two different kinds of
suffix conjugations, nor is there comparative support for different stress pat-
terns of the conjunction we with a Vsuff verb (Joosten 2012, 15). The position
taken in the present article is that ‘we-qatal’ consists of the normal conjunc-
tion we and the normal verbal grammatical morpheme Vsuff.
At the same time it is evident that, with the widened scope of the we-Vsuff
type of clause, something has happened in the Hebrew verbal system that
contrasts with neighbouring Canaanite languages. Meanings that are relatively
rare in main line Vsuff clauses (cf. C.1 above), are frequently used in the we-
Vsuff syntagm, especially the future, habitual and modal meanings.
From the perspective of the ‘normal’ meanings of the Vsuff gram, it is
conspicuous that the ‘new’ and unexpected meanings of the we-Vsuff clauses
are those we would expect of the imperfective gram of the Hebrew verbal
system, the VprefL.
The innovative word order restrictions in Hebrew already discussed above
were the result of a dramatic change that forced the speakers of (Proto-) He-
brew to radically reorganize the syntax of the verb in order not to lose contact
with fundamental distinctions in the system of verbal grams. This radical
change was caused by the loss of short final vowels and the subsequent loss of
the old distinction yaqtul/yaqtulu. The dropping of final vowels led to a partial
loss of the distinction between the short prefix verb and the long prefix verb.
It was lost in the whole regular verb flection, except Hifil.
The nearly complete morphological merger of the VprefS gram and the
VprefL gram, gave rise to word order strategies in order to maintain the func-
tional distinction between the two prefix grams (Gzella 2012, 100).120 As we
have seen above, the solution was to place the VprefS verb in clause-initial
position (except for the conjunction we/wa and except for clauses negated by
ʾal). For the same reason, the VprefL gram could no longer be allowed to
occupy a fronted position. With this restricted word order, and with the use of
initial particles, the distinction between the VprefS gram and the VprefL gram
was upheld, at least temporally, in Biblical Hebrew. It would be understanda-
ble if, after some time, the linguistic instinct sometimes failed to capture the
meaning of the morphological distinction between, say, yå̄ qūm and yå̄ qŏm.
Morphological mistakes could be allowed anyway, because the functional
120 See the introduction, Hypothesis: A word order revolution in Proto-Hebrew.

236 Bo Isaksson
distinction was upheld by word order itself. In this perspective, the morpho-
logical distinction represented an archaism in Biblical Hebrew. We sometimes
encounter forms that are long (VprefL) from a morphological point of view,
but are placed in initial position and are evidently intended to be interpreted as
jussives or indicative perfectives (‘VprefSᴸ’, cf. example (23)).
This is where the we-Vsuff clause fits in. The Vsuff gram in the we-Vsuff
syntagm is always clause-initial, and the conjunction we marks it as being an
addition, an accompanying action. The we-Vsuff clause came in use in cases
when a VprefL clause was no more acceptable. An accompanying action
coded by a we-VprefL clause type would cause confusion with the (usually
purposive) we-VprefS type (with a clause-initial VprefS). And confusion was
avoided by using the we-Vsuff clause instead, presumably with a gradual
widening of its application spectrum. The we-Vsuff clause took over some of
the meanings and functions of the old simple syndetic affirmative VprefL
clause (with no topicalized element). As soon as a negation was needed, or a
topicalized (‘X’) element should precede the verb, the old VprefL clause was
retained: we-lō-VprefL or we-X-VprefL. This fact has made some scholars
conclude that the we-Vsuff syntagm “could not” be split up by a negation, that
the we-Vsuff clause could not be negated. But there was no need for such
‘split’, there were already negated VprefL clauses available that complied
with the word order rule: a we-X-VprefL or we-lō-VprefL could not give rise
to confusion.
The fact that we-Vsuff is an addition means that there is always a clause
that precedes it, and to which it refers, a ‘pre-clause’.121 This does not mean
that the pre-clause is always a main line. It is often not a main line and some-
times it is only a tiny noun clause coding a temporal predication. The we-
Vsuff clause(s) rather often, from a textual point of view122 – codes a main
line.
So, there is nothing mystical about the we-Vsuff clause and it is not “equal
with” a “yiqtol” clause, as is often maintained. And the we-Vsuff syntagm is
not a gram of its own. The we-Vsuff clause codes an addition, the range of
application of which has been widened beyond what we are used to see for
other Vsuff clauses. It is a simple accompanying action, because it is always
121 See the section A cross-linguistic typology of semantic clausal relations, in the intro-
duction.
122 Which is the perspective to be taken when we determine the main line in a text.

affirmative and lacks a topicalized element. Its focus is on the verbal action
itself. This is beautifully illustrated in (98), taken from legal discourse.
(98) Pattern: (kī-VprefL+[we-Vsuff]+[ʾō-Vsuff]) ...;

(kī yiḡnoḇ ʾīš šōr ʾō śɛ [ū-ṭəḇå̄ ḥō ʾō məḵå̄ rō]) ...
‘(If a man steals an ox or a sheep, [and kills it or sells it]),’ (Exod. 21:37;
ESV)
Example (98) shows the protasis construction (marked by parentheses) of a

legal case. As is so common in legal discourse as well as in other protasis
complexes, the protasis contains additional conditions, and in this case the
additional conditions are coded first by a we-Vsuff clause then by an alterna-
tive action. The we-Vsuff clause describes an action that, in this legal case,
must accompany that in the main protasis clause (the kī yiḡnoḇ ʾīš šōr ʾō śɛ)
for the case to be valid. But the disjunctive alternative, signaled by ʾō, is not a
we-Vsuff clause as we would possibly expect, but a Vsuff clause. Since this
clause is completely parallel, the alternatives are symmetric and of equal sta-
tus (Dixon 2009, 30). This ʾō-Vsuff clause has the same meaning and function
as the we-Vsuff clause.
Example (98) shows that the we-Vsuff syntagm is not a conjugation. It also
shows that the conjunction we in a we-Vsuff clause is a ‘normal’ conjunction
we with additive meaning. In the example, the we (in this case the allomorph
ū) marks the following Vsuff as an accompanying action and in the disjunc-
tion clause there is no need to mark this a second time.123
There are instances when the we-Vsuff clause is actually ‘split up’, when
there is no negation involved and no ‘normal’ topicalized element (which
would result in a VprefL clause). This is the case with the adverb gam ‘also;
moreover’ in (99).
(99) Pattern: Ø-lō-VprefL + kī-NCl + [we-Vsuff] + [we-gam-Vsuff] + [we-

Vsuff] + [we-Vsuff];
lō-tiqrå̄ ʾɛṯ-šəmå̄ h śå̄ rå̄ y kī śå̄ rå̄ šəmå̄ h [ū-ḇēraḵtī ʾōṯå̄ h] [wə-ḡam nāṯattī
mim-mɛnnā ləḵå̄ bēn] [ū-ḇēraḵtīhā] [wə-hāyəṯå̄ lə-ḡōyīm]
‘you shall not call her Sarai, but Sarah shall be her name. [I will bless her],
[and moreover I will give you a son by her]. [I will bless her], [and she
shall give rise to nations];’ (Gen. 17:15-16; NRS)
123 A similar example with disjunction is Exod. 22:13 (we-kī-VprefL +[we-Vsuff+ʾō-

Vsuff+[Ø-NCl]]), also within a protasis.

238 Bo Isaksson
In (99) we encounter an instruction to Abraham, but also a promise and a

prophecy. The initial VprefL clause (lō-tiqrå̄ ʾɛṯ-šəmå̄ h śå̄ rå̄ y) is followed by
we-Vsuff clauses describing actions that will accompany the renaming of
Sarah. In this series of four we-Vsuff clauses, the second has an inserted gam
‘moreover’ between the conjunction and Vsuff verb. The meaning of the verb
in this clause is the same as the surrounding we-Vsuff clauses, that is, future
event time.124
C.2.1 The meanings of the we-Vsuff clause

C.2.1.1 Pure accompanying action
A ‘pure’ accompanying action lacks connotations of result and time. It just
describes an action that is done in a certain connection with that of the pre-
clause, not necessarily concomitant with it (but could be so). An example is
(100).
(100) Pattern: wa-VprefS+[we-Vsuff];

wa-yhī hay-yōm way-yizbaḥ ʾɛlqāna [wə-nå̄ ṯan li-p̄ ninnå̄ ʾištō ū-lə-ḵål-
bå̄ nɛ-hå̄ ū-bənōṯɛ-hå̄ må̄ nōṯ]
‘One such day, Elkanah offered a sacrifice. [And at that he gave portions
to his wife Peninnah and to all her sons and daughters.’ (1 Sam. 1:4)
The addition in (100) is ‘pure’, in that there is no relation of result or temporal

succession. The eating of the sacrifice was part of the offering. It is pertinent
to ask why this we-Vsuff was not expressed by a new wa-VprefS (way-yitten).
The answer is that the context is habitual, and habituality is one of the mean-
ings the we-Vsuff ‘took over’ from the imperfective VprefL gram in simple
affirmative clauses. The whole context is habitual, because year after year this
specific offering was performed by Elkanah.
No habituality is perceived in the next example, but instead a future time
reference.
(101) Pattern: ʾal-VprefS+[Ø-Snoun-VprefL+we-Vsuff];

ʾal yippol lēḇ ʾå̄ ḏå̄ m ʿå̄ lå̄ w [ʿaḇdəḵå̄ yēleḵ wə-nilḥam ʿim hap-pəlištī haz-
zɛ]
‘Let no man’s heart fail because of him; [for your servant will go and fight
with this Philistine].’ (1 Sam. 17:32)
124 For a separate study of we-Vsuff clauses, see Isaksson (forthcoming a).

The example shows three clauses. The jussive should be identified as the main
clause. The next clause is asyndetic, so it is not an addition. It relates to the
jussive clause by describing a reason why no man’s heart should fail. This
reason clause is a typical attendant circumstantial clause, meaning ‘in that thy
servant will go’ (the nuance of reason is understood). The continued action
‘and fight with ...’ should have been coded by a simple we-VprefL clause (wə-
yillå̄ ḥem) but, since such a clause was not allowed, the we-Vsuff clause was
used in its stead. For historical reasons, the two clauses have equal status, in
spite of the formal ‘gram switch’ that is displayed in the construction.
In the next example there is neither future time, nor habituality in the
meaning of the we-Vsuff clause (102).
(102) Pattern: wa-VprefS+[we-Vsuff];

way-yōṣē ʾōṯå̄ h məšå̄ rəṯō ha-ḥūṣ [wə-nå̄ ʿal had-dɛlɛṯ ʾaḥărɛhå̄ ]
‘Then his servant brought her out, [and at that he bolted the door after
her].’ (2 Sam. 13:18)
The example illustrates what we would expect from an ‘original’ or ‘prototyp-

ical’ we-Vsuff clause. The we-Vsuff describes an action that accompanies that
in the storyline, and the aspect is perfective with past temporal reference.125
Since the storyline is made up of wa-VprefS clauses, a new wa-VprefS (pos-
sibly *way-yinʿal) would have been unable to stress the close connection
between the two events.
The diachronic scope of Biblical Hebrew, and a reasonable assumption of
a gradual widening of the semantic span of the we-Vsuff clause, justify an
examination of one early and one late example of the usage of we-Vsuff
clauses. The first example, (103), has been adduced by Notarius (2013, 288)
as signifying a diachronic state when there was no special we-Vsuff clause,
that is, with meanings close to a VprefL.
125 Amnon’s servant brings Tamar out of the house after Amnon has raped his sister. And
as an accompanying action the servant bolts the door after her. The same type of
switch is found in Judg. 3:23. This wə-nå̄ ʿal is commonly regarded as an example of
“non-consecutive waw” or “copulative waw” or a textual error. The New English
Translation (2004) remarks, “The Hebrew verb is a perfect with nonconsecutive vav,
probably indicating an action (locking the door) that complements the preceding one
(pushing her out the door)”. Driver (1913) suggests textual error with reference to Ges-
K § 112tt, and so also Joüon and Muraoka (2006 § 119z).

240 Bo Isaksson
(103) Pattern: Ø-Vsuff + we-Vsuff + we-Vsuff + we-Vsuff + we-Vsuff + we-

Vsuff;
då̄ raḵ kōḵå̄ ḇ miy-yaʿăqōḇ, wə-qå̄ m šeḇɛṭ miy-yiśrå̄ ʾēl, ū-må̄ ḥaṣ paʾăṯē
Mōʾå̄ ḇ, wə-qarqar kål bənē šēṯ
‘A star is risen from Jacob, (thereby) a scepter comes forth from Israel; It
smashes the brow of Moab, destroys all children of Seth. Edom becomes a
possession, Yea, Seir a possession of its enemies.’ (Num. 24:17-18)
This famous prophecy of Balaam is commonly held to be a piece of archaic

poetry and is not easily compatible with what standard grammars describe to
be normal usage of we-Vsuff clauses. The introductory Vpref clauses not
quoted here (‘What I see for them is not yet, What I behold will not be soon’)
sets the event time to sometime in the future. The heart of the prophecy (this
example) then shifts to an asyndetic Vsuff clause. The listener perceives that
the shift to a suffix verb raises the intenseness of the prophetic event by pre-
senting it as something definite, as a resultative meaning, situated in the fu-
ture: ‘a star is risen from Jacob’. With a context of prophetic future, the Ø-
Vsuff clause då̄ raḵ raises the imagination of the listener by stating that the
action already occurs in the vision of the prophet. This prophetic future Vsuff
is a well-known phenomenon in prophetic speech, but what is remarkable here
is that the following we-Vsuff clauses seem to have the same meaning and
temporal reference as the initial asyndetic då̄ raḵ. This passage could then be
seen as a confirmation of Notarius’ observation. But there are some further
observations to be made. The Ø-Vsuff clause and the we-Vsuff clauses have
the same event time and the same verbal aspect. The we-Vsuff clauses follow
after an initial asyndetic clause, då̄ raḵ. There is no variation in word-order, no
subject or object is put before the verb. The series of five we-Vsuff clauses is
quite spectacular: it is as if the poet wanted to avoid other positions of the
verb, and for a specific reason. It is possible to interpret the we-Vsuff clauses
as having a semantic relation to the initial asyndetic Vsuff då̄ raḵ: they can be
construed as coding something additive to the initial statement: A scepter is
risen from Israel, the rest are further details that are connected with this future
event. It is not possible to interpret all we-Vsuff clauses as result clauses,
some may be, but not all. If there is a meaning of consequence-result in some
of the we-Vsuff clauses, this must be inferred from an analysis of the context,
it is not explicitly coded by the we-Vsuff clauses themselves. The same holds
for an interpretation of temporal succession. Some of the we-Vsuff clauses

can be pressed to express temporal succession, but not all of them. They just
express actions that accompany the initial Ø-Vsuff clause.
Example (103) is archaic poetry and what we observed concerning the
function of the we-Vsuff clauses could be regarded the starting point of the
internal development of the we-Vsuff clause, that is, (103) shows additions
with ‘normal’ prophetic Vsuff properties.
An alleged example from the diachronic end of the development, close to
the ‘extinction’ of the we-Vsuff clause, is supplied by Joosten (2012, 16). He
gives an example of a “copulative we-qatal”, in (104).
(104) Pattern: (quotation) +we-Vsuff+we-lō-Vsuff;

“......” wə-hɛḥɛ̆rīšū hå̄ -ʿå̄ m wə-lō ʿå̄ nū ʾōṯō då̄ ḇå̄ r
‘"...." But the people were silent and did not say a word in reply.’ (2 Kgs
18:36; TNK)
The we-Vsuff clause in (104) comes directly after a long speech by the com-
mander of the Assyrian army outside the walls of Jerusalem. This coding of
what looks like a narrative thread by a we-Vsuff clause appears to Joosten to
be non-classical Hebrew syntax. We would have expected a wayyiqtol syn-
tagm here. It is hard to interpret this wə-hɛḥɛ̆rīšū with some of the usual func-
tions of the we-Vsuff syntagm: it is certainly not modal, it is not future, it is
not purposive, it is not a result clause, and it is hardly habitual action either.
The people on the walls kept silent, not as a reaction of the speech, but on the
command of king Hiskia. But what could certainly be stated about the we-
Vsuff clause is that it is an action that accompanies the speech of the Assyrian
commander: ‘at that speech the people kept silent’.
In view of the diachronic extremes in the usage of the we-Vsuff syntagm,
it is reasonable to suppose that the we-Vsuff clause went through a develop-
ment from the earliest stage represented by the archaic poetry, a stage which
in this respect resembles the earliest Northwest Semitic inscriptions, also the
Aramaic ones. Then the we-Vsuff syntagm took over some additive usages
from the VprefL gram (the simple ones without negation and without topical-
ized elements) and developed a broad semantic spectrum, including modal
meanings, future, and habituality. The end stage constituted a return to ‘nor-
mal’ Vsuff meanings in addition position, that is additive we plus the Vsuff
gram. It is reasonable to suppose that this renewed ‘normal’ additive narrative
we-Vsuff should be understood in the light of the gradual displacement of

242 Bo Isaksson
narrative wayyiqtol with the qatal and we-Vsuff syntagms in late Biblical
Hebrew.
C.2.1.2 A temporal connotation: ‘then’

The accompanying action easily lends itself to be interpreted as a temporal
‘then’, ‘at that occasion’, when the pre-clause involves an expression of time
or condition or (legal) case. In such instances it is often incorrect to analyze
the pre-clause as a main line clause, and the focal clause (the event at that
time) must be analyzed as main line clause.126 This is evident in (105).
(105) Pattern: [kī-VprefL]+we-Vsuff;

[kī yəḏabber ʾălēḵɛm parʿō lēmōr ‘tənū lå̄ -ḵɛm mōp̄ ēṯ’] wə-ʾå̄ martå̄ ʾɛl
ʾahărōn ...
‘[When Pharaoh says to you, ‘Prove yourselves by working a miracle,’]
then you shall say to Aaron ...’ (Exod. 7:9; ESV)
The temporal clause with an initial kī is a non-main clause in this linking, and
the we-Vsuff clause achieves a temporal nuance because of the semantic
properties of the pre-clause. In (105) there is no nuance of result or conse-
quence involved, but in other contexts this ‘then’ of the we-Vsuff clause ap-
proaches that of the ‘then’ of the apodosis in a conditional clause linking.
The same temporal nuance without any shade of consequence may be
achieved when the pre-clause is coded by a participle, as in (106).
(106) Pattern: Spron-PA+[we-Vsuff]+[we-Vsuff];

ka-ḥăṣōṯ hal-laylå̄ ʾănī yōṣē bə-ṯōḵ miṣrå̄ yim [ū-mēṯ kål bəḵōr be-ʾɛrɛṣ
miṣrayim...] [wə-hå̄ yəṯå̄ ṣəʿå̄ qå̄ ḡəḏōlå̄ bə-ḵål ʾɛrɛṣ miṣrå̄ yim]
‘About midnight I will go throughout Egypt [and then every firstborn male
in the land of Egypt will die ...] [and there will be a great cry of anguish
through all the land of Egypt ...]’ (Exod. 11:4-6)
The example illustrates the habit of forming chains of we-Vsuff clauses that
are syntactical additions to the initial pre-clause, and which often constitute
the textual main line of the passage. In (106) the pre-clause ‘I will go
throughout Egypt’ is certainly not unimportant, but it is the more detailed we-
Vsuff clauses that are focal (Dixon 2009, 6). In a sense, the we-Vsuff clauses
126 For ‘focal clause’, see the introduction, section A cross-linguistic typology of semantic
clausal relations, and Dixon (2009, 3).

could be classified as elaborations of the pre-clause, were it not for the con-
spicuous temporality in the text.127
The conspicuous syntactic applicability of the we-Vsuff clause is illustrat-
ed when the pre-clause is not a finite clause, as the PA in (106), or when the
pre-clause is an infinitive, as in (107).
(107) Pattern: [PREP-VN]+we-Vsuff;

[bə-yallɛḏḵɛn ʾɛṯ hå̄ -ʿiḇriyyōṯ] ū-rəʾīṯɛn ʿal hå-ʾåḇnå̄ yim
‘[When ye do the office of a midwife to the Hebrew women], ye shall look
upon the birthstool.’ (Exod. 1:16; JPS)
Even when the pre-clause is an infinite clause it would be incorrect to analyse

it as being a constituent in the we-Vsuff clause. The conjunction we still
marks a juncture between two clauses, and in Hebrew this kind of clause
combining is no anomaly. The event time is determined by the pre-clause
(future), but there are still two separate clauses involved in the linking. This is
also the case when the pre-clause is a noun clause containing just a temporal
predicate, as in (108).
(108) Pattern: [NCl]+we-Vsuff;

[ʿōḏ məʿaṭ] ū-səqå̄ lūnī
‘[A little more] and they will stone me!’ (Exod. 17:4; NAB; NET)
The pre-clause consists of a temporal adverb as predicate and the we-Vsuff

clause amounts to a temporal ‘then’-clause: ‘when just a little more time (is
gone), then they will stone me’.128
Not surprisingly one or more we-Vsuff clauses may follow an initial im-
perative: ‘Do that, and then ...’.129 Example (109) shows a somewhat complex
127 The brackets could therefore be placed around the pre-clause instead.
128 Examples like (107) and (108) are often taken as “proof” that we-Vsuff is a “conjuga-
tion” of its own. But Exod. 12:3 [bɛ-ʿå̄ śōr la-ḥoḏɛš haz-zɛ] wə-yiqḥū lå̄ hɛm ʾīš śɛ lə-
ḇēṯ-ʾå̄ ḇōṯ śɛ lab-bå̄ yiṯ is a similar example with a prefix conjugation we-Vpref after a
temporal predication (NCl coded by a PrP), and is not a “proof” that we-Vpref is a sep-
arate weyiqtol conjugation (wə-yiqḥū should be analysed as a jussive we-VprefS).
129 It is by no means a rule that an imperative must be continued by a we-qatal clause.
More frequently imperative clauses are concatenated, as in Gen. 27:13, 42:18, 45:17,
Exod. 2:9, 6:11, 12:31, 14:13, 16:33, 17:5, 17:9, and many more. Some, but not all,
such examples of Ø-IMP+Ø-IMP linking are serial verb constructions, some (but pos-
sibly not all) examples of Ø-IMP+we-IMP linking express purpose (J-M § 116).

244 Bo Isaksson
clause structure (in the instruction for Noah’s Ark), which involves impera-
tive, long prefix verb and we-Vsuff clauses with a ‘then’ nuance.
(109) Pattern: IMP+Ø-ADV-VprefL+[we-Vsuff];

ʿăśē lə-ḵå̄ tēḇaṯ ʿăṣē ḡop̄ ɛr qinnīm taʿăśɛ ʾɛṯ hat-tēḇå̄ [wə-kå̄ p̄ artå̄ ʾōṯå̄ h
mib-bayiṯ ū-miḥūṣ bak-kop̄ ɛr]
‘Make yourself an ark of gopher wood making the ark with rooms, [and at
that/then you should cover it inside and out with pitch]].’ (Gen. 6:14)
The main general command in this example is coded by an imperative, ‘make

yourself an ark’. The we-Vsuff clause is not about the rooms (referred to in
the VprefL clause), but, in some way, qualifies the imperative, and concerns
the whole ark, not the making of the rooms: the ark should be covered inside
and out with pitch, an action that should accompany the building of the ark.130
C.2.1.3 Conditional constructions

A conditional clause combining semantically involves an ‘if’-part (‘protasis’)
and a ‘then’-part (‘apodosis’). Since a we-Vsuff clause, as we have seen, may
express a temporal or logical ‘then’-meaning, it is often a suitable choice for
an apodosis.131 An illustrative example is (110).
(110) Pattern: (ʾim-VprefL+[we-Vsuff])+(we-Vsuff);

(ʾim yå̄ ḇō ʿēśå̄ w ʾɛl ham-maḥănɛ hå̄ -ʾaḥaṯ [wə-hikkå̄ hū]) (wə-hå̄ yå̄ ham-
maḥănɛ han-nišʾå̄ r li-p̄ lēṭå̄ )
130 Other examples of temporal ‘then’ we-qatal clauses are found in Exod. 8:12 (IMP+we-
IMP+[we-Vsuff]), 8:23 (ADV-VprefL+[we-Vsuff+[CONJ-VprefL]]), 8:25 (hinnē-
Spron-PA+[we-Vsuff] +[we-Vsuff]), 12:48 (ʾāz-VprefL+[we-Vsuff]), 13:19 (VNabs-
VprefL+[we-Vsuff]), 16:5 (hinnē-Spron-PA+[we-Vsuff] + [we-Vsuff] + [we-Vsuff] +
[we-Vsuff] + [we-Vsuff]), 16:6-7 ([NP]+we-Vsuff+ [NP]+we-Vsuff), 17:11 ([CONJ-
VprefL!]+we-Vsuff+[CONJ-VprefL!]+we-Vsuff), Amos 4:2, 9:13-15 (many cases in
prophetic contexts).
131 This function of the conjunction, “wāw of apodosis”, is attested also in early Akkadian
(Sargonic) royal inscriptions (Kogan 2014, 54). It goes without saying that this and
other designations of we (“waw of coordination”, “inversive waw”, “waw of succes-
sion”, “waw of accompaniment”, “waw adequationis”) do not refer to separate waws,
not even separate “meanings” of waw but are in fact semantic descriptions of the rela-
tions of the clauses that are connected by waw (cf. J-M §§ 115, 117, 150p, 151a, 174h
note 1).

‘(If Esau comes to the one camp [and attacks it]), (then the camp that is
left will escape).’ (Gen. 32:9; ESV)132
Conditional clause linkings are usually more complicated than in (110), but
the example nevertheless permits some important observations. The example
constitutes one finished utterance with just those three clauses. The apodosis
is a we-Vsuff clause (wə-hå̄ yå̄ ...). The protasis consists of a complex of two
clauses, of which the second is a we-Vsuff clause, wə-hikkāhū. Since wə-
hikkāhū is not an apodosis, what does it express? It functions just like other
we-Vsuff clauses. The role of the wə-hikkāhū in the protasis is to add another
condition; the condition becomes more specific by the we-Vsuff clause. The
condition contains two specific events: if Esau arrives, and also attacks the
camp.
The most intriguing observation from example (110) remains, however.
There is no specific syntactic marking of the apodosis. There are admittedly
two we-Vsuff clauses in the example, but only one is contained in the apodo-
sis, the other codes an addition within the protasis. In (110) the protasis is
syntactically marked (by the conjunction ʾim). But we cannot by the syntactic
coding alone determine where the apodosis begins. The beginning of the
apodosis must be inferred from the pragmatic and semantic context, and,
possibly, from the intonation (which is mainly unavailable to us modern read-
ers).
When the apodosis is to begin with a negative clause, the more ancient
construction, with the VprefL gram, is used, and then with asyndesis.
(111) Pattern: (kī-VprefL!+[we-Vsuff]+[we-Vsuff])+(Ø-lō-VprefL);

(kī-yå̄ qūm bə-qirbəḵå̄ nå̄ ḇī ʾō ḥōlēm ḥălōm [wə-nå̄ ṯan ʾēlɛḵå̄ ʾōṯ ʾō mōp̄ ēṯ]
[ū-ḇå̄ hå̄ -ʾōṯ wə-ham-mōp̄ ēṯ ...]) (lō ṯišmaʿ ʾɛl diḇrē han-nå̄ ḇī ha-hū ʾō ʾɛl
ḥōlēm ha-ḥălōm ha-hū)
‘(If a prophet, or one who foretells by dreams, appears among you [and
announces to you a sign or wonder], [and the sign or wonder spoken of
takes place], ...) (then you must not listen to the words of that prophet or
dreamer).’ (Deut. 13:2-4)
132 For clarity, the protasis and the apodosis are enclosed within parentheses in a condi-
tional clause combining.

246 Bo Isaksson
In (111) the protasis is rather complicated with two affirmative addition claus-
es coded by we-Vsuff. The apodosis in this case is negative, ‘then you must
not listen ...’, and in this case there is no need of the innovative we-Vsuff
clause type. A lō-VprefL could be used without risk of confusion.133
C.2.1.4 A result connotation

When time is not the foremost semantic part of the pre-clause, a nuance of
result is often perceived in the we-Vsuff clause, also when there is no condi-
tional linking in the construction.
(112) Pattern: we-Snoun-Vsuff+we-Snoun-Vsuff+[we-Vsuff];

wə-yaʿăqōḇ šå̄ maʿ kī ṭimmē ʾɛṯ dīnå̄ ḇittō, ū-ḇå̄ nå̄ w hå̄ yū ʾɛṯ miqnēhū baś-
śå̄ ḏɛ [wə-hɛḥɛ̆rīš yaʿăqōḇ ʿad bōʾå̄ m]
‘Now Jacob heard that he had defiled his daughter Dinah. But his sons
were with his livestock in the field, [so Jacob held his peace until they
came].’ (Gen. 34:5; ESV)
In (112) Jacob is informed of the raping of his daughter Dinah, but because
his sons are in the fields he can do nothing for the moment. The result (we-
Vsuff wə-hɛḥɛ̆rīš ...) is that he keeps silent until they arrive.134
C.2.1.5 Legal cases

Legal cases often make use of a linking with one or several we-Vsuff clauses,
as in (113).
(113) Pattern: NP+[we-Vsuff+[Ø-Onoun-Vsuff]];

wə-ʿå̄ rēl zå̄ ḵå̄ r ʾăšɛr lō yimmōl ʾɛṯ bəśar ʿårlå̄ ṯō [wə-niḵrəṯå̄ han-nɛp̄ ɛš ha-
hī mē-ʿammɛhå̄ [ʾɛṯ bərīṯī hēp̄ ar]]
‘Any uncircumcised male who will not let himself be circumcised in the
flesh of his foreskin – [that person will be cut off from his people, [be-
cause he has broken my covenant]].’ (Gen. 17:14; CJB)
133 Other examples of conditional clause combining with we-Vsuff clauses involved: Gen.
27:45 (([PREP-VN]+[we-Vsuff])+(we-Vsuff);), 28:20-21 (ʾim-VprefL+[we-Vsuff]
+[we-Vsuff] +[we-Vsuff] +[we-Vsuff]), 32:9 ((ʾim-VprefL+[we-Vsuff])+(we-Vsuff)),
Exod. 4:8 ((ʾim-lō-VprefL + we-lō-VprefL) + we-Vsuff), 12:44 (([NCl + [we-Vsuff])
+ (ʾāz-VprefL)), 1 Sam. 11:3 ((ʾim-NCl-PA)+(we-Vsuff)).
134 Other examples of result nuance in a we-Vsuff clause are: Gen. 17:13
(VNabs+VprefL+[we-Vsuff]), 29:15 (INT-(kī-NCl + [we-Vsuff])), 30:15 (INT-NCl +
[we-Vsuff]), 47:22, Judg. 1:24, Isa. 6:7, Ps. 18:35 (PA+[we-Vsuff]).

(113) is a legal case, the case with the uncircumcised man. We can interpret
the initial noun phrase (NP) with Carl Brockelmann as an ‘eingliedrige Nomi-
nalsats’ (Brockelmann 1956 § 13). It is an imagined situation when an uncir-
cumcised man appears. Even if the phrase is short, typical for legal language,
it is a predication. And after this pre-clause comes the we-Vsuff clause, cod-
ing the action that should accompany the case with the uncircumcised man:
‘on that occasion the person must be cut off’. We could of course translate:
‘an uncircumcised man should be cut off’, and that would give good meaning
and possibly accord well with English legal language. But this is not the way
Biblical Hebrew codes the case. First we have the case, and then by we-Vsuff
clauses is stated the punishment which should or must accompany that case.135
C.2.2 we-Vsuff clauses encoding instruction or procedure

we-Vsuff clauses are used extensively in instruction and procedure, and in
such instances the chain of we-Vsuff usually constitutes a main line. 136 A
simple example of instruction is one of the ten commandments, where the we-
Vsuff is seemingly not a main clause, is (114).
(114) Pattern: ADV-VprefL+[we-Vsuff];

šēšɛṯ yå̄ mīm taʿăḇoḏ [wə-ʿå̄ śīṯå̄ kål məlaḵtɛḵå̄ ]
‘Six days you shall labor [and at that do all your work].’ (Exod. 20:9)
The initial commandment is complemented by another clause coding an ac-

tion that should accompany the first. The we-Vsuff clause is a commandment
in this context, but it is not something that is expected to occur after the six
days or as a result of the six days. Nor is the we-Vsuff clause a purpose
clause. The action of the we-Vsuff is to be performed during the six days. The
we-Vsuff clause means that on the six days all work must be done.137
135 The non-main Ø-Onoun-Vsuff clause is not our main concern in the present section. It
codes a circumstantial clause giving the reason for the punishment (‘since he has ...’),
see C.3.1.
136 I am not concerned, here, with all possible types of discourse that make a conspicuous
use of we-Vsuff clauses, such as procedural, hortatory, expository, or instructional dis-
course. There are semantic differences, and they certainly represent different Sitz im
Leben, but the syntactic are less dramatic. In SBH they can be subsumed under the
heading ‘discourse type with a main line of affirmative we-Vsuff clauses’.
137 Long chains of main line we-Vsuff clauses are found in the instruction on the building
of the tabernacle, Exod. 25-26.

248 Bo Isaksson
(115) Pattern: 10we-Vsuff1 + [Ø-NCl+ we-NCl + we-NCl] 11 + we-Vsuff2 +

[Ø-ADV-VprefL] + we-Vsuff3 + 12 we-Vsuff4 + we-Vsuff5 + [we-NCl] +
[we-NCl]+ 13 + we-Vsuff6 + we-Vsuff7;
wə-ʿå̄ śū1 ʾărōn ʿăṣē šiṭṭīm
[ʾammå̄ tayim wå̄ -ḥēṣī ʾårkō] [wə-ʾammå̄ wå̄ -ḥēṣī råḥbō] [wə-ʾammå̄
wå̄ -ḥēṣī qōmå̄ ṯō] 11
wə-ṣippīṯå̄ 2 ʾōṯō zå̄ hå̄ ḇ ṭå̄ hōr
[mib-bayiṯ ū-mi-ḥūṣ təṣappɛnnū]
wə-ʿå̄ śīṯå̄ 3 ʿå̄ lå̄ w zēr zå̄ hå̄ ḇ så̄ ḇīḇ 12
wə-yå̄ ṣaqtå̄ 4 lō ʾarbaʿ ṭabbəʿōṯ zå̄ hå̄ ḇ
wə-nå̄ ṯattå̄ 5 ʿal ʾarbaʿ paʿămōṯå̄ w
[ū-šətē tabbå̄ ʿōṯ ʿal ṣalʿō hå̄ -ʾɛḥå̄ ḏ]
[ū-šətē tabbå̄ ʿōṯ ʿal ṣalʿō haš-šēnīṯ] 13
wə-ʿå̄ śīṯå̄ 6 ḇaddē ʿăṣē šiṭṭīm
wə-ṣippīṯå̄ 7 ʾōṯå̄ m zå̄ hå̄ ḇ
‘They shall make1 an ark of acacia wood [its length being two and a half
cubits and its width a cubit and a half and its height a cubit and a half].
And you shall overlay2 it with pure gold [– overlaying it both inside and
outside –], and you shall put3 a molding of gold around it. You shall cast4
four rings of gold for it, and put5 them on its four feet, [and at that two
rings are on the one side of it], [and two rings are on the other side of it].
You shall make6 poles of acacia wood, and you shall overlay7 them with
gold.’ (Exod. 25:10-13)
(115) is a short quotation from a long instruction on how to make the taber-
nacle, “the most sacred and important object of Israel’s worship” (NET notes).
The instruction starts with an imperative in verse 2, dabber ʾɛl bənē yiśrå̄ ʾēl
wə-yiqḥū lī tərūmå̄ ‘Speak to the people of Israel, that they take for me a con-
tribution’ (ESV). The main line in (115) is made up of seven we-Vsuff clauses
detailing the different moments in the work. Syntactically, the we-Vsuff
clauses are additions that take over the apodictic nuance of the first clause in
verse 2, actions that should accompany the imperative. From the perspective
of the text, the we-Vsuff clauses make up the core of the instruction. Some of
the we-Vsuff clauses are qualified by (asyndetic) circumstantial clauses, and
in one occasion there is an addition of two noun clauses describing an accom-
panying state. The first, we-Vsuff1, clause in (115) states the making of the
ark of acacia wood, and it is followed by three circumstantial noun clauses
stating the length, breadth and height of this ark. The noun clauses are the

measures needed to perform the action in Vsuff1. Also after we-Vsuff2 (wə-
ṣippīṯå̄ ) more details are given about the action in Vsuff2, but this time the
extra information is given in a dynamic clause with a finite verb: an asyndetic
ADV-VprefL clause (see B.2.1.2). The attendant circumstantial meaning and
the finite verb could be translated ‘in that you overlay it both inside and out-
side’. Finally, after the fifth we-Vsuff clause there follow two syndetic noun
clauses describing states that should accompany the we-Vsuff5, ‘and on that
occasion two rings are on one side, and two (other) rings on the other side’.
C.3 Vsuff clauses encoding a non-main clause

Under this heading non-main Vsuff clauses will be treated that are not of the
we-Vsuff type (C.2).
C.3.1 The asyndetic Vsuff clause (type Ø-X-Vsuff or Ø-Vsuff)138

The asyndetic Vsuff clause in non-main position is basically a clause describ-
ing an action concomitant with the main clause. Such a clause exhibits the
familiar properties of the Vsuff gram. Usually, in narrative prose, the meaning
of the gram is perfective, but also anterior meaning is possible. As usual, a
clausal constituent before the verb is topicalized (the ‘Ø-X-Vsuff’ type of
clause).
An attendant clause with the Vsuff gram may have the functions of reason,
comment, even elaboration. As an attendant clause, it may also code a tem-
poral clause. An attendant perfective clause is found in (116).
(116) Pattern: wa-VprefS+[Ø-PrP-Vsuff]+[we-PrP-Vsuff];

way-ḥappeś [bag-gå̄ ḏōl hēḥel] [ū-ḇaq-qå̄ ṭōn killå̄ ]
‘And he searched, [(in that he) began with the eldest] [and ended with the
youngest].’ (Gen. 44:12)
The ESV marks the non-main status of the Vsuff clauses by English ing-
forms, ‘And he searched, [beginning with the eldest] [and ending with the
youngest]’ (Isaksson 2009, 19-21), but this translation fails to account for the
aspect of the Vsuff verbs. A translation ‘(in that) he began with the eldest...’
would be more precise with its perfective aspect. The concomitant circum-
stantial action is perfective, and describes the same action as in the wa-VprefS
138 Here are discussed instances where X is a topicalized element, but not when X is a
negation.

250 Bo Isaksson
clause, only in more detail. The example shows that an asyndetic Vsuff clause
can be analysed as an elaboration in SBH.139
A corresponding negative clause usually has the negation lō. In such a
clause a topicalized element is placed before the negation, which is positioned
immediately before the verb. With the negation the perfective aspect is neu-
tralized and the action itself is in focus, as in (117).
(117) Pattern: wa-VprefS+[Ø-Onoun-lō-Vsuff]+[we-Onoun-lō-Vsuff];

wa-yhī šå̄ m ʿim-YHWH ʾarbå̄ ʿīm yōm wə-ʾarbå̄ ʿīm laylå̄ [lɛḥɛm lō ʾå̄ ḵal]
[ū-mayim lō šå̄ ṯå̄ ]
‘Moses was there with the LORD forty days and forty nights [without eat-
ing bread] [or drinking water].’ (Exod. 34:28; NIV)
The advantage of using the English ing-form, as NIV does, is that it renders
the circumstantial clause combining also in an English translation, but ESV
with its ‘He neither ate bread nor drank water’, is a more faithful rendering of
the perfective aspect.140
The anterior aspect of an asyndetic Vsuff clause is illustrated well in (118).
(118) Pattern: we-NCl+[Ø-Vsuff];

wə-ʾaḇrå̄ hå̄ m wə-śå̄ rå̄ zəqēnīm bå̄ ʾīm bay-yå̄ mīm [ḥå̄ ḏal li-hyōṯ lə-śå̄ rå̄
ʾoraḥ kan-nå̄ šīm]
‘Now Abraham and Sarah were old, advanced in years. [The way of wom-
en had ceased to be with Sarah].’ (Gen. 18:11; ESV)
The asyndetic Vsuff clause in (118), without topicalized element, supplies a

detail that concerns only one actant in the main clause (NCl). It is not exactly
139 Other examples of attendant perfective asyndetic Vsuff clauses, some of which func-
tion as elaborations, are: Gen. 1:27 (wa-VprefS+[Ø-PrP-Vsuff]+[Ø-ADV-Vsuff]),
13:12 (wa-VprefS+[Ø-Snoun-Vsuff]+[we-Snoun-Vsuff]), 25:18 (wa-VprefS+[Ø-PrP-
Vsuff]), 31:41 (NCl+[Ø-Vsuff+{wa-VprefS}]), 41:11 (wa-VprefS + [PrP-Vsuff]),
41:48 (wa-VprefS+[Ø-Onoun-Vsuff]), Exod. 8:13 (wa-VprefS+[Ø-Snoun-Vsuff]), 1
Kgs 18:6 (wa-VprefS+[Ø-Snoun-Vsuff]+[we-Snoun-Vsuff]), 2 Kgs 3:3 (wa-VprefS +
[Ø-ADV-Vsuff]+[Ø-NEG-Vsuff]). An asyndetic Vsuff comment clause is found in
Gen. 4:20 (wa-VprefS+[Ø-Spron-Vsuff]).
140 Another negative asyndetic example is Exod. 8:27 (wa-VprefS+[Ø-NEG-Vsuff]).

an elaboration, but an attendant circumstance with anterior (pluperfect) aspect.

It could be accurately translated ‘at which the way of women had ceased...’.141
It is not necessary that the asyndetic Vsuff clause is placed after the main
clause, though this is the most common case. An example of a pre-posed non-
main Vsuff clause with anterior aspect is (119).
(119) Pattern: [Ø-Vsuff]+Ø-VprefA;

[hɛḥɛ̆yītå̄ nū] nimṣå̄ ḥēn bə-ʿēnē ʾăḏōnī
‘[You have saved our lives]; may it please my lord, we will be servants to
Pharaoh.’ (Gen. 47:25; ESV)
The pre-posed asyndetic Vsuff clause supplies the reason for the wish ex-
pressed by the Vpref clause, so it could be translated ‘[Since you have saved
our lives] let us find favour before my lord!’ (or. ‘let us be grateful ...’).142
C.3.2 The syndetic X-Vsuff clause

Clauses with an initial we and a clausal constituent before the verb are addi-
tions because of the we, and have a topicalized element (‘X’).143 This kind of
clauses has not undergone the same broadening of the semantic spectrum as
has we-Vsuff clauses.144 What characterizes this kind of addition clauses is the
topicalized fronted element which can be utilized to achieve contrast with a
constituent in the main clause, or, if not contrast, an impression of mutuality:
A, and/but also B. Such constructions are often close to what Dixon terms
‘unordered additions’, which involves “two distinct events which are semanti-
cally or pragmatically related but for which no temporal sequence is assumed”
(2009, 26, 28). The action in the we-X-qatal clause may code a less important,
less salient action which just accompanies the main action, as in (120).
(120) Pattern: wa-VprefS+[we-Onoun-Vsuff];

way-yēṣē ʾălēhɛm lōṭ hap-pɛṯḥå̄ [wə-had-dɛlɛṯ så̄ ḡar ʾaḥărå̄ w]
141 Other anterior examples are: Gen. 43:23 (NEG-VprefS+[Ø-Snoun-Vsuff]+[Ø-Snoun-

Vsuff]), Exod. 14:3 (NCl-PP + Ø-Vsuff), Deut. 32:6 (NCl + [Ø-Spron-Vsuff + {wa-
VprefS}]).
142 Other examples with preposed asyndetic Vsuff: Deut. 32:21 ([Ø-Spron-Vsuff]+[Ø-
Vsuff] + we-Spron-VprefA), 32:39 ([Ø-Vsuff]+we-Spron-VprefL).
143 I do not include negations among the ‘X’ elements. Negative syndetic clauses require a
separate treatment, in C.3.3.
144 And this because there was no need to. we-X-VprefL clauses were retained in SBH.

252 Bo Isaksson
‘But Lot went out to them at the doorway, [and shut the door behind him].’
(Gen. 19:6; NAS)
Both clauses have a perfective aspect, but it is not evident that the Vsuff
clause codes a temporal sequence. The core of wa-VprefS clauses in the story-
line has an implied iconic sequentiality, which is the default interpretation of
the addition clauses in the storyline. When this chain is interrupted by an
additive we-X-Vsuff clause – one of the most frequent digressions from the
storyline in SBH narrative prose – then the default iconic chain is broken, and
the we-X-Vsuff is felt as an accompanying action for which a temporal se-
quence is not assumed.
In most cases the topicalization of the X-element is more salient than in
(120). This is the case in (121), where two objects are emphasized, and at the
same time the clause signals an action that is separate from that in the main
clause.
(121) Pattern: NCl+[we-Onoun-NEG-Vsuff];

ʾiššå̄ qəšaṯ rūăḥ ʾå̄ nōḵī [wə-yayin wə-šēḵå̄ r lō šå̄ ṯīṯī]
‘I am a woman oppressed in spirit; [I have drunk neither wine nor strong
drink].’ (1 Sam. 1:15; NAS)
The example is from direct speech, and the main clause in this case is a noun
clause. The addition has anterior aspect and the wine and the spirit are topical-
ized because the priest Eli has accused Hannah of being drunk. The topicaliza-
tion would have been better rendered in English with ‘neither wine nor strong
drink have I drunk’.
Very typical for narrative prose is that the topicalized element creates an
impression of complementation. The Vsuff clause with the topicalized ele-
ment is a complementary action in (122).
(122) Pattern: wa-VprefS+[we-Onoun-Vsuff];

way-yɛʾsor ʾɛṯ riḵbō [wə-ʾɛṯ ʿammō lå̄ qaḥ ʿimmō]
‘So he made ready his chariot [and took his army with him].’ (Exod. 14:6;
ESV)

Both verbal grams are perfective in (122), and a temporal sequence is not
assumed.145
There may even be a complete mutuality between topicalized subjects as in
the same-event addition of (123), which describes two actions that are con-
comitant (Lunn 2006, 48: “Parallel focus”).146
(123) Pattern: wa-VprefS+wa-VprefS+[we-Snoun+Vsuff];

way-yippol ʿal ṣawwərē ḇinyå̄ mīn ʾå̄ ḥīw way-yeḇk [ū-ḇinyå̄ mīn bå̄ ḵå̄ ʿal
ṣawwå̄ rå̄ w]
‘Then he fell upon his brother Benjamin’s neck and wept, [and Benjamin
wept upon his neck].’ (Gen. 45:14; ESV)
Neither sequentiality nor simultaneity are necessarily assumed in an addition

clause. But as (123) shows, in a certain pragmatic context the Vsuff clause
may be understood as an attendant action, which is illustrated also in (124).
(124) Pattern: wa-VprefS+[we-Snoun-Vsuff];

way-yaʿaś yəhōšūăʿ ka-ʾăšɛr ʾå̄ mar lō mōšɛ lə-hillå̄ ḥēm ba-ʿămå̄ lēq [ū-
mōšɛ ʾaḥărōn wə-ḥūr ʿå̄ lū rōš hag-giḇʿå̄ ]
‘So Joshua did as Moses told him, and fought with Amalek, [while Moses,
Aaron, and Hur went up to the top of the hill].’ (Exod. 17:10; ESV)
We understand that Moses, Aaron, and Hur went up the hill simultaneously
with Joshua’s fighting with Amalek. The rendering ‘while’ in the ESV trans-
lation is not explicitly supported by the syntax, but in the pragmatic circum-
stances it is justified.
The topicalized element in the Vsuff clause very often contrasts with a
constituent in the main clause (Lunn 2006, 48), as in (125).
145 Similar examples of an accompanying action by a we-X-Vsuff clause are: Gen. 2:20
(wa-VprefS+[we-PrP-lō-Vsuff]), 14:16 (wa-VprefS + [we-Onoun-Vsuff]), 15:10 (wa-
VprefS + wa-VprefS + [we-Onoun-lō-Vsuff]), 18:33 (wa-VprefS+[we-Snoun-Vsuff]),
24:46 (wa-VprefS + [we-gam-Onoun-Vsuff]), 24:53 (wa-VprefS+ wa-VprefS + [we-
Onoun-Vsuff]), 26:15 (wa-VprefS + [we-Onoun-Vsuff + {wa-VprefS}]), 27:16 (wa-
VprefS + [we-Onoun-Vsuff]), 28:16 (NCl+[we-Spron-Vsuff]), 33:17 (wa-VprefS-
PrP+[we-PrP-Vsuff]), 37:36 (wa-VprefS+[we-Snoun-Vsuff]), 39:4 (wa-VprefS+[we-
Onoun-Vsuff]), 43:15 (wa-VprefS + [we-Onoun-Vsuff]), 43:22 (wa-VprefS + [we-
Onoun-Vsuff + [Ø-NEG-Vsuff]]), 47:21 (wa-VprefS+[we-Onoun-Vsuff]).
146 Cf. the “canonical instance of Same-event Addition” mentioned by Dixon (2009, 27):
“You are together with me; (and) as for me, I am together with you”.

254 Bo Isaksson
(125) Pattern: wa-VprefS+[we-Snoun-Vsuff];

wa-yhī hɛḇɛl rōʿē ṣōn [wə-qayin hå̄ yå̄ ʿōḇēḏ ʾăḏå̄ må̄ ]
‘Abel took care of the flocks, [while Cain cultivated the ground].’ (Gen.
4:2; NET)
The topicalized Cain contrasts with Abel, and sequentiality is not assumed. It
is not expressly stated, but the two situations are implied to go on simultane-
ously. The main point in the text is not the temporal relation between the
events, but the contrasting activities of Abel and Cain.147
If sequentiality is to be stressed and a sequential adverb be put in topical-
ized position, then a Vsuff clause must be utilized in the storyline (thus
Joosten 2012, 43), as in (126), since a we-ADV-VprefS clause would have
been misunderstood as a VprefL clause.
(126) Pattern: wa-VprefS1-"..." + wa-VprefS2 + *we-ADV-Vsuff1-Snoun +

wa-VprefS3 + *we-Snoun-Vsuff2 + wa-VprefS4;
wat-tōmɛr1 "ma på̄ raṣtå̄ ʿå̄ lɛḵå̄ på̄ rɛṣ!"
way-yiqrå̄ 2 šəmō på̄ rɛṣ
*wə-ʾaḥar yå̄ ṣå̄ 1 ʾå̄ ḥīw ʾăšɛr ʿal yå̄ dō haš-šå̄ nī
way-yiqrå̄ 3 šəmō zå̄ raḥ
*wə-yōsēp hūrad2 miṣrāymā
way-yiqnē4-hū pōṭīpar sərīs parʿō
‘And she said1, "What a breach you have made for yourself!" Therefore
his name was called2 Perez *Afterward his brother came out1 with the
scarlet thread on his hand, and his name was called3 Zerah. *Now Joseph
was taken down2 to Egypt, and Potiphar, an officer of Pharaoh, bought4
him’ (Gen. 38:29–39:1)148
147 Other similar contrast clauses with Vsuff are: Gen. 1:10 (wa-VprefS+[we-PrP-Vsuff]),
4:4 (wa-VprefS+[we-Snoun-Vsuff]), 4:5 (wa-VprefS+[we-PrP-Vsuff]), 25:6 (wa-
VprefS + [we-PrP-Vsuff]), 25:34 (wa-VprefS + wa-VprefS + [we-Snoun-Vsuff]), 27:6
(wa-VprefS+[we-Snoun-Vsuff]), 29:17 (we-NCl+[we-Snoun-Vsuff] ‘Leah's eyes were
weak, but Rachel was beautiful in form and appearance’ ESV), 31:5 (NCl-PA+[we-
Snoun-Vsuff]), 31:47 (wa-VprefS+[we-Snoun-Vsuff]), 32:2 (wa-VprefS+ [we-Snoun-
Vsuff]), 33:17 (wa-VprefS+[we-Snoun-Vsuff]), 35:18 (wa-VprefS+[we-Snoun-
Vsuff]), 37:11 (wa-VprefS+[we-Snoun-Vsuff]), 41:54 (wa-VprefS+[we-PrP-Vsuff]),
Exod. 9:6 (wa-VprefS + [we-PrP-lō-Vsuff]), 9:23 (wa-VprefS + [we-Snoun-Vsuff]
Moses acts on his part and YHWH acts on his part), 1 Sam. 1:22 (wa-VprefS+[we-
Snoun-Vsuff]), 2 Kgs 3:22 (wa-VprefS+[we-Snoun-Vsuff]). Topicalized elements
without contrast: Exod. 9:25 (wa-VprefS+[we-Onoun-Vsuff]+[we-Onoun-Vsuff]).
148 The index numbers refer to sequences of verbs.

A wa-VprefS addition often has a nuance of consequence (related to the pre-

vious clause), which is the case in VprefS2 in (126). This is the reason for the
translation ‘Therefore’ (thus ESV). The we-X-qatal clause is an addition, but
it becomes expressly sequential by the ʾaḥar ‘thereafter’. The first Vsuff
clause with topicalized adverb ʾaḥar takes part in the storyline with a special
emphasis on the sequentiality of the clause, which could not have been ex-
pressed by only a default wa-VprefS.
The second Vsuff clause has a topicalized subject personal noun (Joseph),
and this topicalization re-introduces an actant that has been absent in the nar-
rative during chapter 38 of Genesis. Joseph is now returned to the scene, and
this could not have been done with just a wa-VprefS clause (which can have
no topicalized element). Topicalized VprefS clauses were available no more
in the SBH narrator’s linguistic outfit. Instead, in some instances, Vsuff claus-
es had entered the storyline. What we call a ‘break’ is the result of a new
actant being introduced by a topicalized element in a Vsuff clause (cf. Isaks-
son 2009, 122).149
In sum, the we-X-Vsuff type of clause seems to have retained its mainly non-
sequential character in relation to a SBH storyline. It is most often an accom-
panying action that breaks the default sequentiality of the chain of wa-VprefS
clauses. But there are examples of an intrusion of we-X-Vsuff clauses into the
storyline, which is only to be expected.150 And such is even more the case with
the we-lō-Vsuff clauses that will be discussed in the next section.
149 It has been argued that the meaning of the Vsuff is actually pluperfect. Westermann
(1982, 53) translates (with reference to Ges-K § 142b) “Als Joseph nach Ägypten ge-
bracht worden war, kaufte ihn Potiphar, der Kämmerer des Pharao” (Ges-K argues that
the pluperfect meaning is the result of the preposed subject). This is absolutely possi-
ble. Anterior is one of the expected meanings of a resultative gram like the Vsuff. But
also perfective aspect is. If the meaning is pluperfect, the clause is background and not
part of the storyline. The problem is that this pluperfect is not explicitly stated, it is on-
ly implied from the pragmatic context. And a perfective meaning gives good meaning
as well to the storyline.
150 An example is the special ‘reportive’ narrative style used in the account of the building
of the tabernacle, where the wa-VprefS clauses are often not perceived as by default
expressing temporal succession. In such cases a we-X-Vsuff clause becomes part of the
main line, only that there is a topicalized ‘X’ element in the clause, Exod. 36:33-34.

256 Bo Isaksson
C.3.3 The syndetic lō-Vsuff clause

The indicative VprefS clause was no longer negated in SBH. In the new re-
stricted word order a *we-lō-VprefS clause could be mistaken for a long pre-
fix verb clause. We can expect that a restriction of a narrative main line to just
affirmative clauses was unacceptable, and that the coding of negative storyline
clauses underwent a renewal. The natural new negative candidate to enter the
storyline was the we-lō-Vsuff clause, with perfective aspect (Joosten 2012,
42). When we examine the cases of such clauses, this turns out to be true. To a
greater extent than the we-X-Vsuff type, the we-lō-Vsuff clauses have entered
the SBH narrative storyline and complemented the affirmative wa-VprefS
clauses with corresponding negative clauses.
A simple and clear example of a we-lō-Vsuff in storyline is (127).
(127) Pattern: wa-VprefS+we-lō-Vsuff;

wa-yḥazzeq YHWH ʾɛṯ lēḇ parʿō wə-lō šillaḥ ʾɛṯ bənē yiśrå̄ ʾēl
‘But the LORD hardened Pharaoh’s heart, and he did not let the people of
Israel go.’ (Exod. 10:20; ESV)
In (127) the Lord hardens the heart of Pharaoh, and the result is that he does
not let the people go. Such a result is a typical function of the implied default
sequentiality of a storyline. Both clauses are perfective with an implied past
time reference.
In the next example (128) the actions are not inner attitudes and decisions,
as in (127), but concrete physical events.
(128) Pattern: wa-VprefS+*we-lō-Vsuff+[kī-NCl];

way-yå̄ ḇōʾū må̄ rå̄ ṯå̄ *wə-lō yå̄ ḵəlū lištōt mayim mim-må̄ rå̄ [kī-må̄ rīm hēm]
‘They came to Marah, *but they could not drink the water at Marah [be-
cause it was bitter].’ (Exod. 15:23)
A clear temporal succession is perceived between the wa-VprefS clause and

the Vsuff clause. After they arrived at Marah, they found that they could not
drink the water. It is a result of observation and many translations have tried
to manage the clausal relation by making the wa-VprefS a subordinate tem-
poral clause, ‘When they came to Marah, they could not drink its water’.151
This accounts for the temporal succession but does not faithfully render the
storyline wa-VprefS clause way-yå̄ ḇōʾū må̄ rå̄ ṯå̄ ‘They came to Marah’.
151 Thus NIV, and similarly JPS, ESV, NRS, NAS, NJB, RSV.

The examples abound, but only one more will be given here, the one about
the dove which was sent out by Noah, in (129).152
(129) Pattern: wa-VprefS+*we-lō-Vsuff +wa-VprefS;

wa-yšallaḥ ʾɛṯ hay-yōnå̄ mē-ʾittō lirʾōṯ hă-qallū ham-mayim mē-ʿal pənē
hå̄ -ʾăḏå̄ må̄ *wə-lō må̄ ṣəʾå̄ hay-yōnå̄ må̄ nōăḥ lə-ḵap̄ raḡlå̄ h wat-tå̄ såḇ
ʾēlå̄ w ʾɛl hat-tēḇå̄
‘Then he sent out a dove to see if the water had receded from the surface
of the ground. *But the dove could not find a resting place for its foot, and
returned to him to the ark.’ (Gen. 8:8-9)
In translating the we-lō-Vsuff clause many English versions prefer to replace

the negative clause with a positive one, such as ‘but the dove found no resting
place...’ (CSB). It is understood that the dove searched with no result. The
example shows how an affirmative wa-VprefS clause continues the storyline
after the we-lō-Vsuff.153
Besides the syndetic negative Vsuff clauses that have entered the Hebrew
storyline, there is of course the usual kit of clausal relations where the we-lō-
Vsuff type is involved. Whether the we-lō-Vsuff clause is to be considered of
equal status or not mainly depends on the pre-clause being a Vsuff clause or
not (examples (130) – (132)). When a we-lō-Vsuff clause follows a Vsuff pre-
clause it is usually a clause with equal status that codes an accompanying
action, with meanings that are characteristic of the Vsuff gram, (130).
(130) Pattern: Ø-ADV-Vsuff+*we-lō-Vsuff;

zɛ šå̄ lōš pəʿå̄ mīm hēṯaltå̄ bī *wə-lōʾ-higgaḏtå̄ lī bammǣ kōḥăḵå̄ gå̄ ḏōl
‘These three times you have teased me *and you have not told me.’ (Judg.
16:15)
152 The asterisk marks the discussed new clause when it is part of the main line.
153 Other examples of storyline we-lō-Vsuff clauses in narrative prose: Gen. 26:22,
31:33.34, 38:20, 39:6, 40:23, 45:1, Exod. 1:17, 2:3 (we-lō-Vsuff+wa-VprefS), 6:9,
7:13 (and similar examples in Exod. 7:22, 8:15, 8:28, 9:7, 9:12, 10:27), 7:21.23, 8:14,
9:11, 10:15.27, 11:10, 13:17, 14:20, 15:22, 16:18.20.24, Judg. 2:23, 3:28, 6:10, 8:20.
28.34.35 (wa-VprefS + we-lō-Vsuff + we-lō-Vsuff), 10:6, 11:17.18, 13:21, 14:6.9,
15:1, 16:9, 19:10 (similar examples 19:25, 20:13), 2 Kgs 3:26, Job 2:12.

258 Bo Isaksson
In this direct speech of Delilah she accuses Samson of having teased her. The
meaning is anterior in both Vsuff clauses, and we have no reason to regard the
negative clause as non-main.154
With another type of pre-clause, the we-lō-Vsuff clause will be of non-
equal status and codes a non-main accompanying action. In (131) the pre-
clause is an infinite clause.
(131) Pattern: PREP-VN+[we-lō-Vsuff];

ʿal hasgīrå̄ m gå̄ lūṯ šəlēmå̄ lɛ-ʾɛ̆ḏōm [wə-lō zå̄ ḵərū bərīṯ ʾaḥīm]
‘...Because she sold whole communities of captives to Edom, [and at that
disregarded a treaty of brotherhood].’ (Amos 1:9)
The passage from Amos describes the reasons for God’s judgment, and one
reason is coded by an infinitive construct and a following accompanying ac-
tion expressed by a Vsuff verb. The two clauses describe the same event, and
the Vsuff clause in this case can be argued to have perfective meaning (as in
the translation), but anterior is also possible.155
In (132) the pre-clause is a NCl and the following we-lō-Vsuff clause is
likewise of unequal status, this time with anterior meaning.
(132) Pattern: hinnē-NCl+ [we-lō-Vsuff]+[we-Vsuff]+[we-Vsuff];

hinnē-nå̄ ʾat ʿăqå̄ rå̄ [wə-lō yå̄ laḏt] [wə-hå̄ rīṯ] [wə-yå̄ laḏt bēn]
‘Look here! You are barren [and have not borne children], [but you will be
pregnant], [and will bear a son].’ (Judg. 13:3)
The example is a direct speech of an angel of God to the wife of Manoah. It

starts with a well-known fact for the woman; she is barren and has not borne
children. In this case, the we-lō-Vsuff clause amounts to the same as an elabo-
ration, it describes a state (the NCl) with an accompanying negated dynamic
action (‘you have not borne’).156 Of the four clauses in (132) the first two
154 Similar examples, all with Vsuff clauses and anterior meaning, are: Gen. 13:5.6, 22:12
(with implied causality), Ps. 18:22.
155 There are many similar linkings in Amos. Another is found in Gen. 42:21 (REL +
Vsuff +[Ø-PREP-VN+[we-lō-Vsuff]]) where the Vsuff has perfective meaning.
156 The following two we-Vsuff clauses code accompanying actions that illustrate how the
we-Vsuff clause type has taken over meanings from the no longer acceptable *we-
VprefL clause. The two we-Vsuff clauses express a promise with future time reference.

represent syntactical retentions (NCl Protosemitic, Vsuff Westsemitic), while

the we-Vsuff clauses represent an innovation in SBH.157
C.4 Summary
The Vsuff gram exhibits meanings that are typical of grams of the resultative
trajectory: stative, resultative, anterior, perfective. Diachronically, the Vsuff
increasingly competes with the older VprefS as a perfective gram in past tense
narratives. Typically, the more expressive Vsuff gram introduces a story and
is then followed by a chain of less expressive wa-VprefS clauses.
The we-Vsuff clause in SBH has taken over the functions of the VprefL
gram when there is no need of a topicalized element or a negative clause. This
is the reason why we-Vsuff clauses often exhibit meanings that are close to
those of the VprefL gram. The we-Vsuff is an addition, and as such it can
also, depending on the context, express a result meaning after a conditional
clause.
Other Vsuff clauses, for example we-lō-Vsuff or we-X-Vsuff, have re-
tained meanings that are characteristic of the old Vsuff gram.
We-lō-Vsuff clauses may function as negated storyline clauses, and as
such they supplant otherwise expected syndetic *lō-VprefS clauses which are
unacceptable in Biblical Hebrew due to their being non-clause-initial.
Conclusion
In the present article I attempt to understand the Hebrew Verbal system in the
light of how clauses are linked Biblical Hebrew texts. It is a study based on
raw data collected in a database on linked clauses from prose and poetry in
Standard Biblical Hebrew (SBH) and archaic Hebrew poetry. The perspective
is comparative. Biblical Hebrew is expected to behave as a descendant of
early Canaanite.
A central text-linguistic concept for the analysis in this article is the ‘main
line’. A main line is a text-linguistic concept defined by the pragmatics of the
text. A main line, thus, is a pragmatic concept, but in the specific language it
is coded by specific syntactic patterns. This means that a non-main clause,
157 A similar linking with we-NCl+[we-lō-Vsuff] is found in the previous verse (Judg.
13:2). Isa. 53:3 has a PA+[we-lō-Vsuff] linking in a difficult context (Isaksson 2011).

260 Bo Isaksson
encoding a break from the main line, will also be syntactically marked as a
digression from the main line.
The basic function of the conjunction we/wa is to signal a clause as an ad-
dition in the sense of being an accompanying action in relation to a preceding
clause. In an accompanying action the event time is dependent on that in the
previous clause (the ‘pre-clause’). The conjunction we/wa is not an exclusive-
ly ‘coordinating’ conjunction. The pre-clause and the accompanying action
clause are sometimes of equal status, sometimes of unequal status. If they are
of equal status, they often (but not always) have the same type of predicate
(the same ‘gram’).
The investigation concerns the three basic finite verbal grammatical mor-
phemes (grams) in SBH, the suffix verb (Vsuff), the short prefix verb
(VprefS) and the long prefix verb (VprefL), and how they behave in clause
combining. They are the grams that are most extensively discussed in Hebrew
scholarship.
When the morphological distinction between the VprefS and the VprefL
grams was (partially) lost in Proto-Hebrew, word order became the decisive
strategy to cope with the risk of confusion. The old yaqtul (VprefS) was as-
signed clause-initial position, and the old yaqtulu (VprefL) was placed in non-
initial position. This strategy concerned the affirmative clauses. In negative
clauses, word order has remained free because of a specialization of the two
basic negations. The negation ʾal was confined to the VprefS with jussive
meaning, while the negation lō was used in other cases. This means that an
indicative VprefS could no longer be negated. In its stead, we-lō-Vsuff claus-
es were introduced into a narrative storyline.
The widespread idea that the wa-yiqtol syntagm is a conjugation of its
own, and that the wa-VprefS cannot be ‘split up’, is an ‘optical illusion’.
Word order constraints have replaced negative wa-VprefS with we-lō-Vsuff,
and *we-X-VprefS with we-X-Vsuff. There is only one VprefS gram, and this
gram may occur in three types of affirmative clauses, Ø-VprefS (modal or
indicative), we-VprefS (modal/purposive), and wa-VprefS (indicative).158
The wa-VprefS clause codes an indicative accompanying action. Frequent
inferred functions of wa-VprefS clauses are elaboration, temporal succession,
158 Contrary to the scheme in Joosten (2012, 39) which excludes the modal (jussive)
VprefS and implies that verbal grams in SBH are either indicative (wa-VprefS and
Vsuff) or modal (VprefL and we-Vsuff).

and result. An asyndetic VprefL clause codes a circumstantial event. The non-
main Ø-lō-VprefL clause codes a circumstantial action without a topicalized
element, the non-main Ø-X-VprefL clause has a topicalized element ‘X’.
Syndetic clauses of the types we-lō-VprefL or we-X-VprefL connect as addi-
tions to a previous clause.
The *we-VprefL clause, affirmative and with no topicalized element, had
become unacceptable in Biblical Hebrew because the prefix verb in this case
was clause-initial. It was replaced by the emerging we-Vsuff clause type,
which was also affirmative and lacked topicalized element. This is the reason
why innovative we-Vsuff clauses gradually ‘took over’ meanings typical of
the VprefL gram.
The wide-spread idea that a we-Vsuff clause (with close upon imperfective
meaning) could not be ‘split-up’ is an ‘optical illusion’. There was no need of
a split-up. Equal status clauses of the types we-lō-VprefL and we-X-VprefL
were already available. They represent a syntactical retention in Biblical He-
brew. It is the we-Vsuff that is the new formation. In the main line of proce-
dural discourse, we-Vsuff clauses interact with equal status we-lō-VprefL and
we-X-VprefL clauses.
The we-Vsuff clause type does not represent a separate gram in SBH. It is
a type of clause that expresses an affirmative accompanying action to a previ-
ous clause. As such, it is a clause type that exists in all Central Semitic lan-
guages. What is special to SBH in a comparative perspective is that the we-
Vsuff, due to the word order constraints, has taken over some of the functions
of syndetic imperfective VprefL clauses. The we-Vsuff is never a circumstan-
tial clause (because it is an addition).
A non-main asyndetic Vsuff clause describes a circumstantial action with
meanings typical of the Vsuff gram. It may be anterior, but also perfective.
The non-main we-X-Vsuff clause has not undergone the same broadening
of the semantic spectrum as has we-Vsuff clauses. In relation to a storyline of
wa-VprefS clauses, it usually breaks the default sequentiality of the chain of
wa-VprefS clauses and signifies an accompanying action with a topicalized
element, often with notions of contrast or mutuality or a complementary ac-
tion. It does not normally signal a temporal succession. The typical we-X-
Vsuff clause does not take part in the storyline. There are instances, though,
when the we-X-Vsuff takes part in the storyline, especially when X is an ad-
verb which signals temporal succession (such as ʾaḥar ‘afterward’).

262 Bo Isaksson
The we-lō-Vsuff clause has replaced the no longer acceptable *we-lō-

VprefS. To a greater extent than the we-X-Vsuff type, the we-lō-Vsuff clauses
have entered the SBH narrative storyline and complemented the affirmative
wa-VprefS clauses with a corresponding negative clause. Outside the storyline
the we-lō-Vsuff clause codes an accompanying action with the properties
typical of its predicate (the Vsuff gram).
References
Andersen, Francis I. 1974. The Sentence in Biblical Hebrew. Janua Linguarum 231.
The Hague.
Andersen, T. David. 2000. “The evolution of the Hebrew verbal system”. Zeitschrift
für Althebraistik 13 no. 1: 1-66.
Andrason, Alexander. 2010. “The panchronic yiqtol: Functionally consistent and cog-
nitively plausible?”. Journal of Hebrew Scriptures 10: 1-63.
———. 2011a. “The biblical Hebrew verbal system in light of grammaticalization –
the second generation”. Hebrew Studies 52: 19-51.
———. 2011b. “The biblical Hebrew verbal system in light of grammaticalization –
the second generation”. Hebrew Studies 53: 351-383.
———. 2011c. “The BH weqatal. A homogenous form with no haphazard functions
(Part 1)”. Journal of Nortwest Semitic Languages 37 no. 2: 1-26.
———. 2012. “The BH weqatal. A homogenous form with no haphazard functions
(Part 2)”. Journal of Nortwest Semitic Languages 38 no. 1: 1-30.
Bloch, Yigal. 2007. “From linguistics to text-criticism and back: wayyiqṭōl construc-
tions with long prefixed verbal forms in biblical Hebrew”. Hebrew Studies 48:
140-170.
———. 2009. “The prefixed perfective and the dating of early Hebrew poetry—A re-
evaluation”. Vetus Testamentum 58: 34-70.
Brockelmann, Carl. 1956. Hebräische Syntax. Neukirchen: Kreis Moers.
Bybee, Joan L., Revere D. Perkins, and William Pagliuca. 1994. The evolution of
grammar: Tense, aspect, and modality in the languages of the world. Chicago:
Univ. of Chicago Press.
Bybee, Joan L., and Östen Dahl. 1989. “The creation of tense and aspect systems in the
languages of the world”. Studies in Language 13 no. 1: 51-103.
CJB = Complete Jewish Bible.
Cook, John A. 2006. “The finite verbal forms in biblical Hebrew do express aspect.”
Journal of the Ancient Near Eastern Society 30: 21-35.

———. 2012. Time and the biblical Hebrew verb: The expression of tense, aspect,
and modality in biblical Hebrew. Linguistic studies in ancient West Semitic 7.
Cormack, Annabel, and Neil Smith. 2005. “What is coordination?” Lingua 115: 395-
418.
CSB = Holman Christian Standard Bible.
Dahl, Östen. 1985. Tense and aspect systems. Oxford: Basil Blackwell.
———. 2000. “The tense-aspect systems of European languages in a typological
perspective”. In Tense and aspect in the languages of Europe, edited by Östen
Dahl. 3-25. Empirical approaches to language typology (EUROTYP) 20:6. Berlin
and New York: Mouton de Gruyter.
Dixon, R. M. W. 2012. Basic linguistic theory. Vol. 3, Further grammatical topics.
Oxford: Oxford University Press.
Dixon, R. M. W., and Alexandra Y. Aikhenvald, eds. 2009. The semantics of clause
linking: A cross-linguistic typology. Explorations in linguistic typology 5. Oxford:
Oxford University Press. Reprint, paperback edition 2011.
Driver, Samuel Rolles. 1892. A treatise on the use of the tense in Hebrew and some
other syntactical questions. 3 rev. and improved ed. Oxford: Clarendon Press. Re-
print, Wipf & Stock: Eugene, Oregon, 2004.
———. 1913. Notes on the Hebrew Text and the Topography of the Books of Samuel
with an Introduction on Hebrew Palaeography and the Ancient Versions and Fac-
similes of Inscriptions and Maps. 2nd rev. and enlarged ed. Oxford: The Clarendon
Press.
ESV = The Holy Bible, English Standard Version.
Fischer, Wolfdietrich, and Otto Jastrow, eds. 1980. Handbuch der arabischen Dialek-
te. Porta Linguarum Orientalium N.S. 16. Wiesbaden: Harrassowitz.
Ges-K = Gesenius, Wilhelm, and Emil Kautzsch. 1910. Gesenius’ Hebrew grammar.
Translated by A. E. Cowley. Edited by Emil Kautzsch. 2nd revised English ed. Ox-
ford: Clarendon. Reprint, 1976.
Givón, Talmy. 2001. Syntax: An introduction. Rev. ed. 2 vols. Amsterdam, Philadelph-
ia: J. Benjamins.
Gzella, Holger. 2011. “Northwest Semitic in general”. In The Semitic languages: An
international handbook, edited by Stefan Weninger, Geoffrey Khan, Michael P.
Streck and Janet C. E. Watson. 425-451. Berlin: de Gruyter Mouton.
———. 2012. “Introduction” and “Ancient Hebrew”. In Languages from the world of
the Bible, edited by Holger Gzella, 1-13, 76-110. Boston/Berlin: De Gruyter.

264 Bo Isaksson
Haiman, John, and Sandra A. Thompson, eds. 1988. Clause combining in grammar
and discourse. Typological studies in language 18. Amsterdam; Philadelphia: John
Benjamins.
Halliday, Michael A. K. 2004. An introduction to functional grammar. Edited by
Christian M. I. M. Matthiessen. 3rd rev. ed. London: Arnold.
Haspelmath, Martin. 1998. “The semantic development of old presents: New futures
and subjunctives without grammaticalization”. Diachronica 15 no. 1: 29-62.
Hatav, Galia. 1997. The semantics of aspect and modality. Evidence from English and
biblical Hebrew. Studies in language companion series 34. Amsterdam, Philadel-
phia: J. Benjamins Pub.
Heller, Roy L. 2004. Narrative structure and discourse constellations: An analysis of
clause function in biblical Hebrew prose. Harvard Semitic studies 55. Winona
Lake, Ind.: Eisenbraun.
Hopper, Paul J. 1982. Tense-aspect: Between semantics & pragmatics: Containing the
contributions to a symposium on tense and aspect, held at UCLA, [Los Angeles],
May 1979. Typological studies in language (TSL) 1. Amsterdam: Benjamins.
Huehnergard, John. 1988. “The early Hebrew prefix-conjugations”. Hebrew Studies
29: 19-24.
———. 2005. “Features of Central Semitic”. In Biblical and Oriental essays in
memory of William L. Moran, edited by Agustinus Gianto. 155-203. Biblica et ori-
entalia 48. Roma: Editrice Pontificio Istituto Biblico.
Isaksson, Bo. 2007. “Semitic circumstantial qualifiers in the Book of Judges: A pilot
study on the infinitive.” Orientalia Suecana 56: 163-172.
———. 2009. “Introduction” and “An outline of comparative Arabic and Hebrew
textlinguistics”. In Circumstantial qualifiers in Semitic: The case of Arabic and
Hebrew, edited by Bo Isaksson, 1-35 and 36-150. Abhandlungen für die Kunde des
Morgenlandes 70. Wiesbaden: Otto Harrassowitz.
———. 2011. “The textlinguistics of the Suffering Servant: Subordinate structures in
Isaiah 52,13-53,12”. In En pāsē grammatikē kai sophiā. Saggi di linguistica ebrai-
ca in onore di Alviero Niccacci, ofm, edited by Gregor Geiger and Massimo Pazzi-
ni, 173-212. Collana Analecta: Studium Biblicum Franciscanum 78. Jerusalem;
Milano: Franciscan Printing Press; Editioni Terra Santa.
———. 2013. “Subordination: Biblical Hebrew”. In Encyclopedia of Hebrew Lan-
guage and Linguistics, edited by Geoffrey Khan, vol. 3, 657-664. Leiden: Brill.
———. 2014a. “Clause linking strategies in the narrative and instructional discourse
of Joseph’s speech in Gen. 45: 3-15”. Journal of Semitic Studies 59 no. 1: 15-45.

———. 2014b. “The main line of a biblical Hebrew narrative and what to do with two
perfective grams.” In Proceedings of the Oslo–Austin Workshop in Semitic Lin-
guistics, Oslo, May 23–24, 2013, edited by Lutz Edzard and John Huehnergard,
73-94. Abhandlungen für die Kunde des Morgenlandes 88. Wiesbaden: Harrasso-
witz.
———. 2014c. “Archaic biblical Hebrew poetry: The linking of finite clauses”. In
vik, Sweden, edited by Bo Isaksson and Maria Persson, 109-141. Abhandlungen für
———. Forthcoming a. “The so-called we-qatal conjugation in Biblical Hebrew once
again”. In 11. Mainz International Colloquium on Ancient Hebrew (MICAH), 1.-3.
November 2013: Forthcoming in KUSATU: Kleine Untersuchungen zur Sprache
des Alten Testaments und seiner Umwelt.
———. Forthcoming b. “‘Subordination’: Some reflections on Matthiesen and
Thompson’s article "The structure of discourse and ‘subordination’" and its bear-
ing on the idea of circumstantial clause in Arabic and Hebrew”. In Arabic and Se-
mitic Linguistics Contextualized, edited by Lutz Edzard. Abhandlungen für die
Kunde des Morgenlandes. Wiesbaden: Harrassowitz.
———. Forthcoming c. “Clause combining in the Song of Moses (Deut. 32:1-43). An
example of archaic Biblical Hebrew syntax”. In Biblical Hebrew Linguistics: Pa-
pers from the 16th WCJS, Jerusalem 2013, edited by Tania Notarius and Adina
Moshavi: Eisenbrauns.
Joosten, Jan. 1999. “The long form of the prefix conjugation referring to the past in
Biblical Hebrew prose”. Hebrew Studies 40: 15-26.
———. 2012. The verbal system of Biblical Hebrew. A new synthesis elaborated on
the basis of classical prose. Jerusalem Biblical Studies 10. Jerusalem: Simor.
J-M = Joüon, Paul, and Takamitsu Muraoka. 2006. A grammar of Biblical Hebrew.
2nd rev. English ed. Subsidia Biblica 27. Roma: Gregorian & Biblical Press. Re-
print, 2009 with corrections.
JPS = JPS Holy Scriptures 1917 (English).
Kiparsky, Paul. 1968. “Tense and mood in Indo-European syntax”. Foundations of
Language 4 no. 1: 30-57.
Kogan, Leonid. 2014. “Waw sargonicum. On Parataxis in Sargonic Royal Inscrip-
tions”. Zeitschrift für Assyriologi 104 no. 1: 42-55.
König, F. Eduard. 1881-97. Historisch-kritisches Lehrgebäude der hebräischen Spra-
che. 3 vols. Leipzig: Hinrichs. Reprint, Hildesheim, 1979.

266 Bo Isaksson
Kraus, Hans-Joachim. 1978. Psalmen. 5th ed. 2 vols. Biblischer Kommentar: Altes
Testament 15. Neukirchen.
Lehmann, Christian. 1988. “Towards a typology of clause linkage”. In Clause combin-
ing in grammar and discourse, edited by John Haiman and Sandra A. Thompson,
181-225. Amsterdam; Philadelphia: John Benjamins.
Lunn, Nicholas P. 2006. Word-order variation in biblical Hebrew poetry: Differentiat-
ing pragmatics and poetics. Paternoster biblical monographs. Carlisle: Paternoster
Press.
Meyer, Rudolf. 1966-1972. Hebräische Grammatik. 3rd ed. 4 vols. Sammlung
Göschen. Berlin: de Gruyter.
Müller, Hans-Peter. 1983. “Zur Geschichte des hebräischen Verbs – Diachronie der
Konjugationsthemen”. Biblische Zeitschrift 27: 34-57.
———. 1991. “wa-, ha- und das Imperfectum consecutivum”. Zeitschrift für Althebra-
istik 4: 144-160.
NAB = The New American Bible.
NAS = The New American Standard Bible.
NET = The New English Translation Bible.
Niccacci, Alviero. 1987. “A Neglected Point in Hebrew Syntax: Yiqtol and Position in
the Sentence”. Liber Annuus (Studium biblicum franciscanum) 37: 7-19.
———. 2014. “Background constructions inside the main line in biblical Hebrew”. In
vik, Sweden, edited by Bo Isaksson and Maria Persson, 179-189. Abhandlungen für
NIV = The New International Version.
NJB = The New Jerusalem Bible.
Notarius, Tania. 2013. The verb in archaic Biblical poetry: A discursive, typological,
and historical investigation of the tense system. Studies in Semitic languages and
linguistics 68. Leiden – Boston: Brill.
NRS = New Revised Standard Version Bible.
Nyberg, Henrik Samuel. 1972. Hebreisk grammatik. 2nd ed. Stockholm: Almqvist &
Wiksell.
Pardee, Dennis. 2012. “The biblical Hebrew verbal system in a nutshell”. In Language
and nature: Papers presented to John Huehnergard on the occasion of his 60th
birthday, edited by Rebecca Hasselbach and Na'ama Pat-El, 285-318. Studies in
ancient oriental civilization 67. Chicago: Oriental institute of the University of
Chicago.

Polak, Frank. 2014. “The circumstantial clause as trigger: Syntax, discourse and plot
structure in biblical narrative”. In Strategies of Clause Linking in Semitic Lan-
guages: Proceedings of the International Symposium on Clause Linking in Semitic
Languages 5-7 August 2012 in Kivik, Sweden, edited by Bo Isaksson and Maria
Persson, 191-203. Abhandlungen für die Kunde des Morgenlandes 93. Wiesbaden:
Harrassowitz.
Rainey, Anson F. 1986. “The ancient Hebrew prefix conjugation in the light of Am-
arnah Canaanite”. Hebrew Studies 27: 4-19.
———. 1996. Canaanite in the Amarna tablets: A linguistic analysis of the mixed
dialect used by scribes from Canaan. Vol. 3, Morphosyntactic analysis of the par-
ticles and adverbs. Handbook of Oriental Studies: The Near and Middle East 25.
Leiden: Brill.
———. 2003a. “The yaqtul preterite in Northwest Semitic”. Hamlet on a hill: Semitic
and Greek studies presented to Professor T. Muraoka on the occasion of his sixty-
fifth birthday no. Orientalia Lovaniensia Analecta 118: 395-407.
———. 2003b. “The suffix conjugation pattern in ancient Hebrew tense and modal
functions”. Ancient Near Eastern Studies 40: 3-42.
Revell, E. J. 1984. “Stress and the waw ‘consecutive’ in biblical Hebrew”. Journal of
the American Oriental Society 104: 437-444.
RSV = Revised Standard Version of the Bible.
SCL = The German Schlachter Version.
Sjörs, Ambjörn. Forthcoming. The history of sentential negation in Semitic. PhD diss.
Uppsala.
Steiner, Richard C. 2000. “Does the Biblical Hebrew conjunction –w have many mean-
ings, one meaning, or no meaning at all?”. Journal of Biblical Literature 119 no. 2:
249-267.
Stempel, Reinhard. 2012. “The injunctive in Semitic”. In Studia Andreae Zaborski
dedicata, edited by Jerzy Chmiel, Anna Krasnowolska, Tomasz Plański, Ewa
Siemieniec-Gołaś, Lidia Sudyka and Joachim Śliwa. 523-528. Folia Orientalia 49.
Cracow: Polish Academy of Sciences.
TNK = JPS TANAKH (English).
Tropper, Josef. 1998. “Althebräisches und semitisches Aspektsystem”. Zeitschrift für
Althebraistik 11: 153-190.
Tropper, Josef, and Juan-Pablo Vita. 2010. Das Kanaano-akkadische der Amarnazeit.
Lehrbücher orientalischer Sprachen. Section I: Cuneiform Languages 1. Münster:
Ugarit-Verlag.

268 Bo Isaksson
Van der Merwe, Christo H. J., Jackie A. Naudé, and Jan H. Kroeze. 1999. A biblical
Hebrew reference grammar. Biblical Languages: Hebrew 3. Sheffield: Sheffield
Academic Press. Reprint, 2004.
Verstraete, Jean-Christophe. 2005. “Two types of coordination in clause combining”.
Lingua 115: 611-626.
Westermann, Claus. 1982. Genesis. Vol. 3, Genesis 37-50. Biblischer Kommentar:
Altes Testament I:3. Neukirchen-Vluyn: Neukirchener Verlag.

Clause Combining in Modern Spoken Aramaic

Circumstantial Clause Combining in the Jewish
Neo-Aramaic Dialect of Zakho
Eran Cohen, The Hebrew University, Jerusalem
0. Introduction
This paper discusses circumstantial expressions at different levels in the Jew-
ish Neo-Aramaic Dialect of Zakho.
The Jewish dialect of Zakho (henceforth JZ) is a relatively well attested di-
alect of North Eastern Neo-Aramaic (NENA, see Khan 2007). The most sali-
ent feature of this entire group is that the old Semitic verbal system, found in
the earlier phases of Aramaic, had been lost and the language(s) had under-
gone an extreme restructuring of the verbal system, which now consists of old
participles and nomina actionis. This change created fascinating verbal sys-
tems, each consisting of many forms, whose components are relatively trans-
parent (both synchronically and diachronically). This rich verbal system
makes the description of circumstantial expressions a desideratum, since it is
difficult to predict which forms participate in this function. No such descrip-
tion has been attempted so far of any single dialect.1 The only comprehensive
description of this dialect (Cohen 2012) refers to various aspects of the phe-
nomenon, but not under one heading.
The text used as a corpus for this inquiry includes mainly folktales as well
as other stories; in short, it is basically a corpus of folk literature. Most of it
had been collected by H. J. Polotsky during the 1940s and it is handwritten in
narrow transcription. The corpus typically exhibits a radically different syntax
for each of the textemes, namely, dialogue and narrative. This dichotomy
concerns us here, since various backgrounding devices in narrative are similar
1 An MA thesis is currently being prepared in Jerusalem by Nikolaus Wildner, attempting

to describe the phenomenon in the literary Neo-Aramaic dialect of Urmi.

272 Eran Cohen
to circumstantial expressions at sentence level, but are marked by different

forms or exponents.
These expressions have in common an adverbial expression of a non-
specific semantic nature (namely, it is often difficult to state whether the orig-
inal nuance is temporal, final, or concessive), and characteristically depict the
state of one of the arguments in the “main” clause. I use quote marks because
the circumstantial expression is not always a subordinate clause, and conse-
quently we do not always have a main clause.
1. Sentence level
It seems a good idea to start at sentence level, because this level is valid for
the entire language, independently of texteme or genre. In other words, strate-
gies that express circumstantiality at sentence level are not specific to a cer-
tain environment, and can be found everywhere. For this reason, these expres-
sions have a wider applicability.
At this level there are two types of circumstantial expression – simple and
complex, or rather, unipartite and bipartite. What characterizes these expres-
sions is a strong tendency to show agreement with the argument whose state is
described (termed, in the Arab tradition ḏū [or ṣāḥib] al-ḥāl). This is ex-
plained in Goldenberg (1985, 336-337 [=1998, 185-186]) as the result of the
double relationship this type of adverbial function has, like any adverbial, to
the predicative link, and to the argument whose state it describes.
The first type (unipartite) consists of an adjective, a participle or a ger-
und:

Circumstantial Clause Combining in the Jewish Neo-Aramaic of Zakho 273
(1) ʿād-ıt day ʾēna wḗwā-la

custom-NUC DET spring PST-be-3FS
kud g-šāte mınn-a {qamāya} g-māyes.
each-NUC PRS-drink.3MS/CP from-3FS first.MS PRS-die.3MS
nāš-ıt day bāžer (...)2 g-nabl-ī-wa ımm-ıt gyān-u
3
people-NUC DET town IPFV -bring-3CP with-NUC self-3CP
xa qaṭūsa, g-ma-štḗ-wā-la {qamē-sa}
DET cat-FS IPFV-CAUS-drink.3CP-OBJ.3FS first-FS
‘The custom of this spring was (that) anyone who drinks from it {firstMS}
dies. The people of that town, (when they would come down to the spring
to drink water,) they would bring a catFS with them (and) let itFS drink
{firstFS}’ PT 144
The circumstantial expression ‘first’ shows, by its gender morphemes, the

reference to one of the arguments – in the first case it is ‘anyone who’, which
is conceived as masculine (qamāya), whereas in the second case, it is the cat,
which is feminine in JZ (qamēsa).
(2) ū lıbb-e pıš-le hēš mharhōre

CONN heart-3MS AUX-3MS still GER.fantasize
ū ʾēn-e mēnōxe bıt gawd-ıt ʾıstatt-e
CONN eye(s)-3MS GER.look at body-NUC mistress-3MS
{hādax šulxāya ū-purʾāya} ||šulxēsaFS ū purʾēsaFS
thus naked.MS CONN-uncovered.MS
‘He (lit. his heart) was still having improper thoughts and his eyes were
looking at his mistress’s bodyFS, (||she/it) (being) stark nakedMS (FS)’ PT
786
šulxāya ūpurʾāya lit. ‘naked and uncovered’ (masculine) occurs originally in

the text, while šulxēsa ūpurʾēsa (feminine) is added to the left of the text in
the original manuscript. The former could only refer to the man looking at his
mistress (in both senses), that is, ‘him (being) naked and uncovered’. The
2 (...) stands for a part skipped for relative irrelevance, which is given in the translation
within brackets. See also exx. (15) and (17) below.
3 The glosses generally follow the Leipzig rules. Some combinations are only partly
predictable: 1. The present morpheme k/g- with backshifting (BSH) morpheme
-wa(-) produce an imperfect(ive) (IPFV); both preterites (PST) with -wa(-) result in a
plusquampreterite (PLPT); the future morpheme p/b- (FUT) with -wa(-) yields a special
form which mostly denotes irrealis. The construct state, namely, the bound form, or the
head of the construction, is glossed by NUC (for nucleus).

274 Eran Cohen
latter, on the other hand, could refer to either the mistress or her body (both
feminine singular). The adverbial syntagm contains two adjectival forms in
adverbial function, and the syntactic link is made clear by the form of the
adjective.
The same thing happens when the circumstance is expressed by a partici-
ple:
(3) ʾáw jwanqa qam-dōq-ī́-wā-le gō bāžır

DET youngster PLPT-catch-3CP-OBJ.3MS in town
{ʾısya ʾımmıt parrōke}
PTCP.come.MS with textiles
‘They had caught the youngster in the city, {(having) comeMS with tex-
tiles}’ PT 789–90
The perfect participle ʾısya agrees in gender and number with the youngster.
However, when the expression consists of the gerund, there is no expression
of agreement between it and the described argument:
(4) pıš-lu mēnōxe bıd dō nāša

AUX-3CP GER.look at DET man
{hādax tīwa rıš xa taxta}
thus PTCP.sit.MS on DET chair
ū ṣīwa l- bırk-e
CONN wood to knee-3MS
{ū bīsāya ʾıll-e bıd daw jaʾōza}
CONN GER.come at-3MS with DET ax
‘They began looking at this man, {sittingMS on a chair},
the (piece of) wood on his knee, {and coming at it (=the wood) with the
ax}’ PT 510
Whereas tīwa ‘sitting’ reflects masculine singular, the gerund bīsāya ‘coming’
does not. Nevertheless, both have the same function.4 There is a third bipartite
circumstantial nexus in the example, consisting of a prepositional phrase (un-
derlined).
The negative form of the gerund in this function, formed by the negative
particle la with the infinitive, is seldom attested, and often has the privative
meaning of ‘without’:
4 Note that in some languages, the gerund does inflect for person, e.g., Gǝʿǝz maṣiʾ-o
‘him coming’ maṣiʾ-aka ‘you coming’, etc.

(5) ʾāwa pıš-le bıkxāka ū bıgmāṣa dīd-a

PRON.3MS AUX-3MS GER-laugh CONN GER-smile OBJ-3FS
{la šqal ṭım-ıt xabra
NEG INF.take.NUC essence-NUC word
ū la fham maʿne dīd-e}
CONN NEG INF.understand.NUC meaning POSS-3MS
‘He started laughing and smiling, {not taking in the essence of the
word(s) and not understanding its meaning}’ PT 867
It should be noted that the construction here is different: whereas the gerund
takes a direct object, these negated infinitives take their complement nominal-
ly, in a genitive construction, and therefore they are found in the construct
state and marked accordingly (fham instead of the unbound form fhāma) as
the nucleus of the construction (NUC).
The simple circumstantial expression is occasionally expanded to specify
another part, the theme (that is, the given entity), between which and the
rheme (the new information) there exists a clear predicative relationship, or a
nexus. Compare exx. (6) and (7):
(6) ū msupy-ā-la {hādax ḥmıl-ta}

CONN PRT.deliver-OBJ.FS-3FS thus PTCP.stand-FS
‘And she died (lit. delivered it [i.e., her soul]) {standing}’ PT 853
(7) {ʾāya hādax ḥmıl-ta}, ham mír-rā-le:

PRON.3FS thus PTCP.stand-FS too PRT.say-3FS-DAT.3MS
‘{She standing}, nevertheless she said to him:’ PT 852
The same function is now expressed by a complex form, which is a predica-

tion, or a nexus, but a dependent one, (7): an independent nexus in JZ would
have involved a copula between the personal pronoun ʾāya ‘she’ and the par-
ticiple ḥmılta ‘standing’. The same idea is found in the following pair of ex-
amples, (8) and (9):
(8) mpıq-le xūwe mın dūk-e

PRT.get_out-3MS snake from place-3MS
{dwīqa p-pumm-e xá dıhwa}
PTCP.hold.MS in-mouth-3MS DET gold_coin
‘The snake came out from its place, {holding a gold coin in his mouth}’
PT 453

276 Eran Cohen
(9) mpıl-la dıbba l-ʾarʾa

PRT.fall-3FS bear to-ground
{hēš ʾīz-a duq-ta bıt bāb-ıt dō yāla}
still hand-3FS PTCP.hold-FS at father-NUC DET child
‘The she-bear fell to the ground, {its paw still holding the child’s father}’
PT 752
Here too, ʾīza duqta ‘its paw holding’ is a dependent nexus. Yet another ex-
ample for a bipartite circumstantial syntagm is underlined in ex. (4) (ū-ṣīwa l-
bırk-e ‘and the (piece of) wood on his knee’), where the rheme is a preposi-
tional phrase.
These nexal, or predicative circumstantials are often associated with inal-
ienable body parts of one of the arguments (this type is mentioned and exem-
plified in Polotsky 1996:23-26, under “NON-FINITE PREDICATIVE
GROUPS”).5
Note that these syntagms occur occasionally with the adverb hādax (oth-
erwise meaning “thus”), whose function here is merely to point out a circum-
stantial expression, rather than refer to manner.
The bipartite dependent circumstantial expression can sometimes inter-
change with a finite verbal form that occurs in the very same function and is
hence analyzed as subordinate. Compare exx. (10) and (11):
(10) pıš-la bıkyāpa gō pōẓ-e {ū bōl-a bıʿlāqa

AUX-3FS GER.bend_over in torso-3MS CONN hair-3FS GER.touch
bıd pās-e ū gubʾēn-e}
at face-3MS CONN forehead-3MS
‘She was bending over his torso, {her hair touching6 his face and his
forehead}’ PT 868
5 xa surta d Mart Marjam brun-o go xpaq-o ‘a picture of the Madonna (with) her son in
her bosom’ (Mois de Marie 142, 10 mentioned in Polotsky 1996, 24). Polotsky, howev-
er, does not separate between attributive and circumstantial function with regard to these
syntagms.
6 Theoretically, the substantive and gerund could be analyzed as ellipsis of the auxiliary
copula pıš-la (3FS); however, the substantive bōla is masculine while the auxiliary is 3rd
person feminine, so the probability that this is an ellipsis is not very high.

(11) pıš-la bīsāqa ū bıkwāša …

AUX-3FS GER.go_up CONN GER.go_down
{u ʾēn-a la-g-ʿalq-a ū la-k-tafq-a
CONN eye-3FS NEG-PRS-meet-3FS CONN NEG-PRS-notice-3FS
bıt ču ʾazāla ū ʾasāya}
in NEG goer CONN comer
‘She began ascending and descending … {(while) her eye does not meet
(n)or notices any passer(s)-by}’ PT 855
In ex. (11) there is an asyndetic subordinate clause consisting of two kšāqıl

forms, which basically denote the general present. In this slot, in narrative,
these forms express circumstantial indicative concomitance. Note that the
verbal form in the clause ʾēn-a la g-ʿalq-a ‘her eye does not meet’ in this
function is the syntactic equivalent of the dependent bipartite expression bōl-a
bıʿlāqa ‘her hair touching’. This is one of the reasons to conclude that the
verbal form is subordinate, since it interchanges with a non-clausal unit. The
circumstantial present is found in the following example as well:
(12) sē-lu kutr-u ḥakōm-e {k-ṭāʾe ıl yalunk-u}

PRT.come-3CP both-3CP king-PL PRS-look_for.3CP to children-3CP
‘Both kings came {looking (lit. they look) for their children}’ PT 126
Other forms are capable of occurring in this function – for instance, the
subjunctive as well as the form used in dialogue to denote the prospective or
the future (ex. (13)):
(13) muzvır-ru l-gyān-u {d ʾāse (//b-āse)

PRT.turn-3CP to-SELF-3CP THAT SBJV.come.3CP FUT-come.3CP
l-bēsa} ū ṭrē-lu b-lazzi
to-house CONN PRT.drive-3CP in-speed
‘They turned around {in order (//intending) to go back home} and
drove hastily’ PT 889
This example is taken as is from the text, where the form b-āse ‘intending to
go’ constitutes a close alternative to the purpose clause d ʾāse ‘in order to go’
(the latter is syndetically subordinate, marked by d-). Both verbal forms are
part of the circumstantial paradigm.
Subjunctive forms with backshifting -wa(-) occur occasionally following
matrix verbal forms, which denote the past. However, this “agreement” is not

278 Eran Cohen
consistent, and mostly it is impossible to provide a consistent linguistic ra-

tionale for the function of these -wa forms. In this case, however, with the
prospective p-šāqıl form, it is possible:
(14) qam-dārē-la ʾáy dıste ʾáp-āya rıž dán qapy-e

PRT-put.3MS-OBJ.3FS DET pot too-PRON.3FS on DET log-PL
ū ṭrē-le xmāra {b-āse l bēsa}
CONN PRT.drive-3MS donkey FUT-come.3MS to house
‘He put the pot as well on top of the logs of wood and led the donkey,
{intending to go back home}’ PT 919
(15) xá yōma (…) yōm-ıt ʾrōta wḗwā-la.

one day day-NUC Friday PRT.be-3FS
mpıq-lu mbínnōke drangi {b-āsē-wa l bāžer}.
PRT.go_out-3CP morning late FUT-come.3CP-BSH to town
pıš-la šaḅsa
PRT.become-3FS Saturday
ū ʾāni hēš wē-lu go barīya.
CONN PRON.3CP still COP-3CP in desert
‘One day, (when they were in the city in which they used to sell fabrics,)
it was Friday. They left late in the morning, {intending to go back to the
city}. It became Sabbath (while) they are still in the wilderness’ PT 51–
52
The p-šāqıl-wa form (“future in the past”, mostly used for counterfactual
apodoses) is here circumstantial and prospective, like p-šāqıl (ex. (14)). How-
ever, in all attested cases of the form p-šāqıl-wa as circumstantial, the ex-
pressed intention turns out immediately not to have been realized. For in-
stance, in ex. (15), they are not able to get home, because they cannot travel
during the Sabbath. Another example is ex. (22) below, where the giant who
is about to be slaughtered is not really a giant but rather a little girl. The im-
plications are that there is a modal difference between the two forms: one is
neutral; the other expresses an unrealized intention.
2. Text-level circumstantial expressions

I now move on to the grey area above sentence level, where there are phe-
nomena which express circumstantiality that is mostly related to larger units

than the sentence. Some phenomena are described and explained within the
framework of the text.
Grounding, in narrative, refers to the fundamental opposition between
foreground and background. However, the foreground (or mainline, whose
main function is to move the plot forward) is rather uniform in JZ narrative,
consisting mainly of the two preterites. In addition, one occasionally comes
across narrative events in disguise (until clauses which feature full-fledged
events, for instance).7 On the other hand, the background (off-line material,
or comment mode), has many functions and effects that involve quite a few
exponent types: character description, circumstantial description, topical set-
ting of various forms subdividing the text, etc. For instance, some of the de-
scriptive part at the beginnings of folktales is quite similar in principle to
circumstantial expressions, but its scope of reference is the entire story. In
many cases it is what seems to be an adverbial that has this function; but un-
like an adverbial, which interchanges in its slot only with other adverbial
syntagms (e.g., the interchange of an adverbial clause with a simple adverb),
in this slot this interchange is of a wilder nature – it could be an adverbial, a
bare independent-looking clause, a presentative clause, etc. Not all of the
background is relevant to our discussion, but several phenomena from it defi-
nitely describe the circumstances of an argument.
2.1 Plusquampreterite šqıl-wā-le, qam-šāqıl-wā-le and šqīla wēle

Despite the punctual value usually attributed to them, plusquampreterites do
not represent an event in narrative, but are rather part of the rich array of nec-
essary off-line information. They are often found among other off-line forms,
but not exclusively (ex. (16)):
7 ʾrıq-le basr-u hīl qam-dāwıq-le brōna dīd-a ‘He pursued them until he caught up with
its son’ (82).

280 Eran Cohen
(16) xá yōma ṓha jōtyāra zıl-le l šūla

one day DET peasant PRT.go-3MS to work
ū baxt-e ʾuz-la aw yōma nehra.
CONN wife-3MS PRT.do-3FS DET day laundry
wē-la bıšṭāxa jull-e
COP-3FS GER.hang clothes
{ū bır ḥakōma qímwa-le mın šınsa
CONN son king PLPT.get_up-3MS from sleep
ū ysíqwa-le rıš gār-et bēs-ōhūn}
CONN PLPT.go_up-3MS on roof-NUC house-3CP
ū ʾēn-e nzır-ra ıl day xamsa
CONN eye(s)-3MS PRT.look-3MS to DET maid
ū xrū-le ū mpıl-le ıl dūk-ıt gıyān-e
CONN PRT.faint-3MS CONN PRT.fall-3MS to place-NUC SELF-3MS
‘One day the peasant went to work and his wife did that day the laundry.
(While) she is hanging clothes, {the prince had woken from sleep and
had gone up to the roof of their house}, his eye caught a glimpse of the
maiden and he fainted and fell right there’ PT 499
The preterite forms here generally move the plot forward, whereas the forms
in boldtype (in fact, the entire phrase in curly brackets) report incidents that
had taken place prior to the main line of the story. They are explanatory in
nature, rather than eventive (for instance, if they had been given as preterites).
In this specific case, these forms provide an explanation for the connection
between the prince and the woman doing laundry.
(17) xá yōma musē-lu trḗ jandurm-e dīd-a xá jwanqa (…)

one day PRT.bring-3CP two gendarme-PL POSS-3FS DET youngster
{ʾáw jwanqa qam-dōq-ī́-wā-le gō bāžır
DET youngster PLPT-catch-3CP-OBJ.3MS in town
ʾısya ʾımmıt parrōke}
PTCP.come.MS with textiles
‘One day, two of her gendarmes brought a youngster, (his hair loose and
falling over his shoulders, and his beard barely growing.) {They had
caught the youngster in the city, (having) comeMS with textiles}’ PT 789–
90
Here too, the clause in curly brackets explains where the youngster comes
from, without being a part of the chain of events, or the mainline. Here these
forms are in fact couched in a description – the youngster’s physical appear-

ance. These forms describe circumstances that are previous or anterior to the
reference time of the events. In exx. (16) and (17) the circumstantial infor-
mation explicitly refers to one of the arguments. In the periphery of these
cases, we find similar, explanatory, plusquampreterites, which do not specifi-
cally refer to an argument:
(18) šqıl-lu īxāla d-ōhun ū zıl-lu

PRT.take-3CP food POSS-3CP CONN PRT.go-3CP
mın mbínnōke ū hīl palgıdyōm,
from morning CONN till noon
{dunye šxínwā-la rāba}, qēṭa wē-le
world PLPT.be_hot-3FS very summer PRT.be-3MS
‘They took their food and walked from morning till noon,
{it had been very hot}, it was summer’ PT 41
Note that here, perhaps due to the nature of the verbal lexeme (‘be hot’), no
punctuality is detected.
It is true that the plupreterites/pluperfects šqıl-wā-le, qam-šāqıl-wā-le and
šqīla wē-le show anteriority vis-à-vis the preterites šqıl-le and qam-šāqıl-le.
Actually, that is the only temporal opposition on the narrative plane (except
for subordinate clauses, inside which there is an entirely different set of oppo-
sitions). Note, however, that this temporal opposition is somewhat secondary
(compared, for instance, with the opposition between the same forms in dia-
logue), since the important issue here is not simply when it happened, but
rather that it happened sometime before (or was over by a certain reference
point) and that it is relevant, even crucial, for the understanding of what hap-
pens, expressed as narrative events.
2.2 Presentative constructions: postposed vs. preposed

Neo-Aramaic dialects often have two types of copula, simple and “deictic”, or
“presentative”, which consists of an additional deictic element. The presenta-
tive copula in this dialect occurs mostly in the 3rd person, is not found in sub-
ordinate clauses and has several functions:
1. It is used, mostly in dialogue, as the main copula of compound tenses (to
wit, present perfect and present progressive) in the third person;
2. it reflects the character’s point of view when complementing a verb of
perception; and
3. in the narrative, it constitutes a circumstantial clause.

282 Eran Cohen
Naturally, the 3rd function is the one we focus upon. The presentative cir-
cumstantial clause may both precede and follow the clause(s) to which it re-
fers.
(19) zıl-le basr-e ʾáw ṣīwāya {u

PRT.go-3MS after-3MS DET lumberjack CONN
wē-le bırʾāla mın zdoʾsa lá ʾāxıl-le ʾarya}
COP-3MS GER.shake from fear NEG SBJV.eat.3MS-OBJ.3MS lion
‘The lumberjack followed him {(while) he is shivering out of fear lest
the lion eat him}’ PT 667
(20) tū-la mʿōšē-la {ʾīxāla dīd-a wē-le

PRT.sit-3FS PRT.eat-3FS food POSS-3FS COP-3MS
xa tawırta u ṭḷāhá ʾızz-e}
one cow CONN three goat-PL
‘She sat (and) ate supper, {her food is one cow and three goats}’ PT
412–413
In these two cases, the scope of the presentative clause is local, consisting
only of the preceding clause (‘following’ in ex. (19), and ‘eating’ in ex. (20)).
These cases are quite different from the ones where the presentative syntagm
precedes. In such case, it forms some kind of interdependency with the claus-
es that follow. It is worthwhile noting that in these cases the function is narra-
tive rather than enhancing, namely, hardly any new information is being add-
ed:
(21) ū škıl-le bıd ʾīxāla {wē-le bīxāla}

CONN PRT.start-3MS at INF-eat COP-3MS GER.eat
ū sē-lu ʾıll-e mšalxān-e
CONN PRT.come-3CP on-3MS robber-CP
‘He started eating. {(while) he is eating}, robbers came upon him’ PT
524
The underlined syntagm in ex. (16) is similar (‘she is hanging clothes...’).

These preposed presentative clauses have an anchoring function (note that
they tend to resume previous text parts – in ex. (16) it is the woman doing
laundry, in ex. (21) it is eating), and consequently they link between text parts.
To use Givón’s terms (Givón 1987), the preposed clause(s) serve as a
grounding device, both anaphoric and cataphoric, namely, it has ties with
both preceding and following parts of the text, very much like a topic. The

postposed presentative clause, on the other hand, does not have this function,
and it is far more specific as a circumstantial expression.8 Moreover, the latter
adds new information whereas the former consists of mostly given infor-
mation. For this reason, such examples (such as (16) and (21)) should be con-
sidered a different circumstantial, perhaps an “abstract circumstantial”.
2.3 ay dammıt constructions: postposed vs. preposed

The expression (ʾay) dammıt is the construct state of (ʾay) damma ‘the mi-
nute’, and functions as a temporal conjunction (‘the minute that...’). In the
following example, the (ʾay) dammıt syntagm repeats presupposed infor-
mation and functions as an anchor, as the preposed presentative clauses dis-
cussed in the previous section. In fact, such examples belong to the setting
paradigm. This is the main function of ay dammıt clauses in narrative
(ex. (22)):
(22) bır ḥakōma qam-ma-mpıl-le ṓha kapōra.

son.NUC king PRT-CAUS-fall.3MS-OBJ.3MS DET giant
{ay damm-ıt mpıl-le ṓha kapōra}
DET moment-NUC PRT.fall-3MS DET giant
mōrum-le bır ḥakōma sēpa dīd-e b-qāṭíl-wā-le
PRT.lift-3MS son.NUC king sword POSS-3MS FUT-kill.3MS-BSH-OBJ.3MS
‘The prince knocked down this giant. {As soon as this giant fell}, the
prince raised his sword, intending to kill him’ PT 89
The circumstantial clause in ex. (22) is an explicit resumption of the fall. An

explicit object pronoun in the clause at the end (underlined) stands for the
argument whose state is depicted. However, these clauses often show less
explicit reference. Ex. (23) has a type of resumption, which requires some
inference on the part of the listener:
8 Ramsay (1987) shows this behavior in pre-and postposed conditional and temporal
clauses in English.

284 Eran Cohen
(23) ... ū qam-māpıq-la ay čūčıksa

CONN PRT-take_out.3MS-OBJ.3FS DET bird
ū qam-zābeḥ-la ıl daw gōra.
CONN PRT-slaughter.3MS-OBJ.3FS on DET man
{damm-ıt kuš-le dımm-a rıš daw gōra},
moment-NUC PRT.come_down-3MS blood-3FS on DET man
ṭrıṣ-le pıš-le ṣāx ū qım-le
PRT.heal-3MS PRT.become-3MS well CONN PRT.get_up-3MS
rıš ʾaql-e
on foot-3ms
‘...took out the small bird and slaughtered it over the man. {The moment
its blood dripped over the man},
he healed, became well and stood up on his feet’ PT 103
Note that the dammıt clause tells us of blood flowing (as a direct inference
from slaughtering a bird, as well as in repeating what is said earlier in the
text). These circumstantial clauses are not particularly informative, but rather
functional, viz. of linking what precedes to what follows.
When the dammıt clause follows the clause to which it refers, it has in fact
the microsyntactic adverbial function, which means that the scope of the
“temporal” clause is local, referring to the clause in which it is embedded:
(24) nāš-e k-ṣafn-ī-wa ʾıbb-e

man-PL IPFV-watch-3CP at-3MS
{ʾáy damm-ıt g-māxḗ-wā-le jaʾōza
DET moment-NUC IPFV-strike.3MS-OBJ.3MS hatchet
čār nıkā́ r ṣīwa}
four side (piece of) wood
‘People were watching him {while he would strike (with) the hatchet
around the (piece of) wood}’ PT 509
(25) ḥakōma +rāba qḥır-re

king very PRT.be_sad-3MS
+
damm-ıt mıt-le pālavan dīd-e
moment-NUC PRT.die-3MS acrobat POSS-3MS
‘The king was very sad when his acrobat died’ PT 769
These postposed dammıt clauses are not many, and only occasionally are they
circumstantial (e.g., ex. (24), but not ex. (25), which is a regular temporal

clause). Some of them are similar to ex. (27), in having the verb xzēle in the
clause (‘when he saw’).
When containing presupposed information and being preposed, these
clauses have the same function as the presentative constructions (which are
definitely non-subordinate). When postposed, they have a local scope and
refer to one clause, and are occasionally analogous to the microsyntactic ex-
pressions, compare exx. (26) and (27):
(26) lašš-ıd dṓ gōra mzurzeʾ-la

body-NUC DET man PRT.shake-3FS
{damm-ıt ġzē-le ʾḗ baxta {hatxa mare lıbba
moment-NUC PRT.see-3MS DET woman thus owner heart
ū mare jurʿıta}}
CONN owner courage
‘The man’s body shook {when he saw the woman, {so brave
and courageous}}’ PT 853
(27) u rāba bxē-le ū lıbb-e qız-le

CONN very PRT.weep-3MS CONN heart-3MS PRT.burn-3MS
{bıd ġzāy-ıt ʾarxa dīd-e {bıd dṓ mūjıb}}
in INF.see-NUC guest POSS-3MS in DET manner
‘so he wept a lot and felt pity (lit. his heart burned)
{seeing his guest {in this condition}}’ PT 614
Note that the clause headed by dammıt (ex. (26)) is syntactically equivalent to
the infinitive construction (ex. (27)), which is not the gerund, so it seems (the
gerund is never complemented this way, cf. ex. (5)). Moreover, circum-
stantials which involve seeing may contain yet another potential circum-
stantial as part of the argument structure of seeing verbs (marked by a second
set of curly brackets in both cases).

286 Eran Cohen
2.4 Other “adverbial” constructions: basır

The syntagms headed by conjunction basır ‘after’ are a special case. In
exx. (28) and (29), the two seemingly different cases are related:
(28) sē-le xā gōra ū psıx-le dıkkāna

PRT.come-3MS DET man CONN PRT.open-3MS store
w-īʾıl-le ū šqıl-le xā līra
CONN-PRT.enter-3MS CONN PRT.take-3MS one coin
ū mpıq-le. # {basır mpıq-le}
CONN PRT.go_out-3MS after PRT.go_out-3MS
qam-dāwıq-le ū mír-rē-le
PRT-catch.3MS-OBJ.3MS CONN PRT.say-3MS-DAT.3MS
‘a man came, opened the store, entered, took one coin and came out. #
{After he came out}, he caught him and said to him’ PT 3–4
This “adverbial” clause is functionally similar to an “independent” clause:
(29) wıḷḷa ʾō yāla zıl-le duq-le ʾurx-e

PTCL DET boy PRT.go-3MS PRT.seize-3MS way-3MS
u zıl-le, mpıq-le mın bāžer. # {mpıq-le
CONN PRT.go-3MS PRT.go_out-3MS from town PRT.go_out-3MS
mın bāžer}, zıl-le xá bāžer xēta
from town PRT.go-3MS DET town other
‘So the boy left, took his direction and left, got out of town. # {He got
out of town}, went to another town’ Meehan and Alon 1979 5.5–6.1
In both examples events, which have just been told are repeated immediately
thereafter. The difference is that in ex. (28) the events are repeated within an
adverbial basır clause, whereas in ex. (29) the repetition is packaged as a
seemingly independent clause. Nevertheless, the structures are macrosyntactic
analogues. Such basır clauses are mostly boundary markers, for instance, after
the number sign (#) in ex. (29), a new episode begins, that is, a break in the
text rather than any “enhancing” function, or information regarding the
whereabouts of one of the arguments. Since the information in each case is
repeated, the constructions are of a low communicative value. The same ap-
plies to the independent clause in ex. (29). This type of circumstantiality has a
function of text-boundary.

2.5 (In)dependent constructions

The following pair (exx. (30) and (31)) creates a situation for which it is more
difficult to account:
(30) bale {ʾāni hādax žġīl-e bıd danya šuʾāl-e

but PRON.3PL thus busy-PL in DET matter-PL
ū bıd dan taxmīn-e go dáy x(ıp)parta}
CONN in DET thought-PL in DET dugout
ū ḥakōma tū-le ʾımmıt mıjlıs dīd-e
CONN king PRT.sit-3MS with council POSS-3MS
‘But {(while) they (were) busy with these matters and with these
thoughts in the dugout}, the king sat with his council’ PT 472
(31) {anya žġıl-lu bıd maḥkōye d-ōhun}

PRON.(3)PL PRT.be_busy-3CP in talk POSS-3CP
u ḥakumta ū xıddamta dīd-a
CONN queen CONN servant POSS-3FS
wē-lu bıdrāya rāyi ū tagbir l-xāwxēta
COP-3CP GER.put counsel CONN advice to-RECP.PRON
‘{(while) they were busy with their conversation}, the queen and her
maid are/were consulting each other’ PT 808–809
In both examples the initial syntagm resumes previous information, thus act-
ing as a textual anchor. More specifically, these “circumstantials” function as
markers for the shifting between two parallel episodes: note that no argument
in the second clause is referred to in the first clause. This is textual circum-
stantiality – the first mentioned, parallel episode is used as a circumstantial
lean-on for the second episode. The difference between the examples lies in
the nature of the syntagm and in the relationship with the other clauses in the
example:
Ex. (30) has a dependent nexus, consisting of ʾāni (personal pronoun) and
žġīle (adjective), which together do not amount to a full clause (since a copula
is necessary for that), and all this is followed by a preterite. The relationship is
a unidirectional dependency; the dependent nexus is the dependent part. It
resumes actions reported in the previous episode.
In ex. (31) a preterite (žġıl-lu) is followed by a presentative construction,
which in this case is textually foregrounded. Such a presentative clause, as
we find in ex. (31), does not generally occur alone and is formally interde-
pendent with the first part (the preterite žġıl-lu). Note that normally, it is the

288 Eran Cohen
presentative clause that is considered circumstantial (compare §2.2); in this

case it is impossible: anya žġıl-lu ‘they were busy…’ is the “circumstantial”
syntagm that resumes the previous episode in the text that seems to co-occur
with this episode (compare also ex. (15), with a temporal setting). The follow-
ing table is an attempt to capture this case of skewing:
macrosyntactic circumstantial macrosyntactic foremost entity
ex. (30): syntactically dependent syntactically independent
ʾāniPERS.PRON ... žġīl-eADJ ḥakōma tū-lePRET
ex. (31): syntactically independent syntactically interdependent
anyaPRON žġıl-luPRET wē-luPRESENTATIVE COPULA bıdrāyaGER
One reason for this behavior might be the nature of the verbal lexeme of √žġl,
which is like a carrier verb, namely, a pro-form which can represent another
verbal lexeme, such as the verb do in English. Another peculiar detail is that
both sides of each example are interconnected by the connective u, like the
Arabic fa. This phenomenon occurs in various combinations in JZ (condition-
al structures, free choice quantification, as well as presentatives, see §2.2).
2.6 Copular expressions

Some examples consist of non-presentative copular clauses. These clauses
usually depict a permanent attribute, and in narrative such an attribute always
has an off-line, background function. Some of these cases may be analyzed as
circumstantial (exx. (32) and (33)):
(32) pıš-le ʾáw gōra rāba dōlamánt

PRT.become-3MS DET man very rich
ū mıtū-le yalunke dīd-e ıl madāres
CONN PRT.put-3MS children POSS-3MS to schools
{u ʾāwa pıš-le sawōna}
CONN PRON.3MS PRT.become-3MS old-man
u hḗš g-ēzıl-wa ıl ṣīw-e ta ṣapxāṭır ʾaw jawāhar
CONN still IMPV-go.3MS to tree-PL to for DET diamond
dīd g-yāwíl-wā-le xūwe
REL IMPV-give.3MS-DAT.3MS snake
‘The man became very rich, put his children in schools, {and (although)

he became old}, he would still go to the woods for the diamond which
the snake used to give him’ PT 632
(33) basır qam-mēsē-la baxt-ıt trḗ

after PRT-bring.3MS-OBJ.3FS wife-NUC two
{ū ʾanya trē yalunke… lá wēwā-lu dīd-a},
CONN PRON-(3)CP two children NEG-PRT.be-3CP POSS-3FS
kudyom k-kard-á-wā-lu
everyday IMPV-chase-3FS-OBJ.3CP
‘After he brought the second wife, {(since) these two children … were
not hers}, everyday she would chase them away’ PT 142–143
In ex. (32), we can compare two clauses featuring pıšle ‘he became’: the first
(underlined) is given as an event (‘he became rich’), whereas the second (in
bold) seems to be a circumstantial clause. The common denominator of both
exx. (32) and (33) is the combination of 1. the connective u; 2. a personal
pronoun; and 3. a verb of being. Moreover, the facts revealed are inherent
rather than local in nature (getting old, children originating in another wom-
an), but their relevance to the text where they figure is circumstantial,
namely, concessive (ex. (32)) and causal (ex. (33)).
2.7 Imperfect(ive) forms

Last but not least, Neo-Aramaic is capable of expressing several aspectual
oppositions, and most dialects dispose of at least one form which signals past
imperfective, in our case, it is k-šāqıl-wa. The form in JZ consists of the gen-
eral present k-šāqıl with backshifting morpheme -wa(-). This form is not usu-
ally a part of the subordinate circumstantial clause we discussed under §1 (i.e.,
where we find the forms k-šāqıl, p-šāqıl and p-šāqıl-wa): here they do not
constitute part of another clause but rather join the perfective forms by juxta-
position or interconnection. Compare exx. (34) and (35):
(34) ū ḥakōma tū-le ʾımmıt mıjlıs dīd-e

CONN king PRT.sit-3MS with council POSS-3MS
{ū gı-mbāqır-wa ū gı-mšāwır-wa
CONN IMPV-ask.3MS CONN IMPV-consult.3MS
ʾımmıt wazīr-e dīd-e mā ʾōz-i bıd danya nāš-e}
with vizier-PL POSS-3MS what SBJV.do-3CP in DET man-PL
‘the king sat down with his council {and was asking and consulting his
viziers, what they should do with these people}’ PT 472

290 Eran Cohen
(35) ū tū-lu ū xšū-lu hēš

CONN PRT.sit-3CP CONN PRT.think-3CP still
ū mšōwır-ru bıt xawxét
CONN PRT.consult-3CP in RECP.PRON
‘So they sat down and thought more and consulted each other’ PT 466
In both examples someone is sitting and consulting. Sitting is put into the
preterite that generally denotes punctuality. Consulting, however, is expressed
in different aspects: imperfective (gı-mšāwır-wa) in ex. (34) and perfective
(mšōwır-ru) in ex. (35). The functional difference is that whereas the imper-
fective forms describe, the perfective forms narrate. Ex. (36) tells about a
recently blinded giant chasing the protagonist, who is also the storyteller:
(36) ʾrıq-li urx-ıt xá saʿa

PRT.run-1CS way-NUC one hour
u ʾāwa qam-ʾāwız-la p- xá gāve,
CONN PRON.3MS PRT-make.3MS-OBJ.3FS in one step
{lá-k-xāzē-wa, bale g-māyıx-wa
NEG-IMPV-see.3MS but IMPV-sniff.3MS
ū l-mūjıb myāxa g-mandē-wa gāve}
CONN to-manner smell IMPV-throw.3MS step
‘I ran the distance of one hour, whereas he did it in one step.
{He was not able to see, but he was sniffing and taking a step accord-
ing to the smell}’ PT 394
It is quite clear in this case that the imperfective forms here describe the cir-
cumstances in which the chase took place.
3. Recapitulating and concluding

The circumstantial expression at sentence level is quite uncomplicated in that
all its participants share the same paradigm, or in other words, take the same
syntactic slot and may interchange with each other (see table). The main dif-
ferences between the finite verbal forms in this function are modal, namely,
fact – purpose – intention (realized and unrealized). The various possibilities
of circumstantial expressions at clause level are summarized in the following
table:

CIRCUMSTANTIAL EXPRESSIONS AT SENTENCE LEVEL EX.
perfect participle (3)
unipartite gerund (4)
adjective (1)
adjective
non-clausal (pro)noun perfect participle (7), (9)

syntagms
gerund (10)
prepositional phrase9 (4)

bipartite
subjunctive (purpose) (13)
(verbal subordinate general present (indicative) (11), (12)

forms) clauses
prospective (intention) (13), (14)
prospective (unrealized inten-

(15)
(w/backshift) tion)
At text level, different exponents are found expressing circumstantiality,

which is found mostly in the narrative texteme of folktales. These examples
are all opposed to the preterites standing for narrative events and making up
the mainline of the story. The various circumstantial expressions constitute
off-line information:
9 This structure, with the opposite order (prepositional phrase—noun) is a unipartite

existential circumstantial: mpıq-le garxet aw xūwe {ū-p-pumme xa dehwa} ‘The snake
came out again, {(with) a gold coin in his mouth}’ (443).

292 Eran Cohen
NARRATIVE COMMENT MODE
FORM (TRADITIONAL) NARRATIVE FUNCTION(S) AND

NAME BEHAVIOR
šqıl-le preterites representing narrative events

qam-šāqıl-le
šqıl-wā-le plupreterites
1 qam-šāqıl-wā-le pluperfects (mostly) punctual, explanatory
šqīla wēle
2 k-šāqıl-wa imperfect(ive)
x–wē(wā)le, character- and circumstantial descrip-
3 copular clauses tions
pıšle–x
boundary mark- commute with a

4 basır clause temporal clause er finite, independent
verbal clause
dependent shifting between commute with a

5 constructions absolute phrase parallel stories finite, independent
verbal clause
6 wēle-x presentative con-

structions when prepose when postposed
=grounding CC =local CC
7 ay dammıt clause temporal clause
First, specialized verbal forms – imperfective and plusquampreterite (2 and 1

in the table). To these we may add copular clauses (3) and presentative ex-
pressions (6), all these exponents are non-subordinate, namely, not having a
function inside another clause, but rather forming interdepencies with the
other clauses at play.
Presentative expressions (6) are nevertheless similar to (ay) dammıt clauses
(7) in their sensitivity for order – when preposed they serve as circumstantial
to an entire chain of clauses whereas their postposition means that they are
quite local in scope. Preposed, they tend to have low communicative value
(since they mostly contain presupposed information) and thus serve a more
technical function.
Dependent constructions (5) have been dealt with above, under sentence
level; however, they have been found to occasionally interchange with a
seemingly independent, punctual clause. In this case, they shift the mainline
between different episodes that take place at the same time. They seem to be
related to another type, non-circumstantial but rather textual~adverbial ex-
pression of time – a group of verbs which are used to denote the progress of

time (such as zılle ‘went’ or urre ‘passed’) and constitutes, although it seems
‘independent’, an adverbial anchor:
example and glosses translation structure
ʾur-ra xa šapsa… ‘A week went by...’ “main” clause
PRT.pass-3FS DET week
basır xa šapsa… ‘After a week...’ prepositional
after DET week phase
hīl zıl-la xā šabsa… ‘By the time a week conjunctional
till PRT.go-3FS DET week went (by)...’ phrase
In these examples there seems to be an independent clause here, which has the
same function as the adverbially marked clauses below it. Such substitution
group is analogous to the cases we had in exx. (30) and (31) above, except
that these do not shift between parallel stories but rather are used to mark
various subdivisions in the text (compare also ex. (15)).
References
Cohen, E. 2012. The Syntax of Neo-Aramaic: The Jewish Dialect of Zakho. Gorgias
Neo-Aramaic Studies 13. Piscataway: Gorgias.
Givón, T. 1987. “Beyond foreground and background.” In Coherence and Grounding
in Discourse, edited by Russell Tomlin, 175–188. Amsterdam: Benjamins.
Goldenberg, G. 1985. “On Verbal Structure and the Hebrew verb.” In Language Stud-
ies 1, edited by M. Bar-Asher, 295–348. Jerusalem: Magnes (in Hebrew; English
translation in Goldenberg 1998, 148–196).
———. 1998. Studies in Semitic Linguistics. Jerusalem: Magnes.
Khan, Geoffrey. 2007. “The north-eastern neo-aramaic dialects.” Journal of semitic
studies 52.1: 1-20.
Meehan, Charles and Alon, Jaqueline. 1979. “The Boy Whose Tunic Stuck to Him: A
Folktale in the Jewish Neo-Aramaic Dialect of Zakho (Iraqi Kurdistan).” Israel
Oriental Studies 9: 174–203.
Polotsky, H. J. 1996. “Notes on a Neo-Syriac Grammar”. In Studies in Modern Lan-
guages, Israel Oriental Studies 16, edited by Shlomo Izre’el and Shlomo Raz, 11–
48. Leiden: Brill.
PT= Polotsky, H. J. Zakho Texte 1944-1947 (unpublished)
Ramsay, V. 1987. “The Functional Distribution of Preposed and Postposed ‘if’ and
‘when’ Clauses in Written Discourse.” In Coherence and Grounding in Discourse,
edited by R. S. Tomlin, 383–408. Typological Studies in Language 11. Amster-
dam: Benjamins.

Clause Combining in Epigraphic South Arabian

The Problem of Circumstantial Clause Combining
(CCC) in Sabaean
Jan Retsö, University of Gothenburg
1. Survey of research
In his comparative grammar of the Semitic languages, C. Brockelmann gave a
short survey of what he called Zustandssätze in Semitic (Brockelmann 1913,
501ff.). This syntactic category was defined in semantic and syntactic terms as
clauses which do not bring progression in a narrative but indicate circumstan-
tial conditions to the main event told and are equivalent to predicative com-
plements or temporal adverbials (“…Sätze die nicht einen Fortschritt im Ber-
icht bringen, sondern Nebenumstände der Handlung darstellen und daher für
das Sprachgefühl Prädikative oder Zeitadverbien vertreten”). The marking of
such clauses is made, according to Brockelmann, either by asyndetic attach-
ment to a main clause or syndetic marking by the particle wa- or its equiva-
lent.
It is evident that Brockelmann’s survey was not complete. In his compara-
tive grammar, he often adduces examples from the Modern South Arabian
languages based on the documentation which was available one century ago,
mainly accessible through the documents of the Austrian expedition to South
Arabia in 1898-99, but he never gives examples from the epigraphic South
Arabian languages, which also by then were fairly well documented mainly
through the works of J. Halévy and E. Glaser. The reason was probably that
there was no comprehensive systematic grammatical description of these
languages available.
Such a description appeared for the first time in 1943 written by M.
Höfner. In her grammar, the possible existence of circumstantial clauses is
passed over. She mentions, however, the occurrence of ‘short’ imperfects ‘im
konsekutiven Sinn’ referring to events in the past tense such as e.g. w-ywm
hwṣt-hw … w-yʕqb ḫms1t ḫrfn b-ḍr qtbn ‘when he appointed him … so that

298 Jan Retsö
(so dass) he pursued warfare against Qataban for five years’ (Höfner 1943, 75;
the text is Gl. 481, 2 = CIH 375 = Ja 550). The meaning of ‘konsekutiv’
seems to be ‘causal’ i.e. an event causes another to occur. This is a phenome-
non taken up by later scholars.
The circumstantial clause was introduced to the grammar of Epigraphic
South Arabian by A. F. L. Beeston in his sketch of the grammar of the epi-
graphic South Arabian languages from 1962. Beeston claims the existence of
a construction in Sabaean and Minean similar to the ḥāl-complement in Ara-
bic and gives the following examples:
CIH 548:-4 l-yngs1n s1lḥ-hw w-dmw-m b-s2yʕ-hw
‘if his weapons are defiled, there being blood on his garment’
RÉS 311:1.2 tbʕkrb …rs2w ʕm w-ḏ-rs2wts1 ṣryʕm bn mrqdm s1ṭd ṭly
‘TBʕKRB … priest of ʿAmm – the chief of his priestly college being
ṢRYʕM from MRQDM’.
The first example is from Haram, an area where the texts show some devia-
tions from Standard Sabaean (Stein 2007). The second is Minean. The com-
plements in these two examples are verbless clauses introduced by w- aligning
them with the similar construction in e.g. Arabic and Biblical Hebrew. They
do not say much about the possible ḥāl-construction in Sabaean. As far as the
use of finite verbs in ḥāl-clauses is concerned, Beeston states the following:
The perfect is a narrative tense denoting events in past time and in-
cludes the meanings of both perfect and pluperfect in European lan-
guages. The imperfect conveys a present or future notion, whether this
be in (in main clauses) relative to the drafting of the text, or (in subor-
dinate clauses) relative to the action of the main clause (Beeston 1962,
25).
The last clause then claims the existence of the equivalent of the ḥāl in the
Arabiyya. According to Beeston, the imperfect with w- ‘is frequently not
simply coordinate with the preceding verb, but has a consequential or modal
force’ as in RÉS 3945:2 (Beeston 1962, 61):
hṯb mwy ḏhb-hw rymn w-ykn fnwtm fnwtm
‘[he] repaired [Vsuff] the water supply of his alluvial land RYMN so that
it should be arranged canal by canal’.

The Problem of Circumstantial Clause Combining (CCC) in Sabaean 299
Similarly in RÉS 3945:16:

ʕtb bn ns2n ʔl wḍʔt s2ft-hmw ns1rn ʔlʔltn w-yhrgw
‘he designated [Vsuff] from Nashan those whose dedication to the gods
was ordained, so that they should be slain’.
Beeston also admits that the imperfect in a few cases may function like the
Hebrew consecutive, i.e. as a narrative tense indicating successive events. The
only reference he adduces to Sabaean is to the inscription Ry 535 (= Ja 576),
which contains many such cases which Beeston characterizes as ‘anomalous’.
In his grammar on Sabaean published 20 years later, he describes the dif-
ferent uses of the perfect, one of which is characterized as follows:
[The perfect … may convey] an event anterior to the time of the im-
mediately preceding clause (this being other than the principal verb of
the text). In this case, w-fʕl corresponds to an Ar[abic] ḥāl-structure
wa-qad faʕala and should be rendered by an English pluperfect. This
usage is not always easy to detect, but it has important consequences
for the understanding of a text, since it means that we cannot automati-
cally assume that the sequence of presentation of verbs A w B w C rep-
resents the temporal A → B → C, but on the contrary the temporal se-
quence may be B → A → C, with clause B understood as ḥāl […] The
imperfect may present various degrees of futurity (relative either to the
time of writing or to some anterior event), with or without modal over-
tones (Beeston 1984, 19).
Unfortunately, Beeston does not adduce any examples of subordinate clauses

of this kind with finite verbs as predicate. But it is obvious that he sees the
verbal and clause complementing system in Sabaean being more or less iden-
tical with that in Arabic, i.e. the Arabiyya. Among Beeston’s examples, one
might quote the following one which, without being directly referred to in the
passages quoted above, still could be seen as proof of his assumption:
RÉS 3945:15-16: nḍw gnʔ hgr-hw ns2n ʕd hs2rs2-hw w-hgrn ns2n yhḥrm
bn mwfṭm
‘he tore the wall of his town Nashan until he destroyed it completely, but
he forbade the town Nashan (or: the town Nashan was forbidden) to be
burnt’.

300 Jan Retsö
The tense reference of the verb yhḥrm in this example fits quite well into
Beeston’s model.
Beeston thus discerns three different functions of the subordinate imperfect
in Sabaean: ‘present/future’, ‘consequential’ and ‘consecutive’ (or successive)
relative to the action of the main clause. The main characteristic of the ḥāl-
complement in Sabaean is thus, according to Beeston, identical with the mod-
ern concept of gram-switch, i.e. the breaking of a series of finite verbs with
identical TAM form by another one.
In 1965 Ja. B. Gruntfest published a study called ‘Consecutive Construc-
tions in South Arabian’ in which he tackled the problem that had become
visible in Beeston’s grammar from 1962, viz. that the imperfect seems to have
two functions: as a marker of contemporaneity or rather non-past like its
equivalent in Arabic, and as a narrative tense similar to the ‘imperfect consec-
utive’ (wayyiqtol) in Biblical Hebrew.
According to his study, ‘consecutives’ are found in Minean and Sabaean,
not in Qatabanian. Finite forms in Sabaean have temporal meaning, perfect
designating past tense, and imperfect present/future. Gruntfest based his anal-
ysis on quite a small number of texts which he divided into two groups: one
consisting of early texts, viz. the early Sabaean RÉS 3945, RÉS 3946, Ja 550
(= Gl 481), RÉS 3858, RÉS 4176, and the Minaean RÉS 2975 and RÉS 3022,
the second encompassing fourteen texts, the most important of which are RÉS
3427, Ry 535 = Ja 576 + Ja 577, Ja 574, Ja 601, Ja 602, Ja 633, most of which
are now dated to the Middle Sabaean period.
From the first group we quote the following examples:
Ja 550:2: w-ywm hwsṭ-hw ykrbmlk wtr w-yʕqb b-ktbn bʕly s1bʔ w-ʔs2ʕbn
ḫms1t ḫrfn b-ḍr qtbn.
Gruntfest translates: ‘and when YKRBMLK WTR appointed him and he led
the troop of (against?) Sabaʾ and the tribes [during] five years in the war of
QTBN’.
RÉS 3945:16: w-ʕtb bn ns2n ʔl wḍʔt s2ft-hmw ns1rn ʔlʔltn w-yhrgw
Gruntfest translates: ‘and he appointed those from Nashan concerning whom

order came from the gods and they were killed’. This translation indicates
successive or sequential events: ‘first … and then’. In his comment, however,
he seems to indicate that the relationship between the main clauses with the

verb in the perfect tense, and the subordinate imperfect is that of causality:
YKRBMLK WTR commanded him and because of this (poėtomu) he led the
troop; he appointed people from Nashan and because of this they were killed.
The imperfects thus have a consecutive (or rather causal) meaning as indicat-
ed by Höfner in 1943. We notice the difference between Gruntfest and
Beeston in the interpretation of RÉS 3945:16.
The same structure should be seen in the following examples (the first of
which Gruntfest does not translate):
RÉS 3945:1-2 w-ywm hʕḏb mʕs2rt s1bʔ … w-ytʔmmw w-yḥtẓyw … w-
yns2ʔ… 16-17: w-s1tmḫḍ … w-ymḫḍw
A translation based on the RÉS reads as follows: ‘and when he renewed the
tribal assembly of Saba … so that they obeyed and became successful … and
he appropriated … so that it was handed over’.
RÉS 4176:3 w-hwṣt tʔlb …l-ġrḍ … w-yġrḍw
‘and [the god] Taʾlab commanded … to speed up the work … and they
speeded up’.
RÉS 3945:2 w-ywm ṣdq ʕṯtr w-ʔlmqh ḥg-hmy w-yhṯb mwy ḏhb-hw rymn w-
ykn fnwtm fnwtm w-ḏrym ḏrym.
‘and when he fulfilled the commandment of Athtar and Ilmuqah and re-
paired the water of his rain-watered field RYMN so that it became canals
and terraces’.
RÉS 3945:14 w-ywm ns2ʔ ṯnym mns2ʔm w-ygnʔ gnʔm
‘and when he went out to war a second time and constructed a wall’.
In RÉS 3945:2 the first subordinate consecutive governs a second consecu-
tive. In the two last examples Gruntfest sees a ‘weakening’ of the consecutive
meaning of the subordinate verb, becoming more like a stylistic variation of
the past tense. Gruntfest rejects Beeston’s interpretations of these passages as
subordinate final clauses ‘so that … should’ in accordance with the possible
meaning of finite ḥāl-constructions in Arabic. In Gruntfest’s analysis the
subordinate clauses mark factual consecutive, not modal (see further below).
In his view, modal subordination is marked by the particles l- and k-, not by
w-.

302 Jan Retsö
The most thorough analysis of texts from the second group is given to the
Sabaean Ja 576 + Ja 577 (= Ry 535), in spite of some lacunae one of the long-
est Sabaean texts preserved, written during the joint reign of the Sabaean
kings Ilsharaḥ Yaḥḍub and Yaʾzil Bayyin in the 240ies CE. The verbs in the
narratives alternate between imperfect and perfect in a way in which it is
difficult to see any governing rule at a first glance. Beeston characterized this
text as ‘anomalous’. Gruntfest’s conclusion is that the ‘consecutive imperfect’
by this time has degenerated into an alternative past tense marker used as a
stylistic variation. The explanation of the haphazard use is thus diachronic: it
is a survival of the earlier function of the imperfect as a consecutive, clearly
observable in the text from the earlier group. The old consecutive ‘so that’,
indicating factual consequence, has developed into a narrative past tense.
Gruntfest‘s analysis is important to this discussion since its consequence is
the rejection of the existence of verbal circumstantial clauses in Sabaean. As
we can see from his translations and comments, his term ‘consecutive’ is not
altogether clear. Sometimes it simply implies an event following another;
sometimes it indicates a consecutive or causal relationship between two
events. The evidence from Ry 535/Ja 576 is also problematic since this text is
quite idiosyncratic. Gruntfest himself suggests that large parts of it constitute
an extract from a literary work, a royal chronicle or the like.
In his grammar of the Ancient South Arabian languages from 1966, G. M.
Bauer basically accepts Gruntfest’s analysis (Bauer 1966, 7-77) without dis-
cussing the differences of opinion between Gruntfest and Beeston. He
acknowledges, however, the existence of circumstantial complements but
does not give any concrete or systematic description (Bauer 1966, 112).
The latest scholar who has commented extensively on this issue, N. Nebes,
also adheres to the idea that the finite verbal system in Arabic represents a
system of relative tenses. Nebes’ analysis of Sabaean is based on the concept
of a relative tense system applied on Semitic especially by the German schol-
ars A. Denz for an Arabic dialect (Denz 1971), W. Gross and R. Bartelmus for
Biblical Hebrew (Gross 1976; Bartelmus 1982) and M. Streck for Akkadian
(Streck 1995; cf. Cook 2012, 7ff.). This model is also followed by S. Wenin-
ger in his description of the verbal system in Geez (Weninger 2001).
Nebes presents his view of the Sabaean verbal system in detail in his study
from 1994. This study is based on a corpus of ca. 950 cases of prefix forms in
Sabaean epigraphic texts from ca. 700 BCE to the 6th century CE.

Characterising the suffix form as marking anteriority (Vorzeitigkeit), i.e.

past tense in relation to a reference point that can be located before, together
with, or after the ‘present’ of the text, i. e. Speech-time, the prefix forms are
said to mark simultaneity (Gleichzeitigkeit) and posteriority (Nachzeitigkeit;
Nebes 1994, 204; cf. Nebes 1990, 63; Nebes and Stein 2004, 465).
In a paper from 1990, Nebes had already discussed the circumstantial
clause in Sabaean. It is defined as the equivalent of the Arabic wa-huwa
yafʕalu. The Sabaean equivalent to the Arabic wa-huwa yafʕalu should be w-
hʔ yfʕln which, according to Nebes, is not found in Sabaean (and not in Bibli-
cal Hebrew either, cf. Nebes 1994, 208). What we do find in Sabaean is a
circumstantial clause with nominal constituents as in the Harami text CIH
533:2-4:
bhn qrbh mrʔ ywm ṯlṯ ḥgtn w-hʔ ḥyḍ
‘because on the third day of the pilgrimage a man had approached her,
when she was menstruating´ (cf. Nebes 1990, 65-66).
The absence of verbal circumstantial clause in Sabaean is, according to Nebes,
due to the fact that Sabaean does not possess a finite verbal form denoting
relative simultaneity. The prefix form can only mark simultaneity with the
‘now’ of the text, i.e. it can be an ‘absolute’ present tense of the kind we find
in European languages. In a subordinate clause, however, the prefix form can
only mark posterity (Nebes 1990, 66-67).
A complication is the use of prefix forms in main clauses in what seems to
be continuous narratives referring to the past. Beeston also refers to its use as
presenting a situation envisaged as a consequence or as a concomitant result
of a preceding past-time act (Beeston 1984, 20; cf. Bauer 1966, 77). Nebes’
German term for this is ‘Progressfunktion’ which is defined as ‘activities …
which in a context of actions move forward in time’ (‘Tätigkeiten … die in
einem Handlungszusammenhang zeitlich fortschreiten’; Nebes 1994, 206; cf.
Nebes 1995, 85f., 190 ff., 269 f.). The word progression will be used here as
an English equivalent. By this Nebes explains the use of prefix forms in narra-
tive past context and associates it with the so-called imperfect consecutive in
Biblical Hebrew. This should also be seen as ‘progression’, i.e. indicating
posteriority in relation to an event preceding it in time (Nebes 1994, 208-209).
Nebes sees the verbal system in Sabaean as being very close to the one in
Biblical Hebrew. In his opinion the BH yiqṭol (with or without a preceding
w-) denotes simultaneity or posterity in relation to the textual ‘now’ but only

304 Jan Retsö
posterity in relation to relative ‘now’. A special case is, then, the marking of
‘progression’ in BH. This language has a special form, the so-called imperfect
consecutive, consisting of the apocopated form of the imperfect with a pre-
fixed wa- (+ an originally geminated consonant) marking the ‘Progressfunk-
tion’, viz, posterity in relationship to a preceding event: ‘and then’ or ‘after
that’. The difference between BH and Sabaean is that the latter does not have
any special marking of progression (at least nothing observable in the unvo-
calised texts we have). The Sabaean prefix forms thus only mark ‘progres-
sion’/consecutivity and posterity in subordinate clauses, not simultaneity.
According to Kogan and Korotaev (1997, 235), the imperfect in Sabaean
denotes actions in the future or actions without time restriction, but also ac-
tions in the past regarded as consequences of other past actions. They quote Ja
631:29-31: w-bʕww b-llyn ḥyrt ʔḥbs2n w-yhrgn bn ʔḥbs2n ʔrbʕ mʔnm ʔs1dm
‘they attacked the camp of the Habashites at this night and killed of the Ha-
bashites 400 men’. There are also cases with imperfect without w- indicating
past actions without consecutive meaning such as Ja 631:28: w-l-ṯlṯm ywmm
ybrrn ‘and on the third day they came into the open [to fight]’.
Stein (2003), Nebes and Stein (2004), Stein (2011; 2013) basically follow
Nebes (1990) and (1994). The tense system is seen as a relative one, the im-
perfect expressing simultaneity and posteriority (Stein 2003, 166f; Nebes and
Stein 2004, 465; Stein 2011, 1063-65; id.: 2013, 131-133). The latter leads to
its function as a progressive past. Circumstantial clauses expressing simulta-
neity with verbal predicate, such as the Arabic wa-huwa yafʕalu, ‘cannot be
identified in Old South Arabian with certainty’, only clauses with a nominal
predicate (Nebes and Stein 2004, 476; Stein 2013, 165).
2. Evaluation of the research

A crucial factor, in the discussion about circumstantial clauses in Sabaean,
thus turns out to be the function of the imperfect, the Vpref. We have seen that
scholars have differing opinions on that issue. According to Beeston, it may
indicate contemporaneity with absolute past. Gruntfest sees it as a variant of
absolute past originating from a consecutive function. Nebes claims that it
always marks posterity in the absolute past.
Beeston’s analysis is quite sketchy and he does not give a complete com-
parison between the Arabiyya system and the Sabaean one. This is problemat-

ic since he does not provide any good examples which would relate the two
systems to each other.
Gruntfest’s conclusions relate Sabaean more closely to Biblical Hebrew.
But his analysis leaves the reader confused because one has to assume a rather
haphazard use of the two basic finite verbal forms of the language in some
literary texts in Sabaean. The claim, that Vpref and Vsuff can be used more or
less optionally, is somewhat unsatisfactory. It can be argued that there should
be a difference between causality and successive events: the latter are not
necessarily caused by a preceding event. That this interpretation is not un-
problematic is clear from Beeston’s translation of the example from RÉS
3945:16: ‘he designated from Nashan those whose dedication to the gods was
ordained, so that they should be slain’. Beeston uses another concept viz. that
of intended consecutivity: the sentence does not say that these people were
killed, only that this was the intention. In classical Greek a distinction is made
between intended and factual consecutivity and it might be worthwhile to
keep this distinction in mind.
Consecutivity and causality are categories that establish a kind of link be-
tween events which, as a rule, is absent in a narrative of successive events.
This kind of linking is the one intended by the concept progression used by
the German scholars. Such a linking of events does thus not necessarily indi-
cate any consecutive, causal or final relationship. It goes without saying that
these distinctions can be crucial for an understanding of a text. Another ques-
tion is whether the language codes these distinctions and, if so, in which way.
Nebes has a narrow definition of circumstantial clauses. His definition of
the ḥāl-complement as wa-huwa yafʕalu i.e. S+Vpref does not conform to the
traditional definition which is broader (see below). His claim that the Sabaean
Vpref marks simultaneity only in main clauses, and never in subordinate or
complementary ones, is an allegation which remains problematic. According
to Nebes, Sabaean Vsuff fʕl is ‘vorzeitig’ regardless of the Relationswert.
Vpref yfʕl(n) is ‘gleichzeitig’ and ‘nachzeitig’ in main clauses but not ‘gleich-
zeitig’ in subordinate clauses. As will be shown here, there are facts which
speak against.
If Nebes’ analysis is correct, Sabaean, unlike Biblical Hebrew, would not
make any visible distinction between marking of posterity and progression. A
problem is that Sabaean indeed seems to make such a distinction, viz. by the
employment of the infinitive. Thus, in a middle Sabaean text we find a se-
quence like this (Ja 576:3):

306 Jan Retsö
w-ṯbrw w-hbʕln w-qmʕ w-hs1bʕn

‘and they destroyed [Vsuff] and seized [inf] and subjugated [inf] and
plundered [inf]’.
The perfect ṯbrw is followed by three infinitives, in derived and augmented
verbal forms usually marked with an n-suffix. These forms undoubtedly mark
progression in accordance with Nebes’ analysis and would thus be an alterna-
tive way of making this distinction. Sabaean would thus have two different
ways of marking progression. Although not impossible, this still is somewhat
unsatisfactory. The use of non-finite infinitival forms like finite forms is also
sporadically found in Biblical Hebrew where the so-called absolute infinitive
may be employed in this way. But, unlike the Sabaean infinitive, this is quite
rare in BH and the absolute inifinitive does not replace the consecutive (pro-
gression-marking) but may replace all kinds of finite forms (Gesenius 1910 §
113.4 (y)).
Nebes’ claim that circumstantial clauses with finite verbs do not exist in
Biblical Hebrew also has to be sustained by detailed argument since most
descriptions of that language assumes this to be the case. Nebes’ claim that
Biblical Hebrew has no verbal circumstantial clauses denoting simultaneity
thus remains controversial and cannot be used as a supporting argument for
the absence of circumstantial clauses in Sabaean. Several scholars, if not most
of them, assume the contrary to be the case. There is a plethora of examples in
the Hebrew Bible which look very much identical to the circumstantial claus-
es yafʕalu syntagm in the Arabiyya (see eg. Gesenius 1909, 489-491; Driver
1892, 31-33, 206; Davidson 1901, 185-190; Meyer 1972 III, 92; Andersen
1974, 77ff.; Joüon/Muraoka 1996 § 159; Isaksson 2009; Eskhult forthcoming,
32). The examples given by Waltke/O’Connor (1990, 504) Nebes rejects
without any argument (‘… sind anders zu interpretieren’). Others have argued
against as well (Kuhr 1929, 37; Joosten 2012, 125ff.). The argument against
their existence seems to be that since yiqtol (long Vpref) does not mark pre-
sent tense in main clauses (this being the task of the active participle) it cannot
have this function in CC complements (Joosten 2012). This argument is, how-
ever, doubtful. Most scholars admit that yiqtol can have this function in the
poetic language, especially in archaic poetry. This usage thus belongs to an
earlier stage of the language. It is thus not to be excluded prima facie that it
might have preserved this function in certain syntagms in later prose even if,

admittedly, of rare occurrence. The claim that yiqtol never marks simultaneity
leads to strained explanations of many passages (cf. Cook 2012, 219f).
It is worth adducing a few prosaic passages that by many scholars are seen
as containing circumstantial clauses (Waltke/O’Connor 1990, 504):
wayyåḇoʾ ḥûšay … håʕîr w-ʔaḇšålom yåḇoʾ yrûšålåyim
‘Hushay came into the city while Abshalom was entering Jerusalem’ (2
Sam. 15:37).
wayyånuʕû ʔammoṯ hassippîm … w-habbayiṯ yimmåleʾ ʕåšån
‘The doorposts shook … while the temple was filled with smoke’ (Isa.
6:4).
wayyaʕalû ʔîš yisråʔel hûʾ qåm wayyaḵ bapplištîm … w-håʕåm yåšuḇû
ʔaḥaråyw ʔaḵ lp̄ aššeṭ
‘And the men of Israel retreated. He stood his ground and struck the Phil-
istines … while the troops returned after him but only to strip the dead’ (2
Sam. 23:9-10).
wattiṯpallel ʕal YHWH ûḇåḵō tiḇkǣ
‘She prayed to YHWH weeping’ (1 Sam. 1:10).
These examples have two characteristics which mark them as having a special
function: S+V word order and a gram-switch. They are also characterized by
chiasm. In Classical Greek one would have the mèn … dè construction. One
might compare a modern Arabic translation of these passages which shows
how Arabic can handle different kinds of relationships between events in a
narrative.
fa-ʔatā ḥūšay … wa-ʔabšalūm yadḫulu.
fa-htazzat …wa-mtalaʔat.
fa-ʔaqāma wa-ḍaraba … wa-raǧaʕa.
fa-ṣallat … wa-bakat bukāʔan.
The main narrative line is marked by fa- + Vsuff (‘perfect’). Then there are
two ways of marking simultaneity, either by wa- + Vsuff or wa- + Vpref. It is
obvious that the Arabic translator has interpreted these passages as containing
different kinds of circumstantial clauses. This also shows how the analysis is
dependent on an understandig of the narrative structure of the text. If one
acknowledges the existence of circumstantial clauses with a verbal predicate

308 Jan Retsö
indicating simultaneity (as well as anteriority and posterity) in Biblical He-

brew, this does not immediately disprove Nebes’ analysis of Sabaean but the
support of a parallel in another Semitic language disappears.
From the survey given here, it also appears that the scholars who have
commented on the question about (verbal) circumstantial clauses in Sabaean
have used quite different concepts and definitions. The usual definition of ḥāl
is that it gives a comment on a clause which refers to events or states simulta-
neous or prior to that clause. In the Arabiyya-system the simultaneous event is
coded by the imperfect indicative, the prior event by the perfect form, often
preceded by the particle qad. This is often seen as a kind of gram-switch, i.e.
the breaking of a series of parallel morphosyntactic elements, signaling anoth-
er syntactic-semantic category. The scholars quoted add several other con-
cepts: finality (Brockelmann), consecutivity (Höfner), causality (Gruntfest)
and progression (Gruntfest, Nebes); concepts which should be kept apart.
3. The question of tense/aspect

The problem of tense versus aspect will not be discussed in detail here. Suf-
fice it to quote a clarifying presentation of the categories involved, ultimately
based on the writings of Carlota Smith as referred to by Notarius (2011, 277f.,
cf. Cook 2012, 7-18) and basically followed in this study:
Time in language involves a linkage between three times – Event Time

(ET), Reference Time (RT) and Speeech Time (ST) and is controlled
by by three possible relations (anteriority, coincidence, posteriority)
within two temporal nodes: (1) ST and RT, (2) RT and ET.
There are three patterns of temporal interpretation in discourse – deic-
tic, anaphoric, and sequential: deictic time establishes reference to ST,
anaphoric time refers to another, non-ST, contextually established RT,
and sequential time builds an autonomous temporal succession of
events/situations, usually in chronological order.
The three time patterns bracket together with three groups of situation-
al entities […] eventualities, which include specific events […] and
states; general statives […] and abstract entities.
In addition to the time patterns and situation entities Smith postulates
regularities of text-progression. Text-progression can be temporal
through the temporal locations of eventualities, spatial through the spa-

tial location of states, and metaphorical through the informational

space of the text.
These three temporal dimensions(time in discourse, aspectual entities
in discourse and text progression) shape the linguistic character of five
discourse modes narrative […], report […], description […], infor-
mation […] argument.
Without saying it explicitly, Beeston, in his two studies on Sabaean, assumes

a relative tense system for Sabaean (as well as for Arabic) which is also partly
assumed by Nebes for Sabaean. Translated into the terminology quoted above
this means that the suffix form (the ‘perfect’) locates an event anterior to a
reference time (RT), which in its turn can be coincidental, anterior or posterior
to ST. The prefix form (the ‘imperfect’) locates an event as coincidental with
RT which in its turn may be anterior, coincidental or posterior to ST. This is
thus what Smith/Notarius label anaphoric time. The concept of progression, as
used by the German scholars, would be sequential time in accordance with the
Smith/Notarius model. There is thus a possibility that a language marks ana-
phoric time and sequential time in the verbal system but not deictic time. In a
language with such a system all verbal forms at a first glance can refer to
anteriority, coincidence or posteriority vis-à-vis ST. This has led many schol-
ars to claim that Biblical Hebrew is tenseless language (Notarius 2011, 276).
This is correct only if tense is defined as absolute tense which, however, is an
unnecessary narrowing of the concept.
4. The ḥāl concept: The case of the Arabiyya

It is obvious that the concept of ḥāl complements in the grammar of the
Arabiyya plays a central role in the discussion about circumstantial clauses in
other Semitic languages including Sabaean. Beeston leans heavily on the
Arabiyya-grammar in general in his description of Sabaean and its sister lan-
guages. Nebes’ starting point is the Arabiyya syntagm finite verb + wa- +
explicit subject + imperfect indicative which he claims does not exist in Sa-
baean. Both scholars take their point of departure in morphosyntactics. But, as
we have seen, there are also other concepts involved in the discussion such as
‘consecutive’ and ‘progression’ which are semantic. The discussion is part of
a larger issue, viz. the combining of clauses with finite verbs, which semantic
categories are involved and how they are marked, of which a tentative typolo-

310 Jan Retsö
gy is presented by Dixon (2009). In order to clarify concepts, a short survey of

how the relevant syntagms are handled in the basic descriptions of the gram-
mar of the Arabiyya is a good starting point. But the issue is also connected
with how the verbal system in these languages is understood. The notion of
gram switching and its function, which plays a crucial role in the description
of the phenomenon, is intimately connected with the TAM system. In order to
understand how the relevant phenomena appear in Sabaean, it is necessary to
have some grasp of the position of the Sabaean verbal system within the Se-
mitic languages as a whole. The following is a preliminary sketch of this
background.
In the standard descriptions of the Arabiyya the following syntagms are
relevant to this discussion (cf. Waltisberg 2009, 2-19).
1) Progression
The concept progression (not to be confused with the term progressive) is
often used by German scholars together with, or instead of, consecutive in the
sense of the presentation of successive events in a narrative structure. But,
since consecutive also has other meanings, it is advisable to make some dis-
tinctions. The term progression should be reserved for a series of independent
events that follow each other in a narrative line. In the Arabiyya it is marked
by the particle fa- The fa- indicates the temporal succession of the reported
events (Reckendorf 1895, 463; Reckendorf 1921, 316-320; Fischer 1987
§329; Waltisberg 2009, 5; Isaksson 2009, 39ff. cf. Dixon 2009, 9ff.). It should
be observed that the subsequent verb may be Vsuff or Vpref indicative which
actually may indicate several kinds of progression. Nebes distinguishes
between ‘zeitlicher Progress’ and ‘folgerndes Progress’ (Nebes 1995, 190 ff.).
The latter belongs to the syntagms treated in the following paragraphs:
fa-stabaqnā ʔilā ḏī ḥusumin fa-sabaqnā-hum ʔilay-hi fa-nazala l-ḥusaynu
fa-ʔamara bi-ʔabniyati-hi fa-ḍuribat
‘We raced towards Dhū Ḥusum and we got there before them. Then
Ḥusayn dismounted and gave orders about his tents and they were erected’
(quoted by Isaksson loc. cit.)
2) Juxtaposition
When linked by wa-, the events coded by the verbs are not ordered in a chain.
Events are presented on the same temporal reference point but there is no

temporal structuring between them (Reckendorf 1895, 446-455; Reckendorf

1921, 320ff.; Waltisberg 2009, 5). We here have a case which Smith/Notarius
would classify as spatial textual progress and which is part of Dixon’s catego-
ry Addition (unordered or same-event, Dixon 2009, 26ff.). A suggested term
for this is juxtaposition:
fa-ʔarsala ʕazza wa-ǧalla s-sakīnata wa-hiya rīḥun ḫaǧūǧun … fa-
taṭawwat ʕalā mawḍiʕi l-bayti … wa-ʔumira ʔibrāhīma ʔan yabniya … fa-
banā ʔibrāhīmu wa-baqiya ḥaǧarun fa-ḏahaba l-ġulāmu yabnī šayʔan…
‘God sent the sakīna which is a strong wind … and it encircled the place
of the House … [by then] Abraham was ordered to build … Abraham then
built but a stone remained (= was missing). Then the young man went to
build something else…’ (aṭ-Ṭabarī, Tārīkh I:275).
If verbs coordinated by fa- may be translated with ‘and then’, verbs with wa-
would be ‘and also’, ‘further’. In the example just quoted we see that not only
the nominal circumstantial clause is introduced with wa- but also clauses with
Vsuff. The wa- indicates a pause in the sequence of events telling an addition-
al event (cf. also Reckendorf 1895, 456).
3) Focus
On looking through the examples of ḥāl and ‘consecutive’ constructions with
fa- listed by Reckendorf and others, there appears a category that might be
singled out under its own label. Consider the following (Reckendorf 1921 §
164):
sakata saktatan fa-ʔaṭāla-hā
‘he was quiet for a long time’ (lit.: ‘he was quiet and he prolonged it’)
karra bi-nā r-rikāba fa-ʔasraʕa
‘he made his horse rush towards us with great speed’ (lit.: ‘he made the
horse rush … and speeded it up’).
The verb marked by fa- in these cases hardly implies a subsequent action.
Instead, they indicate a close-up of the main verb, by a more detailed exposi-
tion of the verb. This is equivalent to Dixon’s term elaboration, a subcategory
of Addition (Dixon 2009, 27ff.).
There is a further asyndetic variant of this (Wright II, 288; Waltisberg
2009, 6):

312 Jan Retsö
saǧada ʔaṭāla
‘he prayed for a long time’
ġannā ʔaḥsana
‘he sang beautifully’
qāma saǧada la-hu
‘he arose and prostrated himself before him’
ʔarsala ʔaʕlama ʔabā-hu
‘he sent (and) informed his father’.
4) Factual consecutive
Like the ḥāl concept, the term consecutive as used in the discussion seems to
cover several distinct categories. The definition suggested here is that of
events that state a consecutive relation, i.e. an event which in some ways fol-
lows from another one. The category (equivalent to Dixon’s Consequence)
may be divided in consequences that are factual and those which are modal,
i.e. imagined, wanted, feared etc. (Dixon 2009, 17ff.). It then appears that the
Arabiyya would use fa- for the factual consecutive. The tense in both clauses
is identical which makes it morphosyntactically but not semantically identical
with category 1) above (Reckendorf 1895, 455-466):
ḍarabtu-hu fa-bakā
‘I hit him and then ( = so that) he cried’
ʔamara bi-l-bābi fa-ʔuġliqa
‘he gave an order about the door and it was shut/then ( = so that) it was
shut’
5) Modal consecutive
For the modal, i.e. intended consecutive, the rule given is with fa- + Vpref
subjunctive (Reckendorf 1895, 747ff.).
yā nāqu sīrī ʔilā sulaymāna fa-nastarīḥā
‘O camel! Go to Sulaymān so that we may rest!’
But the subjunctive is not all prevailing. Even Vpref indicative occurs:
karihtu ʔan ʔaḫruǧa-kum fa-tamšūna fī ṭ-ṭīni

‘I loathed to let you go out so that you might walk in mud’

The subjunctive may be ambiguous:
lā taʔtī-nī fa-tuḥaddiṯa-nī
‘you do not come to me 1) so that you might tell me’ 2) ‘so that you tell
me’ (you come but cannot tell).
6) Coordinated circumstantial clause

This syntagm is defined as complements coordinated to a preceding finite
verb with wa- + subj. + finite verb. Unlike the cases treated above, the verb
has a relative tense value with its reference point indicated by the governing
verb (Reckendorf 1895, 553, 556; Fischer 1987, 185f.).
ʔaqbalat ʕīrun wa-naḥnu nuṣallī
‘Camels came while we were praying’
kataba bi-ḏāka ʔilā hišāmin wa-yastaʔḏinu-hu fī-hi
‘He wrote about this to Hisham asking him for permission for it’
Both circumstantial clauses present an activity which is going on when some-
thing new happens. The simultaneity of the situation is coded by the Vpref
verb in the circumstantial clause. It is actually the classical coincidence case.
The difference between the two sentences is that the complement clause
marked by wa- in the first one has a subject different from the governing
clause whereas in the second the subject is identical. We observe that in the
latter case we actually do not have a circumstantial clause according to the
definition since there is no element before the verb.
ʔa-nuʔminu laka wa-ttabaʕa-ka l-ʔarḏalūn
‘Shall we believe you when contemptible men have followed you?’
fa-ntabaha wa-qad šaddū-hu
‘He was informed, in that they had arrested him’.
These clause complexes include circumstantial clauses marking events which
are anterior to the event coded in the governing clause, marked by Vsuff
verbs.

314 Jan Retsö
7) Predicative circumstantial clauses

This syntagm is defined as an asyndetic complement (verbal, nominal), i.e.
complements with no linking element (Reckendorf 1895, 550, 555; Recken-
dorf 1921 § 450; Fischer 1987 § 408, 431). A suggested characterisation of
the complement in these cases is a concretization/singling out or adding a
feature to the governing verb (underlined in the examples):
fī-hi taqūlu ʔuḫtu-hu tarṯī-hi
‘His sister talks about him lamenting him’.
ǧāʔū-kum ḥaṣirat ṣudūru-hum
‘They came to them their bosoms being contracted’ (cf. Q 4:90).
daḫala l-bayta lā yusallimu ʕalay-ya
‘He entered the house without greeting me’.
This construction often has a final meaning largely depending on the seman-
tics of the verb. It occurs especially after verbs of motion:
baʕaṯa ʔilā muʕāwiyata yaṭlubu ṣ-ṣulḥa
‘He sent a messenger to Muʔāwiya asking/in order to ask for armistice’
ḫaraǧtu ʔanā wa-ʔabī nataṣayyadu
‘I went out with my father to hunt’.
This syntagm can be transformed into a noun in the accusative case:
qāma māddan ṯawba-hu bi-yadin = qāma yamuddu ṯawba-hu bi-yadin
‘He stood up straightening his mantle with his hand’.
Summing up this survey, we thus have the following main semantic categories
involved in the discussion about the circumstantial clause:
1 progression ‘and then’

2 juxtaposition ‘and also’, ‘as well as’
3 focussing ‘more precisely’
4 factual consecutive ‘and because of that [it happens]’
5 modal consecutive ‘so that it may/might [happen]’
6 coordinated circumstantial complement = clausal attribution ‘while’
7 predicative circumstantial complement = verbal attribution ‘in that’

They can be classified into the following categories:

Progression (=1): the introduction of new verbal sememes, referring
events occurring in temporal succession, more or less independent of each
other. This is Dixon’s subcategory ‘temporal succession’, one of his tem-
poral linking types (Dixon 2009, 2).
Focus (=2-3): attribution of characteristics to a verb, derivable from the
verbal sememe, belonging to the semantic sphere of the verb, or simply
characterisation of the verb. This would be Dixon’s linking type Addition
subcategorized as ‘unordered addition’, ‘same-event addition’ or ‘elabo-
ration’ (Dixon loc. cit.).
Consecutive (=4-5): events/states following each other in succession but
with a causal link between them, either factual or imagined, Dixon’s link-
ing type Consequence.
Background (= 6-7): commenting the circumstances in front of which the
events and states take place. This is more difficult to integrate into Dix-
on’s linking types since it may belong to Temporal as well as Addition.
The morphosyntactic means available in the Arabiyya for coding these cate-
gories are the following (MC = main clause):
a. MC wa-+pron/subj. + Vpref/Vsuff
b. MC Vpref/Vsuff
c. MC V identical tenses
d. MC fa+Vpref/Vsuff identical tenses
e. MC fa + Vpref subjunctive
f. MC wa+ Vpref./Vsuff. identical tenses
It is 6) and 7) / a) and b) which represent the traditional finite ḥāl-clause in

Arabic. But it is clear that the semantic structures of these two categories are
not limited to them. From the survey of the Arabiyya it appears that there is
no watertight distinction between these categories. On the whole, the tradi-
tional definition of verbal ḥāl-clauses can definitely be questioned (cf. Nebes
1990 n. 6). The circumstantial clause in case 7) often seems to indicate an
addition to the semantic contents of the governing verb, not a specification,
even if the complement can often be located in the ‘semantic sphere’ of the
governing verb. To send someone can often imply asking for something (but

316 Jan Retsö
not necessarily); taking off one’s clothes can imply a following washing (but
not necessarily), and so forth. We are dealing with a kind of attribute to a
verb, an extension of the verbal domain. A complement of the kind described
in 7) can also have a final meaning: ‘he did this in order to…’. There is no
explicit signalling when this is intended.
It might be observed that there is no clear-cut semantic dividing line be-
tween syntagms 3), 5), 6) and 7). The traditional verbal ḥāl-clause thus turns
out to represent two different categories: background comment to a clause and
extension of the domain of a verb. But these distinctions can also be coded by
other syntagms than the traditional ḥāl-clause. The constructions are often
replaceable by each other. The two sentences in 3) above could also be formu-
lated as 7):
sakata saktatan yuṭīlu-hā
karra bi-nā r-rikāba yusriʕu
A clause like ‘God’s messenger asked’ can be formulated in the following
ways:
saʔala rasūlu llāhi yaqūlu
saʔala rasūlu llāhi qāʔilan
saʔala rasūlu llāhi fa-qāla
The fa-construction, followed by a verb in the same tense as the preceding
one, is obviously a variant of the traditional asyndetic ḥāl-clause. It is thus
evident that the different morphosyntactic means used by the language do not
always signal one category only. This is partly due to the fact that it is not
always possible or desirable for the language to make the proper distinctions.
5. The Sabaean verbal system in a Semitic perspective

The existence in Sabaean of clauses with nominal constituents functioning
like the corresponding Arabic ḥāl-clauses seems not to be contested by any of
the scholars (cf. Nebes 1990, 64; Stein 2013, 127f.). The bone of contention is
clauses with a finite verbal predicate. The essential facts are that Sabaean
often uses the finite prefix forms (Vpref) in contexts where one would expect
a suffix form (Vsuff). The use of prefix forms as an equivalent to an absolute
past tense in Sabaean is thus unquestionable. The question is if this phenome-
non lends support to the assumption of the existence of circumstantial clauses

in that language, or the opposite. As far as the Vpref is concerned, we notice

that there are divided opinions among the scholars who have commented upon
it: present tense or progression, i.e. consecutive narrative. The question about
circumstantial clauses is thus closely connected with how the Sabaean verbal
system is understood (cf. Nebes 1994).
The basic outline of the system of finite verbs in Semitic languages can be
classified into the following separate systems:
Terms 1. 2. 3.
a) Akkadian iparras iprus -
b) Northwest-Central Sem. yaqtul-u yaqtul qatal
c) Ethiosem. yəqattəl yəqtəl qatal
The term common to all three variants is term 2) and it can be assumed that it
constitutes a common morphological heritage. Term 1) has been characterized
by different labels: imperfective aspect, durative, cursive, progressive, simul-
taneous, and present tense and a definite analysis will not be made here. There
is widespread agreement that 1) is an innovation in the verbal systems of
Akkadian and Ethiosemitic (Rundgren 1959; Kouwenberg 2011, 85-123;
Cook 2012, 95-120), albeit a very ancient one. We may thus notice that Ak-
kadian and Ethiosemitic have used similar morphological means to create this
term. It may be claimed that the NWCS group shares the innovation but uses a
different morphology. The function of term 1) in all three groups, however, is
more or less the same. Since term 1) is innovative we have to assume a priva-
tive opposition as the result: term 1) is the marked term with supposedly a
clear-cut function; 2) remains the default category. So, if term 1) functions as
cursive-present and so forth, term 2) fulfils the remaining categories, not only
perfective, anteriority, past tense etc., but different modalities like future,
volitive, habitual, and irrealis. It may, however, be observed that, since term
1) tends to intrude into the remaining domains of term 2), its function as a
future tense is widespread.
The languages in groups b) and c) have yet one term, 3), which is also con-
sidered an innovation. The morphology of the term shows its origin in the
nominal system and it is usually seen as an original resultative perfect: ‘I
stand’ =/< ‘I have stood up’, originally limited to verbs of certain semantic
classes but developing towards a general marker of anteriority, perfective
aspect or past tense. It has thus intruded into the semantic/functional field of

318 Jan Retsö
term 2) which is receding on all fronts (Cook 2012, 96-97, 99-105). We can
observe how qatal, not only replaces yaqtul in its anteriority/perfective/past
tense functions, but also as a posteriority/future tense, most evident in the so-
called perfect consecutive in Biblical Hebrew.
There are, thus, two major innovations resulting in the verbal systems we
see in the early Semitic languages: innovation I creating a marked cursive-
imperfective-present form, and II creating a perfective-punctual-past form.
We shall not go into the further developments, especially in NWSC where
yaqtul-u later on is replaced in many functions by the participle qātil/qotel
(Joosten 2012). One should observe the difference between NWCS languages
and Akkadian/Ethiosemitic in creating term 1). NWCS uses the yaqtul (iprus)
but adds a vocalic element. The two resulting variants of the Vpref in NWCS
we shall call long (term 1) and short (term 2) respectively. There is a further
differentiation in verbs from weak roots between long and short Vpref with
alternating long and short theme vowels like the Arabiyya yaqūm(-u) # ya-
qŭm; yabnī # yabnĭ. This differentiation is thus not a common Semitic herit-
age (cf. Revell 1984; Kouwenberg 2010, 476f.; 499). It should be added that
the nun paragogicum, sporadic in Biblical Hebrew (yiqṭlūn etc.) but compul-
sory in the long Vpref in some other languages (Phoenician, Aramaic), should
not be considered a basic common characteristic of the long Vpref.
Worth noting is also that it seems clear that, even if the Arabic complex
belongs to the NWCS-languages, not all forms of Arabic have undergone both
innovations. The imperfects with b-, t- and k-prefixes in many modern dialects
should be seen as representing innovation I, i.e. the creation of term 1 (Retsö
2014). But most dialects do not have these devices and do not use participles
as finite forms either. All known forms of Arabic do, however, have innova-
tion II.
The Sabaean verbal system appears as a binary system with Vsuff opposed
to Vpref. The Vsuff is identical to the qatal-form in NWCS and Ethiosemitic.
The Vpref occurs in two variants: one with a –n(n)-suffix and one without.
These are sometimes called long and short imperfects respectively (Stein
2003, 166), which is unfortunate considering the meaning of these terms in
NWCS where they stand for something else. It is better to use a neutral desig-
nation like Imperfect I for the n-less form, and imperfect II for the one with
the n-augment (cf. now Stein 2011, 1061 and id. 2013, 80).

The Vpref paradigm of the 3rd person looks as follows:

Sing. dual. plur.
Strong roots
Impf. I yqtl yqtl-y yqtl-w
tqtl *tqtl-y tqtl-n
Impf. II ytql-n yqtl-nn yqtl-nn

tqtl-n tqtl-nn tqtl-nn
Roots IIIw/y
Impf. I ybny ybny-y ybny-w
tbny tbny-y tbny-w
Impf. II ybny-n *ybny-y ybny-nn

tbny-n *tbny-y *tbny-nn
The shape of the Vpref from IIIw/y roots leads to the assumption that there is
no visible trace in Sabaean of the short imperfect of the NWCS (cf. Beeston
1983, 16; Stein 2011, 1061; 2003, 192, 194 n241: a form l-yʔt from the verb
ʔty ‘to come’ is found three times in one inscription RÉS 4176 but the same
text also has l-yqny). Even if the South Arabian script has no vowel signs it
could well have made a difference between short and long imperfects of
IIIw/y roots as Ugaritic does (Stein 2003, 194) since final long vowels are
written plene. As long as we do not have evidence of the opposite, the conclu-
sion to be drawn is that the Sabaean imperfect I is the ancient iprus/yaqtul
unaffected by the differentiation processes observable in NWCS. Imperfect I
would thus have looked something like *ya/iqtol – *ya/ibnī – *yi/aqūm. At the
same time Sabaean has the Vsuff forms like all the NWCS languages. It thus
shares these two features with Ethiopic (Geez) but unlike that language it does
not have the iparras/yəqattəl form (Nebes 1994b). Noticeable is that Ethiose-
mitic does not have any variation between long and short imperfects either.
From what we can observe from the writing system, Sabaean stands out as a
fourth type of organisation of the verbal system alongside Akkadian, the
NWCS, and Ethiosemitic. It has innovation II but not innovation I. We should
reckon with a qatal with meaning/function similar to the equivalent in Ethio-
semitic and, to a certain extent in NWCS.

320 Jan Retsö
The Sabaean yqtl can thus be expected to have the function of simultanei-
ty/non-past/future etc. But if the diachrony is as has been sketched here, it can
be expected that it also can have the function of anteriority/past tense etc. in
certain syntagms and/or text types, survivals of an earlier stage of the lan-
guage.
The Sabaean system is in many respects reminiscent of the verbal system
visible in Archaic Biblical Hebrew (Joosten 2012, 411-434; Notarius 2013;
Eskhult forthcoming). Many of the poetic texts in the Old Testament exhibit
features in the verbal system that seem to represent a stage prior to the one
found in Standard Biblical Hebrew (see especially Notarius 2013). In ABH
we find several cases where the long and short Vpref are used in the same
way, both as narratives, modalities and as circumstantial complements. Simi-
lar features are found also in Ugaritic (Greenstein 2006). Another group of
Semitic languages where a similar system is found is, remarkably enough,
several forms of Arabic, e.g. modern Arabic dialects like the dialects of Naǧd
and Tchad and several others which do not have traces of innovation I above,
only II. But also the Arabiyya itself, in spite of having had both innovations,
nonetheless often shows a use of verbal forms strongly reminiscent of ABH
and the dialects mentioned. There are, thus, good typological parallels in other
Semitic languages to the structure of the Sabaean verbal system assumed here
which, then, becomes more likely if one liberates oneself from the traditional
idea that ‘Protosemitic’ must have had a morphology more or less identical
with that of the Arabiyya.
The question that arises is: if the yqtl gram can fulfill all these functions in
the language discernible in ABH, in some Arabic dialects and Sabaean; are
there any means in these languages to separate at least some of the functions,
given that there are no morphological means available? The yqtl would basi-
cally have non-anteriority functions, although it is amply documented that it
may well be employed as a narrative tense, and in many instances it is not
possible to see it as a past progressive or the like, distinguishing it from qtl.
Instead, many of those instances might represent syntactic fossils, survivals of
a stage when the Vpref was the sole finite form in the language. If we assume
that there ever was such a stage in the forebears of Semitic, it is likely that
there still was some morphological differentiation within the Vpref category.
This question will be treated by the present author in a forthcoming study.

6. Clause combining in Arabic: Three examples

We shall adduce three examples from Arabic illustrating some of the intrica-
cies of clause combining, two examples from the Arabiyya and one from a
modern spoken form.
The first Arabiyya example is found in an episode in Ibn Isḥāq’s account
of the battle of Ḥunayn (Ibn Isḥāq 846, 8ff, cf. Nöldeke 1963, 68f):
wa-rasūlu llāhi yaqūlu ḥīna raʔā mā raʔā min-a n-nāsi: ‘ʔayna ʔayyuhā
n-nāsu’ wa-lam ʔara n-nāsa yalwūna ʕalā šayʔin fa-qāla yā ʕabbāsu
uṣruḫ ‘yā maʕšara l-ʔanṣāri yā maʕšara s-sumrati’ qāla fa-ʔaǧābū: ‘lab-
bayka labbayka’ qāla fa-yaḏhabu r-raǧulu li-yaṯniya baʕīra-hu fa-lā-
yaqdiru ʕalā ḏālika fa-yaʔḫuḏu dirʕa-hu fa-yaqḏifu-hā fi ʕunqi-hi fa-
yaʔḫuḏu sayfa-hu wa-tursa-hu wa-yaqtaḥimu ʕan baʕīri-hi wa-yuḫallī
sabīla-hu wa-yaʔummu ṣ-ṣawta ḥattā yantahiya ʔilā rasūli llāhi…
‘The apostle said when he saw the people [in confusion in the battle],
‘Where [are you going] men?’ And I did not see anyone paying heed to
anything, and he said ‘O Abbas, cry loudly: ‘O ʾansār, O comrades of the
acacia tree!’’ and they answered ‘Here we are, here we are!’; then the man
tried to turn his beast but could not do it; then he took his mail, threw it on
his neck, took his sword and shield, got off his mount and let it go its way
and make for the voice until he came to the apostle…’
The series of Vpref from fa-yaḏhabu definitely refers to a succession of

events which take place in the absolute past indicated by the Vsuff fa-ʔaǧābū.
The translation by Guillaume 1955 ad loc. is wrong, but reflects the fact that
the use of the Vpref is contrary to what is expected from the presentation in
most grammars and textbooks of Arabic. One may also notice the introductory
ḥāl-clause which stands in front of the clause with the governing verb.
The second Arabiyya example is taken from the ḥadīths about the ḥaǧǧ de-
scribing the ritual of stone-throwing at Minā (al-Bukhārī: Ṣaḥīḥ: K. al-ḥaǧǧ:
bāb 136, 137):
No. 1748 (ʕAbd ar-Raḥmān b. Yazīd about ʕAbdallāh b. ʕUmar:) ḥad-
daṯa-nā ʔanna-hu ntahā ʔilā l-ǧamrati l-kubrā ǧaʕala l-bayta ʕan yasāri-
hi wa-minan ʕan yamīni-hi wa-ramā bi-sabʕin wa-qāla hākaḏā ramā
llaḏī ʔunzilat ʕalay-hi sūratu l-baqarati

322 Jan Retsö
‘He [ʕAbd ar-Raḥmān] said: He [ʕAbdallāh] came to the big stela having
put the Temple on his left and Mina on his right; he thereby threw seven
[stones] and said/saying: This is how the one on whom the 2nd sura was
revealed threw’.
The Vsuff/perfects have different functions in this passage. The first indicates
the main event: Abdallah (after having performed some ceremonies) arrives at
the big stela. The asyndetic ǧaʕala refers to a situation resulting from his
preparatory movements. He has seen the stela and then takes a position, ensur-
ing that the Temple (= the Kaʿba?) is on his left before he starts the throwing
of stones. This clause is thus a traditional ḥāl-complement. The throwing is
introduced by wa- plus a verb in the perfect followed by another wa- plus a
perfect verb where Abdallah comments the throwing. The use of wa- and not
fa- indicates juxtaposition, i.e. that the throwing as well as the comment takes
place while continuously keeping the Temple to his left.
The ceremony is described in more detail in other ḥadīths:
no. 1750: fa-stabṭana l-wādiya ḥattā ʔiḏā ḥāḏā bi-š-šaǧarati ʕtaraḍa-hā
fa-ramā bi-sabʕi ḥaṣayātin yukabbiru maʕa kulli ḥaṣātin
‘Then he went down the middle of the valley until, when he had come near
the tree, he stood opposite to it. Then he threw seven pebbles saying
Allāhu ʔakbar with every pebble’.
no 1751: … ʔannahu [Abdallāh b. ʕUmar] kāna yarmī l-ǧamrata d-dunyā
bi-sabʕi haṣayātin yukabbiru ʕalā ʔiṯri kulli ḥaṣātin ṯumma yataqaddamu
ḥattā yushila fa-yaqūmu mustaqbila l-qiblati.
‘… that [Abdallah] pelted the closest stela with seven pebbles saying
Allāhu akbar after every pebble; then he went forward to get to a level
ground and stood facing the qibla’.
In 1750 we have two successive events told with fa- + Vsuff : ‘he went down’
and ‘he threw’ and then an asyndetic Vpref indicating what was taking place
while he threw. The Vpref receives its absolute tense value through the pre-
ceding Vsuff. In 1751 it can be observed how the same throwing is expressed
by a Vsuff of the ‘auxiliary’ kāna ‘it/he was’ + a Vpref. The meaning is hard-
ly iterative or habitual but an event in the past. The syntagm kāna yafʕalu
evidently has the same function as faʕala.

no.1752: ḥaddaṯanā… : ʔinna ʕAbdallāh b. ʕUmar … kāna yarmī l-

ǧamrata d-dunyā b-sabʕi ḥaṣayātin ṯumma yukabbiru ʕalā kulli ḥaṣātin
ṯumma yataqaddamu fa-yushilu fa-yaqūmu mustaqbila l-qiblati qiyāman
ṭawīlan…
‘Abdallah threw seven stones at the First stela, then saying Allāhu ʔakbar
on every stone; then he went forward and stood on a level ground facing
the qibla for a long while’.
In 1752 we notice the same Vpref indicating the accompanying event (yukab-
biru) but this time it is preceded by the particle ṯumma, ‘then’, ‘afterwards’.
The following verb is marked in the same way followed by a third one, like-
wise a Vpref, this time introduced by fa-. It is hard to see these Vpref as ḥāl-
imperfects. An alternative is to see them as habitual: every time Abdallah
performed the ceremony of ramy he used to… The problem with this is that
the ḥadīth is finished by the quotation like in no. 1748, an indication that even
this one refers to one specific occasion. We would rather have a series of
progressing events starting with kāna yarmī, i.e. all the Vpref receive their
tense value from kāna which locates the events reported as successive, i.e.
progress, in the absolute past.
The following version of the ceremony describes how the Prophet himself
performed it:
No 1753: kāna ʔiḏā ramā l-ǧamrata l-latī talī masǧda minan yarmī-hā bi-
sabʕi ḥaṣayātin yukabbiru kullamā ramā bi-ḥaṣātin ṯumma taqaddama
ʔamāma-hā fa-waqafa mustaqbilan-i l-qiblati rāfiʕan yaday-hi yadʕū wa-
kāna yuṭīlu l-wuqufa
ṯumma yaʔtī l-ǧamrata ṯ-ṯāniyata fa-yarmī-hā bi-sabʕi ḥaṣayātin yukab-
biru kullamā ramā bi-ḥaṣātin ṯumma yanḥadiru ḏāta l-yasāri mimmā yalī
l-wādiya fa-yaqifu mustaqbila l-qiblati rāfiʕan yadayhi yadʕū
ṯumma yaʔtī l-ǧamrata l-latī ʕinda l-ʕaqabati fa-yarmī-ha bi-sabʕi
ḥaṣayātin yukabbiru ʕinda kulli ḥaṣātin ṯumma yanṣarifu wa-lā yaqifu
ʕinda-hā.
‘It happened that, when he threw [stones] at the stela which is in the direc-
tion of the mosque at Mina, he threw seven stones at it saying Allāhu
ʔakbar every time he threw a stone. Afterwards he posed himself in front

324 Jan Retsö
of it and stood, turning towards the qibla with his hands raised, praying.
He made his standing long.’
‘Then he came to the second stela and threw seven stones at it saying
Allāhu ʔakbar every time he threw a stone. Then he went down on the left
side which is in the direction of the valley; he stood, turning towards the
qibla with his hands raised praying.’
‘Then he came to the stela which is at the gorge and threw seven stones at
it saying Allāhu ʔakbar with every stone. Then he departed, not standing at
it.’
It is close at hand to see this as series of habitual acts, an interpretation which
definitely is possible according to standard Arabic grammar. According to the
sources, however, the Prophet fulfilled the obligation of the ḥaǧǧ only once in
his lifetime, viz. during the ‘pilgrimage of farewell’ which, consequently, was
not only the last but also the first (Peters 1994, 247-248). Even if he might
have taken part in some of the rituals before the call (not reported in any
source), the report in the ḥadīth strongly suggest that the description of the
ḥaǧǧ al-wadāʕ is that of one particular performance.
A striking feature in these Arabiyya examples is the use of the (long) Vpref
in a narrative sequence. We notice that its progress function is usually some-
how marked morphosyntactically which makes it possible to single out the
Vpref used as off-line comments. But the employment of the same form in
sequential narrative and off-line comments is worth observing when we now
turn to the next example. The following are pieces from a story told in the
modern dialect of Central Arabia (Sowayan 1992, 86-92). This dialect belongs
to those which do not have the b-imperfect or its equivalents.
§ 006
wi-tḥafaḏ ̣ ar-rǧāl. ‘So men have memorized old stories.
taʕāṭa l-ʕlūm, They have handed them down to each
other,
ytanāgalōn-h ar-rǧāl min wāḥdin l- passing them on from one to the other,
wāḥid,
wa-l-wāld ywarrṯ-ah lu-wlid-uh. the father bequeathing them to his son’.
The two Vsuff (perfects) are followed by two Vpref which most likely should
be seen as complements of type 7) and 6) respectively, thus a traditional ḥāl-

complement where the Vpref verbs receive their temporal value from the two
preceding Vsuff.
§§ 008-009
raǧǧālin habb luh saʕad b-ayyām-uh. ‘A man who had luck in his lifetime (lit.
for whom luck had blown).
yāṭa an-nuḥūs, allah yarḥam-uh; He walked to the brink of danger, God
have mercy on his soul;
ʕagīdin yraćbin wi-yʕayyir w-yāṣal a bold leader who led mounted men on
al-ʕaduww raids, went raiding on a mount and
reached far-off enemies;
tatlīh ar-rǧāl yōm ad-dinya nāhb w- Men followed him in the days when
manhūb. there was the custom to plunder and be
plundered.’
The Vsuff habb is a general characteristic of the destiny of the hero: he has
had a lucky life. The following Vpref tell what he used to do. They indicate
habitual events during his lifetime, thus having a preterite temporal value.
§§ 039-043
gāl … w-yanhaǧūn miʕ-uh bass ‘He said [‘…’] and only seven left
sabʕih alli ṣmalaw miʕ-uh, together with him who held on to/had
joined him.
akṯar an-nās yadrōn ibin Gdūr wa-la Most people feared Ibn Gdur and did
yadrūn wiš tiṣīr tāl as-sālfih. not know how this affair would end.
w-yićīn ʕala aš-šararāt b-rāf, ḏị lʕin He [Hidlul] attacked ash-Shararat on
ysammūn-uh raf yaḫabrūn-uh ha-l- Raf, a mountain called Raf, these men
ǧimāʕat here in the audience know it.
w-yalga buh, ṭawwal allah ʕimrik, He found on it, may God give you long
sabʕ byūt šarārah life, seven tents of ash-Shararat
w-yāḫiḏ-hum kill ḥalāl-ihum and he plundered them, took all their
livestock.
yōm ḫaḏā-hum w-yiǧīk minćif. When he had plundered them he turned
back.
w-yōm innuh ǧa hal-uh wi-yʕazl When he came/had come back to his
arbaʕ ʕadad arbaʕ an-nāgih people, he set aside four camels- a
number equal to the number of a cam-
el’s legs
w-yaṣlig-hin ʕala al-hirbid. – and drove them straight to al-Hirbid.
gāl: ‘haḏō-lin ʕan nāgit-ak’ w-yiǧūn He said: ‘Take them in place of your
al-bāgyat wi-ytigāsimōn-ah. camel’. The rest of the camels they
divided among themselves.’

326 Jan Retsö
The Vsuff gāl gives the absolute time reference: anterior to speech-time.
There follows a series of Vpref coordinated with w- which likewise must have
the same time reference. Unlike the preceding cases, these Vpref refer to a
series of unique and consecutive events. In a conventional Arabiyya text one
would expect a series of Vsuff coordinated by fa-. Also, many dialects would
probably have preferred Vsuff here.
From these Arabic examples we can see that there is an alternative way of
reporting independent successive events in a narrative chain. Instead of the
usual linking of Vsuff with fa-, Vpref indicative can be used (cf. Nöldeke
1963, 68:i ff.; Nebes 1982, 199ff.). Such a series of Vpref is introduced by a
Vsuff, thus locating the whole sequence mostly anterior to speech-time. If we
now assume that the Vpref should be seen as a gram marking non-anteriority
or even simultaneity, its function in these sequences is that of marking simul-
taneity with a reference point anterior to speech time. It is, thus, not exactly a
praesens dramaticum but a ‘past present’. It is present, i.e. simultaneous, to a
reference point in the absolute past. The Arabic examples indicate that this is
a grammatical device that can be used not only for occasional off-line com-
ments but also when there is a chain of several successive events to be told.
One could say that the whole series of successive events are off-line in rela-
tion to the main verb.
The paradox is that the relationship between the Vpref used in this way and
the governing verb is exactly the same as in a traditional ḥāl-construction.
That construction is also a ‘past present’ if the governing verb is a Vsuff. The
Arabiyya has a device to differentiate between the two, viz. by introducing a
circumstantial clause with asyndeton or S-V word order. But we can see from
some examples quoted above that this is not compulsory: the verb may well
be linked only by a wa-, e.g. kataba bi-ḏaka ʔilā hišāmin wa-yastaʔḏinu-hu fī-
hi ‘He wrote about this to Hisham asking him for permission for it’.
In the Naǧdi dialect we do not see such a clear-cut distinction. Vpref ex-
pressing habitual, circumstance, progression and general absolute present are
not consistently morphosyntactically distinct. The criterion often used for
defining a ḥāl-complement clause, viz. gram switching, is not valid. The lis-
tener/reader is left to his own judgement whether there is a progression in a
mainline or an off-line comment (cf. M. Persson in this volume).
It thus seems possible to state that, in languages which only show innova-
tion II, the Vpref is a general default category that can appear in all kinds of
temporal contexts. It is obvious that it can express absolute anteriority, i.e.

traditional past tense, but that is not an exhaustive description. A possible

solution is to see it as a marker of relative simultaneity. It may then be simul-
taneous to speech time, which makes it equivalent to the traditional present
tense; from there it may develop into a future tense or modal indicator, and
finally, it may be simultaneous to a reference point in the absolute past. In the
Arabiyya, the syntagm Vsuff/pref + a Vpref/suff as a circumstantial comple-
ment is well known and has been thoroughly analysed (Nebes 1982; Premper
2002; Waltisberg 2009). It is usually understood as an off-line comment. But
even in the Arabiyya there are plenty of cases where the Vpref reports succes-
sive events in the absolute past indicated by a Vsuff verb in the context, thus
progress. The same is found in several modern Arabic dialects. It is explicitly
described for the Arabic in Chadic/Northern Nigeria (Roth 1979, 41-52; Ow-
ens 1993, 144) and Yemen (Watson 1993, 74-75) but traces of it are probably
widespread (cf. Feghali 1928, 71ff). The difference between circumstantial
complement and progress may be marked by specific syntactic devices as the
Arabiyya usually does, but this is not necessarily so. When the context is
clear, the language can skip the explicit marking. The main-line – off-line
contrast can be made explicit if desirable but can also be neglected (Persson in
this volume).
Another aspect of this is the status of the off-line commenting clauses.
They are usually seen as subordinate to the clauses of the main-line in a narra-
tive. In the Naǧdi example quoted above it can, however, be asked whether
this is a meaningful statement. The neat distinction between main-line and
off-line is blurred morphosyntactically since there is no consistent marking of
the difference. On many occasions the reader/listener is left to him/herself
(sometime possibly noticing intonation) to decode the textual flow. It is worth
mentioning in this context that the off-line comments in Akkadian, where
contemporaneity is distinctively marked by the iparras-form, are not marked
as subordinate clauses: the verbs lack the –u suffix which is the hallmark of
subordination in Akkadian (Streck 1995).
7. The omnipotent Vpref: The case of Ugaritic

The different forms of Arabic quoted above do not exhibit a consistent use of
Vpref as non-anterior. The use of Vpref in past narratives seems to be a stylis-
tic variant, perhaps to mark a rapid succession of events. Most forms of Ara-
bic have innovation I and II, and the use of Vpref in the Arabiyya in the func-

328 Jan Retsö
tion shown could be a diachronic fossil. Even Archaic Biblical Hebrew does
not show a consistent use of the verbal forms in this manner. The evidence in
both these languages is ‘disturbed’ by the differentiation process between long
and short Vpref in Biblical Hebrew and the grammaticalization of the opposi-
tion between Vsuff and Vpref in the Arabiyya.
Ugaritic is included in the NWCS language group showing both innova-
tions. Since the discovery of the Ugaritic texts it has been observed how fre-
quent the use of Vpref in narratives is (Hammershaimb 1941, 86ff.; cf. Smith
1994, 39). Most scholars have assumed that it is the short Vpref that is used as
narrative, even if the morphology is often not discernible in the vowelless
script. E. Greenstein put a definite question mark at the separate function of
long and short Vpref in Ugaritic (Greenstein 2006). According to him, the
morphological variation exists but has no semantic or syntactic function. It is
only a phonetic/allomorphic variation within the Vpref category (a similar
suggestion for the origin of long and short forms in BH was earlier suggested
by Revell in 1984). The arguments against this analysis by Hackett (2012) are
not decisive. They are based on one quite short text of a special kind of dis-
course and, even if one would agree with the analysis, it might be questioned
how representative this evidence is. Gzella (2010), who supports the idea of a
grammatical differentiation between long and short Vpref in Ugaritic, never-
theless points out that the evidence is not unambiguous. A more unconven-
tional solution would be that we see in Ugaritic the beginnings of a differenti-
ation in the use of the Vpref-forms. The Ugaritic Vpref would thus still to a
large extent have the same multifarious functions as we have seen in the Ara-
bic examples but at the same time shows the signs of a grammaticalization
process of an originally allomorphic variation.
This is a dominant feature in the literary Ugaritic texts and examples can
be found everywhere. Below follow a couple of specimens from the so-called
Baal cycle (KTU I.3.2-22; Smith 2009, 91-123):
ʕbd aliyn bʕl he served mightiest Baal
sid zbl bʕl arṣ waited on the prince, lord of the earth
qm yṯʕr wyšlḥmh he stood, arranged and offered him food
ybrd ṯd lpnwh sliced a breast for him
bḥrb mlḥt qṣ mri with a salted knife, a cut of fatling
ndd yšʕr wyšqynh he stood, served and offered him drink
ytn ks bdh put a cup in his hand

krpnm bklat ydh a goblet in both his hands

bk rb ʕẓm a large imposing vessel
ridn mt šmm a rhyton for mighty men
ks qdš ltphnh aṯt a holy cup women may not see
krpn ltʕn aṯtrt a goblet Athirat may not eye
alp kd yqḥ bḫmr a thousand jars he drew of the wine
rbt ymsk bmskh a myriad he mixed in his mixture
qm ybd wyšr he stood, chanted and sang
mṣltm bd nʕm cymbals in the virtuoso’s hands
yšr gzr tb ql sweet of voice the hero sang
ʕl bʕl bṣrrt ṣpn about Baal on the summit of Sapan
The three introductory Vsuff indicate parallel events which are not sequential.
Then follow three Vpref possibly describing a sequence of events. There are
two more Vsuff in the text with the same function as the three introductory
ones. The Vpref in this passage could well be interpreted as sequential and/or
focus but also as equivalents to the Arabic imperfects in a ḥāl-clause: ‘he
stood, arranging, offering’ etc. More difficult is it in the following example
(KTU 1.4.23-29; Smith 2009, 391-426):
hyn ʕly lmpḫm The skilled one ascended to the bellows
bd ḫss mṣbṭm tongs in the hands of Hasis
yṣq ksp yšlḥ ḫrṣ he cast silver, he poured gold
yṣq ksp lalpm he cast silver by the thousands
ḫrṣ yṣqm lrbbt gold he cast by myriads
yṣq ḫym wtbṭḫ he cast a canopied resting-place
This passage can be read as a sequence describing casting activities even if
some of them do not necessarily have to be seen as a sequence. But the most
natural understanding of the passage is that it describes how a deity ascends,
performs the casting of metals, and finally constructs a resting-place.
The use of Vpref as a narrative progress is thus well documented in Ugarit-
ic (Tropper 2000, 688ff.). There are also numerous cases of Vpref which we
can read as an off-line comment (KTU 1.3.II 17-26; Smith 2009, 127-194):
whln ʕnt lbth tmġyn And look! Anat arrives at her house
tštql ilt lhklh the goddess takes herself to her palace
wl šbʕt tmtḫṣh bʕmq but she was not satisfied with her fighting
in the valley

330 Jan Retsö
tḫtṣb bn qrtm battling between the two towns

tṯʕr ksat lmhr she arranged chairs for the soldiery
tʕr ṯlḥnt lṣbʔm arranged tables for the hosts
hdmm lġzrm footstools for the heroes
mid tmtḫṣn w tʕn She fought violently, looking around
tḫtṣb w tḥdy ʕnt she fought (and) Anat surveyed
tġdd kbdh bṣḥq her inwards swelled/swelling with laughter
ymlu lbh bšmḫt her heart filled with joy
kbd ʕnt tšyt Anat’s inwards with victory
The verbs tʕn and tḥdy in ll. 8-9 can be read as off-line comments (Tropper
2000, 692, 906; the iterative classification of ll.10-11 is doubtful) but also as
markers of progression. The presence or absence of the linking w is not neces-
sary for the off-line intrepretation. There is, thus, no specific marking of cir-
cumstantial complements in Ugaritic. The language possesses a Vpref form
which seems to have the value of non-anteriority. Anteriority seems to be the
value of the qatal-form, i.e. Vsuff., most frequently marking anteriority in
relation to speech time. One of the main functions of the qatal in a narrative is
to locate the series of events in the (absolute) past. The events are then told by
Vpref which, thus, indicate contemporaneity/simultaneity with a Reference
Time which is anterior to Speech Time.
The picture emerging is thus that of a continuum of steps between the
main-line ordering of events and off-line comments. There cannot be any
strict binary opposition between main-line and off-line. Progressive main-line
events can be expressed by Vpref indicating contemporaneity with a RT in the
absolute past. But also off-line comments are expressed by Vpref, this time
contemporaneous with the event time of the main-line. It seems that, in the
examples here, there is no explicit marking of the difference between these
functions of the Vpref. The hierarchy of events is given by the context and the
semantics of the verbs.
The picture given by these examples, especially the Ugaritic ones, shows a
remarkable similarity to Sabaean. The alternation between a Vsuff marking
anteriority, mostly absolute, i.e. past tense, and a Vpref which has a wide
range of functions is the same as we find in Sabaean. The system discernible
in some forms of Arabic, as well as in Archaic Biblical Hebrew, appears fully
exposed in these two languages. The function of the verbs in Ugaritic may
thus throw much light on the system in Sabaean.

8. Analysis of Sabaean texts

Equipped with these insights we can turn to the Sabaean evidence. Three texts
have been chosen from the three main periods of Sabaean for a preliminary
analysis of the possible occurrence of circumstantial clauses with a finite verb.
The first one is the report by the Sabaean ruler Karibʾil Watar about his deeds,
datable to ca. 680 BCE. The second is a description of a series of campaigns
undertaken in the 240s CE by the Sabaean kings Ilsharaḥ Yaḥḍub and Yaʾzil
Bayyin. The last one is the large inscription by Abraha, the Ethiopian ruler of
Himyar ca. 530-550 CE. These three texts are of a narrative character and
represent Old Sabaean, Middle Sabaean, and Late Sabaean respectively. The
main aim of this presentation is to study the interaction in a narrative text
between different verb forms in a narrative text.
RÉS 3945
This is the largest of the early Sabaean texts carved on a wall in the temple at
Ṣirwāḥ consisting of 19 long lines. It is a Tatenbericht by the first empire
builder in Arabia, Karibʾil Watar, and is by most scholars nowadays dated to
ca. 680 BCE. The interpretation is based on the latest translation of the text by
W. Müller (1983).
The text consists of a main clause as an introductory statement with a
Vsuff verb, followed by eleven paragraphs of varying length, describing the
deeds of the ruler. All paragraphs are introduced by the particle ywm which is
a temporal conjunction ‘when’, introducing temporal subordinate clauses,
analogous to the use of Arabic yawma, Biblical Hebrew yôm and Akkadian
inūma. The word means ‘day’ and stands in the construct state to the clause(s)
that follow. The conjunction ywm, in this text, is always followed by a Vsuff
verb. The paragraph as a whole mostly tells about activities with Vsuff verbs
always preceded by w-. The text contains 109 occurrences of Vsuff verbs
enumerating the deeds of the king. Against this stand 13 occurrences of Vpref
verbs. To this come a few cases of finite verbs in relative clauses or clauses
with specified particles such as kḏ. Some passages will be presented that illus-
trate the function of clausal complements. In every instance the subject is the
ruler himself although this is explicit only in the introductory clause.
The syntactic constellation allows for two interpretations. Firstly, all para-
graphs introduced by ywm with following Vsuff verbs are subordinate clauses
to the introductory statement. The whole text is a kind of background exposi-

332 Jan Retsö
tion to the introductory verb hfṭn which is the main event in the text. It follows
from this that the events told in the ywm paragraphs either have been accom-
plished before the hfṭn-event itself or are parts of it. This means that the whole
text should not be read as a report of events and achievements presented in
succession. It is rather a juxtaposition of achievements, somewhat reminiscent
of the so-called Display-texts by the Assyrian kings, thus: ‘These are the
things K. did when he… and when he… and when he…’.
A structure reminiscent of this one is found in the prologue to Hammura-
bi’s law:
inūma Anum ṣīrum šar annunaki When Anum the mighty, king of the
annunaki,
Ellil bēl šamê u erṣetim, šāʔim šīmat Ellil, the lord of heaven and earth, the
mātim determiner of the destiny of the land,
Ana Marduk, mārim rēštim ša Ea to Marduk, the first-born son of Ea

ellilūt kiššat nišī išīmū-šum determined the lordship of all mankind,
ina igigi ušarbiū-šu raised him among the igigi,

Bābilam šum-šu ṣīram ibbiū gave Babylon its illustrious name,
ina kibrātim ušāterū-šu made it preeminent on the earth,

ina libbi-šu šarrūtam dārītam in its midst everlasting kingship,
ša kīma šamê u erṣetim išdā-ša who established its [the kingship] foun-
šuršudā dations like heaven and earth,
inūmi-šu Ḫammurabi … iāti … šumī By then they called me Hammurabi, by

ibbiū my name.
If we see the last inūmi-šu clause as a main clause to which the preceding ones
are subordinated (which is the usual understanding of the text according to the
translations), one could also imagine that clause as an introduction to the
section: ‘The gods called me … when they … when they [had?]… when they
[had?]’. That would mean the structure of our text was in accordance with the
first alternative above.

Secondly, the introductory Vsuff verb is an independent main clause not

coordinated with the following. The following ywm-clause with its Vsuff verb
is subordinate to the following verbs which are in main clauses. The Vsuff
represents a narrative line: ‘These are the things K did. When he … then he …
and then he…’. This is the syntagm we find in the Naǧdi examples quoted
above. It is also identical with the Akkadian one with a temporal clause with
inūma (lit. ‘on (ina) the day when’ (ūma = accusative, construct state) fol-
lowed by one (or several) main clause(s): inūma innamru (verb in the sub-
junctive indicating subordination)… aqbī-kum (verb in the indicative indicat-
ing main clause) ‘When we saw each other … I said to you’ (von Soden 1969,
222). A similar clause structure is found in Biblical Hebrew, e.g.:
Wayhî byôm dibber YHWH ʔel mošē bʔereṣ miṣrayim wayḏabber YHWH
ʔel mošē leʾmor ʔanî YHWH … wayyoʾmer mošē lip̄ nê YHWH
‘When the Lord spoke to Moses in the land of Egypt, the Lord spoke to
Moses saying: ‘I am the Lord…’ then Moses said…’ (Ex 6:28-29).
Both BH and Akkadian usually (but not always) clearly mark the transition
from subordination to main clause level, BH by the use of the wayyiqtol, Ak-
kadian by the use of the subjunctive in the subordinate clause. Such a device
does not seem to exist in Sabaean which, thus, is very reminiscent of the
Naǧdi Arabic example above.
The problem of the interpretation of the Vpref. in our text remains. Are
they off-line comments or do they indicate progression in a main-line or se-
quentiality? The first alternative seems more likely considering the fact that,
with the exception of the first verb which occurs in one special instance in the
text, no explicit subject is found with the verbs. The reason for this is that
agent, i.e. the grammatical subject, is always the ruler himself. This is also the
interpretation by the translators. It is also difficult to argue that the series of
Vsuff must represent sequentiality (pace Stein 2013, 2:47). In some cases they
may, in others not, but there is no grammatical indication of a difference. The
Naǧdi Arabic example shows that such an ambiguity exists.
Four ywm-sections, apart from the introductory statement, are chosen;
three dealing with ‘civilian’ activities, and one report on military activities
against the city of Nashan, present-day as-Sawdāʾ, in the area of Maʿīn.

334 Jan Retsö
/1/ ʔlt hfṭn krbʔl wtr bn ḏmrʕly These are the things that Karibʾil Watar,
mkrb s1bʔ b-mlk-hw l-ʔlmqh w-l- son of Dhamarʿaliy, MKRB of Saba as-
s1bʔ signed/dedicated/presented to Ilmuqah
and Saba when he had become king
The verb that summarizes Karibʾil’s achievements in his introductory state-

ment is ‘give’, ‘assign’ etc. The noun mlk is uncertain: ‘reign’ or ‘become
king’ or ‘acquire something’.
ywm hwṣt kl gwm ḏ-ʔlm w-s2ymm when he established the whole community
of gods and protecting deities
w-ḏ-ḥblm w-ḫmrm and of covenant and favour,
w-ḏbḥ ʕṯtr s2lṯm ʔḏbḥm and slaughtered to A ʿ thtar three sacrifices,
w-nyk ḥwbs1 and NYK (celebrated a hieros gamos
with?) Ḥawbas,
w-ʔlm ʕṯtr and made a banquet for Athtar,
w-ynr b-trhm lighting a fire in TRHM,
w-whb ʕṯtr w-ḥwbs1 ḫlfm and gave robes (?) to ʿAthtar and Ḥawbas.
The first ywm-paragraph is introduced by the so-called covenant formula

referring to some kind of ritual renewal of the Sabaean community. The se-
quence contains five Vsuff verbs connected by w-. The Vsuff verbs do not
necessarily express progression. The translators of the text, (Rhodokanakis in
RÉS, W. Müller 1983), see the following Vsuff as a continuation of the ywm-
clause, thus following alternative 1) above. The main event would thus be
represented by the verb hwṣt which means some religious ceremony, and the
following Vsuff rather function as focussing, i.e. giving details of the contents
of the ceremony. Since we do not know exactly the procedures of this ritual, it
is premature to pinpoint a definite function of the Vpref so far but it is reason-
able to assume that the Vpref has a function different from that of the Vsuff.
Between the fourth and the fifth Vsuff in the sequence we find a Vpref ynr
which is seen as derived from NWR ‘fire’ etc. It can be interpreted in two
ways: either ‘and thereby lighting a fire’ (background off-line comment) or
‘and then lighted a fire’ (progression).
w-ywm hʕḏb mʕs2rt s1bʔ ‘and when he reorganized the assembly of
Saba
w-yʔtmmw in that they obeyed him

w-yḥtẓyw mns2ʔ-hmw k-ʔḥd b-ʕs1y and their military actions/posse

ṣdqm were successful together in performing
right
/2/ w-yns2ʔ ʔs1m l-mtʕ qny-hw and everybody rising to make its property
thrive’.
The next ywm-section, referring to general actions by the rulers, is introduced

by a Vsuff (hʕḏb) and is here followed by three Vpref all coordinated by w-. It
is possible to read them as factual consecutives ‘so that they obeyed him’, ‘so
that they were successful’. It can also be understood as focused, seeing the
activities designed by the Vpref as specifications of the Vsuff. This is also
how the translators render it. The problem is once again the verbal semantics,
especially of the introductory Vsuff. hʕḏb.
Then follows a section dealing with irrigation projects undertaken by the
ruler. Such actions, such as the building and restoration of temples by the
Mesopotamian kings, were important also for ideological reasons.
w-ywm ṣdq ʕṯtr w-ʔlmqh ḥg-hmy and when he correctly had performed the
ordinance of both ʿAthtar and Ilmuqah,
w-yhṯb mwy ḏhb-hw rymn making the water of his irrigated land
RYMN turn back
w-ykn fnwtm fnwtm w-ḏyrm ḏyrm so that a canal became a canal and a
flooded field a flooded field (cf. Beeston
1962, 61).
w-ḥmy ḥrt-hw ʕhl l-mʔwdn bn kḏ and surrounded his water conduit ʕHL to
tḍʔn brḥm l-ḏhbnhn w-l ʔrʕn the border with a wall so that it would not
flow out freely to the two irrigated fields
and to ʔRʕN
w-hṯb zm ḥmy mwtrm and led back the water of the wall-
ḏ-s1twdn bn hwdym surrounded MWTRM which was irrigated
from HWDYM
w-ʕs1y ḥṣṣm w-ḍʕrtm ḏhb mydʕm And acquired ḤṢṢM and ḌʕRTM, the
kl ms1qy zm wtr w-/3+/wqh irrigated field of ?, all the amount of
water from WTR and WQH
w-b-f krbʔl wdy w-tʔw zm wtr w- And by the order of Karib ʾil the water of
wqh WTR and WQH flowed and streamed

336 Jan Retsö
The two Vpref yhṯb and ykn are ambiguous: factual consecutive or focus? All
depends on the semantic content of the Vsuff ṣdq which is not clear-cut. As
far as the Vsuff are concerned, it seems difficult to argue that they represent a
continuous sequence of actions, thus describing one large irrigation project.
The impression is rather that of several projects. On the other hand we must
admit that we do not get a clear picture of the activities described due to our
ignorance of the geography and the terminology employed.
The following two ywm-sections deal with Karibʾil Watar’s military opera-
tions in the Ǧawf area north of Marib. The second one is the longest one in the
inscription.
w-ywm mḥḍ ns2n w-wfṭ ʔhgr-hw w- and when he defeated Nashan and burnt
gbḏ ʕs2r w-byḥn w-kl ʔḏhb-hw b-ʔḥd his city and plundered ʕS2R and BYḤN
mns2ʔm and all his irrigated fields with one
posse
w-ywm ns2ʔ ṯnym mns2ʔm w-ygnʔ And when he mobilized a second posse
gnʔm ḏ-bhw s3wkt ns2n w-ns2qm b-s2ft in order to raise a wall with which
ʕṯtr s2lṯt ḫrfm III Nashan and Nashqum were surrounded
on the order of ʿAthtar for three years
The Vpref ygnʔ seems here to indicate imagined consecutive, i.e. ‘in order to
build a wall’ (pace Gruntfest 1965). The object (gnʔm) to this verb has an
attributive relative clause with a Vsuff verb (s3wkt).
w-hbʕl ns2qm w-bʕḍ-h l-ʔlmqh w-l- And seized Nashqum and its surround-
s1bʔ ings for Ilmuqah and for Saba
w-hrg ns2n ʔlfʔ w-hrs3ḥ s1mhyfʕ w- and killed thousand from Nashan and
ns2n swept away Sumayfaʿ and Nashan
w-hṯb ʔbḍ/15+/ʕ whb-hw mlk s1bʔ l- and returned the areas which the king of
ʔlmqh w-l s1bʔ Saba had given him, to Ilmuqah and
Saba
w-s1tmḥḍ ʔhgr-hw qwm w-gwʕl And appropriated his cities QWM,

w-dwrm w-fḏm w-s2bm w-ʔhgr ʔykm GWʕL, DWRM, FḎM and Shibam and
kl ḏ-qny s1mhyfʕ w-ns2n b-ʔykm the cities of ʔYKM, all that Sumayfaʿ
and Nashan had acquired in ʔYKM.

w-s1tmḥḍ b-bḍʕ-hw l-mhyʕ ʔwṯnn ʕd And appropriated in his area the stretch
wṯn mnhytm l-ʔlmqh w-l s1bʔ of the borderstones until the border of
MNHYTM for Ilmuqah and for Saba
w-s1tmḥḍ ʕḏb ẓlm w-ʕḏb ḥmrt And appropriated the cultivated fields
of the black ones and the cultivated
fields of the red ones
w-fqḥ ṯʕd mlk ns2n w-ns2n bn mwy And opened the irrigation area of the
mḏb king of Nashan and of Nashan from the
waters of [the river] Madhab
w-nḍw gnʔ hgr-hw ns2n And took away the wall of his city
Nashan
ʕd h/16-/s2rs1-hw until he uncovered its fundaments
w-hgrn ns2n yhḥrm bn mwfṭm but protecting the city of Nashan from
burning (cf. Beeston 1984, 19)
w-ʕtb-hw ḫrs2 byt-hw ʕfrw w-ḫrs2hgr- and destined for him the destruction of
hw ns2n his palace ʕFRW and the destruction of
his city Nashan
w-bḍʕ b-ẓhr ns2n s3lʔm ʔfklt And imposed a tribute on Nashan for
w-ʕtb bn ns2n ʔl wḍʔt s2ft-hmw ns1rn priests, and destined those from Nashan
ʔlʔltn w-yhrgw about whom order had proceeded from
the gods to be killed (cf. Beeston 1962,
61; Gruntfest 1965 ad loc.)
The Vpref yhḥrm looks very much like a CCC of the kind we find in Arabic:
‘he broke down the wall … forbidding the burning of the city’. This is the
only case in this text where we find an asyndetic attachment of a Vpref. One
should also observe the SV word-order. The second one, w-yhrgw, is a conse-
quence of the ordering, ʕtb. Some of the Nashanites (bn ns2n) were to be
killed according to divine commandment. The use of the Vpref indicates that
it is an intended i.e. imagined consequence.

338 Jan Retsö
w-ʕtb s1mhyfʕ w-ns2n kḏ yḥwr And destined for Sumayfaʿ and Nashan
1 2 1
s bʔ b-hgrn ns n w-kḏ ybny s mhyfʕ that Sabaeans should settle in Nashan
w-ns2n byt ʔlmqh b-wsṭ hgrn ns2n and that Sumayfaʿ and Nashan should
build a temple for Ilmuqah in the centre
of the city of Nashan
w-s1tmḥḍ mwy ḏ-qfʕn bn /17+/ And appropriated the water of QFʕN

smhyfʕ w-ns2n w-ys1tmḥḍ yḏmrmlk from Sumayfaʿ and Nashan so that
mlk hrmm YḎRMLK king of Haram should ap-
propriate it
w-s1tmḥḍ bn s1mhyfʕ w-ns2n ḥrrtn ḏt And appropriated from Sumayfaʿ and

mlkwqh w-yhmḥḍ nbṭʕly mlk kmnhw Nashan the water conduits of MLK-
w-kmnhw bn ḥrrtn ḏt mlkwqh ln ʔwṯn WQH in order to hand it over to
wṯn krbʔl NBṬʕLY king of Kaminahu and Kami-
nahu, from the water conduit ḎT
MLKWQH to the border stones that
Karibʾil had established
This formula occurs three times in the text: the Vsuff s1tmḫḍ followed by w-
Vpref yhmḫḍ. The interpretation could be that of an intended consequence, i.e.
‘in order to give it’ or a commenting focus: ‘thereby giving it’.
w-gnʔ ns2qm w-hbkl-hw And surrounded Nashqum with a wall
s1bʔ l-ʔlmqh w-l s1bʔ and assigned it to Sabaeans as a settle-
ment for Ilmuqah and Saba
The language employed in RÉS 3945 obviously uses the Vsuff both in se-
quential narrative and non-sequential narrative. It does not seem that the lan-
guage makes any grammatical distinction. On the other hand, it is possible to
see the Vprefs functioning as comments of different kinds. One should beware
of having a too narrow definition of off-line, circumstantial etc. The Arabiyya
evidence teaches us to see the gram-switch, evident in this inscription as a
grammatical signal, marking a distinct syntagm with several different mean-
ings.

Ja 576:1-16+577:1-19
The two texts Ja 576 and Ja 577 are both found on the outer wall of the Aw-
wām temple in Marib. It is most likely that they belong together, constituting
the longest text we have from the reign of the two kings of Saba Ilsharaḥ
Yaḥḍub and Yaʾzil Bayyin, both reigning in the 240s CE. Together it encom-
passes 35 long lines and is, thus, one of the longest texts preserved from pre-
Islamic Arabia.
Like many of the dedicatory texts from the period, this one is divided into
distinct paragraphs. The introduction follows the conventional usage in these
inscriptions when the two kings state how they have dedicated a number of
statues to the god of the Awwām temple in Marib, which is followed by nine
sections introduced by the expression ḥmdm b-ḏt ‘as thanks for’ (576:1, 3,
577:6 (bis), 8, 15, 16, 18 (bis)) thus giving the reason for the dedicatory act.
All the ḥmdm-sections are introduced by a Vsuff verb: hws2ʕ (sections I, II,
IV, V, VI, VIII) or ḫmr (sections III, VII, XI), both verbs meaning the same:
‘grant, bestow a favour’. Section II is the longest, encompassing 19 lines of 35
preserved altogether (= 576:3-16 + 577:1-6), and is a more or less continuous
narrative about the military campaigns of the kings. It is subdivided in smaller
paragraphs by the particles bn-hw or bʕd-hw ‘after that’ or ‘from there’. There
are 15 such sections in 576 and 5 in the following section in 577. Mostly,
however, the narrative is conducted by finite verbs coordinated by w- or f-.
The subordinate paragraphing with the particles bn-hw or bʕd-hw is also
found in some of the other ḥmdm b-ḏt units. The long narrative sections in Ja
576-577 are, according to Gruntfest, a rare specimen of historical writing in
Sabaean, being extracts or imitations of otherwise lost royal chronicles. This
observation is interesting and should be kept in mind when trying to analyse
the syntax.
We shall present an analysis of the first part of the inscription, section I,
i.e. the general introduction, and the first part of section II dealing with a
conflict between the king of Saba, Ilsharaḥ Yaḥḍub, and mainly the king of
Himyar. The text in section I has the conventional introduction reporting that
the two kings have dedicated (Vsuff: hqnyy) statues to the god of Marib as
thanks for (ḥmdm b-ḏt section I) the help that Ilmuqah has given (Vsuff +
[progression] infinitive: hws2ʕ w-hrdʔn) to the kings against all who have
risen (Vsuff: tns2ʔw) against them, people from the north, the south, the sea
and the dry land. This should be seen as a general summary of what is told in
details in the following section. The verbs are Vsuff with one infinitive.

340 Jan Retsö
There follows a subsection of I introduced by w-l-ḏt ‘and also because’.

Here it is told that Ilmuqah has helped (hws2ʕ) to capture MLK king of Kinda
and his ally MRʔ LQS king of ḪṢṢTN in the town of MRB; they gave (Vsuff:
whbw) their sons as pledges; they also gave (whbw) booty. The three main
verbs, as well as those in subordinate clauses, are Vsuff. This episode takes
place somewhere northeast of Marib.
Then follows Section II with a detailed report of the campaigns in the high-
lands west of Marib against Himyar and its allies. The section starts by report-
ing how the god of Saba, Ilmuqah, has helped the Sabaean king to defeat
Ḥabashat, Sahratum and Shammar of dhu Raydan, i.e. Himyar, three enemies
controlling the western mountains and the vast plains both south and north of
the Yaslaḥ pass, who had had some alliance with the king of Saba but had
broken the oath.
Ilsharaḥ and Saba go up from Marib to Ṣanʿā and attack Himyar, Radman
and MḌḤY. They take Shamatan’s land and the city of DLL, the house of
Yahir and the city of Azwar in the land of Qasham. Qasham is the area just
north of the Yaslaḥ pass south of Ṣanʿā, where several encounters with the
Himyarites and their allies take place in the following.
After that Ilsharaḥ is attacked from Qasham; his army goes to the town of
Baʾsan where Shammar’s troops have gone. These troops are defeated. They
then proceed to the plain of DRGʕN. The Raydanites avoid them. They go to
the land of MHʔNF just southwest of the Yaslaḥ pass, taking booty; they go
up the pass of YLRN taking the town of TʕRMN. Then they return to Naʾiḍ, a
town situated on the eastern part of the plain north of the Yaslaḥ pass, which
has been made a base for the Sabaeans. Ilsharaḥ now advances to eastern
Qasham taking the town of Ayḍam, then returning to Naʿiḍ.
Ilsharaḥ then advances back to MHʔNF taking the two towns of ʕṮY in the
western part of their territory. From there they go towards the town of ḌFW
encountering the tribes of MHʔNF and MḎRḤ. They do not take the town.
From there they return to the area around YKLʔ. They meet troops of Ray-
dan/Himyar. In the battle MRḤḌN is crushed but saved by the prince of
YKLʔ. They return to Naʿiḍ. The Himyarites pursue them to wadi NGRRM.
A truce is made with YKLʔ. The Sabaeans return to Naʿiḍ and then to Ṣanʿā.

Section II:1 (= ll. 3-4)

w-ḥmdm b-ḏt hws2ʕ ʔlmqh ʕbd-hw And as thanks because Ilmuqah has
ʔls2rḥ yḥḍb b-ḫrʔn w-s2kr w-nqm ʔḥzb assisted his servant Ilsharaḥ Yaḥḍub in
ḥbs2t w-ḏ-s1hrtm w-s2mr ḏ-rydn w- opposing, defeating and taking venge-
ʔs2ʕb ḏ-ḥmyrm b-ḥbl ḥblw bʕd s1lm w- ance on the units of Habashat,
gzm gzmw S1HRTM and Shammar of Raydan and
the tribes of Himyar because of the
revolt they made after the peace and
oath they had sworn
w-ys3mkw bn hgrn mrb ʕdy hgrn ṣnʕw And they ascended from the town of
lḍbʕ w-hṣrn bʕly s2mr ḏrydn w-ʔs2ʕb Marib to the town of Ṣanʿā in order to
ḥmyrm w-rdmn w-mḍḥym fight and to fall upon Shammar of
Raydan, the tribes of Himyar, Radman
and MḌḤYM
w-yʕdwn mlkn ʔls2rḥ w-ḏ-bn ʔqwl-hw And the king Ilsharaḥ and some of his
w-ḫms1-hw w-ʔfrs1-hw ʕdy ʔrḍ ḥmyrm governors, his infantry and his cavalry
went over to the land of Himyar
w-ṯbrw w-hbʕln w-qmʕ w-hs1bʕn byt And they crushed, seized, subjugated
ḏ-s2mtn w-hgrn dll w-byt yhr w-hgrn and plundered the house of S2MTN and
ʔẓwr b-wṯnn b-ʔrḍ qs2mm b-ywm hġrw the city of DLL, the house of YHR and
the city of Azwar at the boundary in the
land of QS2MM on the day they made
the raid
w-ylfyw bn hnt hgrn mhrgtm w- and brought back from these towns
s1bym w-ġnmm ḏ-ʕs1m spoils, captives and booty in great
numbers
The main narrative of section II is introduced by the phrase ḥmdm b-ḏt hws2ʕ
‘as thanks because [the god Ilmuqah] has shown favour [to his servant the
king]’ by fighting and frightening and taking revenge on the Ethiopians, the
S1HRTM and Shammar of Himyar who obviously had had an alliance (ḥbl)
with the kings of Saba which they had broken (gzmw). Then follows w-
yms3kw ‘and they ascend, go up’ (w + Vpref), viz. from Marib to the high-
lands around Ṣanʿā. This is continued by w-yʕdwn ‘and he marches into’ (w +
Vpref) viz. the land of Himyar. The events there are then described with the
verb ṯbrw ‘they destroyed (w + Vsuff) followed by three infinitives: w-hbʕln

342 Jan Retsö
w-qmʕ w-hs1bʕn ‘seize’, ‘subjugate’, ‘take revenge’. There follows an enu-

meration of cities and regions taken by the Sabaeans which is summarized by
a concluding phrase: b-ywm hġrw w-ylfyw ‘on the day/when they raided
(Vsuff) and LFY (w +Vpref) ‘got’, ‘obtained’ booty from all these cities.
The three Vpref verbs clearly refer to events in the past. It should, howev-
er, be noticed that they both refer to the preliminaries of the action: ascending
into the highlands and marching into the land of Himyar. Then follows the
action: destruction (ṮBR). The use of infinitives after a finite verb to indicate
subsequent activities is a well known fact in Sabaean (Höfner 1943, 63-66;
Kogan and Korotaev 1997, 236). If one has made up one’s mind already, it is
possible to read the sequence as a series marking progression: He granted
them favour – then they ascended – then he marched in – then they destroyed
– first they raided – then they took booty. But this is a very insensitive way of
reading a text. A more natural reading would be to see the ‘granting of favour’
as a summary of the whole enterprise followed by some details of the events,
illustrating/commenting/characterizing the overreaching act of divine favour.
The ascending and marching are preliminaries. The main event is constituted
by the military operations in the land of the enemy which is described by a
series of four verbs indicating a sequence of events. In this case it is possible
to see the two preceding Vpref as not belonging to the main narrative line but
as background. One should also notice Gruntfest’s suggestion that the section
is an excerpt from a literary text, a royal chronicle or the like. This remains a
hypothesis although it might explain several features in this text which is
extraordinary in many ways. One might envisage a preceding clause, now
replaced by the thanksgiving summary, which stated the preparations for the
war and the leaving of the capital told by Vsuff verbs.
The final notice of raiding and taking booty is structured as a sequence
Vsuff + w + Vpref. It can be read ‘they raided and then took booty’ or ‘they
raided so that they took booty’ but also ‘they raided taking booty’. What is
important is that all readings are possible. The last suggestion gives the syn-
tagm more possibilities.

Section II:2 (= l. 4)
w-bn-hw f-hwṣlw ʕdy byn hgrnhn And from there they assembled in [the
area] between the two cities

w-bn-hw f-ybḥḍn mlkn ʔls2rḥ yḥḍb w- And from there Ilsharah Yaḥḍub and
b-ʕm-hw ḏ-bn ʔqwl-hw w-ḫms1-hw w- with him some of his governors and his
ʔfrs1-hw infantry and his cavalry abruptly at-
tacked
w-ybḥḍw ʕdy ḫlf hgrn bʔs1n and they attacked to the town of BʔS1N
1 1 2
w-qdm-hmw ʔs d hys r s mr ḏ-rydn And the soldiers who Shammar of Ray-
bn ʔs2ʕb ḥmyrm l-hʕnn ʕdy wṯnn dan from the tribe of Himyar had dis-
patched in order to protect the boundary
came before them
w-ytqdmw b-ʕm-hmw b-ḫlf hyt hgrn and they attacked them in the region of
bʔs1n the city of BʔS1N
1 1
w-ḫmr-hmw ʔlmqh hs ḥtn hmt ʔs dn And Ilmuqah showed them favour in
ʔḥmrn destroying these Himyarite soldiers
w-yhrgw bn-hmw mhrgm ḏ-ʕs§m w- and they killed a great number of them
tmlyw s1by w-qny hyt hgrn bʔs1n and they took the captives and the
slaves of the city of BʔS1N as booty
Then follow two paragraphs, the first introduced by the syntagm bn-hw f- +
Vsuff, the second by bn-hw f- + Vpref. It has been shown that this syntagm,
i.e. an element in a clause (subject, object, prepositional phrase) followed by
f- + finite verb, is a means of marking the category of progress in several
Sabaean texts (Nebes 1995, 211-213), viz. when it stands as an introduction to
a paragraph. In a past narrative context it is, then, to be expected that the verb
following f- is a Vsuff which also dominates in the material. Another syntagm
is when the f- is followed by an l + Vpref/suff. But in Nebes’ corpus there are
18 cases where we find a Vpref after f-which, according to him, functions as a
narrative past tense, thus in the same way as f- + Vsuff (Nebes 1995, 44-46).
Of these, 13 occurrences are found in Ja 576+Ja 577. This text is datable to ca.
240 AD when the kings Ilsharaḥ Yaḥḍub and Yaʾzil Bayyin are known to
have reigned. Two more texts, with one occurrence in each, are dated to the
same period (Ir 32 and Ry 533), ca. 200 AD. This use of Vpref as a narrative
past tense with the f-prefix is, thus, a phenomenon specific for a small group

344 Jan Retsö
of texts, written during one century under reigns of kings stemming from the
highlands of Yemen (Hamdān, Gurat, Ḥimyar). Nebes’ remarks that this use
of Vpref is reminiscent of the consecutive imperfects in Biblical Hebrew and
that there is evidence that the f- represents some kind of influence from the
north (Nebes 1995, 269 n. 65, 66).
One could, however, suggest another reading. The two Vpref forms of
BḤḌ, ‘to invade’, ‘make an incursion’ are not reporting a sequence. The latter
could be seen as a focus of the former: ‘he invades with his army … indeed,
he makes an incursion even to the border’. The incursion is also a preliminary
that leads up to the confrontation w-qdm-hmw ‘then the soldiers sent by
Shammar of Himyar confronted them’. We would have a presentation similar
to the one in II:1: two Vpref verbs as a preliminary to the main event told by
Vsuff.
There follows a Vpref w-ytqdmw ‘they attack’, followed by a Vsuff w-
ḫmr-hmw, reporting the assistance of the deity. This is followed by another
Vpref: w-yhrgw reporting the consequence of the divine assistance: ‘so that
they were able to kill’. It can be claimed that the Vpref verbs in this sequence
function as a kind of subordinate complement which can well be translated by
past tense but at the same time as subordinate clauses: ‘when the king made
an incursion, an incursion even to the border, [then] he was confronted by
Shammar; when they attacked them they were helped by the deity so that they
killed (could kill) many of them’. Then follows the last event: w-tmlyw ‘then
they took booty’. This would give a main line represented by Vsuff with
comments off-line by Vpref. It should be observed that the comments off-line,
from an extralinguistic viewpoint, are not necessarily background comments.
They could very well be integrated into the main line sequence and told by a
series of Vsuff connected by w-. The paradox is that main line events can be
told as off-line comments just as, in western European languages, main-line
sequences can be told by temporal subordinate clauses. Narrative syntax is not
necessarily iconic.

Section II:4 (ll. 5-7)

w-bnh-hw f-hṣrw ʕdy ḏt mẓʔw ʕdy And from there they set out until they
brrn ḏ-drgʕn arrived at the plain of DRGʕN
w-tflw mṣr ḏ-rydn w-ʔl qdm-hmw And the troops of Raydan went away and
did not precede them
Section II:5
w-b ʕd-hw f-yṣbʔw b-ʕly ʔrḍ And from there they proceeded towards
mhʔnfm w-yhys1rw bḥḍm ḏ-bn the land of MHʔNFM and they dis-
ḫms1-hmw b-ʕly ʔrḍt mhʔfm patched in an attack some of the infantry
against the land of MHʔNFM
1
w-lfyw b-hw mhrgtm w-s by w- And they got from there spoils and cap-
ġnmm ḏ-hrḍw-hmw tives and booty which satified them
Section II:4 reports a series of three sequential events by Vsuff verbs: ‘they
set out’, ‘they arrived’, ‘the [others] went way’. The use of Vsuff for a com-
ment: ‘they did not confront them’ is perhaps connected with the negation.
Otherwise, a Vpref would be the expected form.
Section II:5 has a structure similar to e.g. II:3: two Vpref: ‘they proceed-
ed’, ‘they sent in troops’ followed by a Vsuff: ‘they obtained, got [booty]’. An
idiomatic translation would be: ‘Afterwards, when they proceeded’ or: having
proceeded and sent in troops, [then] they took booty’.

w-bn-hw f-s3mkw mqln ḏ-ylrn w- And from there they went up through the
nḥbw hgrn tʕrmn pass of YLRN and harassed the city of
TʕRMN
w-ḫmr-hmw ʔlmqh hbʕln hyt hgrn And Ilmuqah showed them favour in
tʕrmn w-ylfyw bhw mhrgtm w- taking this city of TʕRMN and they got
ys1byw kl ʔwld w-ʔnṯ-hw w-ymtlyw from there spoils from the dead and took
kl ʔbʕl-hw all its children and women captive and
took as booty all the married men
This section almost looks like a Classical Arabic text. The first two Vsuff
report two successive events followed by a third stating that the operation was
a successful thanks to divine assistance. Then come three Vpref verbs describ-
ing the outcome of the operation, viz. the capture of the enemy and taking of

346 Jan Retsö
booty. The Vpref do not, of course, represent progression but comments to the
main series of events.

w-bn-hw f-ytʔwlnn ʕdy hgrn nʕḍ w- And from there they returned to the city
rtʕ b-ʕm-hw ḏ-bn ḫms1-hmw of Naʿiḍ and some of his infantry was
installed by it
2 1
w-yhṣr mlkn ʔls rḥ yḥḍb w-ʔs d qrbw And the king Ilsharaḥ Yaḥḍub and
b-s3n-hw bn ḫms1-hmw w-ʔfrs1-hmw soldiers who had approached him from
his army and cavalry set out
2 2
w-yhṣrw b-ʕly ms rqt ʔrḍ qs mm And they fell upon the eastern part of
the land of QS2MM
1 3
w-yhs bʕw w-hs ln hgrn ʔyḍmm And they plundered and emptied the
city of ʔYḌMM
2 2
w-ybḥḍw kl ms rqt qs mm And they raided all eastern part of
QS2MM
1
w-ylfyw b-hw mhrgtm w-s bym ḏ- And they got spoils and captives in
ʕs1m great numbers
Section II:8
w-bn-hw f-ygbʔw ʕdy hgrn nʕḍ And from there they came to the city of
Naʿiḍ

w-bn-hw f-yhṣrn mlkn ʔls2rḥ yḥḍb w- And from there the king Ilsharaḥ
ḏ-bn ḫms1-hw w-ʔfrs1-hw ʕdy ʔrḍ Yaḥḍub and some of his infantry and
mhʔnfm his cavalry set out for the land of
MHʔNFM
w-yqmʕw w-hbʕln hgrnhn ʕṯy w-ʕṯy And they subjugated and seized the two
cities of ʕṮY and ʕṮY
w-ylfyw b-hw mhgrtm w-s1bym w- And they got from there spoils, cap-
mltm w-ġnmm ḏ-ʕs1m tives, riches and booty in great numbers

w-bn-hw f-ytʔwlw b-ʕly hgrn ḍfw And from there they returned to the city
of ḌFW
w-ykbnn b-hw ḏ-mḏrḥm w-s2ʕbn And they met there Ḏ-MḎRḤM and the
mhʔnfm tribe of MHʔNFM
2
w-yhbrrw s ʕbn mhʔnfm b-ʕly mqdmt- And the tribe of MHʔNFM came into
hmw w-hs1ḥt-hmw mqdmt-hmw the open against their vanguard and
their vanguard defeated them
ʕdy ḏt ḥml-hmw hgrn ḍfw until the city of ḌFW let them enter
w-yhrgw bn-hmw mhrgm ḏ-ʕs1m And they killed a great many of them
In section II:7 we have here the same structure as in II:3 and II:5. The first
verb is a Vpref ytʔwlnn ‘to return’ followed by a Vsuff w-rtʕ ‘to post’, ‘sta-
tion’ (troops etc.). The Sabaeans, thus having returned to the town of Naʿiḍ,
stationed troops there and made it their base. Then follows the report of two
major military expeditions; one against the land of Qasham (section II:7, 8)
told by five Vpref verbs, then one against the tribe of MHʔNF (II:9, 10). This
is told by seven Vpref. The paragraphs II:7-10, thus, contain only one Vsuff
which, on the other hand, presents perhaps the major event: the establishment
of a military base from which military expeditions are undertaken. We should
notice that the progress, i.e. the sequentiality, is marked by the w-bn-hw f-
construction, not by the verbal grams. The paragraphs are thus ordered in a
temporal sequence, but not necessarily the events told within each.
w-bn-hw f-tʔwlw ʕdy ḫlf hgrn yklʔ And from there they returned to the
region of the city of YKLʔ
w-ykbnn b-hw ḏ-bn ʔqwl ḏ-rydn w- And they encountered there some of the
mṣr ḥmyrm governors of Raydan and the force of
Himyar
w-hbrrw w-tqdmn b-ʕm-hmw w-hs1ḥt- And they came out into the open and
hmw bn mrḥdn attacked them and routed those from
MRḤDN
ʕdy ḏ-ḥml-hmw s3ʔd yklʔ until the ruler of YKLʔ let them enter
1
w-lfyw bn-hmw mhrgtm ḏ-ʕs m And they got from them spoils in great
numbers

348 Jan Retsö
Only one Vpref appears in this paragraph, possibly indicating a circumstantial

event: ‘and from there when they returned to the area of the town Y, thereby
encountering (ykbnn) the Himyarites, they came out into the open and ad-
vanced towards them and routed them until the s3ʔd of the town of YKLʔ
admitted them, [alternatively took them by storm] and then they took booty
from them’.
Section II:12 (l. 9)

w-bn-hw f-tʔwlw ʕdy hgrn nʕḍ And from there they returned to the city
of Naʿiḍ
w-yʔdb-hmw hmt ʔḥmrn k-l-yqdmnn l- And those Himyarites challenged them
mhrgtm ʕdy s1r ngrrm to attack for war trophy in the valley of
NGRRM
w-yhṣrn mlkn ʔls2rḥ yḥḍb w-ḏ-bn And the king Ilsharaḥ Yaḥḍub and
ʔqwl-hw w-ḫms1-hw w-ʔfrs1-hw ʕdy ḏt some of his governors and his infantry
mẓʔw s3d yklʔ and his cavalry set out until the ruler of
YKLʔ arrived
w-ʔl hbrrw hmw ʔḥmrn l-mhrgtm And the Himyarites did not come out
into the open to fight
w-gbʔw w-tʔwlw ʕdy hgrn nʕḍ And they went and returned to the city
of Naʿiḍ
w-bn hgrn nʕḍ f-ytʔwlw ʕdy hgrn And from the city of Naʿiḍ they re-
ṣnʔw b-hwbltm w-mhrgtm w-ʔḫyḏtm turned to the city of Ṣanʿā with animals
w-s1bym w-ġnmm d-ʕs1m as booty, spoils from those killed, pris-
oners, captives and booty in great num-
bers
A new attack by the Himyarites is reported after the Sabaeans have returned to
their base in Naʿiḍ. The challenging (yʔdb) of the Himyarites, and the reaction
(yhṣrn) by Ilsharaḥ Yaḥḍub, are reported by Vpref verbs. The following
events, the arrival of the Sabaeans (mẓʔw), the non-appearance of the Himyar-
ites on the battlefield (ʔl hbrrw), the leaving and return of the Sabaeans to
their base (gbʔw, tʔwlw), are told by Vsuff, the meaning obviously being that
these events should be seen as successive. The whole section about the first
war between Saba and Himyar is concluded with a bn X f- construction report-
ing the final return of the Sabaeans to Sanʿā. This is made by a Vpref
(ytʔwlw), admittedly somewhat difficult to explain. An English rendering of

the passage could mark the verbal forms: ‘they arrived … they (the Himyar-
ites) did not go out … they (the Sabaeans) went/retired and returned to Naʿiḍ
and from that town, then returning to Sanʿā with booty’.
CIH 541
This text is a report by Abraha, the Ethiopian general who took part in the
conquest of Himyar in 525 CE and after some years made himself king of
Himyar. The text is carved on a stele that originally stood in Marib, probably
near the church that, according to the text, was constructed there after the
Ethiopian conquest. The text is written in four columns on the four sides of
the stele which is approximately 2 meters high. It is the longest text preserved
from pre-Islamic Arabia, 136 lines altogether. We choose the introductory
passages dealing with a revolt against the Abraha and the breaking of the dam
in Marib.
ll. 9-17
w-s1ṭrw ḏn ms3ndn k-qs1d ‘They [the Ethiopians] wrote this in-
scription when
w-hḫlf b-gzmn yzd bn Yazid the son of
kbs2t ḫlft-hmw Kabshat, their governor revolted and
violated the oath,
ḏ-s1tḫlfw ʕly kdt w-dʔ kn l-hw ḫlftn he whom they had appointed as gover-
nor over Kinda when they had no gov-
ernor
1 1 1
w-qs d w-ʕm-hw ʔqwl s bʔ ʔs ḥrn mrt He revolted, and together with him the
rulers of Saba, Saḥar, Murrat
ṯmmt w-ḥns2m w-mrṯdm w-ḥnfm Thumamat, Ḥanash, Marthad,
Ḥanif
ḏ-ḫll w-ʔzʔnn ʔqwln of Khalil, the Ya ʾzanites rulers
mʕdkrb bn s1myfʕ w-hʕn Ma ʿdikarib son
of Sumayfaʿ, Haʿan
w-ʔḫwt-hw bny ʔs1lm and his brothers, the sons of Aslam.’
The succession of events reported in the paragraph is that first the Ethiopians
have installed (s1tḫlfw) a ḫlft, a khalīfa, over the tribe of Kinda. Their chief
Yazid has given an oath (hḫlf b-gzmn) to him but later he has revolted (qs1d)
together with other chieftains, some from the old Sabaean aristocracy. The

350 Jan Retsö
writing of the text (s1ṭrw) is then the last event to be told. All events are told
by using Vsuff verbs but not in an iconic sequence. Instead, the events are
reported in reverse order: writing, revolt, oath, viceroy. This is made possible
by the use of the explicit marking of different kinds of subordinate clauses,
most important the particle k which functions as a temporal conjunction.
Worth noting is the clause w-dʔ kn l-hw ḫlftn ‘they did not have a governor’.
This must refer to conditions before the appointment of Yazid, thus an insert-
ed clause giving the background to the appointment. The verb kn (= kāna)
thus refers to a situation anterior to the preceeding s1tḫlfw: they had not had a
khalifa before the appointment of Yazid (cf. Piotrovsky 1998 ad. loc.). This is,
thus, a typical anterior ḥāl-clause in accordance with classical Arabic gram-
mar.
ll. 17-24
w-k-ʔs1y ‘and when he [the king] ordered
grh ḏ-zbrn yʔfqn bqh [m]lkn Garah from Zabran to control, by royal
decree,
b-ms2rqn w-hrg-hw w-s1ḥt[w] the east[ern province] they killed him
and destroyed
mṣnʕt kdr w-yzd gmʕ ḏ-hṭʕ-hw the castle of KDR; then Yazid gathered
[those] who obeyed him
bn kdt w-ḥrb ḥḍrmwt from Kinda and made war against Ha-
dramawt
w-ʔḫḏ mznm hgn ʔḏmryn w-ʕ[...]d and he captured MZN, a freedman from
ʕbrn Dhamar and returned to
Abran.’
Eight more or less successive events are told by Vsuff verbs. One has the
impression that the sequence w-hrg-hw w-s1ḥt[w] ‘they killed and destroyed’
in an older text could have been told by Vpref verbs as a kind of progres-
sion/comment on the preceding. We should observe the asyndetic Vpref yʔfqn,
‘in order to control’, an example of the ḥāl muqaddar of Classical Arabic
grammar. In an older text the introductory k would not occur. Instead one
would expect ywm or w-bn-hw f-.

ll. 24-32
w-wṣḥ-hmw ṣrḫnw s2t[ʔ]w ‘and a cry came to them; they arose
w-gmʕw ʔgyz-hmw ḥbs2t and gathered their armies of Ethiopi-
ans
[w]ḥmyrm b-ʔʔlfm b-wrḫ ḏ-qy[ṣ]n and Himyarites in thousands in the
month of Dhu Qayṣan
ḏ-l-s1bʕt w-ḫms1y w-s1ṯ mʔtm in the fifty-seventh and six hundredth
year
w-s2tʔw w-wrdw and they arose and descended
mqly s1bʔ w-s2ʔmw the passes of Saba and went north-
wards
bn ṣrwḥ ʕly from Ṣirwāḥ towards
nbṭm ʕdy ʕbrn NBṬM to Abran.’
Noticeable here is the use of the verb s2tʔ ‘to rise up’ but which looks like an
auxiliary, such as Arabic ʔaḫaḏa, ǧaʕala. Once again the successive events
are told by Vsuff coordinated by w-.
ll. 32-41
w-k-wṣḥw nbṭm ‘and when they arrived at NBṬM
ḏkyw s1rwt-hmw they sent their army
kdr ʔly to KDR: the ʔLY,
w-lmd w-ḥmyrm the LMD and the Himyarites
w-ḫlyf-hmw wṭh and their vice-commander WṬH
w-ʕwd-h ḏy gdmn and ʕWDH from GDM;
w-wṣḥ-hmw yzd b- and Yazid came to them in
nbṭm w-hʕd-hmw yd-hw NBṬM and gave them his hand
qdmy ḏkyn before the sending of the army’.
We notice the difference between the introduction to this paragraph and the
preceding one. In the latter we have a main clause: ‘and it came to them’.
Here it is clearly subordinated by the particle k: ‘when they arrived…’ Other-
wise the events pass review by Vsuff verbs.

352 Jan Retsö
ll. 41-55
w-k-wṣḥ-hmw ‘And then (k) an appeal came to them
(= Abraha and his men)
ṣrḫm bn s1bʔ for help from Saba
k-ṯbr ʕrmn w-ʕwdn that (k) the dam and its surrouding wall
had been destroyed
w-ḫbs2m w-mḍrft ḏ-ʔfn and ḪBS3M and the side walls of the
distributor
b-wrḫ ḏ-mḏr[ʔ]n in the month of MḎRʔN
w-bʕdn ḏ-l-s1bʕt in the seventh year.
wṣḥ-hmw ḏn ʕhdn After (bʕdn) this report had come to
them
hqdmw b-rdnn they appointed assistants (?)
brṯ ydnn ʕrbn in order to (brṯ) subdue the ʕRB
ʔlht dʔ gbʔw ʕm yzd who had not returned with YZD
w-k-kl-hmw hʕdw ʔyd-hmw And when (k) all of them had assured
their loyalty
w-rhn-hmw b-br[…] and given hostages …
w-s1rwytn ḏ-hḏ[kyw] the army that they sent out
kdr qrnw ʔqwln ʔlht qs1dw to KDR garrisoned the chieftains/rulers
that had revolted’.
Of the ten finite verbs reporting the events, two are Vsuff in main clauses
(marked in the transcription). All the other ones are Vsuff (with one excep-
tion) in subordinate clauses, clearly marked by conjunctions k, bʕdn, brṯ. The
Vpref verb is undoubtedly in a final subordinate clause: hqdmw … brṯ ydnn
‘they sent them out in order to subdue’. We may observe that all events taking
place in the absolute past are Vsuff. Vpref indicating absolute past are not
found here.
There are several more passages similar to this one in the text which, thus,
shows a kind of discourse structuring much more explicit than in the earlier
texts.
Explicit marking of CCC

Brockelmann did not mention the particle ʔənzä in Geez or the kad + partici-
ple construction in Syriac in connection with the Umstandssätze. In these two
languages, these particles obviously function more or less as indicators of

clausal complements equivalent to the traditional ḥāl in Arabic with or with-

our wa-. Geez has a few examples that appear to be CC-complements which
are factual or modal consecutives (Brockelmann 1913, 501ff.):
fännäwomu yəḥoru ‘he sent them out so that they went’

mäṣaʔku ʔəmut məslekä ‘I have come to die with you’
ʔəllä ʕärgu yəsgädu läbäʕal ‘those who had come up to celebrate the
festival’
The multifarious use of the particle k- in Sabaean has been noticed by the
descriptions of the language (Höfner 1942, 167-168; Beeston 1962, 62-63; id.
1984, 50-51; Bauer 1966, 105; Kogan and Korotaev 1997, 239-240; Stein
2003, 207-209; Nebes and Stein 2004, 474-477). According to Nebes and
Stein (2004), it may introduce a verbless circumstantial clause (CIH 541:66-
67):
w-qds1w bʕt mrb k-bhw qs1s1m
‘they held a mass in the church of Marib as there was a priest in it’.
The question remains whether it may introduce a verbal CCC. If so (see Bauer
1966, 105), Sabaean would have a construction similar to the Geez ənzä +
finite verb or the Syriac kaḏ + participle. Let us consider the following exam-
ples:
Ja 618:6 (cf. Ja 627:5, 10, Ja 643:27, Ja 647:16; Ja 735:11)

… bḏt hwfy-hmw ʔlmqh b-tbs2rt tbs2rw b-ʕm-hw k-ymlʔn mnḫt-hmw ḏ-yfd
‘…because Ilmuqah has given them the oracle they asked for so that they
may fill up their basin Ḏ-YFD’ (= consecutive).
Ja 577:9
ṯhb-hmw k-yṣwynn ʔmrʔ-hmw ʔmlk s1bʔ k-hʕs1mw hḫṭʔn w-wʕd-hmw k-
ys2ryn-hmw mlk ḥḍrmwt b-ʕbr ʔmrʔ-hmw ʔmlk s1bʔ…
‘he answered them that he would inform his lords the kings of Saba that
they have continued to commit offences, and promised them that the king
of Ḥaḍramawt would protect them against the kings of Saba’ (= object
clause).

354 Jan Retsö
YM 440 = CCIAS 39.11/06 no. 4, cf. Ja 752:10

w-ḥmdw mqm ʔlmqh k-ḫl-hmw bn qblm ḏ-wdq ʕl-hmw
‘and they praised the power of Ilmuqah because he [had] delivered them
from a calamity which befell them’ (= causal clause).
CIH 313:4-5 (reconstructed)

w-tʔlb ry[m f-wqh-hmw b-ms1ʔl-hw lhmw] w-l-ʔmrʔ-hmw bny ʔʕyn k-
ys1ʕdn-hmw nʕm[tm w-wfym]
‘and Taʾlab of Riyam determined for them and their lords, the Aʾyanites in
order to give them prosperity and safety’ (= final clause).
Ja 643:18-19
w-mlk ḥḍrmwt w-mṣr-hw f-bm-hwt ywmn s1bʔw ʕdy ḫlf hgrn yṯl k-s1fhw
bʕw-hmw krbʔl byn w-mṣr-hw bn hgrn mryb
‘and the king of Ḥaḍramawt and his troops in this very period fought in the
region of the city of Yathull foolishly acting treacherously against Karibʾil
Bayyin and his troops’ (= ḥāl-clause).
The classification of these clauses may be discussed, but the main point is that
the particle k- seems to be used as a general marker of dependent clauses,
regardless of semantics. The texts quoted are from different periods but it
seems that the explicit marking with k- is a feature which is rare in the Old
Sabaean texts, with increasing frequency in Middle and Late Sabaean. The
trend towards explicit marking of different kinds of clausal complements is a
feature apparent in the late inscription CIH 541 analysed above. This language
might thus have been on the same line of development as Geez and Syriac
before it died out.
9. Concluding remarks
The aim of this study has not been to present a complete exposition of CCC
and/or its equivalent based on a corpus extracted from the entire Sabaean
epigraphic material. It has been judged necessary, instead, to pinpoint the
problems before such a project can be undertaken. Since the grammatical and
semantic structure in this language are not yet completely understood (it is not
even certain which morphosyntactic features, if any, should be looked for in

order to collect a corpus), a typological Semitistic perspective based on the

quite spectacular progress made in other areas of Semitic studies concerning
the verbal system and its function during the last two decades should first be
established.
One result from the study of the sample texts plus the general Semitic per-
spective given here, especially of the position of the Sabaean verbal system
within a Semitic context, is that a strict distinction between main-line and off-
line markings is not visible. Also, a strict division between coordination and
subordination should be avoided. The assumption of a strict binary system
blurs the linguistic realities. It is obvious that in many Semitic languages,
ancient and modern, there are many different clausal relations in a coherent
text. It is possible to see how verbal clauses often have different degrees of
dependency on each other. The category labelled juxtaposition above can be
said to represent the highest degree of independency among verbal clauses.
Such an agglomeration of verbal clauses is held together by being enacted on
the same stage, or participating in the same situation, more or less simultane-
ously. When progress is introduced, we have a higher degree of dependency
which can be coded in different ways. A chain of successive events can be
syntactically linked in several ways. We have seen that in many Semitic lan-
guages we find ample use of Vpref verbs in progress-contexts which indicate
a kind of dependency. The same holds for the difference in the Arabiyya be-
tween fa- + Vsuff and wa- + Vsuff. And we have seen that the Arabiyya also
can use fa- + Vpref in a progress function. Arabic dialects can even use wa- +
Vpref in this function, a usage which is more or less normal in Ugaritic liter-
ary language and of which we have ample traces in Archaic Biblical Hebrew
and Sabaean. This usage abolishes the formal distinction between progression
or main-line presentation and circumstantial off-line comments.
The absence of the equivalent of the Arabic wa-huwa yafʕalu construction
in Sabaean pointed out by Nebes does not imply the total absence of clauses
with a similar semantic content and function even if there are few examples of
asyndetic Vpref which appear to be similar to Arabic CCC. Nebes’ claim, that
there are no direct analogies to the Arabic ḥāl-clauses with finite verbs since
the Vpref in a Sabaean subordinate clause never has the temporal value of
simultaneity to a main clause, can definitely be questioned. The claim is
founded upon an understanding of the verbal system of Sabaean which is not
the only one possible. The supporting argument, that no such clauses exist in
Biblical Hebrew, is not tenable.

356 Jan Retsö
All this tells us that it is premature to make too categorical statements

about the clause syntax of Epigraphic South Arabian based on classical Arabic
grammar. Beyond and before the regulated, rule-governed morphosyntax of
that language lies another world, now visible in the documents of the ancient
languages of the Middle East and South Arabia. It seems that we need a new
model of the Semitic verbal system as it appears in some of the earliest docu-
mented languages, including Sabaean. The ḥāl-marking, syndetic and asyndet-
ic, in the Arabiyya stands out as a survival, of an earlier stage when the rela-
tive tense value of the Vpref allowed it to be used as a general finite form both
in narrative sequences and backgrounds as well as modal. The whole system
of clause linking has been systematized in the Arabiyya although, as we have
seen, there is ample documentation of an earlier stage.
The tendency everywhere has been to differentiate the, as it seems, archaic
usage of the Vpref. In Ugaritic we observe the employment of the particle p in
narrative progress sequences which in the Arabiyya has become the general
indicator of progress, just as the wayyiqtol form in BH is a secondary differen-
tiation of the general Vpref. with a similar function. In Sabaean we see the
beginning of a marking of progress by the use of the infinitive instead of a
finite verb which, however, never developed into a full-fledged finite catego-
ry. Just as Arabic, Sabaean tended to develop a special marking of back-
ground clauses by the use of particle k. A close analogy is Geez which in
general uses ʔənzä. This is a development which deserves a separate investi-
gation.
A general, final reflection is that it is unlikely that the use of different ver-
bal grams does not represent semantic distinction. The varying use of Vsuff
and Vpref in the languages discussed here should be understood as represent-
ing different meanings. But at the same time one should be wary of overinter-
pretation. If one assumes that the verbal forms discussed here function in a
relative tense system, one should try to see it as such and perhaps nothing
more. It becomes likely that Vpref, for example, does not mark durative, pro-
gressive, cursive, imperfective, and so forth. In a tense system we should
reckon with temporal distinctions, nothing more. Differences between punctu-
al, semelfactive, stative, durative etc. are of course relevant, but are perhaps
not marked by the Vpref-Vsuff grams. These categories may be located any-
where on a temporal axis. The main factors signalled by the verbal grams are
time location and degrees of dependency.

The linguistic study of the epigraphic South Arabian languages is, in spite
of several groundbreaking works, still in its beginnings. This is true especially
for the syntax of these languages, while, at the same time, the volume of doc-
uments is increasing continuously. This study is a modest contribution to-
wards a broader understanding of these languages.
Summary
The existence, as was assumed by A. F. L. Beeston, of clauses in Sabaean
similar to the traditional ḥāl-construction in Arabic, defined as the kāna
yafʿalu syntagm, was cast into doubt by Y. Gruntfest and N. Nebes. Accord-
ing to the latter the Vpref in Sabaean indicate simultaneity, i.e. absolute pre-
sent tense, only in main clauses. In subordinate clauses it always indicates
future tense. The Vpref can also mark progression, i.e. consecutivity in a nar-
rative chain. Consequently, the kāna yafʿalu syntagm, that is Vpref indicating
contemporaneity in the past tense, does not exist.
Nebes’ analysis turns out to be too narrow and influenced by a view on the
verbal system in Biblical Hebrew, a view which many scholars do not follow.
A closer look into the Arabiyya shows that the ḥāl-construction is only one
subgroup in a set of clause-combining devices which serve as comments,
expansions and off-line comments to a main verbal clause.
A comparison between the verbal systems in Ugaritic, the Arabiyya, and
several modern Arabic dialects shows a much more variegated uses of the
Vpref than usually assumed. The Vpref appears as off-line comments to a
main-line verbal clause, as actual or general present tense in a main clause,
and even coding the main line of a narrative in the absolute past tense. This
multifarious use of the Vpref appears also in Sabaean although in different
degrees during the 1200 years of documentation of the language. In the late
Sabaean texts it looks as if we get a more explicit system of coding subordina-
tion by using specific particles (‘conjunctions’). But the many different uses
of the Vpref in Sabaean, with good parallels in other Semitic languages, an-
cient and modern, should make us cautious in assuming a too limited defini-
tion of its function as well as about the definition of subordinate commenting
clauses. The handling of clause combining in Sabaean turns out to represent a
fairly archaic stage, paralleled by Ugaritic and Archaic Biblical Hebrew, but
also by some modern Arabic dialects, especially on the Arabian Peninsula.
The strict regulation of the Vpref in off-line subordinate clauses according to

358 Jan Retsö
the Arabiyya grammar turns out to be a late stage in the development of the
CCC.
References
Andersen, F. I. 1974. The Sentence in Biblical Hebrew. The Hague/Paris.
Bartelmus, R. 1982. HYH. Bedeutung und Funktion eines hebräischen ”Allerweltwor-
tes” – zugleich ein Beitrag zur Frage des hebräischen Tempussystems. St. Ottilien.
Bauer, G. M. 1966. Jazyk južnoaravijskoj pis’mennosti. Moskva.
Beeston, A. F. L. 1962. A Descriptive Grammar of Epigraphic South Arabian. Lon-
don.
———. 1984. Sabaic Grammar. Manchester.
Brockelmann, C. 1913. Grundriss der vergleichenden Grammatik der semitischen
Sprachen II. Berlin.
al-Buḫārī. 1423/2002. Ṣaḥīḥ al-Buḫārī. Ṭabʕa ǧadīda maḍbūṭa wa-muṣaḥḥaḥa wa-
mufahrasa. Dār Ibn Kaṯīr: Dimašq - Bayrūt.
CIAS = Corpus des inscriptions et antiquités sud-arabes t. I section 1. Louvain, 1977.
CIH = Corpus inscriptionum semiticarum pars IV. Inscriptiones ḥimyariticas et sa-
baeas continens t. II. Parisiis, 1911.
Cook, J. 2012. Time and the Biblical Hebrew Verb. The Expression of Tense, Aspect,
and Modality in Biblical Hebrew. Winona Lake.
Davidson, A. B. 1901. Introductory Hebrew Grammar: Hebrew Syntax. 3rd ed. Edin-
burgh.
Denz, A. 1971. Die Verbalsyntax des neuarabischen Dialektes von Kwayriš (Irak) mit
einer einleitenden allgemeinen Tempus- und Aspektlehre. Wiesbaden.
Dixon, R. M. W. 2009. “The Semantics of Clause Linking in Typological Perspec-
tive.” The Semantics of Clause Linking. A Cross-linguistic Typology, edited by R.
M. W. Dixon and A. Aikhenvald, 1-55. Oxford.
Driver, S. R. 1892. A Treatise on the Uses of Tenses in Hebrew and some other Syn-
tactical Questions. 3rd ed.. Oxford.
Eskhult, M. 2011. “Thoughts on Phrases and Clauses Expressing Circumstance in
Biblical Hebrew Narration.” En pāsē grammatikē kai sophiā. Saggi di linguistica
ebraica in onore di Alviero Niccacci, ofm, edited by G. Geiger. Jerusalem.
———. Forthcoming. “A Discussion on the Date of Job from a Syntactic Perspective”.
Feghali, M. 1928. Syntaxe des parlers arabes actuels du Liban. Paris.
Fischer, W. 1987. Grammatik des klassischen Arabisch. 2nd ed. Wiesbaden.

Gesenius, W. 1910. Gesenius’ Hebrew Grammar, edited and enlarged by E. Kautzsch.

2nd English ed. by A. E. Cowley. Oxford.
Greenstein, E. L. 2006. “Forms and Functions of the Finite Verb in Ugaritic Narrative
Verse.” Biblical Hebrew in Its Northwest Semitic Setting. Typological and Histori-
cal Perspectives, edited by S. Fassberg and A. Hurwitz, 75-102. Jerusalem/Winona
Lakepp.
Gross, W. 1976. Verbform und Funktion: wayyiqtol für die Gegenwart? Ein Beitrag
zur Syntax poetischer althebräischer Texte. St. Ottilien.
Gruntfest, Ja. B. 1965. “Konsekutivnyje konstrukcii v južnoaravijskom jazyke.” Krat-
kije Soobščenija Instituta Narodov Azii 86: Istorija i filologija Bližnego Vostoka,
Semitologija. 129-147. Moskva.
Guillaume, A. 1955. The Life of Muhammad. A Translation of Isḥāq’s Sīrat rasūl
Allāh. Oxford.
Gzella, H. 2010. Review of: A manual of Ugaritic, by P. Bordreuil, and D. Pardee
(Winona Lake 2009). Bibliotheca Orientalis 67, 3-4: 367-371.
Hackett, J. A. 2012. “Yaqtul and a Ugaritic incantation text.” In Language and Nature:
Papers Presented to John Huehnergard on the Occasion of His 60th Birthday, ed-
ited by R. Hasselbach and N. Pat-El, 111-118. Studies in Ancient Civilization 67.
Chicago.
Hammershaimb, E. 1941. Das Verbum im Dialekt von Ras Schamra. Eine morphologi-
sche und syntaktische Untersuchung des Verbums in den alphabetischen Keil-
schrifttexten aus dem alten Ugarit. Kopenhagen.
Höfner, M. 1943. Altsüdarabische Grammatik. Leipzig.
Ibn Isḥāq = Das Leben Muhammed’s nach Muhammed Ibn Ishâk, edited by F. Wüsten-
feld I-II. Göttingen, 1858-60.
Isaksson, B. 2009. “An Outline of Comparative Arabic and Hebrew Text Linguistics.”
In Circumstantial Qualifiers in Semitic: The case of Arabic and Hebrew, edited by
B. Isaksson, 36-150. Abhandlungen für die Kunde des Morgenlandes 70. Wiesba-
den.
Ja = Jamme, A. Sabaean Inscriptions from Maḥram Bilqîs (Mârib). Baltimore 1962.
Joosten, J. 2012. The Verbal System of Biblical Hebrew. A New Synthesis Elaborated
on the Basis of Classical Prose. Jerusalem.
Joüon, P. and T. Muraoka. 1996. A Grammar of Biblical Hebrew. Roma.
Kogan, L. E. and A. V. Korotayev. 1997. “Sayhadic (Epigraphic South Arabian).” In
The Semitic Languages, edited by R. Hetzron, 220-241. London/New York.
Kouwenberg, N. J. C. 2010. The Akkadian Verb and Its Semitic Background. Winona
Lake.

360 Jan Retsö
Kuhr, E. 1929. Die Ausdrucksmittel der konjunktionslosen Hypotaxe in der ältesten

hebräischen Prosa. Ein Beitrag zur historischen Syntax des Hebräischen. Leipzig.
Meyer, R. 1972. Hebräische Grammatik III: Satzlehre. Berlin/New York.
Müller, W. W. 1983. “Altsüdarabische und frühnordarabische Inschriften.” In Texte
aus der Umwelt des Alten Testaments Bd I Rechts- und Wirtschaftsurkunden. His-
torisch-chronologische Texte. 268-282 Gütersloh.
Nebes, Norbert. 1982. Funktionsanalyse von kāna yafʿalu. Ein Beitrag zur Verbalsyn-
tax des Althocharabischen mit besonderer Berücksichtigung der Tempus- und As-
pektproblematik. Studien zur Sprachwissenschaft 1. Hildesheim.
———. 1988. “The Infinitive in Sabaean and Qatabanian Inscriptions.” Proceedings
of the Seminar of Arabian Studies 21: 63-78.
———. 1990. “Gibt es im Sabäischen ‘Zustandssätze’ analog dem arabischen Schema
wa-huwa yaf’alu und wa-huwa fī l-bayti?” In XXIV Deutscher Orientalistentag in
Köln. Ausgewählte Vorträge, edited by W. Diem and A. Falaturi, 61-69. Stuttgart.
———. 1994. “Verwendung und Funktion der Präfixkonjugation im Sabäischen.” In
Arabia Felix. Beiträge zur Sprache und Kultur des vorislamischen Arabien. Fest-
schrift für Walter W. Müller zum 60. Geburtstag, edited by N. Nebes, 191-211.
Wiesbaden.
———. 1994. “Zur Form der Imperfektbasis des unvermehrten Grundstammes im
Altsüdarabischen.” In Festschrift Ewald Wagner zum 65. Geburtstag, edited by W.
Heinrichs and G. Schoeler, vol. 1, 59-81. Beirut.
———. 1995. Die Konstruktion /fa-/ im Altsüdarabischen. Syntaktische und epigra-
phische Untersuchungen. Wiesbaden.
Nebes, N. and P. Stein. 2004. “Ancient South Arabian.” In The Cambridge Encyclo-
paedia of the World's Ancient Languages, edited by R. D. Woodard, 457-487. Ber-
lin/Boston.
Nöldeke, Th. 1963. Zur Grammatik des classischen Arabisch. Wien 1897. Reprint,
Darmstadt.
Notarius, T. 2011. “Temporality and Atemporality in the Language of Biblical Poetry.”
Journal of Semitic Studies 56:2, 275-305.
———. 2013. The Verb in Archaic Biblical Poetry. A Discursive, Typological, and
Historical Investigation of the Tense System. Leiden/Boston.
Owens, J. 1993. A Grammar of Nigerian Arabic. Wiesbaden.
Peters, F. E. 1994. Muhammad and the Origins of Islam. New York.
Piotrovsky, M. 1997. “Les causes d’une disparition.” In Yémen. Au pays de la reine de
Saba’. 219. Paris: Institut du monde arabe.

Premper, W. 2002. Die “Zustandssätze” des Arabischen in typologischer Perspektive.

Frankfurt/M.
Reckendorf, H. 1895. Die syntaktischen Verhältnisse des Arabischen. Leiden.
———. 1921. Arabische Syntax. Heidelberg.
RÉS = Répertoire d’épigraphie sémitique t. VI. Paris, 1935.
Retsö, J. 2014. “The b-imperfect revisited.” In Proceedings of the Oslo-Austin Work-
shop in Semitic Linguistics, Oslo, May 23 and 24, 2013, edited by L. Edzard and J.
Huehnergard, 64-72. Wiesbaden.
Revell, E. J. 1984. “Stress and the waw ‘consecutive’ in Biblical Hebrew.” Journal of
the American Oriental Society 104: 437-444.
Roth, A. 1979. Esquisse grammaticale du parler arabe d’Abbéché (Tchad). Paris.
Rundgren, F. 1959. Intensiv und Aspektkorrelation. Studien zur äthiopischen und
akkadischen Verbalstammbildung. Uppsala /Wiesbaden.
Smith, M. 1994. The Ugaritic Baal Cycle Vol. I: Introduction with Text, Translation
and Commentary of KTU 1.1-1.2. Leiden/New York/Köln.
Smith, M. and W. Pitard. 2009. The Ugaritic Baal Cycle Vol. II: Introduction with
Text, Translation and Commentary of KTU /CAT 1.3-1.4. Leiden/Boston.
von Soden, W. 1969. Grundriss der akkadischen Grammatik. Roma.
Sowayan, S. A. 1992. The Arabian Oral Historical Narrative. An Ethnographic and
Linguistic Analysis. Wiesbaden.
Stein, P. 2003. Untersuchungen zur Phonologie und Morphologie des Sabäischen.
Epigraphische Forschungen auf der Arabischen Halbinsel 3. Rahden/Westf.
———. 2007. “Materialien zur sabäischen Dialektologie: Das problem des amiriti-
schen (“haramischen”) Dialektes.” Zeitschrift der Deutschen Morgenländischen
Gesellschaft 157:13-47.
———. 2011. “Ancient South Arabian.” In The Semitic Languages. An International
Handbook, edited by S. Weninger et al., 1042-1073. Handbücher zur Sprach- und
Kommunikationswissenschaft 36. Berlin/Boston.
———. 2013. Lehrbuch der sabäischen Sprache 1. Teil: Grammatik. Subsidia et
Instrumenta Linguarum Orientis 4,1. Wiesbaden.
Streck, M. P. 1995. “ittašab ibakki ‘weinend setzte er sich’: iparras für die Vergan-
genheit in der akkadischen Epik.” Orientalia N.S. 64: 33-89.
aṭ-Ṭabarī. Tārīkh = Annales Abu Djafar ibn Djarir at-Tabari, edited by M. De Goeje et
al. Vol I. Leiden, 1879.
Tropper, J. 1997. “Subvarianten und Funktionen der sabäischen Präfixkonjugation.”
Orientalia N.S. 66: 34-57.

362 Jan Retsö
———. 2000. Ugaritische Grammatik. Altes Orient und Altes Testament 273. Müns-
ter.
Waltisberg, M. 2009. Satzkomplex und Funktion. Syndese und Asyndese im Althoch-
arabischen. Wiesbaden.
Waltke, B. K. and M. O’Connor. 1990. An Introduction to Biblical Hebrew Syntax.
Winona Lake.
Watson, J. E. C. 1993. A Syntax of Ṣanʿānī Arabic. Wiesbaden.
Weninger, S. 2001. Das Verbalsystem des Altäthiopischen. Wiesbaden.
Wright, W. 1896-98. A Grammar of the Arabic Language I-II. 3rd ed. Reprint, Beirut
1974.

Clause Combining in East Semitic

Circumstantial Clause Combining
in Old Babylonian Akkadian*
Eran Cohen, The Hebrew University, Jerusalem
In memory of Prof. Gideon Goldenberg

A great scientist and a precious man
0. Introduction
It is quite difficult to define circumstantial expressions on syntactic
grounds:1 they are of a different structure and complexity and consequently do
not answer to a limited set of structural criteria. This is the case of Akkadian:
the oldest amply-attested Semitic language does not conform to the accepted
tradition of what a circumstantial expression is in Semitic. This in fact is one
of the basic insights of any syntactic comparison between sibling-languages.
Akkadian is unlike Arabic, which regularly features the indefinite accusative
participle (FĀʿILAN), or the pattern wa–x—YAFʿALU (as well as several other
expressions, see Marmorstein’s contribution in this volume). Nor is it similar
to Biblical Hebrew, which has a pattern similar to Arabic, for instance, waw–
x—QĀṬAL in narrative (e.g., Niccacci 1990, 62–72). There are, of course,
other expressions (see various forms and patterns throughout Isaksson, Kam-
mensjö and Persson 2009, 36-150). Akkadian has its small share of morpho-
logical possibilities, but its main instrument to express circumstantiality is the
use of certain forms within its syntax of chaining.
* I would like to thank Mr. Nikolaus Wildner for his corrections and remarks.
1 Less so on semantic grounds: “Any word or words expressive of some fact subordinate
to the main course of the narrative, or descriptive of some circumstance attaching or ap-
pertaining to the action denoted by the principal verb, may form a circumstantial clause
or secondary predicate: an adverb, a genitive or ablative absolute, a participle or other
word in apposition to the subject—all of which qualify the main action by assigning the
concomitant conditions under which it took place, be they modal, causal, or temporal—
are familiar instances” (Driver 1892, 195).

366 Eran Cohen
0.1 Literature review

The phenomenon of Umstandssätze or Zustandssätze described in the differ-
ent sources is not one but two. One is mentioned or discussed in all the
sources, which describe the literary language (Literary Old Babylonian or
Standard Babylonian – Hecker 1974, Wilcke 1977, Streck 1995), in which
2
IPARRAS forms have some kind of circumstantial function, basically occur-
ring only in the narrative of the epic (and in dream accounts, see Cohen
2006, 64–66). Although related, this phenomenon is not the one on which this
paper concentrates. It has been given ample consideration, despite the fact that
it occurs basically in the literary language and only rarely elsewhere.3 We will
return to it below (§4).
The first “modern” reference to circumstantial expressions in Akkadian is
found as early as 1907:
Wenn dem ersten mit -mâ verbundenen Präteritum oder Permansiv ein
Präsens folgt, so ist der zweite Satz als Zustandssatz, zuweilen auch
mit finaler, kausativer etc. Bedeutung aufzufassen: innabitma ibáḳam
ziḳnášu = er floh, indem er sich den Bart zerraufte; (Meissner 1907,
63)
Thus an attempt was made to equate an Akkadian form to the well-known

Arabic circumstantial clause pattern wa–x–YAFʿALU with a form like IPRUS–
ma IPARRAS. A similar attempt is found further in GAG §159:
“Sätze, die den Zustand kennzeichnen (vgl. die arab. Ḥāl-Sätze!), in

dem die unmittelbar davor genannte Handlung ausgeführt wurde, ste-
hen immer im Prs. auch bei vergangenen Handlungen (s. z. B. aB pī-šu
īpusam-ma izakkaram ana X “er hub an, indem er zu X sprach” [...];
iddi rigma Tiāmat ul utāri kišassa “T. stiess ein Geschrei aus, indem
2 Among other forms which participate in this function—PARIS, non-verbal clauses

(NVCs), UL IPRUS—see Wilcke (1977) and Cohen (2006, 54–63).
3 See, e.g., CH §141–143, where IPARRAS forms have a similar function; in the letters it is
just as rare:
al[p-am] ipṭur–ma {šamm-ī ikkal} [imq]ut–ma imtūt
ox-ACC 3CS.free.PST–CONN plant-OBL.PL 3CS.eat.NPST 3CS.fall.PST–CONN 3MS.die.PF
‘He released the o[x], {it was eating grass}, [it f]ell and died’ (AbB 11, 7:13–14).
Note that the Leipzig glossing rules are applied with the examples; the square brackets
are habitually used in transcribing reconstructed broken entities. For this reason,
curled brackets or bold typeface are used to mark the circumstantial clause (CC).

Circumstantial Clause Combining in Old Babylonian Akkadian 367
sie den Hals nicht umwandte” En. el. IV 71); sie werden meist mit, sel-
tener ohne -ma angeschlossen. Wie es scheint, sind solche Sätze eben-
so wie die gleichartigen, in §158f genannten Finalsätze nur der Dich-
tung eigentümlich, in ihr aber sehr beliebt.” (my emphasis)
Steiner (1985), in his work titled “Umstandssätze im Akkadischen”, is unfor-

tunately the least informative of all other sources: it is an unsuccessful attempt
to prove that IPARRAS forms which follow the connective particle -ma are
synchronically subordinate (by trying to show that the particle -ma actually
subordinates) and thus constitute subordinate Umstandssätze in Akkadian.
First, -ma is not a subordinating, but rather a chaining device (see Cohen
forthcoming). Subordination is quite clearly marked in Akkadian, and -ma
does not necessarily occur with it. Second, subordination is hardly a requisite
for circumstantiality: looking at the jumla ḥāliyya pattern found in Arabic
(e.g., wa–x–YAFʿALU), it is not at all certain that this prototypical syntagm is
subordinate rather than chained. Third, Steiner uses for his inquiry materials
for which he does not distinguish for dialect and genre, some of them serious-
ly outdated.
The other strategy, which is described below, occurs everywhere – in let-
ters, laws and literature. Nevertheless, it is described succinctly in Rowton
(1962, 271–278) and Kraus (1987, 41–45). Consequently, there is much more
to be said about it in this regard – on the syntactic context in which the phe-
nomenon occurs and the forms which participate, which constitute the main
purpose of this article. The fact that this type of circumstantial expression
occurs independently of genre or textual environment (it may occur almost
anywhere) makes it fundamental to Akkadian syntax. The strategy is de-
scribed as a stative PARIS form that is connected forward by the particle -ma
and thus constitutes the circumstantial clause (CC) syntagm. The first to de-
scribe this was Rowton (1962):
In this syntax the clause with permansive describes circumstance in

which the event of the second clause took place. The circumstance of
which the first clause speaks often consists of the state of the subject at
the time of performing action which the second clause speaks of. The
event of the second clause is very often the result of the circumstance
in the first clause, but, as already pointed out by Lambert, this is by no
means always the case.
What we have here is a simple form of syntax in which parataxis is
used instead of subordination. In OB the enclitic particle usually de-

368 Eran Cohen
notes no more than sequence of events when it is affixed to the preter-

ite or the perfect. But when affixed to other tenses it usually denotes a
logical connection of some sort between the two sentences it joins. We
are concerned here only with the combination permansive+present or
preterite (and variants thereof). (Ibid., 272; my emphasis)4
Kraus (1987) discusses “Koppelungen”, which are a wide variety of pairs of

verbal expressions where the first has some kind of modificatory function
with regard to the second.5 His pairs are characterized by a unity of tense,
modus, person and number across the two clauses, to which we can add polar-
ity as well, that is, being affirmative or negative. The type dealt with below,
namely, the stative in circumstantial function, is described ibid. (41–45). Here,
when a stative is the first part of the Koppelung, there is a difference: no unity
is found with regard to aspect (to which we intend to add other characteristics
below: polarity, mood, person and perhaps tense). This kind is described as
follows:
Die Satzkoppelungen dieser Sondergruppe beschreiben Handlungen

und den Zustand ihres Subjekts, der sie ermöglicht, verursacht oder
rechtfertigt. (Kraus 1987, 41; my emphasis)
All these sources speak of the stative as the exponent of circumstances,

mostly in conjunction with one more clause. There is an emphasis on the
paratactic nature of the connection. The aim of this paper is to provide a more
precise characterization of the (macro)syntactic conditions in which this phe-
nomenon occurs and the members of the class, namely, what other forms
belong to this group which signals circumstantiality. This should, in addition,
facilitate the identification of the phenomenon.
0.2 A short typological ID of Old Babylonian

Old Babylonian (henceforth OB) is the oldest attested Semitic language. It is
found written in cuneiform in a huge number of assorted texts – letters, laws,
4 Following Rowton, an addition was inserted into GAG3 (§159): “In Dichtung wie Prosa
finden sich Zustandssätze der Form “Stativ(-ma)+Verbum finitum”, wobei der Stativ die
Umstände angibt, in denen sich die Handlung des zweiten Verbs abspielt”. The syntax is
explained below, §3.
5 The classical example is ul t-atūr–ma ul t-ali-am ‘you did not come up again’ (AbB 7,
178:3), see §5 below.

divinations, epic literature, and more. It is an inflecting language, featuring the

Semitic three case system, as well as root and pattern word formation. The
syntactic features of OB are dwelt upon below, §2.
1. Circumstantial expressions at sentence-level

As is made clear in GAG (§147), we find very few circumstantial examples in
OB within the confines of the sentence, consisting of two almost specialized
morphological strategies, which seem to be based on adjectives:
(1) wēd-ūt-am attalak

single-ABST-ACC 1CS.go.PF
‘I have come alone’ (AbB 13, 149:13)
Note that the circumstantial expression is in the accusative (termed in GAG

§147 ‘der adverbiale Akkusativ des Zustandes’). The first type consists of a
derivation from the adjective 6 followed by the abstract marker -ūt-. The
result is sometimes different, semantically, from the homophonous rare ab-
stract nomen actionis (compare wašbūtum ‘sitting’):
(2) šaql-ūs-su šeʾ-am suḫr-ı̄ –ma šām-ı̄

rare-ABST-GEN.3MS barley-ACC look_for.IMP-FS–CONN buy.IMP-FS
‘It being rare, look for barley and buy (some)’ (AbB 14, 140:31)
(3) ana mīn-im šūm-ū lā bašl-ūs-sunu in-nappal-ū

to what-GEN garlic-PL NEG ripe-ABST-GEN.3MP 3.PASS-dig_out-NPST-MP
‘Why is the garlic dug out unripe (lit. their-unripe-ness)?’ (AbB 12,
25:12–14)
(4) kasp-am marṣ-ūs-su ašaqqal

silver-ACC difficult-ABST-GEN.3MS 1CS.pay.NPST
‘I will weigh the silver with difficulty (lit. its-being-difficult)’ (AbB 9,
61:27)
6 This judgement is based upon the meaning (e.g., šaqlum ‘rare’, rather than ‘weighed’),
with the addition of the derivational morpheme (-ūt-), which forms abstracts. The no-
men actionis (e.g., wašbūtum ‘sitting’) is not particularly related to the verbal adjective,
and is mainly found in objective function: wašb-ūs-su iqb-û-nim ‘They told me (of) his
staying…’ (AbB 13, 21: 13). See also balṭ-ūs-su in ex. (37).

370 Eran Cohen
Note that exx. (2)–(4) include, in addition, explicit reference to one of the
arguments: subject in the case of lā bašl-ūs-sunu ‘(in) their being unripe’,
referring to the garlic, but of the object in the other two cases – the barley (in
ex. (2)) and the silver (in ex. (4)). This reference to one of the arguments in
the clause is considered as a basic feature for circumstantial expressions – that
is, different from what is repeatedly said in the Assyriological grammatical
literature, e.g., Rowton (1962, 272) and Kraus (1987, 41), that the circumstan-
tial generally refers to the subject. Himmelmann and Schultze-Berndt 2005
term this “participant-oriented adjuncts”, where the participants are the
“main” predicate’s core arguments.
The second form is a limited gerundial form, which seems to consist of an
adjective (or lexeme) followed by the unique morpheme -iššī- and a genitive
pronoun, which is functionally similar:
(5) erî-ššī-šu lā illakam

naked-ADV-GEN.3MS PROH 3CS.come7
‘Let him not come naked (lit. to his naked)’ (AbB 12, 178:6’)
This gerundial form may serve as the rheme of the clause8 as well as the cir-
cumstantial expression which depicts the state of one of the arguments and
refers to it (in this case it is the subject) with the attributive pronoun. This
form may occur focused as well:
(6) wēd-iššı̄ -ka–ma tūr

single-ADV-GEN.2MS–PTCL return.IMP.MS
‘Return alone! (lit. to your single)’ (AbB 14, 208:10–11)
This gerund is marked, additionally, by the focus particle -ma.
7 In independent, non-interrogative, non-conditional clauses, the form IPARRAS (otherwise

present-future), when preceded by the negative particle lā, functions as a prohibitive
(glossed PROH). This form is part of the directive domain, the negative counterpart of
the precative (=jussive) and the imperative.
8 E.g., inanna anāku er-îššī-ya
now NOM.1CS naked-ADV-GEN.1CS
u mārat-ki erî-ššī-[š]a
CONN daughter-GEN.2FS naked-ADV-GEN.3FS
‘Now, I am naked and your daughter is naked (lit. to my/her naked)’ (AbB 6, 102:8–9).

Other, sporadic means are the preposition ina with various kinds of ab-
stract nouns, which, together with the negative particle lā, mean ‘without’ and
function as the negative of the gerundial constructions:
(7) šumma aššat awīl-im itti zikar-im šan-îm

IF wife.NUC man-GEN with male-GEN another.MS-GEN
ina itūl-im ittaṣbat
in sleep.INF-GEN 3CS.PF.PASS-catch
‘If a man’s wife is caught lying with another male...’ (CH §129)
This is a very good case showing an infinitive construction as circumstantial:

itti zikarim šanîm ina itūlim ‘lying with another male’ does not explicitly
refer to the wife, but since lying with someone else is a digression mostly with
regard to a (married) woman, the prepositional syntagm must refer to her.
Other examples occur with the negative particle:
(8) ina lā wašb-ūt-i-ya PN...

in NEG sit-ABST-GEN-GEN.1CS
an[a] bīt-im unūs-su uš-t-ērib
to house-GEN equipment-GEN.3MS 3CS.CAUS-PF-enter
‘In my absence, PN9… has brought his equipment into the house’ (AbB 6,
116:7–9)
In this case the form wašb-ūt-um is the nomen actionis, which is here the
equivalent of the infinitive. Note that no explicit relationship is marked be-
tween the adverbial syntagm and the clause, and it is left to be inferred. The
following example refers implicitly in its circumstantial prepositional syntagm
to the subject argument in the clause:
(9) [šeʾ]-um šakin ina lā rīq-ūt-im–ma

barley-NOM STV.put.3MS in NEG empty-ABST-GEN-PTCL
ul uš-ābil-akkim
NEG (1)CS.CAUS-carry-DAT.2FS
‘There is [bar]ley.10 I did not have (it) sent to you only due to lack of
time’ (AbB 11, 40:9–10)
9 PN=proper name; GN=geographical name; CN=canal name.

10 There is no need to interpret šeʾum šakin as an ad hoc counter-factual conditional (had
there been...), as is done in the edition.

372 Eran Cohen
The particle in ex. (9) (as well as in ex. (6) above) is analyzed as the focus
particle -ma.
The ina infinitive constructions, described by Aro (1961, 215-251) as de-
noting temporal or instrumental phrases, are also a strategy for the expression
of circumstantiality. They show varying complexity. The simplest (and rarest
in this function) is what we have in ex. (7) – ina parāsim. In most cases, how-
ever, it is somewhat more complex: a genitive suffix pronoun representing the
agent is appended to the infinitive construction, as in ex. (10):
(10) awīl-um ina lā lamād-i-šu

man-NOM in NEG be_aware.INF-GEN-GEN.3MS
marušt-um imaqqut-aššum
trouble-NOM 3CS.(be)fall.NPST-DAT.3MS
‘Trouble will befall the man (when) unaware’ (YOS X:31 xii:32-35)
In other cases, an object is found preceding the entire infinitive construc-

tion:
(11) šumma šamn-um m-ê ina nadî-ka

IF oil-NOM water-OBL.PL in cast.INF.GEN-GEN.2CS
itb-û 11
3.rise.PST-MP-CONN
‘If the oil, upon your pouring water (on it), rises...’ (CT V pl. 4–6, L.
32, Aro 1961, 239)
One expression with this structure has even become a fixed expression:
ṭuppī/kunukkī (anniam) ina amārim/amārika ‘upon (your) seeing my tab-
let…’.
In addition to the object, other arguments may precede the infinitive:
(12) anāku ana bīt-im ina erēb-i-ya

NOM.1CS to house-GEN in enter.INF-GEN-GEN.1CS
pān-ī-ya ulawwā–ma attatial
face-OBL.PL-GEN.1CS (1)CS.surround.NPST–CONN 1CS.PF.lie_down
‘Upon my entering the house, I wrapped my face and lay down’ (AbB
14, 110:20–21; Aro 1961, 237)
11 šamnum ‘oil’ takes here the plural form of the verb.

(13) šumma alp-um sūq-am ina alāk-i-šu

IF ox-NOM street-ACC in go.INF-GEN-GEN.3MS
awīl-am ikkip–ma uš-t-amīt
man-ACC 3CS.gore.PST–CONN (3)CS.CAUS-PF-die
‘If an ox gores a man to death while passing through the street’ (CH
§250, Aro 1961, 238)
This construction is distinct enough also when occurring with a core argument
preceding the infinitive construction without the genitive suffix pronoun:
(14) šiṭert-am ina šakān-im ašar id-im

document-ACC in put.INF-GEN place.NUC hand-GEN
ašakka[n]-šu
1CS.put.NPST-ACC.3MS
‘Upon submitting the document, I will place it in a suitable place’
(AbB 14, 148: 21–22)
This pattern includes a rare expression of the subject:
(15) m-û ina maqāt-im ana našpak-i-šu l-i-tēr

water-NOM.PL in fall.INF-GEN to silo-GEN-GEN.3MS DIR-3CS-return
‘Water falling, let him return (the grain) to the silo’ (AbB 9, 215:9–11;
Aro 1961, 222)
(16) PN ina alāk-im 2 būr eql-am ... n-u[ka]llam-šu

in go.INF-GEN bur field-ACC 1CP-show.NPST-ACC.3MS
‘PN arriving, we will s[ho]w him the field of two bur...’ (AbB 9, 72:9–
11)
These rare occurrences are very similar in function to the rest of the examples.
The infinitive construction with the argument inserted between the preposition
and the infinitive is basically not found as a circumstantial expression, but
rather as a core argument (Aro 1961, 229).12 By maintaining this close junc-
ture between the preposition and the infinitive, the language made its first step
12 dayyān-am šuāti ina dīn idīn-u en-êm

judge-ACC DEM.OBL.MS in sentence-NUC 3CS.sentence.PST-SUBORD change.INF-GEN
ukann-ū-šu–ma
(3).convict.NPST-MP-ACC.3MS-CONN
‘They will convict this judge in changing the sentence he gave’ (CH §5).

374 Eran Cohen
toward a gerund, which however had not developed any further, as far as we
can tell.
The examples so far all have in common the fact that they constitute part
of another sentence, clearly fulfilling a formal adverbial function within the
confines of sentence-level. However, the relationship with one of the core
arguments is not explicitly marked in all cases (e.g., exx. (11) and (15)).
Before moving on to the heart of the matter, circumstantial clause combin-
ing (given the acronym CCC), one first has to provide an explanation as to the
way the syntax of Old Babylonian Akkadian works.
2. Clause combining in OB syntax

Everyday OB shows differential element orders: in nominal and adverbial
syntagms, modifiers basically follow (e.g., substantive—adjective, preposi-
tion—attribute), whereas in verbal syntagms modifiers precede, namely, the
verbal form takes the last position, following its arguments (subject—object—
verb).
Subordination in OB is a clear, formally marked phenomenon: it always
involves an explicit syntactic nucleus which is often marked as such (either a
substantive, a pronoun or an adverb~preposition~conjunction)13 and the ver-
bal form is basically marked by the subjunctive, namely, the mark of subordi-
nation. The clauses thus marked serve, with their nuclei, as part of another
clause – as substantive, adjective and adverbial clauses:
(17) kīma anāku eppeš-u qibī-šum

as NOM.1CS 1CS.do.NPST-SUBORD tell.IMP.MS-DAT.3MS
‘Tell him that I cultivate (the field)’ (AbB 3, 2:45)
The clause in ex. (17) is an embedded, subordinate object clause.

However, subordination is not the only strategy for clause combining; in
addition to subordination, OB has the asymmetrical connective particle -ma.
This particle creates a series of verbal forms (as well as of other predication
types), a linguistically-pertinent, non-reversible clause sequence which is
characterized by “modal” congruence.
13 They may sometimes constitute the same entity: kīam ‘thus’ (unbound form) vs. kīma
‘like, as’ or ‘that, when, since, etc.’ (bound, nucleus form).

These series, where forms are interconnected via the particle -ma are
termed domains. These are units larger than the clause, which roughly paral-
lel the paragraph in other languages and different genres.
There are three major domain types: indicative, directive and subjunctive
(or attributive), each comprising different forms.14 The forms in each domain
do not interconnect in principle via -ma with other forms (the circumstantial
expressions constitute one notable exception), and thus we have what may be
termed domain congruence (for a full description of this syntax, see Cohen
2014).
2.1 The indicative domain

The indicative domain is termed after the common feature of the forms which
occur in it:
(18) u eql-ī ana errēš-im iddin-ū–ma

CONN field-GEN.1CS to farmer-GEN 3.give.PST-MP-CONN
errēš-um uqallil-anni–ma
farmer-NOM 3CS.humiliate.PST-ACC.1CS–CONN
ana maḫar rabiān-im u šīb-ūt āl-im allik–m[a]
to front mayor.GEN CONN elder-MP.NUC city.GEN 1CS.go.PST–CONN
aw-ât-im anni-āt-im maḫrī-šunu aškun–ma
matter-PL-OBL DEM-FP-OBL front-GEN.3MP 1CS.put.PST–CONN
mār-ī maḫ[ar] rabiān-im u šīb-ūt āl-im
son-GEN.1CS front.NUC mayor-GEN CONN elder-MP.NUC city-GEN
īpul-anni–ma kīam aqbī-šum#
3CS.reply.PST-ACC.1CS–CONN thus 1CS.say.PST-DAT.3MS
‘Moreover, they gave my field to a farmer and the farmer humiliated me,
so I went in front of the mayor and the city elders and stated these mat-
ters in front of them, and my son answered me in front of the mayor and
the city elders and I told him thus’ (AbB 9, 268:11–21)
In the example, there are only forms denoting the preterite, but other forms
may occur in the same domain. There is no obligatory referent continuity in
the verbal forms, and the only mandatory features are basically continuity of
the same mood, in this case the indicative, and the particle -ma, which natu-
rally does not follow the last clause in the chain.
14 There are other, less important domain types, e.g., in pronominal questions, protases,
and apodoses.

376 Eran Cohen
Within the indicative domain, when the chain consists of preterite forms
(as we have in ex. (18)), it tends to end with a special form:
(19) ana PN ṭupp-am uš-ābil–ma meḫer ṭupp-i

to PN tablet-ACC (1)CS.CAUS-carry.PST–CONN answer.NUC tablet-GEN
uš-ābil-am–ma [u]š-t-ābil-akkum #
(3)CS.CAUS-carry.PST-DAT.1CS–CONN (1)CS.CAUS-PF-carry-DAT.2MS
‘I sent PN a tablet, he sent me a response and [I] sent (it) to you’ (AbB 3,
55:30–32)
The formal difference between the final form and the rest of the forms is an
infixed t (ušābil vs. uš-t-ābil). This verbal form has several functions, the
most common of which is to mark off a chain of preterites. In other syntactic
contexts, it has altogether another function.
2.2 The attributive domain

The attributive domain consists of forms marked as such by the subordina-
tive morpheme (glossed as SUBORD). Its function is to mark the clause as the
attribute of a preceding nucleus. Such nucleus may be a noun (which is often
marked accordingly by the so called “construct state”, i.e., the unbound form),
a pronoun or a preposition/conjunction. The noun governed by the nucleus is
marked by the genitive case, whose sole function is to mark it as an attribute.
In fact, the subordinative morpheme can be considered an allomorph of the
genitive case: it marks verbal clauses as attributes, while the latter marks
nominals as attributes, both are found in the same syntactic slot:
(20) #aššum {#kīma dunn-ī ḫabt-u#}

TOP.MARK THAT farm-GEN.1CS STV.rob.3MS-SUBORD
t-ešm-û–ma t-akkud-u–ma
2MS-hear.PST-SUBORD–CONN 2MS-worry.PST-SUBORD–CONN
t-ašpur-am # umma atta–ma
2MS-write.PST-DAT.1CS QUOT NOM.2MS
‘As for (the fact that) you heard {that my farm was robbed} and you
worried and you wrote to me as follows: ...’ (AbB 7, 116:4–6)
This kind of chain is sometimes tricky, since the formal exponent of these
attributive predicative forms is sometimes neutralized (e.g., the last form in
ex. (20), tašpuram, where the subordinative morpheme is masked by the da-
tive morpheme -am, which neutralizes the difference between subordinative

and indicative). Since, however, this domain must follow a nucleus which
marks its beginning (unlike the other two domains), we can identify it fairly
well. Note that in ex. (20) we have two attributive domains, one within the
other, each beginning with a nucleus (aššum and kīma respectively) and end-
ing when there is no more -ma.
2.3 The directive domain

The directive domain consists of directives, i.e., the collective term used here
for jussives, imperatives, cohortatives and prohibitives: these diverse forms
function as a suppletive group found in the same paradigm and accordingly
share a function:
(21) ina ṣāb PN1 u PN2 100 ṣāb-um

from army.NUC and troop-NOM
ittī-šu l-i-llik–ma
with-GEN.3MS DIR-3MS-go–CONN
5 ūm-ī {adi PN1 u PN2 ištu GN illak-ū-nim#}
5 day-OBL.PL until and from 3.come.NPST-MP-ALL15
ina āl-ānī l-i-ptarrik-ū–ma
in cities-OBL.PL DIR-3-ITER.trouble-MP–CONN
ḫarrān-āt-im {ša ī-ten-errub-ā-nim#} [i]šteat ū šitta
caravan-PL-OBL PRON.NUC 3-ITER-come_in-FP-ALL one or two
l-i-dūk-ū–ma l-ī-dur-ā #
DIR-3-strike-MP–CONN DIR-3-fear-FP
‘Let one hundred troops from the troops of PN1 and PN2 go with him,
and let them cause continuous difficulties in the cities for five days {until
PN1 and PN2 come from GN}, and let them strike at one or two caravans
{that come in regularly} so that they be afraid’ (AbB 11, 193:13–23)
Inside two individual directive clauses, one finds in addition two occurrences
of the attributive domain (marked by curly brackets). Note that the attributive
forms of the verb are not interconnected with the directives, but embedded in
the clauses. The notions expressed by the clauses in the directive domain are
1. an expression of will (imperative, jussive); and 2. finality (in non-initial
clauses only). For instance, the second and third directives (underlined) ex-
press will, whereas the fourth directive (līdurā) expresses finality. The latter
15 This morpheme (ventive, allative) is one which, with verba movendi, points in the
direction of the speaker.

378 Eran Cohen
is, furthermore, a notion that needs to be examined for a potential circumstan-

tial function. Two more notions are 3. indirect command, or order content;
and 4. concessive-conditionality (see §3.3.1).
A case of deontic content expression is found in ex. (22):
(22) qibī-šum–ma l-i-llik–ma aḫ-ā-šu

tell.IMP.MS-dat.3MS–CONN DIR-3CS-go–CONN brother-ACC-GEN.3MS
l-itr-am–ma {l[ām]a attalk-u}
DIR-3CS.lead_forth–CONN before 1CS.PF.leave-SUBORD
nikkass-ī-šunu l-ī-puš-ū
account.OBL.PL-GEN.3MP DIR-3.do-MP
‘Tell him he should go and (OR: in order to) bring over his brother so
they can do their accounting {before I (will) have left}’ (12, 44:16–21)
The second and third directives (underlined) are taken to be the notional (ra-
ther than syntactic) content of the order (and consequently not a potential
circumstantial). The attributive domain in this example is clearly marked
(curly brackets).
The directive domain has its own unique complement syntax, as compared
with the other domains:
(23) qi[b]ī–ma ma[mman l]ā udabbab-šu {directive domain}

tell.IMP.MS–CONN PRON.INDEF PROH. 3CS.harass-ACC.3MS
‘O[r]der that (lit. and) n[o one] should harass him’ (AbB 12, 13:17–18)
(24) mamman lā dubbub-šu [i]qtabī-šunūšim {indicative domain}

PRON.INDEF NEG harass.INF-GEN.3MS 3CS.PF.tell-DAT.3MP
‘[He] ordered them that no one should (lit. any[one no]t to) harass him’
(ibid. 12–13)
The same events are referred to in two different domains: Whereas the indica-
tive verbal form in ex. (24) (‘he ordered them’) is formally complemented by
an infinitive in accusative status (‘not to harass’), which is its formal object,
the directive domain in ex. (23) works differently: Here object clauses or
infinitives are rare (compare, for instance, ex. (17)). Normally, the content of
the order is conveyed as a chained (rather than subordinate) clause, using an
entirely different strategy (lit. ‘order and let no one harass him’).

2.4 Domains and their relationship to CCC

The domain syntax, or chaining, presented in the previous sections, is the
main vehicle of circumstantial clause combining in Old Babylonian. However,
like one more paradigm,16 the protasis paradigm in the paratactic conditional
pattern (Cohen 2012, 78–90), it sometimes deviates from modal congruence.
This strategy operates clearly above sentence level, at the chain, or paragraph
level.
3. The circumstantial clause in OB: forms, constraints and

paradigms
The following example is taken from the Eshnunna law collection, but it illus-
trates the spirit of these constructions very well:
(25) {šumma kalb-um šegī–ma}

IF dog-NOM STV.raving.3MS–CONN
bābt-um ana bēl-i-šu uš-ēdī–ma
municipality-NOM to master-GEN-GEN.3MS (3)CS.CAUS-know.PST–CONN
kalab-šu lā iṣṣur–ma
dog-GEN.3MS NEG 3CS.guard.PST-CONN
awīl-am iššuk–ma uš-t-amīt
man-ACC 3CS.bite-PST–CONN 3CS.CAUS-PF-die
‘{If a dog is raving} and the ward authorities notify its owner but (the
owner) does not guard his dog and it bites a man and kills him...’ (LE
§56)
The first form in the legal protasis is šegi ‘it is raving’, which serves as cir-
cumstance to the entire chain of events mentioned. They are all interconnected
via the connective particle -ma. However, despite the fact that the example is
representative, many details need to be verified in order to provide the fullest
picture possible for this phenomenon – forms, location in the chain, polarity
and other issues. Note that the dog in the first clause takes various syntactic
positions in the following clauses: in the second clause it is genitive (‘his
master’), in the third it is accusative (‘does not guard his dog’) and in the last
two clauses, it is the agent (‘it bites a man and kills (him)’). This CC is poten-
16 By “paradigm” I mean a syntactic substitution group, namely, a set of forms that share
the same place in the syntagm and hence the same function.

380 Eran Cohen
tially more extensive in function than what is mentioned in the literature as

describing ‘the state of the subject’ (end of §0.1).
3.1 Forms
This presentation includes an expansion of the paradigm, namely, the group of
forms which participate in this function, as well as some refinement as to the
syntactic conditions where it occurs.
It seems that, in addition to the stative form which denotes the circum-
stantial function, it is now possible to consider several other forms in this
function: First, the negative preterite UL IPRUS, which seems, very much like
the stative PARIS, to constitute, or represent the circumstances. This idea is
tested against affirmative forms in the same position below. Additionally,
non-verbal clauses (NVCs) seem to take part in this paradigm as well.
First and foremost, however, it is the stative form, reinforced by semanti-
cally similar forms: statives “de facto”, or syntactic statives.
3.1.1 Statives
The stative is a predicative form which denotes all kinds of states (Kouwen-
berg 2010, 163-164), as such it naturally expresses circumstances which are
non-eventive. Another important research (Loesov 2011) is a description of
the temporal, aspectual and diathesis-related characteristics of the stative. The
cases of stative with past time dynamics (ibid., 86-88) are viewed semantical-
ly as kind of pluperfects, which does not contradict circumstantiality:17
(26) {alp-ū ša PN ḫalq-ū–ma}

ox-NOM.PL PRON.NUC STV.be_lost-3MP–CONN
ina qāti PN1 u PN2 tamkār-im iṣbat-ū-šunu<ti>–ma
in hand CONN merchant-GEN 3.seize-MP -ACC.3MP–CONN
‘{Oxen of PN having been lost}, they found them in possession of PN1
and PN2 the merchant...’ (Goetze 1958, 28:4–8. Loesov 2011, 287 “Oxen
of PN had been lost, and…”)
It is possible to see here the general sense of circumstantiality vis-à-vis more

specific nuances, such as concessivity:
17 See my other paper in this volume (“Circumstantial clause combining in the Jewish
Neo-Aramaic dialect of Zakho”), §2.1.

(27) {ittī-kunu bāʾer-ūt-am e[pē]š-am kullum-ū–ma}

with-GEN.2MP hunt-ABST-ACC do.INF-ACC STV.instruct-3MP–CONN
t-aprikam–ma ana bāʾer-ūt-im epēš-im
2MS-hinder.PST–CONN to hunt-ABST-GEN do.INF-GEN
[ul t-]addin-aššunūti
NEG 2MS-give.PST-ACC.3MP
‘{They were assigned to d[o] the fishing with you}, but you hindered
and did [not] allow them to do the fishing’ (AbB 11, 112:18–22)
In the examples, there is mostly an explicit resumptive element (marked by an

underline), but not always:
(28) {u inanna šatt-um gamr-at–ma}

CONN now year-NOM STV.finish-3FS–CONN
eql-am ina erēš-im ul t-akm[is]
field-ACC in plow.INF-GEN NEG 2MS-finish.PST
‘{And now the year is over} but you have not fin[ish]ed plowing the
field’ (AbB 10, 96:7’–8’)
In this case the CC does not directly refer to any core argument. The farther
away one is from sentence-level, the less consistent this feature becomes.
Loesov (2011, 287-8, n. 56) claims that the functional difference between
IPRUS and PARIS in chain-initial position is rather one of markedness, where
18
PARIS explicitly signals background. In this framework, the differences be-
tween PARIS forms and affirmative IPRUS forms are viewed as more critical to
the system and they are consequently shown and discussed throughout.
In the following sections, various types of statives in this function are ex-
amined. The first feature to be checked is polarity.
18 “… in narrative passages, the relationship between paris-ma iprus and iprus-ma iprus
chains is that of markedness. The chain paris-ma iprus may suggest explicitly that there
is something “backgrounding” in its first link, while iprus-ma iprus is a default expres-
sion” (Loesov 2011, 287-8, n. 56).

382 Eran Cohen
3.1.1.1 Affirmative stative

(29) {ina CN m-û maṭ-û–ma}
in water-NOM.PL STV.little-3MP–CONN
eqel bilt-i-ni ul ikaššad-ū
field.NUC yield-GEN-GEN.1CP NEG 3.reach.NPST-MP
‘{The water in the CN-canal was in small quantity}, so it cannot reach
our yield’ (AbB 4, 39:8–10)
(30) {atta ana mār-ūt-i nadn-āta–ma}

NOM.2MS to son-ABST-GEN STV.give-2MS–CONN
ana ilik bīti (sic) ab-ī-ka ilteq-ū-ka
to ilkum-service.NUC house.NUC father-GEN-GEN.2MS 3.PF.take-MP-ACC.2MS
‘{You were given up for adoption} so they took you to the ilkum-service
of your (adoptive-)family’ (AbB 7, 125:13–15)
3.1.1.2 Negative stative

Negative statives are not different:
(31) {aḫ-ū-ni ṣēḫr-um aššat-am ul aḫiz–ma}

brother-NOM-GEN.1CP young-NOM.MS wife-ACC NEG STV.hold.3MS–CONN
PN ab-ū-ni aššat-am uš-āḫis-su
father-NOM-GEN.1CP wife-ACC (3)CS.CAUS-hold.PST-ACC.3MS
‘{Our young brother had no wife}, so PN our father married him to a
wife’ (AbB 3, 2:11–12)
(32) {kasp-am ul naši-āku–ma} ukult-am ul ašām

silver-ACC NEG STV.carry-1CS–CONN food-ACC NEG 1CS.buy.PST
‘{I carried no silver}, so I did not buy food’ (AbB 1, 132:7–8)
In addition, most stative forms are passive, and take no direct object. In that
they resemble intransitive lexemes. However, in a small number of examples,
we encounter CC examples which feature active statives; this is verifiable by
their explicit direct object. Exx. (31) and (32) are negative forms with an
explicit accusative object.

3.1.1.3 Active stative forms

Exx. (33) and (34) are affirmative:
(33) {PN annikīam saniq-niāti–m[a]}

here STV.check.3MS-ACC.1CP–CONN
alāk-am ana ṣēr-ī-ka ula n-ileʾʾi
go.INF-ACC to back-GEN-GEN.2MS NEG 1CP-be_able.NPST
‘{PN is checking on us here} so we cannot come to you’ (AbB 9, 88:6–
11)
sanāqum (‘check’) is here an active lexeme, and the speakers are the recipi-
ents. In the following example, the lexeme našûm is active again:
(34) {ṣuḫār-ū-ya šipāt-im naš-û–ma}

servant-NOM.PL-GEN.1CS wool-OBL STV.carry-3MP–CONN
PN īmur-šunūti–ma ... šipāt-im imtašaḫ
3CS.see.PST-ACC.3MP–CONN wool-OBL 3CS.PF.take_by_force
‘{My servants were transporting wool}, PN saw them and ... took the
wool by force’ (AbB 1, 95:7–11)
Note that the propensity toward a causal nuance (exx. (29)–(33)) is not exclu-
sive: like ex. (26), ex. (34) is a neutral circumstantial.
3.1.1.4 Syntactic statives

Syntactic statives are forms which have a stative meaning but do not comply
with the morphological structure of statives. This happens because these
forms, in addition to their meaning (exist, have, know, etc.) do not show any
paradigmatic opposition with other terms, that is, with different functions. For
instance, the verb bašûm ‘exist, be’ occurs almost exclusively in IPARRAS
formation (otherwise denoting non-past), but having hardly any temporal
opposition, it has come to cover the entire temporal spectrum, much like the
stative:
(35) {u šīm-um ul ibaššī–ma} ul attalkam

CONN merchandise-NOM NEG 3CS.exist.NPST–CONN NEG 1CS.leave.PST
‘{There was no merchandise} so I did not leave’ (AbB 12, 53:21–22;
see similarly ex. (55))
The same applies to IPRUS verbal forms without an opposition of other forms,
such as išûm ‘have’ and edûm ‘know’:

384 Eran Cohen
(36) {kasp-am elī-šu īš-û–ma}

silver-ACC on-GEN.3MS (3).have.PST-MP–CONN
nīš-ī-šu ešer itrû
people-OBL-GEN.3MS 10 3.take_away.PST.MP
‘{He owes them silver}, so they took away ten (of) his people’ (AbB 1,
74:14–15; see also ex. (75))
(37) PN mār-ī ištu 8 šan-āt-im iḫliq-anni–ma

son-GEN.1CS since year-PL-OBL 3CS.get_lost.PST-ACC.1CS–CONN
{balṭ-ūs-su ul īdē–ma} kīma mīt-im
live-ABST-GEN.3MS NEG (1)CS.know.PST-CONN as dead-GEN
kisp-am aktassip-šum
funerary_offering-ACC 1CS.ITER.make_funerary_offering-DAT.3MS
‘My son PN disappeared on me 8 years ago, and {I did not know
(whether) he was still alive (lit. his living)}, so I have kept making funer-
ary offerings as if he were dead’ (AbB 13, 21:5–9)
The CC consisting of stative forms is easy to spot. Further below we continue

to explore their additional characteristics, so as to be able to characterize them
as fully as possible.
3.1.2 Negative IPRUS forms

In addition to the statives, there are other forms which participate in this func-
tion in the same syntactic setting. They are analyzed as part of the paradigm.
The first form is the negative preterite, UL IPRUS. This form is shown to be
different from the affirmative IPRUS in Epic Old Babylonian (Cohen 2006, 59-
60), where UL IPRUS is considered part of the background, whereas IPRUS is
part of the foreground. The reason for this is that this form is dramatically less
transitive (using the terms as explained in Hopper and Thompson 1980 as well
as Lazard 2002), since the patient is not actually affected. Consequently, the
effect is similar in nature to less transitive forms, such as other background
forms.
The following pair of examples features the difference between affirmative
and negative IPRUS forms following an initial boundary:

(38) # {mamman ul īkim-anni–ma19}

INDEF.PRON NEG 3CS.deprive.PST-ACC.1CS–CONN
etbiam–ma attalkam
1CS.rise.PST–CONN 1CS.go_away.PST
‘{Nobody deprived me (of anything)}, so I rose and went away’ (AbB
5, 193:8’–9’).
(39) # {PN eqel ab-i-ya īkim-anni–ma}

field.NUC father-GEN-GEN.1CS 3CS.deprive.PST-ACC.1CS–CONN
ana rēd-im ittadin
to soldier-GEN 3CS.PF.give
‘{PN deprived me of my family’s field} and gave it to a soldier’ (AbB 4,
16:8–11)
Negative verbs, even lexemes which otherwise denote telic actions, are never-
theless characterized in a similar manner to the stative. It should be noted how
similar the function of ul īkimanni–ma ‘(no one) deprived me’ is to the sta-
tives mentioned above: it is not part of the chain of actions,20 whereas the
same verb in the affirmative (ex. (39)) certainly is. The following pair is simi-
lar:
(40) # {ana GN gerr-um ul imqut–ma} ul allik #

to caravan-NOM NEG 3CS.fall.PST–CONN NEG 1CS.go.PST
‘{A caravan did not arrive (lit. fall) in GN} so I did not go’ (AbB 2,
77:4–6)
(41) # {PN1 ṣuḫār-ī ... PN2 imqut–ma}

slave-GEN.1CS 3CS.fall.PST–CONN
ubtazziʾ-šu
(3)CS.PF.hurt-ACC.3MS
‘{PN1 my servant ... attacked PN2} and hurt him’ (AbB 2, 115:6–10)
Here too, the verbal form in ex. (40) has only indirect effect (not coming)
whereas in ex. (41) it has a direct effect and hence shows far stronger transi-
tivity.
19 Lit. i-TE-m[a]-an-ni-ma.
20 “A PROTOTYPICAL ACTION ... is an effective volitional discrete action performed
by a controlling agent and actually affecting a well individuated patient” (Lazard 2002,
152).

386 Eran Cohen
Negation in the case of IPRUS is critical, then, since it marks the verbal
form as different in nature from the affirmative form in the chain.
3.1.3 Non-verbal clauses

Since the statives are found somewhere on the scale between full-fledged
verbs and non-verbal clauses (NVCs),21 it makes sense to examine the latter as
well. NVCs turn out to be relevant and both types of NVC (unipartite and
bipartite) are found in circumstantial function.
The crux lies in the interpretation of the particle -ma at the end of the
clause: it may be analyzed as a focus marker (for which see Cohen 2005b, 32-
34), signaling a preceding focus, or as a connective:
(42) {ina qab-î-ni eql-um mād-um–ma}

in command-GEN-GEN.1CP field-NOM much-NOM-PTCL
izni-anniāti
3CS.be_angry.PST-ACC.1CP
‘{On our command the field is abundant}, so he got angry with us’
(AbB 3, 58:13-17)
The non-verbal clause which describes the field may denote either focus on
mādum (as is interpreted in GAG §126e and in Kouwenberg 2000, 34) OR
connection forward, and consequently some kind of logical relation with the
following clause.
In Cohen (2005a, 266, n. 31) there are two previously unexplained exam-
ples, which are now explicable, and are consequently discussed below (exx.
(51) and (52))
3.1.3.1 Unipartites
Unipartite NVCs are clauses consisting of one part only, which is essentially
the predicate (or better, the rheme, namely, the new information). Exx. (43)
and (44) are analyzed in Cohen (2005a:251-253) as a substantive followed by
the focus- or rheme-marking particle -ma (also GAG §126e) and termed
“causal existentials”:
(43) ṣuḫār-um {tidūk-um–ma} ul illikam

servant-NOM combat-NOM–PTCL NEG 3CS.come.PST
21 For a description of NVCs in OB, see Cohen (2005a).

‘The servant, {(because there was) combat}, has not come’ (AbB 7,
55:12–13)
(44) ṣubāt-am {ul gerr-um–ma} ul ušābil-akki

garment-ACC NEG caravan-NOM–PTCL NEG (1)CS.send.PST-DAT.2FS
‘(As for the garment about which you wrote), {(since there was) no
caravan}, I did not send you the garment’ (AbB 7, 11:4–6)
The fact that these examples consisted of an entity which was, at the same
time, both a substantive and a clause, made it difficult to be certain of the
function of the particle -ma. Huehnergard (1986, 238 n. 74) terms such exam-
ples “one-word existential clauses”, and states that the particle “serves to
isolate the word from preceding and following clauses”. This special structure
had to be re-evaluated, in order to be able to provide some explanation for the
particle. A comparison conducted between the standard case of CC in OB (ex.
(46)) and the case in question (ex. (45), which is similar to exx. (43) and (44))
highlights the similarity:
(45) aššum lā t-allikam umma anak[u–m]a

since NEG 2MS-come.PST QUOT NOM.1CS
{midde awāt-um–ma} ul illikam
perhaps matter-NOM–PTCL NEG 3CS.come.PST
‘Since you have not arrived I said “{perhaps (there was some) matter}
so he has not come”’ (AbB 8, 99:12–13)
(46) {maruṣ–ma} ul illikam

STV.be_sick.3MS–CONN NEG 3CS.come.PST
‘{He (was) sick} so he has not come’ (AbB 2, 212:9–10)
The comparison of these two structures suggests that perhaps the situation is
analogous, namely, that both the substantive (which functions as a unipartite
clause) and the stative occupy the same functional slot: both are circumstan-
tial, chained forward by the particle -ma. That is, the particle in this case is not
analyzed as a focus particle anymore, but rather as a connective. The follow-
ing example is similar:

388 Eran Cohen
(47) [x-x-o {ṣ]ibûs-su–ma} [ṭu]rd-aš-šu–ma

need-GEN.3MS–CONN send.IMP.MS-DAT.1CS-ACC.3MS–CONN
[o ṣ]ibût-ī l-ī-puš
need-GEN.1CS DIR-3CS.do
‘{(There is a) need of him}, so [se]nd him to me and let him do my
[n]eed’ (AbB 12, 14:9–11)
The unipartite clause may even express modality, epistemic as well as deon-
tic:
(48) šā[r]-um kīma ištakn-u

wind-NOM as 3CS.PF.settle-SUBORD
{lū mūš-um–ma} n-ušerreb–ma ṭēm-am
PTCL night-NOM–CONN 1CP-let_in.NPST–CONN report-ACC
n-išaparram
1CP-send.NPST
‘As soon as the wi[n]d has subsided, {even if it is night}, we will take
(the barley) in and send a report’ (AbB 7, 84:6’–8’)
(49) kunukk-ī-ka šalm-ūt-im iddin-ū-nim

document-OBL.PL-GEN.2MS whole-MP-OBL 3.give.PST-MP-DAT.1CS
{kīma iddin-ū-nim mūš-um–ma}
as 3.give.PST-MP-DAT.1CS night-NOM–CONN
{ul eptē–ma} rīq-t-am ul amnu
NEG 1CS.open.PST–CONN empty-FS-ACC NEG 1CS.count.PST
‘They gave me your undamaged documents. {When they gave (them) to
me, (it was) night}, so {I did not open (them)} so I could not count the
empty one’ (ABIM 20:61–63)
Comparing the syntagms lū mūšum–ma (ex. (48)) and mūšum–ma (ex. (49)), it
is clear that they are opposed: both are found in the same syntactic slot, func-
tioning as circumstantials. The difference between them is one of modality: it
is realized as concessive-conditionality (in ex. (48)) vs. causality (in ex. (49)).
Note that mostly these unipartites do not show any explicit reference to an
argument, the exception is in ex. (47) (the reference is underlined). This may
be partially attributed to the nature of the unipartite clauses, i.e., which mostly
consist of a substantive.

3.1.3.2 Bipartites
In addition to the unipartite NVCs, which needed a more elaborate analysis,
some bipartite NVCs are found, which helps us to put everything in perspec-
tive. With NVCs that terminate with the particle -ma, as is mentioned above,
the particle may mark the second element as focus:
(50) u eql-am šuāti šan-ûm–ma ibaqqar–[m]a

CONN field-ACC DEM.OBL.MS another-NOM-FOC 3CS.claim.NPST–CONN
taʾīšt-um ka-tt-um–ma #
loss-NOM yours-FS-NOM-FOC
‘Even if (it is) somebody else (who) claims the field, the loss is yours’
(AbB 11, 69:6’-8’)
Ex. (50) ends with a NVC followed by the particle -ma, which is analyzed as
focus for two reasons: first, the letter ends there, so the NVC cannot constitute
a CC for what follows. Second, the two segments šanûm ‘somebody else’ and
kattum–ma ‘yours’ contrast each other, which is a common rationale for focus
marking in OB. In other examples the decision is difficult, but in others focus
is totally out of the question:
(51) umm-u ina maḫrī-ka u mār-u ina maḫrī-ya

mother.NOM in front-GEN.2MS CONN son.NOM in front-GEN.1CS
uḫtalliq-ū–ma
(3).PASS.cause_loss.PST-MP–CONN
{u šū kāṣir-um–ma} šīpāt-im PN iddin-am
CONN NOM.3MS textile_weaver.NOM–CONN wool-OBL 3CS.give.PST-DAT.1CS
‘The mother near you and the son near me were caused a loss, {but he
being a carpet-weaver}, PN22 gave me wool...’ (AbB 7, 187:18–22)
The fact that a relevant person is a textile weaver is reported several lines
before and is not contrasted or contested, so the particle -ma cannot be ex-
plained away as focus. However, as a connective with what follows it makes
perfect sense.
Moreover, the order personal pron—noun (here and in the following ex-
amples) is not the usual order of elements and occurs otherwise only when
there is a special emphasis on the first element:23 It is clearly not the case
22 PN is marked as focus by a focus pattern, see Cohen (2005b, 34-35).

23 See Cohen (2005a, 265) for this order.

390 Eran Cohen
here, which makes it easier to regard this order, with the following parti-
cle -ma, as distinctive:
(52) u aššum PN ša bīs-su maḫrī-ka

CONN TOP.MARK PRON.NUC house-GEN.3MS front-GEN.2MS
{u šū aḫ-ī–ma}
CONN NOM.3MS brother-GEN.1CS–CONN
arḫiš aššas-su u mār-ī-šu [p]uṭram–ma
quickly wife-GEN.3MS CONN son-OBL.PL-GEN.3MS release.IMP.MS–CONN
‘And, as for PN whose house is in front of you, {he is my brother}, so
release his wife and sons…’ (AbB 2, 170:10–15) (see perhaps AbB 14,
92:11–15, where the ‘main’ form is possibly a conditional, as in exx. (74)
and (75))
The bipartite NVC functions as a CC just as the stative.

The following example may belong here as well. In ex. (53) we have diffi-
cult, poetic lines, which may well contain a modal bipartite clause with this
special element order:
(53) {atta lū šamš-um–ma} ṣēt-ka l-u-štaḫan

NOM.2MS PTCL sun-NOM–PTCL heat-GEN.2MS DIR-1CS-REFL.warm
{atta lū erēn-um–ma} ina ṣill-i-ka
NOM-2MS PTCL cedar-NOM–PTCL in shadow-GEN-GEN.2MS
ṣ[ēt-um] ay-y-iḫmuṭ-anni
heat-NOM NEG.DIR-3CS-burn-ACC.1CS
‘{May you be the sun},24 so I can warm myself in your heat; {may you
be a cedar} so the h[eat] does not scorch me in your shadow’ (AbB 9,
228:16–20)
Like the unipartite clause in ex. (48), the bipartite NVC is capable of express-
ing modality. There are other circumstantial NVCs, without personal pro-
nouns:
24 The first syntagm is an enigma as regards the analysis what kind of a lū syntagm this is,
whether deontic (‘be the sun’), epistemic (‘if you are the sun’) or asseverative (‘you are
the sun’).

(54) {ṭupp-āt-u-ya ina qāti ab-ī-ya–ma}

tablet-PL-NOM-GEN.1CS in hand.NUC father-GEN-GEN.1CS–CONN
adi ab-ī lā illikam ul adân
till father-GEN.1CS NEG 3CS.come.PST NEG 1CS.litigate.NPST
‘{My tablets are in my father’s hands}, so before my father comes I
will not litigate’ (AbB 11, 55:23–25)
Note that the CC is compatible with an adverbial clause.
(55) {PN erî-ššī-šu–ma} mimma ina qāt-i-ya

naked-ADV-GEN.3MS–CONN INDEF.PRON in hand-GEN-GEN.1CS
[u]l ibaššī–ma ul ulabbis-s[u]
NEG 3CS.exist.NPST–CONN NEG (1)CS.dress.PST-ACC.3MS
‘{PN is being naked}, but there was [n]othing in my hand so I did not
dress him’ (AbB 12, 178:5–7 and see similarly ibid. 3’–5’)
Note that in ex. (55) there are two circumstantial syntagms – the NVC as well
as the existential clause which follows it.
There is one difficulty with these non-verbal constructions which has not
been addressed: contrary to the principles of the domains (§2 above), where
the same mood is observed throughout the chain, when circumstantial NVCs
are involved, this so-called modal congruence is less strictly maintained. For
instance, in exx. (48), (52) and (65), where the particle -ma interconnects two
different modi on each of its sides. One could have excused it by the mere fact
that these are non-verbal clauses, but it so happens that NVCs in OB are mo-
dally sensitive and express it (by the particle lū). Some of these clauses are
indicative, yet they are compatible with modal clauses in the chain as well as
the other way around.
3.2 Constraints
There are several constraints which characterize CCs in OB: position in the
chain, and reference to arguments.
3.2.1 Position in the chain

Clauses consisting of initial stative forms followed by a connective particle
and another clause have long been recognized as circumstantial, as has been
mentioned in the literature review (§0.1). The following pair is almost a text-
book pair, illustrating the crucial importance of clause order:

392 Eran Cohen
(56) {maruṣ–ma} ul illikam

STV.be_sick.3MS–CONN NEG 3CS.come.PST
‘Being sick (lit. {he is sick and}) he has not come’ (AbB 2, 212:9–10)
(57) ward-am t-addin-am–ma

slave-ACC 2MS-give.PST-DAT.1CS–CONN
ištu t-addin-a (sic) maruṣ
since 2MS-give.PST-DAT.1CS STV.be_sick.3MS
‘You gave me a servant but since you gave (him) to me, he has been
sick’ (AbB 11, 94:22)
Note that the form itself (maruṣ) is not enough to tell us whether it is circum-
stantial or not. The syntagmatic factor, namely, the location in the syntagm is
indispensable: when maruṣ ‘he is sick’ is initial (as it is in ex. (56)) it consti-
tutes the circumstances (and specifically the cause) of the agent in the follow-
ing clause; when final (ex. (57)), it is not circumstantial at all but rather func-
tions as the main message.
The initial place of the stative means its location at the beginning of the
chain, following a syntactic boundary. The following examples follow such a
boundary:
(58) # {PN annikīam saniq-niāti–m[a]}

here STV.check.3MS-ACC.1CP–CONN
alāk-am ana ṣērī-ka ula n-ileʾʾi
go.INF-ACC to back-GEN.2MS NEG 1CP-be_able.NPST
‘{PN is checking on us here} so we cannot come to you’ (AbB 9, 88:6-
11, =ex. (33))
(59) # {inūma t-uṣ-û marṣ-āta–ma}

when 2MS-exit.PST-SUBORD STV.be_sick-2MS–CONN
pī-ya ul ēpuš-akkum ul unaʾʾid-akka
mouth-GEN.1CS NEG 1CS.do-DAT.2MS NEG (1)CS.instruct-ACC.2MS
‘{When you went out, you were sick}, so I did not talk to you, did not
instruct you’ (AbB 1, 8:8–10)
Ex. (59) follows a syntactic boundary, it is the beginning of a chain. The sub-
ordinate clause at the beginning (inūma tuṣû) is a part of the CC clause. How-
ever, all types may actually occur also in mid-chain:

(60) eql-am ina GN ēriš–ma kīma t-īd-û

field-ACC in 1CS.plow.PST–CONN as 2MS-know.PST-SUBORD
{CN sekir–ma} m-û ul ibašš-û
STV.be_closed.3MS–CONN water-NOM.PL NEG 3.exist.NPST-MP
‘I plowed the field in GN, but as you know, {CN is closed} so there was
no water’ (AbB 6, 115:7–9)
The stative form sekir occurs in mid-chain and still functions as a circumstan-
tial to what follows: it constitutes the reason why there is no water. In ex. (61)
the form UL IPRUS is circumstantial, occurring in mid-chain position:
(61) 2 būr eql-am ina kanikt-im iknuk-šum–ma

2 bur field-ACC in seal-GEN 3CS.seal.PST-DAT.3MS–CONN
{adi inanna ina lā rēq-ūt-im ul illikam–m[a]}
till now in NEG empty-ABST-GEN NEG 3CS.come.PST–CONN
eql-am ul t-a[dd]iš-šum#
field-ACC NEG 2MS-give.PST-DAT.3MS
‘He sealed a field of 2 bur for him, {but till now, in the absence of time,
he did not come}, so you did not give him the field’ (AbB 4, 51:9–13)
The form ul illikam–ma in mid-chain is no doubt a circumstantial clause.

However, compared to the last clause in ex. (56) (ul illikam# ‘he has not
come’), it is easy to realize that the last position in the chain cannot host a
circumstantial clause. The last clause in ex. (56) is negative, but does not
constitute a circumstantial clause. The same applies here, in ex. (61), where ul
taddiššum# is not a circumstantial either. The same applies in the following
example:
(62) 1 šiqil kasp-am ušābilam–ma

1 sheqel.NUC silver-ACC (1)CS.send.PST–CONN
{têr-ēt-um ul išlim-ā–ma} ul uṣ-û-nim
omen-PL-NOM NEG 3.be_good.PST-FP–CONN NEG (3).go_out.PST-MP-ALL
‘I sent you one sheqel of silver but {the signs were not good} so they did
not go out’ (AbB 12, 129:5–6)
The initial boundary turns out to be non-pertinent, since mid-chain position is

just as normal for circumstantial function. What is important are the criteria
“clause-final” and “non-clause-final”.
It is worthwhile to note that the prevalent notion expressed in all these
suggested cases of CC is causality. This is true for cases of NVCs as well.

394 Eran Cohen
3.2.2 Explicit reference to the arguments

There are however, cases which are borderline: they consist of stative forms
found in the right syntactic constellation, but do not show an explicit, or al-
most-explicit, reference to one of the arguments in the following clause(s).
The following two cases, both featuring the stative form mād–ma, are a case
in point:
(63) {šeʾ-um ina bīt-im mād–ma} pīḫ-ū

barley-NOM in house-GEN STV.much.3MS–CONN beer_jar-NOM.MP
ul iml-û
NEG 3.be_filled.PST-MP
‘{The barley in the house is plenty} but the beer jars are empty’ (AbB 9,
177:3–4)
In ex. (63), the grain itself is not an argument in the following clause, still, the
beer in the jars, which is the unspecified issue in the following clause, is logi-
cally related to the grain. The nuance here is concessive. A similar semantic
relation exists above in ex. (60), where the canal and the water are closely
related. In ex. (64), on the other hand, the long distance mentioned in the first
clause is merely the reason for the inability to deliver:
(64) {šidd-u mād–ma} ištu maḫrī-ya...

distance-NOM STV.much.3MS–CONN from front-GEN.1CS
šūbul-am ul eleʾʾi
send.INF-ACC NEG 1CS.be_able.NPST
‘{The distance is great} so I cannot send ... from my place’ (AbB 11,
66:21–23)
The difficulty in ex. (64) is real: in these examples we have trouble relating
the state depicted in the circumstantial clause to the following clause. It is
much easier in the following case, where there is not a stative formation, but
rather a non-verbal clause:
(65) {šeʾ-um ša maḫrī-k[a] ula mād-um–[ma]}

barley-NOM PRON.NUC front-GEN.2MS NEG much-NOM–CONN
išar-iš l-i-mḫ[u]r-ū
straight-ADV DIR-3-receive-MP
‘{The barley in your possession is not much}, so let them receive (it)
duly’ (AbB 9, 88:14–16)

The NVC consisting of an adjectival predicate is a rarity. Such NVCs are

generally realized with the stative form (as in ex. (64)). This circumstantial
clause does not explicitly refer to any argument in the second clause, but nev-
ertheless there exists a connection: the barley is in fact the non-specified ob-
ject in the second clause.
The cases with stative forms where no connection exists with the following
clause are not many:
(66) 1 pān šeʾ-am baši-am {nakr-um bāb-am

barley-ACC STV.exist.3MS-DAT.1CS enemy-NOM gate-ACC
ṣabit–ma} šūbul-am ul eleʾʾi
STV.hold.3MS-CONN send.INF-ACC NEG 1CS.be_able
‘I have one pān (of) barley. {The enemy blocks the gate}, so I cannot
send (it). (AbB 9, 160:20–23)
The enemy holding the gate is, like several CC examples, the reason why
barley cannot be delivered, but it is different from the more conventional
cases in that the state is related to the following clause only indirectly.
Yet another type of example, already discussed above (§3.1.3.1), seems to
belong semantically, but turns out to be somewhat different syntactically:
(67) u kīma t-eštenemm-î {nukurt-um–ma}

CONN as 2-ITER.hear.NPST-FS war-NOM–CONN
mamman bāb-am ul uṣṣi
PRON.INDEF gate-ACC NEG (3)MS.leave.NPST
‘and as you keep hearing, {(since there is) war}, no one exits by the
gate’ (AbB 6, 64:15–17)
This example belongs with the group of unipartite NVCs; the entity in ques-
tion constitutes an existant, which, despite the logical relationship with the
following clause, has no direct relationship with any following argument, and
consequently no explicit reference.
3.3 Other forms

Apart from the forms attested so far in the circumstantial slot, there are other
forms which need examination: various functions of the directive forms
LIPRUS as well as of IPARRAS.

396 Eran Cohen
3.3.1 Directive clauses

In §2.3 above, it was shown that non-chain-initial directives may denote final-
ity (as well as indirect commands). The notion of finality as well as some
other functions of the chained directive need to be examined for a potential
circumstantial function – whether it is consistent and whether it is formulable
(thus becoming predictable).
In the first slot in the chain, directive forms usually express the will of the
speaker. At that very slot, however, there exists an additional, rare notion –
concessive conditionality:25
(68) {kasp-am ū šeʾ-am l-i-kill-ū-nikkum–ma}

silver-ACC or barley-ACC DIR-3-have_available-MP-DAT.2MS–CONN
lā t-amaḫḫar
PROH.2MS-accept
‘{Even if they have silver or barley available for you}, do not accept’
(AbB 14, 37:13–15; see also AbB 9, 260:10–15)
Concessive-conditional may be considered a natural candidate for circumstan-

tial expressions. In fact, ex. (68) is perfectly analogous to lū mūšum–ma in ex.
(48), already established as a positive circumstantial.
The notion of finality (e.g., exx. (21) and (22)) is generally circumstantial
in meaning as well; the difficulty with it is that it is not found in the slots
discussed so far (namely, non-final positions), and is rather found in non-
initial positions. Another problem is circumscribing the notion syntactically –
it is deeply related to the verbal lexemes around it, for example, it typically
follows verba movendi. For this reason these directive clauses are excluded.
3.3.2 IPARRAS clauses

Another form, which especially needs to be examined, is IPARRAS, otherwise
the praesens/imperfect. This form, although part of the background in the
language of the epic, is very rarely found in circumstantial function outside it
(see n. 3 above for a few exceptions). One reason for this may be the promi-
nent role the form IPARRAS has in the protasis of the paratactic conditional
pattern (see §7 below). The sole exceptions to this are cases with UL IPARRAS
25 In a previous study (Cohen 2005b, 144–60), I describe this paratactic pattern as one in
which the first member is a directive form (1st or 3rd person), whereas the second mem-
ber may consist of indicative or directive forms.

in this syntactic constellation which occur following the adverb adīni ‘till
now’:
(69) {adīni šeʾ-am anni-am ul akammisam–ma}

still barley.ACC DEM-ACC NEG 1CS.harvest.NPST–CONN
ṭēm-ī ul ašpurak-kum
report-GEN.1CS NEG 1CS.write.PST-DAT.2MS
urram kīma aktamsa ašappar-akkum–ma
tomorrow as 1CS.PF.finish 1CS.write.NPST-DAT.2MS–CONN
‘{I have not yet harvested this barley} so I did not send you my report.
Tomorrow, as soon as I have finished I will send you…’ (AbB 10,
167:12–16)
Note that adīni … ul aparras–ma is not quite a present-future, nor is it pro-

spective, as the other cases of IPARRAS, but rather some kind of present per-
fect. UL IPARRAS–ma (the negated praesens) is often found as a (paratactic)
conditional protasis, which is another paradigm altogether. In addition, cases
of 2nd person negative forms ul t-aprus–ma and ul t-aparras–ma function only
as protases. The difference between the circumstantial and the paratactic prot-
asis paradigms is summarized under §7 below.
3.3.3 CC Paradigm and constraints

The table below summarizes all forms which participate in the circumstantial
paradigm and the various constraints which go along with it:
forward
syntactic location CC paradigm connection
paris
predicative ul iprus
forms adīni ul iparras
non chain-final (concessive) liprus -ma
clause
unipartite: P
NVC bipartite:
S—P and P—S
(modal and indicative)

398 Eran Cohen
4. Epic narrative
In Streck (1995, 35; 53–55; 79-83) it is stated that these circumstantial sen-
tences with IPARRAS have the same function as cotemporaneous temporal
clauses in other genres (inūma, kīma iparrasu ‘when he decides’ etc.). Streck
adds that these cases are unlike the Arabic ḥāl in that the (non-modal) circum-
stantial sentence can precede its main clause and does not have to come with
an explicit connective (mostly the particle -ma). This type of circumstantial
sentence occurs only in the literary texts.
The difficulty with this focus on IPARRAS forms is that in the narrative
parts of the epic, the main arena for Streck’s inquiry, other forms have this
function just as well (e.g., PARIS forms, NVCs and even negative IPRUS forms,
see Wilcke 1977 and Cohen 2006). When attempting to portray a function,
one needs to take into consideration all the forms that are somehow related:
they occasionally form a substitution group, a paradigm.
The circumstantial paradigm as discussed so far is analogous to the phe-
nomenon of setting in the narrative of the epic. It is those circumstantial
pieces of information which serve as background, or informational basis, for
the entire stream of events that follows, rather than for just one clause. This
was already reported by Wilcke 1977 for the beginning of the epic, and is in
fact valid for the entire epic (ex. (70)):

(70) Lambert, Millard and Civil 1969 I, 1–13

line text form
{inūma ilūNOM awīlumNOM
1
{When the gods, (as) man, adverbial
clause
ublūPFV dullaACC izbilūPFV šupšikkaACC}
2
bore corvée-work, suffered toil}
šupšik ilī rabīSTV–ma
BACKGROUND
3
The toil of the gods was great: paris
dullumNOM kabitSTV mādSVT šapšāqumNOM
4
work was heavy; distress was much.
rabūtumNOM anunnakuNOM sebettamACC
5
The great Anunnaki, the seven,
dullamACC ušazbalūIMPV igigīOBL
6 iparras
they would make the Igigi bear the toil.
anu abūšunuNOM šarruNOM
7
Anu, their father, the king,
mālikšunu qurāduNOM enlil
8
their counsellor, Enlil, the hero,
[guz]zalušunuNOM ninurta
9
their chair-bearer, Ninurta,
EVENTS
[u] gallušunuNOM [en]nugi

10
[and] their canal inspector, Ennugi,
qātamACC īḫuzūPFV qātiša
11
held hand in hand; iprus
isqamACC iddûPFV ilūNOM izzūzūPFV
12
They threw a lot, the gods, (and) divided:
anu īteliPF šamêšša
13 iptaras
(and) Anu went up to heaven;
IMPV=imperfective (IPARRAS); PV=perfective (IPRUS)
This genre exhibits a different syntax; for this reason, the forms in narrative
are given different names, which are closer to their actual functions. No gloss-
es are supplied, but the idea seems to be clear: in the background (ll. 1–6) we

400 Eran Cohen
have 1. a temporal clause (ll. 1–2);26 2. three statives (ll. 3–4); and 3. an im-
perfective (l. 6). All these forms are specialized exponents to express the
background, or in other words, information circumstantial to the events them-
selves. The events follow a list of participants (ll. 7–10), looking like a chain –
two preterites (ll. 11–12) followed by a chain-ending perfect (l. 13). The open-
ing of the epic is in fact a larger-scale circumstantial construction.
The table shows the similarities and differences between the two para-
digms:
The circumstantial paradigm vs. background forms in epic narrative:

other genres epic narrative
circumstantial clause paradigm background foreground
ul iprus ul iprus
same
paris ⇐ forms ⇒ paris +iprus
NVC NVC
imperfective IPARRAS has no the main +iptaras
iparras
function in the CC paradigm difference
The grey area denotes the forms commonly used for both the background in
the narrative of the epic as well as for the circumstantial clause in the other
genres. The only difference is the form IPARRAS which is not used in circum-
stantial clauses, perhaps because it is identified with the protasis of the para-
tactic conditional pattern.
5. The phenomenon of hendiadys

Old Babylonian, like Biblical Hebrew, has a mechanism where with specific
lexemes (e.g., ‘return’) two chained verbal forms represent one modified
event. Ex. (71) contains such a classical Koppelung (for which see Kraus
1987), where in the first slot the lexeme târum ‘return’ occurs:
26 This interpretation is found in the original edition (Lambert, Millard, and Civil 1969). I
prefer it because it accounts for perfective forms in the background, which may happen
inside a subordinate setting clause that, only in its entirety, is part of the background.

(71) [arḫ]iš šāpir-ī ṭuppa-šu [an]a PN

[quick]ly supervisor-GEN.1CS tablet-GEN.3MS to
l-i-llikam–ma lā itâr–ma lā idabbub
DIR-3CS-come–CONN PROH.3CS.return–CONN PROH.3CS.speak
‘May my supervisor’s tablet quickly come, and let him not claim again
(lit. let him not return and claim)’ (AbB 12, 92:9’–11’)
The first verbal form, a regular prohibitive,27 does not represent an ‘event’,
but rather an auxiliary function, denoting the notion of recurrence, which is
rendered in English by ‘again’. Note that this cannot, by the current standards,
be considered a function in its own right, but a special serial verb construction
that has to do uniquely with the lexeme târum ‘return’, which may occur with
other verbal forms (namely, in various tenses and modi) as well. This case is
very limited: the verb târum is identical in tense, modus, person, number and
polarity to the clause that follows, and there is a limit to what entities can
occur between the two verbal forms. In the epistolary corpus past-related
forms (preterite and perfect) are in general not used with this expression, and
when they occur with this lexeme they usually are a full-fledged verbal form:
(72) ana ṣuḫr-i-ya attūr–ma

to youth-GEN-GEN.1CS 1CS.PF.return–CONN
[ku]ru[m]mat-ī iṣṣe[ḫ]ir
[fo]od_r[at]ion-GEN.1CS 3CS.PF.dim[in]ish
‘Have I returned to my youth that (lit. and) my [fo]od r[at]ion has di-
mi[n]ished?’ (AbB 2, 150:6–9)
(73) itūr-am–ma aḫat-ki kīma īmur-u

3CS.return.PST-ALL–CONN sister-GEN.2FS as 3CS.see.PST-SUBORD
itbeam–ma ittalka
3CS.rise.PST–CONN 3CS.PF.go
‘She came back and when she saw your sister, she got up and left’ (AbB
2, 103:9–11)
In both examples (72) and (73), despite the full congruence, the first verbal
form represents an independent event.
It is impossible to describe this phenomenon as part of any adverbial or
circumstantial framework because it is impossible to circumscribe the exact
27 See n. 7.

402 Eran Cohen
syntactic behavior, namely, in what precise conditions this phenomenon oc-

curs. It is not part of the paradigm of the circumstantial clause.
6. The paratactic conditional pattern

Conditional protases are sometimes considered as part of the circumstantial
territory (Isaksson, Kammensjö and Persson 2009, 28, 245-246, 260-263), in
view of the often circumstantial function of the protasis. However, in Old
Babylonian the protasis of the -ma conditional pattern (Cohen 2012, 78-90) is
quite different from the circumstantial clause described below. In fact, the
circumstantial expression is compatible with the paratactic conditional pattern,
which a priori means that they constitute different categories. In exx. (74) and
(75) the circumstantial clause precedes the conditional pattern:
(74) u {kallat PN napi-at–ma} (PARIS–ma)

CONN daughter_in_law.NUC PN STV.take_as_pledge-3FS–CONN
ina nakkamt-im t-u<še>ṣṣî-ši–ma
from storehouse-GEN 2MS.CAUS-go_out.NPST-ACC.3FS–CONN
maḫrī-ki l-i-šib
before-GEN.2FS DIR-3CS-dwell
‘Furthermore, {the daughter-in-law of PN is taken as pledge}, so should
you release her from the storehouse, let her stay with you’ (AbB 9,
270:10–15)
(75) {kâti īšū-ka–ma}

OBL.2CS (1)CS.have-ACC.2MS–CONN
šani-am ešeʾʾī–ma ṭāb-kum
another-ACC 1CS.look_for–CONN STV.be_good.3MS-DAT.2MS
‘{Me having you}, should I look for someone else, would it please
you?’ (AbB 9, 226:4–6)
The stative form napiat in ex. (74) and the syntactic stative īšu in ex. (75) both
immediately precede a protasis clause, which in both examples consists of the
form IPARRAS. The following table compares the two paradigms:

form circumstantial clause -ma conditional protasis

functions as a future perfect;
ul iprus-ma functions as a preterite (2nd person is always conditional)
paris–ma the most common form (occurs only once, in the negative)
iparras–ma (only with adīni ul) the most common form
(S)-P–ma part of the paradigm non-occurring
(NVCs)
Note that these two groups have a minimal overlap of forms – UL IPRUS is the
only form found in both paradigms and in those cases it is usually easy to tell
them apart based on the following clause type (for instance, apodoses are
basically non-past). The other forms are specialized: IPARRAS is the default
conditional protasis in this pattern, whereas NVCs and statives are found only
in the CC paradigm.
7. Summary
In the last part, we recapitulate the strategies which are used to express cir-
cumstantiality in Old Babylonian Akkadian. Note how important the para-
digms are, namely, the groups of participating forms in each function, in char-
acterizing the patterns.
The first paradigm is the one used at clause level:
1. CLAUSE-LEVEL CIRCUMSTANTIAL EXPRESSIONS

-iššī-GEN.PRON
-ūt-(GEN.PRON) (all part of the clause)
#(core argument) ina (lā) PARĀSIM/PARSŪTIM (-GEN. PRON)#
The morphology is not very rich, there are two almost specialized exponents
(-iššī- and -ūt-), the latter are found with other functions as well. An interest-
ing syntagm, whose core consists of an infinitive construction with ina, is
central to the expression of circumstantiality at clause-level.
CCs above clause level, which are actually chained forward, seem to be
more common in OB. The following table reflects what we knew before this
inquiry: the stative form in this function has been known for half a century.
The unipartite NVCs are described in Kraus (1984, 43-44), Huehnergard
(1986, 235 n. 61, and 238 n. 74) and in Cohen (2005a, 249-253), but the con-

404 Eran Cohen
nection with the CC stative had not been made. The stative was thought to
occur initially:
2. CIRCUMSTANTIAL EXPRESSIONS ABOVE CLAUSE-LEVEL

S=subject
initial syntactic boundary is # paris–ma (ex. (56))
P=predicate
(initial stage)
considered a must (based on STV.3MS–CONN

examples from the grammat- (both forms deemed non- ⇕
ical literature on Akkadian) related)
existential substantive+
focus particle –ma (ex. (43))
⇓ ⇓ ⇓
syntactic boundary CC syntagm: all forms belong to forward
one paradigm connection
paris
predicative ul iprus
(current stage)
initial syntactic boundary is forms adīni ul iparras

non-pertinent; the CC is in (concessive)
principle a non-chain-final liprus -ma
clause. unipartite: P
bipartite:
NVCs
S—P and P—S
(modal/indicative)
The second table shows the results of the current inquiry: the connection ex-
ists between various types of NVCs (uni- and bipartite alike) and various
predicative forms in addition to the stative. It is now possible to see that the
phenomenon is a rich paradigm, which is both definable and recognizable. Its
place in the chain is anything but chain-final, except for NVCs and concessive
LIPRUS, which are only attested at the beginning of the chain.
References
AbB = Altbabylonische Briefe in Umschrift und Übersetzung. Leiden: Brill 1964–.
ABIM = Akram Al-Zeebari, Altbabylonische Briefe des Iraq-Museum, PhD Dissertati-
on, Westfälische Wilhelms-Universität zu Münster. Münster, 1964.
Aro, Jussi. 1961. Die akkadischen Infinitivkonstruktionen. Studia Orientalia 26. Hel-
sinki: Societas orientalis fennica.
CH = Codex Ḫammurabi

Cohen, Eran. 2005a. “Addenda to non-verbal clauses in Old Babylonian.” Journal of

Semitic Studies 50.2: 247-279.
———. 2005b. The modal system of Old Babylonian. Harvard Semitic Studies 56.
Winona Lake: Eisenbrauns.
———. 2006. “The tense-aspect system of the old babylonian epic.” Zeitschrift für
Assyriologie und vorderasiatische Archäologie 96.1: 31-68.
———. 2012. Conditional Structures in Mesopotamian Old Babylonian. Languages of
the Ancient Near East 4. Winona Lake: Eisenbrauns.
———. 2014. “The domain: a formal syntactic unit above sentence level.” In Strate-
gies of Clause Linking in Semitic Languages: Proceedings of the International
Symposium on Clause Linking in Semitic Languages 5-7 August 2012 in Kivik,
Sweden, edited by B. Isaksson and M. Persson, 233-252. Abhandlungen für die
Kunde des Morgenlandes 93. Wiesbaden: Harrassowitz.
Driver, Samuel R. 1892. A Treatise of the Use of the Tenses in the Hebrew and Some
Other Syntactical Questions. Oxford: The Clarendon Press.
GAG = Wolfram von Soden. Grundriss der akkadischen Grammatik. 3rd ed. Rome:
Pontificium Institutum Biblicum, 1995.
Goetze, Albrecht. 1958. “Fifty Old-Babylonian Letters from Harmal.” Sumer 14: 3–78
& Pls. 1–24.
Hecker, Karl. 1974. Untersuchungen zur akkadischen Epik. AOAT 8. Neukirchen-
Vluyn: Kevelaer, Butzon & Bercker.
Himmelmann, Nikolaus, and Eva F. Schultze-Berndt. 2005. Secondary predication
and adverbial modification: the typology of depictives. Oxford: Oxford University
Press.
Hopper, Paul J., and Sandra A. Thompson. 1980. “Transitivity in grammar and dis-
course.” Language 56.2: 251-299.
Huehnergard, John. 1986. “On verbless clauses in Akkadian.” Zeitschrift für Assyrio-
logie 76: 218-249.
Isaksson, Bo, Heléne Kammensjö, and Maria Persson. 2009. Circumstantial qualifiers
in Semitic: The case of Arabic and Hebrew, edited by Bo Isaksson. Abhandlungen
für die Kunde des Morgenlandes 70. Wiesbaden: Otto Harrassowitz.
Kouwenberg, Norbert J. C. 2000. “Nouns as verbs: the verbal nature of the Akkadian
stative.” Orientalia 69.1: 21-71.
———. 2010. The Akkadian verb and its Semitic background. Languages of the An-
cient Near East 2. Winona Lake: Eisenbrauns.

406 Eran Cohen
Kraus, Fritz R. 1984. Nominalsätze in altbabylonischen Briefen und der Stativ. Mede-
delingen der Koninklijke Nederlandse Akademie van Wetenschappen, Afd. Letter-
kunde, Nieuwe Reeks 47/2. Leiden: North-Holland Pub. Co.
———. 1987. Sonderformen akkadischer Parataxe, die Koppelungen. Mededelingen
der Koninklijke Nederlandse Akademie van Wetenschappen, Afd. Letterkunde,
Nieuwe Reeks 50/1. Leiden: North-Holland Pub. Co.
Lambert, Wilfred G., Alan R. Millard, and Miguel Civil. 1969. Atra-Hasis: The Baby-
lonian Story of the Flood. Oxford: The Clarendon Press.
Lazard, Gilbert. 2002. “Transitivity Revisited as an Example of a More Strict Ap-
proach in Typological Research.” Folia Linguistica 36.3-4: 141–190.
Loesov, Sergey. 2011. “The Suffixing Conjugation of Akkadian: In Search of Its
Meaning.” Babel und Bibel 6: 75–148.
Meissner, Bruno. 1907. Kurzgefasste assyrische Grammatik. Leipzig: Hinrichs.
Rowton, Michael B. 1962. “The use of the Permansive in classic Babylonian.” Journal
of Near Eastern Studies 21.4: 233-303.
Steiner, Gerd. 1985 “Umstandssätze im Akkadischen.” In XII. Deutscher Orientalis-
tentag, ausgewählte Vorträge. Zeitschrift der deutschen morgenländischen Gesell-
schaft, Supplement 6, edited by Wolfgang Röllig, 86–102. Wiesbaden.
Streck, Michael P. 1995. “ittašab ibakki ‘weinend setzte er sich’: iparras für die Ver-
gangenheit in der akkadischen Epik.” Orientalia 64: 33–91.
Wilcke, Claas. 1977. “Die Anfänge der akkadischen Epen.” Zeitschrift für Assyriologie
und vorderasiatische Archäologie 67: 153-216.
Yaron, Reuven. 1988. The laws of Eshnunna. 2nd ed. Jerusalem: Magnes.
YOS X = Albrecht Goetze. Old Babylonian Omen Texts. Yale Oriental Series. New
Haven: Yale University Press, 1947.

Index of terms

408 Index of terms
accompanying action, 10, 182, 313, 317, 320, 326, 327, 330,
185, 186, 187, 188, 189, 195, 350
199, 200, 201, 202, 203, 206, apodosis, 157, 158, 224, 225, 242,
217, 221, 224, 225, 226, 228, 244, 245, 246
229, 230, 236, 237, 238, 239, Arab grammarians, 77, 79, 90,
242, 252, 253, 255, 257, 258, 113, 114, 126, 128, 129, 130,
260, 261, 262 137, 140, 151, 163
active participle, 9, 19, 20, 24, 47, Arabic dialects, 13, 20, 21, 65,
88, 174, 200, 216, 217, 306 320, 327, 355, 357
addition, 10, 181, 182, 185, 186, Arabic vernaculars. See spoken
187, 189, 200, 202, 204, 205, Arabic, Arabic dialects
207, 210, 211, 223, 224, 231, Arabiyya, 10, 298, 299, 304, 306,
236, 241, 242, 245, 246, 248, 308, 309, 310, 312, 315, 318,
251, 255, 260, 311, 315 320, 321, 324, 326, 327, 338,
addition clause, 182, 206, 229, 355, 356, 357
252, 253 Aramaic, 171, 241
additional, 89, 182, 208, 209, 237, aspect, 18, 19, 22, 23, 31, 32, 37,
311 42, 43, 44, 45, 133, 135, 136,
additive, 16, 18, 48, 225, 237, 138, 140, 152, 163, 169, 170,
240, 241, 252 174, 175, 176, 177, 179, 181,
adjacency, 18, 45 187, 195, 197, 218, 220, 225,
adverbial, 16, 39, 46, 58, 70, 74, 229, 232, 233, 239, 240, 249,
75, 80, 84, 90, 112, 116, 126, 250, 251, 252, 255, 256, 308,
127, 128, 130, 131, 132, 135, 317, 327, 368
152, 158, 160, 189, 206, 210, aspectual, 21, 22, 30, 55, 69, 71,
217, 222, 233, 234, 272, 274, 133, 151, 161, 176, 186, 201,
279, 284, 286, 292, 371, 374, 289, 309, 380
391, 399, 401 asyndesis, 18, 19, 24, 39, 40, 45,
adversative, 35, 58, 68, 69, 75, 82, 46, 50, 80, 182, 184, 187, 194,
87 219, 222, 223, 245, 326
Afro-asiatic, 178 asyndetic clause, 17, 19, 24, 29,
Akkadian, 10, 11, 244, 302, 317, 32, 40, 41, 46, 47, 49, 50, 65,
318, 319, 327, 331, 333, 365, 78, 152, 189, 193, 198, 216,
366, 367, 374, 403, 404 240
Aktionsart, 22, 114 attendant CC, 58, 64, 70, 90, 111,
anterior/anteriority, 22, 23, 31, 32, 114, 181, 187, 188, 189, 217,
138, 156, 157, 170, 176, 179, 218, 221, 223, 226, 239, 249,
184, 186, 201, 202, 203, 204, 250, 251, 253
208, 209, 210, 225, 228, 232, auxiliarization, 72, 73
233, 249, 250, 251, 252, 258, auxiliary, 21, 28, 34, 41, 42, 45,
261, 281, 299, 303, 308, 309, 47, 66, 69, 71, 72, 88, 102,

Index of terms 409
108, 134, 135, 138, 322, 351, chained, 367, 378, 387, 396, 400,
401 403
auxiliation, 19, 41, 45, 50 chaining, 152, 365, 367, 379
b(i)-prefix, 30, 42, 81, 82, 83, 85, circumstance, 7, 58, 64, 65, 66,
86, 102, 103 68, 75, 90, 94, 99, 106, 111,
background(ed), 60, 89, 104, 131, 126, 181, 188, 197, 218, 221,
138, 152, 177, 181, 189, 200, 232, 251, 274, 326, 365, 367,
202, 217, 227, 255, 279, 288, 379
310, 316, 331, 334, 342, 344, circumstantial, 7, 58, 365
350, 356, 381, 384, 396, 398, circumstantial clause, 7, 8, 9, 10,
399, 400 15, 16, 17, 18, 19, 24, 25, 26,
backgrounding, 271, 381 27, 28, 29, 30, 31, 32, 24–32,
bədd, 81, 82, 98, 99, 103, 108, 34, 35, 37, 38, 39, 40, 41, 46,
109, 110, 111 47, 48, 49, 50, 55, 56, 57, 58,
Biblical Hebrew, 9, 21, 55, 56, 59, 61, 62, 63, 64, 65, 66, 67,
117, 167, 169, 170, 171, 172, 70, 71, 73, 74, 75, 77, 79, 84,
173, 175, 176, 178, 180, 186, 89, 90, 91, 92, 93, 96, 97, 100,
187, 188, 189, 190, 192, 194, 104, 106, 109, 110, 111, 114,
195, 196, 214, 224, 232, 235, 115, 125, 128, 130, 141, 151,
239, 242, 247, 259, 261, 298, 153, 156, 159, 187, 188, 217,
300, 302, 303, 305, 306, 308, 218, 220, 221, 222, 223, 226,
309, 318, 320, 328, 330, 331, 239, 247, 250, 281, 282, 283,
333, 344, 355, 357, 365, 400 284, 289, 297, 298, 302, 303,
bipartite, 272, 274, 276, 277, 291, 304, 305, 306, 307, 308, 309,
386, 389, 390, 397, 404 311, 313, 314, 315, 316, 326,
body of texts. See corpus (of texts) 331, 353, 365, 366, 367, 374,
bounded, 141, 149, 177, 232, 233, 379, 393, 394, 400, 402
234 circumstantial clause, asyndetic,
Canaanite, 170, 171, 172, 191, 24, 29, 31, 33, 37, 40, 41, 49,
194, 196, 214, 235, 259 63, 66, 129, 130, 131, 201,
causality, 9, 16, 48, 258, 301, 305, 248, 261, 316
308, 388, 393 circumstantial clause, syndetic,
CCC, 8, 10, 297, 337, 352, 353, 24, 25, 26, 28, 35, 36, 39, 45,
354, 355, 358, 374, 379 48, 49, 50, 131, 134, 141, 151,
CCs, 391, 403 152, 158, 162
Central Semitic, 10, 170, 171, circumstantial clause, verbal, 161,
172, 179, 181, 215, 261 163
chain, 195, 222, 247, 252, 255, circumstantial expression(s), 10,
259, 261, 280, 292, 310, 326, 11, 63, 64, 101, 109, 126, 217,
355, 357, 375, 376, 379, 381, 271, 272, 273, 275, 276, 278,
385, 386, 391, 392, 393, 396, 279, 283, 290, 291, 365, 366,
397, 400, 404

410 Index of terms
367, 369, 370, 373, 375, 396, complex predication, 125, 132,
402, 403, 404 133, 140, 143, 161
circumstantial qualifier, 15, 104 concessive, 16, 35, 40, 49, 75,
circumstantiality, 10, 11, 272, 272, 289, 378, 388, 394, 396,
278, 286, 287, 291, 365, 367, 397, 404
368, 372, 380, 403 concomitance, 277
Classical Arabic, 9, 22, 26, 70, concomitant, 147, 152, 199, 219,
108, 117, 125, 126, 129, 132, 222, 228, 229, 238, 249, 253,
133, 138, 145, 157, 160, 162, 303, 365
163, 170, 171, 217, 345, 350, conditional, 9, 16, 38, 59, 60, 65,
356 76, 78, 80, 81, 89, 97, 115,
clause, 57, 127, 173, 175, 297, 116, 157, 158, 224, 242, 244,
370 245, 246, 283, 288, 370, 371,
clause combination(s), 9, 16, 46, 379, 390, 396, 397, 400, 402,
50, 55, 67, 69, 76, 111, 116 403
clause combining, 7, 9, 38, 56, 65, conjugation, 21, 22, 172, 173, 235
77, 89, 130, 173, 190, 214, consecutive, 35, 40, 49, 170, 185,
222, 226, 230, 232, 243, 244, 186, 239, 299, 300, 301, 302,
260, 321, 357, 374 303, 304, 305, 306, 309, 310,
clause hierarchy, 55, 80 311, 312, 314, 317, 318, 326,
clause linking, 7, 8, 15, 16, 17, 18, 336, 344, 353
24, 25, 29, 32, 38, 40, 43, 45, consequence, 57, 181, 187, 199,
46, 48, 50, 57, 58, 77, 80, 81, 229, 240, 242, 255, 302, 303,
82, 85, 86, 87, 116, 181, 211, 304, 312, 315, 337, 338, 344
242, 356 consequential, 298, 300
clause-initial, 172, 190, 191, 192, contemporaneity, 300, 304, 330,
193, 194, 197, 199, 211, 212, 357
214, 215, 223, 224, 235, 236, contrast clause, 131, 152, 182,
260, 261 189, 204, 228, 251, 261
coalescence, 172, 214 conversive theory, 169
coincidental, 9, 140, 141, 143, coordinated, 28, 37, 65, 66, 68,
145, 149, 162, 163, 309 72, 73, 85, 311, 313, 314, 326,
comment (textlinguistic), 66, 70, 333, 335, 339, 351
73, 74, 75, 85, 87, 94, 96, 100, coordination, 25, 65, 66, 68, 86,
106, 107, 152, 156, 181, 189, 133, 182, 183, 188, 189, 244,
202, 204, 207, 219, 220, 249, 355
279, 292, 300, 308, 316, 322, copula, 20, 39, 83, 85, 86, 94, 103,
324, 326, 327, 329, 330, 333, 105, 275, 276, 281, 287, 288
334, 338, 344, 345, 346, 350, co-referential, 151, 153, 154
355, 357 corpus (of texts), 10, 15, 17, 24,
completives, 176 47, 48, 87, 95, 97, 100, 115,

Index of terms 411
125, 140, 149, 271, 302, 343, enhancing, 16

354, 401 enhancing clauses, 16, 48, 90, 131
co-temporal, 9, 22, 36, 133, 140, enhancing function, 100, 286
162 equal status, 63, 106, 184, 189,
Damascene Arabic, 9, 55, 56, 59, 193, 195, 213, 215, 216, 223,
62, 63, 80, 100, 110, 111, 115 224, 225, 226, 228, 230, 231,
database(s), 15, 32, 56, 61, 64, 66, 233, 237, 239, 257, 260, 261
70, 73, 78, 80, 81, 90, 91, 93, Ethiosemitic, 10, 317, 318, 319
96, 97, 102, 103, 108, 111, fa-, 152, 158, 307, 310, 311, 312,
115, 259 321, 322, 323, 326, 355
dependent, 9, 24, 45, 50, 67, 80, faʿala, 126, 133, 138, 140, 145,
84, 127, 131, 134, 151, 157, 146, 149, 150, 156, 157, 158,
158, 159, 160, 162, 184, 188, 159, 160, 161, 162, 163
201, 206, 260, 275, 276, 277, final, 60, 63, 71, 75, 78, 79, 80,
287, 288, 292, 307, 354 88, 89, 107, 108, 110, 111,
desententialized clause, 24, 30– 115, 116, 141, 142, 170, 171,
31, 35, 40, 47, 49 214, 232, 235, 272, 301, 305,
digression, 9, 55, 66, 73, 74, 85, 314, 316, 319, 342, 348, 352,
86, 87, 104, 105, 107, 184, 354, 356, 376, 392, 393, 396,
190, 194, 199, 220, 260, 371 397, 404
discourse, 15, 16, 17, 18, 19, 23, final vowels, 171, 235
26, 28, 35, 41, 42, 48, 49, 50, finality, 41, 79, 82, 110, 111, 142,
55, 68, 73, 104, 158, 173, 177, 308, 377, 396
178, 183, 185, 187, 189, 224, finite verb, 92, 94, 96, 98, 108,
237, 247, 261, 308, 309, 328, 169, 170, 174, 249, 309, 313,
352 331, 342, 343, 353, 356
discrete, 172, 192, 195, 197, 198, finite verbal gram, 176
211, 212, 216, 217, 385 first position. See clause-initial
domain, 9, 125, 126, 132, 133, focal, 58, 180, 181, 242
135, 159, 161, 162, 316, 370, focal clause, 181, 182
375, 376, 377, 378, 379 focus particle, 370, 372, 387, 404
Egyptian Arabic, 8, 15, 17, 18, 33 foreground(ed), 152, 279, 287,
elaboration, 8, 43, 46, 50, 130, 384
144, 182, 189, 193, 198, 199, fronted, 146, 153, 155, 159, 191,
200, 201, 202, 207, 208, 209, 193, 195, 212, 214, 216, 221,
210, 228, 229, 230, 233, 243, 235, 251
249, 250, 251, 258, 260, 311, Geez, 302, 319, 352, 353, 354,
315 356
embedded, 26, 28, 44, 66, 128, general present, 178, 179, 180,
130, 131, 134, 146, 151, 284, 193, 197, 203, 207, 210, 216,
374, 377 217, 277, 289, 291, 357
embedding, 44, 88, 131, 134, 214

412 Index of terms
gerund, 272, 274, 275, 276, 285, ḥāl muqaddar, 71, 110, 111, 141
291, 370, 374 head clause, 16, 27, 28, 48, 59, 60,
gerundial, 370, 371 66, 70, 71, 73, 83, 84, 89, 91,
gnomic present, 177, 178 108
gram switching, 9, 16, 17, 18, 29, hypotaxis/hypotactic, 8, 15, 16,
32, 34, 41, 46, 47, 48, 49, 55, 24, 32, 35, 40, 47, 48, 49, 50,
56, 57, 60, 79, 76–89, 93, 94, 55, 56, 57, 60, 61, 64, 65, 66,
95, 96, 99, 101, 102, 103, 104, 67, 76, 77, 79, 81, 83, 86, 87,
105, 108, 110, 111, 113, 115, 89, 109, 111, 114, 116, 117
116, 186, 199, 201, 202, 203, immediate future, 177
204, 207, 210, 219, 222, 226, imperative, 47, 75, 81, 85, 103,
239, 300, 307, 308, 310, 326, 108, 171, 178, 189, 198, 211,
338 221, 222, 230, 243, 244, 248,
gram(s), 35, 47, 60, 76, 89, 171, 370, 377
172, 173, 175, 176, 177, 178, imperfect, 24, 33, 47, 135, 185,
199, 202, 235, 253, 260, 347, 192, 273, 292, 298, 299, 300,
356 301, 302, 303, 304, 308, 309,
grammaticalization, 19, 20, 40, 318, 319, 324, 396
41, 42, 45, 47, 49, 50, 71, 175, imperfective, 22, 170, 171, 177,
176, 232, 328 178, 179, 215, 219, 224, 225,
grammaticalized, 9, 20, 41, 42, 45, 229, 235, 238, 261, 289, 290,
47, 162 292, 317, 318, 356, 399, 400
grammaticization, 178 indicative, 10, 172, 178, 183, 185,
grammaticized, 177, 178 191, 192, 193, 194, 195, 196,
grounding, 17, 279, 282, 292 198, 199, 201, 202, 207, 210,
Gulf Arabic, 40, 55, 56, 80, 97, 211, 212, 214, 215, 230, 236,
111, 115 256, 260, 277, 291, 308, 309,
ǧumla ḥāliyya, 36, 128, 129, 151 310, 312, 326, 333, 375, 376,
habitual, 22, 23, 24, 34, 177, 179, 377, 378, 391, 396, 397, 404
184, 196, 200, 206, 215, 217, inferred, 9, 16, 18, 31, 48, 58, 60,
219, 223, 225, 228, 230, 231, 184, 186, 240, 245, 260, 371
235, 238, 241, 317, 322, 323, infinite, 174, 175, 210, 243, 258
324, 325, 326 infinitive, 11, 174, 243, 258, 274,
ḥāl, 7, 9, 15, 38, 58, 63, 64, 65, 285, 305, 306, 339, 356, 371,
66, 71, 72, 76, 77, 79, 88, 89, 372, 373, 378, 403
90, 109, 110, 111, 113, 114, inherent time reference, 20, 21
115, 117, 126, 127, 128, 129, injunctive, 178, 192
130, 132, 140, 151, 152, 272, instructional discourse, 182, 183,
298, 299, 300, 301, 305, 308, 247
309, 311, 312, 315, 316, 321, integration, 9, 24, 41, 46, 47, 50,
322, 323, 324, 326, 329, 350, 132, 133, 135, 149, 152, 161,
353, 354, 355, 356, 357, 398 188, 214

Index of terms 413
interlacing, 9, 16, 45, 48 253, 272, 297, 298, 300, 315,

intransitive, 39, 40, 47, 49, 127, 331, 332, 333, 351, 355, 357,
136, 143, 148, 155, 158, 176, 398
382 main line, 10, 55, 66, 70, 73, 74,
iparras, 317, 319, 327, 366, 367, 75, 85, 86, 87, 104, 107, 175,
370, 383, 395, 396, 397, 398, 178, 182, 183, 184, 185, 186,
399, 400, 402, 403, 404 189, 193, 194, 195, 197, 198,
iprus, 178, 317, 318, 319, 366, 199, 205, 212, 215, 216, 220,
380, 381, 383, 384, 386, 393, 221, 222, 223, 226, 231, 232,
397, 398, 399, 400, 403, 404 233, 235, 236, 242, 247, 248,
iptaras, 399, 400 255, 256, 257, 261, 280, 344,
junction, 130 357
juncture, 17, 66, 80, 83, 116, 243, marked, 9, 11, 16, 23, 24, 25, 26,
373 29, 32, 33, 34, 35, 45, 48, 49,
jussive, 126, 172, 178, 179, 183, 50, 57, 58, 60, 75, 76, 78, 79,
185, 189, 191, 193, 194, 196, 80, 81, 82, 83, 84, 85, 87, 91,
211, 212, 213, 214, 221, 226, 101, 102, 105, 110, 112, 115,
239, 243, 260, 370, 377 116, 127, 128, 130, 131, 132,
juxtaposed, 63, 65, 69, 71, 78, 135, 137, 138, 146, 152, 157,
111, 134, 135 158, 180, 182, 184, 186, 188,
juxtaposition, 41, 58, 157, 289, 189, 207, 230, 237, 245, 272,
311, 314, 322, 332, 355 275, 277, 285, 293, 301, 306,
locational(s), 69, 71, 72, 78, 79, 307, 309, 310, 311, 313, 317,
88, 99 318, 323, 324, 327, 347, 352,
long prefix gram, 179, 214 356, 367, 370, 371, 374, 376,
long prefix verb, 9, 172, 192, 193, 377, 378, 381, 389
194, 197, 206, 214, 216, 219, markedness, 35, 55, 57, 76, 381
224, 235, 244, 256, 260 Masoretes, 169
-ma, 367, 368, 370, 372, 374, 375, matrix clause, 9, 46, 149, 152,
377, 379, 386, 387, 389, 390, 153, 162
391, 397, 398, 402, 403, 404 Minean, 298, 300
main clause, 8, 16, 17, 19, 24, 26, modal, 10, 23, 34, 37, 39, 72, 82,
27, 28, 30, 31, 32, 35, 37, 39, 131, 135, 141, 158, 169, 170,
40, 41, 44, 45, 46, 47, 48, 49, 174, 176, 180, 185, 191, 198,
50, 55, 56, 57, 60, 65, 66, 73, 199, 211, 213, 226, 227, 229,
74, 79, 81, 89, 91, 94, 102, 235, 241, 260, 278, 290, 298,
103, 108, 110, 112, 113, 114, 299, 301, 312, 314, 327, 353,
115, 116, 117, 128, 131, 133, 356, 365, 374, 379, 390, 391,
134, 146, 151, 152, 153, 158, 397, 398, 404
174, 180, 183, 184, 187, 188, modifying verbs, 134, 135, 138,
194, 206, 219, 224, 229, 233, 139, 140
239, 247, 249, 250, 251, 252,

414 Index of terms
mood, 44, 60, 80, 116, 170, 178, Old Babylonian (OB), 11, 366,
368, 375, 391 367, 368, 374, 379, 384, 387,
morphological merger, 171, 172, 389, 391, 400, 402, 403
235 optative, 23, 179, 180, 232, 234
motion verbs, 22, 42, 140, 141, parallel, 63, 153, 154, 237, 287,
144, 145 329
mutually-dependent clauses, 157 paratactic, 25, 64, 65, 69, 77, 78,
narrative prose, 178, 181, 182, 81, 87, 109, 116, 368, 379,
183, 200, 201, 217, 226, 249, 396, 397, 400, 402
252, 257 parataxis, 44, 64, 65, 66, 109, 367
negative clause, 173, 191, 194, paris, 366, 367, 380, 381, 397,
214, 225, 245, 250, 256, 257, 398, 399, 400, 402, 403, 404
258, 260, 262 participial, 58, 88, 90, 92, 95, 96,
Neo-Aramaic, 271 98, 101, 112, 128, 131, 137
nexal, 276 participle, 9, 17, 19, 20, 21, 22,
nexus, 146, 274, 275, 276, 287 23, 28, 30, 31, 32, 33, 35, 39,
nomen actionis, 369, 371 40, 43, 47, 49, 50, 60, 77, 79,
nominal clause, 26, 31, 33, 35, 39, 82, 87, 92, 93, 94, 95, 96, 101,
49, 130, 152, 156, 157, 158 108, 113, 126, 130, 133, 135,
non-circumstantial, 292 136, 137, 139, 143, 144, 145,
non-main clause, 8, 9, 18, 27, 29, 148, 149, 151, 152, 153, 154,
32, 39, 46, 56, 57, 60, 64, 70, 155, 156, 158, 159, 160, 162,
72, 75, 76, 77, 78, 79, 80, 81, 163, 175, 187,188, 197, 200,
82, 83, 84, 85, 89, 91, 93, 96, 207, 211, 227, 242, 272, 274,
101, 112, 113, 114, 115, 116, 275, 318, 352, 353, 365
180, 182, 184, 186, 189, 194, passive participle, 19, 113, 179
198, 217, 242, 249 past tense, 20, 22, 169, 170, 178,
non-subordinate, 285, 292 186, 232, 233, 234, 297, 300,
non-verbal clause, 82, 84, 94, 105, 301, 302, 303, 316, 317, 320,
108, 112, 386, 394 327, 330, 343, 344, 357
Northwest Semitic, 170, 171, 172, perception verbs, 29, 135, 145,
214, 241 146
noun clause, 174, 175, 179, 199, perception, concrete, 146, 150
201, 208, 209, 211, 227, 236, perception, notional, 146, 147,
243, 248, 249, 252, 258 148, 149
noun phrase, 27, 28, 173, 247 perfect participle, 274, 291
nucleus, 16, 132, 134, 157, 273, perfective, 22, 170, 177, 178, 179,
275, 374, 376, 377 183, 191, 193, 195, 196, 197,
OB. See Old Babylonian (OB) 201, 202, 209, 212, 218, 226,
off-line, 10, 279, 288, 291, 324, 227, 228, 232, 233, 239, 249,
326, 327, 329, 330, 333, 334, 250, 252, 253, 255, 256, 258,
338, 344, 355, 357

Index of terms 415
261, 289, 290, 317, 318, 399, 188, 251, 255, 282, 283, 285,
400 292
permission verbs, 135, 145, 147 present tense, 17, 21, 204, 226,
Phoenician, 171, 318 233, 303, 306, 317, 327, 357
poetry, 9, 140, 172, 178, 180, 182, present time, 21, 170, 204
183, 193, 195, 196, 198, 207, presentative clauses, 134, 158,
208, 209, 210, 212, 213, 214, 162, 279, 282, 283
217, 218, 220, 230, 231, 232, preterite, 170, 178, 179, 180, 185,
240, 241, 259, 306 280, 287, 290, 325, 368, 375,
polarity, 368, 379, 381, 401 376, 380, 384, 401, 403
postposed, 281, 283, 284, 285, privative, 274, 317
292 progressive, 22, 23, 30, 31, 33, 42,
pre-clause, 182, 185, 186, 187, 69, 76, 77, 84, 94, 95, 102,
189, 200, 202, 203, 204, 206, 133, 140, 158, 163, 174, 175,
208, 210, 211, 212, 225, 226, 177, 215, 216, 281, 304, 310,
227, 228, 236, 238, 242, 243, 317, 320, 356
246, 247, 257, 258, 260 progressivity, 23, 30, 35, 41
predicate, secondary, 127, 365 prose, 9, 15, 48, 125, 172, 173,
predication, 9, 37, 43, 46, 73, 74, 180, 185, 188, 191, 195, 196,
127, 128, 132, 133, 173, 174, 213, 218, 227, 259, 306
210, 236, 243, 247, 275, 374 prospective future, 232
predicative, 20, 112, 127, 128, protasis, 157, 158, 224, 225, 234,
129, 130, 131, 132, 133, 134, 237, 244, 245, 246, 379, 396,
135, 136, 137, 138, 139, 140, 397, 400, 402, 403
141, 143, 144, 145, 146, 148, Proto-Hebrew, 171, 172, 185, 214,
149, 150, 151, 152, 158, 160, 223, 235, 260
162, 163, 272, 275, 276, 297, pure addition, 229, 230, 238
314, 376, 380, 397, 404 purpose, 16, 30, 37, 40, 49, 110,
predicative paradigm, 132, 134, 131, 181, 198, 211, 212, 247,
140, 160 277, 290
prefix conjugation, 18, 19, 24, 25, qad, 9, 76, 133, 135, 138, 139,
33, 126, 133, 171, 172, 243 140, 145, 146, 147, 149, 150,
prefix form, 18, 20, 23, 25, 26, 29, 151, 153, 155, 156, 157, 159,
30, 32, 34, 35, 40, 47, 49, 60, 160, 161, 162, 163, 299, 308,
63, 71, 75, 76, 77, 79, 81, 82, 313
83, 84, 85, 87, 88, 89, 93, 101, qad faʿala, 9, 133, 135, 138, 139,
103, 105, 111, 114, 172, 215, 140, 145, 146, 149, 151, 153,
302, 303, 309 155, 156, 157, 159, 160, 162,
prefixed inflection, 170 163
preposed (clause), 92, 93, 96, 97, Qatabanian, 300
98, 99, 100, 101, 102, 103,

416 Index of terms
qatal, 169, 170, 171, 187, 211, simultaneity, 35, 37, 41, 63, 67,
222, 234, 241, 242, 243, 244, 100, 133, 143, 152, 219, 223,
251, 255, 317, 318, 319, 330 253, 303, 304, 305, 306, 307,
raising, 145, 146, 149, 151 313, 320, 326, 327, 330, 355,
reanalysis, 20, 21, 175 357
reason, 16, 131, 181, 220, 221, simultaneous, 9, 26, 28, 30, 31,
239, 249, 251, 258, 394, 395 69, 83, 84, 93, 97, 106, 114,
reference time, 176, 177, 281, 309 115, 131, 133, 140, 141, 146,
relative clause, 8, 9, 24, 27, 38, 153, 162, 182, 308, 317, 326,
59, 60, 66, 70, 76, 83, 87, 88, 327
115, 196, 233, 331, 336 slot, 8, 10, 133, 277, 279, 290,
relative tense, 22, 176, 302, 309, 376, 387, 388, 395, 396, 400
313, 356 specifying, 37, 38, 40, 49, 78, 116,
relative time, 181, 194 144, 153, 229
result, 60, 81, 85, 89, 151, 181, speech verbs, 135, 151
182, 187, 189, 198, 199, 200, spoken Arabic, 18, 20, 21, 40, 59,
203, 204, 205, 206, 218, 228, 115
240, 246, 256, 261, 303 state verbs, 140, 141
resultative, 23, 31, 32, 133, 139, stative, 22, 24, 94, 137, 138, 143,
140, 163, 176, 178, 179, 202, 145, 179, 356, 367, 368, 380,
210, 232, 233, 240, 255, 317 382, 383, 384, 385, 387, 390,
resumptive, 128, 381 391, 392, 393, 394, 395, 402,
rhetorical organization, 7, 16, 46 403, 404
Sabaean, 10, 297, 298, 299, 300, stative verbs, 22
302, 303, 304, 305, 306, 308, storyline, 48, 50, 173, 178, 183,
309, 316, 318, 319, 320, 330, 185, 189, 191, 195, 196, 201,
331, 333, 334, 339, 340, 342, 202, 208, 209, 210, 217, 218,
343, 349, 353, 354, 355, 356, 219, 220, 229, 239, 252, 254,
357 255, 256, 257, 260, 261, 262
same-event addition, 106, 181, subjunctive, 171, 277, 291, 312,
199, 210, 220, 230, 253, 315 313, 315, 333, 374, 375
semantic relationship, 9, 16, 18, subordinate, 22, 41, 44, 45, 50, 64,
40, 41, 43, 48, 49, 58, 65, 78, 65, 74, 78, 128, 131, 146, 157,
80 256, 272, 276, 277, 281, 289,
sentence level, 10, 272, 278, 290, 291, 298, 299, 300, 301, 303,
291, 292, 379 304, 305, 327, 331, 333, 339,
sequentiality, 152, 252, 253, 254, 340, 344, 350, 352, 355, 357,
255, 256, 261, 333, 347 365, 367, 374, 378, 392, 400
setting clauses, 158, 159, 162 subordinate clause, 22, 41, 44, 45,
short prefix gram, 178, 179, 214 50, 64, 65, 131, 146, 272, 277,
short prefix verb, 9, 175, 178, 183, 281, 291, 298, 299, 301, 303,
184, 193, 197, 212, 235, 260

Index of terms 417
304, 305, 327, 331, 333, 340, 203, 205, 206, 207, 210, 233,
344, 350, 352, 355, 357, 392 238, 240, 255, 256, 260, 261,
subordinating conjunction, 8, 48, 308, 310, 315
55, 78, 80, 116, 183, 184 temporality, 9, 16, 35, 48, 138,
subordination, 17, 49, 65, 68, 88, 146, 243
132, 173, 182, 301, 327, 333, tense form, 21, 169
355, 357, 367, 374 tense switching, 17, 49
subordinative, 16, 376 tenses, 169, 170, 186, 281, 302,
suffix conjugation, 22, 28, 32, 33, 315, 368, 401
126, 133 terminal, 133, 139, 142, 143, 144,
suffix form, 18, 26, 28, 29, 30, 32, 145, 147
40, 47, 49, 60, 68, 76, 77, 79, texteme, 272, 291
83, 84, 86, 87, 89, 93, 94, 101, text-level, 10, 278
103, 105, 114, 316 time frame, 9, 20, 133, 150, 152,
suffix verb, 9, 82, 103, 110, 175, 162
179, 190, 240, 260 topicalized, 25, 99, 172, 189, 190,
suffixed inflection, 170 201, 206, 213, 217, 218, 219,
supporting, 58, 181 221, 223, 224, 226, 227, 228,
supporting clause, 182 230, 231, 236, 237, 241, 249,
switch of clause type, 184, 201 250, 251, 252, 253, 254, 255,
syndesis, 24, 45 261
syndetic, 8, 9, 24, 25, 26, 28, 31, trajectory, 175, 178, 179, 207
35, 36, 37, 40, 45, 48, 49, 50, typological, 10, 71, 130, 320, 355,
63, 65, 116, 129, 131, 134, 368
141, 151, 152, 154, 158, 162, Ugaritic, 10, 170, 171, 319, 320,
185, 195, 198, 199, 211, 212, 327, 328, 329, 330, 355, 356,
223, 224, 225, 226, 227, 228, 357
229, 230, 231, 236, 249, 251, unequal status, 189, 260
256, 257, 261, 297, 356 unipartite, 272, 291, 386, 387,
syntagm, 10, 130, 173, 191, 192, 388, 389, 390, 395, 397, 403,
211, 223, 224, 234, 235, 236, 404
237, 241, 260, 274, 276, 282, unmarked, 9, 16, 19, 20, 23, 25,
283, 287, 288, 306, 309, 313, 29, 34, 35, 48, 49, 57, 60, 77,
314, 322, 327, 333, 338, 342, 87, 116, 131, 178, 179, 184,
343, 357, 367, 371, 379, 390, 185
392, 403, 404 unmarked tense and mood, 178
Syrian Arabic, 55, 60 unordered addition, 181, 187, 315
TAM, 171, 175, 300, 310 we/wa, 10, 184, 186, 188, 189,
temporal linking, 182, 189, 198, 190, 260
199, 203, 204, 315 verbal clauses, 92, 105, 152, 155,
temporal succession, 181, 182, 355, 357, 376
189, 195, 199, 200, 201, 202,

418 Index of terms
verbal complexes, 9, 131, 134, 194, 195, 196, 197, 198, 199,
135, 140, 144, 151, 158, 160, 200, 201, 202, 203, 204, 205,
162 206, 207, 208, 209, 210, 211,
verbal noun, 174, 175 212, 213, 214, 215, 217, 218,
verbal system, 9, 10, 125, 163, 219, 220, 221, 223, 224, 226,
169, 170, 171, 173, 175, 176, 227, 229, 230, 231, 233, 235,
179, 235, 271, 302, 303, 309, 236, 238, 239,243, 249, 250,
310, 316, 317, 318, 319, 320, 251, 252, 253, 254, 255, 256,
355, 356, 357 257, 259, 260, 261, 262
verbicization, 20, 21, 48 Vsuff, 9, 10, 33, 173, 175, 179,
verbless clause, 174 180, 182, 184, 185, 186, 190,
verbs of motion, 19, 24, 39, 41, 191, 194, 195, 196, 198, 199,
110, 135, 147, 314 200, 201, 202, 203, 204, 205,
verbs of posture, 19, 39, 41, 46, 206, 207, 208, 209, 210, 211,
50, 71 218, 220, 221, 222, 223, 224,
word order, 26, 27, 45, 171, 172, 225, 226, 227, 228, 229, 230,
173, 182, 190, 191, 211, 213, 231, 232, 233, 234, 235, 236,
214, 235, 236, 256, 260, 261, 237, 238, 239, 240, 241, 242,
307, 326 243, 244, 245,246, 247, 248,
Vpref, 10, 33, 34, 171, 194, 197, 249, 250, 251, 252, 253, 254,
199, 206, 207, 211, 212, 240, 255, 256, 257, 258, 259, 260,
243, 251, 304, 305, 306, 307, 261, 262, 298, 299, 305, 306,
310, 312, 313, 315, 316, 318, 307, 310, 311, 313, 315, 316,
319, 320, 321, 322, 323, 324, 318, 319, 321, 322, 324, 325,
325, 326, 327, 328, 329, 330, 326, 327, 328, 329, 330, 331,
331, 333, 334, 335, 336, 337, 333, 334, 335, 336, 338, 339,
338, 341, 342, 343, 344, 345, 340, 341, 342, 343, 344, 345,
347, 348, 350, 352, 355, 356, 347, 348, 350, 351, 352, 355,
357 356
VprefL, 9, 171, 179, 184, 190, xabar, 127, 128, 129, 130, 132
191, 192, 194, 196, 199, 203, yafʿalu, 9, 126, 133, 135, 136,
206, 209, 211, 212, 213, 214, 137, 139, 140, 141, 142, 143,
215, 216, 217, 218, 219, 220, 144, 145, 146, 147, 148, 149,
221, 222, 223, 224, 225, 226, 151, 153, 154, 156, 157, 158,
227, 228, 229, 230, 231, 232, 159, 160, 162, 163, 357, 365,
234, 235, 236, 237, 238, 239, 366, 367
241, 242, 244, 245, 246, 247, yiqtol, 169, 172, 192, 194, 196,
248, 249, 251, 254, 258, 259, 227, 236, 306
260,261 zero present, 177, 178
VprefS, 9, 171, 174, 175, 178, zero-gram, 178, 210
180, 182, 183, 185, 186, 187, Zustandssätze, 59, 129, 366, 368
188, 189, 190, 191, 192, 193,


Bo Isaksson - Maria Persson - Clause Combining in Semitic - The Circumstantial Clause and Beyond-Harrassowitz (2015)

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

Bo Isaksson - Maria Persson - Clause Combining in Semitic - The Circumstantial Clause and Beyond-Harrassowitz (2015)

Hochgeladen von

Copyright:

Verfügbare Formate

Abhandlungen für die Kunde

Clause Combining in Semitic:

Deutsche Morgenländische Gesellschaft

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

Bibliographic information published by the Deutsche Nationalbibliothek

For further information about our publishing program consult our

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

Clause Combining in Arabic dialects

Clause Combining in Written Arabic

Clause Combining in Biblical Hebrew

Clause Combining in Modern Spoken Aramaic

Clause Combining in Epigraphic South Arabian

Clause Combining in East Semitic

Index of terms ............................................................................................. 407

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

5 Thus M. Waltisberg, Satzkomplex und Funktion: Syndese und Asyndese im Althochara-

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

Uppsala and Lund in February 2015.

Bo Isaksson and Maria Persson

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

Heléne Kammensjö, Uppsala University

The clause complex is of particular interest in spoken language, be-

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

In their seminal article, dedicated to the issue of hypotaxis as a discourse

The function of tense forms in narrative is frequently not the basic

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

established a priori as past. Rather, tense contrasts may be pressed into

1.2 Data and transcription

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

1.3 Relevant features of spoken Egyptian Arabic

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

1.4 Notes on the EA participle and the prefix forms

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

‘the engine used to be running at night’

Schlonsky (1997, 25-56) makes a similar point concerning the verbicization of

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

Lexical aspect and the participle

Chart 1: Brustad’s simplified model (2000, 171)

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

Unmarked and marked prefix forms

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

Woidich describes the meaning of the bi-form of the prefix conjugation

2. Types of circumstantial clause linking

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

pronoun are generally rather frequent in EA, e.g. in non-narrative passages

2.2 Types of syndetic clauses

2.2.1 The prefix form in the circumstantial clause

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

(9) laGenāhum [wu huṃṃa biyliffu sagāyir ē? wu byišṛábu ḥašīš] (ÄAD

2.2.2 Participle or nominal clause as circumstantial clause

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

clause, as in (14), in which case it must be interpreted as operating at the

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden

The circumstantial clause in (16) has a causal interpretation, supported by the

2.2.3 The suffix form in the circumstantial clause

2.2.4 Unusual head types in syndetic circumstantial clauses

© 2015, Otto Harrassowitz GmbH & Co. KG, Wiesbaden