(Current Issues in Linguistic Theory 87) Bela Brogyanyi, Reiner Lipp - Historical Philology_ Greek, Latin, and Romance. Papers in honor of Oswald Szemerényi II-John Benjamins Publishing Company (1992)

HISTORICAL PHILOLOGY: GREEK, LATIN, AND ROMANCE
AM001STERDAM STUDIES IN THE THEORY AND

HISTORY OF LINGUISTIC SCIENCE
General Editor
E.F. KONRAD KOERNER
(University of Ottawa)
Series IV - CURRENT ISSUES IN LINGUISTIC THEORY
Advisory Editorial Board
Henning Andersen (Los Angeles) ; Raimo Anttila (Los Angeles)

Thomas V. Gamkrelidze (Tbilisi); John E. Joseph (College Park, Md.)
Hans-Heinrich Lieb (Berlin); Ernst Pulgram (Ann Arbor, Mich.)
E. Wyn Roberts (Vancouver, B.C.); Danny Steinberg (Tokyo)
Volume 87
Bela Brogyanyi and Reiner Lipp (eds)
Historical Philology: Greek, Latin, and Romance

HISTORICAL
PHILOLOGY
GREEK, LATIN, AND ROMANCE
PAPERS IN HONOR OF OSWALD SZEMERÉNYI II
Edited by
BELA BROGYANYI and REINER LIPP

University of Freiburg
JOHN BENJAMINS PUBLISHING COMPANY

AMSTERDAM/PHILADELPHIA
1992
Library of Congress Cataloging-in-Publication Data
Historical philology : Greek, Latin, and romance : papers in honor of Oswald

Szemerényi II / edited by Bela Brogyanyi and Reiner Lipp.
p. cm. ~ (Amsterdam studies in the theory and history of linguistic science.
Series IV, Current issues in linguistic theory, ISSN 0304-0763; v. 87)
Includes bibliographical references and index.
1. Classical philology. 2. Romance philology. I. Szereményi, Oswald John Louis,
1913- . II. Brogyanyi, Bela. III. Lipp, Reiner. IV. Series.
PA26.S96H57 1992
480--dc20 92-20724
ISBN 90 272 3586 4 (Eur.)/l-55619-144-8 (US)(alk. paper) CIP
© Copyright 1992 - John Benjamins B.V.
No part of this book may be reproduced in any form, by print, photoprint, microfilm, or
any other means, without written permission from the publisher.
John Benjamins Publishing Co. • P.O. Box 75577 • 1070 AN Amsterdam • The Netherlands
John Benjamins North America ■ 821 Bethlehem Pike • Philadelphia, PA 19118 • USA
SYMBOLAE
OSWALDO SZEMERÉNYI
SEPTVAGINTA QVINQVE ANNOS NATO
GRATO ANIMO
OBLATAE

PREFACE
The Editors take great pleasure in submitting to the linguistic community this second
of three thematically distinct volumes dedicated to Professor Oswald Szemerényi on the
occassion of his 75th birthday. The first volume is entitled "Prehistory, History, and
Historiography of Language, Speech, and Linguistic Theory"; the last volume,
"Comparative-Historical Linguistics: Indo-European and Finno-Ugric".
The present volume mainly contains contributions on the classical languages, Greek
and Latin. In addition to the historical comparative linguistic aspects of these languages,
philological and historical questions are dealt with as well. Consideration of Italic and
Romance topics is also included.
The Editors would like to thank all of the contributors for their patience and
confidence in the project. The deadline for submission of articles was the end of
September 1988; it was therefore as a rule not possible for authors to modify their
contributions or expand their references subsequent to this point.
The word-processing was executed by the company of Joachim Haamann, Aachen,
at the request of our publisher. We are indebted to Mrs. Astrid Haamann for her
dedication and enthusiasm throughout the typesetting procedure.
Heartfelt thanks are due to Mrs. Claire and Mr. John Benjamins for including these
volumes in their publishing program and for generously accepting the costs associated
with word-processing. We would also like to thank Professor E.F.K. Koerner
(University of Ottawa), editor of the series, for his all-round support during the
realization of the project
The Editors of course retain responsibility for all errors and shortcomings.
Freiburg i. Br., September 21, 1991
Bela Brogyanyi and Reiner Lipp

TABLE OF CONTENTS
DEDICATION V
PREFACE VII
TABLE OF CONTENTS IX
I. GREEK LINGUISTICS
1. Francisco R. ADRADOS
Norma y normas en la sintaxis de los casos en griego antiguo 3
2. Riccardo AMBROSINI
Interpretive constraints versus ad sensum anaphora in Classical Greek 15
3. George BABINIOTIS
The question of mediae in Ancient Macedonian Greek reconsidered , 29
4. Alfred AMMESBERGER
Griechisch thés, hés und dós 41
5. James T. HOOKER
Some uses of the Greek imperfect 47
6. Witold MAŃCZAK
Les désinences de grec (φέpέις et (φέpει 67
7. Klaus STRUNK
Greek 1-όλς and related phenomena 77
8. William F.WYATT,Jr.
Aeolic dative plurals in -essi 85
II. GREEK LEXICOLOGY

1. John CHADWICK
HPΥΣ- a Greek ghost-word 99
X
2. Pietro JANNI
On the treatment of some Greek prepositions in modern
language dictionaries 103
3. Giulio LEPSCHY
Fantasia in pantofole 113
III. MYCENOLOGY
1. Mario DORIA
Schiavi in vendita nella Cnosso micenea 127
2. Jean-Pierre OLIVIER
"Cinq" en linéaire A? 135
3. Oswald PANAGL
Mykenische Fossilien im Homertext?
Zur Deutung von άρματοττηγός und άρματροχιή 137
IV. GREEK PHILOLOGY

1. Hans-Christian GÜNTHER
Some remarks on textual problems in Euripides' Iphigeneia in Aulis 147
2. Gregory NAGY
Metrical convergences and divergences in early Greek poetry and songs 151
3. Michael REICHEL
Gräzistische Bemerkungen zur Struktur des Gilgamesch-Epos 187
V. ITALIC AND LATIN PHILOLOGY

1. Patrick CONSIDINE
Some remarks on the Carmen Arvale 211
2. Mario NEGRI
La lingua di Numa 229
3. Maria Luisa PORZIO GERNIA
Tra storia e preistoria linguistica:  valore concettuale della
radice *pāk-nell'Italia antica 267
4. Bernfried SCHLERATH
Zu Isidor VIII, III, 6 293
XI
VI. LATIN AND ROMANCE LANGUAGES

1. Wolfgang RAIBLE
The pitfalls of subordination: Subject and object clauses
between Latin and Romance 299
2. Rudolf WINDISCH
Zwischen Substratomanie und Substratophobie: Die Grenzen eines
sprachwissenschaftlichen Erklärungsprinzips am Beispiel des
Lautwandels von lateinisch  zu französisch ü 339
VIL ROMAN HISTORY

1. Endre FERENCZY
Über das römische Königtum 355
INDEX OF AUTHORS 377

I.
GREEK LINGUISTICS
NORMA Y NORMAS EN LA SINTAXIS DE LOS
CASOS EN GRIEGO ANTIGUO
FRANCISCO R. ADRADOS
Consejo Superior de Investigaciones Científicas, Madrid
Las páginas que siguen parten de una concepción estructural de los casos griegos,
pero de una concepción estructural compleja, que pretende ajustarse a los hechos. Y de
una descripción pancrónica, en su fundamento. Pero esto no obsta para que en una se
gunda fase se descienda al estudio de hechos especiales, diferenciales, en los "dialectos"
en que se organiza la lengua: dialectos diacrónicos, geográficos, sociales o estilísticos o
subdivisiones de estos o bien combinaciones de los mismos. Estos hechos especiales,
que calificamos de "norma" o "normas" dentro del sistema general pancrónico, son preci
samente los que van a atraer nuestra atención en el presente estudio.
Nótese, pues, que nos alejamos conscientemente de dos tipos de descripción sintác
tica: en uno y otro estaría fuera de lugar un estudio como éste. El primero es el de la
Sintaxis histórico-filológica que hemos calificado en ocasiones de "Sintaxis de etiquetas".
Se limita a clasificar de un modo más bien empírico, sin atención a hechos distribuciona-
les u opositivos (por lo menos conscientemente), una serie de usos, sin dar relieve a unos
por encima de otros ni tratar de relacionarlos de algún modo. Mezcla además sincronía y
diacronía y opera con significaciones generales que considera heredadas, como las de
"Genitivo propio", Partitivo y Ablativo dentro del Genitivo y "Dativo propio", Instru
mental y Locativo dentro del Dativo. Las nociones de sistema, norma y uso son ajenas a
este confuso conglomerado que, sin embargo y a pesar de todo, tiene la ventaja de que en
4 FRANCISCO R. ADRADOS
exposiciones de este tipo encontramos colecciones de datos que, convenientemente cla

sificados y organizados, pueden ser útiles para la descripción de la Sintaxis del griego en
general y para la de los diferentes "dialectos".
El segundo tipo de descripción que conscientemente rehuimos, es el que pudiéramos
llamar, para distinguirlo del otro, el de la Sintaxis abstracta y universalista. Arranca de
definiciones previas - que pueden ser muy diferentes entre sí - de lo que "es" cada caso,
como un a priori universal que en ciertas versiones se entiende como perteneciente a la
estructura profunda; en otras, parece, a la parte de la lengua directamente accesible, aun
que en este caso se comprenden mal las diferencias que, real y verdaderamente, existen
entre los casos de las diferentes lenguas. Se comprenderá que, vistas así las cosas, o no
existe hueco para estudios como el nuestro o en todo caso hay que relegarlos a la "super
ficie", con un método indeterminado.
Nuestro, método, estructural pero inductivo, nos lleva, en cambio, a considerar pri
mero cada caso aisladamente, en sus diversas distribuciones y subdistribuciones, en las
que con ayuda de oposiciones se generan definiciones sintácticas que luego se degradan y
semantizan en el detalle; se difunden a veces, con estos sentidos ya secundarios, a otras
distribuciones; o se neutralizan, también. A través de este complejo panorama, que no ex
cluye una noción de unidad del caso a partir de los usos centrales, surge el sistema total.
Pero ese sistema, que quiere ser fiel a la totalidad de los hechos a un nivel
pancrónico, es un sistema bastante abierto y generalizante, que solo el uso concreta en el
detalle. Por ejemplo, el Genitivo indica todas las posibles relaciones de nombre a
nombre, salvo aquellas que se han atribuido a las marcadas con otros determinantes (el
adjetivo, el Ac, el D., también adverbios y giros preposicionales); aunque no se
excluyen ciertas neutralizaciones, relaciones marcadas ya por el G. ya de otra manera.
Pues bien, en la práctica algunas de esas determinaciones no son marcadas por el G. o
apenas lo son: así, el G. de cualidad, frecuente en Latín, lo es muy poco en época clásica
en Griego, salvo en contextos numerales y espaciales muy restringidos.
O véase lo que sucede con el G. dependiente de un verbo, uso bastante frecuente con
ciertos grupos de ellos, sin duda por un proceso de transformación (con inercia que no
cambia el caso en un Ac, como otras veces). Puede verse que, por ejemplo, los verbos
que significan "coger", "tocar", "desear", "amar", entre otros, llevan G., que nosotros no
creemos que sea un partitivo, como habitualmente se dice: es un simple determinante.
LA SINTAXIS DE LOS CASOS EN GRIEGO ANTIGUO 5
Pues bien, verbos como φιλέω, στέργω, ποϑέω de este grupo semántico, llevan siempre
Ac. Ciertamente, también esto era posible dentro del sistema, el Ac. es el determinante
general del verbo. Pero hay un reparto que sincrónicamente al menos es arbitrario: en un
cierto punto no funcionó, simplemente, el sistema de las transformaciones.
O nótese todavía lo que sucede con los D. que llamamos comitativos, que indican
compañia con un verbo de este significado. Su valor es completamente admisible en
sistema para el D.: "en", "junto a", "al lado de", etc. son valores antiguos y habituales de
este caso. Pero solo se da con verbos que formalmente están marcados como de com
pañía: verbos con όμο-, KOΙVO-, etc. o con preverbios como συν-, μετα- y otros
comparables. Y sin embargo, hay excepciones: los verbos de movimiento y otros llevan
un comitativo en contextos militares (sujeto del jefe, D. de las tropas); o bien acom
pañando al D. αύτοίς, etc. Son "normas" dentro de posibilidades del sistema que quedan
sin realizar en gran parte. Evidentemente, había el riesgo de ambigüedad con otros tipos
de D., salvo cuando bien el verbo bien el nombre bien el contexto suministraban datos
para la desambiguación.
Pues bien, dentro de las posibilidades que el sistema ofrece, la "norma" que se crea
se solidifica dentro de uno o varios "dialectos", en el sentido amplio dado arriba a esta
palabra. A veces, ciertamente, en todos o en una mayoría de los textos a partir de un
momento dado: véase lo que decimos más abajo en nuestro apartado 1. Pero otras veces
no: o bien por las razones que sea hay una norma común a varios dialectos, incluso sin
conexión cronológica o geográfica directa, o bien la norma de uno es eliminada por otro
más reciente o éste crea una nueva, antes desconocida, pero admitida dentro del sistema
abstracto.
En este breve estudio vamos a considerar fundamentalmente tres dialectos, que com
parten por supuesto normas comunes del Griego y en cuanto a otras especiales ya coinci
den ya no. El primero es el que llamamos dialecto poético, sin entrar en sus orígenes ni
en los factores que lo unifican ni en sus rasgos internos diferenciales: el constituido por
los textos épicos, sobre todo Homero, por la lírica y la tragedia. El segundo es el de la
prosa ática a partir del último tercio del s. V y que continúa en el IV hasta, digamos, por
poner un limite convencional, la muerte de Alejandro en el 323 .  El tercero es la koiné
popular desde el s. III .  al I d. C, definida por textos como la versión de los LXX,
muchos papiros y el N.T., entre otros.
Claro que añadimos ocasionalmente referencias a otros "dialectos", sobre todo a la

lengua ática coloquial recogida por la Comedia y a la lengua de textos documentales como
son algunos tratados hipocráticos (Epidemias, etc.) y ciertas inscripciones y papiros.
Este trabajo contiene solamente algunos ejemplos de algunas "normas" de mayor o
menor difusión; algunas abarcan a casi todo el Griego, otras a más de un dialecto, otras a
uno solo. Esta es la triple división en que organizamos el material, expuesto abreviada
mente y por supuesto ampliable.
1. Tendencia a generalizar una norma en todo el Griego

Pancrónicamente diremos que no es una norma pangriega, pero sí general; a Veces,
obligatoria ya en ciertos dialectos.
En Griego se suele en términos generales usar los verbos de "decir" con Ac. de la
cosa dicha y D. de la persona a quien se dice: λέγειν TLΤΙΝΊ.Pues bien, este giro, que
por lo demás se encuentra ya en Homero, no es más que el resultado de una regulari-
zación. En Homero se usa todavía, a, veces, el Ac. para indicar la persona de la cual se
habla (IL 1.90 ούδ' ἡν 'Αγαμέμνονα εϊπης) y aquella a la cual se habla (77. 13. 725
Πουλυδάμας ϑρασύν "Έκτορα είπε). Esto prueba que la distinción entre el comple
mento directo y el indirecto solo gradualmente se estableció en estos verbos: en el sistema
más antiguo el Ac. se usaba con toda clase de determinaciones.
No muy diferente es lo sucedido con verbos de "oir" como άκουω la construcción
habitual con Ac. de cosa y G. de persona (de aquel a quien "se oye" algo) es secundaria.
El uso del G. es en principio más amplio: se dice por ejemplo Od. 4.114 πατρός άκουσα
Oyendo acerca del padre', lo que luego se diría περl τοū πατρός. Y se dice también,
con G. del sonido oído, IL 21 Ml κωκυτοū δ ' ήκουσε. O sea: estos verbos llevaban en
G. cualquier determinación, en correspondencia con los nombres verbales que también
eran determinados en G. con varios matices.
También vamos a aludir a otra especialización semejante: con los verbos de "coger",
"tocar", el Ac. significa el todo, el G. la parte, cf. por ej. ya en Homero II. 1.197 ξανϑῆς
δέ κόμης ἕλε Πηλειωνα cogió a Aquiles de la cabellera'. Ahora bien, todavía en
Arquíloco se usa el Ac para la parte (204 χειρα Νεοβουλης ϑιγέΐν 'coger la mano de
Neobula') y el G. para el todo y la parte (24 τέττιγος- έδράξω πτερού has cogido a una
cigarra del ala'). Es bien claro que en sistema no había G. de la parte; había un .  un
G. (según los verbos) del todo y de la parte, la especialización, la norma, es secundaria,

aunque general. Y tiene lugar con ciertos verbos, con otros hay Ac. del todo y de la parte
(77. 24.58 γυναίκα  ϑήσατο μαστόν, etc.).
Añádase, entre otros ejemplos que podrían ponerse, la progresiva extensión del Ac.
como caso del régimen de los verbos, ya aludida más arriba, y que culminó a partir de la
época helenística. En realidad, el Ac. era desde antiguo el determinante verbal, de tipo ge
neral, por excelencia; el G. y el D. le discutieron en un momento dado parte del terreno.
Pero hubo, primero, una especialización verbo a verbo; luego, una difusión máxima del
Ac, hasta desaparecer el D. así como el G. adverbal. No sucedía otra cosa que la imposi
ción sin excepciones en la norma de lo que desde el comienzo estaba en el sistema.
2. Norma que surge en varios "dialectos"

Es muy notable lo que sucede en las oraciones nominales tanto bimembres como uni-
membres. Comenzando por las primeras, a veces aparecen sin construcciones articulares
que las distingan de los grupos atributivos, como en los modelos clásicos o παῖς· κάλος,
καλός- ό παῖς- 'el niño es guapo' frente a ό καλός παῖς-, ό παῖς- ο κάλος- ' el niño
guapo'. Se trata de oraciones nominales sin verbo copulativo, que solo contextualmente
se distinguen de los mencionados grupos atributivos. Pues bien, este uso, que es
absolutamente arcaico y tiende a ser sustituido por la nueva norma, se mantiene en ciertos
textos de poesía, de la lengua popular y del Corpus Hippocraticum. Nos referimos a fra
ses como 77. 10. 551 ἕγγυϑι δ 'ηώς- 'la aurora está cerca', Ar. Ach. άλλ ' οú σχολή
'no hay tiempo', Ηρ. Epid. 3.222 εμετοί πολλοί 'los vómitos fueron numerosos'. Una
cosa semejante sucede con las oraciones unimembres de igual tipo: Od. 1.511 νήσος"
δενδρήεσσα '(hay) una isla boscosa', Hp. Epid. 3.222, έs νύκτα ίδρώς- '(hubo) sudor
de noche', Sokolowski A 58"Hpaι.. δαμαλις- κριτα '(fue llevada) a Hera una novilla
escogida'.
Los mismos textos nos ofrecen, todavía, ejemplos del uso impresivo-expresivo, de
tipo del V., de una serie de N. de nombres y adjetivos. En esta función encontramos el
N. del nombre sin ώ en//. 13,85, etc. αιδώς- 'qué vergüenza!', A. Fr. 207 τράγος· 'ca
brón!'; también en Atico conversacional, cf. Ar. Ra. 652 άνϑρωπος- ιερός- 'un hombre
piadoso!' en koiné: Eu. Io. 17.21 πατήρ 'padre!'.
Nótese que el uso asintáctico, tanto representativo como impresivo-expresivo, del

N., era algo dado desde antiguo: solo muy limitadamente el V. tenía una forma
independiente, se añadía que el N. contiene todavía muchos temas puros. En ciertos
"dialectos", como se ve, se conservó esta posibilidad, mientras que en otros quedó
eliminada. Se impusieron las oraciones nominales con verbo "ser" y construcciones
articulares; y el V. con forma especial. Dos normas diferentes han coexistido, repar
tiéndose los "dialectos". Si las que comentamos aparecen en textos de fechas y estilos
muy diferentes, es porque en ellos rebrota un fondo antiguo, popular, que nunca fue
completamente eliminado, aunque ciertos "dialectos" literarios lo rehuían.
Otras veces se trata solamente de dos "dialectos", el poético arcaico y la koiné, muy
alejados temporalmente y, al menos a primera vista, en el estilo. Las coincidencias se ex
plican de varias maneras, a veces puede decirse que hay una coincidencia por razones ac
cidentales y complejas.
Es notable, por ejemplo, lo que sucede con los giros con V. precedido de la interjec
ción ώ. Es cierto que existe hoy una abundante bibliografía sobre ώ y que no siempre hay
coincidencia en la interpretación de su uso: a veces parece evidente que son razones
métricas o de evitación del hiato las que lo condicionan. Sin embargo, en términos gene
rales se reconoce que los más de los ejemplos homéricos usan ώ con un énfasis especial,
en contextos expresivos y familiares. No lo emplean los mortales ni los dioses cuando se
dirigen a los dioses, salvo cuando muestran impaciencia o violencia: Il. 8.31 ώ πάτερ
ημέτερε Κρονίδη 'padre nuestro hijo de Crono' (Atena a Zeus). Ni entre sí en las escenas
nobles, solo cuando son familiares o sube el tono emocional: Il. 1.158 ώ μεγ ' άναιδες
'gran sinvergüenza' (Aquiles a Agamenón). También para la tragedia suele presentarse
una interpretación semejante, normalmente ώ aparece cuando el V. está determinado
adjetivalmente, incluso dirigiéndose a los dioses: S. Tr. 303 ώ Ζεữ τροπαΐε. Pues bien,
es sabido que no es éste el uso habitual en la prosa ática, en la que ώ se ha difundido
enormemente y ha perdido valor de información; en realidad con los nombres propios lo
expresivo es el uso sin ώ, tipo Σώκρατες. Pues bien, en la koiné desciende estadística
mente la frecuencia de ώ y sube, correlativamente, su valor expresivo: cf. Eu. Matt.
15.28 ώ γύναι, μεγάλη σου ή πίστις 'oh mujer, grande es tu fe'.
Aqui hay coincidencia: el valor original de la exclamation ώ ha sido redescubierto tras
una pérdida del mismo.
Otros "rebrotes" no son exactamente idénticos, pero sí comparables. Uno es el del G.

apositivo, que es poético en frases como Od. 1.2, Τροίης ίερόν πτολίεϑρον a sagrada
ciudad ciudad de Troya', Il. 5. 781 βίη Διομήδεος 'la fuerza de Diomedes' o 'el fuerte
Diomedes', S. Tr. 507 ποταμοῦ σϋένος 'el fuerte río', etc. Estas construcciones
desaparecen, salvo algunos usos muy especiales. Pues bien, en koiné encontramos otras
que no son exactamente idénticas, pero que desarrollan la misma posibilidad de que el G.
determinante se convierta en una mera aposición: 2Cor. 5.5 τòν àρραβώνα τοῦ
πνεύματος la prenda que consiste en el espíritu', PEleph. 5.21 άπαντα τῶν πατρώων
'toda mi herencia paterna'.
A veces podemos comprender la razón de algunas de estas "resurrecciones" que, de
otra parte, como hemos dicho, quedan dentro de las posibilidades admitidas por el sis
tema. Por ejemplo, en poesía arcaica a partir del G. partitivo con superlativos del tipo de
Ar. Ach. 336 τῶν φίλων ΤΟΎΣ φιλάτους 'a los más amigos de los amigos', se ha
creado el llamado G. de encarecimiento en frases como A. Pers. 666 δέσποτα δεσποτãν
'señor de señores', S.  465 ἄρρητ ' άρρητων 'cosas indecibles entre las indecibles'.
Fuerzan un poco, podría decirse, los límites del sistema. Pues bien, tras una desaparición
de estas expresiones a lo largo de los siglos, encontramos en el Griego de los LXX, del
N.T. y de los cristianos expresiones comparables como єis ΤΟΎΣ αιώνας τών αιώνων,
'por los siglos de los siglos', ąσμα ąσμάτων 'el Cantar de los Cantares', ματαιότης
ματαιοτήτων 'vanidad de vanidades'. Aqui el modelo de la recreación de algo que
siempre era posible dentro del sistema griego estaba en expresiones paralelas hebreas y
arameas.
Podrían aducirse muchos ejemplos más, con posibilidad de interpretaciones varias de
la coincidencia entre fechas y estilos diferentes: aunque en general se trata sin más de ex
plotaciones también varias de una misma posibilidad que ofrecía el sistema. Así cuando
tanto en la lengua poética como en la koiné hallamos en ocasiones el Ac. con verbos que
son usualmente intransitivos: en poesía con verbos como δακρύω, χορεύω, ϋειν, en
koiné con otros como ένεργέω, καρτερέω, ϑέλω. O encontramos una determinación del
nombre de persona por nombres plurales también de persona: es el tipo tan citado Il.
11.761 ϑεών ΔÍÍ, Νεστορι δ ' ανδρών 'de entre los dioses a Zeus, de entre los
hombres a Nestor', que encuentra algunos paralelos en prosa ática, pero sobre todo otros
más frecuentes en koiné, así Plb. 22.14 (18).7 των φίλων 'Απελλῇ 'a Apeles de entre
sus amigos', PTeb. 33 Λεύκιος Μέμμιο ς ' Ρωμαίο ς τῶν άπò συγκλήτου Lucio
Memmio romano de los del senado'.
Naturalmente, la distribución dialectal puede variar. Por ejemplo, el Ac. asintáctico se
encuentra en el Corpus Hippocraticum, las inscripciones, la koiné: como en otros
ejemplos puestos más arriba es una posibilidad general que solamente en estos géneros,
menos formalizados y no sujetos a otra norma, se ha abierto camino.
3. Norma que es propia de solo un dialecto

Muy notable es el caso de aquellas construcciones o usos que, perteneciendo al
sistema general de la lengua, solamente aparecen o aparecen muy principalmente en uno
de los "dialectos" de la misma. Diacrσnicamente hablando, puede tratarse, unas veces, de
un arcaismo conservado y luego perdido, otras de una creación dentro de los límites
permitidos por el sistema. A efectos sincrónicos, ambos casos son el mismo.
Veamos algunos ejemplos procedentes del dominio de la poesía arcaica. Uno de ellos
puede ser la ausencia del sujeto allí donde es innecesario por ser equivalente a un τις 'al
guien', 'uno cualquiera'. Es el tipo, luego desaparecido, Il. 22. 199 ώς δ ' έv όυεάρῳ
ού δύναται φευγοντα διώκειν ' como (uno) en un sueño no es capaz de alcanzar al que
persigue'. De haber ejemplos comparables, se hallan en la lengua documental, en que no
han penetrado regularizaciones posteriores que exigen el uso del sujeto.
El D. interno etimológico es tan normal en el sistema como el Ac. interno de igual
tipo; pero a diferencia de éste, solo en poesía se ha difundido; se emplea habitualmente
con un adjetivo para dar relieve a la idea verbal: tipo Od. 3.87 άπώλετο λυγρῷ όλεϑρῳ
Lo mismo sucede con el Ac. de reccón con nombres y adjetivos, que es una transfor
mación normal del adverbial: tipo A. Ch. 23 χοἀς προπομπός, Portadora de libaciones',
por lo demás con cierta difusión. Y con el Ac. expresivo-impresivo, tipo Alc. 10 €με
δείλαν 'desgraciada de mi'. Otras veces, simplemente, la poesía muestra una mayor
frecuencia de ciertos usos, así del D. simpatètico, lo que se relaciona con el contenido de
sus textos.
Muy característico es lo que sucede con ciertos usos locales de los casos que solo en
poesía se mantienen; allí alternan con usos en que van precedidos de preposición, pero
estos son los que luego se imponen casi sin excepción. Se trata del Ac. lativo o de
"dirección hacia", del D. de lugar "en" o reposo y del G. de lugar "desde". Hay que dis-
tinguir éste del separativo en general (con verbos de distanciamiento y origen) que tiene
una mayor difusión; y hay que añadir el uso raro, solo homérico, del G. de lugar en raras
expresiones como II. 6.508 'lavarse en el río'.
No vamos a entrar aquí en la génesis de estos usos casuales, solamente decimos que
son arcaicos. A veces están en relación de transformación con usos adnominales, otras
son fijaciones de antiguos usos asintácticos, otras aún forman parte de un uso más exten
dido de determinación local, así en el caso del G. separativo. Los tres usos principales
llevan un nombre de lugar (de series reducidas de nombres de lugar) o de partes del
cuerpo, a excepción de nombres de persona en pl. en el caso del D. de reposo, en Ho
mero (Il. 1. 68 ); los tres forman un sistema en que se oponen
los tres casos mencionados con valor local; los tres son exclusivos de la poesía, con
excepción de algunos usos fosilizados en Ático y otros dialectos de etc.
Se trata, probablemente, de la diferencia más notable entre la lengua poética y los varios
tipos de prosa que la siguieron, en que las relaciones locales fueron expresadas
exclusivamente con preposiciones en buena medida ayudadas por los mismos casos.
Hubo evolución drástica del sistema en Griego helenístico.
Si pasamos a otro dialecto muy característico e importante, la prosa de la koiné popu
lar, vemos que en él se encuentran realizadas algunas posibilidades del sistema de los ca
sos griegos. A veces se trata de la creación o difusión de un uso posible dentro del si
stema; otras, de la eliminación de otro.
Dentro del primer apartado podemos mencionar el N. asintáctico llamado deno
minativo, equivalente, por decirlo así, a una cita entre comillas. Se cita la frase conocida
del Evangelio: Eu. Io.. 13.13 'llamadme "maestro" '. Hay
luego una difusión de un uso anterior como es el D. de interés de nombres de coas, deter
minando a un nombre. Construido sobre el modelo del de persona, aparece raramente
desde el s. V a. C. (A. Pers. 1022 'un depósito para las flechas',
IG2  1672.68 'clavos para las puertas'); pero es especialmente
importante en koiné, cf. por ej. PTeb. 60 ss, etc. 'semilla
de trigo, lentejas, hierba'. También merece la pena anotar, entre otras cosas, el desarrollo
del G. de cualidad con nombre que indica clase o calidad y adjetivo: uso muy raro
anteriormente y propio ahora sobre todo de la lengua intelectual (con ,
', etc.). Como se ve, hay que hacer distinciones entres los niveles de la
koiné.
Otras veces, en cambio, la koiné lo que hace es acelerar la desaparición de ciertos
usos casuales. Por ej., el Ac. de relación tiende a ser sustituido por un D. de este tipo, el
Ac. de tiempo igualmente por un D.; ya hemos dicho que el Ac. se impone, en cambio, o
tiende a imponerse como caso único de la rección verbal, a expensas del D. y el G. Por
otra parte, disminuye muchísimo el uso del D. instrumental y del de causa, que tienden a
ser sustituidos cada vez más por giros preposicionales, por lo demás existentes desde
antiguo.
Todo esto responde a una breve ejemplificación de cómo las posibilidades de un
sistema tan complejo como es el de los casos griegos, en que alternan usos centrales y
marginales, hay acepciones definidas distribucional, opositiva y semánticamente, hay
también neutralizaciones, el uso no aprovecha todas las posibilidades disponibles ni
mucho menos. Algunas solo poco a poco se abren paso en lucha con otras o, al contrario,
desaparecen tras un período de florecimiento: a veces definitivamente, a veces para re
brotar más tarde aprovechando circunstancias favorables. Otras posibilidades, en cambio,
se realizan inesperadamente en un momento dado. Claro está, nuestra visión no puede ser
sino parcial: una lengua como el Griego antiguo nos es conocida muy imperfectamente:
solo indirecta y parcialmente reconstruimos el uso oral en sus diversos niveles, los dia
lectos geográficos nos son también solo en parte conocidos. Nos es familiar, hasta cierto
punto, la superestructura que son los dialectos literarios, pero mucho peor la base en que
se apoyan y que depuran y modifican.
En todo caso, podemos ver, junto a una norma general del Griego dentro del sistema
del mismo, normas particulares, ya en conflicto, ya sustituyéndose, ya apareciendo o
desapareciendo. Pueden ser propias de un dialecto, de varios o de la generalidad del
Griego, pero no de todo él. Su combinación crea los "dialectos" de que hemos hablado.
Naturalmente, afinando más (aunque nuestros materiales-son más bien insuficientes)
podría llegarse a estudiar cómo la norma se degrada en el uso, es decir, es seguida más o
menos íntegramente y, a veces, alterada o suprimida. Cómo, de otra parte, los "dialectos"
de que hemos hablado la siguen en forma diferente según las edades, los géneros lite
rarios, niveles, etc. Cómo se interfieren gradaciones de frecuencia. Cómo las distintas
normas se combinan variamente unas con otras, según los "dialectos" y los componentes
de los mismos.
En realidad la noción de norma, como un intermedio entre la de sistema y uso (o len
gua y habla, si se quiere), no es más que una abstracción, como abstracciones son por lo
demás en defínitiva el sistema y el uso. Hay toda clase de gradaciones: por eso hemos
hablado, en el título de este trabajo, de "norma y normas". Aun así, la triple división es
un instrumento intelectual que resulta útil para el análisis. Hemos intentado probar esta
validez con un ejemplo concreto relativo al Griego antiguo, ejemplificando con algunos
de los datos y de las ideas que manejamos en un libro en preparación cuyo título es
"Sintaxis estructural del griego antiguoa".
SUMMARY
Not only the concept of system, but also that of norm should be used when studying the Greek
nominal cases. Within a fundamentally unitarian system there are three main dialectal variants which
adopt some partially different norms: that of the poetic language, that of the attic prose and the "koiné".
Some other norms tend to be generalized in all dialects.
INTERPRETIVE CONSTRAINTS VERSUS AD SENSUM
ANAPHORA IN CLASSICAL GREEK
RICCARDO AMBROSINI
University of Pisa
0. In many languages the linear ordering of formal elements can be seen not to coincide
with their progressive informative value. Take for instance the case of German, where the
lengthy syntagm die in der Nähe des Schwarzwaldes liegende schöne Stadt Freiburg
provides a clear illustration of such a discrepancy between the linear sequence of formal
elements and the ordering of referents to be communicated, i.e. their informative
contiguity. In the context of those classical languages most studied by O. Szemerényi (to
whom I take pleasure in dedicating these pages) Classical Greek will represent an all too
familiar example, and indeed the second part of this essay, when taken together with the
observations in the first part, forces us to recognise the existence of implicit organising
mechanisms at work both within and between syntagms in a given sentence. What
follows by way of these reflections will (I hope) be of some interest to the distinguished
scholar to whom they are dedicated.
1. In both our German example and in the syntactically comparable phrase

(coined ad hoc) the formal elements are ordered
differently from, say, in the case of Italian, where a possible analogon would instead be
marked by a left-to-right sequence, viz. la bella città di Friburgo, sita nelle vicinanze della
Selva Nera, in which the latter part qualifies (or better, viewed in terms of a scale of
16 RICCARDO AMBROSINI
cultural and expressional rather than numerical values, quantifies) the former so that any
following information will be syntactically unpredictable on the basis of the structure of
syntagm itself. I should emphasise the expression "syntactically unpredictable infor
mation" for, as I see it, the sole function of the structure of the two lengthy syntagms in
German and Greek is to anticipate and preorganise a relationship, erecting, as it were, the
first pillar of a bridge in the certainty that at the other end of the subsequent flow of
information a second similar pillar will duly be erected. The German die and Greek f)
create expectations which will be satisfied by the connection linking the closing elements
to the opening ones. Or more specifically, the configuration of the closing element in the
German syntagm, Stadt Freiburg, syntactically equivalent to the Latin urbs Roma (where
Greek would have preferred an ethnic genitive in the attributive position)1 might also lead
to the hypothesis that the pivotal element here is Freiburg insofar as it represents the
essential N (or Noun) category in the syntagm as a whole.
The question we shall now raise concerns not so much what historical reasons or
stages led to the establishment of the two different structures, but rather which functions
might be associated with the different typologies that have come to caracterise German
and Classical Greek on the one hand, and Italian on the other. (It may be assumed that
these languages are valid and sufficient representatives of the two distinct typologies
which, however, require not only to be described, but also to be explained). This
difficulty might obviously be overcome by claiming that the two constructions fulfil
essentially the same expressive function, thus casting doubts on the very theoretical
legitimacy of our evaluative question. In what follows we shall attempt to argue the
contrary.
By invoking a (theoretically not unsound) interpretive measure such as the so-called
'X-bar theory' the initial category {Freiburg) may be expanded and, via the (admittedly
no less abstract) 'A-over-A principle', the impossibility of extracting any single formal
element from the German and Greek syntagms may be explained. Actually, it is even
impossible to extract any element from the corresponding Italian syntagm: we may only
mould the elements into a different shape, such as Friburgo, bella città sita nei pressi della
Selva Nera. However, syntagms of the former kind exhibit semantic properties which do
not match those exhibited by syntagms of the latter type. This certainly does not follow
from the fact that the generative grammarians'assumptions, such as the two (essentially
INTERPRETIVE CONSTRAINTS VERSUS AD SENSUM ANAPHORA 17
complementary) we have just mentioned, are able to describe but no to explain. In

elementary algebra too, it is impossible to extract any single term from the formula in
parenthesis, so that the value of the sum itself consists in the totality of the expression: I
may rewrite (a + b) as 2 (a + b)/2 or ab+b2 /b or 4a+2b/2 -a etc. Thus the functional
value of a category N can be split into various acceptable formal series to the extent and
on the condition that their sum always remains equal to their particular functional value.
In this way the comma placed between Friburgo and sita in the first Italian translation
- or the corresponding intonational change in the oral production of a non-defining
relative clause - does not minimally affect the unity of the syntagm, nor does it provide a
way out of its intensional level. It amounts to no more than extensional 'trick' which has
no effect on the structure of a sentence in which both die ... Freiburg and la ... Nera (to
cite only the initial and final elements) are noun phrases with a subject, object or
complement function.
An interpretation provided by generative framework (a necessarily short-sighted one
as will presently become clear) should take almost exclusively into account the fact that
obvious morphological agreements occur within the syntagm, between die, schöne and
Stadt in German, and between la, città and sita in Italian. In the case of a formally
different agreement in a possible English translation, where the attributes liegende and
sita in the German and Italian syntagms might be rendered which is situated, the relative
clause with the verb in the third person singular would render explicit through an
embedded sentence the function fulfilled by the attributes in both German and Italian. It
goes without saying that agreement has a sounder basis in languages such as German or
Classical Greek which possess rich and powerful morphological determining factors by
way of the article and the arrangement of ordered but not always semantically determined
formal elements.
We need only think of the anticipatory-deictic function of pronoun-articles in Epic
Greek
- (II. 16 726 = 17 82)
(Π. 23 374)
to cite only two examples. Further, ad hoc cases can be formed along the same lines as
the following, structurally similar to German:
The following long syntagm from Cicero reveals an analogous construction in Latin:
Illa autem, quae se eius imitatricem esse vult, temeraria atque inconsiderata et plerumque peccatorum
vitiorumque laudatrix, fama popularis2.
The commas inserted by the editors indicate their own particular interpretation, but the
initial illa surely refers to gloria (anaphorically referred to by eius in the context of the
gloria... solida in the preceding sentence); illa is in fact the prolexis of the closing nexus
fama popularis, where the position of the adjective (cf. the more solemn *popularis fama)
may be interpreted as the result of a rhythm-based choice, favoured so much by Cicero
and found in the famous esse videatur. This provides yet another reason for construing
illa as anticipating fama and not simply as replacing the absent gloria.
However, in another exemplum ad hoc, such as audax curru celeriter vectus auriga,
there is no such ambiguity because of the initial audax; but in the variant curru celeriter
vectus audax auriga, curru ... vectus may be interpreted not only as a further qualification
of auriga, but also as a causal, a concessive or a temporal clause, depending on whether
the sentence continues (I) primus evasit, (II) pedibus ire maluit, (ΙΠ) ultimus evasit, for
example. In our original example audax ... auriga can be understood as a series of
qualifications, of attributes of an auriga who would most certainly be rather unskilful if
his charriot went segniter rather than celeriter.
The same applies to our German syntagm featuring Freiburg, where the participle
liegende may certainly not be interpreted in either a causal, a temporal or a concessive
sense. Italian sita and English which is situated are more problematic. All three examples
admittedly share an attributive value, but only in German is this obligatory, and the same
holds for our original and invented examples from Greek. In Italian and in English, then,
such aprosdoketa as Quel bel paesino, sito su un colle ridente, è urbanisticamente orribile
/ That beautiful village (which is) situated on a pleasent hill is horrible from a planning
point of view are not surprising.
Or to put it differently, agreement is syntactically fundamental and semantically
influential on interpretation in such languages as German, Classical Greek and possibly
Latin. Thus, in our ad hoc example means that is "the one who has
suffered", and not the one who "after suffering" or "through having suffered" or "on
account of having suffered" many sorrows, finally returned to Ithaca.
In English and in Italian, on the other hand, the means by which the function of
agreement is expressed does not influence the interpretive process semantically. In the
former languages, which may be termed 'closed-structure languages', determiners fulfil
an essentially quantifying function - the type of function according to which, by saying in
Italian Mille persone hanno acclamato il primo arrivatorIn mille hanno dato la caccia a
quel disgraziato, I can congratulate the first person to arrive or else condole with the
wretched man, but my joy or regret are extralinguistic facts which may, if necessary, be
emphasised by those features that are typical of the so-called 'poetic function' of
language: they cannot be expressed in such a way as to be interpreted (or, more simply,
understood) either immediately or else in advance.
Of course, the same thing occurs in 'closed-structure languages' such as Greek. In
Archilochus (fragment 61), for instance,
the tragic observation of the death of seven wretched men "that we caught by
running" (but lit. "by the feet") and of whom "we are the thousand murderes", can only
be appreciated on the basis of the focusing device. However, the extract will prove
difficult to understand unless we bear in mind both the syntax of the Greek (in which our
"by the feet" is expressed by a 'genitive of the part') and II. 21 564, where the feet are the
swift feet of Achilles. The focused opening contrast between in the
Archilochus extract conveys condemnation of the one-sided slaughter through the 'poetic
function' of language.
In other languages which may be termed 'chain-structure languages' determiners
fulfil both a quantifying function and a 'process function', i.e. a function which depends
on the specific semantic context in which the determiners occur. In a general sense then,
the claim that human language cannot "vary beyond assignable limits"3, even though it
may be an uncontroversial that languages are "infinitely different", amounts to nothing
more than a piece of escamotage by which we simply put an end to the fundamental
question of linguistics itself, namely the relationship between what we mean and what we
are actually able to say given the means to our disposal. The question will hardly find its
solution unless scholars become aware of some basic facts: that essentially descriptive
methods of investigation have no explanatory power, that procedural techniques, until
now only partial and mechanical, cover only a small section of what is linguistically
possible, and that even predicate-logic may provide really useful tools of analysis if it
finally decides to abandon its mechanistic procedures and turns to the study of intensional
rather than extensional procedures, difficult though the former may be.
Turning once again to our analysis of Greek features, which have yet to be evaluated
in terms of the restrictions dictated by the 'A-over-A principle', I should like to draw
attention to one of the many extreme cases represented by two morphologically equivalent
but sintactically different constructions. If we compare, for instance,
and
we see that, despite the identity of the two genitive plurals των φίλων, the former
sentence means "he/she wanted to steal his/her friends' goods", while the latter means
"he/she wanted to steal the goods from his/her friends", because is (at least
preferredly) constructed with the genitive form of the person from whom the thing is
stolen. Any possible ambiguity is averted through the impossibility of extracting
from its embedding structure, which fulfils the positive function of attributing
, as well the negative one of preventing the same from
being directly dependent on The question now arises as to whether Greek,
given its more 'embedding' structure than Latin, does not require a distinction (as Latin
does) in terms of different case markings between the persons to whom the goods belong
and those from whom the goods are stolen. Latin certainly reveals no surface ambiguity
in the opposition volebat amicorum rapere bona versus volebat amicis rapere bona. As a
detailed investigation of the relationship between distribution and case-markings would
take us beyond the scope of the present study, I shall restrict myself in the remainder of
this paper to the discussion of a single aspect of this relationship.
2. Not all problems are quite so clearcut and yet so difficult to resolve. Languages, in
their infinite variety, present us with innumerable subtle distinctions and overlappings in
the light of other measures too, such as, for example, case grammar. Let us consider two
extracts from Xenophon4, which are mutally comparable as they include instances of
indirect speech and a yes/no contrast. In the first extract Xenophon relates the Athenian
strategists' refusal of Alcibíades' proposals:
which may be translated "they ordered him to leave because,

they said, it was they and not he who held command"5. In the second extract we are told
that Lysander had, on his triumph in the Peleponnesian War, kept him close but had not
answered his requests:
(which may be translated "he then ordered him
to go to Sparta because, he said, only the ephors and not he could decide on his
request").
In both passages the indirect speech, which is dependent on an implicit verb of saying
easily inferrable from the verb "to order" - (which I have introduced through the
parenthetic "they said" and "he said") - is constructed with the infinitive and
respectively) plus the nominative in the first part of the contrast in both cases, and
with the infinitive plus accusative in the second part. More specifically, in the first
sentence the accusative form is coindexed with the preceding object clause (i.e. with
αύτόv, which is the subject of απιέναι, and which together with this infinitive forms the
object clause dependent on ). In the second example the accusative τούς
introduces a new thematic element which can simply be alluded to without any
further specification, because no one was allowed to ignore the power of the ephors in
Sparta.
Syntactically speaking, the contrast between the nominative and accusative forms
describes the difference between an object clause whose subject is 'bound' to and one
whose subject is 'free' from coreference with the subject of the main clause. In this sort
of construction the use the nominative is thus symptomatic, as it were, of a binding
construction, in more or less the same way that in the Italian Voglio partire, ma voglio
che tu rimanga (where the repetition of voglio is not merely optional) the use of the
infinitive is likewise symptomatic of a coreferential link between the subjects, while the
use of the construction 'che + an obligatory subject (at least where the singular forms are
identical, except in cases of contextual reference) + the subjunctive' is symptomatic of the
fact that the subjects are non-coreferential.
Thus in Italian a sentence such as Voglio partire, ma che tu rimanga (or even worse,
without tu, ma che rimanga) is hardly acceptable, perhaps to the same extent that the
Lating sentence Volo proficisci, te autem is likewise unacceptable. On the other hand,
while the Italian sentence Dico di partire, ma che tu rimani is agrammatical, the Latin
sentence Dico me proficisci, te autem manere does not sound unacceptable. In point of
fact we should bear in mind that the functional value of cases is not the same in those
languages which have them as it is in Greek or Latin. It is highly plausible that the two
instances of indirect speech in our two Greek sentences would, in the case of Latin, have
required the accusative and infinitive construction, thus giving (Dicebant) sibi imperium
esse, non ei (or more literally, se, non eum gubernare) and {Dicebant) non se deliberare
posse, sed ephoros. The rules for the construction of the 'dicendi' verbs in Latin are
different from those governing verbs of 'willing', while in Greek the two essentially
coincide, with both requiring the infinitive together with a nominative where there is
coreference, and the infinitive plus an accusative where there is no coreference.
Somewhat paradoxically then, we might claim that in this respect Greek ranges with
those languages such as Italian, English and French which dinstinguish between verbs of
'saying' and verbs of 'willing' on the basis of coreference. In this connection it is worth
pointing out that in Modern Greek, Serbian, Rumanian (i.e. in the so called "Balkan
Sprachhund"), but also in colloquial German6, in Persian, and (exceptionally perhaps) in
Latin too (cf. Seneca, De Ben, 7 27, Vellem, non fecissem) the two Italian expressions
Dico di essere and Voglio essere are translated with the same construction as Dico che sei
and Voglio che tu sia, so that even where the subjects are identical such a construction
would, if literally translated, give the equivalent of Dico che sono (about which we shall
say more) and Voglio che io sia. Such a construction is, therefore, the same as the one
normally used, for example, in Italian with verbs of 'willing' where the subjects are not
identical. Two separate parameters are involved in this problem: the one concerning the
specific construction of verbs of 'saying' and verbs of 'willing' (and 'wishing, aiming'
etc.) and their respective lexical projections; the other concerning the function of cases.
Greek (obviously, Classical Greek) and Latin, which both prefer the infinitive
construction, particularly at certain literary levels and during certain historical periods (I
shall therefore disregard the case on and lego quod), neverthless differ with
respect to coreference: they both have the infinitive with the nominative for verbs of
'willing', while the infinitive with the accusative is specific to Latin in the construction of
verbs of 'saying'.
Italian, English, French, etc. distinguish coreferentiality from non-coreferentiality by
means of an infinitive/non-infinitive contrast in the embedded clause. Moreover, when
coreference is stated, the Italian construction 'che + a finite mood' in an embedded clause
following a verb of 'saying' may not only be considered a formal alternative to the
construction 'di + infinitive'; the former has, in fact, different semantic connotations: Ti
dico che sono pronto a tutto and Dico a tutti che sono tuo amico imply different degrees
of emphasis on the truthfulness of the claim. Dico di essere tuo amico or Hai detto di
essere pronto a tutto may be understood as less sincere utterances.
In another similar lexical area someone who assicura che safare una cosa is credited
more than someone who assicura di saperla fare. Or again, I can say Dimostro di aver
letto questo libro when what I say or do depends strictly on the content of the book, but
Dimostro che ho letto questo libro when from my standpoint as agent I provide evidence
in this regard. Given, then, the two construction with dimostro, the former appears less
centrifugal than the latter. If Italian possesses these two forms for verbs of 'saying',
other languages, as we have already seen, make no distinction even with verbs of
'willing'. Coreference therefore lacks any semantic interpretation in these languages.
2.1. Let us return once more to Classical Greek. Cases in Greek appear to have a 'deep'
syntactic function, not to say that Greek reveals traces of 'deep cases'.
It is not easy to say precisely what consequences this might have for the present state
of generative theory. On the one hand, is has often been argued that this theory, based as
it is on English, has largely ignored other culturally relevant languages, thus revealing a
feeble sensitivity for languages which could easily offer the means of verification of
abstract hypotheses; on the other hand, another more serious accusation might be brought
against those scholars who, having detected the existence of traces, have neverthless
failed to consider them with due attention. The 'empty category' which, in our two
examples from Xenophon, acts as the subject of the infinitive clause whose predicate
complements are αυτοί and κύριος respectively, is necessarily located in a case which
fulfils the function of coindexing the trace itself (i.e. the unexpressed subject of the
infinitive 'to be') with the subject of the main clause. Seemingly quite different is the
Latin construction where the reflexive se (or sibi, depending on the translations) is an
index of the coindexing itself (in generative terms, a PRO). However, as we shall see,
the standard form of the infinitive clause both in Latin and Greek takes an explicit or
implicit subject in the accusative.
A clear illustration of the fact that the subject does not require to be made explicit is
afforded by the following extract from Xenophon (already cited) which runs:
(Theramenes)
Here the accusative the object of ι ("he
had kept him up to that moment"), is taken as the subject of the object clause introduced
by ("He ordered him to go to Sparta"). These are binding phenomena in an
intuitive sense of the term and, in a certain way, they show Greek to be even further
removed from structures more easily accessible to a computerised mind. We shall not be
too surprised, then, to find a similar observation - obviously, 'mutatis mutandis' - in
Giacomo Leopardi's Zibaldone, where both the difference between Latin and Italian and
Italian's greater similarity to Greek are emphasised.
Further evidence for the existence of ad sensun references is afforded by a sentence
at the end of the epitaph (notoriously attributed to Lysias and taken from 198 80 of his
work) in honour of the troops sent by Corinthians. The text recounts that the war dead
(subject)
. Here, ως introduces an absolute accusative which coincides neither with the

nominative, i.e. the implicit subject of Φατττονται, nor with the benefactive dative
αυτοί?. On the other hand, the very choice of the accusative absolute - not in fact a very
frequent construction - is difficult to explain in view of the fact that the subjective causal
clause ("because they think that" or "as is, from their point of view, the dead in that war
deserved to be honoured in the same manner as the immortals", where much could be
said about the Greek και which, like the Latin atque, precedes the second part of the
comparison) contains no genitive that might have confused the correct interpretation.
One of the reasons for using the accusative absolute is in fact to avoid confusion with
other genitives which might occur in the same sentence and which cannot be replaced by
another form. Already in Greek, then, it was not possible to have the same case appear
more than once in the same sentence (where 'case' is not intended in the sense of
'grammatical relation'). Hence case languages apparently require equal attention at both
the formal and the syntactically functional level.
The sentence in Lysias 198 80 which follows the one we have just dealt with picks up
the referent of the speech in accusative:
In a variation here of Homer's famous saying, which he criticises, the author
states that "only for them, among men, do I think it is better to have been born". It is
therefore difficult to supply an explanation for the accusative absolute except by reference
to the context. Whereas a genitive absolute would probably have signalled a thematic
contrast and allowed us to think of any war and not the war specifically meant, this
Thucydideanlike extract, by means of the accusative absolute, links the syntagm to the
dative in the title (original or not, as may be the case),
2.2. But if our Greek examples seem to show that what can be said (and actually is said)
represents only a small part of what we mean, there being no biunivocal rules to ensure a
syntactically predictable reading, Greek also contains cases which can be read in this
second sense.
In Plato, Phaedo 66e. e.g.,
the dative in the last embedded
clause is linked to the in the main clause despite the Variatio' which aligns the
temporal clause "when we are dead" with the participial phrase "not while living". "While
living" is itself, in turn, linked to "we shall be given" as if to stress that "that which we
desire and say we are lovers of, namely the possibility of knowing" is precisely what we
desire and love in life. The next sentence, however, may seem surprising since the
connection is in this case completely intuitive:
"for if it is not possible to come to know anything clearly with the body,
then either it is totally impossible to acquire knowledge, or else it is possible after the
death". However, this sentence has no proleptic dative for the last dative to refer to. The
structural similarities in the two sentences, which, moreover, state precisely the same
thing, make it easier for us to refer in the first sentence, which
remains the necessary complement of the two occurences of "it is possible" (namely, '
+ aorist infinitive' and ' + aorist infinitive', where the aorists are
obviously to be read as resultative). But the respective referents are implicit, for nothing
is actually stated explicitly. Moreover, these 'empty categories' implicit in the binding
process come before and not after the marked lexical item and therefore behave in
opposite manner from the bindings studied by generative grammarians.
Many other examples could be cited. Three, however, - one from Greek and two
from Latin - should suffice for a clear illustration of the ad sensum reference to an
accusative noun of which there is, to use a pun, no 'trace'.
The Greek example from Plutarch reads: δοτέ οv...
(but we are not told who "must give the boys a rest"),
("convinced" - but again we
are not told who - "that our entire life is devided into relaxation and tension"). At this
point we should have expected a dative to connect up with the unexpressed agentive
dative of but the possiblility of a wrong agreement between the participle and the
benefactive has led to the choice of the accusative, the generic 'deep case'
not of agreement but of binding - as if, in place of the second verbal adjective, there had
been, say, with the subject of being explicitly
mentioned a few sentences previously, or generically,
The first Latin passage from the Younger Pliny's Panegyricus ("Traiano imperatori
dicatus", to be precise) reads: Magnum est stare in Danubii ripa, si transeas, certum
triumphi (16 2). As in Lysias'epitaph the referent here is a famous one (and furthermore
appears in the title); it is Trajan, or better, the "You" which Pliny uses to adress the
emperor for whom "It is a great feeling to be on the bank of the Danube in the certainty
that, on crossing it, you will find victory". The acc. subject of stare, which is what the
predicate certum is linked to, can be deduced from the second person singular of the
hypothetical clause si transeas - not an impersonal subjunctive but rather a subjunctive of
modal attraction.
The third and final extract is from Cicero and runs Quid melius quam in mediis vitae
laborious obdormiscere et ita coniventem somno consopiri sempiterno?8 Setting aside the
matter of the alliterations in the latter part of the rhetorical question, the subject in the
subject infinitive clause (obdormiscere et ita ... consopiri) is once again left unexpressed.
2.3. The reverse procedure, i.e. the ordering 'full NP - trace', occurs not infrequently in
Greek passages with prolepsis of the subject in the embedded clause, an illustration of
which is once again taken from Xenophon:
Xenophon here anticipates and focuses the subject of embedded clause, thus
transforming it into the object of the main clause and leaving the embedded clause itself
without and explicit subject. The omission is, of course, not particularly striking in view
of the fact that Greek is notoriously 'pro-drop' language. What might instead appear
surprising is the anticipatory device by which the subject of the embedded clause which is
moved into the main clause assumes the case required by the verb of the main clause -
hence the accusative if the verbal construction is accusatival (i.e. if the verb is transitive),
and the genitive or dative according to the requisites of the verbal construction (i.e. where
the verb is intransitive). An English translation, however, might be "Don't you know that
Lycurgus the Spartan would not have made Sparta any different from other cities etc.".
On the other hand, this latter construction which is the opposite of the, one we have just
discussed since it anticipates the subject, occurs (once again in Greek) in the following
quote from Demosthenes10, which
begins with the emphatic, fully focused "Everyone" and continues with "I know that you
wanted the city to be saved".
3. To sum up, our analysis of syntactic structures in Greek reveals the 'deep' as opposed
to the interpretive function of case. In this respect, the function of accusative as the 'deep
case' in surface non-binding proves particularly revealing11. However, as we saw at the
outset, Greek is not the only case where agreements are linked to the specific readings of
the syntagms in which they occur. An implicit syntactic item of information forces us into
an unambiguous semantic interpretation of complex syntagms falling between items
marked at either end by morphological signals, not only in Greek but also in Latin and
German and no doubt in other languages which share these structures. But in Italian,
French and English - and possibly in other languages too - syntactic information used in
the semantic interpretation of 'chain-linked' (and therefore coordinated) items is not
equally unambiguous. The embedding structure favours an attributive reading while the
chain-nexus, although not excluding attributive readings, also admits other values and
this causes a subjective orientation of the interpretive process. If our identication of a
'deep case' free from anaphoric links leaves the syntactician bewildered, then our
discovery of semantic choices enforced by the specific distributional properties of certain
languages should cause him to reflect all the more.
No doubt all things can be expressed in all languages, but - and this does not imply a
return to the problems connected with linguistic relativism - it would appear that some
languages consent to one thing being said and not another. It is by no means an easy
matter to state the precise nature of these limitations - and so I shall not even attempt to
tackle the problem here. However, they look very much like modalities of so-called
'universal grammar' which, if it exists at all, has still to a large extent to be discovered.
NOTES
1 Greek syntax would have required here

2 Tusculanae Disputationes, III, 4.
3 N. Chomsky, Knowledge of Language, New York 1986, 21
4 Historia Graeca, II, 1:26 and 2:17.
5 Syntactic conditions appears different in Italian, where one can say, on one hand, either erano essi or
erano loro a comandare, non tu or non te, but, on the other, non lui is largely preferred to non egli,
and non io excludes *non me.
6 Cf. H.H. Hock, Principles of Historical Linguistics, Berlin-New York-Amsterdam 1986, 495-497.
7 Moralia, 2, 9b.
8 Tusculanae Disputationes, I, 117.
9 Commentarii, IV 4:15.
10 De Corona, 285:117.
11 It goes without saying that both structures have nothing to do with focussing procedures, differently
from English and Italian pronominal structures that have been above pointed out (see n. 5). In Greek,
indeed, the use of accusative forms as subject (obviously, logical, not grammatical subject) in the
construal of substantivised infinitive can be shared as a similar function. Notoriously, the accusative
form as subject of an infinitive - for instance, in 'the fact that men
speak' - does not depend on the case required by infinitive itself as complement: therefore, accusative
occurs as a rule in However, nominative is also
found in such construais, if the subject of the infinitive is coindexed with the subject of the main
clause: it happens, e.g., in sentences as
"he believed to be loved for the fact that he was good" (where both subjects are coreferential). These
cases are analogous to those considered above (see 2.), which propose once again the topic theme of
the contrast between subjectivity and extra-subjectivity (not the same thing as objectivity!).
Accusative may be defined therefore the case of extra-subjectivity and in such a function it is the
opposite not only of the nominative (that is typical, instead, of the coreference between subjects),
but also of the other cases, as it is shown by the above discussion about the extract from Plato,
quoted in 2.2. On the other hand, the construction with the agent of an infinitive in genitive form (as
if infinitive were a noun) has probably very little to do with nominative or accusative as subject of
an infinitive: see, e.g., Demosthenes 19 269 On the
question cf. E. Schwyzer, Griechische Grammatik, 11:368-371, München 1950.
THE QUESTION OF MEDIAE IN ANCIENT MACEDONIAN
GREEK RECONSIDERED
GEORGE BABINIOTIS
University of Athens
0. General
The scholars of Ancient Greek dialects1 have often been concerned with the
occurrence in Macedonian of IE MA *bh, *dh, *gh as M b, d, g (β, δ, γ) instead of TA
ph, th, kh (φ , θ, χ) which appear in the other Greek dialects. The views that have been
stated so far comprise two opposing theoretical interpretations that may be characterised
respectively as (i) the "split theory" (Kretschmer, Hoffmann)2 and (ii) the theory of
"successive evolution" (Hadzidakis)3.
To the above we may add the more recent view by Kalleris4 that, in actual fact, we
should not be speaking of a law but rather of isolated cases, and the quite recent and
contrary hypothesis by Tsopanakis5 that probably we have here a phenomenon of a
general nature as it is attested in a number of Greek dialects. Despite his strong points I
would not agree with the view of Kalleris (who is trying to prove, mainly through
etymological arguments, that we have no change of φ, θ, χ to β, δ, γ but rather initial β, δ,
γ or products of different phonological developments); on the other hand Tsopanakis'
hypothesis, as he himself admits6, is in need of more sound proof. Therefore, I will
confine myself (not without consideration to space available) in the discussion of the two
main "theories" and propose a third, phonological explanation, which I consider
systematically more probable.
30 GEORGE BABINIOTIS
1. The split theory,

Paul Kretschmer was the main proponent of this theory. In accordance with it the MA
series of the IE stops developed in Greek along two different directions (split of the
voiced aspirates):
(a) Law of devoicing of aspirates. According to this the MA of IE changed into TA in
Proto-hellenic.
(b) In accordance with the law of loss of aspiration ("psilosis"), the same series of TA
changed into M in Macedonian Greek before the function of the law of devoicing,
thus coinciding with the inherited voiced stops.
According to the split theory the same series of consonants (bh, dh, gh) split into
two different series: into TA (Greek with the exception of Macedonian) and into M
(Macedonian Greek), i.e.:
Thus, the need to explain the appearance in Macedonian of word-forms like

for and the like, led Kretschmer to the
formulation of his theory and its implications (that the Macedonians spoke a language
related to Greek that was therefore created before the notion of the greek nation had come
into being). This, naturally, caused serious national and ethnological questions, being at
the same time in strong opposition to the rest of the greek structure of Macedonian. Of
course, we need not be concerned here with the greek origin of Macedonian, that is no
longer a matter of controversy, especially after the works of Hoffmann, Hadzidakis and
Kalleris. We will only deal here with the problem of the phonological interpretation of β,
δ, γ instead of φ, θ, χ.
The theory of split was also supported by the major scholar of ancient Macedonian,
Otto Hoffmann. According to Hoffmann, the phonetic status of MA in Greek was
initially (in Proto-hellenic) that of voiceless and not voiced MA7: "die Griechischen
Aspiratae in ältester Zeit nicht Tenues aspiratae, sondern stimmlose Mediae aspiratae,also
THE QUESTION OF MEDIAE 31
nicht stimmlose harte, sondern stimmlose weiche Explosivlaute mit nachklingelndem

Hauchlaute waren."8
The evolution of these consonants in Greek was twofold, or rather threefold, to be
more precise: "Während andere griechische Dialekte schon frühzeitig d°h in th und weiter
in die Spirans verwandelten, haben die Makedonen aus den stimmlosen Mediae
aspiratae stimmhafte Mediae aspiratae oder Mediae gemacht."9 i.e.:
2. The theory of successive evolution

According to this view, whose main proponent was G. Hadzidakis, the inherited MA
of Greek first changed into TA (ph, th, kh), and then developed into M (b, d, g) with the
exception of Macedonian, i.e.:
(i) Law of devoicing

(ii) Law of voicing +
loss of aspiration
We have here the application of two phonological laws, in relative chronology, i.e.
the law of devoicing of MA which affected Proto-Hellenic, and a more recent one, that of
voicing and simultaneous loss of aspiration that affected only Macedonian by diffe
rentiating the system of occlusives of this dialect. That the change to M followed the
change to TA can be proved, according to Hadzidakis,10 by instances such as the
macedonian κεβαλά which presupposes the application of the laws precisely in this order
with only Grassmann's law intervening, i.e.:
IE *ghebhalá
(i) Devoicing of voiced aspirates: khephalá
(ii) Grassmann's law: kephalá
(iii) Change of voiceless aspirates into
M (unaspirated voiced sounds) kebalá
If, according to Hadzidakis, the change into b, d, g had directly occurred from the
inherited *bh, *dh, *gh with a split, then the outcome would be different from the
attested one. In that case, we would have in Macedonian:
i. e. we would have the form instead of the attested

We should note here that in the theory of successive evolution we also have a split
which is dialectal and occurs at a later stage.
3. A review of the two theories

(1) Hadzidakis' criticism, based on the relative chronology of the laws, has shown that
the theory of split provides no explanation for the appearance of forms such as that of
(2) The same is true of Hoffmann's interpretation. We wish to add here two comments:
a) Due to the fact that the MA of IE are by definition voiced, as members of the
system of occlusives, an assumption of "voiceless" MA would cause considerable
phonetic difficulties, b) A double representation in Macedonian of MA as bh, dh, gh
or ("oder") as b, d, g, in a sense of "free variation" (Hoffmann does not indicate
difference in the distribution of either), is not acceptable as (along with the
assumption of voiceless "MA") it implies transition from voiceless segments to
voiced ones, a return, that is, to the initial stage of voicing
(3) This last weakness is also true of the theory of successive evolution. The assumption
of an intermediate stage ph, th, kh (law of devoicing) cannot lead phonetically to an
automatic creation of a series such as the M b, d, g are, that is voiced, unaspirated
stops. The change of MA > M has a prerequisite the change of two crucial phonetic
features, those of voicing and aspiration, and it cannot be effected at a single stage.
Regardless of duration, but in accordance with the relative chronology the change
must necessarily have passed through two strictly ordered phases: (i) the change into
voiced segments (voicing) and (ii) the change into unaspirated segments (loss of
aspiration). In other words, the change of φ, θ, χ into β, δ, γ in Macedonian must
appear as follows:
(i) change to TA (devoicing)

(ii) change to MA (voicing)
(iii) change to M unaspirated
(loss of aspiration)
An acceptance of the change into MA (phase ii), one of the strongest interpretations
deriving from the theory of successive evolution, and partly from the theory of split
(Hoffmann's view), would prove the basic weakness of the respective arguments, as this
kind of phonological "fluctuation" does not appear to be well founded, from a systematic
point of view. As a result, one would seem to prefer the interpretation that suggests an
inverse application of laws (ii) and (iii); loss of aspiration first (iii), and then voicing (ii):
(i) Devoicing
(ii) Loss of aspiration
(iii) Voicing
4. Proposition
Two of the most noted scholars of Macedonian, namely O. Hoffmann and I. Kalleris,
have raised questions concerning the change of φ, θ, χ into β, δ, γ, though from different
points of view and for different reasons. Hoffmann claims that " dass die Mediae β, γ, δ
nicht unbedingt reine Mediae bezeichnen müssen, sondern auch für stimmhafte aspirie-
rerte Mediae stehen können, beweist..."11 He believes that, for Macedonians, the
graphemes β, δ, γ not only stood for the consonants b, d, g but for bh, dh, gh as well.
Kalleris, on the other hand, argues that the case in question is not a case of phonological
laws but of sporadic changes and pertinently remarks that our information on the matter
(inflated and generalised as it might be) comes to us through Plutarch (1/2 century A.D.):
"aucun auteur ou grammairien antérieur à lui [meaning Plutarch] n' en a fait la

remarque."12
Much later, therefore, in the first centuries A.D., and in a rather incomplete and
generalised form, do we begin to have mention of "this phenomenon of the Macedonian
More often than not this comes as incidental information about "idiomatic
words" ι that have been preserved by Hesychius, usually by means of
explanation of similar consonant changes, especially in "Volksetymologie interpretations"
of the grammarians of the time. The way these consonants came to be recorded much
later and the scarsity of reliable sources of information about their precise pronunciation,
led the various scholars to ascribe the phonetic value of the M (b, d, g) of Ancient Greek
to them. However, the matter of the phonetic value of the Macedonian β, δ, γ should be
considered from a different point of view, in close connection with the evolution of MA
in the IE languages in general, especially in the Italic family as well as the Greek one. The
whole question has been studied so far in isolation and in rather fragmentary fashion,
with no reference to the system of occlusives of Greek. We propose to attempt here an
approach to the question from this point of view.
Our position, in a schematic and general fashion, is as follows: we believe that the
aspirated (TA) of (the rest of) Greek in Macedonian changed rather early into voiced
spirants (VS from now on), i.e. into · and not into M occlusives (M: b, d, g), as
the argument goes. Thus we have:
Our reasons for supporting this position are the following:

(i) The phonological distinction of ph : b, th : d, kh : g is of major importance in Greek
as it is of functional and differentiating value. It is not possible, therefore,to accept
a neutralisation of the distinctions which would ensue, if the members of the
oppositions were equated, as follows: ph = b,th = d, kh = g. This would mean for
Macedonian that there would be no distinction any more between the members of
the minimal pairs of the type if ph = b. Thus:
The acceptance of such an assumption would lead to a disorder in the arrangement

of the phonological distinctions of Greek, proving thus the weakness of the
position that TA > M (ph > b).
(ii) Given that the relative chronology in connection with Grassmann's law, forces us
to accept the change of MA > TA for the whole of Greek (Macedonian included) as
certain (if we wish to provide explanations for forms such as the macedonian
κεβαλά), our case automatically connects with the respective evolution in the Italic
family of IE languages. In one, classic for the kind, study of his Szemerényi13
supports and augments Ascoli's theory on the evolution of IE MA in Italian,
concluding that "in view of the unique character of sound change MA > TA, I
cannot but draw the conclusion that both the Greek and "Italic" (comprising Sicel,
Latin, and Osco-Umbrian) changes are causally connected."14
In view of the close connection of this evolution we may move to support the claim
that the further development of TA in Macedonian took an analogous early route to
the change of TA into voiced spirants (VS) and not voiceless ones (SS), i.e.:
Macedonian:
Italian (i):
Latin (ii):
O-U(iii):
The fact that TA served as a transitional phase in Macedonian as well comes in
support of the view of Ascoli-Szemerényi (in contrast with the view of Pisani et al.
who support a direct development of the Italic MA into VS)15
(iii) The evolution of the respective occlusive systems of Italian and Greek, as well as
that of Proto-germanic, move towards a more general restructuring of the system
characterised by a more symmetrical distribution of the consonants through the
creation of respective series of spirants.
GREEK
ANCIENT MODERN
ITALIC
PROTO-ITALIC LATIN-O-U
PROTO-GERMANIC
Examined from a systematic point of view the occlusives of IE show a tendency, due
to the economy of the system, to delete the problematic and unstable MA. This deletion
was effected in stages, either (i) through the change of MA to spirants with the
intermediate systematic stage of their change into TA, or (ii) through their fusion with the
voiced occlusives (M). Process (i) was adopted by languages such as the Greek and,
partly, the Italic one, while process (ii) was adopted by the rest of the IE languages.16
What is typical in the evolution of the occlusive series, as can be seen in the dia
grams, is, according to Martinet's definition,17 the tendency towards a more symmetrical
system, through the creation of one or more series of spirants.
The Greek language evolved along both directions, that is towards the deletion of
MA, and also towards the creation in the system of a respective series of spirants. In
particular, as we propose here, this phenomenon of the change of TA into VS (law of
spirantisation), a general phonological phenomenon in the evolution of Greek, Italic and

Germanic languages, appeared much earlier in Macedonian.
Our position concerning the change in Macedonian of TA > VS (a change that appears
as TA > SS in the Italic family of languages) bears on a more general issue which we
have mentioned earlier and we propose to discuss here at length. The evolution of TA
into SS (Italic) or VS (Macedonian), the phenomenon of spirantisation, comprises
phonological changes of more than one kind, as far as distinctions are concerned: loss of
aspiration, spirantisation in the case of/f/, and voicing in the case of/b/ (v):
Therefore the change of TA >SS>VS may be seen as a twofold/threefold process of

change:
(i) - loss of aspiration
(ii) - spirantisation
(iii) - voicing
Instead of the above process of change:
(i) assimilation
(ii) spirantisation
The overall change in Macedonian appears as follows:
What we have here is the creation of affricates (a combination of occlusive and

spirant) that appear in the so-called "second sound shift" (zweite Lautverschiebung) of
German,18 though with a different course of evolution:
Hadzidakis19 has suggested an analogous evolution for the spirantisation of the conso
nants of Greek in the hellenistic period:
Evidently, the interpretation proposed here is based on the acceptance of the creation
of affricates through a process of assimilation of h with the respective labial consonant
and its further change into a voiced consonant. The next phase is that of the spirantisation
of the affricate with deletion of the occlusive and phonemisation of the voiced spirant.
Finally, the same route of evolution could be proposed for the Italic family for initial
positions, i.e.:
5. Conclusions
Our proposition that the phonetic value of β, δ, γ of Macedonian was that of the
voiced spirants (VS) and not that of the voiced occlusives (M), places the whole question
of the phonological system of consonants of Macedonian on an entirely new basis. Inner
systematic evidence (such as the evolution of the consonants of Greek) as well as
comparative evidence (such as the parallel evolution of MA in the Italic family, according
to Ascoli-Szemerényi), shows, we believe, that the view about the status of β, δ, γ as
being that of b, d, g and especially the explanation that they derive directly from MA of
IE (Kretschmer's theory) should be reconsidered.
Our position in this paper is that Macedonian, an Ancient Greek dialect, existed in an
oral form (quite early the Attic dialect became the official language of the macedonic state
for socio-political reasons), so it did not suffer any effect from a conservative written
tradition and, in this sense, it moved towards spirantisation of TA, thus responding to the
general tendency of the system of Ancient Greek for a more symmetrical consonental
structure with the creation of the series of labial, dental, and laryngal spirants. Moreover,
phenomena such as the alternate appearance of β and F in Laconic inscriptions from the
4th century b.C.
reveal a tendency of some ancient greek dialects for early spirantisation.
Therefore, the ancient Macedonian dialect is by no means a deviation from the

phonological system of Greek posing M for TA. With the early appearance of
spirantisation (VS instead of TA) in it, as we have tried to prove, it became a precursor in
the developments that sealed the phonological structure of the entire Greek language of
the later times, especially during the period of the hellenistic Koine.
NOTES
1 A complete bibliography on the subject, with significant critical comments, may be found in
Kalleris 1976:355 f. On the subject of the evolution of MA this paper has made use of the
conclusions of the brilliant study" The Development of the Indo-European Mediae Aspirarne in Latin
and Italic" by the honoured professor Oswald Szemerényi. Also, we have followed here Szemerényi's
abbreviations for the relevant terminology, i.e.: MA = Mediae Aspiratae (Voiced stop aspirates), TA
= Tenues Aspiratae (Voiceless stop aspirates), VS = Voiced Spirants, SS = Surd Spirants, M =
Mediae (Voiced stops).
2 Kretschmer 1896 and Hoffmann 1906.
3 Hadzidakis 1897,1934.
4 Kalleris 1976:459-60: "de tout cela, il ne subsiste...que les cas de et, apparemment, de
- dire de deux noms propres de formation manifestement grecque; l'anomalie qu'
ils présentent ne saurait, isolée comme elle est, justifier ou étayer la "caractéristique" (ou la "loi") du
macédonien en question, pas plus que la même anomalie ne peut autoriser des conclusions analogues
lorsqu' elle se rencontre ici et là dans des noms propres de la Grèce du Sud; d' ailleurs, les nombreux
noms propres macédoniens, antiques et authentiques, avec sourde aspirée, sont là pour attester que les
Macédoniens prononçaient les sourdes aspirées grecques aussi bien que le reste des Grecs".
Prombonas seems to share this view, though he does not try to provide proof for it (The relationship
between the Macedonian and the Mycenean Dialect and the Proto-hellenic origins of the
Macedonians, Athens 1970:17 [in Greek]).
5 Tsopanakis 1988:93: "The presence of such an extensive phonetic law, uniform in many distant parts
of the modern Greek world, seems to advocate in favour of its antiquity and its extension beyond the
dialect of ancient Macedonia. I feel also that the connection between Macedonian, Homeric and the
ancient Cypriot dialects along with the approximation with the Modern Greek dialectal phenomena,
may be able to break the vicious circle of the Phrygian influence."
6 Tsopanakis 1988:83.
7 For the more general questions on the status of IE MA see Szemerényi 1987:628, note 1.
8 Hoffmann 1906:237.
9 Hoffmann 1906:240.
10 Hadzidakis 1934:237.
11 Hoffmann 1906:240, note 4.
12 Kalleris 1976:358.
13 Szemerényi 1987.
14 Szemerényi 1987:684.
15 Szemerényi ibid.
16 Persian, Lithouanian, Slavic and Celtic; see Szemerényi 1970:50.
17 Martinet 1955.
18 See Szemerényi 1970:50.
19 Hadzidakis 1924,1:128.
20 Buck 1955:47.
BIBLIOGRAPHY
Buck, C D . 1955. The Greek Dialects. Chicago.

Hadzidakis, G.1897. Zur Abstammung der alten Macedonier. Athen.
. 1924. Akadeimika Anagnosmata eis tin Ellinikin kai tin Latinikin Grammatikin. I.-II. Athens.
[Academic Readings in Greek and Latin Grammar.]
. 1934. Kai Palin peri tou Ellinismou ton Archaion Macedonon Glossologikai Erevnai 1:228-264.
Athens. [Again on the "Greekness" of the Ancient Macedonians.]
Hoffmann, O. 1906. Die Makedonen, Ihre Sprache und ihr Volkstum. Göttingen. [= Hildesheim 1974.]
Kalleris, J. 1954-76. Les Anciens Macédoniens. Étude linguistique et historique. I, 1954, II, 1976.
Athènes.
Kretschmer, P. 1896. Einleitung in die Geschichte der griechischen Sprache. Göttingen.
. 1912. Sprache. Einleitung in die Altertumswissenschaft, ed. by A. Gercke, E. Norden. 463-564.
Leipzig.
Martinet, André. 1955. Èconomie des changements phonétiques. Berne.
Szemerényi, O. 1987. The Development of the Indo-European Mediae Aspiratae in Latin and Italic.
Scripta Minora. 11:628-693. Innsbruck.
. 1970. Einführung in die vergleichende Sprachwissenschaft. Darmstadt.
Tsopanakis, A. 1988. Mutations of φ, χ, θ into β, δ, γ (in Cyprus, Macedonia, Homer and Modern
Greek). The History of the Greek Language in Cyprus, Proceedings of an International Symposium,
ed. by J. Karageorghis, O. Masson. 81-97. Nicosia.
GRIECHISCH THÉS, HÉS UND DÓS
ALFRED BAMMESBERGER
Katholische Universität Eichstätt
Schwyzer (1938:800) unterscheidet zwischen griechischen Imperativformen für 2.

Sg. des Typs áges, skhés, eníspes, pieis etc., die "gelegentlich Erweiterung durch -s"
aufweisen, und den formal unveränderlichen Aoristimperativen thés, hés und dós1
Zuletzt wurde dieser Formtyp umsichtig von Strunk 1987 erörtert. Für die historische Er-
kärung von -s geht Strunk im Einklang mit einer Reihe von Vorläufern von der indoger
manischen Sekundärendung -s aus. Die Formen des Typs thés etc. müssen somit im
Kontext des "Injunktivs" gesehen werden.2
Strunk führt aber aus, daß thés etc. "mit Sicherheit nicht unmittelbar alte Injunktiv-
formen" (1987:332) fortsetzen. Von den durch ai. dhãs und dᾶs klar indizierten Injunk-
tivformen idg. *dhē-s und *dō-s könnten in der Tat nur gr. *thēs und *dōs kommen.
Nach der bei Strunk angegebenen Proportion (2. Sg.) skhés: (2. PI.) skhéte = x : théte
konnte X = thés gebildet werden. Für hés und dós ist die gleiche Proportion anwendbar.
Freilich ist hier zu fragen, ob die Vorgeschichte von skhés eindeutig aufklärbar ist.
Die Verknüpfung von gr. ékhõ 'habe' mit der in ai. sáhate 'überwältigt, besiegt' vorlie
genden Verbalwurzel *segh- ist kaum zu bezweifeln.3 Da thematische Aoriste mit
schwundstufiger Wurzel des Typs é-lip-on (Präs. leípō) im Griechischen eine produktive
Kategorie darstellen, konnte ein Aoristparadigma é-skh-o- (Präsens ékh-o-) aufkommen.
Als Imperativ dieses Aoristparadigmas ist *skhe zu erwarten.
42 ALFRED BAMMESBERGER
Im Zusammenhang mit der Hesychglosse âges · âge, phere4 weist Strunk mit Recht
darauf hin, daß âges ein morphologisch regelrechter Präsensinjunktiv sein kann.
Dementsprechend wäre zu dem Aoristparadigma é-skh-o- ein Injunktiv skhés zu er
warten. Als die funktionale Differenzierung zwischen (Imp.) skhé und (Inj.) skhés
verlorenging, konnten beide Formen im Imperativ auftreten.5 Bei thés, hés und dós wäre
dann nach dem Vorbild von skhés das auslautende -s angetreten.6
Obwohl diese Herleitung kaum stringent widerlegt werden kann, lohnt sich doch die
Frage, welche Formen für 2. Sg. Imp. bei Wurzeln des Typs *dhē-, *sē-,7 *dō im
Griechischen aus der Sicht der Sprachvergleichung zu erwarten sind.
Bei athematischen Verbalstämmen sind für 2. Sg. Imp. nach dem Material der Einzel
sprachen zwei Bildungsweisen zu erschließen:8 "eine Variante a mit Endung *-0 und eine
Variante b mit Endung *-dhi", wobei "die erste mit vollstufiger Wurzel, die zweite mit
schwundstufiger Wurzel auftrat" (Strunk 1987:334-335). Wir können dementsprechend
bei der Wurzel *dō- (*doh-, *deh3-) für 2. Sg. Imp. *dehj39 bzw. *dh3-dhi10 postu
lieren.
Als Fortsetzung der endungslosen Form *deh3 > *doh wäre im Griechischen *dõ
bzw. *do11 zu erwarten. Da die Schwundstufe von õ (<*oh, *eh3) im griechischen Pa
radigma der Wurzel dō- als do- (z. Β. [l.PL] édomen) erscheint, ist der Reflex von idg.
*dh3-dhi als gr. *dόthi12 anzusetzen. Gr. *dóthi ist lautlich und morphologisch völlig
parallel mit ithi 'geh'.
Im Gegensatz zu dem intransitiven ithi, das sicherlich häufig vor Pausa auftrat (vgl.
etwa  8 bask' íthi), ist aber zu erwarten, daß der Imperativ zur Wurzel *dō- ein Akku
sativobjekt nach sich haben konnte. Wenn das auf idg. *dh3-dhi folgende Wort vokalisch
anlautete, dann erfolgte im Satzsandhi die Realisierung als *dh3-dhy + V-. Idg. *dh3-dhy
+ V- führte im Griechischen zu *doss + V-13 > dos + V-14
Daß bei der Ausbreitung von -s als Endung für 2. Sg. Imp. die aus dem Injunktivsy-
stem stammende Sekundärendung -s beteiligt war, ist sehr wohl denkbar.15 Daß thés,
hés und dós auf den regelrechten Impeativ auf idg. *-dhil-dhy (mit Verallgemeinerung
der Sandhivariante *-dhy > gr.- s), zurückgehen können, sollte man aber nicht außer
Acht lassen.16
GRIECHISCH THES, HES und DOS 43
ANMERKUNGEN
1 Als Formen für 2. Sg. Imp. auf -s führt Rix (1976:264) dós, hés, thés und skhés an.
2 Strunk zitiert mit teilweiser Zustimmung folgende Äußerung von Hirt: "... ein paar Imperative auf -
s, eníspes, skhés, die man als ganz regelrechte Injunktive ... bezeichnen kann. Ihnen sind wohl thés,
hés, dós nachgebildet." (1912:596). Strunk fragt aber mit Recht, welche Formen denn den
Imperativen thés, hés, dós vorausgegangen sind. Die folgenden Zeilen gehen ebenfalls von dieser
Frage aus.
3 Auf ein athmeatisches Paradigma bei der Wurzel idg. *segh- weist etwa der Optativ ai. sahyãs hin.
Der Imperativ zu einem athematischen Paradigma *segh- wäre theoretisch entweder als *segh oder als
*sgh-dhi anzusetzen. Bei der lautlichen Entwicklung ins Griechische hätte sich jede der beiden
Formen weit vom Rest des Paradigmas entfernt. Daß unter solchen Umständen ein morphologisch
durchsichtiger Injunktiv *sgh-e-s > skhés die Funktion des Imperativs übernahm, ist durchaus ver
ständlich. In dem Abschnitt "Injunktiv und Imperativ" hat Hoffmann auch auf ai. khyas verwiesen,
"woneben kein Ipt. existiert" (1967:256, Anm. 289).
4 Strunk (1987:332) merkt zur Hesychglosse α 470 an, daß die Form âges "leicht innerhalb eines Ver
botssatzes mê âges gestanden haben" kann; ein Grammatiker, der die Form ausgehoben hatte, ordnete
sie dann nach den Sprachregeln seiner Zeit als Imperativ ein.
5 Die bei Wackernagel (1953:176) angegebene Proportion thés:thétõ = X: skhétõ setzt das Bestehen
von thés voraus.
6 Petersen (1931:126) nimmt an, daß dós nach hés und thés gebildet sei. Die Herkunft von hés und
thés erklärt er aber nicht.
7 D i e Frage, ob gr. híêmi zu einer Wurzel i d g . ( v g l . lat. iacere) o d e r ( v g l . lat. serere
gehört, ist im vorliegenden Zusammenhang unerheblich; vgl. Frisk (1960:715).
8 Über die Bildungsweise des Imperativs im allgemeinen siehe etwa Szemerényi (1989:263).
9 Der Imperativ idg. *deh3 > *dō 'gib!' ist in lat. cedo 'her damit, gib her' zu erkennen, während bei dā
'gib!' wohl die Vokalfarbe a von den ursprünglich schwundstufigen Formen (damus, datis, date)
herrührt.
10 Der Ansatz idg. *dh3-dhi kann sich in erster Linie nur auf theoretische Überlegungen berufen. Indi
rekt wird aber idg. *dh3-dhi durch das Material des Indoiranischen indiziert. Dabei kann die Frage, wie
letztlich die Entstehung von ai. dehí (und dhehí) zu erkären ist, durchaus offen bleiben. Jedenfalls
deutet der im Rigveda überlieferte Präsensimperativ daddhi darauf hin, daß dehí wohl nicht ur
sprungsgleich mit dieser Form sein kann, da die unterschiedliche lautliche Entwicklung kaum zu
rechtfertigen wäre; vgl. zu Hoffmanns Herleitung (1956:21) von dehí aus dadzdhí> dazdhí (mit
dissimilitorischen Schwund von -d- in der Folge d-d-), die auch Mayrhofer (1986:111) akzeptiert, ins
besondere Tedesco (1968:2). Tedesco (1968) hatte dafür plädiert, daß dehí als der strukturell regel
rechte Aoristimperativ zu betrachten sei, dessen -e- dann das Resultat einer sekundären Umgestaltung
sein muß. Hoffmann (1967:256) betont, daß im Vedischen bei d(h)ā- "... die 2. Sg. Inj. Aor. dãh und
dhãh häufig in hortativer Funktion belegt ist, aber entsprechende Aoristimperative überhaupt nicht
existieren". Diese für das Indische geltende Formulierung ist deskriptiv richtig. Unter Berufung auf
Hoffmann schreibt Strunk, "daß Imperative der 2. Sg. Aor. zu dhā und dā im Indoiranischen über
haupt nicht nachweisbar sind" (1987:333). Für das Iranische ist jedoch diese Angabe sicherlich
44 ALFRED BAMMESBERGER
unzutreffend. An Stelle des jungavestischen Imperativs dazdi, der zweifellos zu dem reduplizierenden
Präsens *dadā- gehört, erscheint nämlich im Gathischen der Imperativ dᾶidī(Y. 28.6,28.7 [2x], 51.2,
51.7, 51.18); vgl. Bartholomae (1904:711), Insler 1975:349, Kellens 1984:393. Die iranische Form
kann wohl nur auf idg. *dh3-dhi zurückprojiziert werden, wobei sekundär die Hochstufe der Wurzel
eingeführt wurde; Tedesco (1968:7) vertritt die Ansicht, daß bei Wurzeln des Typs dᾶ- sekundär vor
konsonantisch anlautenden Endungen die Hochstufe an die Stelle einer morphologisch berechtigten
Schwundstufe trat. Insbesondere bei den Formen für 2. Sg. Imp. des Typs *dh3-dhi und dhh1-dhi ist
eine derartige Neuerung verständlich. Möglicherweise konnte nämlich der zwischen zwei dentalen
Verschlußlauten stehende Laryngal zumindest unter gewissen Bedingungen (Satzrhythmus?)
schwinden, so daß sich nur aus dem Imperativzeichen *dhi (< *dh(hj)dhi bzw. *d(h3)dhi) bestehende
Verbalformen ergeben hätten. Da im Wurzelaorist ohnehin die Hochstufe der Wurzel in sonst schwa
chen Formen erscheint (z. B. [1. PI.] adᾶma etc.), ist die Bildung von hochstufigen Imperativen idg.
*dhē-dhi und *dō-dhi nicht völlig unerwartet.
11 Bei gr. *do könnte es sich um die Fortsetzung von idg. *doh3 mit Schwund des auslautenden La
ryngeals(vor Pausa?) handeln. Strunk leitet die postulierten Formen des Typs gr. *do dagegen von
einer schwundstufigen Form idg. *dh3 ab. Solche Formen deutet er als Variante ba (= schwundstufige
Wurzel wie bei Typ b, aber Endung -0 wie bei Typ a).
12 Meyer (1879:27) stellt folgende Behauptung auf: "Nach kurzem Wurzelvocal ist das auslautende i des
Suffixes thi abgeworfen und der dann auslautende Dental zum Zischlaut geworden; so in dós (für
dόthi)". Die griechischen Lautgesetze erlauben einer derartige Herleitung von dós nicht.
13 Wenn Kiparskys Regel, daß -Vthi im Griechischen lautgesetzlich -Vith > -Vi ergeben habe
(Kiparsky 1967), zutrifft, dann hätte *dothi zu *doith > *doi geführt. Bei einem denkbaren Neben
einander von *doi (< *dh3-dhi) und dós (<*dh3-dhy) wäre es durchaus verständlich, daß die auf -s
auslautende und somit deutlich als 2. Sg. charakterisierte Form dós verallgemeinert wurde.
14 Pisani 1976 schlug vor, ai. d(h) ehi von *d(h)as + dhi herzuleiten, wobei *d(h)as mit gr. thés, dós zu
identifizieren wäre. Wenn -s in gr. thés, dós auf idg. *-thy zurückgeht, dann kann diese Erklärung
nicht mehr in Frage kommen. Gegen einen entsprechenden Vorschlag von Marsh (1941) hat übrigens
schon Tedesco (1967:2), Anm. 4 entscheidende Bedenken vorgebracht.
15 Hoffmann (1967:256) weist darauf hin, daß in der 2. Sg. Imperativ und Injunktiv zumindest teilweise
komplementär verteilt sind. Daß es sich dabei "wohl eher um eine sprachgeschichtlich ererbte Er
scheinung als um eine sekundäre Regelung" handelt, ist aber trotzdem nicht unbedingt anzunehmen.
Es könnten sehr wohl rein lautliche Schwierigkeiten bei konkreten Formen für 2. Sg. Imperativ dazu
geführt haben, daß die formal durchsichtigen Injunktivformen, die funktional nahestanden, dafür ein
gesetzt wurden. Vgl. noch Anm. 3 zu ai. khyas.
16 Daß der Imperativ im Vokalismus an andere Formen des Paradigmas angeglichen wird, scheint häufi
ger vorzukommen. So ist wohl ai. bodhi nicht mit Strunk (1987:335) direkt in die Grundsprache
zurückführbar. Da für 'wachsen' eine set-Wurzel idg. *bhewə- anzusetzen ist, kann ai. bodhí jedenfalls
nicht auf eine Vorform des Typs *bheu-dhi zurückgehen. Sofern bodhí nicht zu idg. bheudh- gehört,
ist wohl anzunehmen, daß *bhū-dhi (mit regulärer Schwundstufe *bhuə- > *bhū-) zu *bodhi
umgestaltet wurde (Tedesco 1968:4).
GRIECHISCH THÉS, HÉS und DÓS 45
LITERATURHINWEISE
Bartholomae, Christian. 1904. Altiranisches Wörterbuch. Berlin.

Frisk, Hjalmar. 1960. Griechisches etymologisches Wörterbuch. I. Heidelberg.
Hirt, Hermann. 1912. Handbuch der griechischen Laut- und Formenlehre. Heidelberg.
Hoffmann, Karl. 1956. Notizen zu Wackernagel-Debrunner, Altindische Grammatik 11,2. MSS 8.5-24 =
Aufsätze zur Indoiranistik 2.387-402 (1976).
Hoffmann, Karl. 1967. Der Injunktiv im Veda. Eine synchronische Funktionsuntersuchung. Heidelberg.
Insler, Stanley. 1975. The Gãthās of Zarathustra. Leiden.
Kellens, Jean. 1984. Le verbe avestique. Wiesbaden.
Kiparsky, Paul. 1967. A phonological rule of Greek. Glotta 44.109-134.
Marsh, Gordon H. 1941. The voiced sibilants in Sanskrit. Journal of the American Oriental Society
61.45-50.
Mayrhofer, Manfred. 1986. Indogermanische Grammatik. Heidelberg.
Meyer, Leo. 1879. Griechische Aoriste. Ein Beitrag zur Geschichte des Tempus- und Modusgebrauchs im
Griechischen, Berlin.
Petersen, Walter. 1931. The three Greek aorists in -ka. Language 7.125-130.
Pisani, Vittore. 1976. Sanskrit dehí und dhehí. Die Sprache 22.166.
Rix, Helmut. 1976. Historische Grammatik des Griechischen. Laut- und Formenlehre. Darmstadt.
Schwyzer, Eduard. 1938. Griechische Grammatik. München.
Strunk, Klaus. 1987. Ergänzende Beobachtungen zu "Wortumfang und Wortform". KZ 100.323-338.
Szemerényi, Oswald. 1989. Einführung in die vergleichende Sprachwissenschaft. 3. Aufl. Darmstadt.
Tedesco, Paul. 1968. Sanskrit dehi'give'. Language 44.1-24.
Wackernagel, Jacob. 1953. Kleine Schriften. I. Göttingen.
SUMMARY
The Greek imperative type dós may be projected back to the regularly shaped Indo-European imperative
*də -dhi with zero-grade of the root *dō- preceding the imperative marker *-dhi. Preceding a word with an
initial vowel, *də-dhi would be realized as *də-dhy. *də-dhy + V led to Greek *dos(s) and ousted the
sandhi alternant *də-dhi + consonant > Gk. *doihi.
SOME USES OF THE GREEK IMPERFECT
JAMES T. HOOKER
University College London
Having recently helped to edit a selection of Oswald Szemerényi's Scripta Minora, in

which his writings on the Greek language occupy such an important place, I cannot doubt
that he is the most distinguished, and the most original, scholar left to us in that field. We
have to look very far to find a combination of gifts to equal his: linguistic insight married
to a vast knowledge of the literature. If the following brief essay holds any interest for
Oswald, it may arise from my attempt to examine the phenomena of language not as
abstractions but in the context of Greek literary usage. The importance of doing so is one
(only one) of the lessons I have learnt from this great scholar and valued friend.
I
No one, I think, would deny that the imperfect is by far the richest of the Greek
'tenses', in the nuances it can impart to an action or state in the past. These nuances are of
peculiar interest when compared with the various uses of the aorist. But here my concern
is less with the opposition between imperfect and aorist than with the opposition (apparent
or real) between the imperfect and the present. There are, above all, occasions on which
Greek writers use the imperfect to describe what appears to be a present state of affairs.
The standard works on Greek syntax, those of Kühner-Gerth and Schwyzer-Debrunner,
each devote only a brief section to this usage, which is in fact more widespread than they
acknowledge. They explain the usage by suggesting that the speaker looks at a continuing
48 JAMES T. HOOKER
state from a point of view opposite to what is customary. Instead of looking backwards
from the present and indicating the continuation of a state or activity up to the very
moment of speaking ( with present or perfect), the speaker looks forward
from a point in the past (imperfect, sometimes associated with the particle ) and
allows the continuation of the state or activity to be inferred.
There is certainly no logical reason why a given language should not express itself in
the one mode as well as in the other, and I believe that Greek is capable of expressing
itself in both. Nevertheless it may be suggested that greater precision needs to be applied
to a discussion of this topic than is found in the handbooks. They seem to bring under
one heading matters which ought to be kept distinct. Some of the examples they offer do
not, strictly speaking, display a use of the imperfect as present at all: they simply
represent a type of attraction. Again, instances can be found (though these are not very
numerous) in which the proposed explanation is substantially correct: namely, the imper
fect does indicate the speaker's state of mind at a point earlier than the moment of
speaking. Finally there is the (miscalled) 'philosophical' use of imperfect (with or
without ), which is commonly explained on the supposition that there has always
existed some truth or some error which has only just now come to light. It may be,
however, that an explanation of this use can be found by examining those instances
(especially numerous infifth-centurydrama) in which it is hard to perceive any difference
in meaning between the imperfect and the present. If we can find a significant number of
examples in which the imperfect really functions as a present, the usage is so striking that
some explanation must be attempted.
II
First we need to clear the ground by dismissing from the argument the 'imperfect by
attraction'. Kühner-Gerth cite these three passages, among others, to exemplify the use of
the imperfect as present:
1.
2.
SOME USES OF THE GREEK IMPERFECT 49
3.
(Xenophon Anabasis 1.5.6)
In 1. Menelaus 'brought some of his ships to Crete, where the Cydonians lived'; in 2.
Socrates asks Phaedrus: 'Wasn't this the tree you were leading us to?'; Xenophon in 3.
gives the Attic equivalent first of the and then of the All three passages
refer to the past: a past which is more remote in 1. and 3. than in 2. Embedded in each
passage is a verb in the imperfect (respectively stating a general
truth which still applies at the time of speaking: the Cydonians still live where they did;
the tree is still standing; the equivalence of the measures still holds good. But these gene
ral truths are attracted into the past by the past colouring of the whole episode. The
instinct to attract the general statement into the past is a colloquial one, which is found in
modern languages including English and German. In 2., for instance, it would be open to
an English speaker also to put the question indifferently in the form 'isn't this the tree?' or
'wasn't this the tree?' Such a natural type of expression should not be thought to involve
a special use of the imperfect.
ΙΠ
It is true that some examples from the earliest literature can be brought within the
terms of the definition of Kühner and Gerth. When Achilles addresses the embassy with
the words
4.
and again when he says to Patroclus
5.
he is putting himself back in the past, specifically into the situation described in A, when
he assumed his relentless anger on account of Agamemnon's lack of . And when
Patroclus says to him
6.
he likewise refers to the past.1 Hector's dying words
7.
50 JAMES T. HOOKER
also look back to a previous moment, the moment at which he realized the impossibility of
dealing with Achilles (X 123).
By their use of such expressions, the various speakers confront, for some obvious
rhetorical purpose, the present with an identifiable moment in the past. But other speakers
in Homer use the imperfect when no specific time in the past is in contemplation - when,
indeed, they do not seem to be referring to the past at all. Thus in the Odyssey Antinous is
rebuked by both Penelope and Odysseus. Penelope contrasts Antinous' high reputation
with his actual worthlessless:
Odysseus contrasts Antinous' handsome appearance with his lack of sagacity:

9.
In neither instance can the imperfect be understood as referring to the past. Penelope
and Odysseus have never been under any illusion about the true character of Antinous,
and so they cannot be said to have been involved in an 'error' which is now discovered
and corrected.
Similarly artificial, I think, is the attempt to find a past reference in Hesiod's statement
10.
These words do not 'correct' the account of Eris in Theogony 225, as if Hesiod were
to say, 'Eris is of two kinds, then, not (as was thought in the Theogony) of one kind
only'. Such is the explanation advanced in the commentaries of Paley (1883),
Wilamowitz (1928), West (1978), and Verdenius (1985). Sinclair (1932) objected to this
line of interpretation, suggesting that 'the imperfect with expresses what was true all
along and still is'. Sinclair's explanation comes closer to the view of the imperfect, or
rather one type of imperfect, taken below (XI). He is to be commended, above all, for
drawing the reader's attention to the general nature of Hesiod's statement and excluding
any special reference to the past. In Platonic dialogues it is common to refer back (by
means of the imperfect and in other ways) to a demonstration made at a previous stage of
a close-knit argument (cf. nos. 27. and 28.). But it is one thing to allude to a demon-
stration made within a very small circle only a few minutes ago; quite another to 'correct'
a statement in a different poem, composed at a different time, and (at least in my opinion)
for a different audience: I may add that Hesiod's immediately succeeding words,
provide a further indication that there is sometimes no essential
difference in meaning between the imperfect and the present.2
In Homer there is an imperfect (this time not negatived) which can on no account have
any past reference. In the fighting over Patroclus' body, Glaucus becomes indignant with
Hector, whom he accuses of cowardice in falling back before the assault of Ajax:
Glaucus cannot be contrasting the past with the present: his reproach would then be
baseless, for it was Hector's onslaught that just now led to the death of Patroclus. But a
very recent action is usually expressed in Greek, as in Sanskrit, not by an imperfect but
by an aorist (Schwyzer-Debrunner 281-282); and an aorist is duly found at Ρ 166, ούκ
For these reasons it seems necessary to understand as equivalent to a
present.
Hence it appears that, while some of the early examples show an imperfect relating to
a time anterior to the moment of speaking, in others the imperfect cannot readily be
associated with the past but must refer to the present or to the present as well as to the
past. As we shall now see, a similar situation obtains in the Classical language.
IV
I begin with Philoctetes' despairing realization that he is destined never to leave his
cave:
Formulation and meaning are parallel to Hector's exclamation in 7. The precise moment to
which looks back cannot be identified so readily as it could in Hector's case, but
it can hardly be other than the time at which Philoctetes was put ashore.
52 JAMES T. HOOKER
A negative version of this expression is found in the speech of Heracles, contrasting

the happy outcome he expected from the oracle with the disaster which has in fact over
taken him:
The moment recalled by the words is that at which Heracles went to

Dodona.
A third example from Sophocles is brought by Kühner-Gerth under the heading of
what I have called 'imperfect by attraction'; but wrongly in my opinion. In the Oedipus
Coloneus, the Chorus have learnt from the of the presence of an impious intruder.
As they enter they ask two short, impatient questions:
has sometimes been seen (for example by the scholiast) as a simple equivalent of
the present: . An alternative explanation is that contains a real element of
'pastness', since the Chorus have in mind information previously given them (so Kühner-
Gerth 146 and Jebb ad loc). Wunder states that the Chorus could have put the question in
the form only if they had actually been able to see Oedipus, but I do not
know what grounds there might be for this statement. It is, I belive, only by a highly
artificial interpretation that one can detect any past reference in this the scholiast's
comment is right, and we have here an expression parallel to that in no. 10.
V
Looking now at some of Euripides' abundant material, we see within the same play
how a standardized expression may conceal a difference of meaning. At the beginning of
her monologue in Troades, Hecuba forces herself to accept the reality of the situation and
to bear what must be borne after the destruction of Troy:
This is Troy no longer, and no longer are we the rulers of Troy. (Tro. 99-100)
Then, a little later,
May I not groan in my misery for the loss of country and children and hus
band? August pride of my forefathers brought low: you are nothing, then.
(Tro. 106-109)
Hecuba conveys her sense of desolation by pointing the contrast between past and
present.3 It must accordingly be the case that in 109 stands upon the same temporal
plane as the of line 100. I do not think that K. H. Lee can be correct when he
translates how you are nothing after all'. Whatever the negative
with and the verb 'be' may express elsewhere, it does not here convey, as Lee
supposes, a realization of what was true in the past and still is true in the present. It is
only in the present that the splendour of Hecuba's house amounts to nothing; in the past,
of course, it amounted to a great deal. In this instance, then, the imperfect seems to have
purely present reference.
Later in the same play, however, Talthybius uses a similar formulation to express the
realization of a permanent truth:
After all, then, the grand people and those with a reputation for intelligence
are no better at all than persons who count as nothing. (Tro. 411-412)
The usage of Euripides shows that a sudden realization is often expressed by but
since he attaches this particle to the present or imperfect of 'be', to all appearance quite
indifferently, it is hard to suppose that they diverge widely in meaning. In the second
person singular we may compare
(Pylades suddenly appreciates Orestes' dangerous situation) (Or. 755)

with
54 JAMES T. HOOKER
(Copreus says that Iolaus turns out not to be a good prophet on this occasion)
(Heracl.65)
and
(The messenger begins his pert remarks with an exclamation of suprise at

seeing Helen before him: 'you are here, then?')4 (Hel. 616)
and
(The Chorus realize the true character of Medea) (Med. 1279-1280)5

The only factor determining the choice of tense in these four passages is that of metrical
convenience.
VI
The frequency with which Aristophanes attaches a present meaning to an imperfect,
especially in the context of a certain set phrase, suggests that this usage may have
belonged to the Attic vernacular. As is well known, the form often occurs in
Aristophanes when a surprised or impatient question is asked. The question concerns
what is happening now; and two examples show that the verb may equally well be present
or imperfect:
(Bdelycleon asks who is clinging to the donkey) (Wasps 183-184)
(Dicaepolis asks who the devil this is)

(Ach. 156-157)
Such an alternation between present and imperfect, uttered by the same speaker in
respect of the same set of circumstances (which are contemporary with the moment of his
asking the question), demonstrates that the imperfect can sometimes be used as equivalent
to the present. Our first instinct is naturally to suspect that the speaker, in juxtaposing two
tenses which commonly indicate actions or states on different temporal levels, means to
distinguish two temporal levels in these contexts as well. But by no sophistry can we
pretend that Bdelycleon's refers to a different time from his or Dicaeopolis'
from his In apparent ignorance of these facts, particularly of the passage in
Acharnians, MacDowell comments on no. 22:
The reason for the imperfect is obscure. Possibly it is an assumption that the surprising object has
been there for some time, even though the speaker has only just seen it. Or possibly there is a
reference to a past thought resulting in the present surprising sight: 'What was this done for?' 'What
was the point?'
This type of imperfect will remain 'obscure' to anyone who refuses to admit that its
meaning is identical with that of the present tense. It is this identity of meaning that calls
for explanation, not the (wholly illusory) element of 'pastness' which is thought always
to reside in the imperfect.
The Birds too illustrates at several places the use of present and imperfect with the
same meaning:
In each of these examples the variation exhibited is the familiar type involving the verb
'be'. But the herald's speech in the same play affords further material for the study of the
imperfect. A verb other than 'be' is in question; and furthermore two quite different uses
of the imperfect seem to be involved. The herald draws a contrast between the previous
and the present habits of mankind: then they were Laconian-mad, now they are bird-mad.
Without surprise, we observe that the first state is expressed by the imperfect
(1281), followed by a number of other imperfect verbs specifying the
habits characteristics of Laconians. Then, again as would be expected, the second state
receives its description in the present tense, (1284). This bird-madness is
subsequently summed up by the verb (1290); but this imperfect, and the
imperfects which follow, refer to a present state of things, no less than
56 JAMES T. HOOKER
did. So the imperfect was sometimes considererd to express so vividly a present state that
it could be contrasted with an imperfect which refers exclusively to the past
VII
The use of the imperfect by Plato and Aristotle next demands attention. But by way of
preface I may quote a fascinating instance in Herodotus, which I believe helps us to
understand at least one of the functions of the imperfect in philosophical writing.
Book Four of Herodotus contains the long Scythian this in turn containing an
account of the Scythians' warlike customs. The entire account of these customs
(chapters 64-65) is couched in the present (occasionally the perfect). At one point,
however, Herodotus inserts a comment of his own. Having said that many Scythians cut
the skin from their dead enemies' hands to make quiver-covers, Herodotus remarks:
Now human skin is both thick and glossy: the glossiest of virtually all skins,
by virtue of its whiteness. (4.64.3)
We may compare the following comments on this observation:

Hdt. does not actually say that he had seen any such quiver, with a cover of human skin, and the
dpa need not suggest more than 'is as I was told'. - Macan
Die Worte δέρμα - stehen in keinem rechte Zusammenhange mit dem Vorhergehenden
... H. hat wahrscheinlich die Köcherdeckel gesehen, und gründet darauf seinen Schluß, ohne seine
Beobachtung mit deutlichen Worten zu erzählen. Vermutlich schrieb er die ganze Notiz nachträglich. -
Stein.
When Macan tries to detect some 'past' reference in he forgets that, so far as
we know, the whole Scythian is based on hearsay: Herodotus makes no claim that
he himself ever visited the regions or the peoples he describes here. But the comment of
Macan is unapt in another way; for the peculiar property of the human skin is the one
thing that Herodotus did not need to be told. This he knew from his own observation: all
the rest he had to learn from others. I agree with Stein that the sentence just quoted from
Herodotus' Fourth Book represents his own addition to the story, but I do not find it
necessary to speculate whether Herodotus had actually seen such a quiver-cover or not.
The intensely personal type of writing practised by Herodotus excludes the composi
tion of long passages which are purely objective in character.6 So here Herodotus brings
in his own comment (not without introducing some inconcinnity, as Stein observes). But
why does he use the imperfect, in such strong contrast to the present and perfect of the
main account? The reason must be that the nature of the observation differs from that of
all the others: those apply merely in the present, and merely to the Scythians, but the
statement about the properties of human skin is valid without restriction of time or place.
VIII
The diverse uses of the imperfect in Plato cannot all be brought under a single head.
The topic is in fact too extensive to be treated here other than in a very superficial manner,
and I shall distinguish only three types, which ought not to be confused with one another.
One Platonic use of the imperfect closely resembles the Homeric examples collected in
ΙΠ above. Just as Achilles and Patroclus put themselves back into a previous situation by
the use of so does Socrates in Symposium 198d:
I greatly prided myself on the excellent speech I was going to make, as I knew the
truth; but this, it appears, was not the way to make a noble eulogy of anything.
In the second Platonic type the imperfect is used as a kind of shorthand to imply 'as
we saw'. The reference may be particular or more general. A particular reference is indi
cated by the imperfect in Republic 522a:
'Well, is music (the study we seek), to the extent that we have described it?'
'But that', he said, 'was the counterpart of athletics, if you remember'.
58 JAMES T. HOOKER
In this passage the points back to the discussion of music and athletics in 410c-
412b where it was established that the two activities had to be brought into harmony with
each other. An example of a more general type is provided by Crito 47d:
If we do not follow him we shall corrupt and injure that very quality which was
improved by Right and destroyed by Wrong.
Here the imperfects and allude not to a specific pasage of the same
dialogue but to the aggregate of Socratic teaching.
The third type to be considered is the so-called philosophical imperfect, that is to say
the Platonic use of (very often accompanied by to state an evident truth. Such a
statement usually sums up a section of the argument. This imperfect is accounted for on
the supposition that the particular truth obtained all along but has only now been revealed
as a result of the discussion. Although this explanation is widely accepted, I believe it to
be mistaken. The mistake arises from the conviction that whenever the imperfect is used it
must have some reference to the past. Remove that assumption, and the way lies open for
a juster appreciation of the 'philosophical' imperfect. I think it can be shown that Plato's
refers not to a certain time, but to a certain quality, of existence: not that such-and-such
a truth obtained in the past but that it always obtains, outside time. We are helped towards
an understanding of this usage by Herodotus' (no. 25.), which has no particular
past reference. A further indication that the Platonic need not refer exclusively to the
past emerges when this type of is found in company with another imperfect, which
plainly does have past reference. A passage from Meno 97b-c makes the point.
SO. So trae opinion is just as good a guide to right conduct as knowledge is;
and this is a fact we overlooked just now in examining the nature of virtue,
when we said that knowledge alone was the guide to right conduct; but in
fact there is also true opinion.
ME. So it seems.
SO. Then right opinion is just as useful as knowledge.
The precise meaning of Socrates' may be clarified by comparison first with
the preceding and then with the following The difference between
;
' and is obvious. Both are, formally speaking, 'imperfect'; but, whereas
points to a specific stage in the argument (just as the verb did in 27.),
has no temporal reference at all: it expresses an abiding truth. Of course a permanent truth
may be expressed equally well by the present 'tense', as in Socrates' In this
context the sphere of is no more delimited by what is actually 'present' than is that
of by what is actually 'past'. Either the one or the other may indifferently describe a
permanent strate of affairs. In Plato's dialogues the discovery of a particular truth never
comes about suddenly or in a random manner; the chain of inferences which has led to the
discovery is summed up, as usual, by
Many similar Platonic examples could be quoted, but here I mention only one: the
point at which in his long discourse Timaeus touches upon the theory of Forms. He asks
whether 'pure fire' and other such 'pure' qualities have an independent existence, or only
those things which we apprehend by the senses:
But are we making an idle assertion when we declare that an intelligible Form
of everything exists, while in fact it is > nothing but talk? (Tim. 51c)
Here, as in 29., marks the climax: it expresses not 'pastness' but 'reality,
actuality'.
IX
The use of observed in nos. 29. and 30. (and indeed throughout Plato's dialogues)
helps us to understand a phrase which Aristotle invented in order to express a concept of
60 JAMES T. HOOKER
fundamental importance in his metaphysics: This phrase, which is not

used by Aristotle with perfect consistency, glosses 'substance' {Metaphysics
983a27-28); but it later appears {Met. 1028b33-36) that may be applied not only to
but to any of the other three αίτια. In a long passage of Metaphysics Ζ
Aristotle gives some examples to establish in what sense he intends to be
understood: it is the essence of a thing, that in which it is said to be And
'essence' has to be distinguished from 'existence':
Existence is not simply essence but essence considered qualitatively and

quantitatively. {Met 1030a31-32)
Again:
32.
The primary essence is without matter, for it is complete reality.
{Met. 1074a35-36)
It is by such a series of steps that Aristotle builds up a definition of
This remarkable phrase has engaged the attention of Aristotelian commentators at least
since the time of Alexander of Aphrodisias. Three aspects of the formula are open to
discussion: the function of τí, the meaning of the imperfect , and the relationship of the
whole phrase to datival expressions of the type (found in both the
Metaphysics and the de Anima). No satisfying explanation of the has yet been
advanced; and, although there was a close connexion in Aristote's mind between τò τì
(see especially Met. 1031a20-21), a comparison of the
two phrases does not in itself make any easier to explain.
Trendelenburg was the first writer in modern times to attempt a thorough-going ex
planation of the imperfect in As would be expected, his discussion
illuminates at a number of points both the thought and the linguistic usage of Aristotle.
Yet his explanation of does not carry complete conviction. Observing that
, expresses {inter alia) what was in the artist's or craftsman's mind before the act of
creation, Trendelenburg draws up the following scheme:

idea prior to its material realization material realization

Hence for Trendelenburg the can be accounted for in purely temporal terms: it
expresses what 'was' before the 'is' of actuality.9
Trendelenburg's explanation found a number of adherents; but it has also had to meet
the objection that there is insufficient basis in Aristotle's own teaching for the assumption
of a temporal priority of the 'prime mover' in relation to the accomplished fact. A good
summary of the arguments is to be found in Bassenge's papers on the subject.10
Bassenge gives short shrift to the suggestion that in our phrase points to 'duration';
that can be excluded on both linguistic and philosophical grounds. And when the is
explained as an instance of reference to a previous stage in the discussion (as in the
Platonic examples 27. and 28.), the explanation errs in transferring an occasional
Aristotelian use of the imperfect11 to the constant formula expressly constructed in order
to frame a particular concept. This concept can hardly be equivalent to 'that which was
previously' or 'that which was previously demonstrated', since these could have been
expressed adequately without recourse to a special formula. What Aristotle means is
something so fundamental to his thought that it had to be differentiated linguistically both
from the Platonic Form and from everyday 'existence': it is the underlying reality
of a thing which Aristotle chooses to denote by
To conclude this rapid survey of some uses of the imperfect, I mention two passages
of Hellenistic poetry.
In his Fifth Idyll Theocritus describes a contest in singing between the shepherd
Lacon and the goatherd Comatas. The contest proper is preceded by a passage in which
the two contestants exchange insults. It is evident at the very beginning of the poem that
they are well known to each other, since they both mention previous events which have
fomented their mutual dislike. In afinalflurryof insults immediately before the contest:
62 JAMES T. HOOKER
33.
Comatas retorts with
and Lacon's rounds off the passage.
Here, as often in Plato (cf. no. 29.), the imperfect marks the climax: Lacon sums up
the essential character of his rival. The imperfect cannot here (as Gow claims) point to a
long-standing truth only now recognized, since Lacon has been aware of Comatas' nature
all along. His repeats, but in an emphatic and conclusive manner, what
he previously expressed by the present in
Problems of a different kind attend Apollonius' lines transmitted in the following
form:
Can £στι stand here, or must we read , which Schneider proposed and Fränkel
received into his text? Fränkel holds that an imperfect is called for by the phrase
'along their path there stood (not stands) a poplar'; hence either both verbs are
present (as in the model passage  811-813) or the first verb should be (as in Arg.
12
1. 1117). This is not a compelling argument. We have already remarked the alternation
between present and imperfect, both referring to the same state of affairs (e.g. nos. 14.,
16., 22., 23., 24.). Here it is the present, which marks a timeless statement: it
simply points to the existence of the tree, without defining the time at which it existed or
exists; nor is the imperfect strongly marked as a 'past' tense - its function is
merely to describe a perpetual habit of the birds. The author's meaning would not have
been materially different if he had inverted the 'tenses', marking the existence of the tree
with an imperfect and the habit of the crows with a present. This point may be illustrated
by appealing to the usage of a much later poet, namely Musaeus. At the outset of his
narrative proper he makes this statement:
Here the two statements are complementary: * no more refers exclusively to the
present than does < to the past. Both the existence of the cities and their location are put
into the same, timeless frame.
XI
From the above material, restricted in extent though it is, we may now draw certain
conclusions. The first is that the Greek imperfect, by reason of its common use to depict a
past state or the unfolding of a narrative in the past, can refer to the beginning of a chain
of events which culminates in the present: nos. 4., 5., 6., 7., 12., 13., 26. But side by
side with this type of imperfect we find in the literature another, different type. In this
second type, which has not hitherto been fully recognized for what it is, neither tense nor
aspect plays a significant part. The imperfect is used here, with or without , to make a
timeless statement about the existence of some entity. Since this type is properly con
cerned with existence, the verb in the majority of instances is but in the colloquial
style of Aristophanes the usage was extended to other verbs (VI). This 'timeless' was
taken over by the philosophers when they wished to speak of reality, as opposed to
surmise or mere appearence; and an of the same type was incorporated in Aristotle's
to convey the essential nature of a thing.
This use of the imperfect, so far as I know peculiar to the Greek language,13 stands in
an interesting relationship to the present and to the aorist.
Because of its capacity to make timeless statements, the imperfect may be used instead
of, or in company with, the present: nos. 8.-12., 16., 18.-24., 34., 35. It is, I am sure,
only by a loose use of language that such imperfects are deemed 'equivalent' to presents;
nevertheless the imperfect sometimes usurps so many of the functions of the present as to
make the difference between them imperceptible. It is certainly true that in many cases the
'past' reference of the imperfect is either very weakly marked or excluded by the context.
In one passage, disputed since Alexandrian times, the question actually arises whether the
imperfect or the present should be read. When the host of Achaeans is marshalled for the
first time in the Iliad, Athene is present in person, wielding her aegis:
64 JAMES T. HOOKER
36.
Zenodotus preferred the imperfect, but his reason for doing so is not known.
Aristarchus read the present, and his opinion has prevailed: the imperfect, he considered,
is 'out of harmony with the immortal nature of the gods'
ΣΑ). Wackernagel thought that a case could be made for the imperfect, comparing it to the
Homeric type discussed above (no. I.).14 1 prefer to see in (if that is, after all,
to be read) an allusion to a timeless attribute of the goddess; in other words, Aristarchus'
reason for rejecting may turn out to support that very reading.
The 'timeless' imperfect is not used in the same circumstances as the 'gnomic' aorist.
This type of imperfect (most often in the form describes reality divorced from tem
poral considerations. But the gnomic aorist is a true past, using a single example of an
action as a paradigm of other, similar actions at other times (including the present and the
future). Whereas the aorist, and in particular the gnomic aorist, is very much at home in
Homeric similes, the imperfect occurs there only twice:
Nor is it their lot to run across him.

(O 274)
Nor is it her lot to be captured.

(Φ 495)
This intrusion of the imperfect in a sequence of aorists amounts to a comment by the
poet himself: 'but in fact this particular result was excluded for them'. In these two simi
les the desires and hopes of the human actors are contrasted with what is bound to hap
pen, on the timeless plane.
NOTES
1 Euripides employs a similar formulation at IT 369.

2 Cf. no. 35. for a parallel formulation.
3 The sequence of thought is elucidated in Schadewaldt's Monolog und Selbstgespräch, 1926:155.
4 It is unnecessary to suppose, with Kannicht, that the messenger means to convey, by his use of the
imperfect, 'you are here now and have been all along'.
5 There is a little more to say about this example. Towards the end of the Medea, when Medea has
confided to the Chorus her resolve to kill her children, there is a deliberate effort to link this part of
the play with the beginning. For example, the allusion by the Chorus to
(1263) recalls the nurse's in line 2. In a similar way, when the Chorus
describe Medea as 'rock or iron' their words echo those of the nurse (applied to Medea in a very
different sense) at 28-29:
6 A topic ably discussed by W. Aly, Volksmärchen, Sage und Novelle bei Herodot und seine
Zeitgenossen, 1921:263.
7 Not that on all occasions he uses this phrase to denote 'essence'; e.g. at Posterior Analytics 91al3 he
expresses 'essence' by the very term which in the Metaphysics stands in opposition to
8 RhM 2, 1828, 457, 483.

9 'Das, was war das Seyn d. h. das gedachte Wesen vor der Wirklichkeit der Sache' (op. cit. 481).
10 F. Bassenge Philologus 104, 1960, 14-47, 201-222, esp. 30-31.
11 Exemplified by Met. 1071b since (as our argument showed) there are
three kinds of substance'.
12 Noten zu den Argonautika des ApoUonios, 1968:404.
13 Latin occasionally uses an imperfect to express a past state which continues into the present (ef. C.
E. Bennett, Syntax of Early Latin 1,1910, 36-37 and E. Fraenkel, Horace, 1957, 324, n.3); but it has
not borrowed the type of Greek imperfect which indicates 'timeless' existence.
14 Vorlesungen über Syntax I, 1926, 184.
LES DÉSINENCES DE GREC
WITOLD MAŃCZAK
Université de Cracovie
Les opinions concernant l'origine des désinences grecques des 2 e et 3 e pers. sing, du
prés, de l'ind. sont partagées. Voici quelques exemples:
1
1. Brugman estimait que "neben Indik. Injunkt. *ages und Imper. *age gab es in der Zeit
der idg. Ureinheit eine 2. Sg.*aĝei. Diese Bildung hatte, ähnlich wie gewisse Injunktivformen
des Präsensstamms, einen so weiten Gebrauchsbereich, daß sie sowohl adhortativ als auch in
reinen Aussagesätzen angewendet werden konnte. Im Griechischen wurde bei indikativischem
Gebrauch angehängt nach dem Verhältnis von Imper, und wurde
alsdann nach Indik. 3. Sg. neben 2. Sg.
geschaffen. Bei der Verdrängung der lautgesetzlichen Nachkommen von uridg. *agesi *ageti war
Rücksicht auf die Silbenzahl mit im Spiel. Durch Verallgemeinerung jener Formen
gewann man das Verhältnis: in den Singularpersonen einsilbige, in den Pluralpersonen
zweisilbige Endung."
2. Solmsen2 a approuvé l'opinion de Brugmann en ajoutant qu'il était enclin "als die eigentliche
Heimat dieser Formation den Imperativ anzusehen und idg. als Verschmelzung
des letzteren mit dem Imperativ ei... (lat. i...) oder mit der Interjektion ei zu erklären".
3. De l'avis de Meillet3, "le type gr. . ne concorde pas avec celui de skr. bhárasi,
bhárati... et rappelle seulement la 2 e pers. sg. lit. lëkï. Il est possible que ces formes aient été
faites d'après le type en Sans doute n'est-il pas fortuit
que 17 lituanien repose sur un rude et que l'accentuation de gr suppose
une ancienne diphtongue - d'intonation oxytonée, c'est-à-dire ancien
4. D'après Vaillant4, "la désinence -ει du grec trouve son explication dans hitt. -ah-i ... Pour la 2 e
personne du singulier, hitt. -ah-ti doit aider à interpréter les désinences remaniées gr. - εις..., v.
d'intonation rude, supposé par le balto-slave."
5. Une autre opinion est représentée par Schwyzer:5 "so kann (aus
darstellen; Opt.
68 WITOLD MAÑCZAK
, Ind. (lesb.) Nach neuerer Ansicht enthält jedoch ζ. Β.

vor dem zugefügten -ς die indikativisch gebrauchte 'Imperativ'form auf danach
In beiden Fällen ist die Bildung der 3. Person nach der 2. Pers. auffällig."
6. Kurylowicz6 pense que "two phonemic changes take place: the final dental -t and the intervocalic
-s are dropped... In the second person the marked ending is formed from the unmarked one by
subtraction of -s and addition of -i. Hence, once the final -t of the third person has disappeared, we
get proportionally:
It is clear that such a solution would not have satisfied Brugmann or his successors since
proportion (I) would not have been, from the phonetic point of view, admissible."
7. Watkins7 considère que "interne Rekonstruktion des Singulars im gr. thematischen Paradigma
führt... zur Annahme der folgenden historischen Entwicklung:
Il désigne -i comme une "particule facultative".

8. De l'avis de Rix8 les désinences de sont le résultat de métathèses:
9. L'article de Cowgill ne nous a pas été accessible, mais voici ce que Rasmussen9 a écrit à son
sujet: "Cowgill's paper has now appeared in the said congress report (Grammatische Kategorien:
Funktion und Geschichte, hrsg. von B. Schlerath, Wiesbaden, 1985, 99-108). The material
supporting the derivation of Greek- ει from is indeed such that one would like to accept the
sound law proposed by this eminent scholar... :/t/ and /th/ were lost between a posttonic non-
high vowel and/i/."
10 Hoenigswald10 considère que "at one time the active endings for the second and third persons
singular were presumably... *-es-i *-et-i (primary) and *-es* -et (secondary). This bit of paradigm
was disturbed by the Proto-Greek loss of final *-t. At that point an analogic innovation becomes
possible: (e)diōkes: : X; X = ., διώκει, or perhaps better *(e)diõkes:
A second analogic change completes the picture:
Nous n'avons pas l'intention de critiquer les opinions de nos prédécesseurs, mais
voudrions proposer une nouvelle solution à ce problème. Le but du présent article est
d'expliquer la désinence de par ce que nous appelons le développement phonétique
irrégulier dû à la fréquence d'emploi. Depuis de longues années, nous essayons d'attirer
l'attention des linguistes sur le fait que, dans toutes les langues du monde, le dévelop
pement phonétique irrégulier dû à la fréquence est, à côté du développement phonétique
régulier et de l'évolution analogique, le troisième facteur essentiel qui décide de la forme
LES DESINENCES DE GREC 69
des mots: dans n'importe quel texte de n'importe quelle langue, environ un tiers des mots
présente un développement phonétique irrégulier dû à la fréquence11.
En bref, la théorie du développement phonétique irrégulier dû à la fréquence se pré
sente somme suit. Il y a une loi synchronique selon laquelle les éléments linguistiques
plus employés sont plus petits que ceux usités plus rarement. Il existe une sorte
d'équilibre entre le volume des éléments linguistiques et la fréquence. Mais on sait que le
volume des éléments linguistiques n'est pas stable. A cause du développement phonétique
régulier, la longeur des mots peut changer sensiblement, comme le montre la comparaison
de quelques mots latins avec leurs équivalents français:
lat. tē (2 phonèmes) > fr. toi (3) - augmentation de 50 %;
tū (2) > tu (2) - aucun changement;
portam (6) > porte (4) - diminution de 33 %;
illam (5) > elle (2) - diminution de 60 %;
Augustum (8) > août (1) - diminution de 88%.
La fréquence des mots n'est pas stable non plus. Un mot comme sire, très employé au
moyen âge, est maintenant peu utilisé. Inversement, un mot comme chauffeur est plus
employé de nos jours qu'il n'a été dans le passé. Dans cet état de choses, il n'y a rien
d'étonnant à ce que l'équilibre entre la longueur d'un élément linguistique et sa fréquence
puisse être bouleversé. Si un mot devient trop court par rapport à sa fréquence, on le
remplace par une expression plus longue, par exemple août [u] est souvent remplacé par
[ut] ou bien le mois d'août. Mais si un élément linguistique, c'est-à-dire un morphème, un
mot ou un groupe de mots, devient trop long par rapport à sa fréquence, il doit être
abrégé, et il existe plusieurs possibilités: abrègement syntaxique {chemin de fer métro
politain > métropolitain), abrègement morphologique {automobile > auto), abrègement
mécanique (faculté > fac), abrègement graphique {Organisation des Nations Unies >
ONU) ou bien abrègement phonétique: le suffixe germanique -isk aboutit normalement à
-ois, cf. le prénom Franç-ois, mais l'ethnonyme Franç-ais, sensiblement plus employé
que le prénom, présente une réduction de -ois à -aïs; la négation latine nōn, employée
isolément, aboutit régulièrement à non, mais la même négation utilisée devant un verbe (ce
qui constitue un emploi plus fréquent) présente une réduction de non à ne; le groupe de
mots mon seigneur conserve sa prononciation primitive dans le titre ecclésiastique
70 WITOLD MAŃCZAK
monseigneur, mais le même groupe de mots employé fréquemment dans la langue de tous
les jours subit un abrègement en monsieur.
Si le développement phonétique irrégulier dû à la fréquence est assez avancé, il
consiste dans la chute de un ou plusieurs phonèmes, par exemple sapiō > sais, cantāre
habëtis > chanterez, mais quand l'évolution phonétique irrégulière se trouve dans sa phase
initiale, elle ne consiste que dans une réduction partielle du phonème. Voici quelques
exemples:
la quantité de la voyelle peut être réduite, par ex. nostrum > nôtre > notre;
le timbre de la voyelle peut être réduit, cf. le futur fera de faire en regard du futur
régulier plaira de plaire;
l'aperture de la voyelle peut être réduite: a >  >  ou bien a> e>i (par ex. lat. sine a
abouti, en a. espagnol, régulièrement à sen, qui est devenu, par la suite, irrégulièrement
sin);
une consonne palatale peut être réduite en une consonne non palatale, par ex. illōs est
devenu en espagnol los, bien qu'un / mouillé existe en espagnol à l'initiale, cf. lleno.
Il y a six arguments qui témoignent en faveur de la théorie du développement
phonétique irrégulier dû à la fréquence:
I o S'il existe pour la langue et la période en question un dictionnaire des fréquences, on
peut en user parce que la grande majorité des mots subissant un développement
phonétique irrégulier dû à la fréquence se trouve parmi les mille mots les plus
employés. Par exemple, en français moderne, ces mots se présentent comme suit:
1er mille 99 86%
2 e mille 9 8%
3 e mille 4 3%
4e mille 2 2%
5 e mille 1 1%
6e mille - -
2° Si le morphème, mot ou groupe de mots apparaît dans une langue sous une double
forme, régulière et irrégulière, le développement phonétique irrégulier se caractérise
par le fait que la forme irrégulière est en général plus employée que la forme normale,
par ex. notre est plus employé que nôtre.
LES DÉSINENCES DE GREC 71
3° Si les réductions irrégulières ont lieu à l'intérieur d'un paradigme ou d'une famille de
mots, les abrègements irréguliers se produisent plus souvent dans les formes plus
fréquentes que dans les formes moins employées. Parmi les formes ai, as , a, avons,
avez, ont, les formes ai, as, a et ont présentent des réductions parce que le singulier
est plus employé que le pluriel et la troisième personne est plus utilisée que les autres
personnes. It. grande se réduit en gran, tandis que grandezza ou grandeggiare, moins
employés que grande, ne subissent pas de réduction.
4° Si, pour une langue, on dispose à la fois d'un dictionnaire des fréquences et d'un
dictionnaire inverse, il est instructif d'examiner des mots commençant par la même ou
les mêmes lettres ou bien des séries de mots terminés par la même ou les mêmes
lettres. Il s'avère par exemple que, en anglais, il y a 700 dérivés en -iness (comme
happiness), parmi lesquels il n'y en a qu'un seul qui ait subi une réduction, à savoir
business, et business est le mot le plus fréquemment usité parmi les dérivés en -iness.
5° A côté du développement phonétique irrégulier dû à la fréquence, il existe une autre
évolution irrégulière, consistant en des accidents phonétiques connus depuis long
temps sous le nom d'assimilations, dissimilations et métathèses, en des formes
hypercorrectes ou expressives. Toute cette évolution se caractérise par le fait que,
dans des langues différentes, des mots différents la subissent. Chercher < cercher
présente une assimilation, mais il serait difficile de trouver, dans une autre langue, un
mot signifiant "chercher" avec une assimilation. Faible <flēbilem a subi une
dissimilation, mais il ne serait pas facile de trouver, dans une autre langue, un mot
signifiant "faible" avec une dissimilation. Et ainsi de suite. Il n'y a donc aucun
parallélisme entre les irrégularités dites assimilations, dissimilations, métathèses, etc.
En revanche, le développement phonétique irrégulier dû à la fréquence a lieu, dans
des langues différentes, d'une manière plus ou moins parallèle, ce qui s'explique par
le fait que, malgré les différences qui séparent des communautés linguistiques, les
mots les plus employés sont partout plus ou moins les mêmes. Par exemple, le verbe
signifiant "parler" présente dans beaucoup de langues des réductions irrégulières, cf.
fr. parler, it. parlare <rblāre, des formes de narrāre devenu en sarde nárrere
{, nas, nat, etc.), lat. ajō < *agiō (en face du régulier adagium), angl. says, said
(en regard du régulier lays, laid, où la diphtongue persiste), russe gyt < gryt <
govorit, etc.
72 WITOLD MAŃCZAK
6° De deux formes à développement phonétique irrégulier dû à la fréquence, la forme

plus fréquemment usitée existe, en principe, sur un territoire plus vaste que la forme
moins employée. Dans un texte, nous avons trouvé 101 infinitifs en -er, 36 en -ir et
18 en -oir. En consultant I' ALF, on s'aperçoit que l'aire où l'on prononce les
infinitifs en -er sans r est plus grande que celle où l'on supprime r dans les infinitifs
en -ir. Il en résulte qu'il existe un lien entre la fréquence des infinitifs en -er et -ir et la
grandeur des aires où les suffixes de ces infinitifs subissent la chute irrégulière de r.
A notre avis, l'origine des désinences de est comme suit. D'abord, à
l'époque préhistorique, un changement phonétique irrégulier dû à la fréquence d'emploi a
eu lieu dans *(bher)-eti > (chute du t). Après, un changement phonétique régulier
a eu lieu dans *(bher)-esi > (chute du s). Les désinences des 3 e et 2 e personnes
étant devenues homonymes, un changement analogique a eu lieu à la 2 e personne: la
désinence secondaire -s a été ajoutée à -ει , d'où -εις.
A l'appui de la thèse selon laquelle un développement phonétique irrégulier dû à la
fréquence a eu lieu dans la désinence de la 3 e pers. sg. -ει , on peut alléguer deux
arguments:
a. Une réduction irrégulière (chute du t) a eu lieu dans la désinence de la 3e pers. sing.,
tandis que la désinence du pluriel - ουσι(ν) ainsi que celle du duel - ετον, toutes deux
moins employées que - et, ne présentent pas de réductions irrégulières.
b. Parmi les désinences verbales, celle de la 3 e pers. sing, du prés, de l'ind. est la plus
fréquemment usitée. Voilà pourquoi le développement irrégulier de grec - ΕΙ a des
parallèles dans d'autres langues. Voici quelques exemples.
Le développement phonétique irrégulier dû à la fréquence consiste parfois en une
dépalatalisation irrégulière (cf. ci-dessus esp. los en regard du régulier ellos). En v. slave,
la désinence *-ti aurait dû aboutir normalement à *- (comme en v. russe). La désinence
dans une forme v. slave comme s'explique par le fait que, dans * - [ t ' i ] ,
le [t'] a subi une dépalatalisation irrégulière, qui a entraîné automatiquement la substitution
de ü à Ĭ (Ĭ ne pouvant suivre que les consonnes palatales).
Tandis que, dans une forme russe comme , il y a la même dépalatalisation irré
gulière comme dans v. slave , la plupart des langues slaves présentent un déve-
loppement irrégulier plus avancé, à savoir la disparition de la désinence, cf. polonais

pisze.
A l'appui de la thèse d'après laquelle des formes comme v. slave pišetu ou pol. pisze
s'expliquent par un développement phonétique irrégulier dû à la fréquence, on peut
mentionner les arguments suivants:
en v. slave, l'équivalent régulier de la désinence *-ti, c'est-à-dire -tĭ, est attesté plus
rarement que l'équivalent régulier de la désinence *-mi, c'est-à-dire -mĭ, parce que la 3 e
personne est plus employée que la lre;
en bulgare, la désinence de la 3 e pers. sing, a disparu, tandis que celle de la 3 e pers.
plur. (moins employée) s'est maintenue, cf. pece "il fait cuire" mais pekat "ils font cuire";
on observe le même phénomène dans certains parlers russes;
en ukrainien, en biélorusse et dans certains parlers russes la disparition irrégulière de
la désinence de la 3 e pers. sing, a lieu uniquement dans la première conjugaison, tandis
que dans les autres conjugaisons, qui sont moins employées, la désinence en question a
persisté;
dans certains parlers russes, une dépalatalisation irrégulière du t ne s'est produite que
dans la désinence de la 3 e pers. sing, des verbes de la première conjugaison, tandis que
les autres conjugaisons, qui sont moins utilisées, présentent la désinence -t'.
En outre, on peut signaler qu'en v. slave l'évolution phonétique irrégulière de la
désinence de la 3 e pers. est parallèle à l'évolution graphique de la même désinence: dans
les textes v. slaves, bọdetu est parfois remplacé par bọdet ou bọdet
Le fait qu'en lituanien la 3 e pers. du prés, de l'ind., par ex. neša, est, du point de vue
historique, dépourvue de la désinence, s'explique de la même manière que les formes
polonaises du type pisze, c'est-à-dire par un développement phonétique irrégulier dû à la
fréquence, à l'appui de quoi on peut alléguer les arguments suivants:
dans les textes anciens, les verbes thématiques présentent non seulement la désinence
normale -ti, mais aussi les désinences réduites -t' ou-t,
que les terminaisaons -t et -t' doivent leur naissance à la fréquence est prouvé par le
fait que des réductions ont lieu uniquement à la 3 e personne, alors que les désinences des
autres personnes restent telles quelles, cf. miegt "il dort" en regard de miegmi, miegsi; la
conjugaison des verbes athématiques en v. slave constitue un parallèle: dastǔ (avec -tǔ au
74 WITOLD MAŃCZAK
lieu de -tí) en face de damǐ, dasi; le développement de la désinence lituanienne -t'>-t

rapelle une évolution semblable en v. slave et en russe;
dans certains parlers lituaniens, on observe même la chute de ce qui, du point de vue
historique, est un suffixe thématique: ved "il conduit" en regard du littéraire veda.
En ce qui concerne les autres langues baltiques, la situation est semblable, cf. a.
prussien giwa "il vit", qui est comparable à des formes polonaises comme pisze "il écrit".
Goeman et van Reenen 12 ont étudié la chute irregulière du t final dans des parlers
néerlandais et sont arrivés à la conclusion suivante: "For the future we may expect that, if
something of the [t #]-deletion will remain definitely in the Language, it will concern the
items of the present verb tense and the past verb tense with a high token frequency." 
s'agit, entre autres, de formes comme maakt "il fait", où la désinence -t tombe irré
gulièrement.
Angl., néerl, a. saxon is et suéd. är présentent également une chute irrègulière de la
désinence.
Dans des verbes latins comme legit, l'évolution de la désinence *-ti n'a pas été
normale non plus.
A la lumière de tous ces faits, rien d'étrange que la désinence de (Φέpει s'explique par
un développement phonétique irrέgulier dû à la fréquence. D'ailleurs, il en est de même
pour celle de (Φέpω13.
NOTES
1 K. Brugmann, Zur Bildung der 2. Person Singul. Akt. in den indogermanischen, insbesondere den
baltischen Sprachen, IF 17, 1904/1905, 180-181.
2 F. Solmsen, Zur Geschichte des Dativs in den indogermanischen Sprachen, KZ 44, 1911, 174.
3 A. Meillet, Caractère secondaire du type thématique indo-européen, BSL 32,1931,202.
4 A. Vaillant, Le problème des intonations balto-slaves, BSL 37, 1936, 112.
5 E. Schwyzer, Griechische Grammatik, I, Munich 1939, 660.
6 J. Kuryłowicz, The Notion of Morpho(pho)neme, Directions for historical linguistics, Austin, 1968,
77.
7 . Watkins, Indogermanische Grammatik, III, Heidelberg 1969, 123.
8 H. Rix, Historische Grammatik des Griechischen, Darmstadt 1976, 251.
9 J. E. Rasmussen, Miscellaneous morphological problems in Indo-European languages (I-II), LPosn
28, 1987, 62.
10 H. M. Hoenigswald, Some Considerations of Relative Chronology: The Greek Thematic Present,

Festschrift für E. Risch zum 75. Geburtstag, Berlin 1986, 372-373.
11 W. Manczak, Le développement phonétique des langues romanes et la fréquence, Cracovie, 1969;
Slowianska fonetyka historyczna a frekwencja, Cracovie 1977; Frequenzbedingter unregelmäßiger
Lautwandel in den germanischen Sprachen, Wroclaw 1987.
12 . . M. Goeman et P. Th. van Reenen, Word-final t-deletion in Dutch dialects, Vrije Universiteit
Working Papers in Linguistics, No. 16, Amsterdam, 1985,26.
13 W. Mańczak, Le caractère archaïque du type v. ind. bhárãmi, Studia Indo-Iranica, Wroclaw 1983,79-
83.
SUMMARY
Opinion on the origin of the endings of Greek ι is divided. According to the present
writer, first of all, an irregular sound change due to frequency of usage has taken place in the ending *-eti
> - ει (disappearance of t). Then, a regular sound change has occurred in the ending *-esi > - ει (disap
pearance of s). The endings of the second and third persons having become homonymous, an analogic
change has taken place in the ending of the second person: the secondary ending has been added to - ει,
whence -εις.
GREEK AND RELATED PHENOMENA
KLAUS STRUNK
Universität München
1.1. In the literary tradition of Ancient Greek both 'tending goats, goatherd' and
| 'tending kine, cowherd' were used from the Homeric poems onward. On
Mycenaean tablets only the latter compound occurs as qo-u-ko-ro /gwoukolos/, whereas
there is no record of the former. In this early dialect of Greek the meaning 'goatherd'
seems to have been expressed by means of a synonym ai-ki-pa-ta, although some formal
details of this word are not quite clear (see Aura Jorro 1985:135). Whatever may be true
with regard to ai-ki-pa-ta, the non-occurence of an *ai-qo-ro */aikwolos/ in the Linear 
inscriptions does not imply that */aikwolos/ could not yet have been coined elsewhere in
the late second millenium B.C. Such a precursor */aikwolos/ of Homeric αίπόλος- would
evidently square with Mycenaean qo-u-ko-ro /gwoukolos/, Homeric βουκόλος and, in its
last member, also with Mycenaean a-pi-qo-ro /amphikwolos/, Homeric
'handmaid'. Thus */aikwolos/ may have existed in some dialect-variety of Early East or
South Greek, which was not employed by the scribes of the Mycenaean archives.
1.2. After all, the formal and functional parallelism between
(including their derivatives) is obvious. In this respect most details are clear and well
known. In either case we have to do with the -type of active compounds.
Their final member, showing regressive accent on the penultimate syllable according to
Wheeler's law, governs - just like a verbal participle - the first member in the part of a
direct object (see Risch 1974:196-197). As far as ' are concerned,
78 KLAUS STRUNK
their respective last members - are even made up, from an etymological
point of view, by the same verbal noun. It must be derived from *-kwol(H)-o-s with o-
grade of the root *kwel(H)- (cf. Vedic cárati, Latin colo, Greek and
the stem-final suffix -o-. During an early period of alphabetic Greek, the labiovelar
followed by the mid back vowel -o- regularly developed to
'herding horses' etc. This process had not yet worked in Mycenaean Greek, as is made
patent by the preserved labiovelar (or the like1) in the above-mentioned a-pi-qo-ro. In
however, according to an anterior rule, the labiovelar had yielded -κ-
irrespective of its position before -o- due to a preceding -υ- in
'wolf'. Since the third syllabogram of qo-u-ko-ro in Linear  texts turns out to be --,
not -qo-, the adduced change of labiovelars being in contact with  (cf. e-u-ke-to
/eukhetoi/ with -ke- instead of -qe- and Rix 1976:86 § 97 a) must have taken
place earlier in Pre-Mycenaean times.
2.1. Everything is unambiguous so far. But there remains a minor problem with regard to
αϊ πόλος·. In general, of course, the first member of such compounds consists of the mere
noun stem. This is evident in etc.
Hence we should expect, at first sight and from a morphological point of view,
something like or rather as a result of partial assimilation of the
voiced plosive (γ) to the following unvoiced one (π), just like e.g. in
The cluster -κττ- in such an would not have
been impossible for phonotactic reasons, all the less as it was divided up by a clear-cut
morpheme juncture. Within numerous compounds beginning with the cluster ■
occurs on quite similar conditions, for which the lot of entries from up to
in Liddell-Scott-Jones, A Greek-English Lexicon, 515a - 519a, may be
compared. Perhaps -κπ- embracing the morpheme boundary between two members of a
compound was felt to be somewhat more acceptable, if the first member was an adverb
(έκ-) than if it was a noun stem with final dorsal plosive as in αίγ- (αίκ-). But this seems
to hold true not before the later classical language. In the course of that period we meet
with insertion of a so-called 'compositional' vowel -o- in examples such as αίγ-ο-
'goat-faced' Hdt. 2,46, 'goat-slayer' Paus. 9,8,1 etc.2.
Compounds of the type 'steep, sheer' (attested since the Iliad) and
GREEK . AND RELATED PHENOMENA 79
'browsed by goats' (attested since the Odyssey) must be analysed otherwise: in its form
and function their first member represents a dative-instrumental case in
2.2.1. Consequently, there was hardly any phonotactic restraint in Early Greek by which
would have had to be excluded or altered. That means, we find
ourselves confronted with the question what could have given rise to the mutilated stem-
form al- of An idea put forward by Ehrlich has been published by Brugmann
(1904/1905:7) and was adopted by Fraenkel (1910:84 with fn. 1) and Schwyzer
(1939:398), who is referred to by the current etymological dictionaries (Frisk 1960:43;
Chantraine 1968:36). In the formulation given by Schwyzer (I.c.) it runs as follows: 'In
αίπόλος für scheint die Behandlungsweise des absoluten Auslauts (Wegfall
von -γ) in die Komposition übernommen zu sein'. Now, if we do not want to practise
pure paper-linguistics, this assumption raises the problem how we are to envisage a real
basis and the modalities of the suggested transformation in the framework of natural
communication. In other words: Is there truly a minimal free form in the inflection
of (Attic , see Schwyzer 1939:377) and, if so, might it have served as a
functionally plausible model for the actual first member in the compound
Certainly, from a morphological point of view, must be acknowledged as a
vocative singular-form. Yet even he who did not check the word-indices of all Greek
texts including those of the comedy may conclude that because of semantic absurdity
addressing a goat by a vocative cannot, if at all, often be attested. That means, it is
highly improbable that the vocative should have been as natural and frequent in
use as to yield the model for ι
2.2.2. It might be objected that, in a way, the compound γυναιμανης 'mad for women'
instead of could be taken for a revealing counterpart of . For it
likewise seems to have lost the stem-final dorsal plosive of its first member before an
initial labial consonant of its second one (see Schwyzer 1939:583). May that have
happened due to a transfer of the vocative from the simplex-paradigm? In the case
of such an explanation could in fact be right instead of Frisk's (1960:334)
idea, who supposed an analogical formation after (rather different) compounds beginning
with But a possible procedure of this kind in the formation of γυναι
would have been based upon extremely special conditions. The vocative γύναι
was quite natural and frequently used. So, without regard to subsequent texts, we find
80 KLAUS STRUNK
four occurences of ' in the Homeric Iliad and twenty-one in the Odyssey. This
fluency of the vocative γύναι could at least have favoured the formation of γυναιμανης
within the Homeric language. Moreover, in the old Epic the compound itself occurs only
in the vocative singular (II. 3,39 = 13,769):
Here appears in a series of altogether four vocatives. This arrangement

of the verse intended by the Homeric poet may have induced him to create γυναιμανές ad
hoc as an artificial and ephemeral compound with an additional fifth vocative (γυναι-) in
its first member. The position between the feminine caesura (after the so-called 'third
trochee') and the bucolic diaeresis, held by ' in the formulary verse II. 3,39 =
13,769, was preserved in the Homeric Hymn 1,17 (to Dionysos):
Some further occurences of this word and of its derivative ' ' , almost
exclusively in late epic texts, clearly depend upon the Homeric ones, whereas elsewhere
came into usage. To sum up:
seems to be an abnormal and momentary formation, which was shaped
under certain particular circumstances (naturalness of the vocative γύναι, internal
arrangement of the verse II. 3,39 = 13,769). Since similar circumstances do not apply to
αίπόλος, we should stop seeing in its first member the realization of the stem αίγ- in
final position, that means, the vocative. It is not worth while to consider a few allegedly
parallel examples such as 'irascible', supposed to go back to
and (Schwyzer 1939:398-399). In these cases the phonotactic and
morphological details do not sufficiently correspond to those of Further
more, a stem beside 'unmixed, pure', presumed by Brugmann
(1904/1905: 8) in did not exist at all so that Solmsen (1901:30-31) probably
was right in comparing and in deriving the
compound from an underlying sentence (see Chantraine 1968:51
vs. Frisk 1960:58).
3.0. Rebus sic stantibus we cannot help seeking for another way out in the matter of
It is available in the prehistoric sound-shape of this compound, when the initial
plosive of its last member was still a labiovelar (see paragraph 1.1).
GREEK AND RELATED PHENOMENA 81
3.1. According to the evidence mentioned in paragraph 1.2, had lost the high-
round feature of its second and voiceless labiovelar in a Pre-Mycenaean period. Due to a
more general contextual rule this loss turned out to be bound to a position in contact with
u. Since  likewise owned the features [+ high] and [+ round], the special change of
labiovelars in question can be defined in terms of partial dissimilation. In other words:
*gwou-kwólos yielded Mycenaean /g w oukolos/, Greek through partial
w
progressive dissimilation of its *-k - at the outset of the last member from its preceding
stem-final *-ou- in the first member.
3.2. As to a similar process must have happened the other way round more or
less during the same period. That means, became *aik-kwόlos through
partial regressive assimilation of its stem-final dorsal *-g- in the first member to its
following initial labiovelar in the last member. Such a devoicing of a voiced
plosive before a voiceless one is a well known rule both of Indo-European and of Greek
(see paragraph 2.1). Then, by shifting the syllable- or morpheme-boundary, the cluster
merged in Later on, yielded
due to the normal posterior sound change of its labiovelar before the back-vowel
*-o-.
3.2.1. Occasional typological parallels to the supposed shift of the compositional juncture
*aik-kwolos > *ai-kkwólos (first step) and to the subsequent merger *ai-kkwólos > *ai-
-kwólos (second step) may be found in alphabetic Greek, Latin and Romance. They are
able to emphasize the general phonetic possibility of such a procedure.
3.2.1.1. Hermann (1923:132-170) has collected numerous items of syllable-division in
historical Greek inscriptions. There we repeatedly meet with divisions such as
(Dreros), (Malla), (Hierapytna), [πνι]-κτών
(Kos), (Kalymna) etc. (Hermann 1923:157, 161). They probably indicate
spoken shifting of compositional and syllable-junctures (~ 'first step' in paragraph
3.2.1). Furthermore, in an inscription from Eresos (Lesbos) of the late 4th century B.C.
e.g., the well known compound εκκλησία (i.e. ) 'assembly duly summoned'
occurs as (Schwyzer 1923: No. 632 D 9). Since throughout this extensive do
cument geminates prove to have been accurately written (e.g. A 5,
A 21,22,  8-9, D3, . D 20-21
etc.), the single -- of , hardly originated in a scribal error. Instead, it is the final
82 KLAUS STRUNK
result both of boundary-shifting to (first step) and of a subsequent merger in

(second step), just like simple *-k - of our suggested *aikwólos. The Latin loan
w
ecclesia likewise occurs in inscriptions of Gallia Narbonensis as eclesia, and nothing else
but eclesia with -cl- (not -ccl-) can have been the source of the respective words in West
Romance languages such as French église, Spanish iglesia, Italian chiesa etc. (Meyer-
Liibke 1935:253; v. Wartburg 1949:204). Finally; it may be pertinent to quote Meyer-
Liibke (1920:172): 'Die Vereinfachung von ecclesia zu eclesia ... ist in griechischen
Handschriften ebenso häufig zu treffen wie in lateinischen, also vielleicht auf griechische
Rechnung zu schreiben.'
3.2.1.2. Another typological parallel to Greek *aig-kwólos >
may exist in the Latin particle equidem, if it is from
*eg(o)quidem similar to tuquidem (see Leumann 1977:461 § 367.1 Zusatz, with
references). This derivation of equidem is favoured by the fact that Plautus, Terence and
later on notably Cicero had a clear preference for using that particle in connection with the
1st person singular4. Provided that the real starting-point of equidem was indeed
*eg(o)quidem, the development must approximately have been as follows: *egoquidem >
*eg-quidem (syncope) > *ec~quidem (partial regressive assimilation) > *e-cquidem
(shifting of morpheme-boundary) > *equidem (merger of -c- and -qu-).
NOTES
1 Some pertinent questions of labiovelare in Mycenaean and historical Greek, notably with regard to
post-Mycenaean labialization, have been discussed by Szemerényi 1966.
2 See Chantraine 1968:36-37: Un certain nombre de composés présentent au premier terme la forme
αίγο-; ils sont généralement tardifs ...'
3 See Chantraine 1968:30 (s.v. ), Risch 1974:219 § 80 a (concerning
4 See Burckhardt 1935:498-499 (and ThLL V 2 s.v. equidem), Leumann 1939:88-89, and, with further
references, Walde-Hofmann 1938:411, Ernout-Meillet 1967:556-557 (nevertheless, the suggestion that
equidem was based on *eg(o)quidem is rejected by either dictionary).
GREEK AND RELATED PHENOMENA 83
REFERENCES
Aura Jorro, Francisco. 1985. Diccionario Micènico. Volumen I. Madrid.

Brugmann, Karl. 1904/1905. und seine griechischen Verwandten. Indogermanische Forschungen
17.1-11.
Burckhardt, Georgine. 1935. equidem. Philologus 90.498-500.
Chantraine, Pierre. 1968. Dictionnaire étymologique de la langue grecque. Histoire des mots. Tome I.
Paris.
Ernout, Alfred, et Antoine Meillet. 1967. Dictionnaire étymologique de la langue latine. 4ième édition,
2ième tirage. Paris.
Fraenkel, Ernst. 1910. Geschichte der griechischen Nomina agentis auf Erster
Teil. Straßburg.
Frisk, Hjalmar. 1960. Griechisches etymologisches Wörterbuch. Band I. Heidelberg.
Hermann, Eduard. 1923. Silbenbildung im Griechischen und in den andern indogermanischen Sprachen.
Göttingen.
Leumann, Manu. 1939. Lateinische Laut- und Formenlehre 1936. Glotta 27.75-92.
Leumann, Manu. 1977. Lateinische Grammatik. Erster Band. Lateinische Laut- und Formenlehre.
München.
Meyer-Lübke, Wilhelm. 1920. Einführung in das Studium der romanischen Sprachwissenschaft.
Heidelberg.
Meyer-Lübke, Wilhelm. 1935. Romanisches etymologisches Wörterbuch. Dritte Auflage. Heidelberg.
Risch, Ernst. 1974. Wortbildung der homerischen Sprache. Zweite, völlig überarbeitete Auflage. Berlin-
New York.
Rix, Helmut. 1976. Historische Grammatik des Griechischen. Darmstadt.
Schwyzer, Eduard. 1923. Dialectorum Graecarum exempla epigraphica potiora. Leipzig.
Schwyzer, Eduard. 1939. Griechische Grammatik. Erster Band. München.
Solmsen, Felix. 1901. Untersuchungen zur griechischen Laut- und Verslehre. Strassburg.
Szemerényi, Oswald. 1966. The Labiovelars in Mycenaean and Historical Greek. Studi Micenei ed Egeo-
Anatolici 1.29-52 (= Szemerényi 1987:1220-1243).
Szemerényi, Oswald. 1987. Scripta minora. Selected Essays in Indo-European, Greek, and Latin, ed. by
P. Considine and J. T. Hooker. Innsbruck.
Walde, Alois, und Johann . Hofmann. 1938. Lateinisches etymologisches Wörterbuch. Erster Band. 3.,
neubearbeitete Auflage. Heidelberg.
v. Wartburg, Walther. 1949. Französisches Etymologisches Wörterbuch, 3. Band. Tübingen.
AEOLIC DATIVE PLURALS IN -ESSI
WILLIAM F. WYATT, JR.

Brown University, Providence, RI
It is with some trepidation that I approach the subject of the dative plural because
Oswald Szemerényi has himself devoted so much profitable attention to the subject. I
offer this small contribution in hopes that I can perhaps fill in a small corner of the puzzle
which has so productively occupied his time and efforts.1
Anna Morpurgo Davies2 skillfully rehabilitates forms such as - for
theoretically to be expected - as genuine Lesbian forms and not Homeric
loans. Given previous assumptions that all Lesbian, and indeed all Aeolic, datives of
third declension nouns ended in - εσσι, forms such as in Sappho, the only form
of -s- stems to occur in that author, had to be regarded as non-Aeolic and were therefore
held to be borrowings from the Homeric poems. It is obviously a dangerous proceeding
to explain as borrowed the only form occurring in a dialect, but such is the usual belief.
Morpurgo Davies is therefore certainly correct in her assumption of the genuineness of
the form, but she goes astray in providing an accounting of it and in certain other
corollary assumptions. She assumes:
1. in Aeolic
2. '~ e.g., was created after the application of 1)
3. results, as Wackernagel3 held, from the proportion -oi: -oisi = -es : X,
where X = -essi
4. Homeric παντεσσι is an Aeolism of epic, while
86 WILLIAM F. WYATT, JR.
5. is either a) an archaism or, more likely

b) an epic creation based on πάντεσσι.
One will note that not all of these postulates are logically connected, but they are all
related. I am concerned primarily with the first one, and with it therefore the second.
1. Morpurgo Davies posits a phonological rule: -ss- > -s- for common Aeolic, and thus
by regular phonological change. This change seems unlikely to have
taken place, both because of forms such as πάντεσσι, which cannot be invoked here
because of the circularity of reasoning involved; and also - and more importantly -
because -ss- in fact is preserved intact in (e.g.) and is even introduced by
analogy in forms such as (after Given
these facts a purely phonological accounting of the data is contrived and (I should say)
impossible, in spite of Morpurgo Davies' assertion (193) that the evidence for -ss- in
Aeolic dialects is inconclusive and "does not speak for or against the proposed sound
change -ss- > -s-."
Phonological change seems excluded, but morphological change is still a possibility.
One must not a priori assume that because a given allomorph of a morpheme occurs in a
dialect that it is the only allomorph of that morpheme so to occur. The Homeric case is
paradigmatic but perhaps extreme. We must not, however, merely because Aeolic texts of
various sorts contain the allomorph -essi, assume that texts which contain instead the
allomorph -si are somehow aberrant and influenced by other dialects (though that
explanation is not to be totally excluded). In fact -si does occur in Sappho and Alcaeus at
least, and as long ago as 1923 E. Lobel4 observed that: "Datives plural of neuter stems in
-ες which one might have expected to find in the form - εσσι(ν), never have any form but
-εσι(ν), as far as our present evidence goes." He also notes (ibid.) that datives of vowel
stems such as δρύσι or quasi-vowel stems (by this he means forms such as
have simple -si, which he does not regard as the simplification of an original -ssi. In fact
in the Lesbian poets, whatever other alternations there may have been, it seems quite clear
that the allomorph -essi was restricted to consonant stems, and that in vowel stems the
allomorph -si was utilized. We of course tend to regard the s-stems as consonant stems.
The ancients, and in our case more particularly the speakers of Lesbian Aeolic, felt that
forms such as _ were vowel stems very like and In addition, though
AEOLIC DATIVE PLURALS IN -ESSI 87
this is questionable, it seems that monosyllables were or could be regarded as being in the
same class as vowel stems. The rule therefore seems to have been: use the longer form
-essi in polysyllabic consonant stems, but use -si elsewhere. Clearly the evidence of the
poets reveals a certain amount of free variation (in words such as beside πόσι
and and here we may invoke literary influence of various sorts.
Tne important fact is that, as Morpurgo Davies has held, forms such as are
regular in Lesbian and, as I hold, do not result from a phonological simplification of -ssi.
The origin and development of the form can be displayed as follows:
-i- stems -s- stems consonant stems
genitive
dative
dative
The middle term is identified with the left-hand one because of the vowel in the genitive
plural.
J. Chadwick5 argues ingeniously that - σσι in Lesbian is replaced by -σι in s-stems
under Ionic influence in the same way that Lesbian has adopted other Ionic features such
as the - σι for - TL in the third singular indicative active, etc.
πάντεσσι etc. were not similarly shortened because Ionic lacked forms in - εσι in these
forms, having instead. Chadwick's view is not to be dismissed
entirely, for Ionic influence seems to have been pervasive on Lesbian. On the other hand
his explanation cannot be the sole factor, for in fact Lesbian seems to have utilized -si in
-i- and -u-stems, and the form, e.g., πολιεσσι seems late. His explanation will of course
not account for (Alc. F 3(b) 35) or (Alc. Q 1.10). He accounts
only for the s-stems, and that is not enough.6
2. With the status and history of forms such as cleared up, we can now proceed
to which Morpurgo Davies feels is a late introduction to Aeolic. We now see
that there is no need for such an assumption. How one regards the history of this form
will be colored in part by one's view of the history of the Homeric text - to which we
will return below; in part it will be colored by views as to the origin of the form, to which
we will proceed in the next section. An initial obstacle, one that is to me fatal to accepting
a late introduction into Aeolic, is the agreement among all Aeolic dialects in having -
in consonant stems. Of course much depends on what one means by 'early' and 'late'. If
there was such a thing as a period of Aeolic unity, which I rather doubt,7 it must have
existed on the Greek mainland north of the Isthmus, probably prior to the Mycenaean
period.8 Hence forms such as exampled in all Aeolic dialects, must antedate the
Lesbian poets by some six hundred years on the usual chronological assumptions. If
such a dating is late, so be it. Morpurgo Davies is almost certainly correct, however, in
holding - if she does - that the establishment, though perhaps not the origin, of - as
an allomorph of dative plural in consonant stems is not to be dated to a proto-Greek or
common Greek period.9
3. Wackernagel (op. cit.) held that -essi results from the proportion -oi: -oi-si = -es: -es
si, and that -essi was not extracted as such from the -s- stems and extended to consonant
stems. He found the necessary proportion or analogy for such a change missing, and
hence proposed his own proportion. It might be well to note here that he held (375=969)
that the proportion was: Only
later did -εσσι spread to the neuter s-stems. He thus turns the older explanation, whereby
- εσσι spread precisely from the neuter s-stems, on its head. Wackernagel is in principle
correct in faulting the traditional explanation both on general grounds and specifically
because of the (supposed) absence of linguistic probability. His own view, however,
itself encounters severe objections. First of all, one rather fails to see any particular basis
for the analogy, since, after all, -o-l-a- stems are really quite different from consonant
stems. In what way can analogy operate on such incommensurables? Furthermore, and
more fatally, Thessalian and Boeotian lack - and yet have which
means in turn that the -si of Lesbian -o-Ια- stems is an innovation in that dialect and not an
inheritance, an innovation easily motivated by homonomy in those stems between dative
and accusative plural. In fact the best one could expect from an analogy like the one
Wackernagel proposes on the basis of and is the spread of  to form
λογοισι, not the introduction of -es- into I cannot accept Wackernagel's
explanation, and have attempted a refutation of it only because good scholars, like
Morpurgo Davies, have adopted it.
CD. Buck10 has said all that needs to be said about Wackernagel's arguments about
the inadequacy of the older explanation: "That vague but real thing which we call
linguistic sense analyzes forms in various ways according to the momentary point of
view." What can the momentary point of view have been? It seems abundantly clear that
speakers of (certain forms of) Greek were troubled by the lack of morphological clarity in
datives plural such as *potsi (*potši, *possi), or *pantsi (*panïši, *pãsi). As a result of
their malaise they sought and found a longer form of the dative plural which was more
nearly agglutinative in nature and which therefore preserved the morphological clarity of
both stem and ending. Furthermore, theoretical considerations aside, it is clear that they
found an ending -essi ready to hand which could serve, however they analyzed it. That
form doubtless came from words such as just as in northwest Greek -OΙΣ was
taken as such from the o-sterns and added to consonant stems to form πάντοις. In both
cases the morphological problem was paramount and a solution found, a solution which
Wackernagel and linguists generally who are given to (sometimes excessive)
schematization would deplore.
Nonetheless some schematization is necessary. At some point in the history of
Greek, some early point, but probably Mycennaean or sub-Mycenaean, we find datives
of the sort potisi, dakrusi, epessi, potsi, panisi, luontsi with their corresponding genitives
polio:n, dakruo:n, epeo:n, podo:n, panto:n luonto:n. Somehow, and we cannot know
precisely how, potsi, panisi and luontsi became problematic, panisi, luontsi probably
first.11 Again we cannot know precisely what the problem was, but for some reason
Greeks north of the Isthmus wished to create a more perspicuous paradigm which
preserved clarity by preserving the integrity of the stem which was threatened by the
cluster of -(n)tsi. Analogy operated vertically on these paradigms from the -o-stems, as
follows:
panta luonta fila

panto:n luonto:n filo:n
panisi luontsi filois
panta luonta fila
On the basis of these relations was formed the proportion fila :filo:n : filois = panta :
panto'.n : X = luonta :l u o n t o : n :X, where X is pantois, luontois. In fact this development
was probably in origin more widely spread, but was finally restricted to areas which we
ordinarily characterize as Northwest Greek.12 Another group of speakers favored a
horizontal analogical arrangement in that they adopted the dative ending of other
consonant stems.13 Datives available to restore clarity were 1) -si, the original form and
the one that caused the problem, 2) -ssi from the -s- stems, or 3) -isi, -usi-, from the
other third declension nouns. No one of these forms would do because -(s)si would
recreate the same problem, and because either -isi- or -usi- (or -ihi-, -uhi-) would have
been too characterized, belonging uniquely to -i- and -u-stems. This impasse, presented
here as a theoretical problem of linguistic analysis rather than as a need instinctively felt
by native speakers, was broken by the adoption of - εσσι extracted from neuter s-sterns. -
εσσι was positively and unmistakably characterized as dative plural, and its -/e/- vowel
was not so characteristic of any stem class that it would prove distracting. In fact it must,
given the event, have been regarded as characteristic of all consonant stems, thus
allowing us to infer a segmentation This segmentation may have
been widely spread once, but it caught on only in the northeastern area of Greece.
4. Morpurgo Davies holds that, e.g., Homeric πάντεσσι is an Aeolism in epic. Again my
arguments above do not in any way disprove this assertion, but also again in no way is it
necessary. Unfortunately she does not specify what she intends by "Aeolisms of the epic
language." She can mean either relics from an Aeolic phase of the epic language, a phase
which I am not alone in arguing against,14 or she can mean a recent introduction to epic,
but from Lesbian Aeolic, not Aeolic generally, a conclusion which I espoused (op. cit.
144-146). Such an assumption requires the prior existence within Lesbian of the longer
datives in -essi, as in Schematically:
-i- stems -s- stems consonant stems
Homeric
Homeric
after Lesbian
With this change or expansion the Lesbian rule (given above) was reformulated for epic
to read: add -essi to the stem, thus allowing both έπέεσσι and vέεσσι, as well as
πολίεσσι in epic.15 Clearly Homer (or his predecessors) did not formulate what they had
done in rule form, but rather reasoned that, since their own dialect at the time had
and since -essi was a characteristic of old epic and since it occurred both in old epic and
Lesbian old epic and Lesbian, at least in the instance of datives plural, were
identical. From this they concluded that was an old epic form and was not only
useful but to be used. is, therefore, an Aeolic form, but more especially a
Lesbian one, one borrowed or adopted into epic by a poet familiar with the Lesbian
dialect but almost certainly not a speaker of it.16
5. I have assumed above that Homeric is an archaism of epic and not an

innovation based on The necessity of this conclusion should be obvious, the
justification perhaps not so obvious. Some support has been provided in my 1975 article
cited above. The argument for the antiquity of the form will derive from probability. First
of all, Homer does in fact preserve many older forms protected from change by their
position in the verse. Secondly, among these archaic forms there are many which show
-ss- in which later Attic-Ionic has -s-, forms such as τοσσος for * for
latei it would seem most natural to assume that -ss- is preserved in ἕπεσσι as
well. For we know from etymology that the -ss- in this word corresponds with the form
reconstructed for the earliest forms of Greek. The easiest and the best assumption in this
case is that is an archaism of great antiquity.
Morpurgo Davies prefers to believe that is an epic creation based on
This seems quite unlikely, for the proper analogy is simply not there. We have
seen above reason to believe that -essi arose early on in Aeolic, and that Homeric
is likely to derive from the homonymy of Homeric and Lesbian ἕπεσαι. Without that
homonymy it would seem very difficult for the Homeric to have developed. If
Homer (or his predecessors) had built a form on πάντεσσι, it is quite clear that that form
can only have been with -ss- on the basis of the rule: add -essi to the stem.
Hence we are inevitably led to conclude that early epic preserved and developed
after contact with Lesbian Aeolic. Later forms will have developed within the
tradition.
At the time of which we speak epic had the rule: add -essi to the stem of consonant
stems, that is, the Lesbian rule. Later on, within the history of epic, this rule spread so as
to involve all words of the third declension. In part we can follow the progress of this
development and can date it to a period after the loss of digamma. For the form
has the peculiarity in Homer of frequently following datives such as with the
shorter form of the dative, not the longer one which one expects for Homer.17 In fact an
examination of all forms of έπέεσσι in Homer reveals that most either are preceded by a
short form of the o-stem dative plural or occasion elision of a preceding vowel. The
strong suspicion therefore arises that the lines in which these forms occur are
remodelings of earlier lines which had the longer form of the dative plural or hiatus.
Thus:
These two examples contain the unusual form of the dative plural in what seems a
formulaic system of adjective + in the dative plural. It is most likely that at one time
the formula was expressed by which, with the loss of digamma, became
Subsequently - and for whatever reason - was substituted
for this formula. At least one reason mut have been the availability of as a valid
18
form. Similarly with elision:
One may imagine that an earlier form of this phrase (or a phrase like it) took the form
Once established in epic, however, became fixed and
favored, so that, e.g., which occurs ten times, can never be replaced by
The above arguments, starting from the recognition that in Aeolic poets is a
genuine Lesbian form which results from the morphological simplificaton of -ssi allows
us the possibility of accounting for historical developments not only in Lesbian but also
in Homer. The chronology will have been the following:
1.Northeast Greek creates a new dative plural in consonant stems. The ending of this
new form, -essi, is taken over from the equally consonantal -s-stems, as argued
above in 3. This change is common to all Northeast Greek dialects, and must date,
because of inferences derived from Mycenaean, to about 1200-1000 B.C.
2. Homeric Greek adopts the longer dative forms, as in πάντεσσι, from Lesbian Aeolic
by the route given in 4). This development must have taken place after the Ionian (and
Aeolic) migration to the coast of Asia Minor, hence after about 1000 B.C. It could
have been considerably later.
3. Lesbian is replaced by as argued in 1) above. The new Lesbian
rule is now: use -essi in consonant stems, -si elsewhere (including -o- and -a-
stems).19 This change can have taken place at any time, and my formulation of the
rule must allow for optional features, at least in monosyllables. I arbitrarily date it to
ca. 900 B.C.
4. -essi spreads to s-stems in Homer as a result of an extension (and almost certainly
misunderstanding/reinterpretation) of the Lesbian rule to include all consonant stems,
most notably the s-stems, as in The extension of this rule must be dated to
after the loss of initial */w/-, thus to approximately 800 B.C.
From the above one can observe that 5-stems in Greek are the crucial point around
which developments in the dative plural revolve. Sometimes they are regarded as
consonant stems, sometimes as vowel stems. Their position was problematic in that they
partook of characteristics of both classes, and could not be securely assigned to either
one. Uncertainty of classification rather than neat analogical formulae best accounts for
developments in the dative plural of third declension nouns.
NOTES
1 Most notably in The Development of the -o- / -α-stems in the Light of the Mycenaean Evidence,
Proceedings of the Cambridge Colloquium on Mycenaean Studies, ed. by L. R. Palmer &
J. Chadwick, Cambridge 1966, 217-225, and in Strukturelle Probleme der indogermanischen
Flexion: Prinzipien und Modellfälle, Grammatischen Kategorien: Funktion und Geschichte, Akten
der VII. Fachtagung der Indogermanischen Gesellschaft, ed. by B. Schlerath. Wiesbaden 1985,
516-521.
2 Studies in Greek, Italic and Indo-European Linguistics offered to Leonard A. Palmer, ed. by
A. Morpurgo Davies and Wolfgang Meid, Innsbruck 1976,181-197.
3 IF 14, 1903, 373-75 = Kleine Schriften, Göttingen 1953, 967-69.
4 ΜΕΛΗ, Oxford 1925,1.
5 The Aeolic Dative Plural, Festschrift for Oswald Szemerényi on the Occasion of his 65th Birthday,
ed. by B. Brogyanyi, Amsterdam 1979, 207-211.
6 W. Bliimel, Die aiolischen Dialekte, Göttingen 1982, denies that -esi- in s-stems is native colloquial
Lesbian, and asserts that the shorter forms are epic loans, while -essi, segmented -es-si, is the correct
Lesbian form. He also accepts Wackernagel's proportion for the origin of the longer form because
(245-248) he maintains, on quite insufficient evidence, that proto-Aeolic possessed both -oisi and -
ois in the dative plural of o-stems. It is odd that he accepts Corinna's forms as true to dialect, while
asserting that Sappho's equally numerous usages are borrowed from epic.
7 That is to say I do not isolate Aeolic as a unity quite separate from other dialectal areas of Greece,
preferring to think of Thessalian, Boeotian, Lesbian as dialects within the northeastern area of
Greece, an area reached by certain linguistic changes (and not by others) which affected other areas of
Greece; and an area within which certain changes took place which did not reach other dialectal areas.
Cf. my The Prehistory of the Greek Dialects,  101, 1970, 557-632, part. 619-632.
8 J.L. García Ramón, Les origines postmycéniennes du groupe dialectal éolien (= Suplementos a
MINOS 6, Salamanca, 1975) 84 dates the development to the period 1125-1000 B.C. He may well
be correct.
9 Clarity is needed here. Allomorphs can (and do) arise at any time, and some arise at many times. For
a trivial example, the extension of the allomorph -/s/ of the plural in English to the word feet, thus
yielding feets, happens repeatedly among young children until corrected by adults. Hence the
allomorph - of the dative plural arises at any time that forms such as become
problematic within the system of nominal declension. The first such time would have been when
intervocalic -s- passed (by way of -h-?) to -O-.At this point Greek possessed forms such as *olii,
*agroii, *potsi beside *epessi, and hence there were three allomorphs of dative plural: -i, -si, -ssi. -i
had the disadvantage of being homophonous with the dative singular, and -ssi was restricted to -s-
stems, so lacked sufficient generality. As a result, -si was everywhere generalized. On the other hand,
-ssi or possibly even -essi, may have been selected by some persons in some areas for some period of
time. If so, these forms did not catch on generally or at that time, -essi may have arisen repeatedly
but in restricted areas, only to be discarded.
10 CR 19, 1905, 247-250.
11 My guess is that these forms were phonetically [pantsi] but phonemically /pansi/ because there was
no distinction tween /ts/ and /s/ after /n/. Furthermore, it is likely that /Vns/ was realized as so
that the paradigm of would have shown panta panto:n past panta. Though the ending of the
dative was clear enough, the stem on this assumption was not, and it was the clarity of the stem
which required preservaton. And just as -/es/- was inserted in words like was
inserted also in /pant-es-si/.
12 Because the Doric dialects do not share in this innovation, one must assume that they were already
located south of the Isthmus, and this in turn dates the development to 1100 or so at earliest.
13 By vertical analogy, not a good term, I intend an analogy based on a word's paradigm when compared
with the paradigm of other words which bear some resemblance to it. Horizontal analogy operates on
a case (or form generally) within a given class and not across class boundaries. In this case the class
is formed by third declension nouns, and developments here resemble or recall developments in Latin
datives in which -/- spread from -/-stems to consonant stems, and Latin genitives like urbium which
etymologically possessed no -/'-. Wackernagel went wrong in positing that a vertical analogy existed
with the horizontal relation of datives, incorrectly holding that -o-/-a-stems formed a single class
with consonant stems. The spread of - εσσι is rather more like the spread of -i- in Latin. The
analogy is simply not there in Wackernagel's scheme, though an explicit analogy must be present in
such matters, as Szemerényi points out in another connection, citing Rogge, Tractata Mycenaea,
Proceedings of the Eighth International Colloquium on Mycenaean Studies, Held in Ohrid, 15-20
September 1985, ed. by H. Ilievski and L. Crepajac, Macedonian Academy of Sciences and Arts,
Skopje 1987, 349.
14 Cf. my Homer's Linguistic Ancestors of the

University of Thessalonike 14, 1975, 133-147 (with references to Szemerényi's conclusions, similar
to my own, presented at JHS 79, 1959, 193, SMEA 1, 1966, 31-35).
15 This rule may also have spread in Lesbian, though here one does suspect literary influence, for the
relevant forms are late. Cf. E. Schwyzer, Dialectorum graecarum exempla epigraphica potiora,
Leipzig 1923 (repr. Hildesheim 1960) 619.6 (πολ'ιεσσι.) and 634.7 as well as the
cited by Morpurgo Davies 185.
16 Homer clearly was familiar with Aeolic speech in its Lesbian form, as indeed he was also to a lesser
extent with Doric. Whether or not he (or the epic tradition of dactylic poetry) was a Chiote must
remain moot at this point. In favor of Homer as a Chiote cf. D. Gary Miller, Homer and the Ionian
Epic Tradition, Innsbruck 1982, part. 13-38.
17 Cf. P. Chantraine, Grammaire homérique, I, Paris, 1948, 194-96 and 133-34 on
18 Another reason would seem to have been a desire to avoid hiatus before I may have more
to say on this matter elsewhere. For the moment I would observe that Homer preferred the shorter
form of the o-stem dative to hiatus after the longer form.
19 It may be that the development of - from was in fact part of the larger remodeling
within Lesbian of all datives which involved also extending the -a- and -o- stem datives by -i. And
here we may, with Chadwick, assume Ionic influence. For both Thessalian and Boeotian, as well as
other northern Greek dialects, have - ois* / - in the dative plural, thus indicating that -σι was
not present in those dialects. The idea that common Aeolic (Northeast Greek) once possessed - OΙΣι
but that Thessalian and Boeotian adopted - oις / - aιs under West Greek influence, seems
to me absurd. On this entire matter cf. C. J. Ruijgh, Mnemosyne 11, 1958, 97-116, P. Wathelet,
Les traits éoliens dan la langue de l'épopée grecque 243-265 (= Incunabula graeca 37, 1970),
Szemerényi in the Cambridge Colloquium (above η. 1), J. Moralejo Alvarez in Athon: satura
grammatica in honorem Francisci R. Adrados, A. Bernabe et al. Ed., Madrid 1984-. (Read in preprint
form). The dative plural question must in turn be inserted within the larger context of syncretism of
cases, on which most recently see R. Coleman, Early Greek Syncretism, Minos 20-22, 1987, 113
25 (= Studies in Mycenaean and Classical Greek presented to John Chadwick).
II.
GREEK LEXICOLOGY
ΗΡΥΣ - A GREEK GHOST-WORD
JOHN CHADWICK
Downing College, Cambridge
The term 'ghost-word' was invented by T.Skeat in 18661 to describe a word which
has currency in lexica, but was never in actual use in the language. Such words most of
ten arise by mistake, a mis-reading or misinterpretation of a form; but once accepted by a
major dictionary they are liable to be repeated uncritically by other lexicographers. Of
course a ghost-word encountered in a dictionary may be deliberately used by an author,
so that in this case the word springs to artificial life.There are probably a number in the
ancient Greek lexica such as Hesychius, and late authors familiar v/ith these entries, or
more likely their sources, then used them. For instance under the entry Hesychius
remarked: and there are several passages in Hellenistic
poetry where the word is used to mean 'white', though the true meaning of the word in
Homer cannot be deduced from its contexts. But whatever it meant to Homer, by
Hellenistic times it had acquired the ghost-meaning 'white', and this then gained some
limited currency.
Modern dictionaries of ancient Greek are rarely the victims of such errors, for the vo
cabulary has been subjected to intensive study and such forms have been largely elimina
ted. But I believe I have detected a ghost-word, which needs, if not to be banished from
the lexica, at least to be treated with great suspicion until confirmation of its reality is
forthcoming.
100 JOHN CHADWICK
In 1927 the great Austrian linguist Paul Kretschmer visited the Museo Nazionale of
Palermo, and noted some unpublished Greek inscriptions coming from Marsala, the an
cient Lilybaeum. He attributed them to a date of ii B.C., but as we shall see this is plainly
too early. What excited him was the evidence for a new Greek word, and he published a
note on it in Glotta2. He recorded there three short painted inscriptions on small shrines or
naїskoi, in which a dead person was treated as.semi-divine. The clearest example is:
MAPXOC CHMBPONIC ]
There is another repeating the same formula with the personal name But
the interesting example was one in which the name was feminine. It reads quite clearly:
The feminine proves that this is in some sense a dialect text, not
standard κοινή; and it is obvious that HPYC is feminine to . Kretschmer regarded
it therefore as a new dialectal feminine, and sketched an explanation of it as containing ū,
like and therefore the zero-grade of masculine which he reconstructed on
the assumption that the other cases such as ήρωος stood for This theory of the
etymology of can now be confidently discarded, since Mycenaean ti-ri-se-ro-e
(dative) shows no trace of the digamma.
Not surprisingly, the Oxford lexicographers R. McKenzie and H. Stuart Jones recor
ded ἥρνς as a new word and entered it in the ninth edition of Liddell and Scott's Greek
Lexicon, quoting the Glotta article as their source, but prepared to admit doubts on the
position of the accent, since they gave as an alternative. It also penetrated
E. Schwyzer's monumental Griechische Grammatik3. In due course it was entered in
H. Frisk's Etymologische Wörterbuch der griechischen Sprache4 with a reference to an
article by V. Pisani, of which more below; but despite this, no serious doubt of the
word's existence seems to have been entertained by the lexicographer. Likewise it appears
in P. Chantraine's Dictionnaire étymologique de la langue grecque with the sage comment
'sur quoi on ne peut faire que des hypothèses' (p. 417), but again with no doubt
expressed on the word's reality.
Before we examine the evidence more closely we should note a fourth inscription in
the group, unfortunately now hard to read, except for the last six letters. But it is clear that
the personal name is feminine. The text is quoted by various sources in different ways.
E.Gabrici5 confidently gives it complete as:
ΗΡΥΣ - A GREEK GHOST-WORD 101
Others have doubted certain letters, and the most cautious reading is by M.T. Manni
Paraino in her Iscrizioni greche lapidarie del Museo di Palermo, who gives:
remarking that is yet another dialect form for the feminine of This second
variant has not yet penetrated the reference books, and I hope to be in time to prevent it.
There is no reason to believe that the A before belongs to the preceding word;
indeed these last six letters are much clearer, perhaps repainted in antiquity, and stand
separated from the previous text. This encourages me to suggest that we should read
There does not seem to be room for more than two letters after HP[,
whatever the true reading. We have therefore two examples of this strange feminine form.
We must also differ from Kretschmer on the question of date. His dating, which is
unsupported by any evidence, to the second century B.C. is not accepted by later editors,
who have placed these inscriptions in the period i B.C. to A.D. ii.
However, V. Pisani7 must be credited with trying to exorcise this ghost. He pointed
out that the normal feminine is , and suggested that was nothing but an igno
rant spelling of this word. He explained it as due to the development phonetically of ώί to
ωι diphthong and thus to OL, or from to OL and thus to OL. The writing of OL as Τ is
extremely common in late inscriptions, not only from Boeotia where it is as early as iii
B.C. Pisani attempted to bolster this with an interpretation of a gloss, which is much
more speculative. It is of course difficult to find examples of becoming OL, since it is a
very rare group. The long diphthong ωι regularly becomes either ω or OL.
I do not by any means discount Pisani's explanation, but it is clear that it has not con
vinced such eminent specialists as Schwyzer, Frisk and Chantraine. It may therefore be
useful to propound another line of approach. The authors of this group of inscriptions can
hardly have been fully literate in Greek; Lilybaeum is of course in the Punic area, and
Μαρία certainly, and the other female name probably, are not Greek. In any case a literate
Greek would hardly have transcribed the Latin name Marcus Sempronius as
The representation of Latin õ by (accented) ó is evidence that distinctions of

vowel length were no longer observed here. To such an imperfect writer of Greek, the
spoken word was almost certainly unfamiliar, and seeing he probably
102 JOHN CHADWICK
pronounced it as disyllabic, and ignoring the distinction of  and it would have been
indistinguishable from
To write this as is but a short further step. If this explanation is correct, it
would be an example of a mistake due to a spelling pronunciation, a common enough
phenomenon in modern languages, but very rarely demonstrable in ancient times. At all
events, the ghost should be regarded as laid, unless some fresh evidence, such as
an inflected form, is forthcoming.
NOTES
1 Transactions of the Philological Society 1865-7, II 350.

2 Glotta 15, 1927, 306-7.
3 I 346; 479 note 6.
4 I 644 'wohl Neubildung'.
5 Monumenti Antichi (Accademia dei Lincei) 33,1929, col. 53.
6 See further A.M.Bisi, Archeologia Classica 22, 1970, 98-102.
7 1st. Lomb. 73, 1939-40, 490.
ON THE TREATMENT OF SOME GREEK PREPOSITIONS
IN MODERN LANGUAGE DICTIONARIES
PIETRO JANNI
Università di Macerata
The vast and ever more present and alive Greek heritage received into modern
Western languages continues to be treated with the same extreme freedom as in centuries
past, from Humanism to today. The 'Greek' which has increasingly more often
penetrated from scientific terminology and specialized language into common speech and
even into journalism and conversation, is characterised by a grand irreverence in the use
of words and methods of formation, both in formations with suffixes and prefixes and in
forming compounds. This is what has been called "Greek lexicon unknown to the
Greeks" and which Franz Dornseiff, more harshly, has called "gräkoides Esperanto".1
This aims at favouring international comprehension, sacrificing faithfulness to the
language to which it refers and from which it takes its characteristic material.
Just a few examples of each of the categories to which we have referred. In the lexical
field: aboulia (fr. aboulie, it. abulia) ought to continue the Greek αβουλία, well docu
mented in many authors, but has completely altered its meaning, taking that of 'lack of
will-power' instead of 'ill-advisedness, thoughtlessness' or, at most, 'irresolution,
indecision' (according to Liddell and Scott) which was the original meaning in conformity
with the fact that the Greek meant 'will' only in the sense of 'determination,
counsel, design, deliberation'. (And this reminds us that Greek does not have a term
which corresponds to every aspect of the English will or to its correspondents in other
104 PIETRO JANNI
European languages.) A universal term such as biology is the product of modern coining.
It seems to imitate the Greek which, instead, refer to quite a
different area of meaning - to the world of theatre and literature, indicating a realistic and
vivacious portrait of everyday life. For the abstract sense of 'life' as we mean it, the
Greeks would have used, if anything and not
In the field of formations using prefixes and suffixes: anarchy continues exactly the
Greek , but anarchic (fr. anarchique, already found in 1593!, it. anarchico) runs
counter to a precise law of the Greek Wortbildung which prohibits the use of the negative
prefix a(v)- with the suffix -ς. Authentic Greek has only Again, the adjective
2
panoramic, formed immediately after the invention of 'panorama' and its baptism with a
Greek word (1796), applies the same suffix to the nominative form and not to the stem,
as would be correct. A more scrupulous formation would have been *panoramatic.
In the field of compound formation, one example only: the type 'bibliophil', which
goes back to Humanism with some isolated examples and which, from then on, has
become increasingly widespread, formed in complete contrast to the authentic Greek
and to a whole extensive range of parallel compounds where the verbal
element is placed before and not after.
All this is well-known and it would not be worth repeating if it were not just the
preambule to the examination which we wish to conduct of a particular aspect of this
situation: the often inadequate, sometimes frankly mistaken and misleading manner with
which modern language dictionaries refer to Greek to explain origin and meanings of
word derived from that language, either by heredity or by modern coining which employs
Greek elements. Out of the great mass of possible examples, we will choose a few,
relative to words where Greek prepositions appear as prefixes: apo- ana para-.
As we shall see, an attitude of partiality, almost tendentiousness, in the classification
and explanation of items, is common to many dictionaries, misleading the disarmed
reader. This attitude is a result of not recognising an evolution which is, however,
perfectly clear and which ought to be completely intelligible, indeed almost predictable for
whoever is concerned in whatever way with the history of words and the history of
language. This evolution stems from the fact that the Greek prepositions which we have
quoted, like many others words in general, had in the original language a vast range of
meanings and uses, sometimes notably distant from each other. When words in modern
GREEK PREPOSITIONS IN MODERN LANGUAGE DICTIONARIES 105
European languages began to be coined with Greek material, only one of these meanings
usually prevailed. This was accepted by the common linguistic conscience while all the
others were not revived nor used in new coinings. Knowledge of them remained the
patrimony of those who really studied the Greek language and not of those who were
content to accept modern 'Greekoid'. The following situation was therefore created: on
the one hand a generally limited patrimony of words inherited, one way or another, from
the 'true' Greek vocabulary - in this group survive, more or less conspicuously, more or
less clearly, traces of the original use; on the other hand, there are modern coinings whose
history is marked by a kind of 'bottle-neck' - among all the original uses, only one
survives or has come back to life. In recompense, it has become the point of departure for
a sometimes huge quantity of new formations which have made this use familiar to all and
able, at times, to penetrate into common speech, into sub-culture or, indeed, into popular
language itself. A perfectly intelligible situation, as we said before. The fault of the
compilors of almost all our dictionaries is that of not clarifying this and not putting the
reader on the right track. Worse still, hiding this situation with systematic inexactitude
which tends to flatten the historical perspective and to muddle up precisely those things
which most need to be picked out and distinguished.
Let us take one of these prepositions/prefixes - As a prefix in noun compounds
and as a verbal prefix, it has a range of uses and shades of meaning greater than as a
preposition. Placed before a verb, it gives a 'perfective' meaning of fulfilled action, action
approaching its end, action seen in its efficacy, action which succeeds in changing a state.
So, if is generically 'to show, to make known', is more precisely
'to demonstrate actively, to convince'. As in many analogous cases even here the verbal
prefix, frequently used, sometimes becomes pallid and weak, so much so that the
compound has greatly replaced the simple, at least in prose. In fact, and
are more common that their respective simple verbs.
So much for the use of άττο- as an 'empty' verbal prefix. As a 'full' verbal prefix, it
indicates distancing, removal, or receiving something from somebody. Modern
compounds which use the apo- prefix, all, without exception I think, refer to the 'full'
meaning, more concrete and palpable, able to be referred back to the values of derivation,
distancing, separation, privation or simply negation. Botanical terminology offers various
examples with term such as apopetalous and aposepalous, said of flowers which have
106 PIETRO JANNI
their petals or respectively their sepals well distinct and separate from each other.
Mineralogy has apophyllite, the name of a mineral which tends to 'exfoliate', to break into
thin flakes. Physics has apochromatic, said of lenses correct for three colours of the
spectrum and therefore even more 'lacking in colour (parasitic)' than the more common
and less perfected achromatic. All these are modern coined technical and scientific words.
Beside them, a group of words of authentic Greek origin where the primitive 'perfective'
value of the apo- prefix is still clearly recognisable, exists in all European languages. It is
a much less sizeable group but in recompense it includes such better known and more
common words in everyday speech as apodeictic, apologue and apotheosis, none of them
can be connected to 'distancing', 'privation', and so on, meanings. Let us now see how
this series of facts is reflected in one of the most recent and authoritative French
dictionaries (perhaps the most authoritative of all), the Trésor de la langue française
(1974).
Here we find the lemma apo- defined as "élément préfixale". The explanation is
divided into two sections, A and B. The former is further sub-divided into two sub
sections. Under A, we read that apo- expresses the notions of "écartement, séparation,
cessation, négation". (All terms which are to be found punctually in later French
dictionaries.) Under sub-section 1, we read again that apo- "marque l'écartement,
l'éloignement, la modification, la séparation", and apocarpe, garnie, apomixie and
aposporie are given as examples; in sub-section 2, "une valeur privative, négative" is
distinguished (examples: apobiose, apochromatisme, apodactilique, apophylaxie). Section
 refers to a completely different use in which the prefix apo- serves, in chemistry, to
coin names of compounds derived from substances indicated by the root noun:
apocodéine, apoenzyme, apomorphine.
Thus far so good, as long as it is understood that all this is valid for the apo- of
modern compounds and absolutely not to be confounded with the of authentic Greek
compounds. But the Trésor adds a section on 'Etymologie et histoire' where this
confusion appears in its worst possible form. Here, in the most exemplary fashion, the
two deadly sins of almost all modern language dictionaries are committed: 1) there is not a
sufficiently clear and distinct differentiation between real, inherited Greek and modern-
coined 'Greekoid'; 2) things are made worse by trying at all costs to bring Greekoid back
to Greek, i.e. to explain even the most arbitrary and most removed from the Greek
Wortbildung of modern coinings as if everything in them were correct and regular from
the Greek point of view.
Here, this behaviour is made even more obvious by the fact that Pierre Chantraine's
Greek etymological dictionary is quoted, from which, indeed, a correct piece of
information is derived, only to be immediately distorted. It begins saying that apo- is a
"préfixe emprunté au grec από qui exprime l'idée d'éloigner, écarter". (This is already
unilateral and reductive, as we have observed.) Then Chantraine is quoted: "άπο- comme
έκ- marque l'aboutissement du procès... aboutissement (qui) peut être une fin, une
cessation...; enfin le sens peut devenir privatif, négatif." And the editor of the Trésor
explains: "Ces valeurs se retrouvent dans les composés étudiés supra sous A." This way
of interpreting a source is completely superficial because it ignores precisely the key word
"aboutissement du procès", indicating a primary and central value for one from
which many of its uses derive. And this value is still reflected in many words in European
languages (in this case, French), where the presence of apo- is destined to remain
unexplained to whoever consults the dictionary. All the claimed agreement between what
is to be read under section A of the item and what is now to be found in Chantraine is
approximative.
But this is not all. The Trésor quotes Chantraine again: "Dans qq. formation nom.
άπο- semble exprimer la notion 'une espèce de', avec nuance péjorative", and comments:
"Cette valeur est celle étudiée supra sous B." Here, we are to believe that in inventing
apocodéine, etc., the chemists have dug out such a rare and uncertain use of άπο- which a
Greek dictionary like Chantraine's introduces almost as an appendix and in dubitative
form! It is clear, instead, that in the case of modern compounds, the sense of 'derivation'
has been excavated without fatigue from that of 'distancing', 'origin', and 'provenance',
known to all by the modern use of the prefix. And the "nuance péjorative"? The search for
this, at least, has been dropped.
Once having seen the basis upon which the 'Greek' explanation of modern
compounds with apo- is said to be founded, it is only too easy to imagine what to expect
under each single item. With insistent repetitiveness explanations continue to be given of
what there is no need to explain, i.e. all the cases in which the prefix apo- comes under
the impoverished and rigid special applications illustrated under the general item and va
lid only for compounds formed in modern times. Silence, or worse, forced explanations
108 PIETRO JANNI
are given when words of authentic Greek origin reflect various uses of apo- because they
do not fit into the scheme.
The same things are to be found in another large and even more recent French
dictionary: Le Grand Robert de la langue française (1985). Here, too, there is the apo-
lemma, which is more succintly defined in proportion to the minor size of the work:
"Préfixe, du grec apo-, qui indique l'écartement, la provenance, la séparation, la
distinction, l'intervalle, la cessation, la privation ou la négation." Here, too, the
lexicographer is content with being able to demonstrate with examples that apo- has the
meanings already given by him, 'privation' and 'négation' for apo chromatique, apogamie
and apomixie; 'éloignement', from which 'provenance', in the case of apomorphine.
Silence, however, for apocalypse and, with greater reason, for apodictique,
apophthegme, apoplexie and aposiopèse, for which no etymological explanation is given.
Under apothéose, instead, there is an analysis into apo- and theós but the value of apo- in
the compound is not explained and there is no indication of how it is to be connected to
the meanings listed under the general lemma. This desire to explain with rational and
transparent derivations what is, instead, the product of history and which, if anything,
requires a historical note, reaches its apex in the case of apo, "plante fatale aux
chiens", which is to be derived "de apo 'contre' [!] et kuôn [sic]".
We have looked at two French dictionaries to begin with because in this language
Greek neologisms are particularly numerous and because two ample and recently
published lexical works are available; but things are not much different elsewhere. The
Grande Dizionario della Lingua Italiana, founded by Salvatore Battaglia and in course of
publication, does not have an apo- lemma like the two French dictionaries but in
everything else behaves in much the same way. We find both apocatastasi, which is true
Greek, and apocarpo, which is Triballic Greek, coined by modern botanists, classified as
"voce dotta", wihout any distinction being made. Here again, the etymology and value of
apo- is unmentioned in the former case, where it cannot be included in the usual 'origin'
etc., while in the latter case the 'distancing' meaning has not been left out. For apostolo,
the Greek is quoted, while nothing is analysed in the case of apologo,
where the usual 'distancing' and 'separation' are irrelevant. (If things were as described
by the dictionaries, should mean approximately 'nonsense'.)
The case of the Robert's dictionary which defines apo- as 'contre' is paralleled in the
Battaglia dictionary which defines apo- as 'dopo' in the item apotesi, "ultima cadenza,
fine della scena, del coro". The reader is forced to observe, in the end, once again that the
dictionaries never tire of explaining a hundred times over precisely that which needs little
explanation, i.e. that which is most clearly within modern and standardised use, while
nothing is said on other uses, inherited from authentic Greek and still alive in perfectly
French and Italian words, which have the right, no less than the others, to be explained
etymologically, if the dictionary sets itself this task.
It is correct to note at this point that things go very much better in the Oxford English
Dictionary, even if with some imperfections. Here, in the item apo-, a net distinction is
made between "compounds already formed in Greek or others analogous to them" and
"modern scientific words, not on Greek analogies". This is the foundation for any correct
and not misleading or illusory, explanation. Under the item apocarpous, where apo- is
used in a sense that has little to do with real Greek, the fact that it is a modern formation is
clearly underlined. Under apotheosis, the reader is finally put on the right track, at least,
with the explanation that apo-, in compounds, can have the value of "completely".
Returning to the Trésor (1986), let us see how the other prefix para- is treated. This
time, too, there is an entry devoted to it, divided into two sections A and B. In the former
it is said that para- "exprime l'idée de contiguïté, de proximité spatiale ou d'appartenance à
un domaine proche"; in the second, however, we find that it "exprime l'idée de
ressemblance plus ou moins prononcée, souvent trompeuse, de marginalité ou même
d'anomalie". As usual, it is not made clear that in many French word inherited from the
Greek, para- does not only express a false similarity or 'anomaly' but a complete
unlikeness and disagreement, conforming to one of the uses of the Greek π- in words
like πμi which simply means 'illegality' and not 'apparent legality' or 'near-
legality'.
The result, when we arrive at the explanation of single items, is similar to what we
have already seen in the case of apo-; only that which conforms to the rigid and
impoverished scheme valid for modern compounds, is explained, and the rest is silence.
Paraphernal is analysed with a quotation from Liddell-Scott-Jones, explaining that here
para- means "auprès de, à côté de, en dehors de", while no attempt is made to explain
paranoïa.
110 PIETRO JANNI
This time, again, the OED emerges very much better from this test, even if faults are
not lacking here too. But at least under the para-lemma we can read that in compounds it
may have the meaning of "amiss, faulty, irregular, disordered, improper, wrong", or
indeed express "perversion". Even here what we may call excess of etymological zeal
appears when a modern language dictionary wants to explain the semantic developments
internal to Greek itself, and in general does so with unfortunate results, vitiated by the
usual tendency to carry an ample and multiple product of linguistic history back to a
unique origin. One example is the item paralysis, made to refer back to the Greek
explained in its turn as "to loose from beside, disable, enfeeble, f. πaρα- beside
to loose". A doubtful and useless analysis which certainly ought not to be the job
of a modern language dictionary. Even more incautious is what is to be read under the
item paroemia, which seems to want to explain par- in the compound by suggesting as a
gloss "by-word" and following up with "proverb, f. by the way, f. π. by +
way, road". If we look up a specialised dictionary such as Hjalmar Frisk's
Griechisches etymologisches Wörterbuch, we find that the compound is
defined as "semantisch unklar", and it is certainly such for any prudent judge.3 The
modern lexicographer, however, undauntedly forces everything onto the Procrustian bed
of his scheme.
Again, in the OED, parrhesia is explained as a compound of "παρά, beside, beyond
speech". Here it is impossible to understand why such a compound should
mean "free-spokenness, frankness". In fact the correct solution is probably to be found
elsewhere and in the first element of the compound and not παρά should be
recognised.
Here we stop. But not for lack of material. A research such as this could be extended
to the treatment of all the prepositions and all the Greek words in modern language
dictionaries. What has happened to that fortunate prefix meta- since Aristotle's
'Metaphysics' has begun to be quoted4, would be particularly instructive.
Someone could say that it is too much to ask. Modern dictionaries cannot be
exhaustive and scrupulous about Greek. But this is not our criticism which is directed
against an acritical use of Greek. The dictionaries base themselves on an impoverished
and false Greek, fooling the reader with illusory explanations when they ought, instead,
simply note the complex and problematic character of certain facts and refer him to
specialised dictionaries and works for a more complete information.
The small sample of a critique which we have traced here could be continued and
expanded into a whole chapter of our cultural history if we wanted to show how this
acritical use of Greek goes directly back to the mania for Greek in the first etymological
dictionaries of European languages, in their turn children of a more ancient tradition: a
monumental example is Egidio Menaggio's Italian etymological dictionary.5
But we have already exhausted the patience of those who have followed us this far.
The illustrious scholar to whom these pages are dedicated has shown how linguistic
research can rove across continents and centuries of history. We hope he will accept in
homage this small inquiry which roves only between the walls of libraries and the pages
of books.
NOTES
1 F. Dornseiff, Die griechischen Wörter im Deutschen, Berlin 1950 (=Kleine Schriften II. Sprache und
Sprechender, Leipzig 1964, 294-317).
2 On panoramic, see Dornseiff, Die griech. Wörter, 146.
3 S.v. π α ρ ο ι μ ί α .
4 On metaphysics, see the articles of H. Reiner in Zeitschrift für philosophische Forschung 8, 1954,
210-237; 9,1955, 77-99, and . Kremer, Der Metaphysikbegriff in den Aristoteles-Kommentaren der
Ammonios-Schule, Münster 1961, 5 ff.
5 Egidio Menaggio (Gilles Ménage), Origini della lingua italiana, 1669, second edition 1685. On Greek
in Menaggio's Italian etymologies, see M. F. Ferrini in Annali della Fac. di lettere e filos. dell'Univ.
di Macerata 19,1986, 111 ff.
FANTASIA IN PANTOFOLE
GIULIO LEPSCHY
Reading University
Mi propongo di presentare, in un volume per un maestro quale Oswald Szemerényi

del quale sono stato, molti anni fa, allievo (in senso proprio, a University College, a
Londra, e non solo nel senso generale in cui tutti siamo allievi degli studiosi da cui
impariamo), non già un contributo all'indoeuropeistica, che sarebbe un atto di
presunzione da parte di chi, come me, si occupa di linguistica generale e italiana, bensì
qualche appunto che ho raccolto per una lezione sulla storia della parola (e della nozione
di) fantastico per il Corso internazionale su 'Gli universi del fantastico' alla Fondazione
Cini, nell'Isola di San Giorgio Maggiore a Venezia, nell'autunno del 1986, e che spero il
dedicatario della miscellana possa trovare di qualche interesse, anche se mi limito a
mettere insieme fatti noti (sebbene non tutti immediatamente accessibili), e non faccio
nessuna proposta etimologica nuova. Alla fine dell'articolo espongo alcuni dubbi, e
formulo qualche perplessità, a cui le risposte (che non do, e neppure cerco di dare) non
sono forse ovvie.
La famiglia dei termini 'diventare visibile, apparire', 'render
visibile, rappresentare', 'apparenza, immagine',
'capace di rappresentare, di formare immagini', è derivata, con
un ampiamento in τ, da una radice verbale φαν-, che compare nel presente con jod
'render visibile, mostrare, indicare', 'essere visibile, comparire, apparire', e in
greco moderno 'apparente', e 'il fenomeno'. La radice è la
114 GIULIO LEPSCHY
stessa di luce', e viene collegata1 a un indoeuropeo che si ritrova fra

l'altro nel sanscrito bhä-ti 'risplendere, brillare', e che alcuni studiosi (Specht, Fraenkel)
vogliono associare con il 'crescere, divenire' fui, futurum ecc.,
antico slavo byti, russo byt', e forse col latino futuo 'copulare'.
A questo punto è interessante notare che gli indoeuropeisti hanno ricostruito anche
un'altra radice 'parlare', che è alla base del latino fama, fateor, for, fas, fatum,
fabula, facundus, e del greco
affermare, credere', e " 'stregone', 'incantesimo',
'stregare', latino fascinum 'fascino, il fallo, con valore magico come protezione contro gli
incantesimi'. È stato discusso, in particolare a proposito di e fascinum, il
collegamento di questa radice con quella di tipo espressivo onomatopeico *baba-, che
indica il discorso inarticolato, balbettante (anche nelle forme bal-bal-, bar-bar-) e si ritrova
nel latino balbus, balbutio, nell'inglese to babble, greco 'blaterare,
gridare', italiano babbo, inglese baby, antico slavo baba 'vecchia'.
Come ci si può aspettare, la tentazione di accostare, e poi di identificare queste due
radici si rivela irresistibile. Walde e Pokorny nel loro vocabolario comparativo
indoeuropeo citano studiosi come Curtius e Persson, secondo cui la radice che indica
'brillare' e quella che indica 'parlare' sono la stessa, poiché questi due significati si
trovano associati anche in altre radici; Frisk nel suo vocabolario etimologico greco osserva
che queste due radici convergono non solo perché sono fonicamente identiche, ma
anche perché il passaggio da 'illuminare, chiarire' a 'parlare' è semanticamente bene
attestato, come nel latino declaro. E Chantraine nel suo vocabolario etimologico greco
rinvia addirittura ad una radice unica che significa 'chiarire' e 'spiegare, parlare',
osservando come l'ambivalenza semantica si faccia sentire nel greco stesso:
'apparenza, impressione, presentazione, esposizione, significato, e, in retorica, enfasi';
'vista attraverso', 'essere intravisto, degli occhi fra le palpebre, in un
dormiente; sintomo, segno'; " ' sinonimo di ύπερφανiα 'arroganza';
συμφανσις 'brillare insieme (di stelle)' e 'apparenza di incontro, congiunzione'
vanno, almeno in sincronia, con il Liddell e Scott dà due voci separate,
una collegata a che significa 'denuncia, informazione, e l'apparire (di un astro: la
fase)', l'altra collegata a che significa 'espressione, discorso, dichiarazione,
proposizione, giudizio, sentenza'. Anche per il Liddell e Scott dà due voci
FANTASIA IN PANTOFOLE 115
separate, una basata su che significa 'sentenza, decisione; affermazione, giudizio',

e parallela ad 'dichiarazione, predicazione' ( c f r . ' d i c h i a r a t i v o ,
predicativo', nel linguaggio filosofico), e l'altra basata su , che significa 'negazione'
e si oppone a 'affermazione' (e cfr. 'contraddizione'). Questi
valori opposti di ('affermazione' e 'negazione'), che sembrano creare un caso
di enantiosemia (Lepschy 1981), vengono sfruttati da Aristotele che scrive nel De
interpretatione (17 a 25): (la negazione)
la predicazione di una cosa via da un'altra - appunto la sua negazione (Aristotle
122)).
E' notevole che in altre occasioni il Liddell e Scott non presenti due voci diverse, ma
all'interno di una stessa voce faccia risalire due significati distinti ai due verbi in
questione: per abbiamo un significato 'dichiarazione', che viene collegato a
e un significato 'emergenza, ricomparsa', che viene collegato a (per
un significato unico 'esibizione, manifestazione', evidentemente legato a
. Ci sono casi in cui, secondo Chantraine, 'è difficile e probabilmente vano'
decidere se bisogna risalire a per es. 'che appare
da lontano, la cui fama arriva lontano, rinomato, predetto, preannunciato',
'inatteso, imprevisto, indicibile, tremendo'.2
Come si vede, attraverso la radicedifantasia ci siamo collegati alla luce e alla visione,
al parlare e al linguaggio, non senza incontrare, lungo la via, il fallo e il nome del padre:
un vero paradiso lacaniano, a proposito del quale vien fatto di pensare che quelli che
aspirano ad arrivarci farebbero forse meglio a seguire gli irti sentieri glottologici della
grammatica comparativa indoeuropea che non la strada più larga e facile dellafilosofiadel
linguaggio.
Più seriamente, il collegamento fra la luce e il parlare lo trovo illustrato in un brano
suggestivo di un autore contemporaneo fra i più sensibili alle risorse fantastiche del
linguaggio, Luigi Meneghello. In un saggio recente dedicato all'esame di certi aspetti
linguistici della creatività poetica, Meneghello (1986:13-4, 1987:92-3) commenta
un'espressione che aveva usato in passato parlando del suo lavoro: 'vorrei far splendere
quella sgrammaticata grammatica', e scrive:
Quanto poi al proposito di 'far splendere' la mia materia, riconosco che c'è in me il pregiudizio
che il vero splenda, che realtà e splendore siano cose associate. Qui vorrei citare dei versi, anche
perché rappresentano molto bene l'altro polo dei miei interessi,rispettoalla materia dialettale. Mi
116 GIULIO LEPSCHY
riferisco non solo a ciò che scrivo, ma anche a ciò che sento, in generale. A un estremo c'è
l'esperienza paesana, con tutto ciò che comporta; all'altro estremo c'è una passione ugualmente
intensa che non c'entra col fatto che sono veneto, una passione per la poesia. Si tratta di alcuni di
versi più belli che siano mai stati scritti in italiano (e vi dico onestamente che questa bellezza mi
importa non meno di qualsiasi altra cosa), e si dà il caso che parlino appunto dello splendore del vero.
La persona che parla ha appena ascoltato una spiegazione, e la sua mente si schiarisce some il
cielo da cui il vento sgombra le nuvole:
Come rimane splendido e sereno

l'emisperio de l'aere, quando soffia
Borea...
per che si purga e risolve la roffia

che pria turbava, sì che 7 ciel ne ride
con le bellezze d'ogni sua paroffia;
così fec'io, poi che mi provide

la donna mia del suo risponder chiaro,
e come stella in cielo il ver si vide
È un brano che ha per me una straordinaria forza di suggestione, particolarmente alcune parole:
'roffia' per esempio, che io intendo per 'sporcizia, sudiciume' e penso sia la parola 'rufa' del nostro
dialetto, che era scomparsa nella mia memoria tanti anni fa, e recentemente è tornata fuori. Ecco,
questa è una di quelle parole che, potendo, vorrei trovare il modo di far splendere in un contesto
appropriato. "Rufa' mi pare bellissima, con le sue misteriose associazioni infantili: era lo sporco sulle
braccia, sulle gambe, la patina di terra, di fango, di giochi selvatici...
Ho citato questo brano non solo perché mi pare molto bello, ma anche perché presenta
in maniera mirabile l'associazione fra parlare e vedere, fra verità e luce, nel 'risponder
chiaro' di Beatrice. Fino a che punto le connessioni etimologiche rimangano presenti,
nella storia di una parola, è una questione non pacifica, che era stata troppo perento
riamente risolta e spazzata via da molti linguisti, con la frattura fra sincronico e diacronico.
È curioso che uno studioso moderno (Bundy 1927:70) osservi, a proposito di un passo in
cui Aristotele (De anima, 429 a 3) connette 'luce', che tale etimologia è
Ovviamente sbagliata'. A quanto pare non lo è affatto, né diacronicamente, né per la
coscienza linguistica di Aristotele. Ma questo è un problema che ci porterebbe troppo
lontano.
Allo sguardo etimologico che abbiamo dato all'indietro, in direzione dell'indoeuropeo,
aggiungerò ora un'occhiata in avanti, verso il periodo postclassico e moderno. Qui
osserviamo che ha avuto una fortuna notevole e per certi aspetti sorprendente,
all'interno del latino, e poi dei volgari europei, non solo romanzi. In latino phantasia era
penetrato nella lingua popolare, come si vede dall'espressione proverbiale, in Petronio
(38, 15), usata per uno dei commensali di Trimalchione, ora impoverito, ma che soleva
mangiare come un re, e lasciare spandere sotto il tavolo più vino di quanto un altro abbia
in cantina: 'phantasia, non homo' (Petronius 1961:37). Nel latino parlato si deve essere
diffusa una forma (non attestata) senza aspirazione pantasia col significato 'visione',
'paura, angoscia, incubo', e il verbo pantasiare 'avere delle visioni, degli incubi'. Si
pensa subito al Contrasto di Cielo, in cui il corteggiatore lamenta che la sua anima 'dì e
notte pantasa' (Contini 1960:182); e nei vocabolari etimologici troviamo ampia documen
tazione: provenzale antico pantaisar, pantaiar 'sognare, avere un incubo, essere oppressi';
francese antico pantais(i)er, pantoisier 'avere il fiato corto, lavorare faticosamente,
palpitare' (da cui l'inglese to pant 'ansimare'); e varie forme dialettali collegate, francesi e
italiane: dal veneto pantezar 'ansimare',3 al napoletano pandeke 'incubo', all'abruzzese
pandóše 'tosse violenta" (Meyer-Lübke 1935: nn.4274-6, 6458-60; Wartburg 1957:
360-5)
Per scendere dal fantastico al prosaico (come argomento),  per salire dal prosaico al
fantastico (come metodo) possiamo ricordare un articolo in cui Leo Spitzer (1924, e cfr.
1928) cerca, in maniera, come sempre, fascinosa, ma, devo dire, senza riuscire ad essere
molto convincente, di ricavare le pantofole dalla fantasia. C'è da pensare che su un
ingegno sottile e paradossale come quello di Spitzer debba aver agito il piacere della
contraddizione, nel collegare alla fantasia qualcosa di tanto poco tantastico quanto le
pantofole.4 L'italiano pantofola e il tedesco Pantoffel, dice Spitzer, sono presi dal
francese pantoufle, e questo va spiegato all'interno del francese, poiché si è rivelata
inattendibile la tradizionale etimologia greca. Il suffisso -oufle sarà un ampliamento
scherzoso, appartenente alla serie ermoufle 'ipocrita', boursouffler 'gonfiare',
emmitoufler 'imbacuccare', maroufle 'imbroglione' e 'babbeo, balordo', e nel francese
popolare panoufle 'parrucca'. Nel francese meridionale si trova pantouqueto 'moglie 
figlia di contadino' e 'rumore che fanno i tacchi di legno': il significato originario deve
riferirsi a qualcosa di 'pesante, rustico', come in rester pantois 'stupefatto', e francese
meridionale pantés 'rozzo, zotico, villano', panto, e argot pante 'goffo, impacciato', che
appartengono alla serie di pantaia (da p(h)antasiare) 'sognare, ansimare, ansare'. Il
118 GIULIO LEPSCHY
suffisso -oufle si richiama a parole che indicano da un lato l'infagottamento {emmitoufler,

panoufle), e dall'altro la balordaggine (maroufle): le pantoufles erano infatti zoccoli,
rustiche calzature con la suola di legno usate dai contadini.
Se questa derivazione da fantasia non convince, bisogna dire che l'etimologia della
parola non è chiarissima. Spitzer inizia il suo articolo citando un contributo di pochi anni
prima, di Hesseling, che avrebbe mostrato che il leggendario 'tutto
sughero' da cui tradizionalmente si fa derivare pantofola, è un'inven
zione dell'umanista francese Budé. Di fatto Hesseling (1921) citava un testo settecentesco
che attribuiva l'etimologia al Budé, ma dichiarava di non essere riuscito a rintracciarla
nelle opere dell'umanista. In realtà la parola si trova nei Commentarli Graecae Linguae:
ab illis suber dicitur: unde nos pantophellos appellamus crepidas, quarum solum
subere constat: quibus foeminae utuntur, ut proceriores videantur: interdum etiam viri ob
teporem ligni. Cuius dictionis olim me admonuit Ianus Lascaris vir Graecus iudicio
praestans, atque in utraque lingua eruditione eximia praeditus' (Budaeus 1556: col. 288).
Budé aveva finito nel 1529 questi Commentarli che sono una delle opere più importanti
della filogia contemporenea, e in essi Lascaris è l'unico dotto vivente ad essere citato: una
volta a proposito dell'interpretazione di un passo delle Leggi di Platone, e un'altra nel
brano che abbiamo riportato. Che cosa essattamente Lascaris abbia detto a Budé non è
chiarissimo: se cioè gli abbia suggerito di vedere nella parola francese pantoufle
(presentata qui in forma grecizzata),  se gli abbia fornito una parola bizantina, 
neogreca, che starebbe alla base delle forme francese e italiana (cfr. Knös 1945:209-10).
Già Diez (1853:250) aveva respinto l'etimologia greca, e aveva proposto di risalire a
patte 'pianta del piede' per la prima parte della parola, e al suffisso che si trova in
manoufle e emmitoufler per la seconda. Ma l'etimologia cheBudéattribuisce4al Lascaris
era stata, nello stesso anno 1921 in cui Hesseling pubblicava il suo articolo, accolta dal
Boisacq (1921, con un rinvio a Labbé 1661) e si ritrova nel REW di Meyer-Lübke, con
un rinvio a Benedetto Varchi (1570:158), che nell Hercolano citava la parola come
esempio delle discordanze fra gli studiosi di etimologia: 'Pantufola per quella sorta di
pianelle, che hoggi alquanto più alte dell'altre, si chiamano, mule, diriva secondo
cotestoro dal Greco, ma altri d'altronde le dirivano, come il Carafulla5 da piè in tu fola, e
anco può essere, che havendo le cose in sé diverse proprietà, questi ne consideri una, e da
quella la dirivi, e quegli un'altra, e da quella voglia che detta sia', dove si ammirare
l'apertura a considerare una sorta di sovradeterminazione etimologica. Il termine incurio

siva, evidentemente, gli studiosi del Rinascimento. La parola è attestata dalla seconda
metà del 400, e la forma pantophilae si trova citata in quello straordinario vocabolario
domestico che è il De partibus aedium del parmigiano Grapaldo (1494?:n iiir).
A lui rinvia il Du Bois (Sylvius 1531:29) con la forma su cui il Labbé
(1661:372) commenta: 'qui est tres-utile tres-commode, particulièrement aux
gens d'étude'. Anche Juan de Valdés nel suo Dialogo de la lengua (1535: Boehmer 349,
Montesinos 22) citava pantuflos come una parola d'origine greca. Oggi alcuni linguisti
restano fedeli all'etimologia greca di Lascaris e Budé,6 mentre altri si dichiarano incerti,7
 rinviano dubitativamente alla famiglia di patte 'zampa',8 da una base *patta (Meyer-
Lübke n.6301), da cui deriva anche la forma patois 'idioma grossolano, rustico; dialetto',
e cfr. il corso patone 'schiaffo', e presumibilmente il veneziano patón 'colpo, con la mano
aperta, sul volto  sulla testa'.
Forse varrebbe ancora la pena di chiarire i rapporti reciproci (di eventuale priorità e di
direzione del prestito) fra la forma italiana e quella francese,9 e di seguire,
semasiologicamente, i rapporti fra la parola e la cosa: gli zoccoli di legno, rustici, a cui
pensava Spitzer, gli zoccoli alti, urbani,  pianelle, portati da certe donne veneziane (come
si vedono nel Carpaccio, e sono descritti con stupore dagli stranieri che visitano Venezia
nel Cinquecento), fino alle pantofole moderne. Collegare meglio la parola alla cosa
potrebbe anche facilitare un chiarimento etimologico. Occorre dunque un supplemento di
indagine, che converràrimandaread altra occasione.
Desidero invece aggiungere qui qualche domanda, che mi sono posta mentre seguivo i
percorsi, a volte poco lineari, che gli studiosi hanno tracciato per questa parola,
dall'antichità fino ad oggi. Il collegamento fra fantasia e pantofola sembra di primo acchito
poco verosimile, sia fonologicamente, sia semanticamente. Per quanto riguarda l'aspetto
fonologico, dovremmo forse tener presente che non sempre le parole seguono la regolare
trafila postulata dalle leggi fonetiche, e spesso la loro evoluzione è soggetta ad interferenze
che contribuiscono a rendere specificamente individuale e diversa dalle altre la storia delle
singole parole; si troverà un'abbondante esemplificazione scorrendo, per esempio, il
vocabolario etimologico italiano di uno studioso attento agli 'incroci' fra parole diverse,
nel corso della loro evoluzione, come Giacomo Devoto (1966). Forse la cosa non è
troppo preoccupante se pensiamo che perfino in discipline più 'dure', come la fisica, i
120 GIULIO LEPSCHY
fenomeni osservabili non illustrano generalmente una particolare legge quanto piuttosto la
complessa (e non facilmente specificabile) interazione di leggi diverse. Sarà forse vero
che, come dice il proverbio, non cade foglia che Dio non voglia, e che, come dice
Einstein, 'raffiniert ist der Herrgot, aber boshaft ist er nicht'. Certo le singole foglie,
cadendo, non paiono illustrare nel modo migliore la legge della caduta dei gravi; ma
neppure la contraddicono. Così l'evoluzione fonologica di una parola, pur soggetta al
interferenze relative alla sua storia particolare, non contraddice le leggi fonetiche. Più
preoccupante è l'aspetto semantico, se è vero che non è chiaro fino a che punto i rapporti
semantici fra due parole siano classificabili entro categorie rigorose che consentano di
postulare  di escludere una parentela genetica fra i due termini.
Le incertezze sono anche maggiori quando si tratta di decidere se due radici (come i
due *bhā- citati sopra, uno che significa 'brillare' e l'altro 'parlare') vadano trattate come
forme separate  considerate come una radice sola, e se e come vadano collegate ad altre
radici (come *bhu- 'crescere', o *baba- 'balbettare'). Il lettore può avere l'impressione
che attraverso questa nozione di 'collegamento' fra radici rientri per la finestra quel
cratilismo che l'indoeuropeistica aveva, in un lungo e faticoso processo, vittoriosamente
cacciato dalla porta. Forse potrebbe essere utile, per chiarire non solo un punto di storia
della linguistica, ma anche certi percorsi concettuali nella pratica quotidiana dei
comparatisti, uno studio preciso della terminologia usata per descrivere i rapporti reciproci
tra forme di cui si postula la parentela (per esempio, la nozione di 'collegamento' sembra
implicare qualcosa di meno impegnativo di una derivazione genealogica).
Interessanti mi sembrano anche, e non solo dal punto di vista della grammatica storica
e comparata, ma anche da quello della lessicologia e della lessicografia, i criteri che
portano un vocabolario della solidità e dell'autorevolezza del Liddell e Scott, date due
forme omografe di significato diverso, a considerarle in certi casi lemmi distinti, e in altri
accezioni distinte di uno stesso lemma: il lettore può restare sorpreso che il criterio
etimologico non sia sempre quello cruciale, come abbiamo visto sopra per Si
noti che non si tratta dell'indeterminatezza di cui parla Chantraine, per cui può essere
impossibile decidere se una parola va collegata a né della riduzione delle
due radici a una sola, bensì del far risalire due accezioni diverse di uno stesso lemma a
due radici etimologicamente distinte.
NOTES
1 Anna Morpurgo Davies, che ringrazio per i commenti fatti a questo articolo, mi fa notare che il
collegamento, fra le forme con η e quelle senza , è di fatto più problematico di quanto possa apparire
dai vocabolari etimologici.
2 Vedi Walde 1927:105-6, 122-3; rn 1959:91,104-5; Frisk 1970 s.v.
Chantraine 1980, s.v. W alde-Hofmann 1938-54, s.v. fabula, fas Jateor,
fatum, for, fui, futuo; Ernout-Meillet 1959-60, s.v. fas, fuam, for, futuo; Liddell & Scott 1953, alle
voci citate nel testo.
3 Ma, come mi ricorda Paola Benincà, si è proposta anche una derivazione diversa, dà pantex (da cui, al
plurale pantices, l'italiano ρancia): vedi i riferimenti in Marcato 1982:112.
4 Una bella associazione, cratilistica invece che glottologica, con un oggetto apparentemente poco
fantastico come le Ferrovie dello Stato, viene introdotta da Ceronetti 1985:182, per il quale già nella
sigla FF.SS. 'sembra frusciare un fantasma'.
5 Sulla curiosa figura del Carafulla v. Ageno 1959 e Woodhouse 1970. Renzo Bragantini sta
raccogliendo interessanti documenti nuovi sul Carafulla che speriamo di vedere presto pubblicati.
6 Vedi la voce in Prati 1951, Olivieri 1961, Devoto 1966.
7 Vedi la voce in Battisti-Alessio 1950-57 e Migliorini-Duro 1953.
8 Vedi Bloch-Wartburg 1964, Dauzat et al. 1964, Cortelazzo-Zolli 1979-, l'ampia presentazione in
Corominas 1954:645-6, e la difesa della derivazione da patt- in Keller 1958.
9 Secondo i vocabolari la forma francese è attesta dal 1465, quella inglese, come quella tedesca, dal
1494, quella catalana dal 1498, quella spagnola dal 1519; per l'italiano il Battaglia dà citazioni dal
Cammelli (pantofle), dal Bibbiena (pantufole), dall'Ariosto (pantofole), dal Castiglione (pantojfole)
ecc.
OPERE CITATE
Ageno, F. 1959. Un personaggio proverbiale: il Carafulla. Lingua Nostra 20.1-3.

Aristotle. 1955. The Organon, a cura di H.P. Cooke e H. Tredennick, Loeb Classical Library. London:
Heinemann, New York.
Battaglia, S. ed. 1961-. Grande dizionario della lingua italiana. Torino.
Battisti, C , Alessio, G. 1950-57. Dizionario etimologico italiano. Firenze.
Bloch, O., Wartburg, W. von. 1964. Dictionnaire étymologique de la langue française. Quatrième édition.
Paris.
Boisacq, É. 1921. Français pantoufle. Bulletin de la Société de Linguistique de Paris 22.31-32.
Budaeus, G. 1556. Commentarii Linguae Graecae. Basileae: apud Nicolaum Episcopum Iuniorem.
Bundy, M.W. 1927. The Theory of Imagination in Classical and Medieval Thought. University of Illinois
Studies in Language and Literature 12.
Cerominas, J. 1954. Diccionario crítico etimológico de la lengua castellana, vol. 3. Madrid.
Ceronetti, G. 1985. Albergo Italia. Torino.
Chantraine, P. 1980. Dictionnaire étymologique de la langue grecque. Paris.
122 GIULIO LEPSCHY
Contini, G. ed. 1960. Poeti del Duecento. Milano-Napoli.

Cortelazzo, M., Zolli, P. 1979-. Dizionario etimologico della lingua italiana. Bologna.
Dauzat, A., et al. 1964. Nouveau dictionnaire étymologique et historique. Paris.
Devoto, G. 1966. Avviamento alla etimologia italiana. Dizionario etimologico. Firenze.
Diez, F. 1853. Etymologisches Wörterbuch der romanischen Sprachen. Bonn.
Ernout, A., Meillet, A. 1959-60. Dictionnaire étymologique de la langue latine. Histoire des mots.
Quatrième édition. Paris.
Frisk, H. 1970. Griechisches etymologisches Wörterbuch, vol. 2. Heidelberg.
Grapaldus, F.M. 1494?. De Partibus Aedium. Angelus Ugoletus Parmensis Impressor.
Hesseling, D.C. 1921. Spoken. Neophilologus 6.207-17.
Keller, H.-E. 1958. Fr. Pantoufle. In Etymologica. Walther von Wartburg zum siebzigsten Geburtstag
18. Mai 1958, a cura di H.-E. Keller et al. 441-54. Tübingen.
Knös, B. 1945. Un ambassadeur de l'hellénisme - Janus Lascaris - et la tradition gréco-byzantine dans
l'humanisme français. Uppsala: Almqvist & Wiksell, Paris.
Labbé, P. 1661. Les etymologies de plusieurs mots François. Paris.
Lepschy, G. 1981. Freud, Abel e gli opposti. In G. Lepschy. Mutamenti di prospettiva nella linguistica
173-98. Bologna.
Liddell, H.G., Scott, R. 1953. A Greek-English Lexicon. A New Edition. Oxford.
Marcato,  1982. Ricerche etimologiche sul lessico veneto. Rassegna critico bibliografica. Padova.
Meneghello, L. 1986. Il tremaio. Note sull'interazione tra lingua e dialetto nelle scritture letterarie.
Bergamo.
. 1987. Jura. Ricerche sulla natura delle forme scritte. Milano.
Meyer-Lübke, W. 1935. Romanisches etymologisches Wörterbuch. 3. vollständig neubearbeitete Auflage.
Heidelberg.
Migliorini. ., Duro, A. 1953. Prontuario etimologico della lingua italiana. Seconda edizione. Torino.
Olivieri, D. 1961. Dizionario etimologico italiano, 2a ed. Milano.
Petronius Arbiter. 1961. Satyricon, a cura di K. Müller. München.
Pokorny, J. 1959. Indogermanisches etymologisches Wörterbuch, vol. 1. Bern und München.
Prati, A. 1951. Vocabolario etimologico italiano. Milano.
Spitzer, L. 1924. Französische Etymologien. Zeitschrift für romanische Philologie 44.188-200.
. 1928. Zur Methodik der etymologischen Forschung. Zeitschrift für romanische Philologie 48.77-
113.
Sylvius Ambianus, I. 1531. In Linguam Gallicani Isagωge. Parisiis: Ex Officina Roberti Stephani.
Valdês, J. de. 1535. Diálogo de la lengua, a cura di E. Boehmer. 1895. Romanische Studien 6.
. 1535. Diálogo de la lengua, a cura di J.F. Montesinos. 1928. Madrid.
Varchi, B. 1570. L'Hercolano. Fiorenza.
Walde, A. 1927. Vergleichendes Wörterbuch der indogermanischen Sprachen, herausgegeben und bearbeitet
von J. Pokorny, vol. 2. Berlin und Leipzig.
. 1938-54. Lateinisches etymologisches Wörterbuch, 3., neubearbeitete Auflage von J.B. Hofmann.
Heidelberg.
Wartburg, W. von 1957. Französisches etymologisches Wörterbuch, vol. 8, fase. 55. Basel.
Woodhouse.J. 1970. "Carafulleria". Lingua Nostra 31.110-11.
SUMMARY
This article considers some of the proposals that have been made about the etymology of the word
(and related terms), its Latin rendering, and successive Romance forms, including Spitzer's
suggestion of a link with French pantoufle, and Renaissance discussions concerning this word. At the end
some questions are asked about the use of notions like 'connected with' or 'related to', in comparative
studies, and the function of etymological considerations, in dictionaries, to separate homography from
polysemy.
III
MYCENOLOGY
SCHIAVI IN VENDITA NELLA CNOSSO MICENEA
MARIO DORIA
Università di Trieste
Gli archivi di Cnosso contemplano, com'è noto, alcune iscrizioni classificabili come
contratti di vendita di schiavi. Dei due documenti più completi (B 988 e  822; Ai5976
risulta piuttosto mutilo) si è occupato molto recentemente J.-P. Olivier (1), ricavando da
essi tutto eiò che vi si poteva leggere alla luce - aggiungiamo - di una nuova "congiun
zione" resa possibile in seguito alla romanzesca vicenda della scoperta - alla vigilia del
licenziamento delle ultime bozze del I vol. del Corpus delle iscrizioni in Lineare  cretesi -
di 3000 tavolette e frammenti di tavolette nei magazzini del museo di Iraklion.
Come succede abbastanza spesso, i nuovi dati, se da una parte contribuiscono a
rischiarare quello che precedentemente poteva apparire oscuro, dall'altra sollevano nuove
problematiche che è compito nostro tentar dirisolverenel modo migliore possibile.
I fatti nuovi, che riguardano solo una di queste tavolette, la  988, sono i seguenti: a
1. la del documento si è potuto inserire un nuovo frammento, il quale ci consente di
leggere per intero uno dopo l'altro, i due gruppi sillabici pa-qo-si-jo si-ra (anteriormente si
leggeva solamente pa-qo-si[. Lo stesso frammentino ci consente di leggere a 1.1 b per
intero ko-ma-we-to, anziche il mutilo ko-ma[. Importante, per l'economia generale
dell'iscrizione, anche la lettura, nella stessa linea, di VIR 1 [, al posto del vecchio VIR 2 [
il quale creava un certo imbarazzo (ci obbligava a cercare due anziché un solo nome di
schiavo).
128 SCHIAVI IN VENDITA NELLA CNOSSO MICENEA
Dopo un minuzioso esame dei due documenti l'Olivier, molto prudentemente (e

coerentemente con se stesso), giunge alla conclusione che i due testi risultano, ora, suffi
cientemente chiari, a prescindere, tuttavia, dal significato che dobbiamo dare ai due gruppi
sillabici si-ra di  988 (nuovo) e si-ra-ko (noto da tempo) di  822. E' chiaro che diverse
soluzioni ne possono scaturire, a seconda che neghiamo  stabiliamo una connessione tra
questi due gruppi sillabici e a seconda del valore che vogliamo dare ad essi, sia presi
insieme che separatamente. L'Olivier le elenca una per una: alla fine però non si sbilancia
troppo, anche se, a dir la verità, egU dimostri qualche simpatia nel definire si-ra e si-ra-ko
tra loro indipendenti e di ritenere, comunque, ambedue estranei all'antroponimia micenea
(probabilmente nomi di "mestiere" in senso lato, più difficilmente modificazioni - parti
cipi? avverbi? - della forma verbale qi-ri-ja-to Diverso, invece, il
mio parere, in quanto ritengo molto probabile che si-ra e si-ra-ko siano casi flessi di una
stessa parola, senz'altro classificabile come antroponimo. Ma andiamo per gradi.
La prima operazione da fare mi sembra sia quella di tentare un'interpretatio graeca dei
due documenti, a partire dall'analisi che ci ha dato, per l'appunto, l'Olivier:
 822 (l'iscrizione, in un certo senso, la più facile):
1. a po-da-qe-re-si-je-wo
1. b ]pi-ro / si-ra-ko qi-ri-ja-to / / ku-te-ro / ku-ro2-jo do-e-ro VIR 1 [
2. vacai
"Demo]filo (o sim.) .... ha acquistato Kytheros servo di Kyrios, di mestiere (o

funzione) x", UOMO 1["
 988
a pa-qo-si-jo si-ra qi-ri-ja-to
b ka-ra-na-ko / ko-ma- we-to do-e-ro VIR 1[
Pambosios ha acquistate Kranarchos, servo di Komaeis, UOMO 1[.
Di esse, come si vede, non vengono interpretate le parole "chiave" si-ra e si-ra-ko,
date le difficoltà, suaccennate, di determinare con sicurezza la funzione inerente alle due
MARIO DORIA 129
forme. A mio avviso, dette difficoltà sono molto meno gravi di quanto sembri ad un
primo momento. E vediamone il perché.
L'Olivier, come si è detto, nega -  fortemente dubita - che si-ra ~ si-ra-ko siano
antroponimi, in quanto che nell'iscrizione non ci darebbe posto per essi: c'è già un
soggetto, pa-qo-si-jo in  988 e ]-pi-ro (unico nome in - di tutta la grecità micenea)
in  822. C'è poi l'oggetto di qi-ri-ja-to, ka-ra-na-ko in  988, ku-te-ro in  822. Inoltre,
il nome, al gen. del possessore dello schiavo, Ku-ro2-jo in  822, ko-ma-we-to in 
988. Quest'ultimo tratto esclude, poi, qualsiasi possbilità che si-ra-ko possa essere un
genit., come si sarebbe tentati di supporre soprattutto dopo la scoperta della forma si-ra.
E' questo il motivo per cui si è tentati di interpretrare si-ra e si-ra-ko due forme fra loro del
tutto indipendenti.
Il mio parere è, a questo riguardo, completamente diverso: si-ra e si-ra-ko, limitan
domi pel momento al si-ra-ko di  822 (per gli altri si-ra-ko v. avanti), sono forme flesse
di una stessa forma, si-ra al nom. e si-ra-ko al gen.sg. Gli argomenti sono i seguenti:
1. uno di carattere ortografico: si-ra sta a si-ra-ko come to-ro-wi sta a to-ro-wi-ko e come
to-ra sta a to-ra-ke (nom.pl.) Nella notazione del gruppo ks di fine di parola si può
optare  per la notazione di una sola consonante (la prima, tipo wa-na-ka 
per la notazione di ambedue (tipo wo-no-qo-so  per l'omissione totale,
come, per l'appunto, nel caso nostro.
2. pa-qo-si-jo (B 988) non è antroponimo, ma patronimico vero e proprio, secondo un
uso ben noto per Pilo (ad es. nelle ben note iscrizioni o-ka) e qua e là anche per
Cnosso, nonché per altri dialetti greci (e per Omero). Non solo grammaticalmente,
quindi, ma anche nella concretezza delle situazioni proprie di quel mondo lontano,
pa-qo-si-jo è connesso col più semplice Pa-qo-ta (ad es. quindi "figlio
di Pa-qo-ta" a tutti gli effetti (mi si obbietterà che il più semplice pa-qo-ta non è mai
attestato a Cnosso, ma ne vedremo subito il perché).
3. Si-ra-ko (B 822a) ha la stessa funzione di pa-qo-si-jo di  988, quindi ...] philos,
(figlio) di Sillace".
Sulla base di questi pochi e semplici dati siamo in grado - e ciò mi sembra pienamente
lecito - di ricostruire, addirittura, una vera e propria genealogia nei riguardi del perso
naggio Si-ra (che io non esito a leggere
130 SCHIAVI IN VENDITA NELLA CNOSSO MIENEA
1. capostipite
2. figlio del suddetto
3. figlio di Sillace, nipote di Pamboutas.(6)
Appartenendo Pamboutas alla prima generazione è ovvio che esso non venga mai
nominato nelle tavolette di Cnosso: per lui non c'era più posto, per lo meno come membro
attivo, in una società in cui operavano, invece, figlio e nipote/i.
Ma sul nome/personaggio Si-ra-ko e questioni connesse occorrerà fare altre pre
cisazioni, che l'Olivier non ha voluto esporre.
La mia convinzione che si-ra ~ si-ra-ko sia antroponimo nasce dal fatto che il gruppo
sillabico si-ra-ko compare in una funzione che non può essere altro che di NP anche in
altre iscrizioni cnossie, precisamente Ai5977,l e Db8312. Si aggiungano, poi, le attesta
zioni mutile (o apparentemente mutile) si-ra[ di X 7744 e si-ra-[ di Xd 8597. Evidente
mente si-ra[ ci dice assai poco (anche se l'apparato critico di KT4 ci suggerisce la variante
di lettura si-ra-ko). Di più un si-ra[,a che possediamo il si-ra completo di  988,
potrebbe essere, al limite, anch'esso parola completa, quindi anziché diminuire, le per
plessità nei riguardi di questo frustolo di iscrizione aumentano! Maggiori informazioni
ricaviamo da Xd8597, dove siamo sicuri che la parola continuava con un altra sillaba,
quindi, plausibilmente si-ra-[ko (ma è altrettanto possbile un si-ra-[ke  anche un si-ra-
[ki). Dal tutto, comunque, si ricava l'impressione che l'elemento (onomastico) si-ra-(ko)
era, a Cnosso, piuttosto comune. Al polo opposto abbiamo le forme con ]ra-ko possibil
mente mutilo dell'iniziale (KN Dbll5 B, Vc7837); però, l'integrazione si-]ra-ko è
tutt'altro che pacifica, essendo possibile anche a-]ra-ko,  anche, se c'è posto per un
toponimo, da-]ra-ko Praticamente, quindi, da questo -]ra-ko non si
ricava nulla.
La più interessante delle iscrizioni contenenti si-ra-ko è certamente Ai5976 + 8628 +
fr.:
dove si-ra-ko un'altra volta è legato ad un'operazione di compravendita (qi-ri-ja-to). Puro

caso? Non credo. D'altro canto non ritengo prudente integrare -]ko con -r--],
sulla falsariga di  988, data la diversità di posizione di questo -]ko (in Ai5975 prima di
si-ra-ko, in  988 dopo si-ra qi-ri-ja-to). Se -]ko, come sembra, è fine di NP, converrà
MARIO DORIA 131
ritenerlo il nome di un secondo figlio di Sillace, figlio che andrebbe ad inserirsi alla 1.3
della genealogia abbozzata a p. 130.
Quanto a Db8352 si-ra-ko, fuor d'ogni dubbio nomin., è il nome di un pastore cui è
affidato un gregge di 200 capi (ignoto il collector) nella località di ra-ja. Ambiente diverso,
quindi anche nome, per le menoflessionalmente,diverso.
Da tutti questo confronti balzano due considerazioni tra loro in un certo senso anti
tetiche: 1. si-ra-ko è certo NP al nom. in Db 8352, 2. Lo stesso gruppo sillabico si-ra-ko
compare associato ben due volte con qi-ri-ja-to. Una terza volta, però, prima di qi-ri-jato
non c'è si-ra-ko, bensì si-ra. Quindi si-ra -si-ra-ko delle due iscrizioni  e si-ra-ko di
Ai597 sono ovviamente (Tuna al nom.sg., l'altra al gen.sg.) la stessa cosa, mentre si-ra-
ko di Db8352 solamente gli assomiglia, ma come nom.sg., ne è morfologicamente
diverso. Ciò non basta, tuttavia a negare al nostro si-ra-/ si-ra-ko la qualifica di NP,
questo per due considerazioni, una di carattere glottologico e l'altro esegetico: 1. si-ra-ko
(Db) è da leggere Si-ra-kos (nom. sg.), si-ra anche nom.sg. ma atematico, secondo un
rapporto morfologico, piuttosto frequente in greco, dove troviamo abbastanza spesso
alternanti nella flessione dei temi in occlusiva forme atematiche e forme tematiche, cosi
ecc. contro anche δρώψ in
opposizione ad ecc. In altri termini avremo sia si-ra che si-ra-ko
(nom.), tanto più che un è effettivamente attestato per il I millennio
Simon. 34 PMG), da un radicale "strabico" che compare nuovamente a
Pilo nel derivato Si-ri-jo (Sillios). 2. si-ra-ko (nom.) e si-ra (nom.)/si-ra-ko (gen.) non
designano gli stessi personaggi: si-ra-ko di Db è è un semplice pastore (localizzato a ra
ja), si-ra-ko ~ si-ra un personaggio, se non proprio altolocato, per lo meno provvisto di
una solida posizione economica, che gli consente di comparire come compratore nell'atto
di compravendita qui analizzato. Tale sua posizione viene ulteriormente comprovata dal
fatto che non solo lui ma anche il figlio (o due suoi figli) sono interessanti ad un atto di
compravendita e che tutti due (tre) personaggi ci tengono ad essere menzionati con nome
individuale e patronimico (alternativamente possiamo, anche, ipotizzare che in atti del
genere la menzione del patronimico fosse obbligatoria).
Ultima constatazione: qi-ri-ja-to è attestato, ancora, inAi1037(8): lo stato, però, di
conservazione della tavoletta non ci consente di conoscere il nome del compratore.
Vediamo comunque chiaramente che oggetto della compravendita è una 'schiava' (do-e-ra
132 SCHIAVI IN VENDITA NELLA CNOSSO MICENEA
e che questa dichiara di essere consenziente,  per lo meno non dispiaciuta (we-
ka-sa - del fatto di cambiare padrone (Esisteva quindi la possibilità, in detti
contratti, di notazioni accessorie relative ad esso, come il far rilevare che lo 'schiavo'
posto in vendita aveva il diritto di dire la sua in merito alla transazione stessa; forsanco, in
casi limite, di opporsi ad essa). Non saprei dire se questo fatto sia la ragione latente che ha
indotto l'Olivier ad accennare alla possibilità che si-ra ~ si-ra-ko siano termini concernenti
in qualche maniera la modalità dell'acquisto: le differenze d'impiego rispetto il we-ka-sa
sopra ricordato sono, però, troppo vistose, considerata anche la posizione stessa che le
due notazioni assumono nel contesto dell'iscrizione. D'altro canto si-ra-ko compare ben
due volte davanti a qi-ri-ja-to, dando, quasi, l'impressione di formare con questo un
sintagma fisso: resta ad ogni modo da osservare che qi-ri-ja-to, una volta, è preceduto da
si-ra (10), non da si-ra-ko, e che nell'iscrizione Ai5976,1 si-ra-ko ha un rilievo grafico
che sembra staccarlo nettamente dal qi-ri-ja-to che lo segue. Come a dire che di sintagma
fisso non è il caso di parlare.
Concludendo, nulla si oppone acché si-ra ~ si-ra-ko vada considerato come nome
proprio di persona e che il personaggio da esso designeto occupi nel microcosmo delle
operazioni di vendita di una localitá ignota dell'isola di Creta la posizione socio-economica
che abbiamo cercato di delineare.
NOTES
1 J.-P. Olivier, Des extraits de contrats de vente d'esclaves dans les tablettes de Knossos, Minos X,
1987, 479-498.
2 A dire il vero preferirei l'alternativa nome parlante, designazione, quasi, di un aquae
ductor, con riferimento alla tenica di condurre l'acqua da una sorgente, mediante canali, ai luoghi
(pascoli, campi) di utilizzazione.
3 Alla possibilità che si-ra-ko di quest'iscrizione sia un personale accennano ben pochi, fra questi L.
Palmer, The interpretation of Mycenaean Greek texts, Oxford 1963,450 (s.v. qi-ri-ja-to).
4 II rapporto intercorrente tra il N. di persona e il patronimico che da esso deriva non è, tuttavia, sempre
costante e segue una certa casistica. Il non attenersi ad essa può portare, facilmente, l'esegeta a distor
sioni interpetative piuttosto fastidiose. In effetti dalla documentazione in Lineare  in nostro possesso
possono verificarsi queste quattro situazioni, sostanzialmente assai diverse l'una dall'altra:
1. (caso più fortunato): All'interno di un medesimo archivio risultano attestati sia il nome sem
plice che il derivato con funzione patronimica: es. KN ku-ru-me-no e ku-ru-me-ni-jo,
e
MARIO DORIA 133
2. Sempre all'interno di un medesimo archivio sono attestati sia n. semplice che il derivato patro
nimico, ma questo fuoriesce dalla funzione di patronimico vero e proprio, es. KN e-ko-to
valevole solo come n.proprio indipendente, quindi ex-
patronimico), come sopra),
3. E' attestato solo il patronimico (il. n. semplice non esiste, rappresenta una semplice rìcostru
zione), es. PY a-da-ra-ti-jo (a-da-ra-to non è attestato), e-te-wo-ke-
re-we-i-jo (manca e-te-wo-ke-re-we ) ecc.
4. E' attestato solo l'ex-patronimico (il n. emplice non esiste), es. KN i-do-me-ni-jo (mancano sia
I-do-me-ne che I-do-me-ne-u
Il nostro pa-qo-si-jo si inserisce nella categoria (3), dal momento che il semplice pa-qo-ta è
attestato a Pilo, non a Cnosso. Ovviamente, sequendo l'interpretazione dell'Olivier pa-qo-si-jo
apparterrebbe addirittura alla categoria (4).
5 Regolare, come si sa, anche il rapporto venutosi a creare fra -ta ) e-si-jo (-σι,ος ), per le
vicende dell'assibilamento delle dentali sorde nel dialetto meceneo. Cfr. i casi ben noti di ra-wa-ke-ta I
ra-wa-kesi-jo ed e-qe-ta I e-qe-si-jo dove si parte, a dir il
vero, da appellativi. Più aderente al caso nostro la coppia KN a-ke-ta / a-ke-si-jo, in quanto
interessante nomi di persona (per a-ke-si-jo v. ora Killen, Kadmos 24,1 1985,26-33).
6 Per un eventuale altro figlio di da aggiungere alla 1.3 della nostra genealogia v. sopra ρ 130.
7 Cfr. M. Doria Avviamento, Roma 1965, p.245. Ma qui si era arrivati a proporre partendo,
appena, dal genitivo si-ra-ko, non dal si-ra che dormiva nei magazzini del museo di Iraklion!
8 Per un'ulteriore possibile attestazione di v. KN Ai(3) 7735 qi-ri]-ja-to (integrazione
suggerita dal fatto che l'iscrizione appartiene al medesimo 'set' di Ai(3)5976, dove qi-ri-ja-to compare
per intero) Il tutto però (-qi-ri]-ja-to si-qa, 1.2 vacat) non illumina certo le altre iscrizioni qi-ri-ja-to.
Non si è, poi, nemmeno sicuri se l'iscrizione tratta, come le altre, un acquisto di schiavi.
9 Per farci un'idea di cosa possano essere le cosiddette notazioni accessorie potremmo addurre il con
fronto con documenti (semitici) dell' area del Vicino Oriente, ad es. la tavoletta di Mari (N.10 della
raccolta Mission Archéologique de Mari, VIII Textes juridiques, pour J. Boyer, Paris 1952) in cui
figura registrato, perl'appunto, il contratto per l'acquisto di uno schiavo. Il testo, però, è molto più
complesso, in quanto vi sono nominati oltre il compratore, il venditore e, naturalmente, lo schiavo,
anche la data della transazione (giorno, mese ed anno), la clausola di garanzia ("contro chiunque lo
rivendicasse sarà garante Ili Idinnum") e i testimoni (in numero di cinque). Di più, nessuna di queste
notazioni "accessorie" pare ricoprirsi con quelle poche ricavabili dai nostri testi.
10 Al limite (cfr. Olivier cit. p.) potremmo anche considerare si-ra (effettivamente un hapax) errore per
si-ra-ko e arrivare in tal modo alla desiderata omogeneità. Ma l'operazione è - pel momento - una
forzatura e conviene attenersi, d'accordo col principio della lectio difficlior, alla dizione originaria.
EN LINEAIRE A?
JEAN-PIERRE OLIVIER
Université de Bruxelles
Il peut paraître audacieux, voire téméraire, de prétendre essayer d'identifier, dans une
langue qu'on ne lit pas - en l'occurrence le linéaire A - 1 , un mot quel qu'il soit, à plus
forte raison de le "traduire" (le mot pour "total" ou "total général" mis à part, bien
entendu).
Dans le cas qui nous occupe, nous avons peut-être une occasion exceptionnelle de lire
un chiffre du linéaire A.
En effet, cela fait près de trente ans que Maurice Pope a mis en évidence - dans une
démonstration qui n'a sérieusement été contestée par personne - que l'inscription murale
d'Haghia Triada présentement dénommée HT Zd 1562 (pl. I) avait bonne chance de porter
une formule mathématique, plus exactement une progression géométrique d'ordre 1,5, à
savoir «1, 1 1/2, 2 1/4, 3 3/8». Même si l'on ne suit pas l'érudit britannique dans ses
spéculations sur les «réductions de 5 %» dans les tablettes d'Haghia Triada, il est difficile
de ne pas accepter les débuts de sa conclusion déclarant que "This geometrical
progression can hardly have been a devotional or a cathartic outburst. It must be a ready
reckoner of some kind..."3.
Or il s'avère que la formule mathématique de M. Pope est suivie du mot HA 59-57
[TA - JA] et d'un métrogramme. Si l'on interprète ce dernier, comme 1/16 - ce qui est
évidemment loin d'être démontré - on pourrait être autorisé à «lire» "5 1/16", ce qui serait
exactement le terme suivant dans la progression géométrique. Dès lors il n'est pas exclu
136 JEAN-PIERRE OLIVIER
d'attribuer à TA-JA la valeur de «cinq» (bien que les deux signes soient pointés... mais
ces deux signes existent, et peut-être leurs points disparaîtront-ils un jour...).
Cela est-il trop beau pour être vrai? Je laisse aux spécialistes le soin d'en juger.
NOTES
1 J.-P. Olivier, «Lire» le linéaire A?, Le monde grec. Hommages à Claire Préaux, Bruxelles 1975,441-
449.
2 Gorila 4, 132-133.
3 M. Pope, The cretulae and the Linear A accounting system, BSA 55, 1960, 200-210.
ABSTRACT
The wallinscription HT Zd 156 gives us perhaps the name, in linear A, for the number five. The main
problem is that the reading is uncertain.
MYKENISCHE FOSSILIEN IM HOMERTEXT?
Zur Bedeutung von und
OSWALD Ρ AN AG L
Universität Salzburg
Das Problem der sprachlichen Mykenismen im Homertext und - damit eng ver
knüpft - die Frage nach dem Aufschlußwert der homerischen Epen für die politische
Struktur, das soziale Gefüge, das kultische Geschehen, kurzum: für die Lebenswelt des
mykenischen Griechentums im späten 2. Jahrtausend v.Chr. ist fast so alt wie die Linear
-Forschung. Ein kurzer, einem Spezialproblem gewidmeter Aufsatz kann aber keinen
Forschungsbericht zu einem Themenkomplex bieten, in dessen Kernbereich er sich
bewegt und dessen bibliographische Evidenz er eigentlich voraussetzt. So möchte ich nur
in aller Kürze meine Dreischritthypothese der Forschungsgeschichte wiederholen, die ich
an anderer Stelle1 bereits zur Diskussion gestellt habe und demnächst mit neuen Daten
und ausführlichen Literaturangaben zu erhärten hoffe.2
Auf eine Phase großer Zuversicht, die zwischen den Lineartafeln und den Epen
Homers eine kontinuierliche Entwicklung erkennen wollte und die Entdeckung aussage
kräftiger mykenischer Sprachdenkmäler, vielleicht gar von Resten frühzeitlicher Dichtung
oder Geschichtsschreibung nur als eine Frage der Zeit betrachtete,3 folgte alsbald eine
Periode radikaler Ernüchterung. Die Hoffnung auf ergiebige Neufunde hatte sich nicht
erfüllt, und die minuziöse philologische, linguistische sowie realienkundliche Unter
suchung des vorhandenen Textmaterials war nicht nur auf Übereinstimmungen mit dem
138 OSWALD PANAGL
epischen Befund gestoßen. Ganz im Gegenteil: die Gemeinsamkeiten schienen eher

trivialer Natur, die Divergenzen hingegen aussagekräftig zu sein.4 So trat nunmehr die
Zäsur als Modell der Entwicklung an die Stelle des Kontinuums, was der führende
deutsche Mykenologe Alfred Heubeck mit eindrucksvollen Vergleichen und scharfen
Kontrastpaaren belegt:
Aufrechnung von Überschuß und Defizit gegenüber tiefen philosophischen Gedankengängen, trockene
Listen von Personen und Sachen gegenüber unvergänglichen poetischen Entwürfen; Zwang, Auto
kratie, Dirigismus gegenüber Homonymie, Demokratie und Freiheit. Paßt das noch zusammen?5
Will man diese beiden Standpunkte im Sinne der Hegeischen Abfolge wissenschaft
licher Positionen als These und Antithese verstehen, so stellt sich die Frage nach einer
Versöhnung und Verschränkung auf höherer Ebene, also nach einer Synthese. Sie ist
meines Wissens noch nicht als geschlossener und widerspruchsfreier Forschungsansatz
geleistet worden bzw. in all ihren Einzelergebnissen anerkannt. Doch läßt sich immerhin
eine Tendenz ausmachen, die idealtypischen Extreme eines lückenlosen Übergangs oder
eines schroffen Bruches zwischen der mykenischen und der homerischen Epoche in
historischer und sprachlicher Sicht durch ein vermittelndes, gleichsam realtypisches
Modell eines sich in Schüben vollziehenden Wandels zu ersetzen. Wohl rechnet auch
diese Richtung mit einem kulturellen und politischen Umbruch nach der Katastrophe der
mykenischen Paläste, doch wäre eine Schicht palatialer Nachsiedler, die sich mit
zugewanderten Stämmen allmählich zu der griechischen Bevölkerung der archaischen Zeit
amalgamiert hätte, zum sprachlich-ethnischen Bindeglied zwischen den beiden Jahrtau
senden geworden.6
Für den linguistischen und dialektalen Befund bedeutet diese Sicht, daß sich die neue
Gesittung, der zivilisatorische Fortschritt, die veränderte Technologie und das rezente
gesellschaftliche Profil vor allem in einem gewandelten Wortschatz ausdrücken und
widerspiegeln. Diese Dynamik des Lexikons äußert sich nicht bloß in zahlreichen Neu
bildungen, konventionellen Metaphern und Lehnwörtern: Auch der Bedeutungswandel
des überkommenen Wortschatzes verrät den Umbruch in der Welt der Realien. Der
mykenische qasireu ist als vom Landadeligen und Zunftmeister der
Schmiedegilde zum Herrscher und König aufgestiegen. Das Verbum
hat die semantische Spezifizierung eines technischen Vokabels "berädern" (Partizip
Perfekt Passiv: araromotemeno als Attribut eines Wagens) abgestreift und die Allerwelts-
bedeutung "rüsten, ausstatten, verbinden" angenommen.
MYKENISCHE FOSSILEN IM HOMERTEXT? 139
Daß sich im Corpus der homerischen Epen, gleichsam als dem "missing link" zwi
schen der mykenischen Welt und der alphabetisch-archaischen Epoche, fossilhafte Reste
älteren Sprachgebrauchs finden lassen, habe ich an anderer Stelle ausgeführt.7 Ich hoffte
damit einen sprachwissenschaftlichen Beitrag zu jenem vermittelnden, die Extrem
positionen meidenden Standort leisten zu können, der mir spekulativ und bei behutsamer
Betrachtung der Fakten besonders plausibel erscheint. Dabei erhöht es noch die innere
Wahrscheinlichkeit, wenn die lexikalischen Mykenismen der homerischen Sprache vor
allem in Sinnbezirken auftreten, für die der Sachverstand und die technische Fertigkeit der
verflossenen Periode nachhaltige semantische Spuren hinterlassen haben.
Nach allgemeiner Überzeugung hat das mykenische Fachwort amo (=/harmo/) seine
Bedeutung "Rad" in den Linear -Tafeln nicht in das alphabetische Griechisch hinein
fortgesetzt, sondern sich bereits bei Homer zu einer Bezeichnung des Wagens entwickelt.
Da die Pluralform äρματα überwiegt, erkennt man in diesem Bedeutungswandel mit
Recht die Synekdoche 'pars pro toto'. Im Mykenischen gab es bekanntlich die beiden
"Wagen"-Wörter woka und iqija mit gleicher Bedeutung, aber verschiedener regionaler
Verteilung. Das streng komplementäre Auftreten {woka in Pylos, iqija in Knossos) läßt
auf ein ursprüngliches Syntagma woka iqija "Wagen für das Pferd, vom Pferd gezogen"
schließen, dessen Bestandteile sich dialektal verselbständigt haben: woka mit Bedeu
tungsverengung, das feminine Adjektiv iqija durch Ellipse seines Bezugsnomens und
kategorialen Übertritt zum Substantiv.8 Beide Vokabel leben relikthaft im homerischen
Wortschatz weiter: woka durch die anomale Vokalfärbung des s-stämmigen Neutrums
δχεα, (Plur.)9, iqija zumindest im Vorderglied des Kompositums iππιοχάρμης "Wagen
kämpfer".10
Für äρμα(τα) könnte es zunächst eine vereinzelte Stelle geben, an der es als fachlicher
Terminus in der ererbten, etymologischen Bedeutung "Rad" bei Homer belegt ist. Ich
meine damit die Passage I1.2,775f.
140 OSWALD PANAGL
Sie wird üblicherweise so übersetzt (Schadewaldt):

Die Pferde standen bei ihren Wagen ein jedes,
den Lotos rupfend und den sumpfgenährten Eppich.
Die Wagen aber der Herren lagen, gut zugedeckt,
in den Hütten. Und sie, den aresgeliebten Führer entbehrend,
streiften hierher und dorthin durch das Heer und kämpften nicht.
Bei dieser Interpretation werden also die im Freien geparkten Wagen der einfachen
Kämpfer von denen der privilegierten Anführer unterschieden, für die - in
Fortführung moderner Diktion - eine Garage bereitsteht. Mir scheint aber auch eine
andere Auffassung möglich zu sein, wodurch das zuerst genannte άρματα (V.775) in der
jüngeren Bedeutung "Wagen" auftritt, während in V.777 die empfindlichen, kostbaren
Räder der Streitwagen gemeint sind, die in der durch den Groll des Achilleus
erzwungenen Kampfpause vor Troja demontiert und, behutsam verpackt, in den Zelten
der Helden aufbewahrt wurden.
Mag in diesem Beispiel der unvermittelte Bedeutungswechsel von im Abstand
von nur zwei Versen zur Vorsicht mahnen, so scheinen mir die folgenden beiden Belege
mit großer Wahrscheinlichkeit ein Fortleben der älteren Bedeutung "Rad" bei Homer zu
bezeugen.
In I1.4,482ff wird der Todessturz des trojanischen Kriegers Simoeisios im Kampf mit
dem Umsinken einer gefällten Pappel im Sumpfgelände verglichen. Die entscheidenden
beiden Verse dieses Vergleichs sind 485f.
In der Übersetzung von Wolfgang Schadewaldt heißt es an dieser Stelle:

Die aber schlug ein Wagenbauer heraus mit dem braunroten Eisen,
um sie zum Kranze zu biegen für den gar schönen Wagen.
Die Wiedergabe und Interpretation der Berufsbezeichnung des Nomen agentis

als "Wagenbauer" ist zweifellos von der Bedeutung des ersten Gliedes im
alphabetischen Griechisch bestimmt. Die Wurzel *pāg - mit ihrem semantischen Wert
"befestigen, festfügen" scheint sich diesem Befund trefflich einzupassen: Ein Hand
werker, der (den) Wagen fest zusammenbaut, bietet sich als naheliegendes 'Referenz-
objekt' an.
MYKENISCHE FOSSILIEN IM HOMERTEXT? 141
Betrachtet man freilich diese Stelle näher, so fällt auf, daß der Dichter die Tätigkeit des
Werkenden auf die Verfertigung eines Radkranzes beschränkt. Der Wagen
begegnet nur in einem Dativ der Zweckbestimmung als "Empfänger" der
Arbeitsleistung: die verfertigten Räder werden demnach später auf eine Achse montiert
und an dem selbständig und unabhängig gebauten Wagenkorb, der "Karosserie", an
gebracht. Verbindet man diese nüchterne Feststellung und den Wortlaut der Stelle mit der
nachweislichen Bedeutung "Rad" von amo in den Linear B-Texten, berücksichtigt man
ferner das Weiterwirken der mykenischen Kultur auf dem Gebiet des Wagenbaus und
achtet endlich auf die hochgradige terminologische (und praktische) Spezialisierung des
frühgriechischen Handwerks (vgl. kuwanowoko, apukowoko, toronowoko), so
empfiehlt sich eine andere Deutung des Kompositums: Der Fachmann für das Fügen,
bzw. Bauen von Rädern ist gemeint. Er fällt die Pappel, deren biegsames, elastisches
Holz sich zur Herstellung des Felgenkranzes besonders eignet. Aus dem "Wagner", auch
in der Lesart des sprechenden Berufsnamens, wird also der "Radermacher". Daß
Komposita mit ihrer konservativen Semantik der Einzelglieder in gleichsam geschützter
Stellung dem Weiterleben mykenischer Wortbedeutungen besonders günstig ist, war
zuvor schon am Beispiel zu zeigen.
Auch der zweite Fall einer konservativen älteren Bedeutung von in der Ilias
betrifft ein Kompositum. Im 23. Gesang beobachten die griechischen Helden bei den
Leichenspielen für Patroklos den Verlauf des Wagenrennens und geraten über einen
vermuteten Führungswechsel schier in Streit. Endlich zeigen sich die Wagenkämpfer, und
Diomedes erweist sich mit seinem Gespann als der Sieger. In der eindrucksvollen
Darstellung seines Laufes beschäftigen wir uns mit den Versen 503ff.
Die Übersetzung (Schadewaldt) lautet:

Und der Wagen, mit Gold beschlagen und Zinn,
lief hinter den schnellfüßigen Pferden dahin, und keine tiefe
Wagenspur von den Radbeschlägen entstand dahinter
In dem lockeren Staub; die beiden aber flogen dahin.
142 OSWALD PANAGL
Auch an dieser Stelle hat die Nachbarschaft von : in der jüngeren Bedeutung
"Wagen" (V,503) die Übersetzer und Interpreten veranlaßt, auch das Kompositum
als "Wagenspur" zu übersetzen. Eine genaue, den Kontext beachtende Wie
dergabe kann hingegen m.E. nur zu "Radspur" (eigtl. "Radlauf") führen, denn die Räder
sind es ja eigentlich, die mit ihren Reifen den Beschlägen über dem
Felgenkranz, Spuren im Sande ziehen. Wohl ist die traditionelle Übersetzung "Wagen
spur" im Sinne einer Synekdoche (totum pro parte) nicht auszuschließen, doch wird die
pointierte, durch die mykenische Bedeutung von amo gestützte Interpretation "Radspur11
von den Reifen" der exakten, realitätsnahen, Details beachtenden Schilderung technischer
Abläufe bei Homer besser gerecht. Auch in der Lautung des Kompositums
mag man ein Indiz für ein hohes Alter der Bildung erkennen. Die Gestalt des Verder-
gliedes läßt drei Auffassungen zu:
1. Reduktion der 'überlangen' Konsontantengruppe -tir- zu einer Zeit,
als das urgriechische Lautgesetzt -tt- > -st- nicht mehr wirksam war.
2. Haplologische Behandlung eines vorauszusetzenden thematisierten
3. Vorliegen des unerweiterten n-Stammes *harmn- im Vorglied.

Im Falle dieser hocharchaischen Stammbildung würden die morphologische Struktur und
die konservative Bedeutung eines Kompositums gemeinsam den archaischen Charakter
der Bildung belegen und damit einen weiteren Mykenismus der homerischen Sprache
erschließen.
ANMERKUNGEN
1 Panagl 1980:503 ff.

2 Ich habe für das IX. Internationale Mykenologische Kolloqium in Athen (Oktober 1990) einen
Vortrag mit dem Arbeitstitel "Mykenisch und die Sprache Homers: Alte Probleme - neue Resultate"
angekündigt.
3 Ein typischer Vertreter dieses optimistischen Standpunkts war T.B.L. Webster in seinen Büchern
(1958 und 1961).
4 Vgl. Lesky 1967; Shipp 1971; Heubeck 1974a.
5 Heubeck 1974b: 44-57, Zitat S. 47.
6 S. Jalkotzy hat sich in ihren Arbeiten besonders mit dem Problem dieser gebrochenen, mittelbaren
Kontinuität beschäftigt: vgl. Jalkotzy 1977 und 1989.
MYKENISCHE FOSSILEN IM HOMERTEXT? 143
7 Vgl. den in Anm. 1 zitierten Aufsatz.

8 Vgl. dazu ausführlicher Panagl 1985: 283-292, bes. 286f.
9 Dazu Panagl 1982: 251-257.
10 Mühlestein 1955:123f.
11 Die Bildung ' einem agentivischen 'laufend' ist ein gutes Beispiel für ein
Abstraktum 'Laufen' im Übergang zum Resultativum bzw. Konkretum 'Spur' (als manifestes Ergeb
nis); vgl. zu diesem in Komposita besonders häufigen Typus Risch 1974:116ff., Scheller 1951:
44ff.
12 Nicht bloß gleiche, sondern auch ähnliche Silben sind Kandidaten für Silbenschwund, vgl. Schwyzer
1954:262f.
BIBLIOGRAPHIE
Jalkotzy, S. 1977. Fremde Zuwanderer im spätmykenischen Griechenland. Zu einer Gruppe handgemachter

Keramik aus den . III C Siedlungsschichten von Aigeira. (Sitzungsberichte Öst. Akad. Wiss.,
Phil.-hist. Kl., 326.) Wien.
.1989. Frühgriechische Herrschaftsformen in mykenischer Zeit. Jahrbuch der Universität Salzburg
1987-1989.133-151.
Heubeck, A. 1974a. Die homerische Frage. Darmstadt.
.1974b. Mykenologie und Altertumswissenschaft. Probata - Probanda. 44-57. München.
Lesky, A. 1967. Horneros (Sonderausgaben der Paulyschen Realencyclopädie der Classischen Altertums
wissenschaft.) Stuttgart.
Mühlestein, H. 1955. Zur mykenischen Schrift: Zeichen za, ze, zo. Museum Helveticum 12.123f
Page, D.L. 1963. History and the Homeric Iliad. 2nd ed. Berkeley, California.
Panagl, O. 1980. Die homerischen Epen - ein Spiegel der mykenischen Welt. Sprache - Text -
Geschichte, hgg. von P. Stein. 503-526. Göppingen.
.1982. Homerisch Ein verkappter lautlicher Mykenismus. Serta Indogermanica, Festschrift für
Günter Neumann. 251-257. Innsbruck.
.1985. Hippologica Mycenaea. Sprachwissenschaftliche Forschungen, Festschrift für Johannes
Knobloch. 283-292. Innsbruck.
Risch, E. 1974. Wortbildung der homerischen Sprache. 2., völlig Überarb. Aufl. Berlin-New York.
Scheller, M. 1951. Die Oxytonierung der griechischen Substantiva auf -La. Zürich.
Schwyzer, E. 1954. Griechische Grammatik. Bd. 1. München.
Shipp, G.P. 1971. Essays in Mycenean and Homeric Greek. Melbourne.
Webster, T.B.L. 1958. From Mycenae to Homer. London.
.1961. Die Nachfahren Nestors. Mykene und die Anfänge der griechischen Kultur. München-Wien.
144 OSWALD PANAGL
SUMMARY
The autor of this article tries to evidence that the first element of the epic compounds and
represents the Mycenaean technical word amo in its achaic meaning "wheel" (later "chariot").
This term seems to be a further example of Mycenaean 'sediments' in the Homeric poems.
IV.
GREEK PHILOLOGY
SOME REMARKS ON TEXTUAL PROBLEMS IN EURIPIDES'
IPHIGENElA IN AULIS
HANS-CHRISTIAN GÜNTHER
Universität Freiburg i. Br.
If both and are taken together with it is awkward that

inevitably then going with as well - fits only with but not with
one would have to supply another dative like vith the latter
in order to balance As the text stands, it is certainly impossible to separate
from and understand it as an independent complement, but to write
(proposed by Monk) in order to achieve this construction is the wrong way out of the
difficulty. is absolutely faultless in itself3, whereas
hardly seems to be the proper word here to go with . A minute change τ' ού for
removes every offence. is a phrase perfectly in line
with in 333. There is thus no serious lack of clarity in the
neuter but the obvious temptation to connect άδικον with easily accounts
for the corruption of
2. 523
148 HANS-CHRISTIAN GÜNTHER
Markland's minimal emendation is universally accepted by editors. In his

apparaturs Murray proposes a further change to
. Since Euripides places words scanning vvv- - in position 7-10 in the trimeter6
Markland's text cannot be rejected with certainty. However it seems unlikely to my mind
that Euripides should have placed a monosyllabic postpositive (not necessarily creating a
long syllable!) so that the rare and unhandy combination vvv- - results (especially in this
case where äv rather tends to be attached to ). Murray's transposition shows that a
much more convenient possibility would have been available for Euripides to say the
same thing with the same words. Nevertheless Murray's more natural word order would
not be very likely to develop into the text of L. One could think of writing
(Herwerden) but perhaps
is better. This rather exquisite phrase would be prone to
corruption via an interlinear explanation or simply by omission of
after
561
Nauck tried to mend the most apparent fault by suggesting

He thus meets the demand for some modification of (a need especially
pressing with the article in this position) starting from a plausible corruption (ευ to a),
and removal of the article seems to be supported by the reading of Lac, which is τ'
instead of ϑ (the correction cannot be attributed to Lc or to Tr with any certainty). But
even with Nauck's the phrase is suspect. Passive construction
of the internal accusative seems to be absent from Vth century poetry8, and with a
transitive verb this construction is hardly possible at all9. The easiest way to get rid of
at first sight seems to be but this is doubtful
11
metre .I therefore rather suggest
SOME REMARKS ON TEXTUAL PROBLEMS IN EURIPIDES' IPHIGENEIA IN AULIS 149
NOTES
1 For editions and commentaries on the Iphigeneia cited under the author's name only and for my
statements about readings in L(aur. pl. 32,2) see my Textprobleme im Prolog der aulischen
Iphigenie des Euripides, Festschrift für W. Kullmann, WJbb 13, 1987, 59 n.l. Perhaps I may add
that p. 63 η 33.I should have cited Or.1015 (with Willink's [Oxford 1986] note). In favour of
Lloyd-Jones' view of 149-152 (see p. 72) one could point to ASe. 371.
2 A wrong thing for a man (himself) to have (England); but in his version of the second part of the
line (... and one which prevents his friends having any confidence in him) he fails to translate
4 We should have expected something more like αίσχρòν in meaning (England).

5 The first half of the line is corrupt; however it is emended this has no bearing on the parallelism
proposed.
6 Cp. J. Descroix, Le Trimètre Iambique (Paris 1931)142f.
7 Cp. S.Ai. 404, OC 170.
8 The first example I find is Hdt. 9,26 there are few further instances in
Thucydides (e.g. 2,65,11), but the construction seems to be extremely rare before IVth century
prose.
9 LSJ s.v. II. take as middle in active sense. is used instead
of the active E. Hyps, fr.64,101, but this is hardly to justify this assumption here of
without an object.
10 has already been suggested by Monk. I do not know how he
constructes To join it with διάφοροι
parenthesis, would be fatal to the sense.
11 Responsion v-v--vv- is uncertain (cp. Itsumi CQ 32[1982]73; on E.El.438-448 see Denniston
[Oxford 1939] ad loc). One could think of measuring in 546 ( cp. IA
1084 ; for the cho di v-v- - vv- cp. E. Suppl. 1023, Hel. 1338f.t 1454,1468). But
perhaps this is a subterfuge. seems to be an Euripidean neologism, elsewhere in
Euripides (E.HF 402, Hel. 1458, Hyps.fr.  iii 4) the ει is long.
METRICAL CONVERGENCES AND DIVERGENCES
IN EARLY GREEK POETRY AND SONG
GREGORY NAGY
Harvard University
In making a distinction between "song" and "poetry," we must confront a semantic

problem occasioned by our own cultural preconceptions. Whereas the stylized rhythms of
poetry are known to us as "meter," we think of the stylized rhythms of song simply as
"rhythm." This mode of nomenclature is hardly appropriate to the traditions of Greek
lyric, where song operates on principles of "rhythm" that are clearly cognate with the
principles of "meter" in the recitative poetry of, say, Greek epic. In fact, it is common
practice to speak to the "rhythms" of Greek lyric in terms of "meter."
This much said, I shall argue, on the basis of the available metrical evidence, that
poetry can be analyzed as a differentiated derivative of song in early Greek traditions. To
make this point, I choose as illustration two fundamental types of Greek lyric metrical
patterns, the so-called Aeolic and dactylo-epitrite meters. I have two reasons for this
choice, one of which will become clear only as the discussion proceeds. But it is
important to stress, alsready at the start, the other reason: all the attested lyric poetry of
Pindar, with only a few exceptions, is composed in one or the other of these two kinds of
meters.1
I have singled out Pindar as the centerpiece for illustrating the traditions of lyric
poetry, that is, song, because the actual content of this poetry reaffirms both explicitly and
implicitly the thematic connection between epic and lyric, which reflects on the formal
152 GREGORY NAGY
connection between poetry and song. In what follows, I shall argue that these connections
are verified by the form, or better, forms of lyric poetry.
Specifically, I shall argue that the dactylic hexameter, the meter of epic, took shape
from the phraseology of the two basic metrical systems inherited by Pindar, the Aeolic
and the dactylo-epitrite. If the arguments about to be advanced are valid, it would follow
that the three meters of Greek poetry in general, that is, dactylic hexameter, elegiac
distich, and iambic trimeter, are differentiated equivalents of cognate meters found in
Greek song, as in the inherited medium of Pindar.
What follows, then, is a survey of the diachronic relations between the Aeolic and
dactylo-epitrite meters of Greek song on the one hand, and on the other, the dactylic
hexameter, the elegiac distich, and the iambic trimeter of Greek poetry.
Let us begin with a synchronic description of the Aeolic meters.2 Both these meters
and the dactylo-epitrites, which will be described after the Aeolic, are built on the
principle of the colon, which I shall define for the moment simply as a fundamental unit in
the rhythmical structure of song.3 A fundamental type of colon in the system of Aeolic
meters is the so-called glyconic:
xx_oo_o_ = glyc4
Both here and in the other metrical representations that follow, I show the last syllable
of a given metrical unit as uniformly long, not short. This practice reflects a principle that
is at the same time metrical and linguistic: I mean the neutralization, in prepausai position
(that is, in the last syllable of a metrical unit), of the distinction between long and short in
favor of long. This principle is commonly known as brevis in longo.5
The glyconic can be internally expanded by a unit called the choriamb
_oo_ = ch
This expansion may involve one, two, or three choriambs, to form
Similarly, a glyconic can be internally expanded by a unit called a dactyl

METRICAL CONVERGENCES AND DIVERGENCES 153
_oo = da
This expansion may involve one or two dactyls, to form
Now the glyconic

xx_oo_o_ = glyc
has a catalectic variant, called the pherecratic
xx_oo = pher11
By "catalectic" and "catalexis," I mean the shortening of the metrical unit by way of
deleting the last syllable and making the next-to-last syllable the new brevis in longo.
Using the symbol "<" for catalexis, I describe the pherecratic as a catalectic glyconic:
pher = glyc<.
Like the glyconic, the pherecratic can be expanded by choriambs:
Like the glyconic, the pherecratic can be expanded by dactyls:
These metrical types have acephalic variants. The notions of "acephalic" and
"acephaly" can be defined synchronically as the deletion of the first syllable of a given
metrical line. Using the symbol ">" for acephaly, we may say that the glyconic has an
acephalic variant known as the telesillean:16
x_oo_o_ = >glyc = tele17
Also, the pherecratic has an acephalic variant, conventionally called the reizianum:
x_oo = >pher = raiz18
154 GREGORY NAGY
These acephalic variants, the telesillean and the reizianum, can in turn be internally
expanded by choriambs or dactyls, as in
So much for a synchronic description of these Aeolic meters. Turning now to

diachronic perspective, I begin by observing that the mechanics just surveyed have an
Indo-European provenience.21 For example, Brent Vine has demonstrated that cognate
patterns of catalexis and acephaly are at work in Indic poetics.22 Also, I have found that
the relationship between glyc and glyc@ch corresponds, within the Greek evidence itself,
to the relationship between iambic dimeter and trimeter,23 and that there is a cognate
relationship between the Indic dimeter and trimeter.24 Like the Greek dimeter, the Indic
dimeter is an 8-syllable line, or a 7-syllable catalectic variant; also, the Indic trimeter is a
12-syllable-line, or an 11-syllable catalectic variant.25
Let us take a closer look at the relationship between the Greek iambic dimeter and
trimeter:
In this scheme, the abbreviation ia stands for the iamb, x_o_, in the first four
syllables, or opening, of an iambic dimeter, while A stands for the iamb, x_o_, in the last
four syllables, or closing; the sign stands for word-break, that is, word-boundary,
while "&" stands for a combination not marked by any particular pattern of word-
breaking. Also, the sign "+" stands for the addition of an extra opening (ia) or closing
(IA).21 The distinction between opening and closing is a matter of great importance from
a diachronic standpoint. The scheme ia+ia&IA of iambic trimeter stands for a diachronic
pattern of opening plus opening and closing, or extra opening plus dimeter. The reflex of
this 4+8 pattern of syllables in a twelve-syllable line is a word break (|) that leaves a
pattern of syllables. Conversely, the scheme ia&IA+IA of iambic trimeter stands for a
diachronic pattern of opening and closing plus closing, or dimeter plus extra closing. The
reflex of this 8+4 pattern of syllables in a twelve-syllable line is a word-break that leaves a
pattern of syllables. In other words, the synchronic description of Classical iambic
trimeter as having a caesura after syllable 5 or, by default, after syllable 7 corresponds to
a diachronic derivation of iambic trimeter from a combination of an extra opening plus
iambic dimeter or from a combination of iambic dimeter plus an extra closing
respectively.28 We can find a cognate situation in Indic trimeters, where we see an older
pattern 4|8 being displaced by a newer pattern 5|7.29 In early Greek trimeters, moreover,
we find that the pattern 7|5 conceals further subdivision into 4|3|5,30 while the pattern 5|7
conceals further subdivision into 5|3|4.31 I argue for an older pattern 4|8 (4|3|5) in old
iambic trimeter, with the primary alignment of 4+3&5, which is concealed by the newer
pattern 7|5 of Classical iambic trimeter, derived from a secondary alignment of 4&3+5.
While the older pattern 4|8 survives in the guise of 7|5, it is at the same time superseded by
the newer pattern 5|7 of Classical iambic trimeter. To repeat, 5|7 is the normal
configuration of Classical iambic trimeter, while 7|5, still retaining a hidden 4|8 in older
iambic trimeter, is the fallback alternative.32
Besides iambic trimeter, there are other trimeters as well, containing constituents that
are cognate with dimeters other than the iambic dimeter. For example, the element ia can
combine with a dimeter shaped glyc, and here again we can observe the pattern 57 as an
outcome:
In such cases, however, the word-breaking pattern 5|7 is just one of a variety of
possibilities. To use Roman Jakobson's useful distinction between constant and
tendency,34 the 5|7 pattern is a constant in Classical iambic trimeters shaped ia+ia&IA, but
it is only a tendency in archaic trimeters shaped ia+glyc. We can apply an analogous
formulation to the 7|5 pattern in Classical iambic trimeters shaped ia&IA+IA: again we see
a constant, as distinct from a corresponding tendency in archaic trimeters shaped glyc+IA:
So far, we have considered various patterns of reapplied opening and reapplied

closing. But there are also situations where the opening and the closing are inverted, that
is, where a closing is fused with an opening that follows. For an illustration, let us
examine what can happen when two glyconics are run together:
156 GREGORY NAGY
Here I have introduced the symbol "~" to indicate a process that can be described as
"dovetailing."37 In synchronic terms, dovetailing is when the word-break is skipped at
the end of one metrical unit and transferred to the position after the first syllable of the
following metrical unit. As I have argued at length elsewhere,38 this process is analogous
to the synchronic pattern 5|7 of iambic trimeter.
x_o_x|_o_x_o_ = ia+ia&|A
Let us now turn from the Aeolic meters to the so-called dactylo-epitrite meters, the
second of the two major metrical groupings that I have undertaken to survey. A basic unit
of the dactylo-epitrite meters is the so-called prosodiakon:
x_oo_oo_ = pros39
The pros behaves like a constituent of trimeter. Let us compare the pattern traditionally
known as the iambelegos
x_o_x|_oo_oo_ = ia+pros40
with an iambic trimeter that has a word-break after syllable 5:
x_o_x|_o_x_o_ = ia+ia&IA
and with the trimeter formed with a glyconic, again featuring word-break after syllable 5:
x_o_x|x_oo_o_ = ia+glyc41
On the basis of such formal parallelisms in trimeter-formation, I propose that the
prosodiakon (pros) is an offshoot of the Indo-European dimeter, just like the iambic
dimeter and various other octosyllabic meters that are attested in Greek lyric:
x_o_x_o_ = iambic dimeter (ia&IA)42
xx_oo_o_ = glyconic (glyc)
derived from a reconstructed type
xxxxo_o_ = "irregular glyconic"43
x_oo_oo_ = prosodiakon (pros)
derived from a reconstructed type
xxxx_oo_
as attested in the type
xx_x_oo_ = choriambic dimeter44
Accordingly, I withdraw my earlier suggestion that the prosodiakon (pros) is a

resegmented derivative of the glyconic.45 Instead, I propose that both the glyconic (glyc)
and the prosodiakon (pros) are independent derivatives of the Indo-European dimeter.
Each Greek dimeter pattern entails various regularizations of various rhythmical sequen
ces, with the general direction of regularization moving from the closing toward the
opening.46 Thus the glyconic (glyc), as we see in the scheme above, is a regularization of
the "irregular glyconic," in that the fifth-from-last syllable of the "irregular glyconic" has
been generalized as a short, forcing the sixth- from-last syllable to be generalized as a
long to avoid a sequence of three consecutive shorts, which was not tolerated in early
Greek metrics.47 So also with the prosodiakon (pros): we could say that it is a regula
rization of the choriambic dimeter, with the fifth-to-last and sixth-to-last syllables
generalized as two shorts, forcing the seventh-to-last to be generalized as a long to avoid a
sequence of three shorts. What results is the symmetry of a rhythmical sequence
_oo_oo,
framed at each end by one syllable of indifferent quantity (x).48
Besides the various types of Greek dimeter that we have already seen resulting from
various patterns of rhythmical generalization, I now list the following additional types:
trochaic dimeter (tr&TR)49
= trochaic-choriambic dimeter (tr'&CH)50
= choriambic-iambic dimeter (CH&ia')51
= aristophanean (catalectic of above)52
= hagesichorean (hage)53
To posit such a plurality of categories may at first strike us as unnecessary. But the
necessity is indeed there: we are dealing in each case with what amounts to independent
generalizations of rhythmical patterns. I can put it another way from a comparative
standpoint, juxtaposing the patterns of Greek and Indic meter: while Indic versification
158 GREGORY NAGY
tends to tolerate different patterns within a single given flexible meter, Greek lyric tends to
generalize the corresponding patterns as a plurality of separate rigid meters.54 Although
"Greek rigidity in meter is a more advanced phenomenon than Indic flexibility,"55 this
same rigidity is what makes the internal evidence of Greek meter particularly valuable for
the comparative study of Indo-European metrics: the relatively rigid and differentiated
meters of Greek lyric are constants that preserve phases of evolution that have been
blurred by the ongoing tendencies of the relatively flexible and undifferentiated meters of
the corresponding Indic traditions.56
This much said, the diachronic perspective can now help us reduce the proliferation of
categories in Greek lyric. Let us begin by considering again the glyconic (glyc) and its
closest relatives:
= glyconic (glyc)
= hipponactean (hipp)
telesillean (tele)
= hagesichorean (hage)
From the synchronic viewpoint of descriptive metrics, we may describe segments B,
C, and D as derivatives of :  by hypersyllabism,  by acephaly, and D by both
acephaly and hypersyllabism. The diachronic point of view, however, is more subtle.
First of all, let us observe that a 7-syllable unit like  is an inherited variant of A: the
comparative evidence of Indic meter shows that dimeters could be 7-syllable as well as 8-
syllable units, with matching rhythm in the closing.57 Matching rhythm in the closing
creates the perception of a missing initial syllable in a 7-syllable unit that coexists with an
8-syllable unit, whence the synchronic description of acephaly. But the comparative
evidence makes it clear that 7-syllable units are inheritances parallel to, not derived from,
8-syllable units.58 It is possible to offer a similar formulation for matching rhythm in the
openings of coexisting 8-syllable and 7-syllable units, also attested in Indic.59 In this
case, matching rhythm creates the perception of a missing final syllable in the 7-syllable
unit, whence the synchronic description of catalexis. Finally, I would suggest that
matching rhythm in the openings creates the perception of an added final syllable in
inherited 9-syllable units found coexisting with 8-syllable units, whence the synchronic
description of hypersyllabism. (On the subject of functioning metrical perception, I have
been guided by the insights of Cole 1988.) I cite the 9-syllable hipponactean (hipp),
which functions as a dimeter in Aeolic meter.60 Another example is the type
_m_oo_o , 61
attested as a functioning variant of the type
_m_oo_oo_
in compositions attributed to Aleman.62
In light of this diachronic perspective, let us reconsider the process of dovetailing. For
illustration, I turn to our earlier example of two dovetailing glyconics:
xx_oo_oJx_oo_o_ = glyc~glyc63
In synchronic terms, to repeat, dovetailing happens when the word-break is skipped
at the end of one metrical unit and transferred to the position after the first syllable of the
following metrical unit. In diachronic terms, however, what is needed to achieve such a
process of dovetailing is the systematic juxtaposition of phraseology that is rhythmically
shaped hipp with phraseology that is rhythmically shaped tele in order to achieve an
overall rhythmical effect that we may still describe synchronically as glyc-glyc. In
diachronic terms, then, dovetailing can evolve only in situations where the traditional
repertoire features a plurality of rigid meters that allow the perception of synchronic
derivation, one from the other, by way of acephaly or hypersyllabism or both. From the
standpoint of the Aeolic tradition, phrases shaped hipp and tele are functional variants of
phrases shaped glyc by the very fact that they are traditinally combined to produce the
sound-effect of double glyc. For this reason, we may in a given situation designate a hipp
as glyc~ and a tele as ~glyc, provided that we have reason to think that the given
phraseological repertory grew out of a system that produced the effect of dovetailing
glyconics. Following this line of thinking, I shall from here on use the sign "~" before a
given symbol for a given metrical unit to designate an acephalic variant of that metrical
unit, while "~" after a unit will designate a hypersyllabic variant. In other words, "~"
before or after a symbol will indicate that the unit designated by that symbol evolved from
traditionally dovetailed combinations. For example, ia~ can stand for a 5-syllable unit of
phraseology that has evolved in the context of a following 7-syllable unit of phraseology,
as in the iambic trimeter. Conversely, ~ia&IA can stand for a 7-syllable unit of
160 GREGORY NAGY
phraseology that evolved in the context of a preceding 5-syllable unit of phraseology.

Also, from here on I shall write /~ to indicate acephaly that has been disconnected from
any preceding pattern of dovetailing and ~/ to indicate hypersyllabism that has been
disconnected from any following pattern of dovetailing.
Keeping in mind these diachronic considerations, I am ready to streamline my system
of labeling for the dovetailing variants of the Aeolic units. In what follows, the left-hand
column of labels presents alternatives to the right-hand column of equivalent labels:
A xx_oo_o_ = glyc
 xx_oo_o = glyc~ =hipp
 ___ = ~glyc =tele
D x_oo_o = ~glyc~ = hage
A xx_oo = glyc < =pher
 none
 _ = ~glyc < = ~pher = tele < = reiz
D none
A x-oo_o_ = tele = ~glyc above
 __ = tele- = hage above
 ___ = ~tele
D_oo_o = ~tele-
A x_oo = tele <=reiz above
 none
C_oo_ = ~tele < = ~reiz
Dnone
In the case of equivalences like -glyc = tele, what decides the description is the
system of which the unit is a constituent. Let us test this principle on some common
metrical patterns in the Aeolic tradition, starting with the so-called Sapphic Strophe:
_o_x_oo_o =/~ia+tele-/
_o_x_oo_o = l~ia + tele-/
_o_x_oo_o_x_oo = l~ia+tele~tele <64
The primary word-break pattern in the third unit is as follows:
_o_x_oo_o_x|_oo
This word break reflects the evolution of the constiuent phraseology in terms of
tele~tele<, as distinct from tele/tele<, which would have yielded a word-break pattern that
is strictly avoided by the Sapphic Strophe:
65
_o_x_oo_o|x_oo
We turn next from the Sapphic Strophe to the Alcaic:

x_o_x_oo_o_ = ia~tele
x_o_x_oo_o_ = ia~tele
x_o_x_o_x_oo_oo_o = ia + ia~tele~1@da66
In the third unit the strict constraint against the word-breaking pattern
x_o_x_oJx_oo_oo_o
reflects the dovetailing pattern ia~tele.67

What follows are some further examples of Aeolic combinations:
x_o_x|x_oo_o_ = ia+glyc68
_o_xx_oo_o_ = ~ia + glyc69
x_o_x_oo_o = ia + tele~70
We see from the Sapphic and Alcaic Strophe and from these other Aeolic examples
that, just as a glyc allows trimeter formation with a preceding ia that dovetails into it, so
also does the tele, and so on. From a synchronic point of view, the tele functions as an
actual alternative to the glyc, experiencing acephaly on its own like the glyc, as if for the
very first time.
In light of the preceding discussion of dovetailing, we are ready to survey the
building-blocks of the dactylo-epitrite meters. I begin by reviewing the simple trimeter
pattern known as the iambelegos:
x_o_x|_oo_oo_ = ia+pros71
We find also the reverse trimeter pattern in the combination known as the enkomio-
logikon:
_oo_ooJx_o = l~proslia~l72
162 GREGORY NAGY
There is also a type without initial acephaly:

x_oo_ooJx_o = pros/ia-/73
Moreover, there are attestations of the enkomiologikon where we see the results of
dovetailing between the two units:
-oo_oo-x|_o_x = I-pros-I-ia-/14
We may note too this similar pattern:
x_oo_oo_x|_o_o = pros~l~ia& ia<15
Having surveyed these trimeter patterns, we now have a working repertory of the
major building-blocks that constitute archaic dactylo-epitrite meters, as attested most
clearly in the corpus of Stesichorus:
ax_o_ =ia
b x_o_x =ia~
 _o_ = -ia
d _o_x = -ia-
A x_oo_oo_ = pros
 x_oo_oo = pros~
 ___ = ~pros
D _oo_oo = ~pros~
In the dactylo-epitrites of Stesichorus, there are clear signs of parallelism in the
distribution of these parallelel pros and ia segments: for example, both pros- and ia- tend
to be placed in the closing of metrical sequences.76 A typical sequence is the following
from the Thebaid of Stesichorus:
_oo_oo|x_o =/-pros/ia-/ = C+b 77
Besides trimeter formations, we also see combinations of dimeters, as in the
following examples taken from the Nostoi of Stesichorus (PMG 209):
_oo_oo|x_oo_oo = /-pros/pros-/ = C+B
_oo_oo_x|_oo_oo =/-pros-/-pros-/ = B+D78
Although some metricians prefer to distinguish such patterns from dactylo-epitrites by

calling them dactylo-anapests,79 I shall use the term "dactylo-epitrite" to cover both
patterns.
In the immediately preceding example from Stesichorus, we must take note of a
metrical innovation: it is a new metrical license, not yet marked explicitly in the schemes
above for the Nostoi, where one long syllable is optionally substituted for two shorts (m).
As I have argued elsewhere, this metrical innovation, pervasive in the diction of
Stesichorus, is the reflex of the Greek linguistic innovation of contracting short vowels
originally separted by intervocalic *s and *i 8 0 We must also take note of the converse of
this innovation, where two short syllables are optionally substituted for one long. In the
corpus of Stesichorus, this pattern of substitution is restricted to those long syllables that
occur in slots that traditionally allow free variation between a single long and a single
short, as for example in the first syllable of the pros. I shall use the symbols A' and B' to
reflect this metrical innovation:
a __ -ia
b __ = ia~
 _0_ = ~ia
d __ = -ia~
 ___ = pros
' n___
" m___
 __ = pros-
' n__
" m__
 ___ = ~pros~
D _00_00 = ~pros~
This innovation is attested in such compositions as the Thebaid of Stesichorus, where
the pros (as also its variants) allows a new variation in the first syllable: when long, then
optional two shorts instead of the single long; when short, no substitution. I draw
attention to the following example:
_m_oo|n_oo_oo =C+B' =/~pros/pros~/81
164 GREGORY NAGY
The patterns symbolized by A" and B" in the scheme above reflect a further metrical
innovation, attested in such compositions as the Geryoneis of Stesichorus (SLG 7-87),
where the pros allows yet another new variation in the first syllable: only long, with
optional two shorts instead of the single long. I cite the following example:
m_m__oo =pros~/ = B"
m_oo|m_oo_oo_"m_oo = ?/pros"?/ =?/A"I?
m_oo_"m_oo_oo =?"pros/ =?B".82
I indicate with the symbol "?" the sequences m_oo_ and m_oo , which I take to be
derivatives of x_oo_ and x_oo . As I shall argue presently, such sequences as x_oo_
and x_oo result from resegmentations of choriambic internal expansion. For now,
however, the essential point to observe about the patterns in the scheme above is that the
distinction between "m" as derived from the first syllable of the pros and "m" as derived
from the double-short at syllables 3-4 and 6-7 of the pros is blurred. The resulting effect
has aptly been described by one expert as a "river of dactyls."83
This term raises an important question: what is the etymology, as it were, of the
dactyl? In our earlier survey, where the emphasis had been on the synchronic point of
view, we had seen that dactyls (da = _oo), like choriambs (ch = _oo_), figure in the
internal expansion of patterns like the glyconic (glyc), in shapes that have been
represented as glyc@ch, glyc@2ch, glyc@da, glyc@2da, and so on. But now the notion
of internal expansion must be redefined from a diachronic point of view. In the case of
glyc@ch and glyc@2ch, for example, the insertion of one or two choriambs (_oo_ or
_oo oo_) between the opening (xx) and the closing (_oo_o_) of a glyconic (xx_oo_o_)
is modeled on the insertion of a choriamb (_oo_) or iamb (x_o_) between the opening
(_oo_ or x_o_) and closing (_oo_ or x_o_) of choriambic and iambic dimeter patterns.83
In the choriambic and iambic patterns, however, what is synchronically an insertion is
diachronically an opening preceded by a reapplied opening, which in turn may be
preceded by yet another reapplied opening. Both the single and the double reapplied
openings are attested in one particular fragment of Anacreon:
_oo oo oo_x_o_ = CH+CH+CH&ia'
_oo oo_x_o_x_o_ = CH+CH&ia'+IA85
The pattern of reapplied opening _oo_ or openings _oo oo_ that precede _oo_o_o_
is reinterpreted as a pattern of opening _oo_ followed by an internally expanded _oo_ or
_oo oo_. By analogy, the glyc xx_oo_o_, which as we have seen is a functional variant
of the choriambic-iambic dimeter _oo_o_o_, becomes internally expanded as the trimeter
glyc@ch and the tetrameter glyc@2ch. These trimeter and tetrameter patterns have already
been described in the following way from a synchronic point of view: the opening
segment xx("Aeolic base") of xx_oo_o_ (the glyconic) is followed by an internally
expanded _oo_ or _oo oo_. Let us consider the major word-break patterns in
glyc@2ch:
xx oo | oo | oo  = X]Y|Z
xx-oo |oo | oo  = X~Y|Z
xx_oo|_oo |oo_o_ = X|Y~Z
xx_oo |oo |oo_o_ = X~Y~Z86
I draw attention also to a common pattern of word-breaking within sector X of the
glyc@2ch:
xx|_oo__... = X(=V|W...)...
xx_|oo__... = X (=V~W...)...
By now we have seen not one but several dovetailing mechanisms leading to a
"syncopated" _oo_, that is, oo_.
Next, I draw attention to the segments that I have labeled X and X~, xx_oo_ and
xx_oo , within the metrical pattern glyc@2ch. These segments, I suggest, are
analogous to x_oo_ and x_oo , which I have labeled with the symbol "?" earlier on, in
the "river of dactyls" built from prosodiaka, as in the Geryoneis of Stesichorus. I stress
that these patterns are followed by patterns in oo_... . We may compare the presence of
"syncopated" choriambs shaped oo_ in the prosodiaka of dactylo-epitrites, as clearly
attested in Pindar and Bacchylides.87
Having examined the basic configuration oo_ in both the Aeolic and the dactylo-
epitrite metrical traditions, we now have the background that we need for examining the
phenomenon generally known as dactylic expansion, which is attested in both these
traditions.
166 GREGORY NAGY
Let us begin with the Aeolic metrical tradition. The "XYZ" components of glyc@2ch,
I suggest, are cognate with those of a sequence that has earlier been described
synchronically as glyc@2da, that is, a glyconic internally expanded by two dactyls. From
a diachronic point of view, the expanding dactylic patterns are actually resegmented
variants of choriambic patterns. In the metrical schemes that follow, I give some typical
word-break patterns of the type glyc@2da,88 labeling the components marked off by the
word-breaks in terms of the "XYZ" components of glyc@2ch:
xx_oo_|oo_oo_o_ = X|~Y+Z = glyc@2da
xx_oo_|oo_|oo_o_ = X]~Y|~Z = glyc@2da
xx|_oo_oo|_oo_o_ = V|Z*|Z = glyc@2da
xx_|oo_|oo_|oo_o_ = V~|~Z *|Z = glyc@2da
xx_|ooJoo_|oo_o_ =V~|~Y|~Y~Z = glyc@2da
etc.
(The "*" following a symbol designates a sequence where the final syllable is
restricted to a short quantity.) We would expect such equivalences between constituents
of glyc@2ch and those of glyc@2da to be a matter oí phraseological correspondences in
equivalent metrical slots.89
Let us now move from the Aeolic to the dactylo-epitrite traditions. By way of review,
here are the units that we have already surveyed as the basic constituents of dactylo-
epitrite meters:
a x_o_ = ia
b x_o_x = ia~
 __ =~ia
d _o_x = ~ia
A x_oo_oo_ = pros
A'n_oo_oo_
A'm_oo_oo_
 __ =pros~
B'n_oo_oo
B"m_oo_oo
 ___ = ~pros
D _oo_oo = ~pros~
These units are not only prototypical of those found in the earlier dactylo-epitrite
meters of Stesichorus, or the later ones of Pindar.90 They are also identical with some of
the major metrical shapes that consititute the verses of the dactylic hexameter, the elegiac
distich, and the iambic trimeter. We have already observed at length the relationship of the
iambic trimeter to the dactylo-epitrite meters as also to the Aeolic meters; now we may
concentrate on the dactylic hexameter and the elegiac distich. The traditional phraseology
of the dactylic hexameter is distributed in such a way as to leave the following distinctive
patterns of word-breaking:
1. _m_m|m_m_m__ (Penthemimeral caesura)

2. _rn_m_o|o_m_m (trochaic caesura)
3. _m_m_m|m_m (hephthemimeral caesura)
4. _m_m_m_oo|_m (bucolic diaeresis):
Also, there is a constraint against word-breaking of the type
_m_m_m_o|o_m .
This phenomenon is commonly known as Hermann's Bridge.91 Since 99% of
Homeric hexameters have either pattern "1" or pattern "2,"92
we may note with interest
that pattern " 1 " corresponds to dactylo-epitrite formations that we have already seen, of
the type
_m_ooJn_oo_oo =C+B' =/~pros/pros~/93
The pattern " 1 " of hexameter could be described as C+B", corresponding to the C+B'
pattern of the dactylo-epitrite meter immediately above:94
_m_m_|m_m_ =C + B" =/-pros/pros-/
Similarly, the so-called "pentameter" of the elegiac distich could be described as
_m_m_]_oo_oo_ = C+C = /~pros/~pros95
Pattern "4" of hexameter corresponds to a common word-break pattern in

Stesichorean diction: ... w | _ w
Moreover, Stesichorean diction avoids word-breaks of the type
168 GREGORY NAGY
... _o|_o_w
and this pattern of avoidance is directly comparable to Hermann's Bridge.97 Pattern
"3" of hexameter corresponds to yet another common word-break pattern in Stesichorean
diction:
... Jm_w
And there are even sporadic traces of pattern "4":
98
_m_m_o|o_m_oo
Even though these analogues in Stesichorean meter yield examples of all four major
types of word-breaking pattern in the hexameter, we still cannot say that the hexameter is
attested in Stesichorus. In the meters of Stesichorus, a pattern like
_m_oo_m_m_oo ,
which looks on the surface exactly like the hexameter, is a functional variant of
_m_x_m_oo
Here the sequence _x_ is incompatible with the hexameter, which regularly avoids the
dactyl-thwarting pattern _o_." For example, let us consider the following match:100
Stesichorus Oxyrhynchus Papyri 2735.17

for Castor the horse-tamer and for Pollux, good at boxing
Castor the horse-tamer and Pollux, good at boxing

Despite the fact that the match here is both metrical and phraseological, the
Stesichorean line cannot be called a dactylic hexameter because the slot occupied here by
και may allow not only a long syllable (as here at line 17 of Oxyrhynchus Papyrus 2735)
but also a short syllable (as at line 10). Moreover, the Stesichorean line can allow word-
breaking patterns that are altogether foreign to the hexameter. In the Nostoi of
Stesichorus, for example, we find, besides
_m_oo_ |x_oo_oo ,101
the following word-break pattern as well:

102
_m_oo_x|_oo_oo
Thus there is a greater variety of phraseology accommodated by the meters of
Stesichorus than by the hexameter of Homer.103
How, then, does the dactylic hexameter maintain its frame while promoting dactylic
rhythm? The answer has to do with the phraseological heritage of the hexameter. The
phraseology that constitutes the repertoire of hexameter, I submit, comes from the
rhythmical frame of not only the dactylo-epitrite but also the Aeolic meters.
In an earlier attempt,104 I had argued that the framework of hexameter can be derived
from an Aeolic meter.
xx_oo_oo_oo_oo = pher@3da105
In the present work, I shift the emphasis, going beyond my specific argument that the
pher@3da provided the actual metrical frame for the hexameter. Instead, in light of the
preceding discussion of inner expansion by dactyls as a derivative of inner expansion by
choriambs, I offer a more general argument, that the hexameter was shaped by the same
traditional phraseology that had shaped such Aeolic meters as pher@3da, glyc@2ch,
glyc@2da, as well as the dactylo-epitrite meters.106 For the phraseology to be
interchangeable between hexameter on one side and such meters as the pher@3da,
glyc@2ch, and glyc@2da on the other, I posit two innovations on the Aeolic side, taking
as an example the metrical frame of pher@3'da:
107
1) optional replacement of _oo by
2) specialization of the initial "Aeolic base" xx as , with optional
replacement of by _oo.
These innovations would be parallel to those already discussed in the case of the
archaic dactylo-epitrites of Stesichorus:
1) optional replacement of _oo by
2) specialization of the initial x as_, with optional replacement of
_by oo.108
170 GREGORY NAGY
To repeat, the optional synchronic substitution of for _oo can be explained as a

109
diachronic reflex of vowel-contraction. The substitution of for _oo accords with the
theory that phraseological patterns generate metrical patterns which then assume dynamics
of their own and even regulate any incoming non-traditional phraseology.110 There is
internal evidence to show that the pattern is foreign to the second, third, fourth, and
fifth feet of the dactylic hexameter, since it involves phraseological restrictions that do not
apply to the pattern _oo: words with a spondaic ending are shunned, whereas those with a
datylic ending are not. In the first foot of the hexameter, on the other hand, this restriction
does not apply, and words with a spondaic ending are common.111 It bears repeating that
the optional substitution of _oo for in the first foot of hexameter could not have
happened without a pre-existent pattern of optionally substituting for _oo in the other
feet. Then too, in a composition like the Geryoneis of Stesichorus, where we find no
Aeolic base analogous to the first foot of hexameter, we note that phrases containing the
shape can be substituted for those containing the shape _oo but not the other way
112
around. Further, as in hexameter, any sequence shaped that is substituted for _oo
regularly avoids a following word-end; instead, the word-ending is bridged to the
position after the next long syllable.113
We are now ready to examine examples of phraseological interchange between
hexameter and the Aeolic meters, let us begin with an Aeolic pattern that we have already
considered in detail, the glyc@2ch:
xx oo | oo | oo  = X|Y|Z = glyc@2ch
xx_oo |oo_|_oo_o_ = X~Y|Z = glyc@2ch
xx_oo_|_oo |oo_o_ = X]Y~Z = glyc@2ch
xx_oo_|oo__Joo_o_ =X~Y~Z114 =glyc@2ch
Now, let us consider where these shapes could fit within the framework of
apher@3da. Again, the "*" following a symbol designates a sequence where the final
syllable is restricted to a short quantity:
xx_oo_|oo_oo|_oo_ = X|~Z*|Y_|115
Keeping in mind the posited innovation of replacing initial xx with _oo, we find that
the hexameter actually accommodates phraseology corresponding to the glyc@2ch in
exactly these slots:
_oo_oo_|oo_oo|_oo = X|~Z*|Y~|
As a case in point, I cite the hexameters in Hesiod Works and Days 582-593 and
Shield of Herakles 393-401, to be compared with the stichic series of glyc@2ch in
Alcaeus F 347. To indicate the relative metrical position of the phraseology, I propose to
use the symbols for line- initial and line-final position, 1 2 3 4 for the relevant caesuras of
hexameter, and    D for those of the glyc@2ch:
#_oo_oo_|o|o_|oo|_oo # = hexameter
1234
#xx oo_|_|oo_|_|oo_# = glyc@2ch
ABCD
Let us proceed to examine the following correspondences:
xx_oo_|_oo_|oo_o_ = X|Y~Z = glyc@2ch
A D
-oo_oo_|oo_oo|_oo = X|~Z*|Y~| = hexameter
1 4
A (sector X|) in Alcaeus v. 7 corresponds to

τ'| in Works 582. A | D (sector in Alcaeus v. 3 corresponds to 4|
in Works 582 and Shield 393. in Alcaeus
v. 2 corresponds to 1| | 4 in Works 588.116
Now let us consider again the glyc@2da:
xx_ooJoo_oo_o_ = X|~Y+Z = glyc@2da
xx_ooJooJoo_o_ = X|~Y|~Z = glyc@2da
xx|_oo_oo|_oo_o_ = v|z*|z = glyc@2da
xxJoo_oo|_oo_o_ = V~I~Z*|Z - glyc@2da
xxJooJooJoo_o_ = V~I~Y|~Y~Z = glyc@2da
etc.
Again we find that the hexameter actually accommodates phraseology corresponding
to what we find in the glyc@2da, as with the placement of wording shaped AZ* at the
middle of the hexameter:
172 GREGORY NAGY
_oo_oo|oo_oo|_oo_ = X|~Z*|Y~117
There is an interesting alternative pattern, where wording shaped oo_oo_oo at the
end of hexameter corresponds to wording shaped /~Z*/ at the end of one glyc@2da
followed by wording that occupies the Aeolic base in a consecutive glyc@2da:
xx_oo|oo|oo_o_ = X|~Y|~Z = glyc@2da
xx|... = V|...
_oo_oo_|oo_|oo_oo| = X|~Y|~Z*||V = hexameter.118
The same phenomenon occurs in wording shaped /~Z*/at the end of one glyc@2ch
followed by wording that occupies the Aeolic base in a consecutive glyc@2ch.119
Even more common in hexameter is wording shaped for pher sequences, which can
be described as pher* followed by *~pher@da.120
It is important to stress in passing that, just as Aeolic meters influence each other, we
can expect hexameter, even if it is partially inherited from Aeolic meters, also to influence
them.
To sum up, dactylic hexameter can be derived from both the Aeolic and the dactylo-
epitrite patterns inherited by Pindar. Also, elegiac distich and iambic trimeter are cognate
with dactylo-epitrite. Further, as we shall now see, the three basic meters of poetry,
dactylic hexameter, elegiac distich, and iambic trimeter, are not only derived from song
but also differentiated from the corresponding meters of song.
The meters of Greek song, which I shall for the moment call "lyric" poetry, are either
strophic. built on the principle of the colon, the constitutive element of both the relatively
simpler stanzas of monody and the more complex stanzas of the choral medium, or
stichic, built on the principle of the verse.121 Other available terms of description are
"asynartetic," where the divisions between cola are overt, and "synartetic," where they are
latent. 122 It is important, however, not to confuse "synartetic" with strophic and
"asynartetic" with stichic, in that the category of "asynartetic" accommodates not only
verse but also strophes where the colon is clearly delineated; conversely, the category of
"synartetic" does not accommodate strophes where the colon is clearly delineated.
Granted, there are cases where the direction of development is from "asynartetic" to
"synartetic."123 Still, it hardly follows that the constituents of the strophe are built from
the constituents of the verse. As the discussion proceeds, it will in fact become evident
that the direction of development is the reverse: from colon to verse-portion, not from
verse-portion to colon. In any case, "asynartetic" is a cetagory that is roughly half-way
between strophic and stichic.
The notion of "asynartetic" is compatible with that of parakatalogē,a category of song
where both melody and dance become reduced or eliminated altogether. We may now add
that, whereas both melody and dance are bound to a strophic framework, the category of
parakatalogē is associated with an asynartetic or stichic framework.124
The stichic meters of lyric are a transitional point of differentiation from song,
whereas total differentiation is achieved in the stichic meters of non-lyric, that is, of
poetry proper.125 It is the stichic rather than strophic meters of lyric that are actually
attested as usable for extended narrative of a type parallel to epic, composed in the dactylic
hexameter, which is the stichic meter of non-lyric par excellence. A worthy example is
Sappho F 44, a poem with a heroic setting: it is composed in a stichic meter126 that is
clearly cognate with various strophic meters of Lesbian lyric poetry.127 It is such stichic
meters of narrative lyric poetry, conventionally sung to the accompaniment of the lyre,
that are doubtless more closely related than is the spoken meter of hexameter to the format
of the South Slavic guslar who sings to the accompaniment of the gusle.,128
To repeat, the fundamental argument in our examination of lyric meters has been that
the sequences of cola in the strophic meters of song are cognate with the verses in the
stichic meters of poetry. Perhaps the clearest examples that we have surveyed are the
strophic sequences of Stesichorus, which are built from units such as the following:
a __ = ia
b __ = ia~
 _0_ = ~ia
d __ = ~ia~
L ____ = ia&IA
M ____ = ~ia & IA
A ___ = pros
 __ = pros-
" __ - pros-
174 GREGORY NAGY
 ___ = ~pros
D __ = ~pros~
These shapes are not only prototypical of those found in, say, the so-called dactylo-
epitrite meters of a Pindaric strophe: they are also identical with some of the major
components in the meters of poetry, that is, in the dactylic hexameter (CET), in the
elegiac distich (CB"|CC), and in the iambic trimeter (bM).
Using the terminology introduced at the beginning of this presentation, I am saying
that these meters of poetry are not only derived from song but also differentiated from the
corresponding meters of song. As a "finishing touch" of differentiation between poetry
and song, what can even happen is that the meters of song will avoid patterns that have
been appropriated for poetry: thus for example the metrical repertoire of Pindar's dactylo-
epitrite poetry contains the ingredients needed to generate equivalents of the dactylic
hexameter, elegiac distich, and iambic trimeter, and yet it is precisely these patterns of
spoken poetry that the lyric poetry of Pindar, let us call it song, consistently avoids.129 It
appears that the parent forms were covering their tracks, as it were, shading over the
various features that distinguish their respective daughter forms.
With any differentiation of poetry from song through the loss of melody, there would
have to come about a new structural strain in the oral tradition. Melody can be an
important feature in the mnemonics of oral tradition in song, as we know from the studies
of folklorists who scrutinize the transmission and diffusion of song: to put it plainly,
melody helps remember the words.130 We are reminded of the anecdote about the
Athenians captured after the debacle at Syracuse who ingratiated themselves with their
Syracusan captors by singing passages from Euripides: these memorable passages were
evidently parts from choral lyric, not iambic trimeter (as we see from the wording of
Plutarch Nicias 29.3).131 In terms of a differentiation of oral song into oval poetry as
opposed to oral song, I would offer this axiom: with the structural strain brought about by
the loss of melody in poetry, there would come about, for the sake of mnemonic
efficiency, a compensatory tightening up of rules in the poetic tradition.132 This
tightening up would entail an intensification of both phraseological and prosodic
regularities, such as what we see in the formulas and meters of Homer, Hesiod, and the
old elegiac and iambic poets. I would also suggest that the conventional understanding of
"formula," stemming ultimately from Milman Parry's study of Homeric phraseology,

applies precisely to such regularities stemming from the differentiation of oral poetry from
oral song. In other words, what is currently understood as the "formula" is to be seen as
characteristic, more narrowly, only oforalpoetry as opposed to song; in order to account
for the distinct regularities of oral song as opposed to poetry, the concept of "formula"
would have to be considerably broadened.133
For song, it seems most useful to distinguish three dimensions of the formula:
phraseological, rhythmical, and melodic. These three dimension correspond to the
tripartion of melos 'lyric poetry' into logos, rhuthmos, and harmonia in Plato Republic
398d (cf. Aristotle Poetics 1447a21-23). The rhythmical dimension would be represented
by meters and the melodic, by modes. The process of oral composition in song, then, can
be conceived as an interaction of phraseology, rhythms, and modes.
NOTES
1 For a convenient metrical overview, see the analysis of Snell in the SM edition of Pindaric
fragments, pp. 162-174. In only one case, Pindar Olympian 13, is there a coexistence of the two
types within one composition (Aeolic modulating into dactylo-epitrite). For an exceptional case of a
Pindaric song composed in neither Aeolic nor in dactylo-epitrite meters, I cite Olympian 2,
composed in Ionic meters.
2 The following symbols are henceforth in effect: _ = long syllable;  = short; x = anceps = long or
short; m = long or two shorts; n = anceps or two shorts.
3 For more on the term "colon": West 1982:5-6.
4 E.g. Sappho F 96. In this presentation, all fragments of Sappho and Alcaeus are taken from the
edition of Voigt (V). The metrical abbreviations, such as glyc for "glyconic," are modifications of
those found in West 1982, whose taxonomy of Greek meters I shall follow here, with some
exceptions to be noted in due course.
5 See the discussion of Devine and Stephens 1975, especially p. 204.
6 E.g. Alcaeus F 112.
9 E.g. Sappho F 94.
10 E.g. Sappho F 44.
11 E.g. Sappho F 111.1.
12 E.g. Sappho F 140.1 had neglected this poem in Nagy 1974.47, where I claim, wrongly, that there
are no pherecratics to be found with choriambic expansion {FORMAT: brackets over first two _oo_}
176 GREGORY NAGY

16 For the terminology, see West 1982:30.
18 E.g. Sappho 141.
19 E.g. Alcaeus F 130.4.
20 E.g. Sappho F 111.3.
21 Concerning the Indo-European heritage of Greek meters in general, I cite the pioneering studies of
Meillet 1923, Jakobson 1952, and Watkins 1963. Cf. also Schmitt 1967:307-313, West 1973,
1973b, Nagy 1974. For a survey of these works and others (such as Peabody 1975 and Vigorita
1977), see Bowie 1981:16-28. Commenting on the view of Meillet p. 76 that the Indic and Greek
correspondences in metrical patterns are too close to be coincidental, Bake 1957:195 observes that
this view "is equally applicable when comparing the systems of music of the two ancient
civilizations."
22 Vine 1977, 1978.
23 Nagy 1974:37-44. Similarly, the relationship between glyc and glyc@2ch would correspond to the
relationship between iambic dimeter and iambic tetrameter.
24 Ibid.
25 Ibid; also Vine 1977 and 1978. Cf. also Ananthanarayana 1973, who argues that trimeter is derived
from dimeter and who posits for trimeter a basic 11-syllable unit and a derived 12-syllable unit; as
the discussion proceeds, it will be clear that I disagree with the latter point and only partially agree
with the former.
26 E.g. Anacreon PMG 428.
27 In the present work, I have tried to be consistent in using four-letter symbols for whole dimeters
{e.g. glyc) and two-letter symbols for openings or closings of dimeters and trimeters {e.g. ia, ch,
etc.).
28 Nagy 1974:37-44, 279-287; skepticism in West 1974:458, but tentative agreement in West
1982b:296n45. All this is not to say, however, that the dimeter is necessarily older than the trimeter
(see the useful discussion of Vine 1978:175 and 191n7). It is only to say that the constituents of
trimeter are cognate with the constituents of dimeter, and that we can therefore explain the trimeter
in terms of the same constituents that we find in dimeter.
29 Nagy ibid. Cf. West 1982b:295.
30 E.g. Semonides F 7.12 W; Margites F 1 Allen p. 156.
31 E.g. Semonides F 7.3 W. Discussion in Nagy 1974:292-293.
32 For statistics on primary 5|7 and secondary 7|5 patterns in iambic trimeter, see Korzeniewski
1968:45-46.
33 E.g. Alcaeus F 70 and 117b.26 ff. (where the strophe is shaped ia+glyc/glyc@ch).
34 Jakobson 1952 passim.
35 E.g. Alcaeus F 355 and F 360, which show this word-break pattern, as opposed to e.g. F 359,
which does not.
36 E.g. Alcaeus F 360; Bacchylides 18.9 SM.
37 For the term "dovetailing," borrowed from the image of a dovetail joint in carpentry (so West
1982b.295), see Maas 1962.44.
38 Nagy 1974:279-287.
39 E.g. Stesichorus Thebaid 222 (strophe / antistrophe line 5). I refer to the version of the text (Lille
Papyrus 76abc) published by Haslam 1978:32-33, with accompanying metrical analysis. On the
general subject of Stesichorean meter, the work of Haslam 1974, 1978 is essential; for important
parallels illustrated by poetry that is attested in the surviving epigraphical evidence, see Gentili and
Giannini 1977:19-22.
40 E.g. Stesichorus Destruction of Ilion (SLG 88-132) epode lines 6-7 in the colometry of Haslam
1978.
41 E.g., again, Alcaeus F 70 and 117b.24 ff. (where the strophe is shaped ia+glyc/glyc@ch).
43 I say "reconstructed" because actual attestations of initial xxxx reveal distinct trends of patterning,
not completely free variation. One such pattern is initial _oo_ instead of the initial xx_o of
glyconic. E.g. Sappho F 95.6, where _oo_o_o_ functions as a variant of xx_oo_o_ (cf. West
1982.31). For the term "irregular glyconic," see Watkins 1963:203-206. Correcting some of the
scansions offered by Watkins, Itsumi 1982:59n7 says that "Watkins is rash to find in Greek metres
traces of initial unfixed syllables supposedly of Indo-European origin." And yet, Itsumi has not
disproved the existence of these traces; he has simply shown that an initial configuration xxxx is not
attested in all the possible varieties that are expected on the basis of the comparative evidence.
Itsumi assumes, it seems, that rhythmical freedom in the first half of the choriambic dimeter can be
posited only if we find every possible combination of long and short quantitites there. No such
assumption is necessary, since the freedom of the first half is already assured by the simple fact of
variability in its rhythmical patterns, as opposed to the invariability of the patterns in the second
half. Even if the third of the first four syllables in attested choriambic dimeters is regularly long
(Itsumi p. 60), the variety of patterns in the first, second, and fourth syllables still constitutes a
sector of variability in the entire first half of choriambic dimeter, as opposed to a sector of
invariability in the second half. There is a symmetry in the opposition between a variable opening
half and an invariable closing half. I recommend a close reading of Allen 1973:106, with its
enhanced perspective of comparative evidence on the topic of interaction between rhythmically
flexible openings and rigid closings in metrical units, as a counterweight to the limitations imposed
by Itsumi's strictly descriptivist interpretations of available metrical data. On the general tendency of
maintaining a tension between ideal and actual patterns at the beginning of the line and of solving
this tension at the end of the line, see Allen p. 110. Moreover, the notion that an initial pattern of
xx_o in choriambic dimeter precedes the pattern xx_x historically (Itsumi p.71) seems to me
counterintuitive, from the standpoint of comparative metrics. I also disagree with the notion that the
choriambic dimeter may have been "created" by way of changes in the glyconic (ibid.).
44 On the restriction of the initial pattern xxxx to xx_x in choriambic dimeter, see the note
immediately above. The choriambic dimeter is most clearly attested in Anacreon PMG 349, 357;
Corinna PMG 654, 655; Aristophanes Wasps 1457-1461, 1469-1473; Clouds 572. See also
Merkelbach 1967:161-162, who has identified twelve consecutive lines of choriambic dimeter verse
in a fragment of what seems to be the parabasis of the Heroes of Aristophanes. The choriambic
dimeter is also found as the first part of a larger unit known as the Eupolidean verse, as most clearly
attested in the parabasis of Aristophanes' Clouds (518-562). On the recitative (as opposed to sung)
character of the Eupolidean verse, of which the choriambic dimeter is a constituent, see Poultney
1979:140-141. For more on the choriambic dimeter, see Poultney 1979:142-144. Also Itsumi 1982
(who cites neither Merkelbach nor Poultney). The choriambic dimeter and the glyconic are functional
178 GREGORY NAGY
variants in strophic responsion (cf. Itsumi pp. 59, 69). On the principles of strophic responsion, see
West 1982:5.
45 Nagy 1974:294-295.
46 Watkins 1963:203-206; cf. Nagy 1974:30-31.
47 Ibid. That the early Greek metrical law of avoiding three consecutive shorts was an outgrowth of the
language itself is suggested by a residual phonological rule in classical Greek, surviving on the level
of morphophonemics: I am referring to the constraint against three consecutive short syllables in
comparatives and superlatives of thematic adjectives: thus e.g. soph-ō-teros as opposed to lept-o-
teros. See Householder and Nagy 1972:758.
48 To repeat, the last syllable of a line is indifferent in quantity, even though I have been writing it as a
long ( J .
49 E.g. Anacreon PMG 347, 417. See West 1982:57. This pattern is cognate with the Indic dimeter
shaped xxxx_o and known as the trochaic Gāyatrī: see Nagy 1974:170-171.
50 E.g. Sappho F 95.9, 96.7. West 1982.31 recognizes this unit as a variant of the glyc, it is
interchangeable with glyc in e.g. Anacreon PMG 349 (see West p. 57). It can be interpreted as a
derivative of the choriambic dimeter. From here on, the symbol tr'&CH will indicate this particular
variant dimeter.
51 E.g. Sappho F 95.6. West ibid recognizes this unit too as a variant of the glyc. It is clearly a
derivative of the "irregular glyconic." Whereas the glyc generalizes the pattern long-short for the
third and fourth syllables of the opening, resulting in xx_o, the unit under consideration generalizes
the pattern short-long, for the third and fourth and long-short for the first and second, resulting in
_oo_. Whereas the pattern long-short for the first and second is a constant for this unit, the same
pattern is only a tendency for the glyc (cf. West 1982:30). From here on, the symbol CH&ia' will
indicate this particular variant dimeter.
52 E.g. Sappho F 112; Anacreon PMG 385, 386.
53 E.g. Alcman PMG 1.57. For the name, see West 1982:30n3.
54 Nagy 1974:34, with illustrations.
55 Nagy p. 35.
56 Nagy p. 36, with illustrations.
57 Vine 1977.
58 Ibid..
59 For which see again Vine ibid.
61 As we shall see below, the pattern _m is a reflex of xx. {FORMAT: m = longum over biceps}
62 Alcman PMG 1.49 (first type), e.g. 1.35 (second type); cf. also the 9-syllable line in Aleman PMG
39.1.
63 E.g., again, Alcaeus F 360; Bacchylides 18.9 SM.
64 The Sapphic Strophe is the format, for example, of all the compositions in Sappho Book I.
65 In his synchronic metrical description, West 1982:33 provides the following interpretation of the
Sapphic Strophe. For the first two units, he posits >ia + hage. For the third, he hesitates between
>ia+hage+adon or <ia+tele~reiz. Either way, there is a discontinuity with the first two units. To
posit an adonic (adon, or _oo ) is to invoke a segmentation for which I know of no diachronic
explanation. To posit tele clashes with the preceding hage. About his schematization of the third
unit, West remarks (ibid) that it "obscures the essential point that it is a distended form of the first
and second." I hope that the schematization that I offer above helps establish that point.
66 For the Alcaic Strophe, see e.g. Alcaeus F 129. Again, West (ibid.) notes that the third unit is an
"amplification" of the first and second.
67 In the first two units, as West 1982b:296 points out, the word-break predominantly occurs at
x_o_x/_oo_o_. Again, a reflex of ia~tele. There is an analogous type of dimeter in the Indic
tradition, known as the Bhārgavā: see Nagy 1974:180-183.
68 E.g., again, Alcaeus F 70 and 117b.24 ff. (where the strophe is shaped ia+glyclglyc@ch).
69 E.g. Sappho F 98 (where the strophe is shaped glyclglycl~ia + glyc). I write ~ia+glyc instead of
~ia~glyc because the word-break does not necessarily occur after syllable 4.
71 E.g. Stesichorus Destruction ofIli(SLG 88-132) epode lines 6-7 in the colometry of Haslam
1978.
72 E.g. Stesichorus Thebaid strophe / antistrophe line 2.
73 E.g. Stesichorus Thebaid epode line 2.
75 E.g. Archilochus F 168-171 W.
76 Cf. Haslam 1978:56.
77 E.g. Stesichorus Thebaid strophe / antistrophe line 2; also Alcaeus F 383.
78 Cf. Haslam 1974:46.
79 E.g. Haslam 1978.
80 Nagy 1974:49-56. Also Allen 1973:255-259 and 1987:113-114.
81 E.g. Stesichorus Thebaid strophe / antistrophe line 1, as analyzed by Haslam 1978:33. This
variation (when long, then optional two shorts instead of the single long; when short, no
substitution) in the first syllable of pros is analogously extended to the first syllable in the ia
segments as well, as in the Thebaid strophe / antistrophe line 5.
82 E.g. Stesichorus Geryoneis (SLG 7-87) strophe / antistrophe lines 1-5, as analyzed by Haslam
1974:20-22, 31; cf. West 1982-50. From here on, the symbol """ indicates optional word-break.
83 Haslam 1974:32n50.
84 Examples and discussion in Nagy 1974:39-45 (though I now distance myself from the overrestricted
usage there of the terms "symmetry" and "asymmetry").
85 Anacreon PMG 388. The pattern CH&ia' in these combinations is cognate with the choriambic-
iambic dimeter (_oo_o_o_), a unit that functions as a variant of the glyconic.
86 The Aeolic samples of glyc@2ch that I have examined are: Alcaeus F 44, 50, 115a; 296b, 340-349;
cf. Theocritus 28, 30 and Callimachus F 400 Pfeiffer.
87 There is an inventory of attestations in the SM edition of Pindaric fragments, p. 168, and in the SM
edition of Bacchylides, p. xxvii; the sequence oo_ is labeled "d 2 ".
88 I follow the statistics on word-breaking in glyc@2da as compiled by Bowie 1981.35 on the basis of
the following texts: Sappho F 44.5-15, 26, 30-34; 47.2, 48-52; Alcaeus F 38.3-5, 7, 9; 141.3-4;
364; 365
89 This is not to say, of course, that glyc@2da cannot be a metrical frame for shapes other than what
we find in glyc@2ch: for example, we may analyze Sappho F 44.10 as containing phraseology
shaped pher* followed by *glyc.
90 For a useful survey of dactylo-eptitrite patterns in Stesichorus, Pindar, and Bacchylides, see Haslam
1978:54-57. My one basic disagreement with Haslam is that I view the pattern n_oo_oo_ as more
archaic than m_oo_oo_.
180 GREGORY NAGY
91 Maas 1962:60. In Nagy 1974:72, "%.01" should read "%0.1." On the special effects achieved by the
violation of Hermann's Bridge in Hesiod Theogony 319, see Solomon 1985.
92 Cf. Nagy 1974:56; roughly 60% have pattern "4."
93 E.g. Stesichorus Thebaid strophe / antistrophe line 1.
94 Cf. West 1973b:269; Gentili and Giannini 1977:28; Haslam 1978:39-40.
95 In line with my argument that the closing of the "pentameter" is particularly archaic (Nagy 1974:99-
101), it is worth noting that /~pros patterns tend to occupy strophe-final position in Stesichorean
diction; cf. Geryoneis (SLG 7-87) strophe / antistrophe line 9 and epode lines 7-8, as mapped out in
Haslam 1974:11. As for the hexameter of the elegiac distich, its patterns are statistically distinct
from those of the Homeric hexameter: see Greenberg 1985 and 1985b.
96 Data in Haslam 1978.43.
97 Ibid.
98 E.g. Stesichorus Thebaid 230 in the edition of Haslam 1978:33.
99 By "functional variation" here, I mean the same thing as what metricians call "responsion," that is,
where metrical sequences match between strophe and antistrophe. For a survey of patterns where a
sequence with "m" is in responsion with an otherwise identical sequence containing "x, " see Haslam
1978:39-40.
100 Pointed out by Haslam 1974:49.
101 E.g. Stesichorus Nostoi (PMG 209) strophe / antistrophe line 3.
102 E.g. Stesichorus Nostoi (PMG 209) strophe / antistrophe line 4; cf. Haslam 1974.46.
103 If the hexameter tends towards _oo_oo_m_oo_oo at the expense of _oo_oox_oo_oo and of
_oo_oon_oo_oo , this tendency can be expected to parallel a progressive restriction of incoming
phraseology. For vestigial traces of _oo_oo_x_oo_oo in hexameter, where wording shaped
_oo_oo_ is followed by o_oo_oo_, see e.g. Iliad IV 101, XI 697 (cf. West 1982:293n41). The
progressive generalization of _oo_oo_m_oo_oo over _oo_oox_oo_oo in hexamter helps
account for the absence of Homeric introductory phrases of the type as
discussed in Nagy 1974:85. Such phrases would have been ousted by the evolving constraint against
the sequence _o_ produced by fusion of wording shaped _oo_oo_ and o_oo_oo .
104 Nagy 1974:49-102, following up on the suggestions of Wilamowitz 1921:98.
105 Alcaeus F368.
106 On the partially Aeolic heritage of the traditional phraseology in dactylic hexamter, there is further
evidence, beyond what is offered in Nagy 1974, in Bowie 1981, especially pp. 32-46. My
reservation about Bowie's important work is that he undervalues his own important results by not
sufficiently heeding the discovery of Parry 1928b that the word-break patterns of hexameter are not
metrical devices actively deployed by the poet but metrical effects passively reflecting the junctures
where traditional phrases may begin or end (cf. Nagy 1974:57-61). Also, I find some unnecessary
presuppositions about the notions of "formula" and "oral poetry" (p. 41). At Bowie pp. 49-60, there
is a particularly important assessment, with bibliography, of the linguistic evidence for an Aeolic
phase in the evolution of the hexameter. I agree with his statement at p. 55: "in the case of possible
Aeolic forms in Homer, we are not dealing with Lesbian or non-Lesbian, so much as with
specifically Lesbian and generally Aeolic."
107 See Allen 1987:113-114.
108 As we have seen in Stesichorus Geryoneis (SLG 7-87). We have also seen that, in Stesichorus
Destruction of Ilion (SLG 88-132), initial χ still survives as a variant of initial m.
109 Nagy 1974:49-56. Also Allen 1973:255-259 and 1987:113-114.
110 Nagy 1974:196, 216; cf. Allen 1973:258.

111 Nagy 1974:55, with references. While accepting the special status of the first foot of hexameter in
allowing spondaic word-endings, Berg 1978:29 argues for the special status of the second foot as
well, on the grounds that this foot, unlike the third, fourth, and fifth feet, shuns dactylic word-
endings not just spondaic ones (cf. Allen 1973:291-292). He claims that this constraint in the
second foot is conditioned by the oncoming caesura in the third foot. It appears that Berg thinks of
caesura-patterns as diachronic shapers of phraseology; I prefer to follow Parry 1928b in thinking of
phraseology as the diachronic shaper of caesura-patterns (Nagy 1974:57-61). In what follows, we
shall have occasion to consider some phraseological reasons for the constraint against dactylic word-
ending in the second foot: that the third foot is a primary zone for syntactical boundaries in the
hexameter, as reflected by the primary caesuras. When the caesura comes after the long syllable of
the third foot, the second foot can have word-end after the dactyl only if the next word is a
monosyllable. While I am on the subject of Berg's 1978 article, I should note in passing that I am
horrified by his misunderstanding (which he betrays with his bracketed sign "!" at p. 18) of the
English idiom "hit upon" (Nagy p. 148), contained in a deliberately sarcastic sentence of mine that
mocks the topos of the "first discoverer" {ibid); Berg takes my sentence at face value and proceeds to
argue with it earnestly and at length. As the title "Parergon metricum" Der Ursprung des
griechischen Hexameters" suggests, Berg also announces his own etymology of the hexameter (cf.
also Tichy 1981). For a critique of Berg's own proposed solution, see Ritoók 1987:4-5.
112 Cf. Haslam 1974:15. Only at the beginning of cola can oo be substituted for _ in the Geryoneis of
Stesichorus, corresponding to the optional substitution of _oo for in the first foot of hexameter.
113 Ibid.
114 To repeat, the Aeolic fragments that I have examined are: Alcaeus F 44, 50, 115 a; 296 b, 340-349;
cf. Theocritus 28, 30 and Callimachus F 400.
115 Of course we would expect the pher@3da to accommodate a variety of other patterns as well. In the
actual attestations of ρher@3da in Alcaeus F 367 and 368, we find the following configurations of
word-breaking:
116 Such correspondences reinforce the argument of Hooker 1977:80-81 that these Alcaic and Hesiodic
compositions are independently drawing upon cognate traditions. If indeed the pattern of word-
breaking at position 4 in hexameter is conditioned by phraseology shaped /~Z*/ at the close of
glyconics, it seems pertinent that position 4 in hexameter is a common point of syntactical closure.
An extreme example is the end of the Iliad, where the scholia to the Townley manuscript report a
performance tradition where the last line, 804, can be stopped at position 4, to accommodate an
actual beginning of the Aithiopis narrative starting at this same position 4: according to this
tradition, the last word at line 804, 'horse-tamer', as applied to Hektor, is deleted, and a
clause introducing the grand opening theme of the Aithiopis is substituted: ' and
there came an Amazon...'.
182 GREGORY NAGY
117 E.g. in Sappho 44.3 and in Odyssey xv 526; in

Sappho 44.5 and in Homeric Hymn 33.1; in Sappho 44.31 and
in Iliad IX 161.
118Cf. Sappho F 44.7-8, w h e r e c o r r e s p o n d s to hexameter line-final oo_oo
, as in Odyssey iv 511. The expression is apparently secondary to
, as we can see from the expressions and
in Hesiod Theogony 107 and 964 respectively. In other words, it seems that
in certain environments, not the other way around.
119 Cf. Alcaeus F 347.8-9, where corresponds to hexameter line-final
in Hesiod Works and Days 587. We have already examined correspondences between this Alcais
composition and the Hesiodic passage from which this line, Works and Days 587, is taken.
120 As we have already seen, the wording that fills glyc@2da in Sappho F 44.10 can be described as
pher* followed by *glyc.
121 For stikhos in the sense of "verse," cf. Aristophanes Frogs 1239.
122 Cf. West 1982.43
123 Cf. Wilamowitz 1921.421.
124 Cf. Rossi 1978 on Aristophanes Wasps 1528-1537.
125 On the Alexandrian poetic practice of generalizing units taken from the synartetic framework of
strophic son g into the stichic units of verse in poetry, see Rossi 1971:86. For later developments in
the genres of the nomos and the dithurambos, where the principle of strophic responsion is
abandoned, see Gentili 1984.35. As it is pointed out in "Aristotle" Problems 19.15, the
abandonment of responsion entails greater freedom for experimentation in both the rhythms and the
melodies.
126 The meter in questions is glyc@2da, discussed above. All of Book II of the canonical Sapphic
corpus was composed in this meter: Hephaestion 7.7 p. 23.14-17 Consbruch.
127 E.g. glyc@da in Sappho F 94, as discussed above.
128 Cf. West 1973.188: "If there was epic or heroic balladry in (say) 1600 (B.C.), its characteristic verse
was most likely the glyconic [= glyc], whose cognates are used in Sanskrit and Slavic epic." On the
Middle High German evidence for melodic traditions of epic sung in strophic form, see Brunner
1970:160.
129 Cf. Rosenmeyer 1968:230. One apparent exception is at Pindar Pythian 1.92
on which see Gentili and Giannini 1977:17), where part of the sequence
looks like a dactylic hexameter; but it is not in fact a hexameter, since there is no correspondence
with the final word- boundaries of hexameter. Similarly with e.g. Pindar Nemean 9, at the beginning
of each strophe: there is a unit _oo_oo oo_oo matching the hexameter in its initial and final
word-boundaries but clashing with the primary caesura-patterns of hexameter.
130 Cf. Klusen 1969:72-83, cited in an interesting discussion by Rosier 1980:104nl76. As a
counterweight to the notion of gesunkenes Kulturgut in Klusen's work, see Bausinger 1980:41-55.
131 Cf. also Satyrus in Oxyrhynchus Papyri ix no. 1176 fr. 39 col. xix.
132 See e.g. Dale 1968:25-31, especially p. 29, on phraseological and prosodic irregularities in the sung
varieties of dactylic heameter - which are "irregularities" only from the standpoint of the regularities
in the non-sung variety. Cf. West 1982.98, 128 (especially nl25). See also Zumthor 1972:99 on a
comparable situation in medieval European traditions: "Dans les genres non chantées, le
perfectionnement des effets sonores semble une sorte de compensation de la perte de la mélodie." As
for Zumthor's list (ibid.) of formal features that serve to differentiate poetry from song, I should
emphasize that these formal features are all potentially present in song.
133 See Nagy 1979:614-619.
BIBLIOGRAPHY
Allen, W. S. 1973. Accent and Rhythm. Prosodic Features of Latin and Greek: A Study in Theory and
Reconstruction. Cambridge.
. 1987. Vox Graeca: The Pronunciation of Classical Greek. 3rd ed. Cambridge.
Ananthanarayana, H. S. 1973. Basic Metrical patterns in the Poems of the Rg-Veda. Acta Linguistica
Hafniensia 14.155-170.
Bake, A. 1957. The Music of India. New Oxford History of Music. I: Ancient and Oriental Music, edited
by E. Wellesz. 195-227. London.
Barker, A., ed. 1984. Greek Musical Writings. I. Cambridge.
Bausinger, H. 1980. Formen der "Volkspoesie." 2d ed. Berlin.
Ben-Arnos, D. 1976. Analytical Categories and Ethnic Genres. Folklore Genres, edited by D. Ben-Amos.
215-242. Austin.
Bowie, A. M. 1981. The Poetic Dialect of Sappho and Alcaeus. New York.
Cantilena, M. 1982. Ricerche sulla dizione epica. I. Per uno studio della formularità degli Inni Omerici.
Rome.
Cole, T. 1988. Epiploke: Rhythmical Continuity and Poetic Structure in Greek Lyrik. Cambridge, Mass.
Dale, A. M. 1963. Stichos and Stanza. Classical Quarterly 13.46-50. Reprinted in Dale 1969:173-179.
. 1968. The Lyric Metres of Greek Drama. 2d ed. Cambridge.
. 1969. Collected Papers. Cambridge.
Devine, A. M., and Stephens, L. D. 1975. Anceps. Greek Roman and Byzantine Studies 16.197-215.
. 1985. Stress in Greek? Transactions of the American Philological Association 115.125-152.
Figueira, T. J., and Nagy, G. Eds. 1985. Theognis of Megara: Poetry and the Polis. Baltimore.
Gentili, ., and Giannini, P. 1977. Preistoria e formazione dell' esametro. Quaderni Urbinati di Cultura
Classica 26.7-37.
Gentili, ., and Prato, C. Eds. 1979. Poetae Elegiaci. I. Leipzig.
Gentili, B. 1984. Poesia e pubblico nella grecia antica. Da Omero al V secolo. Rome and Bari.
Greenberg, N. A. 1985. A Statistical Comparison of the Hexameter Verse in Iliad I, Theognis, and Solon.
Quaderni Urbinati di Cultura Classica 20.63-75.
. 1985 b. Appendix: Language, Meter, and Sense in Theognis (The Role of agathos 'noble, good').
Figueira and Nagy 1985.245-260.
Haslam, M. 1974. Stesichorean Metre. Quaderni Urbinati di Cultura Classica 17.7-57.
. 1976. Review of Nagy 1974. Journal of Hellenic Studies 96.202-203.
. 1978. The Versification of the New Stesichorus (P. Lille 76 abc). Greek Roman and Byzantine
Studies 19.29-57.
Hooker, J. T. 1977. The Language and Text of the Lesbian Poets. (Innsbrucker Beiträge zur
Sprachwissenschaft, 26.) Innsbruck
184 GREGORY NAGY
Householder, F. W., and Nagy, G. 1972. Greek. Current Trends in Linguistics, edited by T. A. Sebeok
9.735-816. The Hague.
Itsumi, K. 1982. The 'choriambic dimeter' of Euripides. Classical Quarterly 32.59-74.
Jakobson, R. 1952. Studies in Comparative Slavic Metrics. Oxford Slavonic Papers 3.21-66. Reprinted
in Jakobson 1966.414-463. The Hague.
Klusen, E. 1969. Volkslied: Fund und Erfindung. Cologne.
Korzeniewski, D. 1968. Griechische Metrik, Darm Stadt.
Maas, P. 1962. Greek Metre. Translated by H. Lloyd-Jones. Oxford.
Meillet, A. 1920. Sur le rhythme quantitatif de la langue védique. Mémoires de la Sociéte de Linguistique
de Paris 21.193-207.
. 1923. Les origines indo-européennes des màetres grecs, paris.
Merkelbach, R. 1967. Eine Versreihe aus der parabase der 'Heroen' des Aristophanes? Zeitschrift für
Papyrologie und Epigraphik 1.161-162
Nagy, G. 1974. Comparative Studies in Greek and Indic Meter. Cambridge, Mass.
. 1979. On the Origins of the Greek Hexameter. Festschrift Oswald Szemerényi, edited by B.
Brogyanyi. 611-631. Amsterdam. Portions assimilated by Ch.2 above.
. 1981 Another Look at KLEOS APHTITHON. Würzburger Jahrbücher für Die Altertumswissenschaft
7.113-116.
. 1983. Review of Bowie 1981. Phoenix 37.273-275.
. 1986. Ancient Greek Praise and Epic Poetry. Oral Tradition in Literature: Interpretation in Context,
edited by J. Foley. 89-102. Columbia, Missouri. Reworked as Ch.7 above.
Page, D.L. Ed. 1962. Poetae Melici Graeci. Oxford.
. 1974. Supplementum Lyricis Graecis. oxford.
. 1975. Epigrammata Graeca. Oxford.
Palmer, L.R. 1980. The Greek Language. Atlantic Highlands. New Jersey.
Parry, M. 1928. L'épithète traditionnelle dans Homère: Essai sur un problème de style homérique. Paris.
Translated in Parry 1971:1-190.
. 1928b. Les formules et la métrique d'Homère. Paris. Translated in Parry 1971:191-234.
. 1971. The Making of Homeric Verse: The Collected Papers of Milman Parry, edited by A. Parry.
Oxford
Peabody, B.l 1975. The Winged Word. Albany.
PMG. See Page 1962.
Poultney, J. W. 1979. Eupolidean Verse. American Journal of Philology 100.133-144.
Risch, E. 1987. Die ältesten Zeugnisse für Zeitschrift für Vergleichende Sprach
forschung 100.3-11.
Ritoók, Z. 1987. Vermutungen zum Ursprung des griechischen Hexameters. Philologus 131.2-18.
Rosenmeyer, T. G. 1968. Elegiac and Elegos. California Studies in Classical Antiquity 1.217-231.
Rosier, W. 1980. Dichter und Gruppe: Eine Untersuchung zu den Bedingungen und zur historischen
Funktion früher Lyrik am Beispiel Alkaios. Munich.
Rossi, L. E. 1971.1 generi letterari e le loro leggi scritte e non scritte nelle letterature classiche. Bulletin
of the Institute of Classical Studies 18.69-94.
. 1978 (1980). Mimica e danza sulla scena comica greca. Rivista di Cultura Classica e Mediaevale 1-3.
1149-1170. (Miscellanea di studi in memoria di M. Barchiesi.)
Schmitt, R. 1967. Dichtung und Dichtersprache in indogermanischer Zeit. Wiesbaden.
SM. See Snell and Maehler.
Snell, ., and Maehler, H. Eds. 1971. Bacchylides. Leipzig.

. 1975. Pindarus: Fragmenta. Leipzig.
Solomon, J. 1985. In Defense of Hesiod's 'Schlechtestem Hexameter'. Hermes 113.21-30.
Tichy, E. 1981. Horn, und die Vorgeschichte des daktylischen Hexameters. Glotta 59.28-67.
. 1981b. Beobachtungen zur homerischen Synizese. Münchener Studien zur Sprachwisenschaft
40.187-222.
V.See Voigt 1971.
Vigorita, J. F. 1977. The Indo-European Origin of the Greek Hexamter and Distich. Zeitschrift für
Vergleichende Sprachforschung 91.288-299.
Vine, . 1977. On the Heptasyllabic Verses of the Rig-Veda. Zeitschrift für Vergleichende Sprach
forschung 91.246-255.
1978. On the Metrics and Origin of Rig-Vedic na 'like, as'. Indo-Iranian Journal 20.171-193.
Voigt, E.-M. Ed. 1971. Sappho et Alcaeus: Fragmenta. Amsterdam.
W.See West 1971/1972.
Watkins,  1963. Indo-European metrics and Archaic Greek Verse. Celtica 6.194-249.
. 1976. Observations on the 'Nestor's Cup' Inscription. Harvard Studies in Classical Philology 80.25-
40.
. 1976b. Sytax and Metrics in the Dipylon Vase Inscription. In Studies in Greek, Italic, and Indo-
European Lingusticis offered to Leonard R. Palmer, edited by A. Morpurgo Davies and W. Meid.
431-441. Innsbruck.
West, M. L., ed. 1971/1972. Iambi et Elegi Graeci. Oxford.
. 1973. Greek Poetry 2000-700 B.C. Classical Quarterly 23.179-192.
. 1973b. Indo-European Metre. Glotta 51.161-187.
. 1974. Review of Nagy 1974. Phoenix 28.457-459.
. 1982. Greek Metre. Oxford.
. 1982b. Three Topics in Greek Metre. Classical Quarterly 32.281-297.
Wilamowitz-Moellendorff, U. von. 1921. Griechische Verskunst. Berlin.
Zumthor, P. 1972. Essai de poétique médiévale. Paris.
. 1983. Introduction à la poésie orale. Paris.
. 1984. La Poésie de la Voix dans la civilisation médiévale. Paris.
GRÄZISTISCHE BEMERKUNGEN ZUR STRUKTUR
DES GILGAMESCH-EPOS
MICHAEL REICHEL
1. Untersuchungen zur Struktur des ältesten Großepos der Welt, des akkadischen
Gilgamesch-Epos, sind mit Schwierigkeiten verbunden, wie sie bei der Erforschung etwa
der frühgriechischen oder der mittelalterlichen Heldenepik in diesem Ausmaß nicht auf
treten. Der streckenweise sehr fragmentarische Erhaltungszustand der jüngsten und voll
ständigsten (jungbabylonischen) Textfassung aus der Bibliothek Assurbanipals (669-627)
in Ninive - vom Gesamtumfang des Epos fehlt etwa ein Drittel - und die Notwendigkeit,
Textlücken tentativ durch Bruchstücke von Textfassungen aus anderen Jahrhunderten und
sogar anderen Dialekten bzw. Sprachen (Altbabylonisch, Hethitisch) zu ergänzen,
erschweren Aussagen über den Aufbau des Ganzen. Hinzu kommt, daß über bestimmte
Faktoren, die die Komposition der einzelnen Textfassungen während der verschiedenen
Stufen der Tradierung des Stoffes mitbeeinflußt haben können, mangels direkter und
eindeutiger Zeugnisse Unklarheit herrscht: weder läßt sich mit einiger Sicherheit
feststellen, wo das Epos während seiner Wanderung durch die Kulturzentren des Alten
Orients jeweils seinen 'Sitz im Leben' hatte, noch läßt es sich abschätzen, ob und, falls ja,
in welchem Ausmaß die schriftliche Überlieferung durch eine parallele mündliche epische
Tradition beeinflußt wurde.
Angesichts dieser Ausgangslage ist es nicht verwunderlich, daß Forschungen zur
Struktur des Gilgamesch-Epos in der Altorientalistik gegenüber den Erfordernissen der
188 MICHAEL REICHEL
editorischen, grammatikalischen und lexikalischen Erschließung der Texte nur eine unter
geordnete Rolle gespielt haben. So konnte noch K. Hecker1 in seiner 1974 erschienenen
grundlegenden Arbeit von einem 'ersten Versuch' sprechen, die Gattungsmerkmale der
akkadischen Epik (in der das Gilgamesch-Epos natürlich eine herausragende Stellung
einnimmt) zu bestimmen und ihre formalen Gestaltungsmittel einschließlich ihrer 'kom
positionellen Techniken' zusammenzustellen. Nach Hecker hat vor allem J. H. Tigay2 in
seiner umfassenden Untersuchung über Entwicklung und Tradierung des Gilgamesch-
Epos wertvolle Beobachtungen zur Struktur des Werkes beigetragen.
Die auf orientalistischer Seite geübte Zurückhaltung - von Ausnahmen abgesehen -
sollte davor warnen, voreilig Aussagen komparatistischer Natur über den Aufbau des
Gilgamesch-Epos zu machen. Wenn im folgenden trotzdem der Versuch unternommen
wird, bestimmte Strukturmerkmale dieses Werkes ähnlichen Kompositionsprinzipien der
frühgriechischen Epik, wie sie von der Homerforschung seit langem festgestellt worden
sind, vergleichend gegenüberzustellen, so mag dies durch das Interesse gerechtfertigt
werden, das die Beziehungen der frühgriechischen zur altorientalischen Kultur nicht zu
letzt dank den Forschungen W. Burkerts3 und M. L. Wests4 in zunehmendem Maße ge
nießen.5 Von besonderer Relevanz für die Homerphilologie sind dabei "Entsprechungen
[...], die auch direkten literarischen Einfluß der östlichen Hochkultur auf die letzte Phase,
die Phase der Verschriftlichung des Homerischen Epos vermuten lassen."6
Parallelen zwischen dem Gilgamesch-Epos und den homerischen Epen haben die
Forschung seit der Jahrhundertwende beschäftigt.7 Das Interesse galt dabei vor allem
Entsprechungen motivischer Art. So regten zunächst vor allem analoge Elemente im Zug
des Gilgamesch auf der Suche nach dem ewigen Leben und in den Irrfahrten des
Odysseus zum Vergleich an.8 Später wurden auch Parallelen zur Ilias in stärkerem Maße
diskutiert.9 Vor allem sind es die vertretenen Entsprechungen zwischen Gilgamesch und
Achill sowie zwischen Gilgameschs Gefährten Enkidu und Achills Hetairos Patroklos,
die den Homeriker zur Auseinandersetzung auffordern, da die Parallelen hier, anders als
bei den Vergleichen mit der Odyssee, unmittelbar ins Zentrum der Handlung führen und
die Frage nach der 'Originalität' des Iliasdichters aufwerfen, ein Aspekt, dem hier nicht
weiter nachgegangen werden kann.
Neben inhaltlichen Gemeinsamkeiten wurden in letzter Zeit auch Entsprechungen in
der Erzähltechnik der orientalischen und der frühgriechischen Epik verstärkt untersucht,
GRÄZISTISCHE BEMERKUNGEN ZUR STRUKTUR DES GILGAMESCH-EPOS 189
wozu das besondere Interesse der Oral Poetry'-Theorie an diesem Punkt beigetragen ha
ben mag. Zu den gemeinsamen Merkmalen gehören die rein ornamentale Verwendung
mancher Beiwörter, die Formelverse ("whole-line formulae"), ζ. Β. bei der Einleitung
von Reden, die Wiederholung ganzer Versgruppen, Vergleiche, typische Szenen u.v.a.10
Die bislang konstatierten Entsprechungen bewegen sich vorwiegend im Bereich des
Stilistischen und der kleineren Kompositionseinheiten. Vergleiche bestimmter Struktur
merkmale auf einer mittleren und höheren Kompositionsebene fehlen weitgehend, was
mit den eingangs dargelegten Ursachen zusammenhängt. Daher soll im folgenden der
Frage nachgegangen werden, in welchem Ausmaß bestimmte aus Homer vertraute kom
pliziertere Erzähltechniken im akkadischen Gilgamesch-Epos anzutreffen sind. Zu den zu
untersuchenden Phänomenen, die in diesem Rahmen nur in aller Kürze angesprochen
werden können, gehören: 1. Vorbereitung und Vorausdeutung späterer Handlungsab
schnitte; 2. Rückerinnerungen an frühere Teile der Erzählung; 3. Integration von außer
halb der Handlung des Epos liegenden Ereignissen; 4. Retardation; und schließlich 5. die
kompositorische Verwendung von Iterata.
Zum methodischen Ansatz ist Folgendes vorauszuschicken: Die Beschränkung auf
das Gilgamesch-Epos als Vergleichsfolie erscheint dadurch gerechtfertigt, daß nur dieses
Werk mit einem geschätzten Gesamtumfang von 3600 Verszeilen11 eine großepische
Struktur aufweist, wie sie in anderen akkadischen (oder sumerischen) Epen und Mythen,
also etwa im Schöpfungsepos Enuma Elis oder im Erra-Epos schon wegen der viel gerin
geren Länge nicht vorliegen kann. Ein Vergleich mit den relativ umfangreichen ugariti-
schen Mythen und Epen scheint wegen des gegenüber dem Gilgamesch-Epos noch frag
mentarischeren Erhaltungszustandes problematisch.12 Die Beschreibung der Struktur
merkmale erfolgt hier in rein phänomenologischer und synchroner Weise, ausgehend von
dem Zwölftafelepos13 in der ninivitischen Redaktion des 7. Jhds. v. Chr.. Eine
historische Erklärung der Phänomene kann nur von orientalistischer Seite aus geleistet
werden.14 Der Verzicht auf eine Berücksichtigung der historischen Entwicklung des Epos
erscheint gerade unter dem Aspekt möglicher Beziehungen zum frühgriechischen Epos
durchaus legitim, da die ninivitische Fassung auf einen wahrscheinlich seit später mit
telbabylonischer Zeit relativ stabilen Text zurückgeht, der eine großepische Bearbeitung
der älteren Tradition darstellt und mit dem Namen Sin-leqe-unnïnis (12./13. Jhd.?) ver
bunden ist.15 Somit wäre innerhalb eines recht weit gesteckten Zeitraumes potentieller
190 MICHAEL REICHEL
Einflußnahme der orientalischen auf die griechische Epik, nämlich von spätmykenischer
Zeit bis ins 7. Jhd., eine weitgehende textliche Kontinuität des Gilgamesch-Epos anzu
nehmen. Es versteht sich ferner von selbst, daß für die Untersuchung struktureller
Phänomene nur Textpassagen aus ein und derselben Fassung, also hier der ninivitischen,
zu berücksichtigen sind unter Absehung von Ergänzungen mittels Bruchstücken der alt-
babylonischen oder hethitischen Fassung im Sinne einer Gilgamesch-'Harmonie'.16
2.1. Zu den erzählerischen Mitteln Homers, die am stärksten dazu beitragen, die ver
schiedenen Partien der Handlung zu einer organischen Einheit zu verschmelzen, gehört
die Vorbereitung und Vorausdeutung späterer Handlungsabschnitte, sei es durch den
Erzähler selbst oder durch Reden der Götter und Menschen.17 Ein Beispiel für den
erstgenannten Fall ist Ilias IL.688-694 18: der Dichter beschreibt, wie Achill, grollend
wegen der Kränkung durch Agamemnon, sich vom Kampf fernhält, und deutet am Ende
der Passage (694b) den erst viele tausend Verse später (im XIX./XX. Buch) dargestellten
Wiedereintritt Achills in die Schlacht an: "bald aber sollte er aufstehen". An Götterreden,
die auf die Ereignisse späterer Iliasbücher Bezug nehmen, sind die Prophezeiungen des
Zeus in VIII.470ff., XI.186ff. und XV.49ff. hervorzuheben. Die Worte eines Menschen,
nämlich Nestors, in XI.794ff. bereiten den Waffentausch zwischen Achill und Patroklos
im XVI. Buch der illas vor. Dies soll an Beispielen genügen.
Auch das Gilgamesch-Epos kennt "vorgreifende Motive", wie bereits von H. Gress-
mann19 beobachtet wurde, der auf die diesbezügliche Funktion bestimmter Reden und vor
allem der Träume aufmerksam gemacht hat. Als Beispiel sei hier eine Passage aus der I.
Tafel angeführt. Als sich ein Jäger auf den Rat seines Vaters hin (I.iii.l4ff.) an den König
Gilgamesch wendet, da ihm der in der Steppe lebende Enkidu seine Gruben und
Fangnetze unbrauchbar macht, antwortet Gilgamesch ihm mit Worten, in denen seine
Weisungen an den Jäger in eine allgemeine Voraussage übergehen (I.iii.41ff.):20
Geh, führ,  Jäger, mit dir
Eine Dirne nun, die Hure!
Wann denn das Wild herankommt zur Tränke,
Dann werfe sie ab ihr Kleid, sie enthüll' ihre Wollust!
Sieht er sie erst, so wird er ihr nahn:
Doch sein Wild wird ihm untreu, das aufwuchs mit ihm in der Steppe.
In vielfältiger Weise bereiten die Absichtserklärungen Gilgameschs, die Warnungen

Enkidus, die Ratschläge der Ältesten von Uruk und die Gebete von Gilgameschs Mutter
Ninßun, einer niederen Göttin, in Tafel  und III den Zug der beiden Freunde in den von
Chumbaba bewachten Zedernwald vor, wie er in Tafel IV und V dargestellt ist.
Zukunftsweisende Träume spielen im Gilgamesch-Epos eine große Rolle. Freilich
haben sie stark symbolhaften Charakter und bedürfen der Deutung, die für gewöhnlich im
Anschluß an die Erzählung des Traumes durch den Gesprächspartner gegeben wird.21 So
sind in der I. Tafel zwei Träume des Gilgamesch berichtet, die seine Begegnung und
spätere Freundschaft mit Enkidu, der im zweiten Traum durch eine Axt symbolisiert wird,
vorbereiten. Beide Träume werden durch seine Mutter Ninßun interpretiert. Nachdem
Gilgamesch ihr den zweiten Traum erzählt hat, spricht sie (I.vi.l8ff.):
Die Axt, die du sahst, ist ein Mann!
Du gewannst ihn lieb, wie über einem Weib wirst du über ihm raunen.
Und ich werde ihn mit dir gleichstellen.
Er wird zu dir kommen,
Der Gesell, der dem Freund aus der Not hilft!
Insbesondere die IV. Tafel, die den Zug der beiden Freunde zum Zedernwald enthält,
besteht zu einem beträchtlichen Teil aus der Darstellung einer ganzen Reihe von Träumen
Gilgameschs, die alle von seinem Gefährten Enkidu mit Bezug auf den bevorstehenden
Kampf gegen Chumbaba gedeutet werden.
Im Homerischen Epos spielen Träume eine wesentlich geringere Rolle. Als "reprä
sentativ für den homerischen Traumglauben"22 sei hier die Traumbotschaft an Aga
memnon in 77.II.5ff. erwähnt. Vergleicht man die Stelle mit den Traumpassagen im
Gilgamesch-Epos, so fallen eine Reihe von Unterschieden ins Auge: bei Homer erscheint
der Traum personifiziert; er wird von einer bestimmten Gottheit, Zeus, mit einer ihm
wörtlich aufgetragenen Botschaft abgesandt (vergleichbar mit Zeus' Aufträgen an die
Götterbotin Iris in XL 18Iff. u.ö.); sein Inhalt wird nicht allegorisch verhüllt, sondern ist
klar und bedarf keiner Deutung; und schließlich: er ist trügerisch und weckt in Aga
memnon falsche Hoffnungen. Diese Form des Trugtraumes stellt freilich eine 'Brechung'
dar, die das Motiv in seiner ungebrochenen Form, also als Wahrtraum, voraussetzt. Ein
solcher liegt vor in Od. 4.795ff., wo Athene zu der schlafenden Penelope ein Eidolon in
Gestalt ihrer Schwester Iphthime sendet, das ihr die sichere Heimkehr ihres Sohnes
Telemachos verheißt.
192 MICHAEL REICHEL
Das "einzige [...] Beispiel eines symbolisch zu deutenden Traums im homerischen

Epos"23 begegnet in Od. 19. 535ff. Penelope erzählt Odysseus von einem Traum, in dem
zwanzig Gänse in ihrem Haus Weizen fraßen, worauf sie von einem Adler getötet
wurden. Die Gänse stehen hier natürlich für die Freier, an denen der durch den Adler
symbolisierte Odysseus Rache nimmt. Das Schema des allegorischen Traumes mit
anschließender Deutung ist jedoch auch hier nicht mehr in Reinform bewahrt. Der mit
menschlicher Stimme begabte Adler erklärt den Traum - innerhalb des Traumes (!) -
selbst. Der zur Deutung aufgeforderte Odysseus hat dem nichts hinzuzufügen.
Vorausdeutungen durch Götterreden in Form von Prophezeiungen oder Beschlüssen
lassen sich aus den überlieferten Teilen der assyrischen Fassung des Gilgamesch-Epos
nur schwer belegen. Der Vollständigkeit halber sei jedoch auf zwei Passagen aus anderen
Redaktionen verwiesen. So finden sich in altbabylonischen Fragmenten zwei im Wortlaut
identische Prophezeiungen des Sonnengottes Schamasch, der Gilgamesch freundlich
gesonnen ist, und der göttlichen Schenkin Siduri, Gilgamesch werde das (ewige) Leben,
das er suche, nicht finden. In der hethitischen Fassung ist ein Götterrat dargestellt, in dem
Enkidus Tod beschlossen wird. Der Götterrat bildet dabei den Inhalt eines Traumes des
Enkidu, den er Gilgamesch erzählt, wobei der Traum hier nicht in allegorischer, sondern
in unverhüllter Form mitgeteilt wird.24
Sowohl für die Träume als auch für die Reden im Gilgamesch-Epos gilt, daß sie in
der Regel nur ein kurzes Stück weit in die Zukunft weisen; sie bereiten meist nur die
nächstfolgende Geschehensphase vor. Hier ist ein deutlicher Unterschied zu Homer faß
bar, der Vorausdeutungen auch auf Ereignisse gibt, die erst nach einer Distanz, die ein
Mehrfaches vom Gesamtumfang des Gilgamesch-Epos betragen kann, eintreten (vgl.
oben zu/7. II.694b).
Nur an einer Stelle im Gilgamesch-Epos finden sich Vorausdeutungen auf Fernerlie
gendes, die außerdem nicht mit Hilfe von Reden oder Träumen, sondern durch den epi
schen Erzähler selbst gegeben werden: in der Einleitung des Epos werden die Taten des
Gilgamesch verherrlicht, wobei vor allem die im letzten Teil des Epos (inTafel IX bis XI)
dargestellten Wanderungen des Helden angesprochen werden (I.i.35ff.):
Der Wildstier Gilgamesch, der Vollkommene, Ehrfurchtgebietende,
Der da fand die Eingänge in das Gebirge,25
Der dürstete nach den Zisternen am Rand des Steppenlandes.
Der die See überfuhr, das weite, zum Sonnenaufgang hin liegende Meer.
Der die Weltränder ins Auge faßte, überall das Leben suchend,
Der in seiner Stärke gelangte bis hin zum fernen Utnapischtim.
Der eintrat in Städte, an ihre Stätten, die die Sintflut vernichtet hatte.
Daß bereits im Einleitungsteil des Epos konkrete Ereignisse aus seinen Schlußpartien
zur Sprache kommen, unterscheidet das akkadische Epos wiederum vom frühgriechi
schen. So wird ja im Proömium der Ilias (I.1-7) nur ganz allgemein von den verderbli
chen Folgen des Grolls des Achill für das Heer der Achaier gesprochen. Das Odyssee-
Proömium (1.1-10) geht auch auf speziellere Ereignisse wie den Untergang der Gefährten
des Odysseus nach dem Verzehr der Rinder des Helios ein, erwähnt aber keine Ge
schehnisse aus der zweiten Hälfte des Epos. Vielmehr ist die homerische Epik durch ein
"gradual development in the reader's knowledge"26 gekennzeichnet, dergestalt daß deut
lich später liegende Ereignisse zunächst nur stückweise, im weiteren Verlauf der Hand
lung dann mit zunehmender Ausführlichkeit vorbereitet werden. Die Ursache für diese
Besonderheit des Gilgamesch-Epos liegt im hymnischen Charakter des Schlußteils seiner
Einleitung (I.i.27-ii.6).27 So gehört es ja zu den Gattungsmerkmalen des Götterhymnus
(auch des griechischen), die Eigenschaften und herausragenden Taten der gepriesenen
Gottheit aufzuzählen.28
2.2. Das Gegenstück zu den Vorbereitungen und Vorausdeutungen bilden die Rück
erinnerungen an frühere Teile der Erzählung, teils in Form ausführlicher Rekapitulatio
nen, teils als kurze, mehr oder weniger deutliche Anspielungen auf als bekannt Voraus
gesetztes. So erzählt z. B. Thetis, als sie in der Ilias den göttlichen Schmied Hephaistos
um eine neue Rüstung für ihren Sohn Achill bittet, praktisch den gesamten bisherigen
Handlungsverlauf, um ihr Anliegen zu begründen (XVIII.429ff., bes. 444ff.). Eine
kürzere Rückdeutung tritt z. B. in XXII.99-103 auf, wo Hektor vor dem entscheidenden
Kampf gegen Achill an die Warnung seines Mitstreiters Polydamas vor einer Fortsetzung
des Kampfes gegen die Griechen auf freiem Felde zurückdenken muß, die er tags zuvor
in den Wind geschlagen hatte (XVIII.249-313). Die Rückwendung dient an dieser Stelle
dazu, die frühere tragische Verblendung Hektors in Erwartung seines Todes noch einmal
in Erinnerung zu rufen.
Auch im Gilgamesch-Epos gibt es in beachtlicherem Umfang derartige Rückerinne
rungen. Als Gilgamesch in Tafel X den Flutheros Utnapischtim aufsucht, um mit seiner
Hilfe das ewige Leben zu erlangen, wird er von ihm gefragt, warum er 'Harm in seinem
194 MICHAEL REICHEL
Gemüt' habe. In seiner Antwort in X.iv.50-v.22 spricht Gilgamesch von seiner Trauer
um Enkidu und rekapituliert, indem er von ihren gemeinsamen Taten erzählt, große Teile
der bisherigen Handlung des Epos: so erinnert die formelhafte Beschreibung seines Ge
fährten als "Enkidu, mein Freund, der flüchtige Maulesel, der Wildesel des Gebirges, der
Panther der Steppe" (X.v.7) an die in Tafel I beschriebene Herkunft Enkidus aus der
Wildnis (vgl. I.iii.29: "ein Mann, der vom Steppenland gekommen") und an die in Tafel
II beginnende Freundschaft mit ihm. In X.v.10 erzählt Gilgamesch von der gemeinsamen
Tötung Chumbabas im Zedernwald (dargestellt in Tafel V), in X.v.9 - die Reihenfolge im
Bericht weicht hier leicht von der in der epischen Handlung ab - von der Erlegung des
Himmelstieres (dargestellt in Tafel VI). Die Worte "es hat ihn [sc. Enkidu] ereilt die
Bestimmung des Menschen" in X.v.13 nehmen auf den Tod Enkidus Bezug, der in dem
verlorenen Schlußteil der VII. Tafel dargestellt gewesen sein muß. Die Rekapitulation
wird umso ausführlicher und detaillierter, je geringer der zeitliche Abstand des referierten
Ereignisses und je größer die emotionale Beteiligung an ihm ist. So schildert Gilgamesch
in eindringlichen Worten seine Trauer um Enkidu (X.v.l4ff.):
Um ihn weint' ich sechs Tage und sieben Nächte,
Ich gab nicht zu, daß man ihn begrübe,
Bis daß der Wurm sein Gesicht befiel.
Mir graute vor meines Freundes Aussehn,
Ich erschrak vor dem Tod, daß ich lief in die Steppe!
Das, woran Gilgamesch hier erinnert, ist wiederum nicht erhalten, hat aber vermutlich
in einer Lacuna der Tafel VIII gestanden (vgl. den Anfang von Tafel IX: "Gilgamesch -
um Enkidu, seinen Freund, weint er bitterlich, läuft herum in der Steppe."). Ebenso wie
die rekapitulierenden Worte der Thetis in IL XVIII.444ff. (s.o.) den Zweck haben, ihr
Ansinnen an den Gott Hephaistos zu motivieren (vgl. den Schluß ihrer Rede
XVIII.457ff.), so stellen auch Gilgameschs Schlußworte den Übergang zur Haupterzäh
lung her, indem sie den Grund seines Gangs zu dem von den Göttern privilegierten Heros
Utnapischtim angeben: es ist die Furcht, das gleiche Schicksal wie Enkidu zu erleiden
(X.v.21f.):
Enkidu, mein Freund, den ich liebte, ist zu Erde geworden!
Werd ich nicht auch wie er mich betten
Und nicht aufstehn in der Dauer der Ewigkeit?
Ganz ähnliche, ja im Wortlaut vielfach identische Rekapitulationen von Gilgameschs

Erlebnissen mit Enkidu finden sich in Gilgameschs Totenklage um seinen Gefährten
(VIII.ii.8ff.), in seinem Gespräch mit der Schenkin Siduri (X.i.35ff., hier ohne Nennung
Enkidus) und in seinen Worten zum Fährmann Urschanabi (X.iii.l5ff.).
Eine andere an Rückerinnerungen reiche Passage begegnet in Tafel VII, wobei zu
gleich ein anderer Typus der Rückdeutung auftritt: die Verfluchung. Als Enkidu, den Tod
vor Augen, auf dem Krankenlager liegt, verflucht er die Dirne (VII.iii.6ff.), durch deren
Beilager er in Tafel I vom animalischen zum menschlichen Wesen geworden war, ja er
verflucht sogar den Jäger (VII.ii.37ff.), der ihn damals in der Steppe erblickt hatte
(I.ii.42ff.) und anschließend zu Gilgamesch gegangen war, der ihm darauf befohlen
hatte, die Dirne zu Enkidu zu führen. Bemerkenswert ist es, wie Enkidu nicht allein an die
jüngstvergangenen Ereignisse wie die Tötung Chumbabas und des Himmelsstieres als
den unmittelbaren Anlaß seiner Not denkt29, sondern weit zurückliegende Ereignisse aus
der Zeit noch vor seiner Begegnung mit Gilgamesch anspricht, Ereignisse, die in einer
langen Kausalkette, deren Glieder hier nicht thematisiert, sondern implizit vorausgesetzt
werden, zu seiner jetzigen Lage geführt haben. Eine detaillierte Rekapitulation von Ereig
nissen der Tafeln I und II wird auch durch die Antwort des von Enkidu angerufenen
Sonnengottes Schamasch in VII.iii.35ff. gegeben:
Warum, Enkidu, verfluchst du die Dime, die Hure?
Die dich götterwürdige Speisen essen ließ,
Mit feinstem Bier, wie es Königen ansteht, dich tränkte?
Mit vornehmer Kleidung dich kleidete,
Und Gilgamesch dir, den herrlichen, als Gesellen zu eigen gab?
2.3. Während kürzere epische Gattungen wie das Heldenlied sich in aller Regel auf
ein Thema beschränken, bietet das Großepos Gelegenheit, mehrere miteinander
verbundene Themen darzustellen. Ein weiteres Merkmal, das das Großepos vom
Heldenlied unterscheidet, ist die Möglichkeit, auch Geschehnisse, die außerhalb der
eigentlichen Erzählhandlung liegen, innerhalb des Epos mehr oder weniger ausführlich
zur Sprache zu bringen. Das Phänomen sei hier als Integration von außerhalb der
Handlung des Epos liegenden Ereignissen bezeichnet. Beim Gilgameschepos wird man
natürlich in erster Linie an den ausführlichen Sintflutbericht des Utnapischtim in Tafel XI
denken. Die Darstellung der Flut kursierte im alten Mesopotamien und benachbarten
196 MICHAEL REICHEL
Gebieten als selbständige Erzählung in verschiedenen Versionen, deren bekannteste

(neben der biblischen Fassung) das babylonische Atrahasῖs-Epos ist.30 Die Darstellung
der Flut wurde, zumindest in dieser ausführlichen Form, möglicherweise erst spät in das
Gilgamesch-Epos integriert.31
Als Beispiel einer umfangreichen, in der 1. Ps. erzählten Erinnerung, die die Haupt
handlung unterbricht, läßt sich Utnapischtims Bericht auf den ersten Blick mit den Apo-
logen (Bücher 9-12) der Odyssee vergleichen. Die Parallele reicht freilich nicht sehr weit.
Odysseus' Bericht bildet einen unverzichtbaren Bestandteil der Haupthandlung, der
lediglich 'phasenverschoben' mitgeteilt wird, und Odysseus selbst ist die zentrale Gestalt
des Epos. Das trifft auf Utnapischtim im Gilgamesch-Epos nicht zu. Auch liefert seine
Erzählung nicht die Vorgeschichte der Haupthandlung, sondern ist durch ihre paraineti-
sche Funktion mit ihr verknüpft. Als typologische Parallele im homerischen Epos bieten
sich somit weniger die Apologe der Odyssee als vielmehr das in der Rede des Phoinix
enthaltene Meleagerparadigma in Buch DC (Vv. 527-599)32 und vor allem die Jugend
erzählung Nestors in Buch XI (Vv. 670-762) der Ilias an. Wie Utnapischtims Worte
darauf abzielen, eine Änderung der inneren Einstellung Gilgameschs zu bewirken, so
sollen auch Phoinix' und Nestors Worte Achill zu einer Änderung seiner Haltung be
wegen (im XL Buch auf dem Wege über Achills Hetairos Patroklos). Die Sprecher be
dienen sich sowohl der direkten Aufforderung (GE X.v.37ff. und Il. IX496ff., 513ff.,
600ff. sowie XI.656ff.,790ff.) als auch der indirekten Mahnung mittels der in die
Vergangenheit ausgreifenden Erzählung. Utnapischtim will Gilgamesch durch die Schil
derung seiner Erlebnisse während der Flut deutlich machen, "that his own attainment of
immortality was due to a set of unique, unrepeatable circumstances"33. Phoinix will durch
die Erzählung des Sinneswandels Meleagers, ebenso Nestor durch die Darstellung seiner
Taten im Kampf der Pylier gegen die Eleier deutlich machen, daß Achill seinerseits die
Pflicht hat, für die Achaier in der Schlacht einzustehen. Während Phoinix' und Nestors
Erzählungen unmittelbar paradigmatische Funktion haben (Achill soll genauso handeln,
wie Meleager - letzten Endes - und Nestor es taten), muß Utnapischtims Bericht ex
contrario auf die Haupthandlung bezogen werden (Gilgamesch soll einsehen, daß er nicht
erringen kann, was Utnapischtim zuteil geworden ist).
Neben derartigen ausführlichen Digressionen erscheinen bei Homer en passant zahl
lose, meist kürzere Anspielungen auf Ereignisse, die vor Beginn der Handlung stattge-
funden haben. Überwiegend handelt es sich dabei um Motive aus dem troischen Sagen
kreis, wie sie in den Kyklischen Epen dargestellt waren34, z. B. den Raub der Helena
(IL IIL442ff., xm.626f. u.ō.), die Sammlung der Achaier in Aulis (IL II.303ff.), die
Ermordung Agamemnons (Od. 1.35f. u.ō.) u.v.a.. Auch im Gilgamesch-Epos wird in
ähnlicher Weise, wenngleich selten, auf Geschehnisse angespielt, die bereits vor dem
Einsatz der Handlung abgeschlossen sind. Das gilt vor allem für den Bau der Stadtmauern
Uruks und des Tempels für Anu und Isentar durch Gilgamesch, deren Existenz bereits in
der ersten Tafel wie auch später vorausgesetzt ist (abgesehen von den Versen im Prolog
I.i.9ff. vgl. I.iv.37, 44; VI. 157; XI. 303ff.).35
2.4. Eines der besten Mittel zur Erzeugung von Spannung gerade bei einem groß
epischen Gedicht beträchtlichen Umfangs ist die Retardation.36 Besonders gut läßt sie
sich am dritten in der ilias dargestellten Schlachttag (Bücher XI-XVIII) beobachten, wo
die Dominanz des einen Themas, des Kampfes, ermüdend wirken müßte, würde der
Dichter die Handlung nicht in eine Reihe vorantreibender und retardierender Momente
kunstvoll zergliedern. Der von Zeus verheißene Triumpf Hektors (XL.187-194 ~ 202-
209), der den Zielpunkt der Kampfhandlung bildet,wird immer wieder aufgeschoben: zu
nächst durch die Aristie Agamemnons, später durch die Unterstützung einzelner Götter
für die achaische Seite und durch die Aussendung des Patroklos. Das Prinzip der Re
tardation durchwaltet die ganze lliashandlung, sowohl innerhalb der größeren als auch der
kleineren Darstellungseinheiten.
Schwerer als für das homerische Epos läßt es sich für das Gilgamesch-Epos beurtei
len, ob bei der Unterbrechung der Handlungsbewegung auf ein bestimmtes Telos hin
durch aufschiebende Momente ein beabsichtigter Effekt vorliegt oder andere Gründe aus
schlaggebend sind. So hat z. B. die Einschaltung der Sintfluterzählung in der XI. Tafel in
dem Moment, da Gilgamesch hofft, den Schlüssel zum ewigen Leben zu finden, rein de
skriptiv betrachtet, zweifellos retardierende Wirkung. Zugleich dürfte sie aber auch um
ihrer selbst willen dargestellt werden, wie ihr kunstvoller Aufbau zeigt, und nicht zuletzt
hat sie in Hinblick auf Gilgamesch eine bestimmte parainetische Funktion, wie oben aus
geführt wurde. Ähnliches gilt für andere Digressionen und Episoden, z. B. wenn Gilga-
meschs ausführliche Antwort auf das Werben der Göttin Ischtar (VI.24-79) seine
schließliche Ablehnung aufschiebt, aber zugleich auch durch die Darstellung des un-
198 MICHAEL REICHEL
glücklichen Schicksals ihrer früheren Geliebten begründet. Auch die Zerdehnung von
Situationen durch Wiederholung von Passagen dürfte weniger um des retardierenden
Effektes willen als wegen des in sumeyrischer wie in akkadischer Erzählung so beliebten
Stilmittels der Wiederholung als solcher, vor allem in steigernder oder variiernder Form,
erfolgen. Trotz aller Vorbehalte drängt sich jedoch der Eindruck auf, daß zumindest im
Schlußteil des Epos, bei Gilgameschs Suche nach dem ewigen Leben in Tafel IX bis XI,
das letztliche Scheitern Gilgameschs möglichst lange hinausgeschoben und seine trügeri
sche Hoffnung möglichst lange aufrecht erhalten wird, damit seine Verblendung in ihrem
vollen Ausmaß sichtbar werden kann. Gressmann 37 hat den Weg des Helden in treffen
den Worten geschildert:
Zunächst geht die Wanderung zu Ut-napištim verhältnismäßig flott von statten. Gewiß ist das Ende
der Welt auch für den antiken Menschen sehr weit entfernt, uns aber scheint es, als wäre Gilgameš gar
zu schnell dort angelangt. Von da an verlangsamt sich sein Weg um so mehr, je näher er dem Ziele
kommt. Station folgt auf Station und immer ferner rückt die Mündung der Ströme: Skorpionriesen,
Wanderung durch den Schacht des Gebirges, Sabîtu. Noch kürzer werden die Abstände, die Hindernisse
türmen sich gradezu, seitdem Gilgameš den Fährmann erreicht hat: es sei erinnert an die Zertrümme
rung der Steinkisten, an den Ersatz durch Stangen, an die pfeilschnelle Fahrt über das offene Meer, an
den Bau der Hängebrücke über die Wasser des Todes und endlich die glückliche Landung. Der Aufent
halt bei Ut-napištim ist der Höhepunkt des Ganzen und wird darum besonders ausführlich geschildert.
Die retardierenden Momente sind vor allem hier so zahlreich und so wirksam, daß die Aufmerksamkeit
der Hörer bis zuletzt aufs äußerste gespannt wird.
2.5. Ein methodisch voraussetzungsreiches Phänomen ist die kompositorische Ver

wendung von Iterata. Will man sie bei Homer nachweisen, so ist für jeden einzelnen Fall
durch Untersuchung des Sprachgebrauchs und des Kontextes der Stellen zu klären, ob
die Iterata in einer bestimmten poetischen Absicht gebraucht werden oder ob sie an
verschiedenen Stellen unabhängig voneinander auf traditionelles Formelmaterial zurück
gehen. So ist z. B. das zweimalige Auftreten des Verses "viele kraftvolle Seelen (bzw.
Häupter) warf er dem Hades vor" in IL 1.3 und XI.55 sicher nicht als formelhaft im
strikten Sinne zu erklären 38 : an beiden Stellen ist vom Willen des Zeus die Rede, dessen
Erfüllung in 1.5 angekündigt wird und vom XL Buch an zur Entfaltung kommt; es liegt
also ein "absichtlicher Rückgriff"39 vor, der diesen Handlungszusammenhang augenfällig
machen soll. Nicht nur einzelne Verse, sondern auch ganze Versgruppen werden bei
Homer nach längerem Abstand wörtlich wiederholt, wobei in vielen Fällen der spezifische
inhaltliche Bezug zur Handlung des Epos dafür spricht, daß es sich nicht um erebte
Versatzstücke handelt, sondern Passagen desselben Werkes bewußt wiederaufgegriffen

werden.40
Wiederholungen von Versen und Versgruppen sind in der akkadischen Literatur ge
nerell so häufig, daß - selbst wenn man die Frage nach einer vorgängigen oder zur Zeit
der Abfassung des Zwölftafelepos noch lebendigen mündlichen Tradition ausklammert -
die Unterscheidung zwischen rein stilistischem Gebrauch der Wiederholung einerseits
und ihrer planvollen, kompositorischen Anwendung andererseits oft schwer zu treffen ist.
Am häufigsten treten Iterata hier freilich nach kurzem Abstand auf, und zwar in viel rei
cherem Maße als bei Homer, wo dieses Phänomen auf die vergleichsweise seltenen Fälle
von 'Erteilung und Ausführung bzw. Übermittlung eines Auftrags' beschränkt ist. Da
neben gibt es aber in der akkadischen wie in der griechischen Epik Wiederholungen von
Versen und Versgruppen über große Distanz mit dem Zweck, kompositorische Zusam
menhänge sichtbar zu machen. Das beste Beispiel im Gilgamesch-Epos ist zweifellos das
Wiederaufgreifen der Verse Li. 16-21 des Prologs am eigentlichen Ende des Epos in
XI.303-307, wo Gilgamesch den Fährmann Urschanabi voller Stolz auffordert, die
Stadtmauer Uruks zu prüfen. Eine der verschiedenen Interpretationen des Epos, die
durchaus Plausibilität beanspruchen kann, sieht in diesen Worten den Reflex einer inneren
Entwicklung des Protagonisten41: Gilgamesch hat erkannt, daß er mit seinem Streben
nach Unsterblichkeit das dem Menschen gesetzte Maß überschritten hat, und findet den
Sinn seines Daseins nunmehr in der Erfüllung des 'Menschenmöglichen', vor allem,
indem er seinen Herrscherpflichten nachkommt durch die Fürsorge für seine Untertanen,
von der die von ihm erbaute schützende Mauer Uruks Zeugnis ablegt. Die besondere Po
sitionierung der Versgruppe an den herausragenden Stellen am Anfang und am Ende des
Epos und ihr Bezug auf seine zentrale Aussage machen es wahrscheinlich, daß hier eine
beabsichtigte Beziehung vorliegt.
Als weiteres Beispiel kompositorischer Wiederholung sei das in Tafel IV dreimal be
zeugte (i.44f., ii.46f., iv.l9f.) und zwei weitere Male in Textlücken zu postulierende
Verspaar zitiert:42
Nach zwanzig Doppelstunden nahmen sie einen Imbiß ein,
Nach dreißig Doppelstunden schickten sie sich zur Abendrast.
Das fünfmalige Auftreten der beiden Verse in Tafel IV dient dort zur Binnenglie
derung eines kleineren Erzählabschnitts innerhalb des Zugs Gilgameschs und Enkidus
200 MICHAEL REICHEL
zum Zedernwald. Die gleichen Verse erscheinen jedoch noch ein weiteres Mal, und zwar
kurz vor Ende des Epos (bzw. der Tafel XI) in XI.283f. und 300 b -301 a bei der Schil
derung der Rückkehr Gilgameschs, der sich vorübergehend in den Besitz des unsterblich
machenden Lebenskrautes gebracht hat; diesmal ist die Person in seiner Begleitung nicht
Enkidu, sondern Urschanabi. Für die Annahme, daß das Iterat hier eine kompositorische
Funktion hat, seien die Worte Heckers 43 zitiert. Nach seiner Auffassung schließt die
Wiederholung des Verspaars in Tafel XI
den Bogen zu dem so hoffnungsvoll begonnenen ersten Abenteuer des Gilgameš eben dem Zug gegen
Huwawa in Tf. IV, dem jetzt die resignierende Heimkehr folgt. Der direkte, durch den Wortlaut der
Floskel gegebene Anklang wird im übrigen noch durch sachliche Parallelen gestützt, denn abgesehen
von der Dauer der Reisen entsprechen auch die Lokalitäten einander. Der Verlust des Krautes 'Der
Greis wird ein junger Mensch' spielt sich an einem Brunnen ab, 'dessen Wasser kühle war', einem
Brunnen also, wie ihn Gilgameš im stereotypen Teil der Traumerzählung in Tf. IV zu graben hat.
Steht dort die formale Geschlossenheit eines motivischen Komplexes im Vordergrund, so bewirkt die
repetitorische Verwendung der Floskel hier in Tf. XI in erster Linie eine Verdeutlichung der inneren
Zusammenhänge des epischen Geschehens: in der formalen Verknüpfung erweist sich der Verlust des
Verjüngungskrautes als ein keineswegs zufälliger Ausgang des Abenteuers des Gilgameš, sondern als
die endlich eingetretene Realisation einer all seinen Plänen und Taten von Anfang an mitgegebenen
tragischen Bestimmung. [Hervorhebung von mir, MR]
Auch in der sprachlichen Technik der Anwendung iterierter Versgruppen sind Paral
lelen zwischen der homerischen und der akkadischen Epik feststellbar. Für die Iterata der
frühgriechischen Epik hat Strasser detailliert aufgezeigt, wie "eine Formulierung den
Erfordernissen einer neuen Erzählsituation durch Änderungen bei den Verbformen sowie
den Personalpronomina angepaßt werden" kann. In ganz ähnlicher Weise stellt Hecker bei
Wiederholungen von Versgruppen in akkadischen Epen "eine durchgehende Anpassung
der pronominalen und modal-temporalen Gegebenheiten an die neue Situation" fest.44
3. Die vorausgehende nur wenige Punkte berührende Untersuchung einzelner Struk

turmerkmale bestätigt die Auffassung des Gilgamesch-Epos als eines "consistent and
well-ordered whole, with unified structure and themes" 45 . Wie die vergleichende Gegen
überstellung gezeigt hat, sind auch im bedeutendsten akkadischen Epos bereits bestimmte
erzähltechnische Mittel im Ansatz verwendet, die im homerischen Epos so wesentlich zur
inneren Geschlossenheit der Handlung beitragen. Mit dieser Feststellung soll keineswegs
der immer noch sehr große Qualitätsunterschied in der Struktur des akkadischen und des
homerischen Epos überspielt werden. Trotz des viel größeren Umfangs - die Ilias ist mit
rund 15000 Vv. (ohne Dolonie) etwa viermal, die Odyssee mit rund 12000 Vv. etwa
dreimal so lang wie das Gilgamesch-Epos - ist die Verflechtung einzelner Handlungs
stränge und die Verklammerung weit voneinander entfernter Handlungspartien hier viel
kunstvoller durchgeführt. Auch operieren Ilias- und Odysseedichter mit einem viel
größeren Personeninventar (allein in der Ilias erscheinen etwa 350 Eigennamen). Die
komplizierte Technik der Schauplatzwechsel und der Erzählung paralleler Handlungs
stränge bei Homer ist in dieser Form ohne Beispiel.46 Dasselbe gilt für die ständigen Be
rührungen zwischen menschlicher und göttlicher Hanlungsebene, wobei der umfang
reiche Götterapparat noch in zwei Parteien gespalten ist.47 Einer der wesentlichsten Un
terschiede ist auch, daß zumindest in der Ilias - in der Odyssee ist die Situation wegen der
Apologe etwas anders - nur ein sehr kurzer Zeitraum zur Darstellung kommt; im ganzen
sind es 51 Tage, von denen nur wenige (genauer gesagt: 15 Tage und fünf Nächte) aus
führlich geschildert werden.48 Das Gilgamesch-Epos stellt dagegen zeitlich weit ausein
anderliegende wichtige Stationen aus dem Leben des Helden dar. Somit nähert es sich der
Gattung des biographischen Heldenepos, die in der Ilias nicht und auch in der Odyssee
nur mit Abstrichen vorliegt. Parallelen dazu im frühgriechischen Raum lassen sich in
verlorengegangenen, z. B. um die Gestalt des Herakles oder Ödipus zentrierten Epen
vermuten. In der außergriechischen Epik begegnet diese Gattung etwa im angel
sächsischen Beowulf, im byzantinischen Volksepos Digenis Akritas oder im altspani
schen Cantar de Mio Cid.
Welche Ergebnisse lassen sich nun aus dieser vergleichenden Gegenüberstellung für
die Homerphilologie gewinnen? Hier ist zu unterscheiden zwischen 1. Folgerungen für
die Frage genetischer Beziehungen zwischen altorientalischer und frühgriechischer Epik
und 2. Erkenntnissen typologischer Natur über die Bedingungen der Entwicklung groß
epischer Dichtung. Zunächst zum ersten Punkt
Burkert weist zurecht darauf hin, daß sich im Bereich der stilistischen Elemente kein
Beweis einer direkten Abhängigkeit führen lasse49, und fordert dafür den Nachweis von
Entsprechungen "komplexer Strukturen".50 Daß die Übernahme solcher komplexen
Strukturen über die sprachliche Grenze hinweg möglich war, zeigen z. B. die Nachdich
tung des churritisch-hethitischen Sukzessionsmythos bei Hesiod oder, um beim bislang
herangezogenen Vergleichsmaterial zu bleiben, die wohl kaum zufällige Parallele
zwischen GE VI. 1-91 (bes. 80ff.), wo sich die Liebesgöttin Isentar bei ihrem Vater Anu
202 MICHAEL REICHEL
und ihrer Mutter Antum über die ihr von einem Sterblichen, Gilgamesch, widerfahrene
Kränkung beschwert, und II.V.330-431 (bes. 370ff.), wo sich Aphrodite bei ihrer
Mutter Dione beklagt, daß ein Sterblicher, Diomedes, es gewagt hat, sie zu verwunden
(später wird sie auch von ihrem Vater Zeus getröstet).51 In diesen Fällen sind die
Strukturen mit einem bestimmten Inhalt oder Motiv ausgefüllt, während die hier
beschriebenen Gemeinsamkeiten allgemeine Strukturprinzipien (auf einer makrostruk
turellen Ebene) betreffen, die inhaltlich zunächst unbesetzt sind und ganz unterschiedlich
angewandt werden können. Jedoch sind die Parallelen auch in diesem Bereich so
bemerkenswert, daß sie in der weiteren Diskussion um die Beziehungen der griechischen
zur vorderorientalischen Epik zumindest im Zusammenhang mit anderen (motivischen,
religionsgeschichtlichen, sprachlichen etc.) Argumenten der Berücksichtigung wert
erscheinen, mögen sie für sich allein genommen auch nicht tragfähig genug sein.
Von noch größerer Bedeutung für den Homerforscher dürfte demgegenüber die Er
wägung der Bedingungen sein, unter denen sich die hier angesprochenen strukturellen
Techniken im Gilgamesch-Epos entwickelt haben. Die mit Homer annähernd zeitgleiche
ninivitische Fassung setzt mindestens ein halbes Jahrtausend schriftlicher Überlieferung
in großepischer Form voraus, falls der Gilgamesch-Stoff schon in altbabylonischer Zeit in
einem umfangreichen Dichtungswerk behandelt wurde, sogar ein ganzes Jahrtausend.
Zieht man auch die sumerischen Einzellieder heran, so ist die schriftliche Tradierung von
Gilgameschsagen schon für die Zeit um 2000 v. Chr. nachzuweisen, und es ist nicht aus
zuschließen, daß die Erwähnung des Himmelstieres auf einem in Ebla gefundenen Tafel
fragment aus der Zeit um 2500 v. Chr. ebenfalls bereits in den Bereich der Gilgamesch-
Mythen gehört.52 Die Schrift ist in der langen Überlieferungsgeschichte des Gilgamesch-
stoffes nicht allein Medium der Tradierung, sondern zugleich Mittel zur Redigierung des
Stoffes, zur Neuformulierung und Neustrukturierung von Passagen, zur Einfügung neuer
Teile, neuer Gedanken usf., wie Tigay in seiner Studie gezeigt hat. Die Rolle, die die
Schriftlichkeit beim Zustandekommen dieses "integrated and well-structured epic"53
gespielt hat, sollte denen zu denken geben, die in der um ein Vielfaches komplexeren
Struktur der homerischen Epik das Produkt einer rein mündlichen Komposition sehen
möchten. Zwar steht Homer, anders als der babylonische Dichter der kanonischen
Fassung des Gilgamesch-Epos, Sin-leqe-unnïni, nicht in einer jahrhundertealten Tradition
schriftlicher Epik, doch mußte die im Vergleich zur Keilschrift leichtere Verwendbarkeit
der Alphabetschrift den Prozeß ihrer Nutzbarmachung für die Zwecke des Dichters we
sentlich erleichtern und zeitlich abkürzen.
Die schon aufgrund innerer Kriterien wahrscheinliche Annahme, daß die homerischen
Epen schriftlich komponiert wurden, gewinnt somit auch auf dem Wege des Strukturver
gleichs mit dem kunstvollsten Epos des Alten Orients an Plausibilität.
ANMERKUNGEN
1 Hecker (1974: V). Von besonderem Interesse für die im weiteren verfolgte Fragestellung ist Kap. III:
41-67.
2 Tigay (1982). (Für diesen Literaturhinweis sowie für die hilfreiche Diskussion einzelner Fragen bin
ich Herrn Dr. Konrad Volk, Freiburg, zu Dank verpflichtet).
3 S. vor allem Burkert (1984) (mit ausführlichen Literaturangaben).
4 West (1966,1971 und 1978). Zu Hesiod s. auch Walcot (1966).
5 Eine wichtige Rolle spielen in diesem Beweiszusammenhang auch die lexikalischen Entlehnungen
des Griechischen aus dem semitischen und dem altanatolischen indogermanischen Sprachraum, deren
Umfang in der Vergangenheit oft unterschätzt wurde. Vgl. dazu Szemerényi (1974).
6 Burkert (1984:13).
7 Die m. W. älteste von zahlreichen einschlägigen Arbeiten Jensens erschien bereits 1902.
8 Vgl. etwa Ungnad (1923), wiederabgedruckt in: Oberhuber (1977:104-137, s. bes. 133ff.); in jüngerer
Zeit Gresseth(1975).
9 Vgl. Webster (1956:104-116, bes. 114f. und 1960:93-126, bes. 113ff.); zuletzt Miller (1982:16-19);
Burkert (1984: 92-95). Zurückhaltend äußert sich Kirk (1980:245f.).
10 S. die zusammenfassende Darstellung bei Burkert (1984:106-110); ferner Stella (1978:362-391). An
älterer Literatur sei hier genannt: Bowra (1952: passim); Dirlmeier (1955). Die Existenz typischer
Szenen wird für die akkadische Epik mit guten Gründen von Hecker (1974:180, s. auch 61) bestritten.
11 Nach de Liagre Böhl (1957-1971:368). Der postulierte Umfang des Gilgamesch-Epos liegt damit in
der Größenordnung von dreien der bedeutendsten mittelalterlichen Heldenepen, des angelsächsischen
Beowulf (3182 Vv.), der altfranzösischen Chanson de Roland (4002 Vv.) und des altspanischen Cantar
de Mio Cid (3720 Vv.). Möglicherweise handelt es sich bei diesem Umfang um eine Konstante, die
das Maximum darstellt, das eine einzelne Person beim mündlichen Vortrag binnen eines Tages (mit
Unterbrechungen) bewältigen kann.
12 Immerhin erkennt Röllig (1978:264) im ugaritischen Baal-Epos "eine Komposition mit großem
Konzept, sorgfältiger Beziehung der Einzelepisoden zueinander und innerhalb des Gesamttextes."
13 Die XII. Tafel wird im folgenden nicht berücksichtigt, da sie kein organischer Bestandteil der Hand
lung des Epos ist. Ihren eigentlichen Schluß findet die Erzählung mit Tafel XI, wie auch das Wieder
aufgreifen der Vv. I.i.16-21 vom Anfang der I. Tafel am Ende der XI. Tafel (XI. 303-307) zeigt.
14 Vgl. dazu Tigay (1982:passim). Unsicherheit herrscht in der Frage, ob die Bruchstücke der altbabylo
nischen Fassung bereits einem Überlieferungsstadium entstammen, in dem die älteren selbständigen
Einzellieder des Gilgameschstoffes, wie sie in sumerischer Sprache erhalten sind, zu einem längeren
204 MICHAEL REICHEL
Epos mit durchgehender Handlung vereinigt waren. Tigay rechnet schon für die altbabylonische Zeit
mit einem "unified epic on a grand scale" (242), ähnlich Oberhuber (1975:13ff.), während v. Soden
in: Schott/v. Soden (1982:6) Skepsis ausdrückt.
15 Für eine systematische Übersicht über die Tradierung des Gilgamesch-Stoffes vgl. Oberhuber
(1975:22).
16 Kritik am grundsätzlichen Konzept einer Gilgamesch-Harmonie bei Oberhuber (1975:19f.); vgl. auch
Schmökel (1971:17).
17 Vgl. vor allem Duckworth (1933); Schadewaldt (1943); Hellwig (1964).
18 Iliasbücher werden im folgenden mit römischen, Odysseebücher mit arabischen Ziffern zitiert.
19 Gressmann (1911:176).
20 Bei Zitaten aus dem Gilgamesch-Epos bezeichnen große römische Ziffern die Tafel, kleine römische
ggf. die Kolumne, arabische die Verszeile. Die Verszählung, die dt. Übs. und die Schreibung der
Eigennamen durchgängig nach Schott/v. Soden (1982) Kursivdruck weist auf unsichere Lesung oder
Deutung oder auf Ergänzung von Textlücken. Die maßgebliche Textausgabe ist immer noch
Thompson (1930).
21 Oppenheim (1956: bes. 206ff.).
22 J. Hundt (1935:44). Auf die Problematik der typologisch recht unterschiedlichen Traumschilderungen
im homerischen Epos (vgl. bes. IL XXIII.62ff., Od. 4.795ff., 6.15ff., 19.535ff., 20.87ff.) kann hier
nicht näher eingegangen werden. Wichtig ist die Unterscheidung von Außenträumen, bei denen der
Traum "ein Subjekt, eine Person eigenen Wesens und Lebens außerhalb des Schlafenden" ist (Hundt:
1935:43), und ¡nnenträumen, bei denen der "Traum als Traumzustand verstanden ist" (ebda.). Bei
Homer überwiegt deutlich der erste Typus, während die Traumdarstellungen im Gilgamesch-Epos,
soweit erkennbar, durchweg dem zweiten Typus entsprechen. Auch werden die Träume hier als
Regelfall durch den Träumenden selbst nachträglich berichtet, während sie im homerischen Epos
meist aus der Perspektive des epischen Erzählers mitgeteilt werden.
23 Hundt (1935:86).
24 Dt. Übs. der beiden Passagen bei Schott/v. Soden (1982:80f. bzw. 62).
25 Die Vv. 36f. betreffen nach Tigay (1982:151) "tangential aspects of the expedition to the Cedar
Mountain", womit dann auch auf Handlungsteile aus der ersten Hälfte des Epos (Tafel IV und V)
Bezug genommen wäre.
26 Duckworth (1933:39, Anm. 93).
27 Vgl. Tigay (1982:150f.).
28 Auch Gilgamesch, ein historischer König von Uruk in der ausgehenden Frühdynastisch II-Zeit (zwi
schen 2700 und 2500 v. Chr.), wurde schon bald nach seinem Tode, vielleicht sogar schon zu Leb
zeiten, vergöttlicht; vgl. Tigay (1982:13ff.); Falkenstein (1957-1971:359f.).
29 Seine Worte in VII.i.38ff. scheinen auf den Zug in den von Chumbaba bewachten Zedernwald anzu
spielen, vgl. die Bemerkung bei Schott/v. Soden (1982:63).
30 Text und Übs. bei Lambert/Millard (1969).
31 So Tigay (1982:238f.); anders de Liagre Böhl (1957-1971:367).
32 Die Parallele schon bei Webster (1960:113f.). Speziell zu Homer vgl. Willcock (1964).
33 Tigay (1982:239).
34 Grundlegend dazu Kulimann (1960 und 1981).
35 Vgl. Tigay (1982:6).
36 Vgl. schon Goethe über das "Erforderniß des Retardirens, welches durch die beyden Homerischen Ge
dichte überschwenglich erfüllt wird" (an Schiller am 19.4.1797; hier zitiert nach E. Grumach, Goethe
und die Antike. Eine Sammlung, I. Bd., Berlin 1949, S. 147). Zum Prinzip des 'Aufschubs' in der
Ilias s. Schadewaldt (1943:passim, bes.15): "Der Dichter täuscht wohl gar den Hörer über die Weite
des Weges, der ihm bevorsteht. Wie einem Wanderer erscheint dem Hörer dann das Ziel in der Ferne
zuerst zum Greifen nah, dann näher und näher, und entschwindet zwischendurch wohl wieder seinen
Blicken. Zumal ein Dichter, der auf weite Sicht arbeitet, erregt so durch Vorblick und Aufschub
Spannung und hält sie lebendig. Auch vermag er dadurch der Handlung Vielfalt und Fülle und zu
gleich ein zielbewußtes Fortschreiten zu geben." Vgl. damit die unten zitierten Worte Gressmanns
über das Gilgamesch-Epos (m.Anm. 37).
37 Gressmann (1911:177; zu Retardation allgemein s .110, 177-179). Es sei hier betont, daß die durch
keine Quellen gestützte Vorstellung Gressmanns vom mündlichen Vortrag des Epos und seiner
Wirkung auf das Publikum im Schlußsatz der zitierten Passage über den hier verfolgten Ansatz hin
ausgeht, in dem Retardation weder 'rezeptions-' noch 'produktionsästhetisch', sondern rein text
immanent als eine Verzögerung der Handlungsbewegung auf ein bestimmtes Ziel hin betrachtet wird.
38 So (im Anschluß an M. Parry) Lord (1960:143 und 291, Anm. 2).
Schadewaldt (1943:41).
40 Vgl. die Beispiele und ihre Diskussion in der neueren Literatur bei Strasser (1984); Mueller
(1984:149-158: "Repetition and Contextual Surplus"); Gordesiani (1986:72-84).
41 Vgl. etwa Stamm (1952), in: Oberhuber (1977:293-311, bes. 303f.). Eine Wandlung Gilgameschs in
dieser Richtung würde auch zu dem erschlossenen Sinn der stark fragmentarischen Rede
Utnapischtims in X.v.37ff. passen (s. Schott/v. Soden: 1982:89-92).
42 Vgl. Hecker (1974:57)
43 Hecker (1974:59f.). Vgl. auch Tigay (1982:7): "The thematic and structural integrity of the epic is
supplemented by a number of motifs and phrases which echo through it" (mit weiteren Beispielen).
44 Strasser (1984:46, mit Beispielen aus Homer); Hecker (1974:159, mit Beispielen aus dem Gilga
mesch-Epos und anderen Epen); vgl. auch Strasser (1984:47) und Hecker (1974:158) über die
Anpassung von Iterata durch den Austausch von Eigennamen. Anders als bei Homer unterliegen
jedoch in der akkadischen Epik die durch Boten übermittelten Aufträge keiner sprachlichen Variation.
Botschaften werden entweder bereits in angepaßter Form aufgetragen oder aber in unangepaßter Form
ausgerichtet (Hecker 1974:160).
45 Tigay (1982:5).
46 Vgl. Zielinski (1899-1901); Hellwig (1964); Krischer (1971:91-129). Gerade im Bereich der Schau
platzwechsel und Szenenverknüpfungen läßt sich ein deutlicher Unterschied zwischen der homerischen
und der akkadischen Epik fassen. Für letztere sind - auch außerhalb des Gilgamesch-Epos - abrupte,
sprunghafte Übergänge charakteristisch; innerszenische wie auch szenenverbindende Übergänge
unterbleiben häufig. Hecker (1974:50-67) hat diese Phänomene unter dem Begriff der 'epischen
Paralipse' beschrieben und erklärt. Vgl. dagegen zu Homer Krischer (1971:97): "Der Hörer hat [...] nie
das Gefühl, unvermittelt und willkürlich aus einem fesselnden Zusammenhang herausgerissen zu
werden, sondern glaubt immer, vom weniger Bedeutsamen zum Bedeutsameren geführt zu werden."
Für ein anderes mit den Schauplatzwechseln eng zusammengehörendes Phänomen, die doppelsträngige
Erzählhandlung, hat Burkert (1984:108) auf die Parallele zwischen der Erzähltechnik des
Odysseedichters, der ja die Ereignisse auf Ithaka bzw. während der Reisen des Telemachos und die
gleichzeitigen Erlebnisse des Odysseus in eine erzählerische Abfolge bringen muß, und dem Anfang
des Gilgamesch-Epos aufmerksam gemacht, wo der Erzähler zunächst vornehmlich Enkidu folgt, um
dann in den Worten der Dirne den um Gilgamesch zentrierten Handlungsstrang aufzugreifen (I.v.23-
vi.28).
206 MICHAEL REICHEL
47 Eine Parteiung der Götter findet sich auch im Gilgamesch-Epos schon in nuce: vgl. die Diskussion
zwischen Anu, Enlil und Schamasch über die Bestrafung Gilgameschs und Enkidus wegen der Tötung
des Himmelsstieres und des Chumbaba; die Passage ist allerdings nur in einem hethitischen Fragment
überliefert (bei Schott/v. Soden 1982:62f.). Vgl. auch Kullmann (1956:148, Anm. 1).
48 Vgl. das 'Szenarium' derIlias bei Latacz (1985:136-151).
49 Burkert (1984:110).
50 Burkert (1984:85).
51 Vgl. Gresseth (1975:14 und Anm. 24, u. Burkert (1984:92-95).
52 Vgl. Tigay (1982:16, Anm. 65).
53 Tigay (1982:10).
LITERATUR
Bowra, C.M. 1952. Heroic Poetry. London 2 1961; dt. Heldendichtung. Stuttgart 1964.
Burkert, W. 1984. Die orientalisierende Epoche in der griechischen Religion und Literatur. SB Heidelberg.
Dirlmeier, F. 1955. Homerisches Epos und Orient. RhM 98. 18-37. (=Ausgewählte Schriften zu Dich
tung und Philosophie der Griechen. Heidelberg 1970. 55-67.)
Duckworth, G.E. 1933. Foreshadowing and Suspense in the Epics of Homer, Apollonius and Virgil.
Diss. Princeton.
Falkenstein, A. 1957-1971. Gilgameš A. Nach sumerischen Texten. Reallexikon der Assyriologie und
Vorderasiatischen Archäologie. Bd. 111:357-363. Berlin/New York.
Gordesiani, R. 1986. Kriterien der Schriftlichkeit und Mündlichkeit im homerischen Epos. (=Studien zur
klass. Philologie 19.) Frankfurt/M.
Gresseth, G.K. 1975. The Gilgamesh Epic and Homer. CJ 70.1-18.
Gressmann, H. 1911. Das Gilgamesch-Epos, neu übs. v. A. Ungnad u. gemeinverständlich erklärt v. H.
Gressmann. (=Forschungen zur Religion und Literatur des Alten und Neuen Testaments 14.)
Göttingen.
Hecker, K. 1974. Untersuchungen zur akkadischen Epik. (=Alter Orient und Altes Testament, Sonderreihe
Bd. 8.) Kevelaer/Neukirchen-Vluyn.
Hellwig, Brigitte. 1964. Raum und Zeit im homerischen Epos. (=Spudasmata 2.) Hildesheim.
Hundt, J. 1935. Der Traumglaube bei Homer. (=Greifswälder Beiträge zur Literatur- und Stilforschung 9).
Greifswald.
Jensen, P. 1902. Das Gilgameš-Epos und Homer. ZA 16. 125-134.
Kirk, G.S. 1980. Griechische Mythen. Ihre Bedeutung und Funktion. Berlin. (= Dt. Übs. von: The Nature
of Greek Myth. Harmondsworth 1974.)
Krischer, T. 1971. Formale Konventionen der homerischen Epik. (=Zetemata 56.) München.
Kullmann, W. 1956. Das Wirken der Götter in der Ilias. Untersuchungen zur Frage der Entstehung des
homerischen 'Götterapparats'. Berlin.
.1960. Die Quellen des Ilias (Troischer Sagenkreis). (Hermes-Einzels. 14.) Wiesbaden.
,1981. Zur Methode der Neoanalyse in der Homerforschung. WSt, N.F., 15 (94). 5-42.
Lambert, W.G. / Miliard, A.R. 1969. Atra-hasïs. The Babylonian Story of the Flood. Oxford.
Latacz, J. 1985. Homer. Eine Einführung. München/Zürich.
de Liagre Böhl, F.M.Th. 1957-1971. GilgameS B. Nach akkadischen Texten. Lexikon der Assyriologie
und Vorderasiatischen Archäologie. Bd. 111:364-372. Berlin /New York.
Lord, A.B. 1960. The Singer of Tales. Cambridge/Mass.
Miller, D. Gary. 1982. Homer and the Ionian Epic Tradition. (=Innsbrucker Beiträge zur Sprachwissen
schaft 38.) Innsbruck.
Mueller, Martin. 1984. The Iliad. London/Boston/Sydney.
Oberhuber, K. (Hrg.). 1977. Das Gilgamesch-Epos. (= WdF 215.) Darmstadt.
,1975. Gilgamesch. In: Oberhuber (1977:1-22).
Oppenheim, A.L. 1956. The Interpretation of Dreams in the Ancient Near East. Philadelphia.
Röllig, W. 1978. Die ugaritische Literatur. W. Röllig (Hrg.), Altorientalische Literaturen. (= Neues
Handbuch der Literaturwissenschaft Bd. 1.) 255-271. Wiesbaden.
Schadewaldt, W. 1943 (= 2. Aufl.). Iliasstudien. Leipzig1 1938. (Nachdr. Darmstadt 1966.)
Schmökel, H. 1971 (= 2. Aufl.). Das Gilgamesch-Epos, eingef., rhythm, übtr. u. m. Anmm. vers.
Stuttgart1 1966.
Schott, A./v. Soden, W. 1982. Das Gilgamesch-Epos, übs. u. m. Anmm. vers. v. A. Schott, neu hrg. v.
W. v. Soden. Stuttgart2 1958 (mehrere Auflagen).
Stamm, J.J. 1952. Das Gilgamesch-Epos und seine Vorgeschichte. Asiatische Studien 6. 9-29; wiederab
gedruckt in: Oberhuber (1977:292-311).
Stella, Luigia Achillea. 1978. Tradizione micenea e poesia dell' Iliade. (=Filologia e Critica 29.) Roma.
Strasser, F. X. 1984. Zu den Iterata der frühgriechischen Epik. (= Beiträge zur klass. Philologie 156.)
Königstein/Ts.
Szemerényi, O. 1974. The Origins of the Greek Lexicon: Ex oriente lux. JHS 94.144-157.
Thompson, R. C. 1930. The Epic of Gilgamesh: Text, Transliteration and Notes. Oxford.
Tigay, J. H. 1982. The Evolution of the Gilgamesh Epic. Philadelphia.
Ungnad, A. 1923. Gilgamesch-Epos und Odyssee. Breslau; wiederabgedruckt in: Oberhuber (1977:104-
137).
Walcot, P. 1966. Hesiod and the Near East. Cardiff.
Webster, T.B.L. 1956. Homer and Eastern Poetry. Minos 4. 104-116.
.1960. Von Mykene bis Homer, München/Wien. (= Dt. Übs. von: From Mycene to Homer, London
1958; New York2 1964.)
West, M. L. 1966. Hesiod. Theogony. Edited with Prolegomena and Commentary. Oxford.
.1971. Early Greek Philosophy and the Orient. Oxford.
.1978. Hesiod. Works and Days. Edited with Prolegomena and Commentary. Oxford.
Willcock, M. M. 1964. Mythological Paradeigma in the Iliad. C1Q, N.S., 14 (58). 141-154.
Zielinski, Th. 1899-1901. Die Behandlung gleichzeitiger Ereignisse im antiken Epos. Philologus. Suppl.
8. 405-449.
SUMMARY
This paper deals with the general structure of the Akkadian Gilgamesh epic in the Ninivite recension. It
starts out from the question whether some of the compositional techniques that help to unify the Homeric
poems and that have been examined in Homeric scholarship occur in the Gilgamesh epic as well. These
208 MICHAEL REICHEL
techniques are: 1. preparation and anticipation of events taking place later in the epic (by the poet himself
or by speeches of the characters, by dreams etc.); 2. recapitulation of earlier events in the epic; 3.
integration of events lying outside the action of the epic (e.g. the Meleagros paradeigma in the Iliad, the
flood story in the Gilgamesh epic); 4. retardation; and 5. compositional use of verse repetition (including
repeated verse groups). The fact that all of these compositional devices can be found in the Gilgamesh epic
constitutes a further point of comparison between the oriental and the Homeric epic. Besides motif
parallels and stylistic features (e.g. fixed epithets, whole-line formulae) questions of structure should also
be taken into account in the discussion of the relationship of the early Greek epic to Near Eastern epic
poetry. In the case of the Gilgamesh epic these structural techniques were developed in the course of a
long written tradition. It is, therefore, likely that the Homeric poems, which make use of the same
techniques, on a much more sophisticated level, are also the result of written, not oral composition.
V.
ITALIC AND LATIN PHILOLOGY

SOME REMARKS ON THE CARMEN ARVALE
PATRICK CONSIDINE
University College London
The Carmen Aruale is a chant of the Fratres Aruales, an ancient Roman religious
brotherhood, moribund in later Republican times (before Augustus it is mentioned only
by Varro De Lingua Latina 5, 85) but revived by Augustus and known from the Acta
Fratrum Arualium over a period of two and a half centuries from 21 . - 241 A.D.,
during which time the principal ceremony was in honour of Dea Dia at her grove five
miles from Rome1. The Carmen invokes the aid of Mars for his worshippers and is
contained in the Acta for 28th May 218 A.D., preserved on a marble tablet2. The last
seven lines of the tablet (lines 31-38) are given below. Particular attention must be drawn
to Gordon's warning that 'this is one of the worst written, least legible, Latin
inscriptions, especially in the Carmen' (Gordon 1983:160) and that 'we still consider it
impossible to present the correct text of the Carmen from this inscription with any
assurance of certainty; not merely the lettering but even the word divisions are uncertain'
(Album 42)3.
I offer here a few comments on reading and interpretation4, in the hope that they may
be of some small interest to the distinguished scholar who is honoured by the present
volume, and who initiated my interest in Latin inscriptions twenty five years ago with a
series of lectures which are a valued memory and inspiration still.
An impression of the appearance of the text on the stone was provided by Ritschl's
reproduction of 1862, which was justly praised by Calvo (1957:388). For the details see
212 PATRICK CONSIDINE
CARMEN ARVALE 213
Gordon Album 41-6 (also Introduction with the excellent photograph PLATE 49).
The drawing below is based on Gordon's publications; Ritschl's reproduction is given for
comparison. It is evident from the introduction (libellis acceptis) and conclusion (libellos
receperunt), that in 218 AD the priests read their chant from a book, written, no doubt, in
cursive script; the extreme inaccuracy and many ambiguities of the text are in part due to
errors in transcription from the imperfectly understood exemplar, itself the product of a
centuries old tradition in the course of which the chant had become corrupted by
misunderstandings both by readers and listeners, like the Saliorum carmina uix
sacerdotibus suis satis intellecta (Quintilian 1.6.40). Cf. Bréal (1881).
We have no hope of restoring the original, but we can hope to get behind our present
text to something more like the original. Despite the generally wretched state of the text, it
has the great advantage that every phrase in it is given three times, so that it is to that
extent self-correcting. It is also of course possible that this triple phraseology had earlier
served to entrench error; for a mistake once made in one of the three versions could then
be treated as the norm against which to 'correct' the other two.
In the following transcription5 I have inserted word gaps of which I believe there is
little or no doubt (even in one or two cases where the word division has been questioned;
see below). In at least nine6 cases the engraver has marked divisions with a point, but
there are few if any cases of word division deliberately indicated by a space. The main
problems are in lines 33-5.
31 IBI SACERDOTES
32 ÇLVSISVCCINCTI LIBELLIS • ACCEPTIS CARMEN DESCINDENTES
TRIPODAVERVNT IN VERBA HAEC ENOS LASES IVVATE
33 [.] NOS LASES IVVATE ENOS LASES IVVATE NEVELVAERVE
MARMA SINSIN CVRREREINDLEORES NEVELVERVE MARMAR
34 [..] NSIN . CVRREREINPLEORIS NEVELVERVE
MARMAR - SER.INCVRREREINPLEOIVS:SATVR . EVRERE MARS LIMEN
35 [..].E STA • BERBER SATVR • FVFERE MARS LIMEN SALI
STA BERBER SATVR FV FERE MARS LIMEN SAH SIA BERBER
36 [...] YNIS ALTERNEI ADVOCAPIT CONCTOS SEMVNIS ALTERNEI
ADVOCAPIT CONCTOS SIMVNIS AITER ... ADVOCA ...
37 [ ] OS ENOS MARMOR IVVATOENOS MARMOR IVVATO ENOS
MAMOR . IVVATO TRLVMPE TNVMPE TRIVMPE TIVM..

38 [....} MPE POST TRIPODATIONEM DEINDE SIGNO D..Q
PVBLLCL INTROIER ET LIBELLOS RECEPERVNT
The text of the Carmen itself in lines 32-7 consists of five series of letters each of
which is twice repeated (for SINS-SINS-SER. in 33-4 see below); so the damage to the
stone and the fact that many individual letters are illegible, ambiguous or altogether
omitted, are less of an obstacle to establishing the text than they would otherwise have
been. Thus in 33 LVAERVEMARMA can confidently be emended to
LVERVEMARMAR in the light of the two repetitions of the phrase; in 35.]E - SALI -
SAH (Gordon: ] LE-SALL- SNI) can similarly be interpreted as SALI (which must have
been intended by the engraver in the second and even the third occurrence, just as
TRIVMPE is no doubt intended by TNVMPE in 37; cf. the variant letter shapes in the
drawing). Similarly in 34 it is probably safe to replace FVRERE by FVFERE in the light
of 35, and in 36 to restore AITER ... as ALTERNEI. Nevertheless some generally
accepted readings are less certain, and the following facts should be mentioned:
1. In line 33 'neither CIL nor Hübner has a pt. between in and currere, and all the
editions cited read sins incurvere here and in line 34, but in two of the
(apparently) three occurrences of the phrase the break seems to be before currere'
(Gordon Album 46). It is in fact certain that the engraver intended currere, not
incurrere. From which it seems to follow that he also intended SINSIN, i.e. sin sin
not sins in; see below.
2. In the third occurrence of the same phrase (line 34), the reading SERS is generally
adopted. But the fourth letter looks as if it might be a defective T or F or P (Gordon
SERPLN); it does not look at all like an S, unless indeed it be understood as an
absent-minded attempt at a cursive S. It is true that with this engraver almost anything
is possible, there are other signs of cursive influence, and other examples of his S
would be barely recognisable in isolation; but rarely can such scholarly ingenuity have
been expended on explaining a form for the existence of which there is so little
evidence (e.g. Radke 1981:1l0f.: sers < *serīs < sesīs < se-sə-īs). Pisani (1975:3)
was surely right to see in SER. a corruption of SINS resulting from a misreading of,
or in, the cursive exemplar (for his explanations of SINS see note 7 below).
CARMEN ARVALE 215
3. The sequence INPLEORES in 33 is followed in 34 by INPLEORIS, and then by

INPLEOIVS (Gordon PLEORIS 'which looks quite as much like pleoius)'.
4. In 35 BERBER and in 36 ADVOCAPIT are clear only in the first occurrence and are
unintelligible in the third (even the first ADVOCAPIT was - understandably - read by
Ritschl as ADVOCADIT. Cf. note 11).
The text on the stone seems therefore to be a careless reproduction of the following:
1 (31-2): ibi sacerdotes clusi succincti libellis acceptis carmen descindentes tripo-
dauerunt in uerba haec:
2 (32-33): enos Lases iuuate (ter)
3 (33-34): neueluerue Marmar sinsincurrereinpleoris (ter)
4 (34-35): satur fufere Mars limen sali sta berber (ter)
5 (36-37): Semunis alternei aduocapit conctos (ter)
6 (37): enos Marmor iuuato (ter)
7 (37-38): triumpe triumpe triumpe triumpe triumpe
8 (38): post tripodationem deinde signo dato publici introiere et libellos
receperunt
Lines 3-4 are usually further analysed as follows (e.g. CIL, Norden, Gordon):
3: neue lue rue Marmar sins incurrere in pleores
4: satur fu fere Mars limen sali sta berber
However, other interpretations are possible7. The most substantial discussion of the
Carmen since Norden (1939) is that of A. García Calvo (1957), who agrees with much of
4 but with hardly a word of 3 and 5. Against the prevailing view that the Carmen is
addressed to Mars in his capacity of god of agriculture and the wild, Calvo believes that
he is invoked as the god of war (cf. note 11). It is not possible here to discuss all the
arguments, but it will be instructive to set out Norden's text and interpretation of 3-5,
followed by Calvo's. The following versions are distilled from their respective articles,
and English translations are added:
Norden:
neue lue(m) rue(m) Marmar sins incurrere in pleores
Und nicht Seuche Schaden lasse einlaufen in die Menge!
Nor plague and pestilence, Mars permit to advance against the masses!
satur fu, fere Mars limen sali, sta berber

satt sei, wilder Mars! auf die Schwelle springe! stehe dort!
Be satisfied, wild Mars! Leap the threshold! stand there!
Semunis alternei aduocapit(e) conctos

Die Semonen wechselweise werdet ihr herbeirufen, sie alle.
The Semones in turn shall ye call, all of them.
Calvo:
ne uelue rue Marmar! sin sin currere, imple oris!
No des vueltas! cae, oh Marte! deja, déjanos correr! llena los oídos!
Stop turning, prostrate yourself! Let us, let us run! fill your ears!
 Mars! (i.e. Mars-worshipper)
satur fu, fer e Mars limen sali, sta berber
hártate! lánzate, oh Marte! salta el umbral! en pie, voceador!
Be satiated! attack,  Mars! Leap the threshold! Stand, crier!
Semunis, (s)al! (s)terne, i! advoca! pet conctos!

Semune, salta! derriba, ve! Ea, llama a tu lado! arremete con todos
(a todos los tuos) (los enemigos)
Semunis, leap! overthrow, go! Call to your side! Attack all!
(your supporters) (your enemies)
Or:
Se muni sal!
Salta lejos del muro!
Leap far from the wall!
That the chant accompanies a procession and dance is clear from lines 32 and 38.
Calvo more precisely thinks of the worshippers dancing around a sacred trophy ('un
CARMEN ARVALE 217
trofeo armado con los ancilia'); they interrupt the dance with a cry 'Stop turning! Throw
yourself to the ground!' and resume it with 'Let us run! Fill the ears (sc. with the clash of
the bronze weapons they were brandishing)'. There follows a call to Mars to leap out
upon his enemies. Calvo notes that the thirteen imperatives in his version (uelue, rue, sin,
imple, fu, fer, sali, sta, sal, sterne, i, aduoca, pet, Norden recognises only three, fu, sali,
sta) are all presents, and so contrast with the final iouato, with its marked future reference
- 'ayúdanos siempre'. Which if any of them offer a reading superior to that of the general
consensus must be a matter for detailed discussion; Calvo's interpretation is given here to
illustrate the extent of the disagreement and uncertainty which still attach to the
interpretation of this text. My own view is that his reading of the Carmen as a war chant
of twenty five words, the Arval Brethren being in effect a branch of the Salii, is not an
improvement on Norden's view that it is a (twenty word) prayer for the crops, which
shows a good deal of formal, if not substantial, Greek influence. But students of the
Carmen are very much in Calvo's debt for the quality of the critical judgement which
characterises his important article, and for some stimulating individual suggestions.
1. ENOS
There is nothing to be said for the old view that enos is an old form of nos (comparing
Greek έ-μέ) and it was rightly rejected by Lindsay and Norden. There is no other
evidence for it, and where known Latin forms can be read and make sense, they should
be preferred to speculative reconstructions. It is less easy to choose between e nos and en
nos. En, preferred by Mentz, is well attested with imperatives in Classical Latin, and is in
common use as an interjection with the affective function of engaging the hearer's
attention, which is precisely its function here. E, although ancient, has no existence apart
from such forms as ecastor, edepol and equidem (cf. Osean etanto, Umrian etantu) and is
simply asseverative; en is therefore to be preferred. Although the writing of doubled
(long) consonants of the pronunciation with two letters was general by the first century
, there can be no difficulty about assuming the retention of the earlier (until late  
invariable) spelling in the Carmen, which in the same phrase preserves the ancient (TV 
and earlier) Lases for Lares. Bréal (1881:376f.) suggested enom (cf. Umbrian enom =
turn).
2. NEVELVERVEMARMARSINSINCVRREREINPLEORIS
This sequence presents the most difficult questions of word division in the Carmen.
The usual division into neue lue(m) rue(m) Marmar sins incurrere in pleoris is open to at
least four objections, viz. (i) neue 'neither, nor' is a very improbable first word of an
invocation to Mars, following the invocation of the Lares, and it is possible that
NEVELVE is ne uelue rather than neue lue (ii) if LVE is a word, it and RVE are then on
the face of it imperatives, sc. lue and rue8; and while there is no objection in principle to
assuming orthographic omission of final - m, which is very common (and particularly
easy before Marmar), neither is it immediately obvious that it must be assumed in this
case. The asyndetic *lues rues (cf. Greek adduced by Norden and others) is an
attractive reading, but certainly not the only possibility (iii) the form *sins does not exist;
such expansions as *sine and (worse) *sinas are mere guesses (despite Radke's confident
assertion to the contrary, 1981:109), and the sequence SINSIN at once suggests that
SINS is a ghost word generated by dittography (iv) although in pleoris is (despite
objections that have been made) perfectly possible archaic Latin for in plures, it is
semantically awkward9, and a division imple oris should be considered as a possibility.
Clearly these questions are inter-dependent; Calvo's analysis is to be preferred to
others in so far as it divides the line into two self-contained halves, each containing a
simple address to the god (of the same kind as the following line), expressed in simple
Latin words and using imperative forms, as expected in such an address. His last word
*oris for auris 'ears' (with rustic  for au) will, I believe, generally be found implausible;
but if imple is an imperative, then a clue to what followed may be contained in the second
of the three occurrences of the phrase 'which looks quite as much like pleoius' (Gordon
Album 46; see above). The sequence OIVS could be metathesis of OVIS (cf.
ALTERNIE, if rightly read for ALTERNEI in line 36; see note 5), and OIVS in cursive
script might well be mis-read as ORIS, which in turn, being interpreted as the second half
of a form pleoris, was replaced by the correct consonant stem form pleores, whatever the
precise developments which led to the three distinct spellings appearing on our stone. The
phrase *imple ouis would make good sense as 'make fertile the sheep' (OLD impleo: 4 to
make pregnant); Mars is associated with the sheep by the sacrifice to him of sheep, bull
and pig on the Campus Martius at the Ambarvalia in May (suouetaurilia, OLD s.v., with
CARMEN ARVALE 219
refs.), and by the fact that the principal sacrifice at the meeting of the Arval brethren was
that of a sheep on the second day; the idea that the god should give increase to the animal
that is to be sacrificed to him is a natural one. See further below.
Cuny, followed by Mentz, proposed to take PLEORIS as *plōris, a by-form of
*plõ-ro-/plã-ro- seen in Irish lãr, English floor etc.: 'das wäre also genau die arva, die
Fluren, die man sonst in dem Gedicht vermisst' (Mentz 1952:220). But there are so many
objections to this suggestion that it would be less exceptionable simply to restore *arua.
Certainly a reading *imple arua would be semantically ideal in the Carmen Aruale; the title
of the brotherhood and of the hymn itself leads one to expect the word in the text, as
Mentz observes; it would fit here better than anywhere; the present text is uncertain; and
ARVA has the right number of letters, of which the second and third each occurs in the
right position in one or other of the three forms on the stone (and cursive a and  could
easily be confused). But these indications hardly amount to even aprima facie case that
original arua was replaced by ouis. It may have been (or the two may even have co
existed, either within the chant or as alternative versions of the kind so common in
modern hymns); but I can suggest no clear line of development, and although it would
solve difficult problems to assume original arua, we cannot assume it.9a
If we treat IMPLE as the imperative imple and omit one SIN, then the simplest
interpretation of the first part of the line would be ne uelue rue(m), 'do not bring back
pestilence' (so Ferri), with retention of early Latin uel- instead of modernisation to uol-;
the meaning would be close to that given in OLD uoluo: 2 to bring round (events, seasons
etc.) in due course, and the existence of a noun rues (replaced by ruina) seems to be
independently guaranteed (Norden 121: CGL VII 216), and is assumed by the many
scholars who read neue lue(m) rue(m).
There remains a serious objection to the assumption that SINSIN is due to
dittography. Not that the engraver evidently recognised a word boundary before
CVRRERE, for it is possible, as implied above, that he anticipated Calvo in regarding
SIN as an apocopated imperative sin(e), in which case an accidentally reduplicated
SINSIN could easily be taken as emphatic and generalised (but see Pighi 158 on
apocope). The difficulty is rather that, if sins goes, so does the rhythm of the phrase,
which indeed becomes acephalous; and that INCVRRERE is left floating without an
anchor. This second difficulty is not very serious; apart from the possibility of an
imperatival infinitive, or an old deponent by-form of curro, a word division incurre

reimple is perhaps possible. Although the double prefix is not otherwise attested in this
form, redimpleo is well known, with regular classical red- for re- before a vowel. Is the
first objection fatal? I am aware of skating on thin ice as I go on to suggest that not merely
is SINSIN the result of dittography, but that this part of the inscription has been affected
by haplograpy also, and indeed that the two may be interconnected: that the sequence
MARMARSINSIN is a reshaping, (perhaps a deliberate correction made when -SIN- was
interpreted as an old imperative sin) of MARMARMARSIN, and that in an earlier form
the text ran:
NE VELVERVEM MARMAR MARS INCVRRE RE IMPLE OVIS
Make not pestilence return, Mars hasten hither, reimpregnate the
 Mars! sheep! for re<d>imple)
If it be objected that the suggestion is absurd, because Mars does not in fact make the
sheep fertile, it will be enough to recall the words of Dumézil (1966:234) in commenting
on Cato de Agri Cultura 141: 'Mais priant Mars, voulant intéresser Mars, le paysan le
traite le plus possible en agent, amplifie le plus possible l'intervention qu'il sollicite. En
conséquence il fait de Mars le sujet de tous les verbes...' The juxtaposition of ancient
Marmar and modern Mars is striking but both forms appear in most interpretations
(Radke's dissentient view is discussed below) and the combination of archaic and modern
in early Latin texts is common.
3. SATVRFUFEREMARSLIMENSALISTABERBER
The widely accepted division satur fu fere Mars limen sali sta berber has been
challenged by Radke (1981:111 f.), who objects to Norden's rendering because of 'drei
Schwierigkeiten die bisher nicht beachtet wurden', namely, in ascending order of
significance, that it does not specify of what Mars is satur, that it provides no pronominal
element to which the reduplicated -ber is appended (in contrast to o Haber, ollaner), and
that the vocative of Mars should be Mar, as shown by Marmar. Radke meets his own (not
very cogent) objections by positing satur fu, fere Mar, slimens; ali sta berber, i.e. 'Be
sated,  Mars, with strife; set it elsewhere'. For the noun *slïmen (from *stlïmen) he
compares classical līs 'law suit', earlier stlīs [cf. the phrase decemvir stlitibus iudicandis],
CARMEN ARVALE 221
with a -men formation like germen [<*gen-men] beside gens, and treats *slïmens as a
syncopated form of the genitive *slîmenes. This interpretation bristles with difficulties.
Radke sees three difficulties in Norden's interpretation; I see six in Radke's. The
existence of a suffix -ber is quite uncertain, for the forms ollaber and ollaner occur only in
a formulaic ollaber/-ner arbos in a difficult passage in Varro LL 7:8, and a suffix -ber is
nowehere else attested (Palmer's reference 1954:65 to our berber in elucidation of ollaber
is no more illuminating than Radke's reverse procedure. Without further evidence there
can be no presumption that the two are related, that berber contains a suffix -ber, whether
reduplicated or not, or that a suffix -ber exists in Latin). Both the formation of *slimen
with -men and the form of the putative genitive are assumed ad hoc, and the latter
especially must be suspect. One must also view with reserve the reduplication of a suffix;
the startling tmesis with which it is associated; the unusual transitive use of stare', and the
fact that the proposed utterance runs quite counter to the rhythm of the rest of the chant -
and indeed in my view is simply prose, despite Radke's claim to the contrary.
The sequence then seems to consist of eight Latin words, of which six are very
common and two are the transparent fu and the baffling berber. There
are no difficulties of morphology or syntax, and yet, quite apart from berber, the meaning
of the words has been as much discussed as that of the apparently much more intractable
preceding section. Of course much depends upon one's view of the nature of the god
Mars; I assume here only that he was god of the wild as well as of war, and that 'Mars ist
auch Gott des ver mit seinem Monate beginnt das ver' (Eisenhut Kleiner Pauly 5:
1182). Bearing in mind the words of Dumézil quoted above, a possible translation of the
first four words would then be 'Be abundant, wild Mars', an appeal to Mars as lord of the
wild places, who, himself untamed, has power to tame, to protect the fields and provide
for the people's needs in the way of crops; cf. OLD satur 2: {of places) rich in stock or
produce. (This does not necessarily imply that the god is in origin a vegetation deity in the
sense denied by Norden 1939:138). The limen will then perhaps be not the threshold of
dwelling, temple or city gate but the boundary between the wild and cultivated land, i.e. it
would be used here in the sense of classical limes (as has sometimes been suggested; see
e.g. Calvo 433).
The sequence BERBER has, as Norden observed (1939:142), attracted more
discussion than almost any other in the Carmen. It appears to be (i) a reduplicated form or
(ii) an assimilateci one, like barba or (iii) a late

spelling confusion of  and b as in uerbex = ueruex 'wether' (see below). If it is a
product of assimilation, it is evidently in origin a compound, and a possible first element
would be uer 'spring'. Since the festivals of Mars began with the advent of the spring,
and since he gives his name to the month which heralds the season, it is tempting to think
of uernifer 'spring flowering' used by Martianus Capella (early fifth century AD; I 1
coronis uerniferis florentia limina sertis) and to posit a similar earlier compound
'spring-bringing', whence berber. Or one
might think of uerber, not in the sense of 'lash, rod' but in that of the derivative uerbēna
'foliage, branches', especially as used in religious ritual: von
Grienberger 1906 160, followed by Mentz 1952 223, suggested uerber with initial ft- for
u-, as often in popular Imperial texts, with orthographic ambivalence following phonetic
convergence, e.g. biclus = uiclusluitulus, or berbeces/uerbeces - uerueces in the Acta (cf.
Väänänen 1981 50). A more remote possibility might be *werwi-feros 'protection-
bringing', formed to *weru-'protection' (Sanskrit varu- etc.; IEW 1161) with the root
*wer- seen in Latin uereor 'fear', originally 'be on one's guard against', and regular Latin
adaptation of the Indo-European -u stem to an -i stem.10
4. SEMVNISALTERNEIADVOCAPITCONCTOS
Heavy weather has been made of this line by commentators, and in particular there
has been a curious reluctance to accept that the sequence ADVOCAPIT is what it appears
to be, i.e. that it represents aduocabit and means 'he will call'. Apart from Calvo's
version above, Pisani has ad vo(s) capite (see note 7), Warmington invents a form
'advocapite = advocate since all the other sentences are imperative', and even Norden
finds a non-existent second plural future in -te in advocapit(e): see Pighi 158 on the
'entirely hypothetical' nature of such an apocope.
Assuming that the Ρ is intended in all three occurrences11 (see above), its being used
for  is on a par with other sporadic interchanges of letters for the voiced and voiceless
stops from the early Empire onwards, e.g. SCRIPIS for SCRIBIS at Pompeii (CIL IV
1623); that is to say, it is a reverse spelling testifying to an incipient devoicing of
intervocalic voiceless stops, a process apparently complete by the fifth century (Väänänen
1981 57). The idea that the Ρ may be a very ancient reflex of the absence of signs for
CARMEN ARVALE 223
oiced stops in the Etruscan-derived alphabet (voice being non-phonemic in Etruscan) is a

nuch less likely explanation. The view that the P is actually evidence for the antiquity of
he text (Norden 117f.; earlier Bréal 1881:376) I find altogether implausible.
There can be no reasonable doubt that the meaning of the four words is 'he will call
apon the Semones in turn', sc. to play their part in protecting the fields and crops. The
subject of the verb is conceivably Mars; alternation between second and third person
forms in prayer is not unprecedented, and is a constant feature of, for example, the
psalms of the Old Testament. But this assumption raises some problems, or at least
questions, not least about early Roman religion and ritual; and it seems simpler to interpret
the words as a rubric 'one will invoke the aid of the Sowers in turn', the subject not being
expressed, as often in early Latin.12 Here no doubt it would be the priest (originally the
king?) or whoever is leading the prayer. When (or even whether, in ancient times) the
words came to be felt to be part of the chant, it is impossible to say. The principle is that
of the incorporation of a gloss in a text, and it may be that aduocabit, from being an
instruction to the priest, came to be felt to be a statement about the activity of the god.13
NOTES
1 For the basic facts see the article Fratres Arvales in the Oxford Classical Dictionary,Oxford 1970, or
(much fuller) Arvales Fratres in the Kleiner Pauly, Lexikon der Antike, Munich 1975. A recent study
with much bibliography is Ronald Syme, Some Arval Brethren, Oxford 1980.
2 The damaged tablet was found in 1778 and is now 'attached to the wall of the sala a croce greca in the
Vatican (inv. no. 215)': see Gordon 1983:160. The text is in CIL VI 2104 a, Gordon's Album No.
276. The Carmen Aruale is separately treated in CIL I 2 2, Gordon 1983 and many other places. See
bibliography.
3 Lommatsch 1918 'Scriptura lapidis pessima est' and 'Minutis litteris neglegentissime insculptum
carmen, quod neque quadratarium neque etiam sacerdotes tunc intellexisse consentaneum est'. The
average height of the letters is stated by Gordon to be 8-9 mm. Ritschl's 1862 reproduction was full
size.
4 The vexed question of the metre must be considered on another occasion. That it is of Saturnian type
is agreed by all. But that it is composed in Saturnians directly comparable with those of Livius
Andronicus, Naevius and other third to second century  sources is to my mind far from certain,
though often assumed. The fundamental question remains that of the nature of Saturnian verse in
general; whether it is syllabic, accentual, quantitative or hybrid, the structure of the lines, the origin
of the metre are interlocking questions to which we seem to be as far from an agreed answer as ever.
Pasquali's theory {Preistoria della poesia latina, Florence 1936) of a 'sintesi da elementi greci', of
Greek lyric elements borrowed in the sixth century  from the Greeks of Southern Italy and adapted
to narrative in the fourth, was congenial to Norden with his view that the Carmen Aruale 'ist nach
griechischem Vorbild geformt, wahrt aber römische Art', a carmen 'graecanicum'. It has been favoured
by others and a Greek origin recently endorsed by A.S. Gratwick {Cambridge History of Classical
Literature II, 1982, 57), whose stated reasons for his view are quite inadequate and are clearly shown
to be so in the survey to which he refers by T. Cole {Yale Classical Studies, 1969 [not 1972]. Cole's
own preferred view of the Saturnian as 'a predominantly syllabic Indo-European metre partially
though never wholly transformed by the application of principles borrowed from Greek versification'
is closer to Leo {Der Saturnische Vers, Göttingen 1905), who proposed a complex quantitative
analysis but held that 'alles erweist mit voller Sicherheit seinen latinischen, das heisst italischen
Ursprung.' This view of a native origin constitutes a little common ground with proponents of once
fashionable (and perhaps still in large measure correct) syllabic/accentual interpretations of the metre
as presented for example by W.M. Lindsay (a Saturnian line comprised a hemistich of seven syllables
with three accents followed by one of six syllables with two accents: American Journal of Philology
15, 1893, 139 ff., 305 ff.; cf. The Latin Language Oxford 1894:1281, 159; but in Early Latin Verse,
Oxford 1922, l0f. the Saturnian is an ignis fatuus and Lindsay does not mention his earlier work);
H.W. Garrod's Note upon the Saturnian Metre {Oxford Book of Latin Verse 1912:505 ff.) is still of
interest. Some important surveys are: M. Barchiesi Nevio epico Padua 1962:310 ff.; W. Beare Latin
Verse and European Song, London 1957:114 ff.; T. Cole sup. cit.; . Luiselli II verso saturnio
Rome 1967; Radke 1981 Beobachtungen zum Saturnier 54-68 with 187-193. With reference to the
Carmen Aruale: Norden 1939:278 ff.; Bickel 35 f.; R.  Tanner, The Arval Hymn and Early Latin
Verse, Classical Quarterly 11, 1961, 209 ff.
It will be evident that it is perilous to invoke the metre in support of readings or emendations in the
Carmen Aruale. Cf. Bréal 1881: 377 2 and generally on principles of interpretation.
5 This transcription neglects the sporadic use of the i longa (there are some clear cases, including a
cluster of five in line 32, but others are made uncertain by the very uneven lettering) and of the left-
facing L (line 35, LIMEN bis). The letters I and L seem to be used inter-changeably. Gordon's Album
transcription (which is somewhat more cautious than that of the Introduction) differs from the above
in that it restores the beginnings of lines, reproduces the i longa and reversed L, has a number of
additional dots under letters, and reads SERPLN-PLEORIS in 34 (SER.IN - PLEOIVS above; see
further in text), SNI STA in 35 (SAH SIA above), AITERVIE ADVOCNIT in 36 (AITER...
ADVOCA...) and DATO in 38 (D..Q). Other differences are trifling. Cf. next note.
6 Radke's comment that the scriptura continua 'wird lediglich durch acht (nicht sieben, wie E. Norden
a.a.O. 117 sagt) einzelne Punkte unterbrochen' ist misleading. Gordon recognises nine, of which
those before the first SATVR and first BERBER (34-5) are very faint, but the state of the text is such
that one cannot be certain of the precise number intended by the engraver.
7 For an analysis very different from those discussed in the text above see Pisani 1975 and 1981: ne
uelueru e Marmar sins incurrere in pleores = ne uoluere  Mars semper incurrere in plures; satur
fufere, Mars = satis fuere, Mars; limensalis taberber are two dvandva compounds, 1. lim- + en-sali-s
'hunger and earthquake' (Gk.\λιμós\ Osean limu 'hunger' + 'leaping within, convulsion'), 2. tabes
'wasting, decay'+ *berber 'destruction', neuter noun in -er from *bherdh-, cf. Gk. πέθω 'waste,
destroy'. So: 'Do not wish, O Mars, always to be falling upon more people; there have been enough
famines, earthquakes, waste and destruction.' For ADVOCAPIT Pisani proposes ad vo capit(e), sc.
'you (two) Semones take (us) all under your protection (after Mars has stopped plaguing us), vo being
dual, the 'tmesis' like that of sub uos placo. The 1981 article re-interprets sins as a very early Greek
loanword (σιvos 'hurt, harm'). Most of this seems to me to be rather recherché.
CARMEN ARVALE 225
8 Whether or not R. M. Ogilvie (The Romans and their Gods, London 1969, 88) is right in saying that
a procession like that of the Ambarvalia 'was called a lustration from luere (to loose) because it was
believed that it freed the fields from harmful forces' (there is no general agreement about the
etymology of lustrum), the possibility that LVE is the imperative lue should be considered; one
might then think of a metathesis of the first two letters of this section of the hymn, and read en nue,
lue, rue Morman! 'Favour (us), free (us), hasten (to us),  Mars!' Here nue would be the imperative of
the root of numen, that most characteristic of all Latin words for divine power; the simplex *nuo is
not otherwise attested, but adnuo is specifically used of the gods granting favour (e.g. deus annuat,
oro; Silius Italicus Punica 3:115), and the root meaning 'nod' is also present in innuo. The early
spelling ENVE for *en nue might particularly easily be read NEVE = neue in the repetitions of the
sequence after the preceding imperatival expression.
9 The passages usually quoted from Plautus Trinummus 35 and 294 to justify the meaning 'the people'
seem to me to be inadequate for the purpose. At 294 the meaning is 'the dead' as opposed to the
living, at 35 plures is contrasted with pauciores, and both uses are evidently caiques on the Greek ol
πλεονες. Such a caique in the Carmen must be highly questionable, despite Norden's advocacy of
formal Greek influence. If PLEORES is to be read as pleores = plures, it does not in my opinion
make good sense and is unlikely to be original.
9a Bréal (1881:379) surprisingly dates 'e mot qui vient après luem est arve (pour arves)', but offers no
supporting argument.
10 Still more adventurous would be to see in berber the Latin parallel to Greek θυρωρός 'door-keeper',
Homeric perhaps Cypriot θυραFωρός from Indo-European *dhwer-/dhwor-/dhur- 'door' +
*wer-/wor- 'guard, ward' (DELG s.v the development would be *dhweri-weros > *fer(i)-
wer(o)s > *ferßer > *berber (cf. above on barba). The meaning 'guardian of the gate' would fit well
with limen sali in a 'military' interpretation of the text - not that there is any reason why different
functions of the god should not be combined in even a short prayer. Bréal 1881:378 took it as a
divine name Berber parallel to Marmar.
11 It is indecipherable in the third, difficult to make out in the second, and was read as D by Ritschl in
the first; and looked at through a magnifying glass the first two look very like  on the published
photograph (cf. also the remarks by Mentz 224). However, this impression is not confirmed for the
first by the reproduction of a squeeze in the Album or by enlargements, and Professor Gordon has
kindly informed me in a letter of 25. Oct. 1987 that he and Mrs. Gordon 'have never doubted the Ρ
reading' and have no evidence to support B. He also points out that the indices to CIL 6 and 8 'show
examples of  for Ρ and vice versa, some of them between vowels'. It seems clear that, although
ADVOCAPIT is to be read at least once in the Carmen, it is simply a variant spelling of aduocabit.
The Ρ of triumpe beside Greek is therefore quite different. On this word and the noun
triump(h)us see L. Bonfante Roman triumphs and Etruscan kings JRS 60, 1970, 49-66; also in Out
of Etruria, Oxford 1981 (BAR International Series 103), 103-110 as Etruscan Influence in Early
Rome: The Latin Word Triumphus. Cf. the same author's Etruscan Life and Afterlife Warminster
(Aris and Philips) 1986:261 f., with refs.: the Romans 'celebrated their military triumphs with
Etruscan dress and music, marching to the three beat rhythm they called the tripudium or triumphus.'
(ib. 262, where the connection of an Etruscan war dance with 'the dance traditionally performed by the
Salii, the warrior priests of Rome', is consonant with Calvo's general approach to the Carmen
Amale.
12 For example, in the early legal texts: in the Twelve Tables, which originated in V , sentences like
si nox furtum faxsit, si im occisit, iure caesus esto 'if (some one) commits a theft, (and) if (the
householder) kills him, he (the thief) shall have been lawfully slain' (Table VIII: Ernout 120,
Warmington, 111:182) are normal.
The original function of the words was precisely stated by Cirilli (1913:104 echoing Bréal
1881:377): Elle [la phrase] ne fait pas partie du chant arvale, mais du dispositif. Elle doit être lue: «Il
implorera l'un après l'autre tous les Semones». This view has been developed by Lida Bernardini,
Studi Classici e Orientali 5, 1956, 82-5 (with comparative material on chants and rubrics) and by S.
Ferri 1954 in the same volume), according to who the second half of 3 and 4 were also rubrics.
13.1 experienced a not dissimilar conversion of instruction to chant nearly a quarter of a century ago. At
the Good Friday liturgy a long series of intercessions were each followed by the formula flectamus
genua, with which the priest invited the congregation to kneel and pray; after a short period of silent
prayer he said levate 'rise'. On the occasion in question the congregation used levate as a reponse to
the priest's flectamus genua, until after the third or fourth repetition he explained the meaning and use
of the word.
SELECT BIBLIOGRAPHY
Other references are given ad hoc, esp. in notes 4 and 11. Bibliographies are provided by Lommatsch and
Degrassi in CIL, Gordon Album 41 and Radke 1981 114. Cf. Paladino 12f., 195-202.
Bréal, M. 1881. Le chant des arvales. MSL 4.373-381.
Bickel, Ernst. 1940. Carmen arvale. Rheinisches Museum für Philologie 89. 28-36.
CIL I 2 Corpus Inscriptionum Latinarum. I: Inscriptiones Latinae Antiquissimae, editio altera ed. E.
Lommatsch, 1918; Fase. I, No. 2:369-70; Fase. 2 Addenda, 1931:717; Fase. 3, Addenda altera,
1943:831. [See especially the substantial Fase. 4 Addenda tertia 1986 ed. A. Degrassi (2 voll., texts
and plates) 1:855.]
Cirilli, R. 1913. Les prêtes danseurs de Rome. Paris.
DELG. Chantraine, P. 1968-80. Dictionnaire Etymologique de la langue grecque. Paris.
Dumézil, G. 1966. La religion romaine archaique. Paris.
Ernout, A. 1973. Recueil de textes latins archaiques Paris.
Ferri, S. 1954. Il Carmen Fratrum Arvalium e il metodo archeologico. Latomus 13. 390-393. (See further
Studi classici e orientali 5,1956, 87-106.)
Garcia Calvo, A. 1957. Una interpretación del carmen arval. Emerita 25. 387 ff.
Gordon, A.E. 1951. Greece and Rome 20. 85-8. [Contains text of Acta for 28th. May, 218 AD and
PLATE 3b.]
.Album of dated Latin inscriptions 4 parts in 7. Berkeley/Los Angeles 1958-65, 3. 41-46 (No. 276 =
CIL VI 2104 6-38) with PLATE 130 (squeeze).
Illustrated Introduction to Latin Epigraphy. University of California 1983:160-161 (No. 75) with
PLATE 49.
Lindsay, W.M. 1897. Handbook of Latin Inscriptions. London.
Mentz, F. 1952. Zum Carmen Arvale. KZ 70. 209-227.
CARMEN ARVALE 227
Nacinovich, M. 1933-34. Carmen arvale . 2 vol. [I have not had access to this work; it is widely agreed to
be remarkable for industry, scholarship and quite untenable conclusions ('ein Phantom' Norden 109,
Pighi 195 note 1).]
Norden, E. 1939. Das Kultlied der Arvalbrüderschaft. Aus altrömischen Priesterbüchern. 109-293. Lund.
OLD. Oxford Latin Dictionary. 1968-82.
Paladino, Ida. 1988. Fratres Arvales: storia di un collegio sacerdotale romano, (esp. 195-203.) Roma.
Palmer, L. R. 1954. The Latin Language. London.
Pighi, G. B. Review of Norden. 1941. Aevum 15. 155-161.
Pisani, V. 1975. London. Testi latini arcaici e volgari. 3rd. ed. 2-5. Turin.
. 1981. Un grecismo nel Carmen Arvale. Parola del Passato 36. 243-244.
Radke, G. 1981. Archaisches Latein. 106-114. Darmstadt.
. 1972. Antike. Romanitas 11.189-264. (239-251 on the Carmen Aruale).
Grienberger, Th. von. 1906. Das Carmen aruale. Indogermansiche Forschungen 19.140-171.
Vaanänen, V. 1981. Introduction au latin vulgaire. 3rd.ed. Paris.
Warmington, E. H. 1967. Remains of Old Latin. IV: Archaic Inscriptions. 250-253.
"nonnulla nomina in utraque lingua habent
radices, ut arbores quae in confinio natae in
utroque agro serpent"
"quae usque radices in Oscam linguam egit"
LA LINGUA Dl NUMA
MARIO NEGRI
Università di Macerata
1. La lingua di Numa*
1.1. La distinzione fra "protosabino" e "sabino", formalizzata da Giacomo Devoto in un
breve saggio apparso in "SE" 391, è di fondamentale rilievo per il problema cui sono
dedicate queste pagine. Consente essa infatti di render conto storico delle più gravi
contraddizioni, interne ed esterne, connesse con la tradizione sabina, delle quali riferisce
nel complesso il Durante2. L'articolazione interna, così istituita, che distingue una fase
arcaica, appartenente all'VIII/VII s. a.C, quindi all'età regia, rispetto all'altra, recenziore
e correlata con l'età della più antica repubblica, proietta infatti tali contraddizioni sulla
diacronia tormentata del Latium uetus, dando luogo a un tentativo di storicizzazione degli
eventi affatto assente p. es. nel Latin Language del Palmer3  negli Elements dialectaux
dell'Ernout4, dove l'identificazione "sabino" = "italico" (e, dunque, "non latino"), in una
visione che organizza su questi due poli il conflitto linguistico fra gl'Indeuropei dell'Italia
antica centro-meridionale, appare portata alle sue conseguenze estreme. La tesi che qui si
cercherà d'illustrare, ancorché in parte se ne discosti, nelle sue linee maestre s'innesta
tuttavia sulla dicotomia devotiana.
230 MARIO NEGRI
1.2. Ciò che, nella tradizione antica, è attribuito o, secondo una strategia ciclica,
attribuibile al sabino mostra una complessa e, dal punto di vista della lettura immediata,
contradditoria serie di equipollenze - subiettive e obiettive - con altre entità -  nozioni -
linguistiche, che alla loro volta intrattengono vicendevolmente relazioni diverse. Da un
lato - son cose note - il sabino par talora coincidere con la lingua dei ueteres, degli
antenati dei Romani: è il caso p. es. delle forme non rotacizzate. Dall'altro, invece,
sembra trovare nel falisco, in quanto favella latina difforme rispetto a quella urbana, il suo
partner più prossimo: è il caso difIh iniziale5. Né mancano tratti francamente italici: fra i
quali è del maggior rilievo *kw > p (o, sincronicamente, p vs lat. qu)6, e poi coincidenze
lessicali come cascus: casnar, Mamers, Leucesie ecc. Come render conto di tutto questo?
1.3. Non, come si è detto, con la semplicistica riduzione del sabino a italico, e ciò almeno
per due buone ragioni, l'una linguistica, l'altra storica. La ragione linguistica è la sicura
presenza in sabino di tratti non italici: l'esito non fricativo delle MA (o, sincronicamente,
la presenza di occlusiva contro la fricativa italica), l'alternanza fIh in sede iniziale e,
soprattutto, quell'area del lessico che è in totum latina salvo che per la marca formale
sabina: fasena, ausum, lupus (?) e così via. La ragione storica sta invece nella difficoltà,
che continuo a ritenere insuperabile, d'incontrare degli Italici in età così alta intorno a
Roma: l'orizzonte laziale, popolato da Sabini, Romani ed Etruschi, è chiaramente quello
della Roma regia nel racconto liviano. Ma neppure è lecito ridurre il sabino a mera
dimensione mitica del latino arcaico per l'obiettiva - e subiettiva - presenza dei tratti di
tipo italico, di cui sopra ho accennato. Fra i quali, per quel che vale, è da ricordare la
comune base onomastica, *sabh- di Sabini, Sabelli, Samnites, Samnium1.
1.4. Per il mio vedere, la frattura, posta dal Devoto, fra "protosabino" - la lingua dei
Sabini in età regia - e "sabino storico" - che ne rappresenta al più l'estrema diluizione
nell'area ormai italicizzata circostante a Roma - è più profonda di quanto non sia apparso
all'illustre studioso. Giacché il "protosabino" - certo dal punto di vista subiettivo, com'è
evidente p. es. in Liu. Hist. 1.13, Uerg. Aen. 7.709, Lue. ap. Cic. De Fin. 1.9, Inc. De
Praenom. 1.3. = 323 F., e particolarmente significativo P.F. 59 L. "dici mos erat
Romanis in omnibus sacrificiis precibusque: populo Romano Quiritibusque, quod est
Curensibus, quae ciuitas Sabinorum potentissima fuit" (per lafiliazionedi Roma da Reate
nel pensiero varroniano vd. F. Della Corte, L'idea della preistoria in Varrone, in
LA LINGUA DI NUMA 231
Congresso internazionale di studi varroniani, Rieti 1976, 111 ss.), ma son convinto non
meno da quello obiettivo - appartiene alla storia interna del latino di Roma. Solo si tratta
di un ramo non dominante, e la corrente del suo apporto si affievolisce sempre più nel
tempo, e in età tardo-repubblicana e classica par sopravvivere appena, e però non più a
Roma ma, almeno per quanto è giunto a noi, a Faleri.
1.5. Dove, in verità (vd. quanto detto nella n. 5), s'intreccia con tendenze nuove, e con
una diversa dimensione storica in cui s'inscrivono i rapporti e i conflitti linguistici con
l'italico. Come se, con l'esser divenuta la città metropoli, il rapporto arcaico città-
campagna si fosse arricchito di un terzo livello, quello del suburbio - l'ambiente delle
Atellane - di modo che nei nuovi apporti non è più sentita una valenza arcaica, bensì
piuttosto volgare, com'è p.es. il caso del salapūtium 'cazzabubbolo' di Catullo, volgare e
per il significato (cf. prae-pūtium) e per l'aspetto formale, che ha epentesi, e quindi marca
osca,  del sīfilāre di Nonio9.
1.6.1 caratteri e l'autonomia del sabino rispetto all'osco e d'altronde al latino possono
solo parzialmente essere posti in prospettiva genetica (il sabino è italico? Rappresenta
davvero, giusta la dottrina del Devoto, il versante linguistico dell'avanzata nel Lazio
degl'Italici inumatori - Numa inumato in Plut. Num. 22 - qui interagenti con i
"Protolatini" pur essi inumatori e con gli Etruschi incineratoli, di modo che di qui origina
la triade Titienses, Ramnes, Luceres, alle origini di Roma?, cf. p. es. Uarr. LL 5.55),
bensì più utilmente in prospettiva interferenziale. Se pure il sabino fosse stato - né lo
escludo - originariamente una favella italica (del che parrebbero far fede quelle notizie che
riconducono a origine sabina ethnē italici, come p. es. Plin. NH 3.110 "[Picentes] orti
sunt a Sabinis uoto uere sacro" (vd. anche Fest. 424 L. = P.F. 425 L. "Sacrani appellati
sunt Reate orti, qui ex Septimontio Ligures Siculosque exegerunt; dicti Sacrani quod uere
sacro sint nati"),  Uarr. LL 7.29 "ab Sabinis orti Samnites...", Strab. 5.4.2
P.F. 437 L. "Samnites ab hastis appellati sunt,
quas Graeci appellant...siue a colle S amnio, ubi ex Sabinis aduentantes
consederunt", e così via), pure all'età documentaria questi caratteri italici paiono
indistricabilmente avvinti a caratteri di tipo latino, com'è d'altronde vero l'inverso, che il
latino più arcaico ha già interagito col sabino: lupus ma tesqua, con esiti per così dire
incrociati. In questa prospettiva, credo, possono avviarsi a soluzione non poche di quelle
232 MARIO NEGRI
contraddizioni, di cui si diceva, emergenti dalla documentazione, fra le quali è simbolica

quella Saflni: Sabini. Ciò che importa è che, per il mio vedere, quanto si riesce a
decantare da quella materia nuova che è già il prodotto di reazioni linguistiche pregresse
ha una sua identità, diversa da quella paleolatina e pure diversa da quella paleoosca. Di
fatto, l'azione dei contatti "speciali" col latino su di una lingua forse -  probabilmente -
italica10 ha dato luogo a un risultato che, almeno ai nostri occhi, appare nuovo rispetto
alle sue componenti. Questo chiamo qui la lingua di Numa.
2. Labiovelari
2.1. L'esito labiale delle labiovelari in sabino è mostrato da argomenti del tutto saldi. In
tal modo, applicando la strategia ciclica di cui già si è fatto cenno, potremo considerare la
presenza di labiale < labiovelare (o, già si è detto, di labiale vs. labiovelare) alla stregua di
argomento indiziario per sospettare la possibile origine sabina della parola in cui questa
appare. E', come appresso vedremo, il caso di lupus.
2.2. Il nomen del re Numa, Pompilius, di suo padre Pompiusi/Pompo e di sua figlia
Pompilia (Plut. Num. 21), corrispondono a lat. Quin(c)tilius, Quin(c)tius1.
2.3. L'aspetto formale del nome del 'lupo', e dei derivati d'ambito sacrale Lupercus,
Lupercãlês ecc., è estraneo alla diacronia del latino (ci attenderemmo*luquos, -uus, come
equos, -uus)2. DELL, s.u. lupus: "Le p latin représente la labio-vélaire indo-européenne,
la forme venant sans doute de parlers osco-ombriens. Comme bōs, ce serait un de ces
mots sabins qui se sont introduits dans la langue de Rome". Nega ciò il Pisani3, giacché
"per lupus i Sabini dicevano anticamente hirpus", cfr. P.F. 106 M. 228 L.: "Irpini
appellati nomine lupi, quem hirpum dicunt Samnites", e Serv. Aen. XI 785: "lupi
Sabinorum lingua hirpi vocantur"... sì che "è probabile che ci troviamo di fronte a un
tabù linguistico, riguardante un animale così pericoloso per le greggi e per il bestiame
minuto... Il p latino è... la sostituzione del suono oscoumbro che si sapeva corrispondere
al proprio qu..., in modo da ottenere la desiderata deformazione del nome infausto". E'
evidente che tali argomentazioni sono viziate dalla prospettiva, per me erronea,
dell'identità storico-linguistica fra sabino e italico  oscoumbro, intese queste come aree
linguistiche caratterizzate in prima istanza negativamente rispetto al latino di Roma da
certe marche formali (-ƒ- per -b-, p per qu, ecc.) e, in via di principio, indistinguibili al
loro interno: è l'italico nebuloso da cui qui, e più volte precedentemente4, cerchiamo di
estrarre i frustuli del sabino. Di modo, che l'appartenenza di hirpus all'osco non solo non
osta all'attribuzione di lupus al sabino, ma anzi in qualche modo la conforta, giacché, per
così dire, elimina un pretendente - l'osco appunto - a tale attribuzione. Che hirpus sia di
tradizione osca (sannita) è fuor di dubbio. Vd. p.es. Strab. 5.4.12
P.F. 93 L. "Irpini appellati nomine lupi, quem
5
irpum dicunt Samnites", ecc. . Peraltro, il terionimo è presente e come toponimo e come
etnico in terra osca. Di modo, che il von Planta, a proposito della notizia di Servio sopra
citate6, avanza la possibilità di sostituire a "Sabinorum" "Samnitium"7. Ma forse, come
vedremo, questa violenza alla tradizione serviana potrebbe essere, senza costo eccessivo,
evitata.
Non sarà ora inutile affrontare il problema rileggendo la tradizione antica incentrata
sui Lupercãlia. Il termine, già in antico oscuro8, e certo solo paretimologicamente da
riconnettersi con arceō (Sem. ad Aen. 8.343 «et Euander deo gentis suae sacrauit locum
et nominauit lupercal, quod praesidio ipsius nominis lupi a pecudibus arcerentur»), trova
ora una brillante etimologia nel quadro nuovo, istituito dal Peruzzi, dei rapporti
protostorici fra la Grecia micenea e il Lazio preromano9. Di modo, che Lupercālēs altro
non sarebbe che l'adattamento in latino di un che varrebbe 'Arcadi-lupi'
 'Arcadi dei lupi'. Il qual culto rappresenterebbe l'introduzione di quello di Πάν
sul Palatino a opera di Evandro10. L'aspetto anomalo, costituito dall'uscita -ālēs
rendente -adēs, si spiegherebbe non invocando la cosiddetta '7 sabina", per cui vd. 4.,
bensì perché" un composto greco *wlukw-arkădes,... nella tradizione linguistica latina
che ha lupus, doveva essere adattato come
(donde poi lupercālēs per effetto dell'accento iniziale)". Tale resa sarebbe stata in qualche
modo necessaria e tale da non doversi "spiegare con l'alternanza latina dll..., che del resto
il più delle volte può attribuirsi a falsa analogia. Infatti, se si escludono tardi grecismi,
non esistono nomi latini in -ădi-,..11 Sicché *wlukwarkădes doveva essere assimilato ai
nomi... in -āli- plur. -ālēs (= italico -āli-: osco fertalis, umbro uerfale ecc.)"12. In effetti,
l'attribuzione del culto lupercale all'ambiente sabino va contro al dato mitografico, che
connette coerentemente tale culto con Romolo e Remo (vd. p.es. Ou. Fast. 2.361-6). E
tuttavia non mancano riferimenti, ancorché indiretti, al sabino: Uarr. LL 5.85, 6.13
"Lupercalia dicta, quod in Lupercali Luperci sacra faciunt. Rex cum ferias menstruas
234 MARIO NEGRI
Nords Februariis edicit, hunc diem februatum appellat; februm Sabini purgamentum, et id
in sacris nostris uerbum <non ignotum: nam pellem capri, cuius de loro caeduntur puellae
Lupercalibus, ueteres februm uocabant>, et Lupercalia Februatio, ut in Antiquitatum
libris demonstraui". La medesima connessione Lupercali-: febru(u)m in Cens. 22.13.14
"est februum quidquid piat purgatque et februamenta purgamenta, item februare purgare
est et purum facere. Februum autem non idem usquequaque dicitur; nam aliter in aliis
sacris februatur, hoc est purgatur. In hoc autem mense Lupercalibus, cum Roma
lustratur, salem calidum ferunt, quod februum appellant, unde dies Lupercalium propre
februatus et ab eo porro mensis Februarius uocitatur". D'altronde, l'appartenenza dei riti
lustrali, in cui verte febru(u)m, all'ambiente culturale - se non linguistico - sabino è in
Macr. Sat. 1.13.3 "lustrari ... eo mense ciuitatem necesse erat, quo statuii [Numa] ut
iusta Dis Manibus soluerentur". Inoltre P.F. 75-6 L. "Februarius mensis dictus, quod
turn, id est extremo mense anni, populus februaretur, id est lustraretur ac purgaretur, uel
a Iunone Februata, quam alii Februalem, Romani Februlim uocant, quod ipsi eo mense
sacra fiebant, eiusque feriae erant Lupercalia, quo die mulieres februabantur a lupercis
amiculo Iunonis, id est pelle caprina; quam ob causam is quoque dies Februatus
appellabatur... id uero, quod purgatur, dicitur februatum", Uarr. LL 6.34, P.F. 49 L.,
Uarr. ap. August. CD. 18.17, ecc. Si ricordi infine che, giusta Ou. Fast. 2.267-8, Liu.
Hist. 1.20, fu Numa stesso a istituire l'ufficio del Flamen Diālis, cui competeva il
sacrificio priscō more13.
Epperò, posto che non sia inammissibile l'idea che Numa abbia organizzato, e dato i
nomi, a un complesso cultuale istituito da altri, e che dunque effettivamente il nome dei
Lupercali sia di origine sabina, e che così dunque sia possibile, quandanche ovviamente
in via d'ipotesi, dare un'identità positiva alla fonte non latina del nome del 'lupo', resta
tuttavia il problema di conciliare ciò con la notizia serviana, di cui sopra si è detto, ossia
che i Sabini per 'lupo' non dicevano lupus bensì hirpus (il qual termine, se la notizia
fosse tenuta per degna di fede, apparterrebbe al novero delle isoglosse lessicali sabino-
osche, del tipo cascus: casnar). Invero, una soluzione, che certo non s'impone, e che
tuttavia continuo a giudicare preferibile rispetto alla violenza suggerita dal von Planta14,
potrebb'essere questa: secondo una rilevante schiera di notizie15 i Luperci, prima di
utilizzare il fango, usavano coprirsi, per realizzare la loro metamorfosi rituale, con le pelli
dei capri sacrificati, elafebruātio delle fanciulle consisteva nello sferzarle con pelli di
capro appunto (Yamiculum Iunonis) e non di lupo. Così p.es. Seru. ad Aen. 8.343
"Remum et Romulum ante urbem conditam lupercalia celebrasse eo quod quodam
tempore nuntiatum illis sit latrones pecus illorum abigere: illos togis positis cucurrisse
caesisque obuiis pecus recuperasse: id in morem uersum, ut hodieque nudi currant. Non
nulli propter sterilitatem hoc sacrum dicunt a Romulo constitutum, ideoque et puellae de
loro capri caeduntur, ut careant sterilitate et fecundae sint: nam pellem ipsam capri ueteres
februm uocabant". Non si saranno così intrecciati nella notizia serviana sul nome sabino
del 'lupo' il nome romano-sabino del 'capro', hircusí/fircus, e quello italico del 'lupo',
hirpus, in modo tale da creare un "sabino" hirpus 'lupo', all'interno  comunque in
connessione coi riti lupercali, dove il lupo e il capro, predone e preda appaiono, e ab
antiquo, contaminati, tanto che i sacerdoti devoti di Lupercus, il dio-lupo qui lupos arcet,
realizzavano la loro teriomorfia lupina pellibus caprinis cincti?
Se tutto ciò è vero, vi potremmo scorgere non prove ma indizi linguistici e storico-
antiquari a confermare l'impressione antica della sabinità del lupo, i cui echi sono già
forse in Orazio, Od. 1.22.9-12.
2.4. In "PdP" 112, 29 ss. il Peruzzi ha affermato l'origine sabina, per la presenza di p <
*kw, di simpulum P.F. 455 L. "uas paruulum non dissimile cyatho, quo uinum in
sacrificiis libabatur; unde et mulieres rebus diuinis deditae simpulatrices". Vd. anche
simpuuium16. Parimenti sabino potrebbe essere popa, se connesso con coquö, popïna
ecc., secondo ritiene il Devoto, Storia, 82 (ma cf. etr. pupa ecc.? DELL, s.u.). Inoltre
brütus, ove ne venga accolta la parentela (omologia) con grauis, conclamata in P.F. 28 L.
"brutum antiqui grauem dicebant", vd. DELL, s.u.
2.5. Fest. 496 L. "Tarpeiae esse effigiem ita appellari putant quidam in aede louis
Metellinae, eius uidelicet in memoriam uirginis, quae pacta a Sabinis hostibus ea, quae in
sinistris manibus haberent, ut sibi darent, intro miserit eos cum rege Tatio; qui postea in
pace facienda cauerit a Romulo, ut ea Sabinis semper pateret". Fest. 464 L. "...] xum
Tarpeium appell[...]tis, qui ob sepultam Ta<rpeiam> ... eum montem Sabinis pro [...
...]nominatus est; uel i[... L. Tarpeius Romulo [ ]tas uirgines aduersa... [.". Inoltre
Uarr. LL 5.41, Liu. Hist. 1.11. Vd., per le vicende del colle Tarpeo, occupato da Tito
Tazio e poi denominato Campidoglio (! vd. 4.3), Della Corte, art. cit., e R. Gelsomino,
Warrone e i sette colli di Roma, ib., II, 379 ss.
236 MARIO NEGRI
Tarp- rappresenta verisímilmente la resa sabina della base Tarqu- di Tarquitiae Scalae,
Fest. 496 L., Tarquinius ecc.: "[Il nome] della rupe Tarpeia in Campidoglio mostra un
suffisso sabino e un'alterazione sabina della radice mediante la introduzione di un p al
posto di q: cioè la t r a s p o s i z i o n e di una parola costituita da una radice iarq e da un
qualsiasi suffisso di derivazione aggettivale. La forma che così nasce può dunque
identificarsi con Tarquinia se vogliamo dare una trascrizione latina qualsiasi, con Tarchna
 simili se preferiamo una ipotetica forma etnisca. La forma anteriore, con q, si sarebbe
conservata nella formula Tarquitiae scalae. Storicamente ci rendiamo ben conto dell'una e
dell'altra forma pensando alla dinastia di re etruschi che hanno regnato durante il VI
secolo; dei quali restò il tempio di Giove Capitolino. Da quel tempo dovrebbe derivare la
forma di "monte Tarquinio"; che poi sarebbe stato tradotto in forma sabina per esempio al
tempo dell'occupazione da parte del sabino Appio Erdonio"17.
2.6. Contrastano tuttavia con queste unanimi testimonianze, il cui spessore mi pare non
possa revocarsi in dubbio, di sabino p < *kw alcune forme, che appresso cito. Ps. Acr.
ad . . 1.14.19. "t(h)esqua loca deserta et difficilia lingua Sabinorum sic dicuntur",
Is. Or. 15.12.3 L. "tescua quidam putant esse tuguria, quidam loca praerupta et aspera",
tesca in Uarr. LL 7.10-11 "quod addit templa ut si<n>t 'tesca', aiunt sancta esse qui
glossas scripserunt. Id est falsum... loca quaedam agrestia, quae alicuius dei sunt,
dicuntur tesca ...Quare h<a>ec quo<d>tesca dixit [seil. Accius], non errauit, neque ideo
quod sancta, sed quod ubi mysteria fiunt attuentur, tuesca dicta", P.F. 489 L. "tesca loca
augurio designata. Cicero aspera ait esse et difficilia", Acc. 554 R 3 , ecc. (Yiter
dovrebb'essere tesquos > teseos, su cui poi l'intera flessione, come deus) 18. Uarr. LL
6.57 "hinc dicuntur eloqui et reloqui19 in fanis Sabinis, e cella dei qui loquuntur". Uarr.
LL 5.51 "collis Quirinalis, <quod ibi> Quirini fanum. Sunt qui a Quiritibus, qui con
Tatio Curibus uenerunt [ad] Roma<m>, quod ibi habuerint castra". Restano peraltro
oscuri - ma credo paretimologici - i rapporti fra Quintes ecc. e curis (qui-DELL s.u.l.)
'hasta': Ou. Fast. 2.475 "proxima lux uacua est, at tertia dieta Quirino: /qui tenet hoc
nomen, Romulus ante fuit, /siue quod hasta curis priscis est dieta Sabinis / (bellicus a telo
uenit in astra deus), /siue suum regi nomen posuere Quirites, /seu quia Romanis iunxerat
ille Cures", P. F. 43 L. "curis est Sabine hasta. Unde Romulus Quirinus, quia earn
ferebat, est dictus; et Romani a Quirino Quirites dicuntur. Quidam eum dictum putant a
Curibus, quae fuit urbs opulentissima Sabinorum", Fest. 304 L. "Quirinalia mense
Februario dies, quo Quirini fiunt sacra: idem stultorum feriae appellantur, [antequam in
eum commigrarent fere Sabini Curibus uenientes post foedus inter Romulum et Tatium]
quod qui dem suorum fornacalium sacrorumt cognominanti-, eo potissimum rem
diuinam faciunt", ib. (= P.F. 305 L.) "Quirinalis collis, qui nunc dicitur, olim Agonus
appellabatur, ante quam in eum commigrarent fere Sabini Curibus uenientes post foedus
inter Romulum et Tatium ictum ...", ib. "quirina tribus a Curensibus Sabinis
appellationem uidetur traxisse", Uarr. LL 6.68 "quiritare dicitur is qui Quiritum fidem
damans inplorat. Quirites a Curensibus; ab is cum Tatio rege in societatem uenerunt
ciuitatis. Ut quiritare urbanorum sic iubilare rusticorum", P. F. 55 L., Uarr. ap. Dion.
Hal. 2.48.4
2.7. Sull'esito labiale di *kw mi pare che non sia dubbio. Pompilius (il suffisso pare fra
l'altro caratteristico dell'ambiente sabino,  quantomeno a questo non estraneo, cf. il
nome della moglie sabina di Romolo, Hersilia)20 attesta per il mio vedere il fatto in modo
perentorio, e così Pompius/Pompō. Giacché continua a parermi difficile immaginare degli
Osci, e degli altri Italici, nelle vicende connesse con la veneranda antichità in cui si
proiettano le origini di Roma, il ritrovare una forma con -p- in Tarpēia (il cui
coinvolgimento con l'ambiente sabino è peraltro conclamato dalla tradizione - il padre Sp.
Tarpēius, comandante della rocca, sarà secondario sul nome della figlia?-) consente,
secondo verisimiglianza, di utilizzare anche questa forma per illustrare l'esito labiale (cf.
Tarquinius: va da sé che qui più che di esito si tratta di corrispondenza, ma la cosa è di
modesto rilievo in questa sede). Non concordo quindi con l'idea del Devoto, secondo cui
la sabinizzazione di Tarpēia apparterrebbe a una fase seriore rispetto all'età romulea.
Quanto a lupus, i motivi della possibile attribuzione al sabino sono stati diffusamente
esposti, riferendo della documentazione. Contrastano con sab. ρ < *kw non tanto eloqui
e reloqui, di cui in verità le fonti non affermano in alcun modo la qualità sabina, ma solo
la natura di verbi connessi con la liturgia sabina, né Quirinalis ecc., della cui sabinità
nessuno parla, ma solo dell'esser stato il luogo d'insediamento dei Sabini provenienti da
Cures, né Quirīnus, solo paretimologicamente accostato a curis 'hasta', quanto piuttosto
tesqualtesca11. E' la prima delle contraddizioni interne alla tradizione sabina in cui
c'imbattiamo, e la cui ragione, secondo il mio vedere, è inquadrabile soltanto in chiave
interferenziale.
238 MARIO NEGRI
3. Medie aspirate
3.1.II problema del trattamento delle MA in sabino appartiene con evidenza alla maggior
questione delle sorti di questi suoni nell'Italia antica, che qui, peraltro, non potrà essere
neppur accennata. Di fatto, una celebre notizia festina, riferita ad alpus, ci conserva la
testimonianza più salda di uno stadio T(A) < MA. Personalmente1 non credo che un
"italico" (qui inteso come somma di latino, venetico, oscoumbro [e sabino!]) MA > TA si
sia mai avverato, ma la questione non ha qui grande rilievo. Quanto invece mi preme di
mostrare è la chiara autonomia, in questo caso, del sabino rispetto all'osco (e, ancorché
meno coerentemente, all'umbro).
3.2. P. F. 4 L. "album, quod nos dicimus, a Graeco, quod est est appellatum.
Sabini tarnen alpum dixerunt. Unde credi potest, nomen Alpium a candore niuium
uocitatum". Se questa notizia festina che, comunque sia, mi pare non possa esser
semplicemente posta in non cale, dovesse essere, pur con le necessarie cautele, tuttavia in
qualche modo accolta 2, le si potrebbero aggiungere altre testimonianze, quali più, quali
meno significative, mostranti MA > T(A) in ambiente italico e, ora potremmo
congetturare, sabino (giacché in nessun'altra lingua dell'Italia antica3 un fatto del genere
par testimoniato). Di tali documenti danno ampio resoconto, pur da punti di vista
differenti, il Bonfante e il Campanile4. Qui solo riferisco dei meno disperati, e cioè
rutilus, Uarr. LL 7.83 "...aurei ... rutili, et inde eti<a>m mulieres ualde rufae rutilae
dictae", se veramente appartiene alla famiglia di ruber, rūfus, di modo che avremmo qui,
nelle parole per 'rosso', testimonianza completa dei possibili esiti italici, e cioè ruber
(*rudh-), rōblgō, -us, rōbus P.F. 325 L. "robum rubro colore et quasi rufo significari, ut
bouem quoque rustici appellant, manifestum est", e d'altronde (??) Rudiae (con sonora) e
(??) raudus (|o|u) Uarr. LL 5.163 "aes raudus dictum", P. F. 339.L., Fest. 320-2 L. =
P.F. 321 L.; rūfus, Rufuli, vd. oltre, con fricativa sorda come in ou.; rutilus, Rutuli (??)
con T(A?). Fest. 320 L. = P. F. 321 L. "rutilum (mss. -iura) rufum significat. Cuius
coloris studiosae etiam antiquae mulieres fuerunt; unde traxerunt cognomina Rutilias, ut
indicat frEquenter Afranius", Fest. 358 L. "rutilae canes, id est non procul a rubro colore,
immolantur." Avremmo così:
*reudh-/roudh-/rudh-
rub-lrōb- (con M < MA, e *dh > b condizionato in latino.);
rud- (con M < MA, ma non *dh > b: mess. (?), ven.);
ruf- (con fric, sorda < MA: osco(umbro))
rut- (con T(A?) < MA: sabino?).
Fest. 316 L. = P.F. 317 L. "Rufuli appellabantur tribuni militum a consule facti, non a
populo; de quorum iure quod Rutilius Rufus legem tulerit, Rufuli, ac post Rutuli sunt
uocati". Credo che l'idea del Bonfante5, di tenere per sabina la forma con -ƒ- (Rufuli), si
fondi solo sulla cronologia relativa dei due omologi, nella notizia festina acclarata dal
post. Epperò non mi sento di accoglierla. Quanto a Liternum ('città libera'??, cf. lat. līber
(Liternum: líber come rutilus: ruber), ecc., si tratta di testimonianze di ben lieve spessore.
Di assai maggior rilievo sarebbe invece - ma il condizionale è d'obbligo visto lo spessore
della testimonianza - scirpus Fest. 444 L. se questo trovasse un omologo "italico"
nell'oscuro skerfs dell'iscrizioncella arcaica da Poggio Sommavilla (vd. 1.2), vd.
Prosdocimi in Storia, LXXXVI ss., inoltre A.A.V.V. in Civiltà arcaica dei Sabini II,
Roma 1974, 45 ss, e Cristofani, ib., III, Roma 1977, 97 ss. Ne scrive acutamente la
Marinetti in "SE" 49, 124 n. 20, (e vd. anche «SE» 51, 167-70), cogliendo se non m'
inganno l'intima contraddizione di una "sabinità di Roma (o a Roma?)" con -p-, laddove
il sabino "autentico" (che per la Marinetti è francamente italico) ha -ƒ-, secondo appunto il
modo italico. Quanto a me, ancorché meno certo della corrispondenza scirpus: skerf-, se
questa fosse vera, non potrei che trovarvi, come nei Safìni sudpiceni, una significativa
conferma delle linee di storia linguistica che ipotizzo per il Latium uetus. Per Sapinia
(tribus) vd. Prosdocimi, in «Dialoghi di Archeologia» 1987, 57.
3.3. Ma contro l'idea, che si regge invero su di un solo caso, e tuttavia, per il mio vedere,
di notevole rilevanza, di MA > T(A?) in sabino, e cioè alpus, sembrerebbero parlare
alcuni fatti. In primo luogo teba (e fundus tebanus?? Vd. Schulze, Gesch. lat. Eigenn.,
531), sabino in Uarr. RR 3.1.6, sembrerebbe mostrare un esito di tipo latino, e cioè MA
> M. Uarr. RR 3.1.6. "lingua prisca et in Graecia Aeolis Boeoti sine adflatu uocant Collis
t[h]ebas; et in Sabinis, quo a Graecia uenerunt Pelasgi, etiam nunc ita dicunt, cuius
uestigium in agro Sabino uia Salaria non longe a Re<a>te, miliarius cliuus cum appellatur
240 MARIO NEGRI
t[h]ebae". Ove questa forma sia posta a confronto, il che non pare illegittimo, con tifata,
P.F. 503 L. "tifata iliceta. Romae autem Tifata Curia. Tifata etiam locus iuxta Capuam", e
credo anche con calabr. tifa 'zolla'6, avremmo, anche in mancanza di etimologia
indeuropea (giacché a questa, dal punto di vista della diacronia, appartengono le MA),
comunque riprodotto il rapporto b: ƒ dei ben noti7 omologi latini e italici. Di Loebāsius,
che probabilmente non appartiene al problema di cui qui è questione, si dirà oltre (vd.
6.3). Quanto a Tibur se, secondo la dottrina del Kretschmer, va con Tifernum8, e posto
che sia sabino, la situazioneriprodurrebbepuntualmente quella di teba : tifata.
Inoltre, ma solo possibilmente, nebrundines: P.F. 157 L;"Nefrendes arietes
dixerunt, quod dentibus frendere non possint. Alii dicunt nefrendes infantes esse nondum
frendentes, id est frangentes. Liuius (trag. 38): "Quem ego nefrendem alui lacteam
inmulgens opem". Sunt qui nefrendes testículos dici putent, quos Lanuuini appellant
nebrundines, Graeci Praenestini nefrones". Inoltre Fest. 342-4 L.: "Rienes
quos nunc uocamus, antiqui nefrundines appellabant, quia Graeci eos uocant.
Plautus in Satyrione (113): "Male tibi euenisse uideo; glaber erat tamquam rien" = P.F.
343 L. "Rienes antiqui uocabant nefrundines, quia Graeci eos νεφρούς dicunt. Plautus
(frag. 113): "Glaber erat tamquam rien".
La glossa "lanuvina" nebrundines ha posto, e del resto pone, problemi, e cioè:
1) quale è la fonte ie. di "lanuvino" -br- prenestino -ƒr- = gr. - φρ -? 2) In che rapporti
stanno le forme nefrundines:nebrundines?3) Qual è infine lo spessore dell'attribuzione al
dialetto di Lanuvio, di cui pochissimo (ma almeno un'altra cosa interessante) sappiamo?
Per quel che attiene alla prima questione, abbiamo da un lato il greco, che potrebbe
risalire tanto a *-bhr- quanto a *-gwhr-; alcune forme germaniche che chiedono invece *-
gwhr- {DELG, s.u. e le tre forme italiche, nefrones, nefrundines e nebrundines
che, dal punto di vista delle continuazioni dirette, cioè trattandole in prospettiva
rigidamente verticale, chiederebbero, almeno a prima vista, *-hh- (o, forse, *-dh-).
Infatti, considerando il rapporto -b-: -ƒ- alla stregua di quello fra trattamento latino e
trattamento italico, in latino la -b- non può - sempre dal punto di vista dei rapporti verticali
- essere altro che il riflesso di una MA labiale (o dentale). Per l'impossibilità di "lat." -br-
< *gwhr- vd. Pariente, "EM" 11, 1943, 338 ss., e la bibliografia lì citata, anche se, va
detto, una volta ammesso un latino fric. sorda < TA < MA (secondo la classica teoria
dell'Ascoli), l'analogia di -br- < -sr-, p. es. in funebris < *funes-ri-, non renderebbe
inverisimile -br- < -fr-. Ma contro la teoria ascoliana il Pisani ha portato, se non
m'inganno, argomenti decisivi: vd. Le unità intermedie, di chi scrive, alle pp. 32 ss. Ma,
evidentemente, la soluzione del problema eccede la dimensione verticale.
Assumendo in via d'ipotesi che esista un comun denominatore ie. per queste parole
dei 'reni'  dei 'testicoli', la soluzione piú economica è quella che pone un archetipo con
MA lbv., che consente di non separare, contro verisimiglianza, le forme germaniche dal
greco: si noterà che le perplessità del DELG sono dovute esclusivamente alle forme
italiche. Ora, un gruppo *-gwhr- in italico dovrebbe continuare legittimamente con -fr-,
cf. u. vufru T.I. II b 21 = LIA 60, se < *wogwh-. Col qual esito si accorda il prenestino
nefrõnes e  "arcaico" (ma credo solo come tipo lessicale rispetto a rien) nefrundines.
Ancorché diversamente verisimili possono essere qui ricordate le ipotesi del Pisani
(p.es. in Storia della lingua latina, 98 e 163), secondo il quale la b di nebrundines
rappresenterebbe l'estrema memoria in una favella italica - ché a questo ambito lo
studioso milanese attribuisce la glossa di Lanuvio - della fase b < *gwh, e del Devoto,
Antichi italici, 94, che, vedendo in nebrundines una latinizzazione di nefrundines, giusta
il consueto rapporto b: f, caratterizzante il latino e per converso l'italico, offre della nostra
glossa una spiegazione certo più che possibile. Quanto all'idea del Pariente, "EM" 11,
338 ss., essa è inficiata dal falso presupposto secondo il quale deve negarsi il sabino lbv.
>lab.
Io qui mi limito a rilevare, senza però azzardare un'attribuzione, che la glossa di
Lanuvio, posto che effettivamente risalga verticalmente a un archetipo con *-gwh-, ha
l'aspetto che ci attenderemmo in sabino. Del resto la stessa solidarietà - comunque sia da
intendersi - lanuvino-sabina sembrerebbe manifestarsi nella parola mānus per 'bonus':
cfr. Macr. Sat. 1.3.13 "Lanuvini mane pro bono dicunt" e d'altronde le attribuzioni di m.
ai Carmina Saliaria in Fest. 132 L. "manuos in carminibus saliaribus Aelius Stilo
significare ait bonos", inoltre P.F. 109 L. 'Matrem Matutam antiqui ob bonitatem
appellabant..." ib. "in carmine Saliari Cerus manus intellegitur creator bonus", e
d'altronde Uarr. LL 6.4 "bonum antiqui dicebant manum" - vd. però 11.2, n.l 9 .
3.4. Ancora va ricordata l'esigua schiera dei termini che mostrerebbero -ƒ- (da MA 
rispetto a lat. M), secondo il modo italico. Invero, il più10 degno di attenzione par essere
Farfarus. Seru, ad Aen. 7.715 "Fabarim...quem dicit...per Sabinos transit et Farfarus
dicitur: Plautus dissipabo te tamquam folia Farfari"11. La testimonianza serviana è
242 MARIO NEGRI
dunque esplicita nell'affermare che, in territorio sabino, l'omologo di un termine (latino?)

con -b- aveva un'-/- e ciò, fuor di dubbio, contrasta, non diversamente da come
contrastano Pompilius e d'altronde tesqua, con quanto si è fin qui visto. Ma, qui, forse la
contraddizione potrebbe risolversi già all'interno della documentazione: giacché vi si
afferma che, ai tempi in cui Servio scrive, quello che nell'età di Enea e di Evandro si
chiamava Fabaris ora è detto Farfarus. Avremmo qui null'altro - come, del resto, nel
rapporto Sabini : Safìni - che la testimonianza della diluizione del sabino in quella
dimensione linguistica italica, volgare e suburbana, accerchiante Roma fin, credo, dal
V/IV s. a.C. Di modo che l'esito italico - ed è il fatto rilevante, a ben vedere - con
fricativa sorda pare potersi escludere per il sabino12. Quanto al contrasto di cui sono
protagonisti alpus e d'altronde teba, mi pare legittimamente interpretabile allá stessa
stregua di quello fra ρ e qu di Pompilius : tesqua.
4. Lo scambio dll
4.1. Sullo scambio fra d e / all'interno del lessico latino - il fenomeno comunemente noto
come '7 sabina"1 - converge una notevole massa documentaria, diretta e indiretta. Gioverà
tuttavia rilevare che solo in due casi abbiamo attribuzione esplicita del fenomeno
all'ambiente sabino. Di modo, che pare a me non del tutto condivisibile la sicurezza con
cui il Conway per quanto so primamente2, e altri dopo di lui3, hanno fatto propria l'idea
del sabinismo di l/d, che credo in prima istanza essere appartenuta a Varrone (o,
quantomeno, che ci è documentata più anticamente in Vairone).
4.2. Uarr. LL 5.123 "item dictae lepestae, quae etiam nunc in diebus sacris Sabinis uasa
uinaria in mensa deorum sunt posita; apud antiquos scriptores Graecos inueni appellari
poculi genus quare uel inde radices in agrum Sabinum et Romanum sunt
profectae"; Uarr. ap. Non. 877 L. "lepistae. Uarro de Uita Populi Romani lib. I: "ut fere
habent aeneum illi, qui uenditant oleum. Lepistae etiamnunc Sabinorum fanis
pauperioribus plerisque aut fictiles sunt aut aeneae" "; P. F. 102 L. "lepista genus uasis
aquarii". Posto che veramente, come si afferma in DELL s. u. lepista, avessimo qui a che
fare con un "emprunt oral au grec allora l'invenzione varroniana (ché tale
sarebbe in questa prospettiva) di un greco (così in LEW, s. . lepista)
parrebbe chiedere d'essere spiegata all'interno di una precisa opinione di Varrone in
merito a questi fatti, nel senso di "sentire" un rapporto fra l'occlusiva del greco e la
liquida del sabino. E tuttavia non va taciuto che mancano argomenti saldi a ipotizzare
quest'intervento varroniano: giacché, se da un lato la presenza di
potrebb'essere indizio a favore della sua autenticità, d'altra parte proprio e
e d'altronde avrebbero potuto fornire a Vairone il punto di
partenza per la sua creazione.
4.3. Arnob. Adu. Nat. 3.38 "Nouensiles Piso deos esse credit nouem in Sabinis apud
Trebiam constituios", Liu. Hist. 8.9. "Lares, Diui Nouensiles, Di Indigetes". In Uarr.
LL 5.74 la forma con d: "Nouensides a Sabinis". In Mar. Uict. 6.26 K. le due varianti:
"Nouensiles siue per L siue per D scribendum. Communionem enim habuit L littera cum
D apud antiquos, ut dinguam et linguam et dacrimis et lacrimis et Kapitodium et
Kapitolium et sella a sede et olere ab odore: est et communio cum Graecis, nos lacrimae,
illi Nouensiles autem, quos Graeci
post nouendii a considendo, id est eadem sede praediti", Uarr. ap. Arnob. 3.38
"(Nouensiles) nouenarium numerum tradii Uarro, quod in mouendis rebus potentissimus
semper habeatur et maximus".
4.4. Come già in quest'ultimo passo, abbiamo inoltre non poche attribuzioni di forme con
/ in luogo di d attesa (e, ma più raramente, l'inverso) ai ueteres. P. F. 41 L. "casilam
antiqui pro casside ponebant"; Uarr RR 3.9.19 "ad hanc rem electis maximis gallinis, nec
continuo his, quas Melicas appellant falso, quod antiqui, ut Thetim Thelim dicebant, sic
Medicam Melicam uocabant"; P.F. I l l L. "Melicae gallinae, quod in Media id genus
auium corporis amplissimi fiat L littera pro D substituta"; Fest. 64 L. "delicare ponebant
pro dedicare", P.F. 61 L. "delicata dicebant dis consecrata, quae nunc dedicata. Unde
adhuc manet delicatus, quasi lusui dicatus", Ps. Plac. 4.59.4 L. "delica[ue]re: déferre;
quod et dedicare dicebant pro commertio litterarum"; Fest. 460 L. "seliquastra sedilia
antiqui generis appellantur, D littera in L conuersa; ut etiam in sella factum est, et
subsellio, et solio, quae non minus a sedendo dicta sunt", P.F. 461 L. "seliquastra genus
sedilium", Uarr. LL 5.128 "ab sedendo appellatae sedes, sedile, solium, sellae,
siliquastrum"; Mar. Uict. 6.9 K. "nos nunc et aduentum et apud per D potius quam per R
scribamus, aruentum et apur, et linguam per L potius quam per D, et praesidium per D
potius quam per L"; P.F. 96 L. "inpelimenta inpedimenta dicebant"; P.F. 133 L. "Médius
fidius compositum uidetur et significare louis filius, id est Hercules, quod Iouem Graece
244 MARIO NEGRI
et nos Iouem, ac fidium pro filio, quod saepe antea pro L littera D utebantur", e cf.
anche Cat. αρ. Dion. Hal. 2.49.2; Fest. 188 L. "odefacit dicebant > antiqui ab odore,
<pro olfacit>", P. F. 189 L. "odefacit dicebant pro olfacit...".
4.5. Si vedano inoltre: Mar. Uict. 6.8 K. "Gn. Pompeius Magnus et scribebat et dicebat
kadamitatem pro calamitate", Is. Or. 1.27.14 "L autem litteran interdum pro D littera
utimur, ut... . 'calamitatem' pro 'cadamitatem'; a cadendo enim nomen sumpsit
calamitas"; P.F. 18 L. "ancaesa dicta sunt ab antiquis uasa, quae caelata appellamus,
quod circumcaedendo talia fiunt. Ipsum quo que caelare uerbum ab eadem causa est
dictum, D littera cum L permutata"; P.F. 60 L. "dacrimas pro lacrimas Liuius saepe
posuit, nimirum quod Graeci appellant item dautia, quae lautia dicimus, et dantur
legatis hospitii gratia", Gloss. Ansil. L DR 13 "dacrumis lacrimis..."; Fest. 334 L.
"rediuiam quidam, ali reluuium appellant, cum circa unguis cutis se resoluit", P. F. 335
L. "rediuia siue reluuium dicitur, cum circa ungues cutis se soluit, quia duere est
soluere"; Uarr. LL 6.83 "littera commutata dicitur odor olor, hinc olet et odorari et
odoratus et odora res...»
4.6. A queste testimonianze dirette potrebbero da ultimo aggiungersi le non poche forme
latine che, sulla base della comparazione, interna  esterna, mostrano lo scambio dll:
solium ma sedeō (vd. Uarr. LL 5.128; Fest. 386 L. "solia appellantur sedilia, in quibus
non plures singulis possint sedere; ideoque soliar sternere dicuntur, qui sellisternium
habent, et solaria uocantur Babylonica, quibus eadem sternuntur" = P.F. 387 L.)
ecc.; lēuir; soleō ma (?)4 sodālis, ecc.
5. Rotacismo
5.1. Giacché obiettivamente il rotacismo appartiene al IV s. a.C., l'antichità della
documentazione cui in prima istanza ci riferiamo renderebbe in ogni caso impossibile
rinvenirne tracce in sabino (cf. esemplarmente Uarr. LL 7.27 "... f<o>edesum foederum,
plusima plurima, meliosem meliorem, asenam arenam, ianitos ianitor.") Di modo che,
contrariamente al Palmer1,riterremoqui indistinguibili i documenti sabini e quelli latini. E
la stessa prospettiva credo debba attribuirsi anche agli estensori delle notizie esaminate, ai
cui occhi doveva in questo caso apparire una coincidenza completa fra il uetus e il
sabīnum.
5.2. Uarr. αρ. Uel. Long. 7.69 . "nos non tam propter illas causas...harenam
dicimus>, quam propter originem uocis, siquidem, ut testis est Uarro, a Sabinis fasena
dicitur, et sicut S familiariter in R transit, ita F in uicinam adspirationem mutator"; per
ausum e Auselii vd. 6.6. Per Loebãsius 6.3.
5.3. In Uarr. LL 5.74 si ha un elenco degli dei sabini cui Tito Tazio ha dedicato are:
"Feronia, Minerua, Nouensides a Sabinis. Paulo aliter ab eisdem dicimus h<a>ec: Palem,
Uestam, Salutem, Fortunam, Fontem, Fidem. E<t> ar<a>e Sabinum linguam olent, quae
Tati regis uoto sunt Romae dedicatae: nam, ut annales dicunt, uouit Opi, Flor<a>e,
Uedioui Saturnoque, Soli, Lunae, Uolcano et Summano, itemque Larundae, Termino,
Quirino, Uortumno, Laribus, Dianae, Lucinaeque; e quis nonnulla nomina in utraque
lingua habent radices...". Per quanto interessa qui a noi, rileviamo che sia Larunda 'mater
Lamm', sia Lares appaiono con -r- pur se "Sabinum linguam olent" laddove ai ueteres è
attribuita la forma con -s- in Uarr. LL 6.2 "sic, inquam, consuetudo nostra multa
declinauit a uetere, ut ab t solu solum, ab Loebeso Liberum, ab Lasibus Lares". La
forma con -s- è, com' è noto, fra l'altro nel Carmen Aruale ("enos Lases iuuate")2.
5.4. Quanto a Casinum, veramente Varrone LL 7.29, contrariamente a quanto afferma
l'Ernout, Eléments, 58-9, non ne sostiene la natura sabina, ma si limita a supporla,
giacché "hoc enim [i.e. oppidum (.)] ab Sabinis orti Samnites tenuerunt", all'interno
della dimostrazione dedicata al significato di cascus 'uetus', e alla sua origine sabina. Vd.
la notizia in extenso in 11.2.
6. Dittonghi
6.1. Scrive l'Ernout: "D'après un renseignement fourni par Varron, L.L., VII, 96,
"rustici pappum (dicunt) mesium non maesium". De la phrase de Varron, il faut
rapprocher la glose de l'abréviateur de Festus "Maesius lingua Osca mensis Maius. Osci
enim a regione Campaniae, quae est Oscor, uocati sunt". P.F., 109, Th. P. M. von
Planta, Gramm, osk.-umbr. Dial., II, 207 a démontré que le mesius de Varron, comme le
maesius osque de Festus sont un même mot, dérivé à l'aide du suffixe -yo- d'un
comparatif *ma(h)-is- (cf. le mais de la table de Bantia). Les rustici dont parle Varron
emploient mesius avec le sens de pappus, de même qu'en latin maior s'emploie avec le
sens de "âgé, plus âgé". Si l'on songe que les Sabins maintiennent -s- intervocalique (cf.
246 MARIO NEGRI
le nom propre Casinum donné comme sabin par Varron, L.L., VII, 29) et réduisent la
diphtongue ai à ē, et si l'on se rappelle que la plupart des citations dialectales faites par
Varron sont tirées du vocabulaire sabin, c'est aux rustici du Samnium qu'on attribuera le
mesius équivalent au pappus latin..."1. Dallo stesso punto di vista, e cioè da un abbaglio
prodotto da una non retta considerazione delle conseguenze sull'analisi della
documentazione dell'equipollenza - possibile, non necessaria - fra notizie attribuite ai
ueteres, ai Sabini e d'altronde ai rūstici, afferma il Palmer che "the different treatment of
the diphtongs provides yet another series of words of presumed Sabine origin. Thus
from ou, au and ai the Latin equivalents ū, au, ae contrast with Sabine ō, ō, ē
respectively..."2
La documentazione è invece inequivoca nel negare questa presunta precocità sabina
nella monottongazione dei dittonghi.
6.2. AL Troviamo qui subito l'unica attribuzione esplicita di una forma con
monottongazione al sabino. Uarr. LL 5.97 "<h>ircus, quod Sabini fircus; quod illic
fedus, in Latio rure <h>edus, qui in urbe ut in multis A addito <h>aedus". Varrone
registra qui dunque tre livelli di corrispondenza, e cioè:
latino rustico: <h-> δ (<h>ēdus)
sabino ƒ- δ (fēdus)
latino urbano: <h-> ae (<h>aedus).
La notizia principale è riferita al consonantismo iniziale (e infatti si rileva <h>ircusl
fircus), ma la corrispondenza sab.ē:lat. urb. ae (sentita sincronicamente come "aggiunta"
di a all'ē rustico") è esplicita. Notevole - ma, per chi scrive, non inatteso - il rapporto lat.
rust. <h->: sab./-. Il versante del uetus riferito al nome del 'capretto' è coperto da due
notizie: Uel. Long. 7.69 K. "haedos dicimus cum adspiratione, quoniam faedi [mss.
foedi] dicebantur apud antiquos" (qui probabilmente = Sabinos, per/- e per oe, vd.
appresso), e Fest. 74 L. "faedum antiqui dicebant pro haedo". Anche se la situazione è
indubbiamente complessa, potremmo azzardare, in coerenza con il quadro linguistico
generale, la situazione seguente:
sabino faed- (ai).foed- per ipersabinismo.
lat. rust. hed-
lat. urb. haed-
Mesium è, come si è visto, attribuito ai rūstici, e non ai Sabini. Il mancato rotacismo non
costituisce, secondo il mio vedere, qui indizio probante, giacché potrebbe doversi in
qualche modo all'osco maesius. Così pretor e Cecilius in Luc. 9.10 "Cecilius pretor ne
rusticus fiat"3. Né Non. 894 L. "laeuir (mss. leuir) dicitur frater mariti, quasi laeuus uir",
né P. F. 102 L. "leuir est uxori meae frater meus", attribuiscono il nome del'cognato' ad
ambiente sabino (le glosse, ove si consenta ai dubbi espressi in 4. sulla sabinità di dll,
sono sostanzialmente oltreché formalmente adespote: contra Ernout)4. Qui la presenza di
dittongo antico è solo provata dalla comparazione: sscr. devar-, gr. δαήρ ecc., vd.
DELL, s.u. lēuir.
La monottongazione precoce di ai è invece ben documentata in iscrizioni extraurbane5
e in falisco6.
Una via affatto diversa da quella qui proposta è stata invece percorsa dal Peruzzi
("PdP" 106, 15 ss.) che, sulla base delle forme deuas comiscas di TL A 16 (da Roma),
inoltre P. F. 24 L. "aestimias aestimationes [codd. aestimatinibus]", infine anabestas
CIL I2 2.969, ha supposto un trattamento sabino ai > ā. Vd. però Prosdocimi, in Storia,
LXXX ss. (perché oi > oe ma non ai > ae?) e Lazzeroni, "SSL" 5,65 ss.
6.3. OI. Il trattamento sabino dovrebb'essere, come ha mostrato il Peruzzi7, ed è
d'altronde ampiamente confermato dalla documentazione, oe. Così Oenotrus - per quanto
è indipendente dal modello greco - re sabino in Uarr. ap. Seru. ad Aen.. 1.532 "Oenotria
... dieta est uel a uino optimo, quod in Italia nascitur, uel ut Uarro dicit ab Oenotro, rege
Sabinorum", in Plin. NH 3.99 e Dion. Hal. 1.11.3 fratello di Peucetius (eu! vd. oltre)8.
Loebāsius, quali che siano i suoi rapporti obiettivi con Līber, ne è, in qualche modo,
subiettivamente il teonimo corrispondente in sabino9: "quamuis Sabini Cererem Panem
appellent, Liberum Loebasium, dictum autem, quia graece λοιβή dicitur res diuina". Alla
qual forma, quandanche in modo non del tutto perspicuo, si collegano loebesum e
loebertotem di P.F. 108 L. "loebesum et loebertatem antiqui dicebant liberum et liber-
tatem. Ita Graeci λοι,βήν et λεißειv". Vd. inoltre Fest. 224 L. "priuiclioes, priuis id est
singulis" e 224 L. "Pilumnoe poploe in carmine saliari Romani"; P. F. 76 L. "fescemnoe
uocabantur, qui depellere fascinum credebantur"; P. F. 17 L. "ab oloes dicebant pro ab
illis; antiqui enim litteram non geminabant"10.
Ancorché indirettamente, anche la forma foedos dei mss. di Uel. Long. cit. in 6.2,
potrebbe render testimonianza di sabino oi > oe. Giacché dal punto di vista della fonetica
248 MARIO NEGRI
storica qui il dittongo oe è comunque ingiustificato, coglie il sospetto che foedus sia,
come già si è accennato, una forma ipersabinizzata, con due marche che, evidentemente,
dovevano essere sentite appunto per caratteristiche sabine:/- per h- e oe appunto. Ma
mentre f- per h- è anche semplicemente uetus, vd. 8.3, oe non è tale p. es. in P.F. 17 L.,
come ha finemente rilevato il Peruzzi, vd. ad n.10, ove si rileva che gli antiqui "litteram
non geminabant", di modo che / della forma registrata oloes corrisponde a // di illīs
classico, ma non si fa cenno di oe(s) per -īs classico: "segno che -oe... -oes erano
morfemi correnti nei testi arcaici che si potevano ancor leggere al tempo di Verno Fiacco.
Ma in quali testi? Le più antiche epigrafi latine che possediamo presentano oi, non oe..I
grammatici ricordano terminazioni con oi "ex libris antiquis foederum et legum"...; non
citano invece alcuna desinenza con oe. Dunque, le terminazioni con oe non erano
caratteristica generale dei testi arcaici, ma solo di alcuni, e non avevano rilevanza per gli
studiosi della grammatica latina. Si ha insomma un'altra conferma che quel digramma è
peculiare di un gruppo di documenti (e documenti non epigrafici), ed a tale gruppo
appartengono i carmina saliari". In tale visione, hoedos urbano sarebbe sorto per il
semplice procedimento di scambiare/- con h-, lasciando immutato il resto della forma,
per rappresentare in tal modo il rapporto fra parola sabina e parola latina, secondo quanto
sarà detto appresso11.
6.4. EU. L'accertata appartenenza all'ambiente sabino del Carmen Saliare [p. es. Uarr.
LL 7.3 "quorum si Pompili regnum fons in Carminibus Saliorum neque ea ab
superioribus accepta, tarnen habent DCC annos...", Enn. Ann. 120 V. ap. Uarr. LL
7.43] offre un'importante testimonianza in ordine al trattamento di eu. Ter. Scaur. 7.28
. "...ut [la notizia riguarda le grafie del tipo quom/cum] Numa in Saliari Carmine:
cuine ponas Leucesiae praetexere monti
quot ibet etinei de is cum tonarem".
Giacché in sabino il nesso di dentale +  sembra dar luogo a una sibilante (vd. 7.1.2.),
Leucesiae non sarà, come vuole l'Ernout, "sans doute graphie fautive pour Leucetie"12,
bensì l'attesa resa in sabino del teonimo corrispondente al Lūcetius osco (e latino?). P. F.
102 L. "Lucetium Iouem appellabant, quod eum lucis esse causam credebant", Seru. ad
Aen. 9.567 "sane lingua Osca Lucetius est Iuppiter". Contrasta in verità con la
testimonianza offerta da Leucesie quella di Lūcii in Uarr. LL 6.5 "crepusculum a crepero:
id uocabulum sumpserunt a Sabinis, unde ueniunt Crepusci nominati Amiterno, qui eo

tempore erant nati, ut Luci prima luce in Reatino". Ma le due testimonianze hanno
evidentemente assai diverso spessore: il gentilizio appare comunque in forma
modernizzata (e/o latinizzata?).
E' naturalmente di grande interesse por mente al fatto che in area falisca si danno
alcune forme con eu conservato13. Se, come credo, Leucesie è realmente una forma
sabina, questo e il falisco conserverebbero nell'Italia antica eu contro la regola latina (e
italica) per cui eu > ou: già (?) iouxmenta nel cippo del Foro. Eu appare comunque
nell'iscrizione da Poggio Sommavilla: vd. Prosdocimi, Storia, LXXXIX.
6.5. OU. Per quanto se ne possono lecitamente distinguere le vicende da quelle di eu,
l'unica forma "sospetta" dal nostro punto di vista dovrebb'essere, per quanto so, rōbīgō,
già in Plaut. Stich. 230 robiginosam, poi ben attestato anche come teonimo, p. es. in Ou.
Fast. 4.907, masch. in Uarr. LL 6.16 "Robigalia dicta a Robigo; secundum segetes huic
deo sacrificatur, ne robigo occupet segetes", RR 1.1.6., P. F. 325 L., Seru. ad Georg.
1.151 "robigo autem genus est uitii, quo culmi pereunt... Inde et Robigus deus et sacra
eius decimo Kl. Maias Robiginalia appellantur. Sed haec res abusiue robigo dicitur; nam
proprie robigo est, ut Uarro dicit, uitium obscenae libidinis, quod ulcus uocatur", ecc.
Ma nulla, né dal punto di vista documentario né, ora lo possiamo affermare, formale,
impone e neppur consiglia di vedere nel nome della 'ruggine' una forma sabina. Si
tratterà piuttosto, come d'altronde è reso verisimile dall'ambito culturale cui il termine
appartiene, di una forma rustica, con õ come losna per luna a Preneste (TL A 22).
6.6. AU. Oltre al gentilizio Clausus, per cui vd. 7.1, la conservazione del dittongo è
conclamata in P.F. 8 L. "aurum ... a Sabinis translatum putant, quod illi ausum
dicebant", cf. invece per il trattamento rustico Fest. 196 L. "orata genus piseis appellatur
a colore auri, quod rustici orum dicebant, ut auriculas, oriculas", = P.F. 197 L., e in
Auselii, P. F. 22 L. "Aureliam familiam ex Sabinis oriundam a Sole dictam putant, quod
ei publice a populo Romano datus sit locus, in quo sacra faceret Soli, qui ex hoc Auseli
dicebantur, ut Ualesii, Papisii pro eo, quod est Ualerii, Papirii". Questa notizia festina
suggerisce, per quanto oscuramente, un rapporto fra il gentilizio sabino e il nome del
sole. Sōlè, giusta Uarr. LL 5.74, già cit. in 5.3, una delle divinità sabine cui Tito Tazio
dedicò are. La sua origine sabina è confermata da LL 5.68 "Sol uel quod ita Sabini (mss.
250 MARIO NEGRI
Sauini, vd. 1. η. 12), uel <quod> solus ita lucet, ut ex eo deo dies sit": ma la lettura qui
riportata15 è frutto di emendamento, giacché la tradizione manoscritta ha "sola". In tale
situazione testuale l'intervento del von Blumenthal16, "Sol <uel a s> auel...", onde il
"sabino" sauel, ancorché ovviamente ipotetico, appare del maggior interesse, giacché
sauel a un tempo conforta e alla sua volta è confortato dalla variante Sauīni, vd. quanto
detto in 1. η 1217.
7. Assibilazionedei gruppi di dentale + y

7.1. -dy- > -s-. Uerg. Aen. 7.706-9 "Ecce Sabinorum prisco de sanguine magnum/
agmen agens Clausus magnique ipse agminis instar,/ Claudia nunc a quo diffunditur et
tribus et gens/ per Latium, postquam in partem data Roma Sabinis", Seru. ad Aen. 7.706
"nam Clausus, Sabinorum dux, post exactos reges, ut quidam dicunt, cum quinque
milibus clientium et amicorum Romam uenit, et susceptus habitandam partem urbis
accepit: ex quo Claudia et tribus est et familia nominata", Liu. Hist. 2.16 "Attius Clausus,
cui postea Appio Claudio fuit Romae nomen... ab Inregillo... Romam transfugit", Tac.
Ann. 4.9"... Sabina nobilitas, Attus Clausus ceteraeque Claudiorum effigies", Suet. Tib.
1 "patricia gens Claudia... orta est ex Regillis oppido Sabinorum. Inde Romam recens
conditam cum magna clientium manu conmigrauit, auctore Tito Tatio consorte Romuli
uel, quod magis constat, Atta Claudio gentis principe, post reges exactos sexto fere
anno"1; basus in CGL V.170.2 'rufus niger' se va con badius, vd. DELL, s.u. basus e
badius, Devoto, Storia 82-3. Ma cf. Medius Fidius = Zeύs IIíστιος in Dion. Hal.
2.49.2, vd. anche 4.4; trimodia in Schol, ad Hor. Sat 1.1.53 "cumerae dicuntur uasa
minora quae capiunt quinquE siue sex modios, quae lingua Sabinorum trimodiae
dicuntur", e fondatore di Cures2.
7.2. -ty- > -s-. Leucesie, cf. "osco" Lucetius, vd. 6.43. Ma cf. in
Uarr. ap. Dion. Hal. 1.14.1. Inoltre Met<t>ius Curtius [mss. curcium] Sabinus in Uarr.
LL 5.149.
8. Lo scambio h/f in posizione iniziale

8.1. La complessa documentazione riferita al fenomeno in esame è organizzabile in
quattro sezioni, tuttavia variamente intersecantisi, e cioè: a) attribuzione ai Sabini b)
attribuzione ai ueteresC)attribuzione ai Falisci d) adespote. Nonostante che il fatto sia

largamente attestato in etrusco, per quanto so nessun grammatico ne fa cenno.
8.2. Uarr. LL 5,91 fircus, vd. oltre1;febru(u)m ecc., vd. 2.2; Uel. Long. 7.69 K. (qui
in extenso) "item non nulli harenam cum adspiratione, siue quoniam haereat, siue quod
aquam hauriat, dicendam existimauerunt; aliis sine adspiratione uidetur enuntiandam. Nos
non tarn propter illas causas quas supra proposuimus harenam dicimus, quam propter
originem uocis, siquidem, ut testis est Uarro, a Sabinis fasena dicitur, et sicut S
familiariter in R transit, ita F in uicinam adspirationem mutatur. Similiter ergo et haedos
dicimus cum adspiratione, quoniam faedi dicebantur apud antiquos: item hircos, quoniam
eosdem aeque fircos uocabant. Nam et e contrario quam antiqui habam dicebant, nos
fabam dicimus".
8.3. Ter. Scaur. 7.11 . "ubi illi [antiqui] ... F litteram posuerunt, nos H substituimus,
ut quod illi fordeum dicebant, nos hordeum, fariolum quem nos hariolum"; Uel. Long.
7.81 K. "antiquorum consuetudo respicienda est, quibus moris fuit pro hac adspiratione
litteram dicere F. Itaque harenam iustius quis dixerit, quoniam apud antiquos fasena erat,
et hordeum, qui fordeum, et, sicut supra diximus, hircos, quoniam firci erant, et haedos,
quoniam faedi"2; P.F. 74 L. "faedum antiqui dicebant pro haedo, folus pro holere, fostim
pro hoste, fostiam pro hostia"3; P.F. 91 L. "horreum antiqui farreum dicebant a farre"; P.
F. 91 L. "horctum et forctum pro bono dicebant".
8.4. Ter. Scaur. 7.13 . "quem antiqui fircum, nos hircum, et quem Falisci habam, nos
fabam appellamus, et quem antiqui fariolum, nos hariolum"; Seru. ad ken. 7.695
"Faliscos Halesus condidit. Hi autem, immutato H in F, Falisci dicti sunt, sicut febris
dicitur quae ante hebris dicebatur, Formiae quae Hormiae fuerunt, nam
posteritas in multis nominibus F pro H posuit", cf. P.F. 73 L. Evidenti riscontri
epigrafici sono LIA 143 Efe ma lat. hic (hec); LIA 150 foied ma lat. hodiē, ecc.
8.5. Inoltre gli adespoti Haunii 'di agrestes', vd. Fauni; P.F. 91 L. "hanula pania
delubra, quasi fanula"; Uarr. RR 2.5.6. "quae sterilis est uacca, taura appellata, quae
praegnas, horda. Ab eo in fastis dies hordicalia nominatur, quod tum hordae boues
inmolantur", LL 6.15 "Fordicidia a fordis bubus; bos forda quae fert in uentre; quod eo
die publice inmolantur boues praegnantes in curiis complures, a fordis caedendis
Fordicidia dieta", P.F. 91 L. "horda praegnans, unde dies, quo grauidae hostiae
252 MARIO NEGRI
immolabantur, Hordicidia". Uarr. LL 5.113 "filum, quod minimum est hilum", Uarr. LL
6.53 "<fariolus> ac fatuus dictum" (?) se è variante di hariolus, haruspex4.
9. Il passaggio kw > 
9.1. Alcuni termini attribuiti ai Sabini sembrerebbero mostrare il passaggio kw > k, forse
condizionato da  seguente (come in latino): per tesqualtesca vd. 2.6; ih. per fìrcus (cf.
hirpus? e Hirpi Sorani?)1; inoltre Sancus (?), vd.l1.2; per Fircellius vd. 8.2.
10. Chiusura delle vocali medie avanti a -rC-

10.1. Mancano argomenti per la sabinità di erC- > irC- (e del parallelo orC- > urC-, vd.
Ernout, Eléments, 62), giacché fircus è "sabino" per/- contro h- dell'urbano hircus, e
hirpus non è sabino, come del resto ursus è "lucano" in Uarr. LL 5.100: si tratta, credo,
di un tratto palesemente rustico, cf. Mircurius ecc., Uarr. ap. Uel. Long. 7.77 K. "mium
et commircium quoque per I antiquis relinquamus, apud quos aeque et Mircurius per I
dicebatur, quod mirandarum rerum esset inuentor, ut Uarro dicit. No stris iam auribus
placet per E, ut et Mercurius et commercia dicantur", stircus di TL A 251 ecc. (Ernout,
Eléments, 60 ss.).
11. Lessico
11.1. A conferma dell'obiettivo spessore storico del sabino, e del suo non essere mera
variante (stilistica? sociolinguística? psicologica?  che altro?) del latino, gioverà da
ultimo rilevare come un certo novero di notizie, anziché informarci di un tratto formale
sentito come caratteristico, sul tipo dì faedus/haedus, ci conservi i frustuli di un lessico, e
d'altronde di un patrimonio onomastico, teonomastico, toponomastico sabino autonomo
rispetto a quello latino e, anche se forse in misura minore, a quello italico.
11.2. Fest. 456 L. "scensas < Sabini dicebant, quas > nunc cenas. Quae autem < nunc
prandia, cenas > habebant, et pro ceni < s uespernas >", P. F. 457 L. "scensas Sabini
cenas dicebant. Quae autem nunc prandia sunt, cenas habebant, et pro cenis uespernas
appellabant" = P.F. 47 L.; Plin. NH 14.37 "pumula"; Strab. 5.4.2 "
cf. anche Fest. 228 L. "picum auem quidam dictum putant a Pico rege Aboriginum";
Lyd. De Mens. 1.21 "[Ν.]·
Fest. 510 L. "uernae, qui in uillis uere nati, quod tempus duce natura feturae est, et tunc
rem diuinam instituent Marti Numa Pompilius pacis concordiae obtinendae grada inter
Sabinos Romanosque, ut uernae uiuerent ne uincerent. Romanos enim uernas
appellabant, id est ibidem natos, quos uincere perniciosum arbitrium Sabinis, qui
coniuncti erant cum populo Romano", P. F. 511 L. "uernae appellantur ex ancillis ciuium
Romanorum uere nati, quod tempus anni maxime naturalis feturae est"; Uarr. LL 6.28
"idus ab eo quod Tusci Itus, uel potius quod Sabini Idus dicunt", P.F. 56 L. "cumbam
Sabini uocant earn, quam militares lecticam, unde uidetur deriuatum esse cubiculum", ma
P.F. 44 L. "cumbam Graeci appellant"; Seru. ad Aen. 7.517 "Sabini lingua sua
nar dicunt sulphur. Ergo hunc fluuium ideo dicunt esse Nar appellatum, quod odore
sulphureo nares contingat", vd. anche Enn. Ann. 265 V.; Uarr. RR 1.14. 3 "aggeres
faciunt sine fossa: eos quidam uocant muros, ut in agro Reatino"; Gloss, nom. 32 Labh.
Plin. NH 15.3.13 "[oliuam Sergiam] Sabini
regiam uocant"; Seru. ad Aen. 12.538 "sane sciendum cupencum Sabinorum lingua
sacerdotem uocari"; Uarr. LL 6.5 "secundum hoc dicitur crepusculum a crepero: id
uocabulum sumpserunt a Sabinis, unde ueniunt Crepusci nominati Amiterno, qui eo
tempore erant nati, ut Luci prima luce in Reatino; crepusculum significai dubium; ab
eo res dictae dubiae creperae, quod crepusculum dies etiam nunc sit an iam nox multis
dubium", LL 7.77 "crepusculum ab Sabinis, et id dubium tempus noctis an diei sit";
Uarr. LL 7.46 "cata acuta: hoc enim uerbo dicunt Sabini" e le testimonianze enniane lì
riportate (Ann. 391 V., 529 V.); Uarr. LL 7.28-9 "in Carmine Priami quod est "ueteres
Casmenas cascam rem uolo profarier", primum cascum significat uetus; secundo eius
origo Sabina, quae usque radices in Oscam linguam egit. Cascum uetus esse significat
Ennius quod ait: "Quam Prisci casci populi tenuere Latini". Eo magis Manilius quod ait:
"Cascum duxisse cascam non mirabile est/ quoniam cariosas conficiebat nuptias". Item
ostendit Papini epigrammation, quod in adulescentem fecerat Cascam: "Ridiculum est,
cum te Cascam tua dicit amica, / fili Potoni, sesquisenex puerum./ Dic tu illam pusam: sic
fiet "mutua muli": nam uere pusus tu, tua amica senex". Idem ostendit quod oppidum
uocatur Casinum (hoc enim ab Sabinis orti Samnites tenuerunt) et nostri edam nunc
254 MARIO NEGRI
Forum Uetus appellant. Item significai in Atellanis aliquot Pappum senem, quod Osci
casnar appellant" P.F. 41 L. "cascum antiquum", ib. "casnar senex Oscorum lingua"; e
cf. Sauf. αρ. Seur, ad Aen. 1.6. "Saufeius Latium dictum ait, quod ibi latuerant incolae,
qui quoniam in cauis montium uel occultis cauentes sibi a feris beluis uel a ualentioribus
uel a tempestatibus habitauerint Cascei uocati sunt, quos posteri Aborigines
cognominarunt, quoniam † aliis ortos esse recognoscebant. Ex quibus Latinos etiam
dictos"; Seru. ad Aen. 3.235 "Sabini et Umbri quae nos mala, dira appellant"; Gell. NA
11.1 "uocabulum autem ipsum «multae» idem M. Uarro in XXI Rerum Humanarum non
Latinum sed Sabinum esse dicit, idque ad suam memoriam mansisse ait in lingua
Samnitium, qui sunt a Sabinis orti", P. F. 127 L. "multam Osce dici putant poenam"
Fest. 126-8 L. "multam Osce dici putant poenam quidam" cf. Uarr. LL 5.177, 7.29.;
Uarr. LL 5.106-7 "circuii, quod mixta farina et caseo et aqua circuitum aequabiliter
fundebant. Hos quidam qui magis incondite faciebant uocabant lixulas et similixulas
uocabulo Sabino: quae frequentia Sabinis"; Uarr. ap. Macr. Sat. 3.18.13 "nux tarentina
dicitur quae ita mollis est ut uix attrectata frangatur. De qua in libro Fauorini sic reperitur:
'item quod quidam Tarentinas oues uel nuces dicunt, quae sunt terentinae a tereno, quod
est Sabinorum lingua molle, unde Terentios quoque dictos putat Uarro ad Libonem
primo"; per lepesta vd. 4.; Uarr. LL. 5.73 "Mars ab eo quod maribus in bello praeest, aut
quod Sabinis acceptus ibi est Mamers", Fest. 116 L. "Mamercus praenomen Oscum est
ab eo, quod hi Martern Mamertem appellant" = P. F. 117 L., ib. "Mamers Mamertis facit,
id est lingua Osca Mars Martis, unde et Mamertini in Sicilia dicti, qui Messanae habitant",
Fest. 150 L. "Mamers ... qui lingua Oscorum Mars significatur"; per febru(u)m vd. 2.3;
per h-/forctus vd. 8.3; Uarr. LL 5.66 "Aelius Dium Fidum dicebat Diouis filium, ut
Graeci Διόσκορου Castorem, et putabat hunc esse Sanc[t]um ab Sabina lingua et
Herculem a Graeca", Lyd. De Mens. 4.90 "
August. CD 18.19 "Sancus primus rex Sabinorum: sed Aeneam,
quoniam quando mortuus est non comparuit, deum sibi fecerunt Latini. Sabini etiam
regem suum primum Sancum ... rettulerunt in deum", Ou. Fast. 6.213-6 "quaerebam,
Nonas Sanco Fidione referrem,/ an tibi, Semo pater; tum mihi Sancus ait/ 'cuicumque ex
istis dederis, ego munus habebo:/ nomina terna fero: sic uoluere Cures/", ecc.; Porph. ad
Hor. Epist. 1.10.49 "Uacuna in Sabinis dea" ecc.; Plin. NH 14.38 "uinaciolam soli
nouerunt Sabini"; Schol. ad Hor. Sat. 1.1.53 "cumerae dicuntur uasa minora quae
capiunt quinque siue sex modios, quae lingua Sabinorum trimodia dicuntur"; per curis
vd. 2.6.; per sōl vd. 6.6; per ausum vd. 6.6.; per tesqua vd. 2.6; per hirpus vd. 2.3.; per
ēloquium e reloqui vd. 2.6.; per teba vd. 3.3.; Uarr. LL 5.159 "Uicus Cyprius a cypro,
quod ibi Sabini ciues additi consederunt, qui a bono ornine id appellarunt: nam cyprum
Sabine bonum"1, Uarr. LL 5.97 "poreus, quod Sabini dicunt aprunu<m> porcu<m>"(6);
Gell. NA XIII.23 "sed 'Nerio' a ueteribus sic declinabatur quasi Anio: nam perinde ut
'Anienem' sic 'Nerienem' dixerunt tertia syllaba producta. Id autem, siue 'Nerio' siue
'Nerienes' est, Sabinum uerbum est, eoque significatur uirtus et fortitudo. Itaque ex
Claudiis, quos a Sabinis oriundos accepimus, qui erat egregia atque praestanti fortitudine,
'Nero' appellatus est", Lyd., De Mens. 4.60
Suet., Tib., 1.1 "Inter cognomina

autem et Neronis assumpsit, quo significatur lingua Sabina fortis ac strenuus" (vd. anche
DELL, s.u. Nerō); gloss. Abol.L 2 ST 23 = CGL V.515.62 "stolones: ramuli maioribus
succrescentes ita dicuntur a Sabinis" (ma vd. Uarr. RR 1.2.9); Seru., ad Aen. 7.684
"Sabinorum lingua saxa hernae uocantur" (ma vd. Schol. Veron, ad Uerg. ib. "Marsi
herna dicunt"), P.F. 89 L. "Hernici dicti a saxis, quae Marsi herna dicunt"; per Panem
vd. 6.3; Fest. 492 L. = P.F. 493 L. "Talus praenomen erat Sabinorum"; spurium sabino
ap. Plut. Quaest. Rom. 103. Uarr. ap. Inc. De Praenom. 1.3 = 323 F. "Uarro Simplicia
in Italia nomina fuisse ait existimationisque suae argumentum referí, quod Romulus et
Remus et Faustulus neque praenomen ullum neque cognomen habuerint. Qui ab eo
dissentiunt aiunt matrem eorum Ream Siluiam uocatam, auum Siluium Numitorem,
fratrem eius Amulium Siluium, ac superiores Albanorum reges Capetum Siluium,
Agrippam Siluium posterioresque duces Mettium Fufetium et Tutorem Cloelium uocatos.
Nec contenti his ad Sabinos transgrediuntur: Titum Tatium, Numam Pompilium. In hunc
modum Uarronis sententia subruitur ... Romanos autem arbitrandum est maxime ab
Albanis et Sabinis multiplicandorum nominum consuetudinemque traxisse, quoniam ab
illis orti sunt..."; Uarr. ap. Seru. ad Aen. 7.657 "Uarro ... dicit in gente populi Romani,
Sabinos a Romulo susceptos istum accepisse montem [Auentinum], quem ab Auente,
fluuio prouinciae suae, Auentinum appellauerunt"; Uarr. ap. Inc. De Praenom. 1.4. =
324 F. "Ancum praenomen Uarro e Sabinis translatum putat"; ib. (325 F.) "Appius ab
256 MARIO NEGRI
Atto eiusdem regionis [Sabinae] praenomen"; ib. (344 F.); "Titus e Sabino nomine Tito
fluxit; P.F. 41 L. "Calpurni a Calpo, Numae regis filio, sunt oriundi"2.
12. Conclusioni linguistiche

12.1. Se il corpus così raccolto e organizzato presenta, come mi auguro, solo
manchevolezze nel dettaglio, potremo azzardare sulla base di esso qualche conclusione
linguistica. Sottolineo il carattere strettamente interno alla prospettiva linguistica e non,
com'è comune fare ragionando di sabino, a un tempo linguistica e storico-antiquaria, che
è naturalmente e, spero, chiaramente correlato alla qualità - e ai voluti limiti - del tentativo
fatto in queste pagine.
12.2. La prima considerazione - che, peraltro, è ai miei occhi di speciale importanza - è la
seguente: la documentazione - riferita al sabino delle glosse - è sostanzialmente coerente al
suo interno, ciò che pare prova del fatto che tale documentazione rappresenta il ricordo di
una realtà obiettiva, storica, com'è una lingua appunto. Se il sabino fosse, come taluno
ritiene, nient'altro che l'ipostasi mitica di una varietà del latino le notizie sarebbero fra
loro, credo, ben più incoerenti di quanto non sia il sabino che obiettivamente emerge da
quanto fin qui si è letto.
12.3. La seconda considerazione, che ovviamente si riallaccia alla prima, e in qualche
modo la precisa, è questa: né il latino coevo, né l'osco, né il falisco - né tampoco il greco!
- sono in grado di render conto dei tratti legittimamente attribuibili al sabino. Questa
seconda considerazione, correlandosi alla prima, consente di riaffermare la natura
sostanzialmente autonoma del sabino - nell'età in cui ci è documentato, e cioè in quella
regia - sia dal latino sia dalle lingue italiche, senza tuttavia pregiudizi in ordine alle
vicende pregresse rispetto alla situazione documentaria (vd. 1.4.6). Converrà ora
riconsiderare partitamente e in prospettiva di bilancio dialettale tali tratti.
12.4. Il sabino coincide con il latino (e in special modo col latino coevo) nei seguenti
tratti: 1) mancato rotacismo; 2) scambio dll 3) ai conservato (?) (ove non si accolga ai > ã
del Peruzzi); 4) au conservato; 5)f- per h-. Se ne discosta invece in: 6) p < * kw; 7) p <
*-bh- (se teniamo per buona la testimonianza di alpus); 8) oe per oi; 9) eu conservato; 10)
una certa porzione del lessico; 11) -dy-, -ty- (?) > s.
12.5. Dei tratti così individuati, solo 1), 3), 4), 6) ricorrono in osco, 6) in umbro. Per
quanto attiene a 10), le coincidenze sabino-osche sono di notevole rilievo. Per 11)
sarebbe esplicitamente differenziale Lucetius : Leucesie, che è pero contraddetto
soprattutto dalla varietà bantina. Per quanto attiene all'umbro in ordine a 7), vd. 3.1. 2) è,
per comune dottrina, d'ambiente umbro. Per un suo inquadramento più ampio vd.
Bottiglioni, SE 17, 315-26.
12.6. Il falisco coincide col sabino in 5), 9). Va però detto che le peculiarità dell'alfabeto
falisco c'invidiano di decidere se q di LIA 151 fifiqod, risp.  di LIA 148 fifiked
(*dheiĝh-) rappresentino ilriflessodi MA > T  MA > M (come in latino). Esplicitamente
differenziale è però 11), cf. LIA 150 foied <*fo-diēd(Ernout, Eléments, 82), con -dy- >
12.7. Quanto al greco, che qui si prende in considerazione soprattutto in ossequio alla
tradizione dell'origine greca dei Sabini1, oe per oi è grafia nota in beotico, che però non
mostra la labiale - pur se dialetto eolico - nella seconda sillaba del numerale per 'cinque',
né ovviamente in quella del nome del 'lupo'. I dittonghi vi sono in genere ben conservati
(ma proprio in beotico paiono particolarmente disposti a monottongarsi!). Per quanto
attiene all'esito delle MA, vero è che la p di alpus potrebbe rappresentare TA, ma, se ciò
fosse, sarebbe singolare che non fosse giunta fino a Verrio Fiacco, e da lui ai suoi
epitomatoli, la "sensazione" della coincidenza del sabino alpus /alphus/ col greco λς
/alphos/. Ma quasi nessun elemento speciale del lessico sabino trova confronto in greco.
Inoltre il sabino -s- si conserva dove in greco > -/2-, né peraltro il greco mostra particolare
tendenza a scambiare dll 2 e, soprattutto, ignora completamente lo scambio f-Ih- (che
coinvolge fra l'altro un fonema del tutto ignoto al greco di quell'età).
12.8. Le conseguenze linguistiche del quadro che così emerge confermano, credo con
una certa chiarezza, l'assunto di partenza, e cioè che, per l'età regia cui si riferisce la
documentazione, il sabino è una lingua, benché ovviamente non estranea a quelle
indeuropee della stessa area ed età, sostanzialmente non riconducibile a nessuna di
queste, e quindi autonoma. Se tali conclusioni vengono accolte, resterà il compito di
conciliare questo quadro linguistico con la tradizione delle origini "italiche" e/o "greche"
dei Sabini.
258 MARIO NEGRI
NOTE
AL CAP. 1
* Nonostante l'evidente assonanza, il "latino di Numa Pompilio" di Emilio Peruzzi (PdP 106,15 ss.) e
la "lingua di Numa" non sono la medesima lingua. Di fatto, "nel definire 'sabina' questa varietà di
latino" il Peruzzi ha "usato il termine non come qualificazione linguistica ma solo come etichetta
culturale: il latino che corrisponde a una determinata cultura di Roma arcaica..." giacché "non
sappiamo nulla del sabino, tranne pochissime parole attribuite a quella lingua (e forse qualche volta a
torto) da alcuni autori antichi e la superficiale impressione tratta dalle fonti classiche che i romani
sentissero i sabini come un popolo affine" (art. cit. 34-5). La mia opinione è diversa, come cercherò
di mostrare nelle pagine che seguono. Mi è caro però quiricordareil mio debito, metodologico e
culturale, nei confronti dell'illustre studioso.
1 Vd.anche la Storia della lingua di Roma, Bologna 1983 (I ed. 1940) 80 ss. e Gli antichi italici,
Firenze 1969 (I ed. 1931) 85, 111, 140. Per l'identificazione storico-archeologica vd. Pallottino, in
Civiltá arcaica dei Sabini, Roma 1973, 33 ss.
2 I dialetti medio-italici, in Lingue e dialetti dell'Italia antica, Roma 1978, 789-824.
3 Soprattutto 37-8.
4 Soprattutto 58-9.
5 Lo stesso falisco, anche se non è questione che possa essere se non accennata, non è esente da
contraddizioni interne: di particolare momento, in quanto si collega a una problematica su cui
converrà tornare, pare a me quella connessa con il trattamento delle MA interne, se è vero che nella
"Coppa di Cerere" (LIA 151) coesistono i due trattamenti, quello di tipo latino in ƒ<if>iqod ma
l<o>uf<ir:> conI'-f-italica del più tardo efiles di LIA 146 F, forma su cui il Devoto ravvisa come
l'effetto di una "pressione italica" (Gli antichi italici, 131 ss). Vd. anche, di chi scrive, Latino
arcaico, latino rustico e latino preromanzo, Milano 1982,69-70.
6 Nell'ambiente si danno infatti frequenti casi in cui più che di vera evoluzione fonetica hanno da
vedersi casi di trasposizione. Vd., di chi scrive, ASGM 19, 27 ss.
7 Vd. Devoto, Gli antichi italici, 103-4. Non concordo però con l'illustre studioso quand'egli riguarda
la forma Sabini alla stregua di una latinizzazione di un originario Safīni. Sono piuttosto convinto
che il rapporto Sabini: Sqfīni riproduca il contrasto fra etnico nella sua forma "protosabina" (per
usare la terminologia del Devoto) ed etnico nella sua forma "italicizzata" (cioè sabina recente). Un
discorso a parte chiede invece la variante Sauīni,ricorrenteuna decina di volte in uno dei manoscritti
del De Lingua Latina (LL 5.41, 5.66, 5.68, 5.74, 5.97, 5.107, 5.123, 6.5, 7.46, 7.77). Mi pare
evidenteI'improponibilità dell'idea che la variante con  ([w]? [v]?) rappresenti la conservazione nel I
s. a.  dello status [b] < *-bh-, di modo che comunque il rapporto fra Saulnus e Sabīnus dovrebbe
appartenere al quadro delle alternanze blv di ambiente volgare. Restano però aperte tre questioni, e
cioè: a) si tratta di grafie non varroniane, manifestanti il passaggio b > v, come Vivia per Vibia a
Pompei, ovvero di forme autenticamente varroniane? b) se si tratta di grafie non varroniane, perché il
passaggio b > v si manifesterebbe, almeno per quantoriguardala tradizione del De Lingua Latina,
soltanto nell'etnico Sabīnus? c) se si tratta invece di grafieriproducentil'originale varroniano, perché
Varrone avrebbe utilizzato due varianti dello stesso etnico? La spiegazione del Poucet, Les origines
mythiques des Sabins, in Etudes Etrusco-Italiques, Louvain 1963, 155 ss., in part. 195-6, secondo
cui l'etnico Sabìnus avrebbe offerto minor resistenza al passaggio b > v del resto del lessico a causa
dell' accostamento Sabìnus: quia uitem seuit, davvero non convince, anche se non saprei proporne
una migliore. A meno che la variante con  non si connetta in qualche modo con la notizia, che è
però frutto d'integrazione, sull'origine sabina del nome del 'sole', saueV. in Uarr. LL 5.68, vd. 6.6.
Comunque sia, il tipo Sauīnus costituisce, almeno per me, un problema aperto. Per l'etnico S. si
veda sempre Ribezzo, RIGI 14, 59 ss., Rix, BNF 8, 127 ss., e ora R. Giacomelli, Graeca Italica,
Brescia 1983, 19 ss., Wallace, The Sabellian Languages, Ann Arbor, 1986, 1 ss, Prosdocimi in
Dialoghi di Archeologia, 1987, 57-8. La stessa alternanza blv nel fitonimo Sabina/ulna, per cui vd.
André, Les noms de plantes ..., Paris 1985, 223.
8 Così finemente il Devoto, Storia, 82 "Pure, altra è la situazione dei primi [scil, bōs, lupus ecc.],
legati alla vita dei campi, rappresentanti il vocabolario degli immediati dintorni di Roma e quindi
secondo ogni presunzione antichissimi, altra è quella dei due aggettivi [scil, brūtus e bardus] che si
adattano altrettanto bene all'ambiente linguistico delle Atellane, dunque sarebbero arrivati in Roma in
età più recente. Senza alcun dubbio appartengono a questo tratto, non sabino ma osco, termini come
popina ..., botulus 'sanguinaccio'; e il nome campano della 'falce' secula come tale ricordato da
Vairone..."
9 Cf. ASGM 19, 27 ss.
10 Vd. ora le importanti riflessioni di Prosdocimi, in Dialoghi di Archeologia 1987, 53-64. Per la
questione della solidarietà sabino-medioadriatica vd., di chi scrive, Sabino delle glosse e sabino delle
epigrafi, in Episteme, Roma 1990, 137-44.
AL CAP. 2.
1 Vd. M. G. Bruno, I Sabini e la loro lingua, Milano 1962. Le voci "sabine" di Vairone sono raccolte
anche in J.W. Donaldson, Varronianus3, Cambridge-London 1860, 128-39 e in J. Collart, Varron
grammairien latin, Paris 1954, 229-43.
2 Emout, El., 195 (cf. anche, come suggerisce lo studioso francese, hircus). Dal punto di vista del
latino, naturalmente, *kw di *lukwos e * w di *e wo s dovrebbero essere coalescenti (ma cf.
Sull'annosa e futile questione del nome del 'cavallo' rimando a quanto ho scritto in II latino fra le
lingue indeuropee, Milano 1985,17-8.
3 Storia, 163-4.
4 Vd., di chi scrive, Latino arcaico, soprattutto 25 ss.; Lupi Sabinorum lingua hirpi uocantur, Acme
36, 199-203; Due note sabine, Acme 36, 21-6. Alcuni passi e idee presenti nei miei lavori citati
sono stati qui rifusi e riutilizzati.
5 La ricchissima documentazione in Peruzzi, Aspetti culturali del Lazio primitivo, Firenze 1978,
31 ss.
6 E cioè "lupi Sabinorum lingua hirpi uocantur".
7 Gramm. Osk.-Umbr. Dial., Straßburg 1897, II, 591.
8 Vd. DELL e LEW, s.u. Lupercus.
9 In numerosi scritti, culminanti nelle Origini di Roma, Bologna 1970-3, negli Aspetti e da ultimo in
Mycenaeans in early Latium, Roma 1981.
10 Aspetti, 7 ss.
11 A eccezione di retrogradis (/-gradus) in Apuleio (Aspetti, 31).
12 Aspetti, 31.
13 Aspetti, 11, cf. Enn. Ann. 122 ss. V. ap. Uarr. LL 7.46. Vd. anche Uarr. LL 5.84-6.
14 Vd., di chi scrive, Acme 35,199-203.
15 Per cui vd. Aspetti, 8 ss.
16 DELL, s.u. simpuuium. Inoltre piãreV.
260 MARIO NEGRI
17 Devoto, Storia, 75. La connessione fra la rupe e i Tarquinii è palese in Liu. Hist. 1.55. Nega con
argomenti privi di valore, soprattutto dal punto di vista metodologico, il carattere sabino di Tarpēia
(nonché, in generale, il sabino *kw > p) il Pariente, Emerita 11,1943, 338 ss.
18 Leumann, Lat. Laut- und Formenlehre 2 , München 1977, 31. In Handbuch der lat. Laut- und
Formenlehre, Heidelberg 1914, 221, il Sommer pone all'origine un *tweskwa. In DELL viene
proposto possibilmente un confronto con airi, terc (<*tersk-).
19 In Goetz-Schoell "eloquium ac reloqui". La questione è qui priva di rilievo.
20 Vd. Bruno, I Sabini, 78.
21 Vd. anche 9.1.
AL CAP. 3
1 Vd. per questo, di chi scrive, Le unità intermedie. 1. L'italico, Milano 1983, 32 ss., e la bibliografia
lì citata.
2 Rispetto p. es. all'accoglimento senza riserve che ne fa il Peruzzi in Mycenaeans, 14 ss.
3 Per o. hipust della Tabula Baritina vd. LIA 9 e Le unità intermedie, 32 ss.
4 Risp. in Emerita 2, 79-110 e in SSL 8, 16-130.
5 Emerita cit., 87.
6 REW 8731.
7 Vd., di chi scrive, ASGM 19, 27-37. Nomostante Giacomelli, Graeca Italica, 124, dubito fortemente
che sussista un rapporto fra la forma "sabina" teba e il toponimo (che ora il miceneo sembra
testimoniare con lbv.), che, proprio in ambiente eolico, appare con l'aspirata iniziale e, palesemente,
appartiene a un patrimonio toponomastico di ambito mediterraneo (uso il termine solo come
qualificazione geografica), secondo mostra il suo ricorrere anche in Egitto e in Cilicia.
8 Einleitung in die Geschichte der griechischen Sprache, 193.
9 Lo stesso ragionamento - fondato sulla spia fonetica di -gwh- > -b- - potrebbe farsi per febris, se <
*dhegwh-ri-s, ma vd. giustamente DELL, s.u. febris.
10 Trafere e vefere di Apul., De Not. Asp., 94, posto che non siano mere invenzioni, appartengono
piuttosto alla casistica dello scambio hlf. Montebufo, per cui vd. Bruno, I Sabini, 116, non ha titolo
alcuno a essere tenuto per sabino. Più interessante invece è il toponimo Μήφυλά in Dion. Hal.
1.14.3 (ma Medullia in Liu. Hist. 1.38  Maebula, vd. Bruno, I Sabini, 110??), che credo
interpretabile alla stregua di Farfarus. Vd. anche, da questo punto di vista, Strab. 5.3.1.
e Uarr. ap. Dion. Hal.
1.14.1
cui si potrebbe inferire una
possibile recenziorità degl'insediamenti urbani del reatino. In SE 43, 119 ss., il de Simone ha
sostenuto l'origine "umbro-sabina" delle varianti con -ƒ- del nome del Tevere del tipo *Tifaro-
(omologo secondario - latino *Tibaro > Tiber, etrusco ϑihvare). La "sabinità" è però riconducibile
alle idee esposte in 1.3-6, e per essa vale quanto lì detto.L'Ofens, contrariamente a quanto affermato
dal Devoto, Storia, 82, non appartiene ad area sabina, cf. Fest. 212 L. =P.F. 213 L. "Oufentinae
tribus initio causa fuit nomen fluminis Ofens, quod est in agro Priuernate mare intra et Tarracinam.
Lucilius (1260) "Priuerno Oufentina uenit, fluuioque Ofente"...". Vd. anche Uerg. Aen. 7.745. Per il
sito di Farfa vd. anche Bruno, I Sabini, 36, e le considerazioni di Moraldi, in Dialoghi di Archeologia
1987, 7-15.
11 L'accostamento fra il fitonimo e l'idronimo, per cui vd. Plin. NH 24.15.35, è oscuro. Vd. anche P.F.
78 L.
12 Non mi nascondo che un problema spinoso è costituito dal nome del dittatore albano, coevo di Tulio
Ostilio, Mettius Fufetius (Enn. Ann. 126 V., Metioeo Fufetioeo, Liu. Hist. 1.23) con -ƒ- interna. La
questione ovviamente non interessa il sabino (ancorché Mettius sia praenomen del sabino Mettius
Curtius in Liu. Hist. 1.12), ma parrebbe alludere alla presenza di un tratto marcatamente italico nel
Lazio dell'età regia, ciò che contrasterebbe con l'idea, più volte, e anche qui, da me espressa,
dell'estraneità degl' Italici rispetto alle vicende della Roma regia (a meno che la -ƒ- sia connessa in
qualche modo con l'area etnisca). Vd. anche Pallottino, in Civiltà arcaica dei Sabini I, Roma
1973,33.
AL CAP. 4
1 Che per il Devoto, p. es. Storia, 80 ss., rappresenta la più cospicua manifestazione della diffusione
nel IV s. in Roma di quella che l'illustre studioso chiama "moda sabina", e a cui dovrebbe attribuirsi
il definitivo accoglimento in latino delle varianti prodotte per commutazione automatica, sul modello
odor: olēre, del tipo lacrima in luogo di dacru(li)ma.
2 IF 2, 157-67.
3 P.es. Petr, BB 25, 127-58; Ernout, Eléments, 80-1; Palmer, Latin Language, 37-8; Devoto, Storia,
84-6. Più cauta la posizione dello Schrijnen, KZ 46, 376-80. Non prende posizione il Leumann, Lat.
Laut- und Form., 155-6. Contrari all'attribuzione del fenomeno al sabino il Sommer, Handb., 176-7,
e in parte il Peruzzi, Aspetti, 31. Corretto inquadramento del fenomeno in Bottiglioni, SE 17,
315-26.
4 Così il Kretschmer, KZ 31, 420.
AL CAP. 5
1 Latin Language, 37: "Sabine was further distinguished by its retention of intervocalic -s-...".
2 Vd. DELL, s.u. Lär. Sulla scorta di Uarr. αρ. Amob. Adu. Nat. 3.41 "Uarro... antiquorum sententias
sequens Laruas esse dicit Lares, quasi quosdam genios et functorum animas", e di forme quali Larunda
'mater Larum', Lara ecc., vi si avanza la possibiltà che le attestazioni con -s- del teonimo (come già
si è detto, nel Carmen Aruale, in Uarr. LL 6.2 vd. sopra, e in Quinci. Inst. 1.4.13) siano prodotto
d'iperarcaismo. Il che non mi pare impossibile.
AL CAP. 6
1 Eléments, 58-9.
2 Latin Language, 37. Vd. anche Devoto, Storia, 83.
3 Vd. Ernout, Eléments, 58.
4 Eléments, 191.
5 P. es. Victorie TL A 15. Vd. Emout, Eléments, 58.
6 P. es. efiles, efile di LIA 146 F. Vd. G. Giacomelli, La lingua falisca, Firenze 1963, 119-21.
7 Aspetti, 53 ss.
8 Su Oenotrus vd. Poucet, Origines, 191 ss.
9 Vd. DELL, s.u. Líber. Peruzzi, Aspetti, 166-8: "... Loebasius... non è la forma sabina del teonimo
Liber" bensì "un epiteto del dio, perché -s- esclude un nesso con Líber < *leudheros, che ha r
originario (come appunto loebertatem) ... Loebasius è derivato da *loibā con il suffisso
-asio-. Di Loebasius... non è dubbio il carattere sabino; anzi, esclusivamente sabino, perché né
l'epiteto né la forma *loibã da cui esso deriva sono usati a Roma. Ciò conferma una volta di più che
le voci saliari con e non rappresentano una fase intermedia del passaggio romano oi > ū, che si
262 MARIO NEGRI
sarebbe conservata in alcune forme latine... bensì uno sviluppo di oi caratteristico del sabino, 
almeno del sabino arcaico e di quello religioso ... (e perciò conservatore e pur esso arcaico)".
10 Vd. per queste attestazioni Peruzzi, Aspetti, 161-5.
11 Un'altra forma utile potrebb'essere moeros, ancorché in Uarr. RR 1.14.3 appaia la forma con 
"aggeres faciunt sine fossa: eos quidam uocant muros, ut in agro Reatino": moeros è invece in Enn.
Ann. 419 V. "matronae moeros complent spectare fauentes" (altrove mü-, Valhen 280) e in Uarr. LL
5.141 "oppida quod operi muniebant, moenia; quo moenitius esset quod exaggerabant, aggeres dicti,
et qui aggerem contineret, moerus. Quod muniendi causa portabatur, munus, quod sepiebant oppidum
eo moenere, moerus". Non sarebbe inutile la considerazione puntuale delle varianti dei manoscritti
con oe.
12 Eléments, 194.
13 P.es. neuen deiuo da Ardea III s. a.C, LIA 149 (non credo che il numerale 'nove' sia presente in
Nouensidēsl-lēs, secondo suggerisce possibilmente il Pisani). Vd. per l'intera questione del
trattamento di eu in falisco R. Giacomelli, Problemi di storia linguistica del latino dialettale. 1:
Ricerche falische, Firenze 1978, 28 e già RIL 109, 393-8. Vd. però la critica di Prosdocimi in Storia,
LXXVIII.
14 Vd., per la documentazione, Ernout, Eléments, 221.
15 Accolta sia in Goetz-Schoell sia in Kent.
16 In IF 53,117-22. L'integrazione del von Blumenthal è accolta dal Pisani in Paideia 14, 168-81, dove
si propone "sol a <sa>uel quod ita Sauini". Va da sé che, essendo sauel forma ricostruita, non può in
via di principio essere utilizzata per dimostrare alcunché. Comunque si sia, anche sol con ogni
probabilità non rappresenta una forma con monottongazione, vd. DELL, s.u. sōl (<*swōl), e ora
Bammesberger, KZ 98,111-13.
17 Per quel che vale, au appare conservato in aunom LIA 54, vd. ad l. nonché Wallace, 72-3.
AL CAP. 7
1 Vd. Bruno, I Sabini, 81. E'probabilmente del tutto casuale che sia stato un Claudio come il
personaggio qui ricordato, e cioè Appio Claudio Cieco, colui che, giusta Pomp. Dig. 1.2.2.36 "R
litteram inuenit". Va ancora notato, a proposito delle sorti del dittongo au, che la variante Clodius
appare soltanto in età tardo-republicana. Per quanto attiene alla priorità del praenomen Attius (Atta)
vs. Appius, conclamata in Liu. Hist. 2.16, non va dimenticata l'opinione della Marinetti, SE 49,
123, e Res Publica Litterarum 5.1,169-81 secondo cui si avrebbe "un rovesciamento nella tradizione
che assegna il nome Atta al sabino, mentre Appius sarebbe forma latina: in realtà l'originaria forma
sabina è Appius, e ciò è provato dapprima autonomamente nel latino, e confermato poi grazie al
l'apporto sudpiceno..." (cf. apaes, apauis, ecc.]) Se ciò fosse vero rientrerebbe bene nella mia visione
dei rapporti fra sabino e italico il fatto che il praenomen del volsco Tullius in Liu. Hist. 2.35 ecc. sia
Attius e non Appius.
2 Vd. Bruno, I Sabini, 82.
3 Lo stesso in Marsi, cf. Martius?
AL CAP. 8
1 E cf. Fircellius Pāuō Reātinus, interlocutore di Vairone nel De Re Rustica, vd. Bruno, I Sabini, 83-
4.
2 Codd. fedum, foedwn, Keil ad 1. Vd. la nota seguente.
3 Codd. anche foedum e hedo, edo, Lindsay ad 1. La frequenza della grafia oe anetimologica può essere
riguardata dal punto di vista di cui si è fatto cenno in 6.3.
4 Qui può aver giocato la paretimologia sufari, a meno che f'ariolus non sia effettivamente
appartenente a questa famiglia (posto ovviamente che sia accolto).
AL CAP. 9
1 Ma il problema è se il termine è davvero sabino  se il suo carattere sabino è limitato alla/- iniziale.
Nel primo caso, il termine per 'capretto' potrebbe essere utilizzato anche a proposito del trattamento
delle lbv.
2 Vd. Durante, LDIA, 792.
AL CAP. 10
1 Che ricorre - significativamente! vd. 1.6. - accanto a forme quasi osche del tipo fundatid, parentatìd,
proie citad (per *proiecatid), vd. TL, ad l.
AL CAP. 11
1 Dunque i Sabini per 'bonum' dicevano cyprum, i ueteres mānum giusta Uarr. LL 6.4 "bonum antiqui
dicebant manum", vd. però 3.3.
2 Solo per amore di completezza, dato lo spessore inconsistente della testimonianza, ricordo che,
riprendendo una idea dello Zander, il Pisani, TL 37, ha accolto la lettura uero nel tormentato ricordo
del Carmen Sahare in Uarr. LL 7.26, confrontando quindi questa forma "sabina" dal punto di vista
lessicale e da quello morfologico con u. ueru 'portas'.
AL CAP. 12
1 Per la tradizione sulle origini dei Sabini vd. Poucet, Origines, inoltre Berard, La Magna Grecia7,
Torino 1963,453-6.
2 Gli esempi più copiosi del fatto sembrano appartenere a una tradizione di tipo popolare, vd. da
ultimo, di chi scrive, KZ 96, 61 ss.
ELENCO DEI PASSI CITATI
Acc. 554 R3;

Arn. Adu. Nat., 3.38, 3.41;
August. Cd 18.17, 18.19;
Cens. 22.13.14;
Cie. De Fin. 1.9;
264 MARIO NEGRI
Dion. Hal. 1.11.3, 1.14.1, 1.14.3, 2.48.4, 2.49.2;

Enn. Ann. 120 V., 122 ss. V., 126 V., 265 V., 391 V., 419 V., 529 V.;
Fest. 64 L., 74 L., 116 L., 126-8 L., 132 L., 150 L., 188 L., 196 L., 212 L., 224 L., 228 L., 304
L., 316 L., 320-2 L., 334 L., 342-4 L., 358 L., 386 L., 424 L., 444 L., 456 L., 460 L., 464 L.,
492L., 496L., 510L.;
Gloss. Abol. L 2 ST 23 = V 515.62; Corpus Gloss. Lat.
Gloss. Ansil. L DR 13 = V 63.9 Corpus Gloss. Lat.;
Gloss. Nom. 32 Labh. = II 197.53 Corpus Gloss. Lat.;
Gell. NA 11.1, 13.23;
Hor. Od. 1.22.9-12;
Inc. De Praenom. 1.3, 1.4;
Is. Or. 1.27.14, 15.12.3;
Liu. Hist. 1.11, 1.12, 1.13, 1.20, 1.23, 1.38, 1.55, 2.16, 2.35, 8.9;
Luc. 9.10;
Lyd. De Mens. 1.21, 4.60, 4.90;
Macr. Sat. 1.3.13, 1.13.3, 3.18.13
Mar. Uict. 6.8 K., 6.9 K., 6.26 K.;
Non. 877 L., 894 L.;
Ou. Fast. 2.267-8, 2.361-6, 2.475, 4.907, 6.213-6;
P. F. 4. L., 8 L., 17 L., 18 L., 22 L., 24 L., 28 L., 41 L., 43 L., 44 L., 47 L., 49 L., 55 L., 56 L.,
59 L., 60 L., 61 L., 73 L., 74 L., 75-6 L., 76 L., 89 L., 91 L., 93 L., 96 L., 102 L., 108 L., 109
L., I l l L., 117 L., 121 L., 127 L., 133 L., 157 L., 189 L., 197 L., 213 L., 305 L., 317 L., 321
L., 325 L., 335 L., 339 L., 343 L., 387 L., 437 L., 455 L., 457 L., 461 L., 489 L., 493 L., 503
L., 511 L.;
Plaut. Stich. 230;
Plin.NH 3.99, 3.110, 14.37, 14.38, 15.3.13, 24.15.35;
Plut. Num. 21, 22; Quaest. Rom. 103;
Pomp. Dig. 1.2.2.36
Porph. ad Hor. Epist. 1.10.49;
Ps. Acr. ad Hor. Epist. 1.14.19;
Ps. Plac. 4.59.4 L;
Quinct. Inst. 1.4.13, 1.5.56;
Seru. ad Aen. 1.6, 1.532, 3.235, 7.517, 7.657, 7.684, 7.695, 7.706, 7.715, 8.343, 9.567, 11.785,
12.538; ad Georg. 1.151;
Strab. 5.3.1, 5.4.2, 5.4.12;
Suet. Tib. 1.1;
Tac. Ann. 4.9;
Ter. Scaur. 7.11 ., 7.13 ., 7.28 .;

Uarr. LL 5.41, 5.51, 5.55, 5.66, 5.68, 5.73, 5.74, 5.84-6, 5.97, 5.100, 5.106-7, 5.113, 5.123, 5.128,
5.141, 5.149, 5.159, 5.163, 5.177, 6.2, 6.4, 6.5, 6.13, 6.15, 6.16, 6.28, 6.34, 6.52, 6.57, 6.68,
6.83, 7.3, 7.10-1, 7.26, 7.27, 7.28-9, 7.29, 7.43, 7.46, 7.77, 7.83, 7.96; RR 1.1.6, 1.2.9, 1.14.3,
2.5.6, 3.1.6, 3.9.19;
Uel. Long. 7.69 ., 7.77 ., 7.81 .;
Uerg. Aen. 7.706-9, 7.709, 7.745.
TRA STORIA E PREISTORIA LINGUISTICA:
IL VALORE CONCETTUALE DELLA RADICE *pk- NELL'ITALIA ANTICA
MARIA LUISA PORZIO GERNIA

Università di Torino
0. Nello sviluppo della radice l'italico1 e il latino presentano consonanze particolari

che, rilevanti sul piano linguistico e culturale, distinguono queste lingue dalle altre lingue
indoeuropee. Fino ad ora il problema non è stato studiato esaurientemente; in particolare,
non è stata presa in considerazione l'ipotesi di un contatto interlinguistico tra l'italico ed il
latino. Gli strumenti metodologici di cui può disporre il linguista non sono stati applicati
in tutta la loro varietà e potenzialità. Ne consegue che la storia di *PÃK- nell'Italia antica è
rimasta in parte ignorata, in parte assorbita in prospettive metodologiche che l'hanno
illuminata solo parzialmente, se non deformata del tutto.
Si cercherà di interpretare l'isoglossa italico-latina rappresentata dai derivati di
ricorrendo a nozioni fondamentali della linguistica storica e utilizzando metodi di varia
ispirazione, spesso antitetici, ma integrabili e quindi tali da contribuire, pur da ottiche
diverse, alla soluzione di un problema particolare di linguistics storica.
L'analisi si svolge seguendo questo schema:
1. definizione della famiglia lessicale che ha sviluppato nell'italico e nel latino, e
delle strutturazioni semantiche ad essa connesse
2. revisione critica dellaricostruzioneproposta dai dizionari etimologici
3. interpretazione dell'isoglossa italico-latina e della sua rilevanza culturale e ideologica
nella preistoria, protostoria e storia linguistica dell'Italia
268 MARIA LUISA PORZIO GERNIA
1. La radice si presenta particolarmente produttiva nei dialetti italici, ma non in

modo uniforme perché i derivati sono attestati in umbro e nei dialetti medioitalici, mentre
in osco compare una sola forma, di dubbia e difficile interpretazione.
Nelle preghiere, recitate nelle cerimonie dedicate alle varie divinità umbre e conservate
nelle Tavole Iguvine, ritorna insistentemente l'invocazione futu fos pacer pase tua, 'esto
fauens propitius pace tua' dove sono attestate le forme derivate da pase 'pace' e
pacer 'propitius'. Pase è ablativo di un tema radicale che compare anche nel latino
e indica la benevolenza della divinità che si cerca di impetrare con le cerimonie e i
sacrifici. Pacer è nom. sg. del tema *pakri- (nom. plur. pacrer VI b 61) ed è l'attributo
della divinità 'benevola, ben disposta'. I due termini compaiono quasi sempre nello stesso
contesto2. Pacer è documentato da solo nella formula fos sei pacer sei 'fauens sis
propitius sis' (VI a 23; ricorre pure, pressoché identica, in VI b 7, 26).
Solo in un luogo (VI a 20: uasor uerisco treblanir porsi ocrer pehaner paca ostensendí)
nelle Tavole Iguvine è attestata la parola paca, che presenta notevoli difficoltà
interpretative. Se i linguisti sono concordi nell'attribuire paca alla radice non lo
sono altrettanto sul suo significato e sulla sua funzione. La maggior parte di essi riconosce
in paca un ablativo sing. femm. usato come preposizione (cfr. lat. causa, gratiā), collegato
al genitivo precedente con il senso 'ocris piandae causa'. Il Devoto (1962:174), preferisce
interpretare paca come ablativo di un nome astratto *paka- che avrebbe il valore di 'id
quod inter deos et homines firmum est', cioè ritus, e si riferirebbe ad ostensendi (fut. in
-s-)3. Traduce quindi paca con rïtë4. Il passo suonerebbe dunque 'uasa quae ad portam
Trebulanam acris piandae rite ostendentur'.
L'interpretazione del Devoto è accolta dalla Viano (172), che pone in rilievo il valore
sacrale del termine, che fa riferimento alle intenzioni e ai sentimenti della comunità dei
fedeli che intendono propiziarsi la divinità con riti e preghiere. L'analisi della famiglia
lessicale di nel suo complesso porterà elementi di conferma a questa ipotesi
etimologica.
A causa della natura delle fonti, ovviamente in umbro non sono documentate
formazioni onomastiche.
Nei dialetti medioitalici è più volte attestato il tema aggettivale *pakri- 'propizio'. In un
gruppo di iscrizioni è riferito alla divinità, nella formula aisos pacris 'di propitii'.
Compare nella iscrizione nota come Bronzo di Rapino (Pisani 1964, n. 52), trovata in
TRA STORIA E PREISTORIA LINGUISTICA 269
territorio marrucino (r. 1 aisos pacris 'di propitii'; r. 11 pacrsi 'ut propitia sit'). La
medesima formula aisos pa(cris) si trova in una sors su dischetto di piombo rinvenuta in
territorio frentano, in località assai prossima al confine marruccino dove si trova Rapino
(Poccetti 1979:82 s.). Si ripete ancora in un cippo augurale marso (Pisani 1964, n. 5):
esos nouesede pesco pacre 'Di Nouensides templum propitii'. Pare dunque che si tratti di
una formula sacrale stereotipa usata in particolare nell'ambito delle pratiche divinatorie, in
cui l'invocazione agli dei assume un carattere propiziatorio (Poccetti 1979:83).
In due iscrizioni sepolcrali peligne l'aggettivo *pakri-, nel senso di 'ben disposto,
benevolo' è riferito, invece che agli dei, alla buona disposizione degli uomini, che si
fermano a leggere l'iscrizione tombale. Una proviene da Sulmona (Pisani 1964, n. 46,
r. 5) peo [i]s [ri]d (ablativo avverbiale dal tema *pakri-) 'piis propitie'. L'altra (Pisani
1964, n. 47, r. 6) è l'iscrizione sepolcrale di Vibia Petiedia: eite uus pritrome pacris puus
ecic lexe lifar 'Ite uos in laetitiam beneuoli quos hoc legisse libeat'.
Nell'area dei dialetti medioitalici la radice è attestate anche nell'onomastica: pel.
Ρacia, mars. Pacuies; in forma abbreviata marso, pel. pa.y pel. p.
L'osco assume una posizione particolare rispetto agli altri dialetti italici. La radice
■ è riconosciuta in PRUPUKID, che nel Cippo Abellano (r. 2) compare accanto al
nome di un funzionario. In PRUPUKID non sono chiari né la struttura morfologica

(<*pro-pak-іo-  *pro-pakid) e la -U- della radice5, né il significato. E'comunemente
interpretata come ablativo avverbiale, variamente tradotto ('ex antepacto', 'ex praefinito',
'pro pacto' ecc.). M. Vestricio, il magistrato abellano cui il termine è riferito, sembra cioè
essere un funzionario scelto 'in seguito a convenzione'.
Interessante è la proposta del Corssen (1864:162), ripresa dalla Viano (1953:173), di
attribuire a pro- valore locale (attestato in italico e in latino: ad es. o. PRÜFATTED, u.
PRUZURE, PRUSEKATU ecc.; lat. prõscribere, prōmulgāre ecc.). L'o. PRUPUKID
assumerebbe così il significato di 'deliberazione pubblica' e porrebbe in rilievo il carattere
pubblico dell'accordo relativo alla designazione del funzionario di Nola. Il termine osco,
se accettiamo l'ipotesi, assumerebbe un valore particolare per la storia del concetto di
'patto' nell'antichità (1953:173): l'osco cioè avrebbe indicato il patto stipulato dai
rappresentanti della città con un termine derivato da come in latino (v. pactum e
pāx). Nonostante le difficoltà interpretative, restano indubbie la pertinenza giuridica di
PRUPUKID e il suo collegamento con la radice La radice è molto vitale
nell'onomastica osca: PAKIS, PAAKUL, PAKULLIIS; in forma abbreviata

., ., PK.
1.1. La radice sviluppa in latino una ricca famiglia lessicale, costituita da forme
verbali e nominali. Il verbo *pacere, attestato nelle XII Tavole, è un termine del diritto
privato e si riferisce alla riconciliazione che segue ad una lite tra privati (I, 6 rem ubi
pacunt, orato 'nel caso che concludano l'affare con un accordo, si proceda'; VIII, 2 ni
cum eo pacit talio esto 'se non conclude un accordo con lui, si applichi la legge del
taglione'. Già in Plauto è sostituito da pacīscor.
Le forme nominali sono numerose. Pāx, in quanto nome radicale, presenta una
struttura morfologica antica; è un nome di azione che indica originariamente il fatto stesso
di concludere un accordo tra due contendenti (solo in un secondo tempo passerà ad
indicare lo stato di quiete che ne deriva). L'uso del termine in senso giuridico sembra
antico: fin dall'età arcaica pax indica l'atto della 'convenzione fra due stati, che ha lo
scopo di porre fine alle ostilità (cfr. Ennio in Varr. 1.1. 7, 41 orator sine pace redit).
Da pāx deriva il verbo pācō-ās 'pacificare, sottomettere', attestato a partire da
Cicerone e continuato nelle lingue romanze (ad es. it. pagare).
Nell'iscrizione di Dueno si legge chiaramente pakari, infinito deponente  passivo, di
norma interpretato con ä (pǎcāri). L'Ernout-Meillet (s.u. *pacõ) lo indentifica invece, in
forma dubitativa, come voce di pācō, -ās.
Oltre a pāx, sono documentati i termini paciō-ōnis, attestato da Festo (296,35) come
forma antica di pactiõ-õnis e già usato da Plauto, e pactum-ī 'convenzione, patto' (neutro
del participio pactus di pacīscor) che compare nelle  Tavole.
Applicati già in età arcaica alla lingua del diritto sia privato che pubblico, i termini
latini conservano tuttavia qualche traccia di un antico valore sacrale. Pāx esprime anche il
concetto religioso della 'benevolenza degli dei', come dimostra l'uso dei sintagmi pax
deum  diuom e tua pax riferito alla divinità (su queste formule v. sotto § 3.3), che
appartengono alla tradizione della lingua sacrale. Pāx, in questa accezione, è strettamente
connesso con l'u. pase.
Sono stati inoltre rilevati (Viano, 1953:178 s.) contatti tra paclscor e spondeō, termine
della lingua sacrale che indica l'atto di 'prendere un impegno solenne'. Frequentemente
infatti paclscor sostituisce spondeō quando si tratta di un impegno solenne7. Le
connessioni tra pacis cor e spondeō risultano evidenti anche dal fatto che il latino usa
pactum per indicare il contratto redatto nella forma della stipulatio. Nell'esempio
tramandato da Plauto (Poen. 1157 AG. Tuam mihi maiorem filiam despondeas. HA.
Pactam rem habeto. AG. Spondesne igitur? HA. Spondeo) spondeō si riferisce
all'espressione orale dell' impegno solenne che si assume per quanto riguarda il contenuto
del patto, indicato invece da pacīscor (v. Viano, loc. cit.). Inoltre il sostantivo pactum,
che esprime una nozione fondamentale del diritto romano, è inteso come sinonimo di
conuentio (da *con-uenīre)come dimostra il fatto che i due lessemi sono spesso accostati
 interscambiabili. L'identità dei due termini e delle nozioni che essi esprimono dimostra
che in pactum sopravvive il concetto della reciproca disposizione benevola dei contraenti
(Viano, 1953: 177 s.).
A questo punto si può avanzare qualche ipotesi su pakari, attestato nell'iscrizione di
Dueno, un documento assai discusso per quanto riguarda sia la natura stessa sia
l'interpretazione del testo.
Il problema di pakari è discusso dalla Viano (1953:174 ss.) che nota come nei
numerosi tentativi di interpretazione siariconosciutocostantemente il carattere magico del
documento; muovendo da questo punto di vista riprende l'ipotesi che la uirco sia
Proserpina e che si tratti di uno scongiuro rivolto a divinità infernali. A sostegno di
un'interpretazione sacrale adduce vari argomenti: analogia con il Carme S aliare e con le
tabellae defixionum, in particolare con la defixio osca di Vibia. In questo contesto anche
pakari assumerebbe un valore sacrale: non semplicemente 'riconciliarsi' ma, come
suggerisce la Viano (1953:175), 'placare, rendere benevolo' con un riferimento specifico
alla divinità. L'ipotesi trae conferma dal valore della radice conservato nelle lingue
italiche e nel latino.
Dal confronto della famiglia lessicale italica e latina si deduce che i significanti
raggruppati intorno a ■ presentano isomorfia solo nel caso del tema radicale
rappresentato dal lat. pax e dall'u. pase; non è da escludere l'ipotesi che un tema in -ā-,
supposto nell'u. paca e poi scomparso nel latino, sia alla base del lat. pakari. Essi
ricoprono però strutturazioni semantiche analoghe: non si riferiscono alla cultura
materiale, ma alla sfera etico-religiosa; presentano come nucleo semico fondamentale il
concetto 'essere ben disposto'.
Dall'esame che fino a questo punto è stato condotto sui documenti, nell'ambito della
storia linguistica,risultanoevidenti e profonde le corrispondenze tra il latino e l'italico: nei
due ambienti culturali e nelle due tradizioni linguistiche pur nella diversità degli
sviluppi derivazionali, assume una funzione assai rilevante nella sfera sacrale-giuridica,
con un valore che non è riscontrabile in altre lingue indoeuropee.
Per proporre un'interpretazione di questa isoglossa occorre anzitutto procedere
all'analisi delle ipotesi prospettate dai dizionari etimologici, che raccolgono il risultato
dell'applicazione del metodo comparativo tradizionale alla ricostruzione della preistoria
della famiglia lessicale che qui ci interessa.
2.1 dizionari etimologici8, sia quelli che riguardano le lingue indoeuropee in generale sia
quelli specifici di lingue particolari, non senza difficoltà e contraddizioni riferiscono con
cordemente i termini latini ed italici che abbiamo analizzato ad una unica radice che pre
senterebbe un'alternanza di occlusiva sorda e sonora9: il
cui significato originario sarebbe 'conficcare, piantare in terra, fissare'.
Le voci latine ed italiche sono così collegate ad unaricchissimafamiglia lessicale nella
quale viene incluso un numero molto elevato di voci. Sempre nell'ottica di una radice
unitaria, la connessione di pāx con pangere è stata recentemente riaffermata10.
Il quadro etimologico preistorico, che si delinea su questo sfondo, si presenta assai
vasto e complesso sul piano sia formale che semantico. Le voci italiche e latine sono
interpretate secondo una prospettiva, tipica della ricostruzione etimologica indoeuropea,
che tende a riportare all'unità la pluralità delle forme attestate. Questa reductio ad unum,
che ha dominato a lungo la linguisticaricostruttiva,consiste nel raccogliere e raggruppare
sotto una unica radice, cioè un unico significante originario il maggior numero possibile di
significanti. Questo tipo di ricostruzione che pone sullo stesso piano, in modo
indifferenziato, gli elementi da comparare, opera sul presupposto dello schema
genealogico; inoltre, fondato esclusivamente sull'analisi del significante, lascia in ombra
l'aspetto significativo del segno sia per quanto concerne il valore della forma originaria sia
nella tradizione delle lingue storiche considerate.
Poiché l'estensione dell'analisi alla preistoria linguistica rappresenta un criterio
metodologico spesso determinante per la soluzione di problemi di storia linguistica, anche
per l'interpretazione dell' isoglossa italico-latina sarà utile proporre una nuova
organizzazione dei dati che i dizionari etimologici, pur nei limiti sopra indicati, mettono a
disposizione dello studioso; si dovrà, in particolare, approfondire l'analisi con l'applica-
zione di procedimenti metodologici che consentano di delineare prospettive cronologiche,

fenomeni innovativi e i loro centri di irradiazione, contatti e interferenze linguistiche fra
tradizioni diverse.
2.1. Sia dal punto di vista semantico, sia da quello formale pare più convincente
ricostruire due radici distinte, , ciascuna delle quali
dotata di un proprio significato. Questa ipotesi era già stata avanzata in passato dal
Meringer (1904:176; 1907:311 s.) che distingue *PĀG 'battere con l'ascia, costruire una
intelaiatura' riferita alla tecnica della carpenteria e PÄK 'intrecciare, legare'. L'ipotesi non
convince sul piano semantico perché, troppo condizionata dall'ottica 'Wörter und Sachen'
e 'Geschichte der Sachen', privilegia l'aspetto tecnico del lavoro manuale; giustamente,
pertanto, è concordemente rifiutata dai dizionari etimologici.
Anche la proposta della Viano (1953:169 ss.), che distingue *PÃG-con significato
materiale 'piantare, congiungere' e *PÄK-, di estensione più ridotta, con un significato
spirituale 'placare, rendere benevolo', non è esauriente. Soprattutto, come si vedrà, non è
verosimile che il significato spirituale supposto per *PÃK- sia il più antico, e che le forme
di *PÃK- con significato materiale abbiano carattere recenziore, in quanto si sarebbero
sviluppate per contatto con la famiglia di *PÃG- (come cercheremo di dimostrare, è il
valore sacrale di *PÃK- che assume un carattere innovativo).
L'ipotesi di due radici distinte può reggere se si individuano altri nuclei semantici
originari, diversi per ciascuna delle due radici, nei quali si possano conciliare gli esiti delle
lingue storiche, molteplici a livello di significato e di significante.
Per *PĀG-/P G- si propone il significato 'piantare, conficcare', per
'legare, unire'. Dalla considerazione dei dati raccolti dai dizionari etimologici, i derivati di
*PÄG- non solo si distinguono da quelli di *PÄK- sul piano del significante, ma
delimitano una sfera semantica diversa; inoltre non è identica la distribuzione geografica
delle due famiglie lessicali in area indoeuropea. Durante lo sviluppo storico,
particolarmente complesso, delle due radici, hanno avuto luogo fenomeni di contatto
intralinguistico e interlinguistico, processi di estensione semantica, di simbolizzazione e
astrazione, nonché di contaminazione per contiguità semantica.
2.1.1. Lo schema dei valori semantici sviluppati da *PÄG-/P G- 'piantare, conficcare'

sarebbe il seguente: -
a1 'pala', 'palo' (=ciò che si conficca)

A. /piantare, conficcare/ 2 'pietra di confine, confine'
3 'recinto'; 'chiudere, recintare'
b1 'commettere, costruire'; 'commessura,

incastro, impalcatura, telaio'
b2 'essere  rendere solido, rigido,

compatto'; 'coagulare'
b. /fissare/
b3 'pattuire'
b4 'bloccare, arrestare'; 'trappola, calappio'
Le formazioni derivate da *PÄG- appaiono così distribuiete nelle lingue indoeuropee (le
lingue sono citate incominciando da quelle in cui lo sviluppo è maggiore; i lessemi sono
raggruppati secondo i nuclei semantici rappresentati nello schema):
Greco:
A 'pianto, conficco
a 1 'supporto';
b 'fisso', 'fissato';
b1 'commetto, costruisco', 'commessura, impal
catura, telaio', 'costruito', 'compaginazione';
b2 'rendo compatto, condenso', 'indurito, coagulato',
'ghiaccio, brina', 'rupe, roccia' (=ciò c h e è fissato, duro?,
v. Frisk s. u.), 'gelo, freddo', 'ghiacciato',
'solido, compatto', 'cacio fresco', (dor.
'coagulato, condensato', 'latte coagulato, formaggio',
'gelo', 'solido, compatto', 'gelo, ghiaccio'
'coagulamento', 'brina rugiada';
b ' 'concludo un patto';
b4 'laccio, trappola'; 'rete per uccelli'.
Latino:
A pangō 'pianto, conficco';
a1 p āla < *pāg-slā- 'pala', pālus 'palo', prōpāgēs 'piantone',
pastinum 'zappa';
3 pāgus '(villaggio, territorio rurale delimitato da confini)'11;
b pangō 'fisso';
b1 pagina 'vite a spalliera', compingō 'commettere, riunire', compāctiō
'commessura', compāges compāgō 'commettitura', impāgēs 'traversa
della porta';
b pangõ 'concludo un patto'.
Germanico:
3 sass. a.fac, a ted. a. fah 'recinto' (ted. mod. fach); sass. a.fœc 'cinta,
chiusura'; fris. a. fek, fak 'scompartimento'; ags. fœc, ingl. a. fœk 'epoca,
spazio di tempo'.
Slavo
b1 ceco, pol., russo paz 'incastro, giuntura'; slov. pâž 'incastro, assito'.
Baltico:
1 lit. pozas 'incastro'.
Tocario:
a1 toc. A pyākäs 'palo, colonna'12.
Nell'italico la radice non è documentata.
La radice si distribuisce dunque nell'area indoeuropea centrale e occidentale13. Nel
greco presenta il massimo sviluppo, nel sistema derivazionale morfologico e come
numero di campi semantici. Il latino è caratterizzato pure da un ricco sviluppo di *PÃG-.
Le corrispondenze formali con il greco sono molte e pangō ,
e-pig- e pactus, e pãgus). Sono scarse però le connessioni sul piano
semantico, se si escludono le coincidenze, probabilmente per un processo avvenuto
indipendentemente nelle due lingue, tra e pangõ (b), v. sotto § 2.2, e l'uso
nella lingua della carpenteria (bl). Nel latino *PÄG-si sviluppa nella lingua tecnica
dell'agricoltura; mancano derivati nella lingua della caccia (b4) e tutti quelli connessi con il
nucleo semantico b2, assai produttivo nel greco. Nel greco mancano i valori semantici che
abbiamo indicato come a2 e a3. Questi valori 'confine, recinto, recintare' costituiscono
invece una calzante isoglossa latino-germanica (lat. pāgus, sass. a. fac, a. ted. a. fah).
Nello slavo e nel baltico sopravvivono solo alcune tracce nella lingua della carpenteria.
2.1.2. Lo schema dei valori semantici sviluppati da *PAK- 'legare, unire'
sarebbe il seguente:
A. /legare, unire/ 'fune, legame, laccio, catena'
'chiudere'; 'gancio, piuolo, chiavistello (=ció che serve

per chiudere  a cui si lega qualcosa)'
'prendere, catturare'; 'cattura, preda'; 'laccio, trappola'
B. /fissare, fermare/
'commettere, costruire, congiungere';
'impalcatura, incastro'
'convenire, andare bene, bello'
'adattare'
'ben disposto'
Le formazioni derivate da *PAK- appaiono così distribuite nelle lingue indoeuropee (si
usano gli stessi criteri applicati nel paragrafo precedente)14:
Germanico15:
b2 isl. a. fā ingl. a. fōn, fris. a. fā, got. as., a. ted. a.fāhan, as. fangan 'pren-
dere, catturare'; isl. a. fengr. ingl. a.fan 3 , fris. a. fang, got.fengr,fang, a.
ted. a. fang 'cattura';
b ingl. fan,fris.Ά. fōga, a.fōgian, a. ted. a fuogen, mod. fügen 'con
giungere, adattare', a. ted. a. fuoga, mod. fuge 'commessura, incastro';
b4 isl. a.fagr, ingl. a..fœgr,fœger, mod. fair, as., a. ted. a. fagar 'splendente,
bello';
b4" got. fagrs
Latino ed italico:
La radice sviluppa in particolare i valori semantici indicati come b4", con specifica
connotazione giuridico-sacrale (v. sopra, § 1).
Greco:
Le voci greche seguenti si possono attribuire senza difficoltà a *PÃK-:
b e b 1 'chiudo, lego saldamente', 'fermare, fissare';
'piccolo chiavistello' (il lat. pessulus 'chiavistello, catenaccio' è
un grecismo).
Indoiranico:
A av. pas- 'fissare l'un l'altro, congiungere';
a1 scr. 'fune, legame' (den. 'lega',pāśáita- 'legato'), av.
'con catene'; pers. m. fšīh 'legame, catena'.
La concordanza tra le aree laterali, costituite dall'indoiranico e dal germanico occidentale,
attesta che il significato più antico è /legare, unire/.
La radice *PĀK- presenta il massimo sviluppo nel germanico dove, accanto a
significati che si riferiscono alla sfera materiale, è attestato però anche un valore
simbolico-metaforico nella formazione aggettivale *pakro-, che assume un particolare
rilievo per il nostro problema. Essa è rappresentata dall'ingl. à. fœgr faegr, as., a. ted. a.
fagar 'splendente, bello' e dal got. *fagrs adatto, conveniente'. La forma
gotica ricorre al nom. ntr. sing, fagr in L. 14, 35 ed è riferito al sale che, se diventa
scipito, nih du air pai, ni du maihstau fagr ist
'non è più adatto né per la terra né per il concime'. La qualità
indicata dall'aggettivo in questo passo è dunque attribuita ad una sostanza materiale. Ma la
forma negativa dell'aggettivo, *unfagrs, attestata in L. 6, 35 è chiaramente riferita agli
uomini: 'jah wirpip mizdo izwara managa, jah wairpip sunjus hauhistins, unte is gods ist
paim unfagram jah uselijam
la
vostra ricompensa sarà grande e voi sarete i figli dell'Altissimo, che è buono con gli
ingrati e con i cattivi'. Si profila quindi chiaramente la connessione del gotico col tema
italico *pakri-. Il gotico e l'italico, all'interno della comunità linguistica dell'Europa
centrale, attribuiscono alla forma aggettivale derivata da *PAK- un particolare valore

morale 'ben disposto', che l'italico connota in senso sacrale.
2.2 La riorganizzazione dei valori concettuali e l'analisi della distribuzione dei derivati
confermano a nostro avviso l'ipotesi che *PÃG- e *PÃK- siano due radici distinte e che
abbiano una preistoria diversa.
Non si può escludere che tra le due famiglie lessicali abbiano avuto luogo fenomeni di
interferenza all'interno di singole tradizioni linguistiche. Un caso è costituito dall'uso del
lat. pangō nel senso traslato di paclscor (v. i sintagmi confoedus, amicitìam). Per quanto
riguarda il latino l'accostamento trapangōe *pacere, paclscor può essere stato facilitato
dal fatto che il part, pactus, da cui deriva il sostantivo pactum, appartiene tanto a pangō
quanto a *pacere e paclscor. Per tutta la tradizione classica perdura la consapevolezza di
un contatto tra i derivati di *PÄG- e *PÄK- (v. Viano 1953:177).
Contatti tra le due aree semantiche possono essere dovuti a fenomeni di contiguità
semantica poiché le singole radici hanno generato tratti semici comuni. In questo modo si
spiega come da due nuclei diversi /piantare/ e /unire/, attraverso un tratto /fissare/, si possa
avere come esito comune /commettere, costruire/ e /bloccare, arrestare/ attestato in
ambedue i gruppi.
Che /fissare/ sia il nucleo generatore di /costruire/ e /essere solido/ è dimostrato dal gr.
nome d'azione che dal significato 'atto di fissare, consolidare' ha assunto i due
significati diversi 'compaginazione' e 'coagulamento'; inoltre dal gr. nel suo
duplice valore 'costruito' e 'coagulato'.
Che da /fissare/ si sia generato oltre a 'coagulare' anche 'bloccare' è documentato dal
gr. che significa 'latte coagulato, formaggio' e 'rete per gli uccelli'. La massima
gamma di sviluppi si condensa nel gr. , 'pianto', 'fisso', 'costruisco', 'con-
denso' che estende la sua area semantica pure a valori simbolici con il senso 'pattuisco'.
Anche in questo caso viene confermato il carattere innovativo del germanico rispetto
alle altre lingue indoeuropee, in particolare a quelle del gruppo occidentale che in una certa
epoca della preistoria dovettero costituire una unità culturale, di cui l'italico e il germanico
sembrano formare il nucleo16.
I rapporti del germanico con il latino e quelli del germanico con l'italico si configurano
in modo diverso. Mentre i primi appaiono poco rilevanti e limitati alle lingue nordiche, la
maggior parte delle parole comuni solo all'italico e al germanico sono estese a tutti i
dialetti germanici. Si deduce che l'italico è stato a contatto con il germanico in un'epoca in
cui già tendeva ad una certa unità; poiché le connessioni riguardano termini di cultura
elevata, giuridici, politici e religiosi, tra oscoumbro e germanico devono essere esistite
relazioni di notevole rilievo culturale (Molinari 1965:391 s., 397 s., 400). Tra i dialetti
germanici, l'italico appare più vicino al gotico (Molinari 1965:392 s.).
3. L'innovazione, costituita dalle forme aggettivali17 comuni all'italico e al germanico (in

particolare al gotico), presenta le stesse caratteristiche delle isoglosse individuate dalla
Molinari. Appare quindi probabile l'ipotesi che lo sviluppo in senso morale della radice
*PÄK- 'essere ben disposto' appartenga originariamente alla tradizione linguistica italica,
che le dà una connotazione sacrale, e che processi di contatto e interferenza l'abbiano
estesa al latino dopo la discesa degli Osco-Umbri in Italia18.
Delle tre soluzioni che a priori si prospettano per l'isoglossa italico-latina (innovazione
interpretabile come reazione di un sostrato comune, innovazione latina estesa per contatto
all'italico19, innovazione italica estesa per contatto al latino) l'analisi della preistoria
linguistica fa dunque propendere per l'ultima: risultano evidenti sia il carattere innovativo
dell'isoglossa italico-latina rappresentata da sia la sua collocazione nel quadro
storico dell'Italia antica sia il suo irradiamento dall'area italica. Del resto, una lunga
tradizione di studi ha dimostrato che la maggior parte delle isoglosse che collegano il
latino all'italico si sono formate in Italia. Molti fenomeni linguistici attestano infatti
l'esistenza, anteriormente all'espansione romana, di un clima storico omogeneo, di aree di
affinità linguistico-culturale, diverse per età ed estensione.
3.1. Se prendiamo in considerazione la distribuzione dei derivati di nell'Italia
antica, risulta evidente che il valore sacrale è attestato nel latino, nell'umbro e nei dialetti
sabellici; è assente invece nell'osco, dove alla radice è attribuibile una sola forma, non
molto chiara sul piano semantico (v. sopra).
Questa distribuzione mostra una singolare coincidenza con quella di un'altra radice,
che esprime un concetto fondamentale della religiosità, cioè 'purificare,
compiere un sacrificio a scopo di purificazione'20.
E' assai diffusa nell'umbro (dove forma *piãom 'purificare' e pihaclom 'sacrificio a
scopo di purificazione'), nei dialetti minori (marr. peai 'piae', pel. pes 'pius', volsco
pihom 'pium') e nel latino dove troviamo forme corrisponenti a quelle italiche (pius,
piāre, piaculum). In osco presenta invece scarse tracce (è documentata in un epiteto, poco
chiaro, di Giove nella Tavola di Agnone, DIÚVEÍ PIÍHIÚÍ, e nell'avverbio πεhεδ 'pie',
attestato in una iscrizione lucana Pisani 1964, n. 7).
La corrispondenza delle forme italiche con il gr. τίνω 'pagare, espiare', ποινή
'punizione, prezzo fissato per l'espiazione', scr. 'vendicare, punire' e
'retribuzione' consente di ricostruire per un valore fondamentale 'espiare,
compensare, pagare l'ammenda'. Solo l'italico connota la radice in senso sacrale. Le
forme latine corrispondenti con p- iniziale, in luogo della labiovelare propria del latino,
provano fenomeni di interferenza tra italico e latino, in particolare l'origine italica dei
termini21.
Sia denunciano un'origine italica e riflettono un
rapporto di affinità culturale e linguistica certo assai antica: vi sono coinvolti l'umbro in
primo luogo, i dialetti sabellici e il latino, in posizione marginale e ricettiva rispetto ai
fenomeni che si irradiano dall'italico.
Caratterizzata dalla stessa distribuzione (mancano però attestazioni nei dialetti medio
italici) e dall'affinità semantica con la radice *FAU- che esprime
l'atteggiamento di benevolenza della divinità 'essere ben disposto, favorevole'. Tanto in
umbro quanto in latino appartiene alla lingua religiosa. Ad essa si connette l'u. fos, fons
'fauens', foner n. pl. 'fauentes', aggettivo che compare nella stessa formula in cui è
attestato pacer; il gen. foner 'fauentis', attestato in VIa 20, 23, 33, 36, dal Pisani (1964:
186) è attribuito ad altra radice e tradotto 'precationis'. La forma umbra deriva da *foni <
*fau-ni- (il Walde-Hofmann, s.u.faueō, attribuisce all'umbro il vocalismo o, *fou-, che è
presupposto pure dal Peruzzi 1977: 90). Il lat fauēre ha pure una connotazione religiosa:
cfr. Enn. ann. 291 placata Iuno coepit fauere Romanis; Ον. met. 6, 325; 9, 281 faueant
tibi numina; epist. 3.88 Marte fauente. I dati non ci consentono di determinare quale sia la
tradizione linguistica originaria cui appartiene. L'etimologia non è sicura:
la corrispondenza con lo si. a. gověti 'religiose uereri' è ambigua e incerta sul piano
fonetico, cfr. Ernout-Meillet s.u.faueō; esistono inoltre difficoltà per quanto concerne il
vocalismo della radice latina e umbra dove alternano a ed o, il cui rapporto non è chiaro22.
In ogni caso, la coincidenza della distribuzione geografica e la pertinenza sul piano
semantico indicano che questa radice ha assunto, in un'area determinata dell'Italia antica,
uno specifico valore funzionale all'interno di un certo contesto ideologico-culturale.
Altre implicazioni interessanti possono derivare dall'esame di alcune radici che
esprimono nozioni fondamentali dell'ideologia religiosa dell'Italia antica. In Porzio Gernia
(1961:133 ss.)23 è stata posta in luce la posizione reciproca delle aree latina, umbra ed
osca per quanto riguarda la funzione in esse assunta dalle radici *AIS- *SÄK-
/FËS-. La radice *AIS- è assai vitale in umbro, in etrusco e nei dialetti medioitalici; è
quasi assente in area osca, manca in latino (Porzio Gernia 1961:107). *SÄK-, invece,
mostra un grande sviluppo in osco e in latino24; è attestata in etrusco e presenta poche
tracce in umbro (Porzio Gernia 1961:111 ss.). Anche è assai produttiva in
osco e in latino, mentre in umbro è attestato l'unico termine *fesnã- (ib. 166 ss.)25. Per
quanto concerne si è notata sopra l'area di diffusione che
comprende l'umbro, il latino e i dialetti medioitalici.
La posizione dell'osco appare chiara. Lo sviluppo di da cui
l'osco trae la maggior parte dei suoi termini sacrali, sostituisce le originarie radici italiche
delle quali nei testi epigrafici oschi non sono
documentate che rare e isolate tracce. Anche in questo caso si manifesta il carattere
innovativo del lessico osco rispetto a quello umbro, carattere indubbiamente dovuto
soprattutto al contatto con la grecità (Prosdocimi 1978a:555); nel caso delle radici *SÄK-
e a causa delle singolari concordanze con il latino si può ricorrere all'ipotesi
di una interferenza del sostrato di tipo latinoide sull'osco (Porzio Gernia 1961:134 ss.):
l'osco avrebbe assunto in età antica queste radici e avrebbe dato loro il più ampio
sviluppo, che solo in parte coincide con quello del latino di Roma.
Il lessico umbro, più conservatore di quello osco, nella lingua sacrale mantiene
elementi dell'originario patrimonio italico come La
produttività di nell'onomastica italica conferma l'antica estensione a tutta l'area
italica: Pacuuius è riconosciuto dalla tradizione come prenome osco (v. Schulze 1966:476
s.). Nell'osco, obliterata nell' uso della lingua sacrale, resta nella voce
PRUPUKID il cui valore giuridico  riflette un uso antico, caratteristico dell'italico,
oppure rappresenta una innovazione osca forse già dovuta ad un influsso latino (Viano
1953:174).
3.2. I rapporti preistorici e storici che sono intercorsi tra l'italico e il latino
anteriormente all'espansione romana si articolano in fasi complesse26. In età preistorica e
nella prima fase della sua storia il latino di Roma ha assorbito apporti linguistici italici che,
come nota il Terracini27, riflettono aspetti elevati di cultura. Di primaria importanza e
dotata di particolare prestigio è la corrente sabina28 che è vitale in registri particolari, come
la lingua giuridica e soprattutto religiosa. Anche un elemento formale come l'allitterazione,
che nella tradizione latina connota i registri più alti, continua una tradizione italica (Durante
1958:77, 81). saranno da attribuire ad una cultura italica antichissima29,
sommersa  alterata dalla crisi generale del V sec.30.
I derivati latini di sono già attestati nei più antichi documenti epigrafici (VI-V
sec. a.C). Si noti che il vaso di Dueno, che ci ha conservato pakari, presenta evidenti
italismi: il dat. -oi (Duenoi), l'uso tipicamente italico della posposizione (fed endo); la
formula stessa di questa iscrizione presenta analogie con la maledizione di Vibia (Viano
1963:175 s.). Del resto, già agli albori della tradizione letteraria latina la famiglia lessicale
di mostra un grado assai avanzato di strutturazione lessicale: in Plauto pāx, pacō,
pacīscor, pacificō, pactiō, pactum (quest'ultimo già depauperato semanticamente nella
locuzione quo pacto = quo modo, v. LODGE s.u.). Altrettanto sviluppata appare in
Plauto la famiglia di
LODGE 1924, s. uu.).
(di cui l'osco conserva solo alcuni resti) si mantengono nella
tradizione umbra e latina e nell'area dei dialetti medioitalici; rappresentano la
sopravvivenza e la continuazione, ricca di informazioni sul piano ideologico e storico,
dell'antica religiosità italica. Il piaculo, cioè il rito di purificazione di colpe, e l'auspicio
sono cerimonie antichissime (Devoto 1967:201 s.). Esistono strette corrispondenze tra il
rito umbro e quello romano, in particolare tra il piaculo e la lustrazione delle Tavole
Iguvine e le cerimonieritualiricordateda Catone (a.c. 134-141): alla base stanno le stesse
categorie concettuali e ideologiche (Prosdocimi 1978b: in particolare 600 s., 609, 636 s.,
641).
La protostoria italica non può essere disgiunta dalla protostoria del latino. La latinità
sarà allora da considerare "in una protofase, quella della koiné culturale italica (anteriore al
V sec.) alla quale è da ascrivere quello che chiamerei diasistema semantico e di cui le
Tavole Iguvine rispetto al latino di Roma rappresentano un caso tipico" (Prosdocimi

1978a:555).
3.3. Nell'ideologia che caratterizza la ritualità romano-italica una funzione primaria è
assunta dalla preghiera (v. Devoto 1967:201; Prosdocimi 1978b:605). Il formalismo
testuale, cioè il carattere formulare delle preghiere, in questo contesto assume una
rilevanza particolare e implica l'esistenza di una tradizione assai antica.
Profonde analogie sono state notate tra le formule iguvine e i carmina latini (Durante
1958:69 ss.; Prosdocimi 1978b:600) che riflettono affinità culturali assai antiche. Sono
particolarmente calzanti le corrispondenze tra le formule connesse al piaculo tramandate da
Catone (a.c. 134-141, passim) e i passi iguvini relativi al piaculo e alla lustrazione. I
mezzi formali sono identici: l'anafora, l'allitterazione, l'omoioteleuto e il chiasmo. Il
discorso è scandito in membri ritmici ed è tessuto con simmetrie, ripetizioni, assonanze,
coppie allitteranti. Nella formula della preghiera sacrificale, recitata in tutta la sua
estensione per il sacrificio a Giove Grabovio (T. I. VI a 23-55), si identifica una struttura
ritmica complessa in cui si sono interrelati fattori prosodici e semantici, combinati in modo
da formare e concatenare dicola  unitàritmico-elementari(Prosdocimi 1978b:600). Come
rileva il Prosdocimi (op. cit., p. 601) esiste una stessa struttura che non investe
semplicemente il piano rituale, ma consiste in "una comune testualità, intesa come
modello-brogliaccio da riempire e modificare secondo i riti". Inoltre, all'interno della
stessa tradizione la prescrizione si differenzia dalla preghiera per la diversa funzione del
testo: le preghiere sono comunque ritmiche, le prescrizioni possono essere ad andamento
ritmico  prosastico (Prosdocimi 1978b:633). Per quanto concerne la formula umbra che
ci ha conservato le forme pacer e pase (v. sopra § 1) si presenta come un dicolon
allitterante: futu fons / pacer pase tua {tua non ha autonomia ritmica poiché forma un'unità
semantico-sintattica con pase)31;sono rilevanti anche le ripetizioni sinonimiche. In VI a
23; b 7,26, dove la formula compare al congiuntivo, gli espedienti formali sono diversi; il
ritmo si fonda sul parallelismo dato dallaripetizionedi sis: fons si / pacer si{r).
Nella tradizione latina sono conservate tracce dell'esistenza di stilemi fissi che
semanticamente e formalmente corrispondono a quelli individuati nella tradizione umbra.
Così il sintagma pace tua (su cui v. sopra, § 1.1), che corrisponde esattamente all'umbro
pase tua,è formula fissa tradizionale nella lingua religiosa latina. Ricorre frequentemente in
Plauto, che la riferisce a divinità particolari (Amph. 1127; Rud. 698; Trin. 837; Curc.
270; Merc. 678 ecc.; ν. anche Virg. Aen. 10, 31 sine pace tua.). Nel sintagma pax deum
(diuom) l'uso del genitivo deum (ad es. PI. Poen. 253; Liv. 3, 5, 14; 7, 2, 1) o diuom
(Lucr. 5,1126; Verg. Aen. 3, 369) rivela che si tratta di formula antica, cristallizzata nella
lingua religiosa; indica la benvolenza data dagli dei32. Frequentemente pax è unito a
propitius, coppia allitterante equivalente a u. pacer pase: ad es. Plauto, Merc. 678-80
Apollo, quaeso te, ut des pacem propitius...meoque ut parcas gnato, pace propitius; Trin.
837 M tuapax (Neptune) propitia foret praesto (cfr. anche Merc. 953 s.).
Nelle formule di preghiera latine, nelle invocazioni alla divinità propitius è usato da
solo (Cat. r.r. 141 Mars te precor, ut sies propitius', v. pure Plauto Amph. 1090, As. 781,
Aul. 1090, Curc. 531, 557, Mi. 701 ecc., cfr. Lodge, s.u.). Spesso ricorre in coppia
sinonimica con uolens: così nelle precationes patris familias in Cat. a.c. 141, 134 Mars
pater precor quesoque uti síes uolens propitius', PI. Curc. 89 festiuae fores, potate, fite
mihi uolentes propitiae', Liv. 7, 26 Si diuus, si diua esset, qui...uolens propitius
adesset33. E' evidente il parallelismo con il formulario propiziatorio umbro futu fons
pacer 'esto fauens propitius' e fos sei pacer sei 'fauens sis propitius sis'. Tenendo poi
conto del rapporto rilevato tra il lat. pakari del vaso di Dueno e le forme italiche, diventa
evidente il parallelismo tra cosmis attestato nella stessa iscrizione e l'u. *pakri-: entrambi
gli aggettivi esprimono il costante atteggiamento della divinità (Viano 1953:175).
La pietās, che deve essere attributo dell'uomo, è strettamente correlata con la pāx,
concessa dagli dei. La nozione espressa da pius (=purifìcato, perché ha compiuto i sacri
doveri verso gli dei) è condizione gradita dagli dei perché concedano la pax34. In latino
anche pūrus 'puro', accanto a pius, è accostato a pāx: così nelle formule dei Feziali,
attestate da Livio (1, 32): puro pioque duello querendas (res) censeo (cfr. Pisani 1960:
58), in cui si insiste sull'atteggiamento che deve essere pūrus e pius, cioè rispettoso di
ogni regola morale imposta dalla divinità, v. anche PI. Amph. habui expur()gationem:
facta pax est.
3.4. La connessione che, rappresentata da stata individuata tra il

latino e l'italico non deve essere ristretta nell'ottica del 'prestito' che spesso si limita a
definire la lingua di partenza e la lingua di arrivo senza prendere in considerazione il
contesto culturale e ideologico.
Da quanto si è detto si deduce che (ed anche
nonostante l'oscurità etimologica, v. sopra § 3.1), con i valori concettuali sottesi,
contengono valori semantici assai rilevanti nell'ideologia religiosa dell'Italia antica. Nella
tradizione linguistica latina ed italica abbiamo di fronte una strutturazione di determinati
campi semantici: da un lato si articolano intorno allo stesso significante, cioè la famiglia di
con i valori ad essi connessi; dall'altro lato sono
però inglobati in significanti diversi. In latino, come attesta l'uso nelle formule di
preghiera, propitius e uolens sono omologhi all'u. *pakri-: all'interno dello stesso
contesto ricoprono la stessa area semantica. Propitius che è la forma più usata e più
strettamente connessa con pāx in sintagmi allitteranti (v. sopra § 3.3); l'omologia con
*pakri-risultaevidente non solo dall'analogia dei contesti ma anche dall'etimologia: è un
aggettivo composto che si riallaccia alla radice di petō (cfr. Ernout-Meillet 1959, s.u.
propitius) e contiene quindi il valore 'dirigersi verso, a favore di qualcuno',
concettualmente assai vicino a quello di *pakri-. Benché fauēre appartenga ad una
antichissima tradizione (v. Peruzzi 1977:119), nei documenti latini arcaici non è attestata
la forma fauēns.
In latino accanto a pius di origine italica, esiste ρūrus che ha le stesse valenze
semantiche e appartiene alla tradizione latina piú antica. Ha la stessa radice del scr.
pavitár- 'quello che purifica', pütáh 'purificato' ecc. (cfr. Ernout-Meillet 1959, s.u.): è
uno dei molti termini della lingua religiosa che, perduto altrove, si è conservato nelle
estreme aree laterali costituite dal latino e dal sanscrito.
Il momento storico che ha determinato il contatto tra il latino e l'italico si perde dunque
nella protostoria del latino. Sulla radice che esprime una sfera concettuale
fondamentale dell'etica e della religiosità degli antichi Italici (la buona disposizione che
lega l'uomo all'uomo e, nell'ambito sacrale, la divinità all'uomo), il latino ha nel corso dei
secoli sviluppato, autonomamente e indipendentemente dall'italico, una ricca famiglia
lessicale che, pur conservando tracce dell'antico valore religioso, si connota in senso
giuridico-politico; alla sua base sta originariamente il significato 'patto, accordo che si
fonda sulla buona disposizione dei contraenti'35. Il processo di specializzazione in senso
giuridico del primitivo valore sacrale, o sacrale-giuridico, è tipico del latino e si manifesta
anche in altre parole36.
4. Nel problema trattato sono affiorate le due prospettive metodologiche dominanti nella
ricostruzione linguistica: la prospettiva genealogica, che compara i dati e li riduce ad una
formula unica originaria e la ricostruzione che prende in considerazione la nozione di

contatto interlinguistico e quindi la formazione di aree di affinità.
Sia nella ricostruzione della storia e preistoria dell'Italia antica, sia in quella delle
preistoria indoeuropea dell'italico, i due orientamenti, pur antitetici e contrastanti37, si
sono affiancati in un confronto dialettico e articolato. I dizionari etimologici, in cui prevale
il primo orientamento metodologico, ci hanno offerto, opportunamente classificato e
ordinato, il materiale su cui fondare una interpretazione concretamente storica. In
particolare, definire i dialetti in una prospettiva genealogica  nei fasci di isoglosse che ne
individuano la posizione, rappresenta il presupposto necessario dell'indagine volta a
ricostruire aree di affinità culturale e linguistica. Il problema del rapporto tra i due
orientamenti è particolarmente evidente nell'Italia antica, dove "la natura documentale ha
ben presto permesso di affiancare alle classificazioni genealogiste su base esclusivamente
linguistica... discorsi pienamente 'storici'." (Prosdocimi 1978a:546).
L'interpretazione della distribuzione geografica dei fenomeni è tuttora euristicamente
produttiva. Tanto nell'area indoeuropea quanto nell'Italia antica costituisce traccia di
storia: non solo è utile per definire la posizione in senso genealogico di una lingua  di un
dialetto, ma contribuisce, come si è visto, a determinare cronologie relative, centri di
innovazione, aree di affinità, resti di fasi arcaiche38.
Il tema qui trattato comporta problemi di storia linguistica, che concernono il periodo
in cui, sia pur scarsi, esistono documenti, e problemi di preistoria linguistica, in cui
l'indagine, finalizzata alla ricostruzione di fasi remote, si concentra su quel valore fossile
che il linguaggio, in quanto tradizione, contiene intrinsicamente39.
Il linguista che studia la storia linguistica dell'Italia antica deve spesso risalire la
tradizione storica delle lingue fino a toccare problemi che concernono le vicende
preistoriche delle tradizioni poi confluite in Italia. Prescindendo dall'esame del quadro
preistorico non sarebbe stato possibile interpretare storicamente la connessione italico-
latina con l'ipotesi di un contatto culturale che ha il suo punto di origine nell'italico.
Preistoria e storia linguistica non sono dunque separate rigidamente e antiteticamente, ma
si confondono senza soluzione di continuità40. Nel nostro caso si impone la nozione di
protostoria dell'Italia antica, quale anello tra preistoria e storia linguistica.
La ricostruzione non deve limitarsi alla comparazione dei significanti, ma orientarsi
verso i valori significativi del segno, attraverso i quali si recuperano valori ideologici e
concettuali (Ramat 1984:121 ss.). Una ricostruzione orientata in tal senso fa emergere
strutturazioni lessicali e semantiche collegate ad un determinato significante, ma anche a
radici differenziate. L'etimologia della parola non è pertanto storia di un singolo elemento
lessicale, ma si inserisce in un quadro assai vasto e complesso. In questo contesto il
metodo comparativo conferma la sua validità euristica: l'analisi di sviluppata entro
i presupposti sopra rilevati, ci ha portati ad individuare schemi ideologici e culturale che,
considerati non staticamente ma nei rapporti dinamici fra tradizioni linguistiche diverse, ci
hanno condotti sul terreno storico concreto.
NOTE
1 Per 'italico' si intende l'osco, l'umbro e i dialetti medioitalici (o sabellici).

2 La formula ricorre in VI a 30, 33, 40, 42, 50, 52; VI b 11, 13, 32, 34, 61; VII a 14, 17, 31, 50.
Compare al plurale in VI b 61 f ututo foner pacrer pase uestra 'estote fauentes propitii pace uestra'.
3 Si noti che il genitivo finale è attestato anche da solo (v. ad es. VI a 8,18 s.).
4 L'ipotesi del Devoto è accettata dal Pisani (1964:142) che traduce paca con 'ex pacto' e compara la
forma umbra con il scr. āā.
5 Interpretata come influsso della consonante labiale (come nell'o. praefucus)  come derivata da *pro-
pak-w- con sincope e con sviluppo di -u- per anaptissi.
6 Se, come ritiene la Sordi (1985a: 12), sono autentici i testi conservatici da Dionigi d'Alicarnasso per
il foedus Cassianum (6, 95, 2 ss.) e da Polibio (3, 24, 6) per il secondo trattato fra Roma e
Cartagine, la parola pāx: doveva essere attestata in luogo di ειρήνη nel trattato stipulato agli inizi del
V sec. a.C. dai Romani con i Latini e nel trattato con Cartagine.
7 Ad es. nel contratto matrimoniale, espresso indifferentemente con i sintagmi pacīscīo dēspondēre
filiam, puellam, cfr. PI. Aul. 241, Ci. 601, Poen. 1157, Trin. 1155, 1183 ecc.; Liv. 1, 2; anche
pactum assume talora il valore di 'promessa per le nozze'. V. Viano 1953:178 s.
8 V. in bibliografia. Cfr. pure Devoto 1952:280; Milani 1985:24 s.
9 Secondo l'Ernout-Meillet 1959, s.u. *pacõ, l'alternanza di sorda e sonora non creerebbe difficoltà, se
si tiene conto della coesistenza di forme atematiche e di presenti con infisso nasale. Cfr. Chantraine
1968-1970, s.u.
10 V. Sordi, 1985b:147 (cfr. Sordi 1985a:12). Secondo la studiosa pāx designerebbe il 'conficcamento
del chiodo' nella porta del tempio (cfr. Liv. 7, 2 e 3), un rito che simboleggia l'alleanza conclusa con
la divinità che concede benessere e pace. L'ipotesi è accettata anche dalla Milani (1985:25) che,
notando come sia esclusivo il rapporto tra l'umbro e il latino, uniche lingue a dare al tema *ā- il
significato 'pace', suggerisce, come pura ipotesi e senza approfondire il problema che esula dal tema
da lei trattato, che possa trattarsi di un prestito dal latino all'umbro.
11 II Bonfante 1937-38:57 prospetta l'ipotesi che ρãgus nel senso di 'villaggio' 'centro abitato' possa
essere sorto in un ambiente di palafitticoli che edificavano su pali le loro abitazioni.
12 V. Duchesne-Guillemin 1940-41:159 che collega la voce a *PĀG/PĀK-. Per il significato, nel quadro
qui proposto, la forma deve essere attribuita a *PÄG-. Difficoltà semantiche rendono dubbia
l'attribuzione di toc. A paci 'destra' suggerita dallo studioso (loc. cit.).
13 Per quanto riguarda il ser. pajráh, il cui significato non è chiaro, la connessione con *PÄG- sarebbe
ammissibile solo se si accettasse il significato 'solido, duro', che è assai dubbio (cfr. Mayrhofer, s.u.).
L'arm. p'akem 'chiudo piantando un piuolo', proposto dal Meillet (1935:110) continuerebbe però una
forma con sorda aspirata iniziale.
14 Per le difficoltà semantiche e formali che sollevano non si possono prendere in considerazione I'irl,
ant. āil 'gradito' (<*pokli-) e accai (< *pakni-, 'legame'?). Cfr. Walde-Hofmann 1938, s.u. pacíscō.
15 La famiglia lessicale rappresentata da got. faheps 'gioia' fagìnom 'rallegrarsi' (cfr. a. ted. a. gifehan,
ags. gefeōn 'rallegrarsi', fægen 'lieto' ecc.) presuppone una radice germanica feh-: fez che tanto per il
vocalismo quanto per il significato non pare collegabile con *PẴK- (v. Feist 1939, s.u. faheps).
16 Bonfante 1938, 1976; Molinari 1965:343.
17 La corrispondenza tra got. *fagrs e u. *pakri- è notata dal Bonfante 1938:365 e dalla Molinari 1965:
388.
18 A questo proposito sono significative le conclusioni della Molinari (1965:397): "Perciò, dato che tra
latino e osco-umbro non ci sono stati probabilmente contatti particolari al di là di quelli nella
penisola, possiamo affermare che le isoglosse 'europee' più probabilmente sono state acquisite da
entrambe le lingue separatamente, quando tutte e due si trovavano al di là delle Alpi a contatto con i
popoli dell'Europa centrale; e che invece le isoglosse solo col germanico, dato il loro carattere
eterogeneo, possono con più sicurezza essere considerate imprestiti dal latino all'osco-umbro, 
viceversa, di parole originariamente comuni ad una sola di queste lingue e alle lingue germaniche".
19 Ipotesi prospettata, dalla Milani, v. sopra n. 10.
20 Su questo problema v. Porzio Gernia 1961:118 s.
21 V. Pisani 1964:127; 1960: 53 s.;Porzio Gernia 1961:118 s.
22 Per questo problema v. Peruzzi, 1977:88 ss., dove si prospetta l'ipotesi, convincente perché sorretta
da solidi criteri metodologici, che il vocalismo a rappresenti in faueō-foueō, come in molte altre
voci, una corrente tipica dell'area costiera del Lazio; il vocalismo  sarebbe invece caratteristico
dell'area orientale e settentrionale (le forme umbre fo(n)s faner presupporrebbero quindi o).
23 L'articolo, benché sia da rivedere in alcuni punti, pone in rilievo profonde divergenze tra osco e umbro
nell'ambito del lessico sacrale.
24 Assume però un diverso sviluppo semantico rispetto al latino: è comune solo l'aggettivo . 
(=lat. sacer); l'osco trae numerose formazioni verbali che presentano una notevole divergenza
semantica rispetto al lat. sacrāre. Di norma il significato è 'fare un sacrificio', talora 'celebrare con
sacrifici'; SAKARAKLUM (< *sacra-tlo-) indica 'complesso sacro'.
25 L'area dei dialetti medioitalici è caratterizzata dalla compresenza di tutte le radici.
26 Per un quadro generale del problema, corredato dalla bibliografia, v. Porzio Gernia 1982:12 ss.
27 "Per l'epoca arcaica la presenza di una vera e propria pressione linguistica dell'italico su Roma è
innegabile... Cronologia e geografia, che sono tutt'uno, ci attestano ciò in modo troppo evidente.
Troppi italismi del latino sono, come avrebbe detto il Meillet, termini di civilizzazione e ci provano
quindi che in certi campi e in certe epoche questo predominio dell'italico indubbiamente ci fu...
Questa corrente italica non fa che continuare, agli albori della storia di Roma, una complessa
situazione preistorica" (Terracini 1981:223 s.).
28 Studiata in particolare dal Peruzzi (v. le opere citate nella bibliografia; cfr. Porzio Gernia 1982:15).
29 Su di essa v. Prosdocimi 1978a:551 ss.
30 Nel V sec. la situazione muta radicalmente. Gli Oschi scendono nelle sedi storiche e sviluppano
indipendentemente caratteri culturali e linguistici, dove assume un ruolo determinante il contatto con
la grecità e con lingue probabilmente di ceppo latinoide (v. sopra, § 3.1). Il latino continua a
mantenere una posizione marginale e subisce un intenso influsso di tipo umbro, in particolare
partecipa a molte innovazioni fonetiche e morfologiche nelle quali sono coinvolti, sia pur
parzialmente, l'etrusco e il falisco (monottongazione, riduzione delle consonanti finali, processi di
trasformazione della declinazione. V. Porzio Gernia 1982:17 ss.).
31 In VI b 61 è ovviamente al plurale.
32 Sugli sviluppi del concetto di pax nel periodo classico e sulla rinnovata vitalità del senso religioso
della pax deorum in età augustea, v. Viano 1953:180 ss. e Sordi 1985b:146 ss.
33 Cfr. Servio, ad Aen. 1, 733: esse uelis secundum etruscam disciplinam; sic enim dicunt uolens
propitius sis.
34 Ad es. PI. Ru. 26 facilius si qui pius est a dis supplicans... inueniet ueniam sibi, Poen. 1255
nostram pietatem adprobant decorantque dei.
35 Sugli sviluppi di pax nel latino classico v. Viano 1953:179 ss. Il carattere politico della religione
romana è stato ripetutamente rilevato; al centro del rapporto fra politica e religione, tipico del mondo
romano, starebbe il concetto di pax deorum (cfr. Sordi 1985b:146 ss.).
36 Si veda iūs (Devoto 1958:100 ss.; Benveniste 1969,11:113); scelus (ib. p. 108; cfr. Viano 1953:169
ss.); rēx (Benveniste 1969,11:9 ss.); spondeē (ib. pp. 209 s., 214 s.; v. sopra § 1.1).
37 Sul problema del rapporto tra i due criteri v. Terracini 1957:241 ss.
38 Sull'applicabilità della geografia linguistica allo studio delle lingue antiche v. le pagine, ricche di
suggerimenti metodologici, di B. Terracini 1957:20 ss.
39 Terracini 1949:25, 36 e 45; 1957:169.
40 Sul rapporto tra preistoria e storia linguistica v. Terracini 1957:241 ss., 1981:231.
BIBLIOGRAFIA
AA. VV. 1978. Lingue e dialetti dell'Italia antica, a cura di A. L. Prosdocimi. (Popoli e civiltà dell'Italia
antica, vol. VI.). Roma.
AA. VV. 1985. La pace nel mondo antico, a cura di Marta Sordi. Milano.
Benveniste, E. 1969. Le vocabulaire des institutions indo-européennes. 2 voll. Paris.
Berneker, E. 1924. Slavische etymologisches Wörterbuch. I e II (H. 1). Heidelberg.
Boisacq, E. 1923. Dictionnaire étymologique de la langue grecque. II ed. Heidelberg-Paris.
Bonfante, G. 1976. Contributi allo studio della posizione dialettale del germanico. Filologia e critica,
Studi in onore di V. Santoli. I. 27-31. Roma.
. 1938. Les rapports linguistiques entre germanique et osco-ombrien. Revue des Etudes Indo-
européennes 1. 353-376.
. 1937-38. Tracce di terminologia palafitticola nel vocabolario latino? Atti del R. Istituto Veneto di
Scienze, Lettere ed Arti 97. 53-70.
Chantraine, P. 1968-1970, Dictionnaire étymologique de la langue grecque. I-II. Paris.
Corssen, W. 1864. Zum oskischen Dialekt. Zeitschrift für vergleichenden Sprachforschung 13. 161-201.
Devoto, G. 1967. Gli Antichi Italici. (I ed. 1931). Firenze.
. 1962. Origini indeuropee. Firenze.

. 1958. Parole giuridiche (1933). Scritti Minori. I. 95-118. Firenze.
. 1962. Tabulae Iguvinae. III ed. Roma.
De Vries, J. 1961. Altnordisches etymologisches Wörterbuch. Leiden.
Duchesne-Guillemin, J. 1940-41. Tocharica. Bulletin de la Société Linguistique 41.140-183.
Durante, M. 1958. Prosa ritmica, allitterazione e accento nelle lingue dell'Italia antica. Ricerche
linguistiche 4.61-98.
Ernout, A. - Meillet, A. 1959. Dictionnaire étymologique de la langue latine. IV ed. Paris.
Falk, H. S. - Torp, A., 1960. Norwegisch-dänisches etymologisches Wörterbuch. 2 voll. Heidelberg.
Feist, S. 1939. Vergleichendes Wörterbuch der gotischen Sprache. III ed. Leiden.
Frisk, H. 1960 ss. Griechisches etymologisches Wörterbuch. 2 voll. Heidelberg.
Holthausen, F. 1963. Altenglisches etymologisches Wörterbuch. II ed. Heidelberg
Klein, E. 1967. A comprehensive Etymological Dictionary of the English Language. 2. voll. Amsterdam-
London-New York.
Kluge, F. 1960. Etymologisches Wörterbuch der deutschen Sprache. XVIII ed. a cura di W. Mitzka.
Berlin.
Lodge, G. 1962. Lexicon Plautinum. Hildesheim. [=Leipzig 1924.]
Mayrhofer, M. 1956 ss. Kurzgefaßtes etymologisches Wörterbuch des Altindischen. A Concise
Etymological Sanskrit Dictionary. 2 voll. Heidelberg.
Meillet, A. 1935. Les sourdes aspirées en armenien. Bulletin de la Société de Linguistique 36.109-120.
Meringer, R. 1904. Wörter und Sachen, I. Indogermanische Forschungen 16.101-196.
. 1907. Wörter und Sachen, V. Indogermanische Forschungen 21.277-314.
Miklosich, F. 1886. Etymologisches Wörterbuch der slavischen Sprachen. Wien.
Milani, . 1985. Note sulla terminologia della pace nel mondo antico. AA.VV. 17-27.
Molinari, M. V. 1965. Relazioni tra il lessico germanico ed i lessici latino ed oscoumbro. Memorie
dell'Istituto Lombardo -Accademia di Scienze e Lettere, Classe di Lettere, Scienze morali e storiche,
28:3.339-402.
Onions, C.T. 1966. The Oxford Dictionary of English Etymology. Oxford.
Peruzzi, E. 1977. Aspetti culturali del Lazio primitivo. Firenze.
. 1966. Il latino di Numa Pompilio. La Parola del Passato, 15-40.
. 1970-1973. Origini di Roma. I-II. Firenze.
. 1967. Sabinismi dell'età regia. La Parola del Passato, 29-45.
Pisani, V. 1964. Le lingue dell'Italia antica oltre il latino. Torino.
. 1960. Testi latini arcaici e volgari. II ed. Torino.
Poccetti, P. 1979. Nuovi documenti italici a complemento del Manuale di E. Vetter. Pisa.
Porzio Gernia, M.L. 1982. Il latino e le lingue indoeuropee dell'Italia antica. Atti del Convegno 'Alle
origini del latino' della Società Italiana di Glottologia, Pisa 7-8 dicembre 1980. 11-26. Pisa.
. 1961. Rapporti tra il lessico sacrale osco e latino. Archivio Glottologico Italiano 46.97-138.
Prosdocimi, A. L. 1978a. Le lingue italiche. AA.VV. 1978. 543-558.
. 1978 b. L'umbro, ib., 585-788.
Ramat, P. 1984. Della realtà dell'irrealtà scientifica. Incontri linguistici 9.115-126.
Schulze, W. 1966. Zur Geschichte Lateinischer Eigennamen. II ed. Berlin.
Sordi, M. 1985a. Introduzione: dalla 'koinè eirene' alla 'pax romana'. AA.VV. 1985. 3-16.
. 1985b. "Pax deorurn' e libertà religiosa nella storia di Roma. Ib., 146-154.
Terracini, B. 1957. Conflitti di lingue e di culture. Venezia.
. 1981. Di che cosa fanno la storia gli storici del linguaggio? Storia dei tipi BENIO e NERBA nel
latino volgare. Linguistica al bivio. Napoli. [= Archivio Glottologico Italiano 27, 1935, 133-152;
28, 1936, 1-31, 134-150.]
. 1949. Guida allo studio della linguistica storica. I. Roma.
. 1957. Questioni di metodo nella linguistica storica. Pagine e appunti di linguistica storica. 1-40.
Firenze [= Atene e Roma, N.S., 2, 1921, 31-47, 99-116.]
Turner, R. L. 1966. A Comparative Dictionary of the Indo-Aryan Languages. London.
Vasmer, M. 1953 ss. Russisches etymologisches Wörterbuch. 3 voll. Heidelberg.
Viano, M. 1953. Contributo alla storia semantica della famiglia latina di "pax". Atti dell'Accademia delle
Scienze di Torino 88.168-183.
Walde, A. 1927. Vergleichendes Wörterbuch der Indogermanischen Sprachen. Herausgegeben und
bearbeitet von J. Pokorny. 3 voll. Berlin und Leipzig.
Walde, A. - Hofmann, J. B. 1938. Lateinisches etymologisches Wörterbuch. 3 voll. Heidelberg.
ABSTRACT
The paper deals with the historical interpretation of the linguistic and cultural isogloss represented by
the lexical group derived from the root
In the Italic dialects and in Latin, these words, besides being formally parallel, have a sacral and
juridical value which is not attested in the other Indo-European languages. Traditionally, in etymological
dictionaries, these words are connected, together with pangō etc., to a root pãg*-/pəg-, alternating with
*pãk-/pək-, and meaning 'fix', 'drive into'. However, it seems more likely to reconstruct two different
roots: *-/, 'bind', 'join' and *pāg-/pəg-, 'fix', 'drive into'.
The root *pãk-/pək- is very productive in Germanic; the adjectival form pakri-, attested in Umbrian,
corresponds, formally and semantically, to the Gothic word *fagrs. It is well known that, among Western
Indo-European languages, the Italic languages have a number of isoglosses in common with the Germanic
languages (in particular with Gothic), which are not found in Latin.
It is likely that one has to do with a Germanic and Italic innovation, which entered Latin in the
course of contacts which took place in Italy. Most Italic-Latin isoglosses have in fact developed in Italy,
as a consequence of cultural relationships. The distribution of the words derived from *pk- in the Italic
dialects clearly shows that Umbrian was the center of diffusion. The area in which *pãk- is attested is the
same in which one finds the root This confirms the existence of cultural and
linguistic affinities among Latin, Umbrian and Sabellic.
In the course of centuries, Latin has developed a rich lexical family, whith juridical connotations,
from the root *pk-, which originally expressed a fundamental concept in the religious ideology of ancient
Italy: the favourable attitude of divinities toward men.
ZU ISIDOR VIII, III, 6
BERNFRIED SCHLERATH
Freie Universität Berlin
Aeneis VIII, 185 ff. erläutert König Euandros dem Aeneas das Jahresfest der Er
schlagung des Cacus durch Hercules und sagt:
non haec sollemnia nobis,
has ex more dapes, hanc tanti numinis aram
vana superstitio veterumque ignara deorum
imposuit: ...
Servius sagte zu Vers 187 folgendes:
VANA SVPERSTITIO superstitio est timor superfluus et delirus, aut ab aniculis dicta superstitio,
quia multae superstites per aetatem delirant et stultae sunt: aut secundum Lucretium superstitio est
superstantium rerum, id est caelestium et divinarum, quae super nos stant, inanis et superfluus
timor. VETER VMQVE IGNARA DEORVM duo dicit: non ideo Herculem colimus, aut quia
omnem religionem veram putamus, aut quia deos ignoramus antiquos. cautum enim fuerat et apud
Athenienses et apud Romanos, ne quis novas introduceret religiones: unde et Socrates damnatus est
et Chaldaei vel Iudaei sunt urbe depulsi.
Es ist reizvoll zu sehen, in welcher Weise Isidor VIII,  , 6 den Servius adaptiert.
Die Stelle kehrt bei ihm in folgender Form wieder:
Superstitio dicta eo quod sit superflua aut superìnstituta observatio. Alii dicunt a senibus, quia
multis annis superstites per aetatem delerant et errant superstitione quadam, nescientes quae vetera
colant aut quae veterum ignari adsciscant. Lucretius autem superstitionem dicit superstantium rerum,
id est caelestium et divinorum quae super nos stant; sed male dicit. (Die identischen Wörter in beiden
Texten sind hervorgehoben.)
294 BERNFRIED SCHLERATH
Servius hat sicher seine Erklärungen von superstitio nicht "etymologisch" im techni
schen Sinn gemeint, d.h. er hat z.B. die gleichen Buchstaben in super(fluus) und ti(mor)
nicht als ein bestätigendes Indiz für die begriffliche Verwandtschaft dieser Wörter mit su
perstitio angesehen. Isidor, der hier timor durch observado ersetzt, mußte aber zweifellos
die Erklärung des Servius als Etymologie ansehen. Warum er timor durch observado er
setzt, können wir nur vermuten. Ganz sicher nicht deshalb, weil er an der formalen Seite
(der Gleichsetzung der Buchstaben ti in superstitio und timor) etwas auszusetzen hatte.
Für Isidor genügt bereits ein einziger identischer Buchstabe für eine etymologische Ver
knüpfung. Auch nicht deshalb, weil er später eine andere Etymologie (alii dicunt...) als
möglich hinstellt. In der Antike können in aller Regel mehrere Etymologien nebeneinan
der bestehen, wie ja auch in der Tat ein Begriff mehrere inhaltliche Verknüpfungen
zuläßt. Der Grund für die Ersetzung von timor könnte darin liegen, daß in der Sprache
Isidors das Wort superstitio die Konnotation eines (motivierenden) individuellen timor
verloren hat und mehr zu einer Bezeichnung eines abergläubischen Brauchs {observado)
geworden ist. Doch dafür spricht nichts. Man vergleiche etwa die beiden aus dem 3. Jh.
stammenden Passagen des Corpus Juris D 28, 7, 8 pr. timidi metu divini numinis usque
ad superstitionem (Ulpian) und D 48, 19, 30 quo leves hominum animi superstidone
numinis terrentur (Modestinus). - Eher scheint mir möglich, daß die Ursache im Wort
timor liegt, das durch Ausdrücke wie timor Dei eine neue Dimension gewonnen hatte, die
es für eine Verbindung mit Aberglauben schlechter tauglich machte (vgl. Röm. 3, 18; 2.
Cor. 7, 1; Eph. 5, 21; Phil. 1,14).
Ferner fügte Isidor zu observado noch superinstiuta hinzu, das wegen des -stit- sich
etymologisch empfiehlt und auch gut zur Kennzeichnung eines christlichen (haeretischen)
Aberglaubens paßt (vgl. superstitio haeretica Cod. Theod. 16, 5, 10).
Im nächsten Satz ersetzt Isidor ab aniculis durch a senibus, wahrscheinlich weil er
das anlautende s- in dem Anlaut von superstitio wiedererkennt und er möglicherweise mit
dem Maskulinum senibus beide Geschlechter meinen kann. Dadurch kann er das
folgende auf die aniculae bezogene multae nicht mehr so brauchen und formt es nicht
ungeschickt zu multis annis um. (Sonst wird freilich genau wie bei Servius der
Aberglaube den alten Weibern, nicht den alten Männern zugeschrieben, vgl. Cic. div.
2,125 : supersddonem imbecilli animi atque anilis; nat. deor. 2,70: supersddones aniles).
ZU ISIDORVIII, III, 6 295
Die bei Servius folgende Erklärung des Lukrez rückt Isidor an das Ende. Er verwirft
sie ausdrücklich und zwar ganz klar deshalb, weil man dann die christliche Religion unter
superstitio subsumieren könnte. Durch die harte Änderung von (rerum)... caelestìum et
divinarum in caelestìum et divinorum (also "das Himmlische und das Göttliche") unter
streicht er noch die Unvereinbarkeit mit christlichen Vorstellungen.
Eine ähnliche Veränderung nimmt Isidor mit dem Vergilschem veterumque ignara
deorum vor. Er ändert supers titio ... ignara in senes ... ignari und läßt deorum weg, weil
er den Hinweis auf Heidnisches vermeidet, wenn er nicht die Gelegenheit zu einer aus
drücklichen Ablehnung -wie etwa VIII, IX, 14, wo es um divinano, divinus und
divinitas geht - ergreift. Darum muß er veterum, wie auch das vorausgehende vetera
colunt zeigt, als nicht näher bestimmtes kollektives Ntr.Pl. nehmen.
Isidors Erklärung findet ihren Niederschlag in verschiedenen alphabetischen Glossa
ren des frühen Mittelalters unter superstitio: superfluitas religionis (Placidus Cod. Par. =
Corp. gl. V, 154, 13) superflua observado (Glossae Nom. = Corp. gl.II, 594, 18) su
perflua institutio (Gl. Cod. Sangall. 912 = Corp.gl. IV, 289, 18), superflua aut superba
institutio (Gl. affatim = Corp. gl. IV, 570, 55) oder mit Heranziehung des Vergilschen
vana, das sich ja bei Isidor nicht findet: vana observatio (Gl. ab absens = Corp. gl. IV,
424, 7).
Wenn also Sueton, Domitian 1,2 die schlechteren Handschriften das besser bezeugte
variae superstitionis in vanae superstitionis verändern, steckt nicht nur die Erinnerung an
Verg. Aen. VIII, 187 vana superstitio dahinter, sondern Vergil ist noch mit allerlei
Schulweisheit verquickt.
Will man das Verfahren Isidors allgemein charakterisieren, so müßte man sagen, daß
er die antike Etymologie nicht durch neue Aspekte bereichert oder vertieft hat. Genau wie
auf anderen Gebieten auch ist seine Rezeption eher oberflächlich. Aber er hat sie von
Elementen gereinigt, die für das Christentum zu anstößig waren.
Seine große Leistung ist die Systematisierung des gesamten Wissens. Damit hängt
unmittelbar zusammen, daß ihn Antiquarisches weniger interessiert und er deshalb auch
dort am ehesten seine Quellen verkürzt. (Auch heute wie immer ist - natürlich auf anderen
Ebenen - die ausgewogene Berücksichtigung von Diachronie und Synchronie eine der
schwierigsten Aufgaben der Forschung). Ein ähnlich überzeugendes System wie das Isi
dors findet man in der Antike auf dem Gebiet des Rechts in den Institutionen des Gaius
296 BERNFRIED SCHLERATH
wieder. Der erste Antrieb zur Aufstellung des Systems des Rechts war nicht, daß man
durch die systematische Darstellung die Systemhaftigkeit des dargestellten Gegenstandes
beweisen und illustrieren wollte, sondern es war der aus der Notwendigkeit der Schule
geborene pädagogische impetus. Wir können annehmen, daß es bei Isidor ähnlich war.
SUMMARY
Servius' explanation of Verg. Aen. VIII, 187 vana superstitio veterumque ignara deorum is the source of
Isidor etym. VIII, III, 6 {superstitio). Isidor's wording is indebted to the tendency to avoid expressions
which could be offensive for a christian reader.
VI.
LATIN AND ROMANCE LANGUAGES

THE PITFALLS OF SUBORDINATION
SUBJECT AND OBJECT CLAUSES
BETWEEN LATIN AND ROMANCE
WOLFGANG RAIBLE
University of Freiburg
1. Some Essentials
The present contribution deals with language change1. In the case outlined below this
amounts to a change in the techniques of subordination. This change is connected with a
change in language type. My considerations will be based upon six essentials:
1. Every language has to cope with a basic problem. The message a speaker wants
to transmit to a hearer is a multi-dimensional whole. In order to be communi
cated, this whole has to be taken to pieces, pieces which are then strung onto the
uni-dimensional thread of discourse. It is the task of the hearer to put these
pieces together, to rearrange them in such a way as to form a new pluri-
dimensional whole. This aspect was recognized very clearly by a philosopher of
French enlightenment, Etienne Bonnot de Condillac (1715-1780). The crucial
problem in this domain is that it is not possible to find an optimal solution in the
endeavour to linearize the pluri-dimensional. The advantages a solution A might
have on a level a always entail disadvantages on a level b.2 Nevertheless,
speakers again and again make attempts at solutions which are intended to make
possible, on all grammatical levels, one central task: "la liaison des idées"
(Condillac), the re-combining of pieces to wholes.
300 WOLFGANG RAIBLE
2. This leads us to a second essential: As has been said, speakers dissociate wholes
into parts and they string these parts onto the thread of discourse. Now, as the
speaker knows the whole, we may suppose that he knows what he wants to be
transmitted and he intends this to be understood by the hearer. Speakers should
be interested to provide conditions that enable hearers to rearrange parts to
wholes or, in other words: a system in which multi-dimensional wholes are
decomposed and linearized has to be a system which serves, above all, the
interests of the hearer.
Phenomenologists like Edmund Husserl have seen the problem this way3 René
Thorn quotes in this context a maxim taken from the field of jurisprudence:
looking for the nknown author of, say, a crime, the first question usually is:
Cuiprodest?orIsfecit, cui prodest: Hearers shape language-systems - because a
language system serving in the first place the interests of speakers (speakers, for
instance, like systematical economy) has anything but a promising future.
3. Dependency grammar seems to be the most powerful model for sentence
grammar, at least for Indo-European languages. In the model of dependency
grammar, the verb is the ordering center of the sentence. Verbs are highly
unsaturated elements which have to be complemented by nominal elements or
noun groups. Using the metaphor of theater favoured by Tesnière: The verb is
the nucleus of a little drama prompting a certain number of roles for actants. For
the sake of the hearer, it is the task of the grammatical system to make clear, by a
scale of grammatical means, which actant plays which grammatical role.
(Scholars disliking dependency grammar as a typically French manifestation of
Prague school structuralism can perhaps derive comfort in predicate calculus as
proposed by Hans Reichenbach and fostered by generative semantics: predicate
calculus and dependency grammar are largely isomorphous.)
4. It is useful to distinguish the tasks of oral and written discourse. The range of
functions brought about by these two types of communication are quite different.
This has been outlined e.g. by Edward Sapir, more recently by Talmy Givón.4
Written communication lacks the aid of the situational context. Thus it has to be
much more precise, explicit, and exact than oral communication. For instance, it
is well-known that the transition of a language community from oral to written
THE PITFALLS OF SUBORDINATION 301
culture triggers the development of new syntactic devices. This has been shown,
with reference to the development of Old Spanish, by Georg Bossong. He
argues that syntactical complication was made necessary, among others, by the
translation of texts belonging to the written Arabic culture.5 The special
conditions of written communication should be borne in mind when we consider
the subordinative devices of Latin.
The following two essentials concern directly the history of language, resp. the
tory of Latin.
5. From the point of view of the history of language or of language change, the
status quo of a language always represents the simultaneousness of the non-
simultaneous, "die Gleichzeitigkeit des Ungleichzeitigen".6 If we proceed from
the assumption that language systems have to solve problems which underly
cognition and human communication, and if we assume that for certain problems
there exist scales of techniques forming a continuum, continua of this kind will
be "the space in which language changes occur". In language change one
technique will be replaced by another, but all the techniques, those growing
obsolete, the flourishing ones, and the techniques bound to be the main devices
of future, are present in the synchrony of the language system.7 Those different
techniques that coincide in the synchrony of language will belong to different
registers or different styles. The techniques growing obsolete usually have high
stilistic value and are typical for written language. The techniques advancing in
productivity, i.e. the techniques of the future, are typically oral, initially they
belong to colloquial and familiar speech.
6. Latin, by virtue of its Indo-European origins, seems to be a language of the
SOV-type. According to essential no. 3,I should speak of an 'AlA2V-language'
(Al = first actant), but in the domain of Indo-European languages, the term
'SOV' will be sufficient.8 The Romance offsprings of Latin - which are SVO-
languages - show that, from Latin to Romance as well as in Latin itself, there has
been a major change in the ordering of the main constituents of sentences. This
change in the linear order of the actants and the verb has to do with a major
number of rearrangements in the grammatical system.9
302 WOLFGANG RAIBLE
2. Some Facts Worth Attention

After this prologue in the heaven of - relatively - pure theory, I would like to come back to
earthly reality and I shall begin with some rather plain facts. If we take an English,
German, or a Romance example of a subordinate clause, our feeling of native or trained
speakers tells us that subordinate clauses have to be placed on the right hand side of the
main verb: "I wish that he come", "je veux qu'il vienne", "ich will, daß er kommt", and
so on. The common explanation for the emergence of the (later) conjunctions daß, that is:
"Ich (S) will (V) das (O): er kommt". This presupposes SVO-ordering. A subordinate
clause at the beginning of the sentence is unusual, sometimes even impossible. Yet our
Middle-European linguistic instinct is powerless against Latin reality. There, the opposite
principle seems to be at work. The "normal" place of subordinate clauses, especially in
older stages of Latin, is on the left hand side of the main verb. If we wanted to say in
Latin "It will be easy to ascertain what judgement you will pronounce when I shall have
notified what is the matter", we would probably write:
Facile erit statuere quam sententiam dicatis, cum aperuero, quid agatur.
Cicero writes instead (in hisfifthPhilippica (V,6)):

Quid autem agatur, cum aperuero, facile erit statuere quam sententiam dicatis...
The tendency to anteposition of subordinate clauses is most apparent in the writings of the
Roman historians. In this respect, Julius Cesar is an excellent source. A self-evident
(however somewhat extreme) example is taken from Bellum Gallicum 3,3 (the skeleton of
the main clause is printed in italics):
His nuntiis acceptis Galba, cum neque opus hibernorum munitionesque plane essent perfectae neque
de frumento reliquoque commeatu satis esset provisum, quod deditione facta obsidibusque acceptis
nihil de bello timendum existimaverat, Consilio celeriter convocato sententias exquirere coepit.
The anteposition of subordinate clauses seems to be a phenomenon common to Italic

languages and dialects.10 Apart from passages in the Roman historians it is particularly
current in legal texts.11
As far as subordination is concerned, there is another phenomenon which seems
unfamiliar to us: Latin provides the possibility to subordinate clauses without a subor
dinating conjunction. In such cases, subordination is marked only by the mood of the
finite verb (subjunctive), i.e. by means of a verbal form. As a rule, we find, especially in
older documents, tokens of subordinate clauses in anteposition. I quote a few examples:
amicus ueniat licet;
in tabulas publicas referatur facito (Lex Iulia Municipalis, 44 B.C.);
facias volo.
It should be noted, however, that the Rhetorica ad Herennium (4,10,14) qualifies cases
like "satisfacías oportet" as specimens of a simple style.12
3. The (Relatively) Simple Explanation of the Facts Mentioned

In the discussion stimulated by Joseph H. Greenberg's highly original article on word-
order typology, much has been said on word-order. Especially scholars like Winfred P.
Lehmann have established entire lists of phenomena characteristic for languages with the
word-order types 'SOV' and 'VSO'. The only problem is that there is no valid
explanation for the fact that e.g. the ordering type 'SOV' combines with postpositions and
with the order 'genitive-noun'. Lehmann or Theo Vennemann only give, to term it with
Bernard Cornile, "formal explanation(s), without any further consideration of the
question why this particular explanation should hold".13
This "why" is (relatively) easy to grasp if we consider the first three of my essentials
(i.e. those concerning the problem of linearization, of hearer-orientedness, and of
dependency grammar). I will illustrate this by means of an example. My example is the
Latin version of the sentence 'Brutus, the nephew of Cato, killed Cesar in the building of
the senate'. In Latin, this would be something like
Brutus nepos Catonis Caesarem in domo senatus necavit.
A (simplified and slightly modified) stemma of Tesnière's would render this as follows:
FIGURE 1
304 WOLFGANG RAIBLE
Apart from the conception already mentioned of a dichotomy between the verb and the
actants, this stemmatic representation involves one further assumption that has to be made
explicit: the ordering of the actants/circumstants from left to right. Indo-European
languages have a so-called subject-conjugation, which implies that every finite verb has to
have an implicit or explicit "first actant", whereas not all verbs entail a second actant, and
so forth. Beyond any doubt it would be to the benefit of the hearer if, on the level of the
actants, linearization began with the actant which is always present, then follow with the
actant which is next in frequency, and so on.
Our task is now to linearize this stemma, i.e. to string it onto the thread of discourse.
It is quite obvious that there are always two extreme solutions for this kind of problem:
the one going from the top to the bottom, the other from the bottom to the top. The first of
these extreme types would result in:
Necavit Brutus nepos Catonis Caesarem in domo senatus
{verb) {subject) (object) (place)
In this first case threading begins with the structural center of the sentence, i.e. with the
verb 'necavit'; then it continues with the actants, going from left to right, and, within the
particular actants or circumstants, always from top to bottom. - In the other extreme
solution, threading takes the opposite direction, beginning with the first actant and ending
with the structural center of the sentence. This time, we get something like
Catonis nepos Brutus Caesarem senatus domo in necavit
{subject) {object) (place) {verb)
If we did not take into consideration the indication of the place where this murder
happened, we could characterize the two extreme types of actant and verb ordering with
'VOS' viz. 'SOV'. Obviously these types exhibit, beside the order of the actants, further
specific characteristics relating to the inner structure of the actants: The type which threads
downwards leads to the order 'noun-genitive' (resp. 'noun-adjective'), whereas the type
'SOV', threading upwards, results in the order 'genitive-noun', 'adjective-noun', i.e.
'Catonis nepos' and so on. Since subject or object clauses are clauses acting as substitutes
for the respective actants, they should be treated in the same manner. In other words:
Within the type 'SOV' we obtain the characteristic order 'clause-sentence', whereas the
type 'VOS' shows the order 'sentence-clause'.
Up to this point one can deduce these considerations with a perfect logic particularly
from chapters 6 to 14 of Lucien Tesnière's Eléments de syntaxe structurale. What I have
called the 'direction of threading' is, in Tesnière's terms, "le sens du relevé linéaire".
Since for Tesnière the verb is the center of the sentence, he calls the extreme type which
threads from the top to the bottom the 'centrifugal type'. The other (which ends the
threading with the verb) then is the 'centripetal type'.14
We know from the work of Tesnière, of Greenberg, and others that the extreme types
exist and that they possess the features which can be deduced from the stemma.
Nevertheless, this type of explanation seems to have a serious shortcoming: The types in
question show further characteristics which seem to go beyond the explanatory power of
the stemmatic method or, more specifically, beyond the range of the essentials no. 1 and
no. 3. It is e.g. a well-established fact that SOV languages - when using grammemes on
the actants in order to mark the grammatical role which actants have with respect to the
verb - show a high affinity to postpositions or "case endings". In VSO languages, we
meet with the exact reverse. If they mark case relations on the nouns, they prefer
prepositive signals. The reason for these preferences is so obvious and so trivial that its
discovery was perhaps beyond the scope of linguistics proper: it was René Thorn's merit
to set it forth. Since it is the function of case endings or prepositions to make clear to the
hearer the grammatical role of a particular actant regarding the ordering center of the
sentence, i.e. the verb, and since it is the role of such elements to help the hearer
(essential no. 2) to establish the correct linkage between the actants and the verb, the
linking elements have to be placed where such elements are most significantly placed:
between the entities that are to be linked.15
Once more, what holds for the actants is also valid for clauses. Here, too, the linkage
element is optimally placed between the parts which have to be linked together. Thus we
can expect a linkage element between the verb and the clause(s) in the case of VSO; in the
case of SOV, instead, the linking element comes to stand between e.g. the object clause
and the main verb of the sentence. If we mark the linking element (which establishes, for
the hearer, the specific relation between the actant [or the clause] and the verb) by an
asterisk, the extreme types will show the following characteristics:
306 WOLFGANG RAIBLE
SOV SVO
post-determinative case-marking pre-determinative case-marking
(if there is case-marking on the (if there is case-marking on
nouns at all), that is S* O* V the nouns), that is V *S *0
the subordination marker of the the subordination marker of
clause tends to be placed at the the clause tends to be placed
end of the clause (cl* V) at the beginning of the clause
(V*cl)
generally: nominal and clausal generally: nominal and clausal
post-determination pre-determination
As for the internal organisation of actants we shall have:
genitive-noun noun-genitive
adjective-noun noun-adjective
There could be deduced further characteristics of the two types.16
It appears that Latin shows a considerable number of the "ideal" characteristics of the
type which uses SOV for the ordering of the actants and the verb. This is not very
surprising for those who subscribe to the thesis that the source of Indo-European
languages was a language belonging to this SOV type. Nevertheless, Latin shows some
important "irregularities": Latin nouns have post-determining case-markers, but they have
too - should we say already? - prepositions. Postpositions are (already) rare; moreover,
Latin has (already) conjunctions, i.e. is subordination markers at the beginning of
clauses. This fact - which will turn out to be an instance of my essential no. 5, the
simultaneousness of the non-simultaneous in language - will be of special interest for the
problem under consideration. But before this point can be treated, I shall discuss a few
shortcomings of the two types characterized as 'extreme', 'pure', or 'ideal'.
4. The Shortcomings of the Extreme Types

In my first essential I have already mentioned that there will never be a satisfactory
solution to the objective of linearizing the pluri-dimensional. To put it in the words of
Lucien Tesnière 17:
Il y a (...) antinomie entre l'ordre structural, qui est à plusieures dimensions (réduites à deux dans le
stemma), et l'ordre linéaire, qui est à une dimension. Cette antinomie est la "quadrature du cercle" du
langage. Sa résolution est la condition sine qua non de la parole.17
The two extreme solutions outlined above may be consistent and coherent solutions from
the point of view I have adopted. Nontheless, they involve serious problems for the
hearer. I want to deal only with the problems characteristic of the "consistent" SOV type.
Here the most important problems originate - precisely in clausal subordination. In order
to show this I shall slightly modify my stemma: By adding a relative clause to the object
of the sentence we obtain the following stemma:
FIGURE 2
If we linearize this stemma according to the SOV principle, i.e. "from the bottom to the
top", we get:
Brutus - Gallos - vicit -
Then would follow the subordinator for the relative clause (qui), then Caesarem, and so
on. This solution has an obvious flaw: it is a trap for the hearer. Having decoded the
perfect SOV sequence Brutus Gallos vicit ... he will have to cancel what he has
understood up to this point. Antinucci has shown this problem in connection with relative
clauses, but it exists for all types of subordinate clauses with a finite verb, the "normal"
order of clauses in an "ideal" SOV system being CI* V. Since the clause structure will
recursively be SOV, this means: ... (SOV)* V.
308 WOLFGANG RAIBLE
How do speakers of SOV languages react in order to spare the hearers those
apparently inevitable pitfalls of subordination? If we have a look at Turkish as a relatively
pure SOV language, we can discern two kinds of solutions:
(a) If the form of the verb closing the clause is grammatically different from the
corresponding verbal form of the main sentence, there is no difficulty at all: such
verbal forms are non-finite verb forms, e.g. infinitives or participles.
(b) Whenever the verb of the subordinate clause in such a language is as finite as the
verb of the main sentence, special forms of finite verbs have to be introduced. In
Turkish, this is the case for the so-called modus conditionalis. A somewhat
similar device exists in Japanese in the so-called 'conditional form'. Still more
typical for Japanese seems to be the difference between sentence-closing syn-
tagms and syntagms that do not close sentences, e.g. conjunctional forms of the
verb, and so on.18
After this brief glance at Turkish (and Japanese)19, we immediately understand the
"motivation" of the subordinative devices in Latin as an offspring of the Indo-European
type 'SOV'. We see for instance why the construction called ablativus absolutus
harmonizes with this type of ordering of the actants and the verb. Let us take an example:
Ariovisto victo Caesar Lugdunum rediit.
Although this construction typically exhibits the same linearization principles as other
clauses or sentences (the verb has a final position), there is a double security against
misinterpretation by the hearer: The verbal form is
- non-finite and
- the entire construction is marked as a sub-unit by case congruency, the case in
question not being used very frequently to mark actantial relations between a
noun group and the finite verb. (Other languages use other cases, for instance
the genitive; however it seems to be important that this case should not be one
that usually marks the first, the second, or the third actant.)
Moreover, we can see that constructions like the accusativus cum infinitivo or accusativus
cum participio harmonize perfectly with the 'SOV' type of ordering. I add another
example from the 5th Philippica (V,l 1):
Haec se ex commentariis C. Caesaris (...) agere dicebat.

This type of construction, too, fits into the picture of CI* V ordering, and again there is a
double security against misinterpretation by the hearer: (1) the non-finite verbal form
closing the clause (in this case: agere), and (2) the avoidance of the nominative case
(which could be mistaken for the subject nominative of a main clause). The nominative is
replaced by an accusative - a fact resulting, as in the quoted sentence, in a subordinate
clause marked by two accusative forms, haec and se.
Latin does not only make use of infinitive and participle constructions in order to cope
with the pitfalls of subordination created by SOV ordering of the actants and the verb; it
also uses the second one of the devices which can be observed in Turkish where special
finite forms of the verb are used as indicators of subordination. Latin, like other Indo-
European languages, developed a so-called subiunctivus for this purpose.20 Since the
subjunctive is also a grammatical mood, I will first explain the function of grammatical
moods in general in order to illustrate the special functions of this particular mood.
5. Communicative Responsibility and Grammatical Moods

For the domain of 'mood', I will adopt a definition proposed by Klaus Heger. He
brought the concept of "kommunikative Regresspflicht" into the linguistic discussion, that
is to say 'communicative responsibility'.21 It is of major importance for the hearer to
know whether or not the speaker assumes the responsibility for the information he wants
to communicate. This requirement is so important that languages usually make its
fulfilment obligatory by means of grammatical devices. The grammatical means to tell the
hearer that he may rely upon what has been said is, as a rule, the indicative mood. A
mood like the French Conditionnel, on the contrary, tells the hearer that the speaker
refuses to assume a communicative responsibility, especially that he conveys this
responsibility to a third person ("M. Untel aurait dit telle ou telle chose"). Thereby the
speaker rejects his communicative responsibility which can often coincide with a legal
responsibility, i.e. a responsibility which could be "actionable". By other moods, the
hearer understands that the speaker is not sure of what he says. This is, for instance, the
sense of French or English modal future ("M. Untel sera malade"). Here the third person
of future tense is something like a modus potentialis. To take still another mood: the
imperative excludes by definition all responsibility on the part of the speaker: In the case
310 WOLFGANG RAIBLE
of the imperative, this responsibility remains, so to speak, with the hearer. It is up to him
to carry out what he is asked to do. (The speaker cannot take over the responsibility for an
action which lies in the future and which has to be carried out by the hearer.)22
There exists an entire scale of modal possibilities, the one pole being, for instance,
modus indicativus, the other modus imperativus. Between these poles, the other
grammatical moods have to be located, i.e. moods such as potentialis, optativus,
hypochoristicus, promissivus, and so on. An important point for the adaptability of
grammatical moods is the fact that, normally, the indicative represents the unmarked pole
of the scale, while the other moods become increasingly marked, the pole of the highest
markedness usually being occupied by the imperative. That is to say that, e.g., the
indicative can be used for imperatives ("I command you to be silent"), whereas with the
imperative form "be silent" the speaker will never be able to express what is the proper
function of the indicative, i.e. that is to say that he assumes his communicative
responsibility.23
6. Advantages and Disadvantages of the Latin Subordinating Techniques

In section 4 it has been outlined that SOV languages use subordination markers at the end
of the subordinate clause. One of the devices used in this domain may be a special mood
(modus conditionalis in Turkish, "conjunctional forms" of the verb in Japanese). If in
such a system a grammatical mood is used to indicate subordination, this mood will
necessarily have two functions:
- a syntactical function (it serves as a subordination marker for the hearer);
- and a modal function.
The character of this modal function can be logically deduced: If the indicative has to be
located at the one end of the modal scale and if it signals that the speaker takes over the
communicative responsibility, and if subjunctive is a grammatical mood, then the modal
meaning of a modus subiunctivus can only be a restriction of the speaker's
communicative responsibility. Thus in languages with such a mood the subjunctive is
usually the least marked grammatical mood after the indicative; this means that the
subjunctive can never take over the function of the indicative (else it would lose its raison
d'être). But (in subordinate clauses) it is able to take over the functions of all other
grammatical moods and modal possibilities. This is why we find, in Latin grammars,
coniunctivus iussivus, horiativus, voluntativus, dubitativus, optativus, potentialis,
concessivus, Irrealis, prohibitivus and so on.24 This multiplicity of facets only reflects the
(relative) unmarkedness of the subjunctive. The restriction of the communicative
responsibility of the speaker applies to all the above-mentioned modal manifestations.
As has been said, the raison d'être of the subjunctive mood in a language of SOV type
is to be seen in the fact that a finite verb in subjunctive mood, placed at the end of the
anteposed clause, is, for the hearer, an unmistakable signal of subordination. This is
why, at least in Older Latin, the unique subjunctive, placed before the main verb, is
largely sufficient to convey the information of subordination. The restriction of
communicative responsibility which goes with this form is, one might say, a "necessary
evil" due to the fact that in a SOV language mood is used to indicate subordination.
So far our considerations have shown that the subjunctive is the solution for a
problem which is particularly virulent in SOV languages. However, as has already been
pointed out, there is no perfect solution in the domain of discoursive linearity. Hence this
kind of solution, too, cannot but create new problems. It is precisely the genuine Janus-
facedness of the subjunctive which will turn out to be a congenital defect. This will
become clear by the following consideration.
Object and subject clauses are governed by the finite verb of the main sentence. In
terms of tagmemics, they are clauses filling actant slots. The finite verb in subjunctive
mood indicates to the hearer that such clauses play the role of actants. Now it is always
possible to make semantically explicit informations which a language would normally
express by grammatical means. Thus, instead of using the imperative, one can express the
same content by means of a semantically imperative verb ("be silent" turns into "I
command you to be silent"). The finite verb of the main clause makes explicit a command,
the content of which depends, in this English example, on the main verb and has the form
of a non-finite object clause ("to be silent").
The following scheme serves to illustrate the four classes of main verbs governing
possible subject or object clauses. Let A be the set of all verbs admitting a subordinate
(subject or object) clause. Let  be the sub-set comprising all verbs which make explicit
that the speaker assumes his communicative responsibility, i.e. epistemic verbs as 'to
know', 'to have seen', 'to be sure', etc. Let  be the sub-set of verbs making explicit the
312 WOLFGANG RAIBLE
contrary of , that is, i.e. making semantically explicit that the speaker does not assume
the communicative responsibility ('to ignore' etc.). There remains an intersection between
 and  which I call D. It stands for the verbs which are not unequivocal as to the
question whether or not the speaker assumes his communicative responsibility. The
respective verbs are 'to believe', 'to think', 'to hope', etc.
FIGURE 3
Now I come to the problems indicated above. With a main verb of sub-set B, the speaker
semantically assumes his communicative responsibility. If, for instance, I say "I know" in
the main sentence while having a finite verb in subjunctive mood in the subordinate object
clause, this would result in a modal contradiction. In the main sentence, the speaker
semantically assumes his communicative responsibility; in the subordinate object clause
he reduces this responsibility. This is why Classical Latin has no subjunctive construc
tions when object clauses depend on main verbs of sub-set  (and D, which, in Latin, is
part of B). These are the so-called verba sentiendi et dicendi (and also: sciendi). In
Classical Latin, these verbs nearly always have object clauses with the Acl type of
subordination.
The cause of this situation is evident: In a language of the SOV type of actant and verb
ordering, constructions like the Acl prove to be a far more efficient and powerful
instrument of subordination. A Latin finite verb is defined in terms of (a) person, (b)
number, (c) genus verbi, (d) tense, (e) affirmation/negation, and (f) mood. Now an AcI
form is marked merely with respect to five of these categories:
- person (dico me/te venisse);
- number (dico me/nos venisse);
- affirmation/negation (dico me venisse/non venisse);
- tense (dico eum venirelventurum esse/venisse);
- genus verbi (facere/fieri; facturum esse/factum iri; fecisse/factum esse).
That is to say that AcI lacks one information which is expressed by all finite Latin verb
forms: it lacks modal information. The AcI construction is the non-modal form of a Latin
verb. In the domain of subject or object clauses it can as it were be used as a "Jack of all
work". While AcI, in Classical Latin, is the only resort to obtain subject and object
clauses when the main verbs belongs to the sub-set  of my scheme, it can also be used,
without any difficulty, for the entire range of subject and object clauses. Thus it is applied
to sub-set  and to the remaining set (A-(B+C)). Exceptions are due to the linguistic
norm, not to the constraints of language system.25 The same system seems to have been
valid also for Umbrian.26
314 WOLFGANG RAIBLE
7. What Happens if SOV Ordering Changes to SVO?

It has been pointed out that Latin reveals a considerable number of techniques
characteristic of SOV languages: case suffixes, the order 'genitive - noun' ("senatus
consultum" against "pater familias"]), subordinate clauses placed in front of the finite
main verb, subordinators in clause-final position (infinitives, participles, subjunctive).27
However, as has already been mentioned at the end of section 3, Latin also shows
characteristics which are far from typical of languages with SOV ordering: it has not only
case suffixes but also propositions (in Classical Latin, there subsist only some few
postpositions); Latin has already a great number of subordinating conjunctions which are
placed at the beginning of subordinate clauses: this is a device that fits very well into the
'actant-verb' ordering of the types 'VSO' or 'SVO'. As to the ordering itself, it is
perfectly true that the SOV type is the rule. However, although the first written Latin
testimony, the so-called Fibula Praenestina with its SVO ordering seems to be a fake,
there is SVO ordering in Latin, too. This shows that Classical Latin was a language that
underwent a crucial change in 'actant-verb' ordering, a change the different results of
which are manifested in the Romance offsprings of Latin.
In the domain of subject and object clauses, the transformation process results in a
multplicity of techniques. According to what has been outlined in essential no. 5, this
scale of subordinating techniques is "the space in which language change occurs". That is
to say that within this scale the techniques of past, present, and of future stages coexist.
This coexistence and the overlapping of the techniques is shown in the following
synopsis. (Table 1).
It is based on some hundred pages of the learned Lateinische Syntax und Stilistik by
Johann Baptist Hofmann and Anton Szantyr. It should perhaps be added that the quoted
grammar is far from supporting the kind of view proposed in the present contribution.28
I The signal for subordination is The signal for subordination is ]

at the end of the clause at the beginning of the clause
A.c.I. subjunctive ut/ne/an/quin quod/quia/quoniam I

alone + subjunctive + subj. or indicative
a) verba voluntaria
e.g. volo (Old Latin volo volo (Old Latin) volo (e.g. Middle
with identical Latin)
subject)
e.g. opto opto, exopto
b) verba cernendi
e.g. curo (only with curo curo (Old Latin) e.g. statuo (e.g.
different subject in decretum est Marsilius of Padova)
Cicero) (Old Latin)
cerno (Old Latin) cerno (e.g. Chronicle
of Fredegarius)
c) verba rogandi
postulo (Old Latin) rogo rogo
peto, impero, iubeo postulo pos tulo
(Old Latin)
d) verba efficiendi
veto (Old Latin) veto (Horace,
sino (Old Latin) sino sino CIL)
prohibeo (Old Latin) prohibeo prohibeo prohibeo (Late Latin) I
facio (seldom in facio (Old Latin) facio
Classical Latin)
e) verba sentiendi
et dicendi
e.g. scio (Old Latin) scio (Old Latin)
video (Old Latin) diff. meaning video (e.g. Petronius)
dico (Old Latin) diff. meaning dico (e.g. Quintilian)
credo (Old Latin) credo (e.g. credo (Middle Latin)
Hieronymus)
puto (Old Latin) puto (Late Latin)
non dubito (Old Latin) non dubito (Late Latin)
dubium non est (Old Lat.) dubium non est dubium non est (Late
quin Latin)
f) verba pollicendi
e.g. promitto, polli- promitto, polliceor
ceor (Old Latin) (e.g. Gregory of Tours)
iuro iuro (e.g. Gregory of
Tours)
e) verba affectuum
e.g. gaudeo (Old Latin) gaudeo (Old Latin;
with quod/quia)
Idoleo (Classical Latin) doleo (Class. Latin)
h) verba impersonalia
e.g. accidit (Old Latin) accidit (Cicero) accidit (Cicero)
Licet (Old Latin) licet
[oportet (Middle Latin) oportet oportet (Late Lat.;
mirum est (Old Latin) mirum est (Jur.)
necesse est (Middle Lat.) necesse est (Old necesse est (Old
Latin) Latin)
opus est (Old Lat.)
Table Part of the scale of Latin techniques for subordinating

subject and object clauses
316 WOLFGANG RAIBLE
The Acl construction was an adequate solution for the SOV ordering type, but where
it has survived in Romance languages, it has been reduced to a marginal existence. The
reason for this near-disappearance is evident: A subordinating device placing the
subordinator at the end of the subject or object clause may be ideal for SOV ordering but
not for the ordering types 'SVO' or 'VSO'. In order to survive in the Romance
languages, the AcI should have had the subordinator in initial position. (This would have
meant VSO ordering in the subordinate clause whereas in Latin subordinate clauses SOV
ordering subsists even much longer than in main clauses.29) An object clause opened by
an accusative, i.e. by a case that is normally attributed to the object, would be a solution
of little use to the hearer. The AcI went through a particular elaboration by the constraints
of written culture (see essential no. 4) and its use was even expanded (contrary to the
norms of Classical Latin) in Late Written Latin. (For this extension, see the synopsis.)
Nevertheless, this type of construction had no chance to survive in spoken language.30
Which one of the subordinating techniques featuring in the synopsis will be the
successor of AcI? As far as the unique subjunctive is concerned, its possibilities are
restricted. In this case, too, a greater range of possibilities could only have subsisted if
SVO ordering in the main sentence had combined with VSO ordering in subordinate
clauses. Then the subjunctive, functioning as a subordinator, would have been in clause-
initial position. But it has to be repeated that the tendency in Latin subordinate clauses to
stick to SOV ordering is even stronger than in main clauses. There is, therefore, no
alternative - unless the subordinate subject or object clause consists of one element only,
the finite verb. This device by which the finite verb of the subordinate clause is brought
into the position of being at the same time the first and the last element in the clause,
subsisted in Old French, and it still exists in Spanish and Italian.
The technique which makes use of the conjunction ut seems to be in perfect harmony
with the SVO type - but there appear to be no traces of ut in the Romance languages. This
has been adduced to "phonetic weakness" of the conjunction. Be that as it may - the true,
structural causes emerge as we look at the synopsis and as we take into account the
shortcomings of the mood (subjunctive) governed by ut. Thus any combination with the
verba dicendi et sentiendi is excluded. This is made clear by the very occurrences in
which ut-clauses are combined with verba sentiendi: in these cases the signification of the
verb changes ('dicere, ut ...' means 'to command', etc.). Since the verba dicendi et
sentiendi are the domain par excellence of AcI, it must have been exactly this group of
verbs that created the greatest difficulties wherever the AcI had to be replaced by another
subordinating technique. On the other hand, it is of interest that we have occurrences of
scire + quod already in the plays of Plautus, that we have dicere + quod since the time of
Quintilianus, videre + quod in the Satiricon of Petronius. It must be noted, too, that - as
has been pointed out e.g. by Hofmann and Szantyr - "the possibility of parataxis has
always been preserved" with respect to the great group of verba sentiendi et dicendi.31 (It
goes without saying that if there were enough space, this technique should be added to the
above synopsis.)
The remaining solution was the construction with quod/quia/quoniam. Our synopsis
makes visible that this technique could be used for the entire range of AcI cases. Even
more: This type of construction could also be used in cases where no AcI construction
was to be replaced, i.e. in the domain of consecutive and temporal clauses. Apart from
being an adequate technique for object clauses in the 'actant-verb' ordering type 'SOV',
this technique presents one further advantage: the choice of subjunctive (with its structural
disadvantages) was not obligatory. At the same time it should be clear that the technique
making use of the conjunctions quod/quia/quoniam as subordinators for subject and
object clauses was considered to be relatively "simple" or "low" style in written culture
(cp. the end of essential no. 5).
8. The Mood Used in quod-Sentences: an IntriguingPicture

The conjunction quod used as a surbordinator at the beginning of the subject or object
clause seems to be an adequate solution for the problem in question. In a language
developing a SVO order we could even expect that a unique subordinator at the beginning
of the clause should be sufficient. Such a language, one might think, should be able to
dispense with the subjunctive mood which, as we have seen, only creates further
problems. Nonetheless, especially beginning with the turn from the 2nd to the 3rd
century, we are confronted with an intriguing picture: the number of constructions with
quod and subjunctive mood is increasing. The fact in itself is not intriguing: for the
hearer, two markers of subordination are better than one, anyway. What is intriguing is
that there are many 'scandalizing' cases - scandalizing from the point of view of modal
318 WOLFGANG RAIBLE
contradiction. Q. Cervidius Scaevola, one of the Digest authors who wrote at the end of
the 2nd century A.D., quotes a fictitious testator as follows:
Scio enim, quia valde me bene ames. (Digests 44,7,61,1)
Here, it is conceivable that Scaevola wanted to characterize the style of certain testators -
the "analytic construction" is not very frequent in the writings of Digest authors.
Nevertheless, we find more examples of this type. Ulpian writes:
Nev enim ignotum est, quod multa contra iuris rigorem pro libertate sint constituía. (Digests
40,4,24,10)
There are even some instances of "sciendum est, quod..." or "notandum est, quod...". A
further example of Ulpianus:
Praeterea in hac actione notandum est, quod reus contumax per in litem iusiurandum petitoris damnati
possit "ei iudice quantitatem taxante". (Digests 10,4,3,2)
What is intriguing in these examples is the fact that, all of a sudden, the modal
contradiction seems to be irrelevant. If the speaker says "scio", this means that he knows
what he says and that he, therefore, can assume his communicative responsibility. The
subjunctive mood in the subordinate clause contradicts this. The same holds for the
subject clause in the case of "nec ignotum est, quod...". I will give another example. We
have the same scandalizing cases in the astrological work of Firmicius Maternus (who
died after 360 A.D.).
- Scire autem nos convenit, quod nulla in hoc loco gaudeat stella nisi sola Luna (Math. II,19,9);
- Scire itaque nos principe in loco oportet (...) quod ad imaginem speciemque mundi formam
hominis (...) Deus ille (...) perfecerit (Math. III, prooem. I).
The same facts can be observed for instance in the famous Peregrinado Aetheriae. The
authors of the 3rd, 4th, partly even of the 5th century had apparently a strong tendency to
generalize the use of the subjunctive mood in subject and object clauses. In the work of
Lucifer of Cagliari (who died around 371 A.D.) we find more than 60 quod-clauses with
subjunctive against 8 cases with indicative.32
9. Transitory Phenomena
If we want a satisfying explanation for the intriguing picture outlined in section 8, we will
have to speak once again of the change in the ordering of the actants and the verb. In this
domain we have two valuable articles written by J J. Schlicher on the style of Caesar and
the Roman historians. These studies are valuable because, among others, they treat the
phenomenon of ordering within the frame of one and the same type of texts.33 As has
already been mentioned, Caesar is an extreme adept of SOV ordering. Nevertheless, there
is a certain development. Schlicher tells us:
A second means of more even distribution[hemeans: besides the possibility to make two coordinated
sentences out of one complex sentence], which flew directly in the face of the periodic sentence, was
that of placing the participle phraseorablative absolute not before, but after the dominant verb: "de
tertia vigilia ad hostes contendit, eo minus veritus navibus quod in litore molli atque aperto deligatas
ad ancoras relinquebat (BG 5,9).34
In Cesar only two percent of the clauses are placed behind the finite verb. In the writings
of Cornelius Nepos and Sallustus already about ten percent of the clauses, especially
participles, are placed after the main verb. In Curtius Rufus and Tacitus more than 20
percent of the clauses are postposed to the verb. Commenting on this phenomenon
Schlicher mentions the characteristic feature that the clauses after the verb are considerably
more frequent in reported speech than in the narrative passages of the respective authors.
In Latin, the change in the ordering of the verb and the actants does not seem to have
been a catastrophic event. It must have rather been a lengthy process lasting for
centuries.35 So it is interesting to note that Paolo Ramat counted only 32 cases of SVO
ordering against 54 instances of the "old" type 'SOV' in the Chrestomathy of Pompeian
inscriptions collected by Ernst Diehl.36 If we analyze the Peregrinatio Aetheriae (5th
century), SVO ordering is already dominant, but by no means predominant. This means
that we have a long period of transition in which the "old" ordering (SOV) certainly was
no longer "normal" - but at the same time not yet considered "unnatural". On the other
hand, the "new" SVO ordering was not yet normal, but not unnatural either. The same
holds for subject and especially for object clauses. The position in front of the finite verb
was no longer the normal position, but it was far from being anomalous. Likewise the
position after the finite verb was neither normal nor anomalous.
The consequences are evident: In languages with SOV ordering the subordinate clause
precedes the dominant finite verb and the subordinator is placed at the end of the clause (if
subjunctive mood functions as subordinator, we get cases such as "in tabulas publicas
referatur facito"). In languages with VSO or SVO ordering the dominant finite verb
precedes the subordinate clause. The subordinator (conjunction) has to be placed at the
320 WOLFGANG RAIBLE
beginning of the subordinate clause ("doleo, quod..."). Now during the transition period
clausal subordination is, so to speak, overcharacterized. Both the conjunction and the
modal subordination (subjunctive) are used as subordinators in order to serve the interests
of the hearer. In the case of the verba sentiendi et dicendi, especially the epistemic verbs
whose special structure will be explained later, this syntactical over-characterization takes
place at the expense of the modal contradiction. Thus for a certain period Latin undergoes
the same phenomenon which the Romance languages have to face in a special case: In the
Romance languages subject and object clauses have their normal place after the dominant
finite verb. Once they are placed in front of the main verb, i.e. once they exhibit a position
the hearer is not accustomed to, they show the same double syntactical characterization by
means of the introducing conjunction in combination with the subjunctive mood ("je sais
que Pierre est malade" vs. "que Pierre soit malade, je le sais").37
10. The New Technique Takes Shape

In the 3rd, 4th, and 5th centuries, we have a number of texts which show an increasing
amount of SVO ordering. At the latest by the 6th century, Latin seems to return to the
"old" SOV ordering of the verb and the actants. Yet this should be specified: it holds for
written Latin. The reason for this development is evident: The gap between spoken and
written Latin has become too wide, written Latin is increasingly taught and assumes
prestige of a learned language (in the Middle Ages 'grammatica' therefore was
synonymous with 'Latin'). The written culture proves to be very conservative, tenacious,
and resistant. The gap between spoken and written language creates, among others,
linguistic consciousness. For a large part, the linguistic observer of the 20th century must
base his conclusions on indirect evidence.
A part of this indirect evidence can be derived from the fact that in the 6th century
(and partly before) the subordinating technique using quod, quia, quoniam has become
perfect. The "scandalizing" cases have disappeared. I intend to show this with the help of
the Historia Francorum by Gregory of Tours (538-594). Gregory is at the same time a
good example for the growth of linguistic consciousness. For example, he tells us that he
is aware of the fact that he writes "bad" Latin, although he stands in the tradition of Latin
written culture. Thus, three out of four subject or object clauses are construed with the
AcI. As has already been mentioned, in (written) Late Latin the use of Acl extends even to
verbs which do not belong to the "classical" AcI verbs. This strong Acl tradition is still
characteristic of parts of Middle Latin - although this subordinating technique has become
obsolete in spoken language. What is therefore interesting in the work of Gregory are not
the AcI cases, but the remaining 25 percent of subject and object clauses. These are the
225 cases I have examined. 219 out of these 225 subordinate clauses are placed after the
dominant verb. The position of the subject and object clauses has definitely changed.
The following table shows the conjunctions used by Gregory (there are still some
cases of qualiter) and the grammatical mood which appears in the subordinate clauses.
conjunction used
quod quia
grammatical 121 = 54% 104 = 46%

mood used in the
subordinate clause
subjunctive 103 = 85% 10 = 10%

indicative 18 = 15% 51 + 43 = 90%
FIGURE 4
First I have to explain why there are two numbers for 'quia + indicative': In Greek there
exists a use of the conjunction hóti which corresponds roughly to a colon and an opening
quotation mark in our written language. 43 of the instances with quia belong to this type.
(Accordingly, in 33 out of the 43 cases the dominant verb is dicere) - From this table it
becomes immediately evident that there is a strong tendency (but merely a tendency) for
quod to combine with the subjunctive; the reverse is true for quia. It has still to be
322 WOLFGANG RAIBLE
mentioned that the verbs going with quia form a concise group (corresponding to sub-set
 on p. 313), whereas the occurrences with quod are scattered over a large number of
generally not very frequent verbs.
The quantitative analysis can only indicate tendencies - for verbs that are frequent in
the main clauses are construed both with quod and quia, and nearly always both with
indicative and subjunctive. Max Bonnet who wrote an outstanding work on the Latin of
Gregory of Tours in 1890, points out that in sentences which otherwise are absolutely
identical, we find sometimes the indicative, sometimes the subjunctive mood, both after
quia and quod. Apparently, the situation keeps on to be bothering.
However, a thorough analysis will show that Gregory is perfectly systematic (and
that seemingly identical sentences are not identical): Gregory uses subjunctive mood as a
supplementary subordinating signal - but whenever this leads to a modal contradiction
between the meaning of the dominant verb and the subordinate clause, the indicative
replaces the subjunctive.
The following scheme (Figure 5) is a more sophisticated version of the scheme

outlined on page 314-315. The big circle represents all verbs which govern subject and
object clauses. The two intersecting circles within the big circle have become more
complicated. There is a distinction between epistemic verbs strictu sensu (gr. epístasthai
means 'to know'), i.e. verbs like 'to know' and their opposites, and epistemic verbs in a
larger sense, i.e. verbs which signal, by virtue of their meaning, whether or not the
speaker will take over his communicative responsibility. These are verbs such as 'to say',
'to claim', 'to maintain', 'to assert' (or their opposites). - That the grammatical persons
are taken into account is a further refinement in this scheme. The role of grammatical
persons will become evident in the case of epistemic verbs. - The hatched area of the
scheme is the domain of the subjunctive in subordinate subject or object clauses. This is
as it were the "normal" solution.
FIGURE 5
324 WOLFGANG RAIBLE
The following cases can be distinguished:

(a) Epistemic verbs - There are some verbs which, by virtue of their semantics,
always imply that the speaker assumes the communicative responsibility. This is why
these verbs are always formed with the indicative: since this implication is of a semantic
nature, it does not matter at all whether the verb is finite or infinite, whether it is in the
first, in the second, or in the third grammatical person, provided that they be used in their
positive (affirmative) form: therefore, to avoid modal contradiction, we have 'I know',
'you know', 'he knows' with the indicative mood. But when I say "I ignore", the normal
subjunctive can follow, whereas "you ignore" implies that I know, so we expect the
indicative in the subordinate clause. It is not worth while to allege the numerous "normal"
cases in the indicative with verbs of the family 'scire' and 'novisse'. Instead, I shall quote
examples with negative epistemic verbs. (All quotations are from Gregory's Historia
Francorum, 2nd edition in the Monumenta Germaniae Historica. Readers who are
acquainted only with Classical Latin should keep in mind that Gregory writes in 6th
century Latin.)
... declamare plurima de rege coepit  dicere, quod filii eius regnum capere non possint, eo quod
[because] mater eorum ex familia Magnacharii quondam adscita regis torum adisset, ignorans, quod,
pratermissis nunc generibus feminarum, regis vocitantur liberi, qui de regibus fuerant procreati
(228,9).
When the author says of someone that he ignores something, this implies that the author
himself does not ignore it. He therefore has to use the indicative in order to avoid a modal
contradiction.
Sed  memor fuit Childeberthus [but I, the author, know it], quod, quotienscumque aduersus fratrem
suum egit, semper confusus abscessit (148,10).
(b) Epistemic verbs in a larger sense, first person - If the main verb is an epistemic
verb in the broad sense and if it is in the first person (thus implying the speaker of the
sentence), we obtain the indicative mood in the subordinate clause in order to avoid modal
contradiction.
Audio frater et episcope, quia regis gratiam non habes, ideoque nec nostram caritatem uti
poteris, priusquam regis indulgentia merearis (223,3).
Verum enim dico tibi, quia, edam si regi epulum cupias praeparare, fercula regalia
conponere possum, ne quisquam a me melius (113,10).
Nam in ventate loquor, quia vidi per somnium a basilica sancti Vincenti veniente virum
inluminatum, tenente manu caereum et domus negutiantum (401,10).
Videmus, quia te valde diligit pars maxima populorum ( 139,4).
(c) Epistemic verbs in a larger sense, second and third person - The true modal
conflict can only arise between the speaker of the entire sentence and the content of the
subject or object clause. With the group of verbs under discussion, the subordinate clause
may thus contain the "normal" subjunctive if the verb of the matrix sentence is in the
second or in the third person or in form of a participle referring to a second or third
person. I shall quote some examples with participles:
Detrahebant enim et de episcopo, dicentes, quia illius dolo et haec turbatae et mona-
stirium fuerit derelictum (464,1).
Sed audientes, quod scilicit tali fuissit affectus interitu... (423,19).
Dum autem Chrodigildis regina Parisius [in Paris] moraretur, videns Childeberthus, quod
mater sua filius [ace. pl.] Chlodomeris, quos supra memoravimus, unico affectu diligeret,
invidia ductus  metuens, ne favente regina admitterentur in regno, misit clam ad fratrem
suum Chlothacharium regem dicens... (117,17).
This group includes a bulk of cases with verbs such as adserebant, ait, audiunt, audivit,
certa est, certi sunt, dicunt, iuravit, promittunt, videt and so on.
(d) Epistemic verbs in a larger sense: "exceptions" - There are cases where the first
person is implied by the context while the verb stands actually in the second or third
person. Some examples will make this clear:
Ad haec ego: bene ais, quia, cum egressus fuerit ab homine spiritus et iacuerit corpus
mortuum, non cogitat de his quae in mundo relinquit [you say this, but I would say this,
too] (498,15).
Potasne ["don't you think - I at least do so"], quia non est super quisquam de filiis
Chlothari regis... (158,15).
Renuntiate rege vostro, quia rex sum, sicut et ille [here the first person is expressed in the
object sentence] (110,20).
(e) Role of the position of the subordinate clause - The "normal" position for the
object and subject clauses has become the position after the main verb. Among 225
examples only six are exceptions. In five of these six occurrences the main verb is an
epistemic verb. Hence it will be particularly interesting to observe how the modal conflict
is solved here. If a modal contradiction is avoided, the hearer disposes of only one mark
326 WOLFGANG RAIBLE
of subordination (the conjunction), whereas the unusual position should motivate further
syntactic security nets. However, should the syntactic security decide the issue, the modal
conflict would be inevitable. Here a few examples:
Quod me Chlothacharius pater meus exosum habuerit, habetur incognitum nulli. Quod
autem ab eo vel deinceps a fratribus sim tonsuratus, manifestum est omnibus... (357,15).
Quod Dominus in adsumpto hominem mortuus fuerit ac resurrexerit, non ambigo (496,24).
Quod fuerim amicus regis Chilperici, negare non potero, non tarnen contra utilitatem regis
Childeberthi haec amicitia pullulavit (511,4).
The last instance I am going to quote shows one of those epistemic verbs that do not have
implications for the speaker when used in the first person, whereas they have implications
when used in the second or third person. Obviously this has to be the verb 'nescire' or
'ignorare':
Quod sim iniustus, tu nescis (219,18).
11. Some Romance Tendencies

In the preceding section I did not consider the 75 percent of cases in which Gregory uses
the technique of the past in the domain of subject and object clause marking, the AcI,
which survives in written texts. Instead I dealt with the remaining 25 percent in which he
uses the actual technique of spoken language (it will have been noticed that a great number
of the examples quoted in the preceding section are specimens of direct or reported
speech). Although it has been pretended that these remaining 25 percent present a rather
chaotic picture, a closer examination has revealed that the situation is everything but
chaotic. In reality a new technique of subordination with clear-cut features has emerged.
And moreover, we can see the outlines of at least three further developments.
1 . I n the first type of development, the two or more subordinating conjunctions
which can be placed at the beginning of subject or object clauses (quod, quia,
qualiter, eo quod) merge to one conjunction, the well-known que. There are
sporadic instances of written que in texts of the 8th century. In this type of
development the subjunctive mood keeps its function as a second subordination
signal (the third being the post-position of the clause).
This is the solution which was largely adopted in Western Romance. The disadvantages
are evident: Whenever there is modal contradiction, the "normal" subordinating
subjunctive has to be replaced by the indicative. This is a problem I have dealt with on
various occasions.38 There are some minor differences between Italian on the one hand
and French/Spanish on the other: in Italian the verbs located between knowing and not-
knowing, i.e. the verbs of belief, are treated as verbs of not-knowing, whereas in French
and Spanish the same verbs belong to the verbs of knowing.39 This type of solution too
entails disadvantages: different matrix sentence verbs govern different con-structions. The
affirmative form of certain verbs is construed in another mood than the negative one ("je
crois qu'il est malade" vs. "je ne crois pas qu'il soit malade"). Nevertheless this solution
also makes possible a number of subtle differences for which, in other languages, two
different conjunctions are needed ("il ne savait pas que Pierre était malade" ['that', 'dass'
- i.e. I,the speaker, know it] vs. "il ne savait pas qu'il fût malade" ['whether', 'ob' - i.e.
ƒ, the speaker, do not show whether I know it or not]).
In the work of Gregory of Tours we observed a tendency to use quia with the
indicative, quod with the subjunctive. This forms the basis for a second type of solution:
2. While one and the same conjunction is used with subjunctive in the first type and,
in cases of modal contradiction, also with the indicative, in the second type one
conjunction is used with indicative, another with subjunctive. Thus the second
type of solution provides the conjunction with a supplementary information. This
may result in the reduction of modal distinctions (subjunctive vs. indicative) in the
verb of the subordinate clause.
This is the solution prevailing in Eastern Romance languages and in Modern Greek. It
also allows the modal distinction in the verb to be transposed from the end of the verb
(suffix) to the head. If in this case modern Greek should have served as a model for
Roumanian and for dialects in Southern Italy and Sicily, one should again take into
account that Greek had to cope with the same problems to re-organize its subordinating
system according to a change in the type of ordering of the actants and the verb. Identical
problems sometimes may create the same solutions. Be it as it were, speakers of
Roumanian or of the South Italian dialects would not have developed or adopted this
solution had the system of their languages not been compatible with the device.
In both types, anyway, subjunctive mood has become a second or third order
subordinating device - the most important signal being the conjunction at the head of the
clause. Consequently, the subjunctive can be reduced in the verbal system. For instance,
we can observe a tendency to reduce the tenses of subjunctive mood. In his grammar of
Romance languages, Wilhelm Meyer-Lübke wrote (, § 665):
328 WOLFGANG RAIBLE
The Romance languages are not overly creative in the domain of the subjunctive, they have rather the
tendency to reduce the range of its applications; in spite of occasional contrary tendencies they might
come nearer and nearer to the end which has almost been reached by some dialects, i.e. to abandon
wholly the subjunctive mood.
It has repeatedly been mentioned that in written texts of the 3rd, 4th, and 5th century we
can observe an increasing tendency towards SVO ordering. This tendency holds for main
clauses, not for subordinate clauses. In subordinate clauses, the speakers stick to SOV
ordering. This may be the motive for a third solution:
3. According to the 'SVO' type and to solutions 1 and 2, subordination is marked by
a conjunction, too. But contrary to solutions 1 and 2, a difference in ordering -
SOV as against SVO - is used as an additional signal for subordination.
This solution has been adopted in the Germanic domain. In main sentences we find SVO
ordering:
Peter (S) sieht (V) seinen Vater (0).
If this sentence is imbedded in a matrix sentence, the imbedded sentence continues to have
the "old" ordering SOV:
Ich glaube, daß Peter (S) seinen Vater (0) sieht (V). 40
In this Germanic solution the different ordering of the verb and the actants in the
subordinate clause replaces the subjunctive mood as a signal for subordination.
Nevertheless, the subjunctive survives in the Germanic domain in a function different
from the one it serves in Romance languages: in the Romance languages the domain of
reported speech has, as a rule, been separated from the domain of subjunctive (an
exception being Reto-Romance). In German, on the other hand, speakers use subjunctive
mood in order to mark reported speech without a subordinating conjunction:
Er glaubte, sein Vater sei nach Hause gekommen.
In this context German has, at least in written language, established an additonal

difference between subjunctive I and subjunctive II. Thus German written language
affords the luxury of a special subjunctive for reported speech.
It should perhaps be added that there are also authors who, unlike Meyer-Lübke, see
an increase in the use of subjunctive mood in the Romance languages. Behind this vision
there is the idea that the speakers of a language make no or little use of subjunctive mood
when the language is still "young", but that the use of this mood is to become more
frequent as the language gets more sophisticated or "grows older". For Spanish, this
conception has been put forward by Antoni Badía y Margarit. His view is based, among
others, on the fact that in the Latin civil law documents written in Spain in the 8th, 9th,
and 10th century great use is made of quod + indicative. But Juan Bastardas Parera, on
whom Badía relies, did not see that these indicatives are due to the nature of the governing
verb. In the highly formulaic civil law documents this is nearly always a sentence
beginning with
omnibus non est incognitum, quod... (or the like).
This is one of the epistemic verbs which, in order to avoid a modal contradiction, are
construed with the indicative and not with subjunctive in the subordinate clause.41
12. Some Concluding Remarks

On various occasions Hansjakob Seiler has told us that language typology has to make
visible the features which give a language system its structure and which hold it together.
He refers as well to a remark of the late Emile Benveniste,
Le donné linguistique est un résultat, et il faut chercher de quoi il résulte.42
On this background a few concluding remarks are in order for discussion.

1. The conception I have outlined with the help of the typological view and the first
three of my essentials is likely to reduce the value of super-, supra-, and adstratum
theories in language change.43 On the one side, identical problems can (but do not have
to) lead to identical solutions, even without language contact; on the other hand, the
contact between languages with different problems and different solutions is likely to be
without consequences, at least without direct consequences: According to the translation
program developed by Alfonso el Sabio, it were the entirely different subordinating
techniques of written Arabic that triggered and accelerated the further development of the
subordinating techniques in Spanish.
2. Considering the first three of my essentials, one might presume that an "optimal"
language system should be homogenously constructed in the sense of word order
typology. Apart from the fact that - as has already been mentioned at the beginning - an
ideal solution for the linearization of pluridimensional entities will never be found, there
330 WOLFGANG RAIBLE
are at least four reasons to believe that features fitting into the ideal picture will always
occur mixed with other features.
(a) The essentials outlined at the beginning imply the notion of syntactic hierarchy.
On the level of the verb and the actants, it is to be expected that speakers tend to
organize the morphological markers in a more or less homogenous way. But the
actants may have a more or less complicated inner structure (noun + adjective,
noun + genitive, noun + relative clause, and so on); clauses have by definition a
structure of their own. On this second level (and on further levels of syntactic
organization), it is unlikely to find the same principles rigorously applied. The
ordering of the actants and the verb in German clauses is a good example. In this
case different ordering is one of the signals that make clauses stand out against
sentences.44
(b) A second reason for the inhomogeneity of solutions which can be observed in
particular languages is the phenomenon outlined in my essential no. 5: The status
quo of a language always represents the simultaneousness of the non-
simultaneous. We have seen that different techniques of subordination coexist,
side be side, in the synchrony of Latin. Some of these techniques conform to the
new principle of ordering, others do not. It usually takes a very long time until
out-dated solutions disappear totally. Especially the language of written culture
favours techniques growing more and more obsolete in spoken language.
(c) So far I have merely spoken of different levels of syntax. But syntax can be seen
as opposed to morphology, morphology as opposed to phonology. To a certain
degree these different systems have an existence and an evolution of their own.45
This may create interferences and irregularities. For instance Latin A-declension is
the prototype for the further evolution of noun-marking in the Romance
languages. The O-declension followed this evolution with a certain delay, thus
creating heterogeneities in historical stages of some Romance languages.
(d) In an earlier contribution I have given some thought to the relationship between
analogy and anomaly in language. In this context I have applied to language the
Heraclitean picture of the bow with its diverging ends, maintained in a proper
balance by the string. One end of the bow of language is analogy, the other
economy. The string which keeps together both ends corresponds to the goal-
orientedness of a language system: it is made for the sake of communication.
Apart from symbolizing the inner tension and the dynamics of a language system,
this picture shows very well the role of economy as a factor of anomaly in
language.46
The four points adduced as reasons for inconsistencies in terms of word order typology
should not be interpreted as factors leading to anarchy or confusion in language systems.
There is after all a great amount of order in every language system.
The Cologne research group on language universais and language typology worked
out the concept of dimension. Dimensions are answers to problems of communication
faced by every speaker in every language. The dimensions of the Cologne research group
always imply two opposed principles and a scale of techniques arranged between these
poles. One should now think that the entire scale of techniques could be available at any
moment for all language systems. Instead, languages (or speakers of languages) make
typical choices. As regards the dimension of 'Apprehension', Ulrike Kölver has made the
following point:
(...) The scale also comprises a certain range of techniques which largely are mutually exclusive on
the plain of individual languages, that is, they are specific for particular groups of languages.47
In the present context this means that choices made e.g. on the level of the linearization of
the actants and the verb reduce the possibility of choices within the "nominal" or
"actantial" domain (or in the domain of "ajouts", as René Thorn would put it). Choices
entail other choices, and this leads to the "clustering" of techniques and to a typological
profile.
3. The considerations outlined in this paper might perhaps be a contribution to a
different picture of language and language change. With the help of the line of thought
proposed here, language history and language change might lose their aspect of
contingency; to a certain extent they might become truly understandable. At the same time
it should be clear that a true understanding of language change makes indispensable a
complementary universalistic view of things, since we have to know the continua - and
the whole scales of appropriate techniques which belong to them - in order to understand
the meaning of a choice or a clustering of techniques. This combination of historical with
general linguistics has always distinguished the thinking of Oswald Szemerényi.
NOTES
1 This contribution was conceived mainly in 1982/83 as part of a research project financed by the
Deutsche Forschungsgemeinschaft. A first version was presented at the International Interdisciplinary
332 WOLFGANG RAIBLE
Conference held at Gummersbach/Cologne, Germany, Sept. 18-23, 1983, but not published in the
respective volume (Seiler/Brettschneider 1985), mainly owing to its proportions. The study is
situated at the intersection of two lines of thought. The first one are the considerations I have given
the problem of what I call "the regularity of the irregular". They have resulted in a series of articles
devoted, among others, to subordination. The second line of thought is language change as related to
language typology. The confluence of these lines of thought makes out of this contribution, among
others, the historical foundation for the work I have done on the subject of the Romance
subjunctive. - Thanks to Cynthia Stäbler, née Dermarkar, the quality of my high-school English has
been considerably improved.
2 Cp. to this effect Lüdtke 1980.
3 . Raible 1980.
4 For the difference between spoken and written discourse, cp. the stimulating contributions by Koch/
Oesterreicher 1985 and Ludwig 1986. See also Raible 1987
5 Cp. Bossong 1979.
6 Cp. for this notion my article "Relatinisierungstendenzen", forthcoming in Lexikon der
romanistischen Linguistik, vol. I (article no. 100).
7 For the idea of dimensions manifesting themselves in a scale of techniques, I refer to Hansjakob
Seiler and his research project on language universais and typology. Cp. for instance the
"Introduction" in Seiler 1986.
8 I refer to an important contribution by Klaus Heger (Heger 1982).
9 As regards the hypothetical Indo-European original language as SOV- langage, see for instance
Braunmüller 1982 and Schrodt 1983.
10 See e.g. Hofmann/Szantyr 1965.
11 It should be clear that there have always been exceptions due to the iconic representation of contents,
especially in the case of hypothetic periods ("if... then") or consecutive constructions.
12 In some of the Romance languages the possibility of subordination by means of the unique
subjunctive subsists, but only under very restrictive conditions. The most important of these
conditions is the (immediate) post-position of the subjunctive. Cp. ital. "credo venga". The Latin
ordering "venga credo" is not possible.
13 Comrie 1981.
14 For somewhat similar considerations, against a background of Generative Semantics, see Antinucci
1977.
15 Cp. Thorn 1973, here p. 242. Later on, somewhat similar ideas were put forward by the Hungarian
fenno-ugrist P. Hajdu. - It is striking that ideas with a particularly fruitful impact in a branch of
science often come from outside the discipline. In linguistics, apart from René Thorn (mathematics
and topology), good examples are Francesco Antinucci, Karl Bühler (psychologists) or Joseph H.
Greenberg (anthropology).
16 The same principle of linkage could apply to the marking, for the hearer, of the relation between two
nouns, for instance within an actantial syntagm: In a 'genitive-noun' ordering, the linkage element is
optimally placed at the end of the first noun; in the reverse ordering 'noun-genitive', the optimal
position is at the head of the 'genitive'. A further characteristic of the type 'SOV' are postdetermining
question particles - e.g. Latin -ne which is added to that part of the sentence which, in a particular
case, is in question.
17 Tesnière 1959, livre I, ch. 7, § 3.
18 For Japanese, see e.g. Lewin 1959, § 118 ("Flexionsformen des Verbs"), § 213 ("Die
Satzverbindung"), §§ 214, 216. - For Turkish I consulted Lewis 1965. For the special forms and
problems of relative clauses the stimulating thesis of Lehmann 1984 should be consulted.
19 Eugenio Coseriu has underlined the superficial, if not arbitrary character of a typology based on
SOV-ordering. Strictly speaking, the Japanese categories S, O, V would not correspond to, say,
Latin S, O, V (Coseriu 1983). On the level discussed here this is not the problem. What is at issue
is the ordering of the actants and the verb, not of the categories S, O, V as categories of a particular
language. See essential no. 6, p. 301.
20 The Latin grammarians saw this subordinative function very clearly.
21 Heger 1976, § 6.1.; . Heger 1977:1 and 14 (§ 3.2.). - For a further elaboration of this concept see
Ludwig 1988, esp. chapters I and III.
22 Verb forms for which the speaker assumes his communicative responsibility are those which can be
contradicted. This means that, in terms of logic, they imply truth values.
23 See Raible 1983.
24 See Hofmann/Szantyr 1965.
25 Within the framework of an "ideal" SOV language, the only weak point of the AcI is the fact that
there may appear two forms in the accusative. Therefore misunderstandings are not entirely excluded.
Nevertheless, in the case of ambiguity, passive transformation will help. One would not say
"Camillum Brennum vicisse satis constat"; in Latin, this sentence should read: "Brennum a Camillo
victum esse satis constat".
26 I refer to the material exposed by Helmut Rix (Rix 1976). In Umbrian, as attested by the Tabulae
Iguvinae, we have the same devices for subordination: subjunctive mood and infinite constructions.
The explanation proposed by Rix has two shortcomings. (1) Rix gives the infinitive a modal
function ("L'infinitif et l'indicatif caractérisent l'événement rapporté par la complétive comme réel ou
vrai - ou encore nient sa réalité ou vérité -, sans donner lieu à doutes" [p. 224]), whereas, in our
view, the infinitive, being modally unmarked, takes on the modal information required by the
context (Jack of all work). (2) He fails to see the modal conflict which may emerge from the
semantic nature of the governing verb ("Une détermination du mode du verbe subordonné par la
nature du verbe régissant n'existe pas en latin, dont la syntaxe est très proche de celle de l'ombrien: le
même auteur dit, naturellement pour exprimer des choses différentes, dicere ... Siculos ... questos
esse (Cic. Verr. 3,103) et dixisse, ut ... scriberet (Cic. Att. 11,7,2)..., [p. 222]). The point is
"naturellement pour exprimer des choses différentes". There is a considerable difference of meaning
between the type "pronuntiaverunt recte curatum non esse" (prusikurent kuratu rehte neip eru) and
the type "itidem dicito eant" (sururont deit etaias). Thus there is no basis for the following
conclusion: "Le fait que dans une complétive dépendant d'un verbe de la classe sémantique de 'dire' se
trouve ici [first type] l'infinitif et dans d'autres cas [second case] le subjonctif exclut la possibilité
que le mode de la complétive soit régi par la classe sémantique du verbe régissant." - There is only
one rule which has to be respected: to avoid modal contradiction (see also note 28).
27 For some of these characteristics, see Adams 1976, Ramat 1980, Andersen 1979. - A very good
survey of the problems occurring in the field of word order typology and word order change is given
by Harris 1984. This discussion comprises such factors as perception problems (put forward e.g. by
Antinucci) and a conflict between strategies (esp. on pp. 188- 192). Harris is not familiar with the
proposals made by René Thorn and the idea of syntactic hierarchy does not seem to play a prominent
role in his discussion.
334 WOLFGANG RAIBLE
28 József Herman has proposed, on a smaller scale - i.e. only for the verbs dicere and dolere - the same
kind of synopsis, but he omitted the possibility of subordination by the unique subjunctive and the
constructions with ut (Herman 1963:34sq.) - Some valuable ideas have been developed on the
subject under discussion by Anna Machtelt Bolkestein (especially 1976; see also 1976, 1977). In the
domain of verba dicendi the author establishes a distinction between three groups of verbs: (1)
neutral verba dicendi which, by virtue of the following complement clause, may be interpreted as
imperative verbs (with the ut-technique) or as declarative verbs (with the Acl-technique). (2)
Semantically imperative verbs which can be constructed both with the ut-technique or Acl[because,
in my interpetation, there emerges no modal contradiction when the ut-technique is used whereas the
Acl-technique can do no harm at all.] (3) Declarative verbs which, in my interpetation, only are
compatible with Acl-technique. - Unfortunately, the author confines herself to the group of verba
dicendi and to only two of the possible techniques.
29 The niche for former Acl constructions are the verbs of perception and the causative construction
with Tacere' (used with Acl only in Late Latin). A typical example in French is the construction "je
le vois venir" or "je vois venir Jean". There is a strong tendency not to separate the finite verb of the
main clause from the infinitive of the "subordinate clause". This brings the non-finite form
signalling subordination to the beginning of the subordinate clause.
30 For the Acl and its development, see especially Perrochat 1932.
31 Hofmann/Szantyr 1965, § 298 (p. 528). See also section V of Raible 1983.
32 As to the "intriguing picture", see e.g. Hofmann/Szantyr 1965, § 312, p. 577: "Der Modusgebrauch
bei quod, quia und quoniam und nach den verba sentiendi et dicendi unterliegt im Spätlatein starken
Schwankungen und ist zusammengefaßt noch immer nicht untersucht." - Kiss 1982 speaks of a
"crisis of subjunctive in subordinate clauses". - Ever since the doctoral thesis of G. Mayen (Mayen
1889), Greek influence has been alleged in order to explain the new preference to use the conjunction
quod for subordination. This may not be wrong insofar as speakers of the ancient Greek language had
to solve the same problem of subordination - and in a former SOV language answers to problems
may be similar. But this is not the point in the case in question, because the subjunctive mood
characteristic of Latin cases is no problem in Greek - Greek uses a conjunction at the head of the
subordinate subject or object clause (hóti, hôs), the mood being the indicative, sometimes the
optative.
33 In another article I have shown - in the domain of Latin - the importance of discourse traditions for
the interpretation of language phenomena (see Raible 1985).
34 Schlicher 1936; see also Schlicher 1933.
35 One of the first valuable contributions to this topic was made by Richter 1903. More recently, Koll
1965 gave a survey essentially based on the chrestomathy Sermo vulgaris latinus of Gerhard Rohlfs
(21956). Although Koll has only counted one to two pages of each of the texts we find in this
chrestomathy, the study merits attention because the author takes into account the relevant studies
written on this topic in the first part of the century.
36 See Ramat, quoted above in note 27, p. 188.
37 For this aspect see Raible 1983. For the conception of the two- or threefold syntactic security nets
("filets de secours") one of which is subjunctive mood, I am indebted to Jorgen Schmitt Jensen
(Schmitt Jensen 1970).
38 E.g. Raible 1980 and 1983.
39 This is one of the results of the doctoral thesis of Renate Tretzel ("Glauben heißt nicht immer
wissen oder: Der Konjunktiv in abhängigen Subjekt- und Objektsätzen"), submitted at the
University of Freiburg in 1984.
40 For the history of subjunctive in German, see also Richard Schrodt 1983.
41 Badía y Margarit 1953; Bastardas Parera 1953.
42 Benveniste 1966:99-118. The sentence is on p. 117.
43 Cp. to this respect for instance Lüdtke 1987.
44 In the article cited above Thom 1973. René Thom did already show that different principles of
ordering on different syntactical levels are a good means of distinguishing levels: "Il importe (...) que
les ajouts soient systématiquement perçus comme des ajouts, inférieurs en stabilité au verbe central.
Un des meilleurs moyens d'y parvenir est de leur conférer une typologie opposée à celle de la phrase
nucléaire. En effet, comme le montre par exemple la transformation interrogative, (vient-il?: VS), le
renversement des typologies déstabilise une phrase en produisant sur l'auditeur un effet de choc. De
ce point de vue, l'inversion de la typologie des ajouts libres est une sorte de tribut que l'ajout libre
paie au verbe central, en signe de vassalité. On pourrait, je crois, montrer que la règle d'inversion de
la typologie des ajouts a pour effet d'augmenter la stabilité sémantique d'une phrase, en diminuant la
possibilité d'apparition d'ambiguité." - Recently, Wulf Oesterreicher has developed the same line of
thought in an important contribution (Oesterreicher 1988). In his view, the phenomenon I have
called, in a series of contributions, "the regularity of the irregular", constitutes one of the factors
creating heterogeneity.
45 This conception has been developed especially by Wolfgang U. Dressier on various occasions.
46 Raible 1980, 1983. - On the basis of similar thoughts Otmar Werner comes to the conclusion that
natural language systems tend to be a good mixture of "regular" and "irregular" devices. See e.g.
Werner 1987.
47 Kölver 1985.
BIBLIOGRAPHY
Adams, J.N. 1976. A Typological Approach to Latin Word Order. Indogermanische Forschungen 86.70-
99.
Andersen, Paul Kent. 1979. Word order typology and prepositions in Old Indic. Festschrift für Oswald
Szemerényi zum 65. Geburtstag, hrsg. von Bela Brogyanyi. 23-34. Amsterdam/Philadelphia.
Antinucci, Francesco. 1977. Fondamenti di una teoria tipologica del linguaggio. Bologna.
Benveniste, Emile. 1966. La classification des langues. Emile Benveniste. Problèmes de linguistique
générale, vol. 1.99-118.
Badia y Margarit, Antony. 1953. El subjuntivo de subordinación en las lenguas romances y especialmente
en iberorománico. Revista de filología española 37.95-129.
Bastardas Parera, Juan. 1953. Particularidades sintácticas del latín medieval. Barcelona/Madrid.
Bossong, Georg. 1979. Probleme der Übersetzung wissenschaftlicher Werke aus dem Arabischen in das
Altspanische zur Zeit Alfons' des Weisen. Tübingen.
Braunmüller, Kurt. 1982. Syntaxtypologische Studien zum Germanischen. Tübingen.
Comrie, Bernard. 1981. Language Universals and Linguistic Typology. Syntax and Morphology. Oxford.
336 WOLFGANG RAIBLE
Coseriu, Eugenio. 1983. Sprachtypologie und Typologie von sprachlichen Verfahren. Allgemeine
Sprachwissenschaft, Sprachtypologie und Textlinguistik. Festschrift für Peter Hartmann, hrsg. von
Manfred Faust. 269-279. Tübingen.
Harris, Martin B. 1984. On the causes of word order change. Lingua 63.175-204.
Heger, Klaus. 1976. Monem, Wort, Satz und Text. Tübingen.
.1977. Modalität und Modus. Zeitschrift für romanische Philologie 93.1-14.
.1982. Nominativ - Subjekt - Thema. Fakten und Theorien. Beiträge zur romanischen und allgemeinen
Sprachwissenschaft (Festschrift Stimm), hrsg. von Sieglinde Heinz und Ulrich Wandruszka. 87-93.
Tübingen.
Herman, József. 1963. La formation du système roman de conjonctions de subordination. Berlin.
Hofmann, Johann Baptist - Szantyr, Anton. 1965. Lateinische Syntax und Stilistik. München.
Kiss, Sándor. 1982. Tendences évolutives de la syntaxe verbale en latin tardif. Debrecen.
Koch, Peter - Oesterreicher, Wulf. 1985. Sprache der Nähe - Sprache der Distanz. Mündlichkeit und
Schriftlichkeit im Spannungsfeld von Sprachtheorie und Sprachgeschichte. Romanistisches Jahrbuch
36.15-43.
Kölver, Ulrike. 1985. The Notion of Scalarity in Language Comparison. Language Invariants and Mental
Operations. International Interdisciplinary Conference Held at Gummersbach/Cologne, Germany,
Sept. 18-23, 1983. 53-57. Tübingen.
Koll, Hans-Georg. 1965. Zur Stellung des Verbs im spätantiken und frühmittelalterlichen Latein.
Mittellateinisches Jahrbuch 11.241-272.
Lehmann, Christian. 1984. Der Relativsatz. Typologie seiner Strukturen, Theorie seiner Funktionen,
Kompendium seiner Grammatik. Tübingen.
Lewin, Bruno. 1959. Abriß der japanischen Grammatik. Wiesbaden.
Lewis, G.L. 1965. Teach Yourself Turkish. London.
Ludwig, Ralph. 1986. Mündlichkeit und Schriftlichkeit. Felder der Forschung und Ansätze zu einer
Merkmalsystematik im Französischen. Romanistisches Jahrbuch 37.15-45.
.1988. Modalität und Modus im gesprochenen Französisch. Tübingen.
Lüdtke, Helmut. 1980. Kommunikationstheoretische Grundlagen des Sprachwandels. Berlin.
.1987. Überlegungen zur Methodologie der lateinisch-romanischen Sprachgeschichtsforschung. Text-
Etymologie. Untersuchungen zu Textkörper und Textinhalt. Festschrift für Heinrich Lausberg zum
75. Geburtstag, hrsg. von Arnold Arens. 382-391. Stuttgart.
Machtelt Bolkestein, Anna. 1976. The Relation Between Form and Meaning of Latin Subordinate Clauses
Governed by 'verba dicendi'. Mnemosyne.
.1976. A.C.I. - and ut-Clauses with verba dicendi in Latin. Glotta.263-291.
.1977. Part II: The Differences Between Free and Obligatory ut-Clauses. Glotta 55.231-250.
Mayen, G. 1889. De particulis quod, quia, quoniam, quomodo, ut pro c.  inf. post verba sent, et
declar. positis. Kiel.
Oesterreicher, Wulf. 1988. 'Konsistenz' als typologisches Kriterium. Romanistik, Sprach typologie und
Universalienforschung. Beiträge zum Freiburger Romanistentag, hrsg. von Wolfgang Raíble. 223-
262. Tübingen.
Perrochat, Paul. 1932. Recherches sur la valeur et l'emploi de l'infinitif subordonné en latin. Paris.
Raible, Wolfgang. 1980. Edmund Husserl, die Universalienforschung und die Regularität des Irregulären.
Wege zur Universalienforschung. SprachwissenschaftLiche Beiträge zum 60. Geburtstag von
Hansjakob Seiler, hrsg. von Gunter Brettschneider und Christian Lehmann. 42-50. Tübingen.
.1980. Regel und Ausnahme in der Sprache. Romanische Forschungen 92.199-222.
.1983. Knowing and Believing - and Syntax. On Believing. Epistomological and Semiotic
Approaches. De la croyance. Approches épistomologiques et sémiotiques, hrsg. von Herman Parret.
274-291. Berlin/New York.
.1985. Nominale Spezifikatoren ("Artikel") in der Tradition lateinischer Juristen oder Vom Nutzen
einer ganzheitlichen Textbetrachtung für die Sprachgeschichte. Romanistisches Jahrbuch 36.44-67.
.1987. Konzeptionelle Schriftlichkeit, Sprachwerk und Sprachgebilde. Zur Aktualität Karl Bühlers.
Romanistisches Jahrbuch 38. (Forthcoming.)
Ramat, Paolo. 1980. Zur Typologie des pompejanischen Lateins. Wege zur Universalienforschung.
Sprachwissenschaftliche Beiträge zum 60. Geburtstag von Hansjakob Seiler, hrsg. von Gunter
Brettschneider und Christian Lehmann. 187-191. Tübingen.
Richter, Elise. 1903. Zur Entwicklung der romanischen Wortstellung aus der lateinischen. Halle.
Rix, Helmut. 1976. Subjonctif et infinitif dans les complétives de l'Ombrien. Bulletin de la Société de
Linguistique de Paris 71.221-239.
Rohlfs, Gerhard. 2 1956. Sermo vulgaris latinus. Tübingen.
Schlicher, J.J. 1933. Non-Assertive Elements in the Language of the Roman Historians. Classical
Philology 28.289-300.
.1936. The Development of Cesar's Narrative Style. Classical Philology 31.212-224.
Schmitt Jensen, J0rgen. 1970. Subjonctif et hypotaxe en italien. Une esquisse de la syntaxe du subjonctif
dans les propositions subordonnées en italien contemporain. Odense.
Schrodt, Richard. 1983. System und Norm in der Diachronie des deutschen Konjunktivs. Der Modus in
althochdeutschen und mittelhochdeutschen Inhaltssätzen (Otfrid von Weissenburg - Konrad von
Würzburg). Tübingen.
Seiler, Hansjakob - Brettschneider, Gunter. Eds. 1985. Language Invariants and Mental Operations.
Tübingen.
.1986. Introduction. Seiler, Hansjakob. Apprehension. Language, Object, and Order, part III: The
Universal Dimension of Apprehension. 1-15. Tübingen.
Tesnière, Lucien. 1959. Eléments de syntaxe structurale. Paris.
Thom, René. 1973. Sur la typologie des langues naturelles: essai d'interprétation psycho-linguistique. The
Formai Analysis of Natural Languages. Proceedings of the First International Conference, hrsg. von
Maurice Gross, Morris Halle und Paul Schützenberger. 233-248. The Hague/Paris.
Werner, Otmar. 1987. Natürlichkeit und Nutzen morphologischer Irregularität. Beiträge zum 3. Essener
Colloquium über Sprachwandel und seine bestimmenden Faktoren vom 30.9. bis 2.10.1987 an der
Universität Essen. 289-316. Bochum.
ZWISCHEN SUBSTRATOMANIE UND SUBSTRATOPHOBIE
Grenzen eines sprachgeschichtlichen Erklärungsprinzips
am Beispiel des Lautwandels von lat. u zu frz. [ü].
RUDOLF WINDISCH
Der Keltologe Ernst Windisch hat in „Gröbers Grundriss der Romanischen Phililo-
gie" den „Einfluß des Gallischen auf das Romanische" untersucht. Er kolportiert hier
(Gr. Gr. I 2 , 1904/06:371-404) eine gegen Ende des 19. Jahrhunderts gängige - und wie
es scheint - längst anerkannte Lehrmeinung:
Es ist selbstverständlich, daß die lateinische Sprache im Munde der Römer und im Munde der
Eingeborenen je nach dem Lande einen dialektisch verschiedenen Charakter annehmen mußte (...).
Der Einfluß des Keltischen auf das Lateinische hat begonnen, sowie man auf keltischem
Sprachgebiet anfing lateinisch zu sprechen. Der Einfluß von Sprache auf Sprache äußert sich in der
Entlehnung von Wörtern, wichtiger aber ist der Einfluß, den der gallische Mund mit der ihm eigenen
Artikulation auf die fremde Sprache, die er sich aneignete, ausgeübt haben muß. Wenn wir eine
fremde Sprache lernen und sprechen, so sind wir geneigt, die Laute derselben nach unserer Art und
Weise auszusprechen und unsere Redensarten oder unseren Stil in der fremden Sprache wiederzugeben.
Ähnliches dürfen wir auch bei den Galliern voraussetzen.
Wir wollen überprüfen, wieweit diese Meinung auch für den Wandel von ü > ü gilt,
der innerhalb der Romania vor allem in Frankreich, in einem Teil des rätoromanischen
Gebiets sowie in den galloitalienischen Mundarten Oberitaliens eintritt.
Alle die genannten Gebiete unterlagen einst mehr oder weniger starkem keltischem
Einfluß. Der Wandel ü > ü wird daher dem keltischen Substrat zugeschrieben. Bleiben
wir der Einfachheit halber im zuerst genannten Bereich der Galloromania, wo die kel-
340 RUDOLF WINDISCH
tische Besiedlung, und damit der sprachliche Einfluß des Keltischen, am stärksten ge
wesen sein dürfte. Im Französischen spiegelt die Vertretung von lat. ü durch frz.[y],
Grapie u, z.B. lat. muru(m), duru(m) > frz. [myr], [dyr] usw. ganz offensichtlich eine
Entwicklung wider, die im Keltischen, beispielsweise im Kymrischen, über die im Fran-
zösichen erreichte Stufe hinaus bereits früh den Endpunkt dieses Palatalisierungs-
vorganges erreicht hat, nämlich i, z.B. kymr. din < dunum 'Burg'; die Stadt 'Lyon' <
gallisch Lug(u) = dūnum oder kymr. dir 'Stahl' < (ferrum) durum; auf diese Parallele
hatte zuerst Hugo Schuchardt, dann aber Isaia Graziadio Ascoli in einem seiner
sprachwissenschaftlichen Briefe, 1881, hingewiesen.
Es war vor allem Ascoli, der die These von der 'ethnischen Einwirkung' des Kel
tischen auf den Wandel ū > ü durch folgende drei Punkte zu untermauern suchte:
1.wegen der räumlichen Übereinstimmung der Bereiche, die den Wandel ü > ü mit dem
früheren keltischen Siedlungsgebiet zeigen;
2. wegen der sog. 'inneren Kongruenz', d.h. i < ü (über ü) in den heutigen keltischen
Dialekten;
3. wegen der 'äußeren Kongruenz', d.h. ü < ū auch im nichtromanisch/ ehemals
keltischen Gebiet, z.B. in den Niederlanden.
Lassen wir - als Kontrast zu Ascoli - zunächst die Gegner der Substratthese zu Wort
kommen. Besonders energisch spricht sich W. Meyer-Lübke in mehreren Aufsätzen zwi
schen 1919 - 1927 gegen keltischen Einfluß aus. Im wesentlichen führt er folgende 4
Argumente an:
1. Es gibt (romanische) Dialekte, in denen ü erscheint, ohne daß keltisches Substrat
vorliegt, so vor allem im Vegliotischen und auch im Albanischen;
2. bewahrt die Emilia, mit der einst keltischen Hauptstadt Bononia/Bologna fast
durchweg u;
3. Fehlen von ü im Katalanischen, einem Gebiet, das zweifellos gallisch beeinflußt
war;
4. Fehlen der Palatalisierung von  in frz. Wörtern wie cul < culus, cuve <
cupa, cure < cura, cuivre < cuprum, bei denen wie im Falle von lat. c e, i >
altfrz. [ts] hätte eintreten müssen, wenn das ū tatsächlich schon [y] gewesen
wäre.
ZWISCHEN SUBSTRATOMANIE UND SUBSTRATOPHOBIE 341
Aber die Einwände des großen Meisters sind nicht in jedem Falle so stichhaltig, wie
es zunächst erscheinen mag. Zum letzten Punkt, Nichtpalatalisierung von  vor [y], wäre
zu sagen, daß dem Nicht-Eintreten der Palatalisierung keine so große Bedeutung beizu
messen ist: das sich aus  entwickelnde ü braucht nicht unbedingt bereits denselben
palatalen Wert wie e,i gehabt zu haben, um.eben schon an dieser altfrz. Palatalisierung
von  vor e,i z.B. cera > [tsire] teilzunehmen.
Meyer-Lübkes Befund scheint - was einen 2. Einwand betrifft - für den Bereich von
Bologna im Prinzip zuzutreffen. Es scheint nun müßig, Gründe für diese Ausnahme zu
suchen. Wir haben diesen Tatbestand einfach zu registrieren. Wäre damit aber die ganze
Keltenthese schon hinfällig? Dieser Ansicht kann auch Meyer-Lübke nicht gewesen sein:
Es hieße sonst von einem sprachlichen Erklärungs-Prinzip - hier dem Wirken des Sub
strats - zu viel zu verlangen, wenn es in allen Fällen, in denen die einmal formulierten
Bedingungen gegeben sind, tatsächlich auch eintreten müßte.
Walther v. Wartburg, der gewichtigste Apologet von Ascolis Keltenthese, geht im
wesentlichen auf die beiden Einwände Meyer-Lübkes ein, wonach 1.) sich „das ü
vielerorts nur unvollständig durchgesetzt (habe)" und 2.) daß „die geographische Über
einstimmung mit dem alten Keltengebiet (fehle)". Zumindest wegen des letzten Argu
ments Meyer-Lübkes fühlt sich v. Wartburg zu einer minuziösen Auflistung aller
Schweizer und Oberitalienischen Dörfer und Landschaften verpflichtet, für die sich - als
ū-Abweichung im ü-Gebiet nach Kritik von Meyer-Lübke - doch noch ein ü finden oder
historisch rechtfertigen läßt: anhand des AIS gelingt es v. Wartburg, ein vollständiges
Bild der ü-Verbreitung zu geben. Sie ist dichter, als es Meyer-Lübke seinerzeit wissen
konnte. Ist damit aber etwas für oder gegen die Keltenthese gewonnen? Sicherlich eine
Stärkung der von Ascoli behaupteten räumlichen Übereinstimmung. Oder anders
ausgedrückt: Es ist einfacher, die sprachlichen Fakten, d.h. die heutige ü-Verteilung mit
dem keltischen Siedlungsgebiet in Deckung zu bringen, als umgekehrt aus der ū/ü-
Verteilung sichere Rückschlüsse auf die historische Ausbreitung und Dichte jene
keltischen Gebietes zu ziehen.
Andere Romanisten, z.T. auch Vertreter der Allgemeinen Sprachwissenschaft aus
beiden Lagern reklamieren mit nicht weniger starker Entschiedenheit Beachtung und Zu
stimmung für ihren Standpunkt - allerdings ohne jene Faktenkenntnis- und Diskussions-
342 RUDOLF WINDISCH
breite vorzuweisen. So sagt z.B. Hans Rheinfelder - im übrigen ein Meister der histori
schen französischen Grammatik - zum Lautwandel ū/ü:
Die frühere Meinung, die Kelten hätten das ihnen fremde lat.ūdurch ü ersetzt (Lautsubstition), ist
unbewiesen. Die gelegentlich immer wieder versuchte Erklärung des ü aus keltischem Substrat (...)
läßt sich durch nichts rechtfertigen.
Mit einem solchen apodiktischen Urteil ist uns hier wenig geholfen - das Warum der
Verurteilung der Keltenthese hätte uns interessiert. Aber auch andere bedeutende
Romanisten, wie etwa Gerhard Rohls, haben unserer Ansicht nach ihr Urteil über die
Substratthese nicht sehr viel besser untermauert: so mag sich Rohlfs zwar wundern, daß
die 'keltische Theorie' immer noch ihre 'hartnäckigen Verteidiger' findet - gemeint ist v.
Wartburg - aber Rohlfs' an dieser Stelle (1957:495 ff.) vorgebrachten Einwände, auf die
wir hier nicht eingehen wollen, können die bereits angemeldeten Bedenken gegen die
Substratthese nicht ernstlich erschüttern.
Welche Seite hat nun Recht? Gehen wir dieser Frage noch ein wenig nach! Listet man
die konträren Argumente auf - wir konnten das Für oder Wider nur summarisch auf
zählen -, so gewinnt man den Eindruck, als wäre unser Problem sowohl vom Faktischen
als auch von der Methode her erschöpfend behandelt und die Bewertung, ob die Sub
stratthese richtig ist oder nicht, letztlich nur noch der subjektiven Einstellung des ein
zelnen Forschers unterworfen - in der Tat eine nicht annehmbare Kontrollinstanz. So hat
nun diese Frage - sie wurde in der letzten Zeit ohnehin kaum noch aufgegriffen - ein Sta
dium erreicht, das einem sich-im-Kreise-drehen ähnelt: der Ausgang der Diskussion hat
sich - nach dem Eingehen auf die jeweiligen Gegenargumente - im Sande verlaufen,
ganz so, als ob eine Lösung unmöglich wäre.
Einen solchen Lösungsversuch wollen wir hier auch gar nicht wagen, höchstens eine
Schritt hin in diese Richtung. Dieser Schritt beruht auf dem Vorwurf einer erstaunlich
ungenügenden Berücksichtigung bestimmter sprachlicher Fakten, die bisher kaum in
unsere Frage einbezogen waren: 'erstaunlich' erscheint uns dieser Mangel deshalb, weil
es schließlich eminente Vertreter einer Richtung - für ihre Sorgfalt im Umgang mit den
sprachlichen Fakten berühmt - waren, die sich ausgiebig mit unserem Problem befaßt
hatten. Wie läßt sich die leichtfertige Behauptung belegen, Positivisten bei der Vernach
lässigung von Fakten ertappt zu haben? Wir glauben, durch den schlichten Hinweis auf
die ungenügende Deutung der lautlichen Verhältnisse des festlandkeltischen Gallischen
im Bereich des palatalen ü. So wurde nicht einmal geklärt, ob das Gallische überhaupt ein
ü kannte. Unter diesem Gesichtspunkt bedarf eine erneute Behandlung der keltischen
Substratthese wohl keiner besonderen Rechtfertigung mehr. Einen expliziten Hinweis auf
diese Vernachlässigung der Fakten und der sich daraus ergebenenden explikativen Fehl
leistungen findet man - soweit wir die bisherigen Beiträge zu diesem Thema überblicken
- nur bei Tagliavini:
Zwar weiß man nicht, ob das Gallische je ein Phonem ü besaß, aber die Tendenz des mittelalterlichen
und modernen Keltisch,  in i zu verwandeln (z.B. air. dún 'oppidum', aber kymr. din; ebenso in den
Entlehnungen aus dem Lateinischen: lat. cupa > kymr. cib 'Tasse'; usw.) setzt eine Zwischenstufe ü
voraus.
Vor dem Hintergrund einer so unsicheren Kenntnis der für unsere Diskussion not
wendigen Fakten mutet die Heftigkeit des Für und Wider, die scheinbare Stringenz der
gezogenen Schlüsse und die Sicherheit ausstrahlende Verurteilung gegensätzlicher Auf
fassung schon überraschend an. Halten wir fest: erstens scheint nicht völlig sicher, ob
das Keltische den Wandel von ü zu i, dürus > dir überhaupt über die Stufe ü hat laufen
lassen - dieser Stufe kommt wohl eher aus Gründen lautphysiologischer 'Natürlichkeit'
auf dem Weg der Palatalisierung von ü zu i Wirklichkeit zu; zweitens: kommt der Wandel
ū > i lediglich im inselkeltischen Britannischen, also im Kymrischen, Bretonischen und
Cornischen vor? Weiter, gibt es keinen Hinweis auf ein ü im Gallischen selbst, wie man
doch zunächst erwarten sollte? Die Tatsache, daß sich die Verfechter der Substratthese
hier schwer taten, entbindet die Gegner allerdings nicht, ihrerseits zur Aufklärung des
Sachverhaltes beizutragen. Wenn nun die historischen Grammatiken des Keltischen auch
für das festländische Gallische überhaupt kein ü belegen, wäre damit die These vom kel
tischen Einfluß nicht unwiderruflich hinfällig? Bevor wir uns dieser scheinbar zwingen
den Folgerung anschließen, untersuchen wir zugunsten der Substratthese doch noch ge
nauer, wo bzw. in welchen Fällen im Keltischen - immer vorausgesetzt, der Wandel von
ü > i ist über ü gelaufen - überhaupt ein Einfluß auf die Galloromania hatte ausgehen
können!
Wie wir im folgenden noch sehen werden, liegt die genannte keltischen Entwicklung
ü zu i zeitlich eindeutig früher, als der Beginn des Wandels  zu ü im Französischen.
Zunächst einmal: die Entwicklung im Keltischen selbst ist komplizierter, als sie in den
romanistischen Beiträgen - wenn überhaupt - angedeutet wird. Diesen Sachverhalt ver
deutlicht das folgende Schema mit seinen 3 unterschiedlichen Ausgangspunkten und
344 RUDOLF WINDISCH
seinen beiden zeitlich verschobenen Entwicklungsstufen: Stufe 1) das gemeinbritannische

Ergebnis, Stufe 2) die keltischen Einzelergebnisse:
Aus der genaueren Erklärung dieses Schemas ergeben sich folgende Punkte: in der
romanistischen Diskussion werden gewöhnlich die bereits erwähnten lateinischen Lehn
wörter des Keltischen aufgezählt, die die Entwicklung der I. Reihe mitmachen: lat. durus
> kymr. dir, cupa > cib, cupella > cibell, muscu(u)lus > misgl, also die Elemente, in
denen lat. ü im Keltischen durch i vertreten ist. Nun sind es aber in der Tat nur diese
wenigen lateinischen Wörter mit langem ū, die in diese Entwicklungsreihe gehören.
Das gros der lateinischen Lehnwörter - etwa zwischen 500-600 - wurde zwischen
dem 1.-4. Jahrhundert übernommen. Die Mehrzahl jener mit langem ü wurde aber nicht -
wie durus, cupa oder musc'lus - mit den auf idg. ü zurückgehenden keltischen Wörtern
identifiziert, sondern mit dem gemeinbrit.  aus idg. oi (vgl. Reihe II). Dieses aus idg. oi
entstandene und mit lat. ū aufgefüllte brit.  fiel dann seinerseits mit einem brit. 
zusammen, das über gemeinbrit.  etwa zwischen dem 1. bis Ende 3. Jahrhundert aus
den idg. Diphthongen au, eu, ou entstanden war (Reihe  ) . Dieses aus 3 verschiedenen
Quellen entstammende gemeinbritannische u, also aus idg. oi und lat. ü (vgl. Reihe II)
und aus idg. au, eu, ou (vgl. Reihe III) wurde schließlich, ebenso wie idg. ü (vgl.
Reihe I) über *uzu i (vgl. Stufe 1 und 2 des Schemas).
Wie man in Reihe  leicht erkennt, hat sich bei der Entwicklung von idg. au, eu, ou
auf dem Weg zu brit.  auf der Vorstufe  noch die Möglichkeit zur Aufnahme von lat. õ
ergeben, allerdings nur õ im Wortinnern: vlat. cõsilium > mittelkymr. cusyl, altcorn.
cusul, mod. bret. kuzul. Die Entwicklung gemeinbritannisch ü aus idg. oi, sowie aus au,
eu, ou und lat. ü und õ, also Reihe II + III, die dann zunächst zu *ü führte, konnte erst
eingetreten sein, als die Entwicklung von idg. ü > i (vgl. Reihe I) mit den wenigen frühen
lat. Entlehnungen vom Typ dürus > dir bereits abgeschlossen war. Eigentlich haben die
Romanisten, soweit sie überhaupt keltische Beispiele zitierten, nur diese letztere, d.h.
ältere Entwicklung idg. ü über ü zu i (Reihe I) im Auge gehabt, da diese Entwicklung in
der Anpassung der lat. Beispiele wohl am leichtesten durchschaubar war. Als Beleg für
Substratwirkung im Romanischen weist sie dennoch - aus bereits angedeuteten Gründen
- mindestens 2 gravierende Mängel auf: 1.) ist die Entwicklungsstufe ü < idg.  bereits
erreicht, bevor das gros der lat. Lehnwörter ins Inselkeltische übernommen wurde; 2.) ist
die Stufe ü nicht belegt, sie kann nur erschlossen werden, allerdings nicht nur aufgrund
lautphysiologischer Evidenz. In der romanistischen Diskussion werden diese drei
unterschiedlichen Entwicklungsstränge, die zwar zum selben Ergebnis führen, aber
chronologische Differenzierungen aufweisen, nicht auseinandergehalten.
Wir haben nun für diese 3 Entwicklungsstränge die Stufe ü genannt. Es ist ein er
schlossenes ü. Den Nachweis für diese Lautung können wir hier nicht erbringen. Wir
stützen uns auf die historischen Grammatiken des Keltischen, die die Gewähr für die
sprachgeschichtliche Existenz bzw. lautgesetzlich korrekte Rekonstruktion dieses Lautes
bieten. Wir können uns hier nur auf deren Untersuchungen und Ergebnisse berufen.
Auch in den ältesten keltischen Belegen läßt die Graphie - was uns nicht überrascht - kei
ne direkten Aufschlüsse über bestimmte Lautungen zu: wir hätten es sonst auch kaum mit
dem vor uns liegenden Problem zu tun - zumindest nicht in diesem Umfang. Auch die
heutige Graphie weist keine eindeutig phonematische Schreibung auf: einerseits haben
wir für die lat. Beispiele, die die Entwicklung von idg.  mitgemacht haben, heute durch-
346 RUDOLF WINDISCH
gehend i, also kymr. kib, dir usw., andererseits kymr. Llun (< L ūnae); kymr. bret. mur
(< mūrus) usw., wo die alte Graphie  freilich immer noch für das längst palatalisierte
Endergebnis í steht; ähnlich ist die Situation im älteren Französisch, wo die Graphie
zunächst nur unzureichender Reflex der lautlichen Entwicklung dieses Palatilisie-
rungsvorgangs ist.
Bleibt noch die entscheidende Frage nach der relativen und absoluten Chronologie
des Wandels: idg. ü (vgl. Reihe I des Schemas) erreicht seine Endphase i längst bevor die
übrigen beiden Reihen, etwa ab dem 6. Jahrhundert, das gemeinsame  in palatale Rich
tung überhaupt erst nach ü zu verschieben begannen. Oder anders ausgedrückt: die Ende
des 5./6. Jahrhunderts in die Bretagne ausgewichenen Bretonen konnten nur ein i, näm
lich das aus idg. ū mitgebracht haben, aber kein palatales ü, da dieses eben schon zu i ge
worden war, und das 2., neue, aus gemeinbritannisch  entwickelte ü sich wohl erst -
wie gesagt - ab Mitte des 6. Jahrhunderts, d.h. nach Abwanderung der Bretonen,
allmählich in die palatale Richtung verschoben hatte. Wir können nicht umhin, das
Problematische der auf einer so scharfen zeitlichen Abgrenzung beruhenden Folgerung
selbst herauszustellen.
Es wäre jedoch falsch, die so gezogenen Zeitgrenzen durch fließende Übergänge
nach vorwärts oder - je nach Bedarf - nach rückwärts gleichsam zu 'entschärfen', um den
Bretonen den Anfang der Palatalisierung von gemeinbrit.  > ü noch auf den Weg zum
Festland mitgeben zu können.
Ob nun der französische Wandel  > ü schon im 7. oder erst im 8. Jahrhundert
eintrat, wie die Mehrzahl der Gegner der Keltenthese einwendet, spielt dann keine Rolle
mehr: der Anschluß an das noch lebende Gallische ist nicht zu gewinnen - zwischen des
sen Aussterben Ende des 4. Jahrhunderts und dem Eintreten des Wandels liegen minde
stens 3 Jahrhunderte. Beharrt man aber auf gallischem Einfluß, muß man wieder das
recht geheimnisvolle Weiterleben, eine Kontinuität der Substratwirkung einführen, wofür
uns gerade die Beweise fehlen. Wie auch immer - das Bild von dem unter dem
zugefrorenen Eis des Baches weiterfließenden Wasser - hier die weiterlebende keltische
Stubstratwirkung unter der galloromanischen Sprachschicht - hat nicht nur für den auf
Fakten beharrenden Positivisten etwas Geheimnisvolles, Mysteriöses an sich.
Unserer Ansicht nach weicht W. v. Wartburg der Diskussion dieses Punktes aus,
wenn er das von ihm selbst angesprochene Poblem der zeitlichen Lücke zwischen dem
Aussterben des Gallischen und dem auch von ihm für das 7. Jahrhundert noch nicht
nachweisbaren Vorkommen des Wandels  > ü auf die an sich wichtige Unterscheidung
zwischen dem Eintreten des Wandels und dessen Abschluß verschiebt.
Es bleiben (...) vielleicht 3-4 Jahrhunderte zwischen dem Beginn des Lautwandels im Gallischen und
seinem Abschluß im Galloromanischen, wie ihn Gamillscheg ansetzte, und in dieser Zeit mag sich 
allmählich und wohl mit langem Verweilen auf den Zwischenstufen (...) zu ü bewegt haben.
Das hieße nämlich, um diese Lücke zu schUeßen, hätte das Galloromanische des 7./8.
Jahrhunderts den Anstoß zur Palatalisierung noch von dem im 3./4. Jahrhundert ster
benden Gallischen erhalten und diese Tendenz hinübergerettet: eine Möglichkeit der Er
klärung, der wiederum das Odium des Mysthischen anhaftet - ein Weiterleben nach dem
Tode. Dagegen sprechen unter Umständen folgende Fakten: einerseits haben wir im 3./4.
Jahrhundert eben schon kein kelt.-gall. ü mehr, sondern bereits ein i < idg. u, aber auch
kein neues ü, da sich dieses - wie unser Schema zeigte - erst im 5./6. Jahrhundert ent
wickelte, als das festlandkelt. Gallische bereits längst ausgestorben war und die von den
britischen Inseln eingewanderten Bretonen es offensichtlich noch nicht mitgebracht haben
konnten.
Woher also das ü in der Galloromania? Ist es das Resultat einer vorgegebenen
Tendenz zur Palatalisierung? Oder beruht es vielleicht doch auf bretonischem Einfluß, da
dieser Dialekt nun tatsächlich mit seinen früheren inselkelt. Nachbarn, dem Kymrischen
und dem Cornischen, dasselbe palatale Ergebnis i aus ü aufweist?
Drei Antworten waren möglich, erstens, die einfachste: das Bretonische hatte mit dem
aus idg. ü zu gemeinbrit. Zeit bereits entwickelten i das analoge Vorbild; zweitens,
warum sollte das Bret, nicht spontan, disponiert zur Palatalisierung, aus  ein ü und dann
i entwickeln, wenn beispielsweise auch das Albanische das ū der lat. Elemente alle zu ü,
z.T. sogar zu i entwickelt hat (lat. palüdem > alb. pül usw.); drittens, die spekulative,
mystische: die im Keltischen angelegte Tendenz zur Palatalisierung wirkt sich in jedem
Fall dort aus, wo sie im Keim schon angelegt war! Vielleicht rückt diese uns verborgene
Substratwirkung eines Tages das frz. ü näher an das i, um den keltischen Prozess auch
hier zum Abschluß zu bringen? Anzeichen dafür wird man im heutigen Französisch aber
(noch!) nicht ernsthaft suchen wollen!
348 RUDOLF WINDISCH
Gegen die zuvor angedeutete späte bretonische Beeinflussung auf dem Festland, also
gegen eine räumlich bedingte Adstratwirkung, sprechen ihrerseits folgende Fakten: im
Französischen erscheint ü nicht vor Mitte des 8. Jahrhunderts; im Norden, Normandie,
im Osten (östl. Lothringen, Burgund, in Wallonien z.T. heute noch nicht), spät, erst zwi
schen dem 10.-13. Jahrhundert; in Nordfrankreich, dem möglichen Kontaktgebiet zum
Keltischen also mit Sicherheit später also im Zentrum! Wie will man diese zeitlichen und
räumlichen Divergenzen in Übereinklang bringen? Schwerlich - aber trotzdem können
wir uns noch nicht zu einer Aufgabe der Substratthese in unserer Frage entschließen: die
Indizien für ihre Wirksamkeit - d.h. räumliche Übereinstimmung des Wandels mit kelt.
Siedlungsgebiet, innere Übereinstimmung des Wandels mit dem Keltischen selbst -
scheinen uns doch zu auffällig, als daß sie purem Zufall zugeschrieben werden könnten;
andere lautliche Erscheinungen, wie z.B. die Reduzierung der ursprünglichen lat. Dop
pelkonsonanz -pp-, -tt-, -kk- wie in cǔppa, gutta, bǔcea zu einfachem -p-, -t-, -k- sowie
die Assimilation des Konsonantennexus lat. -ct-, factum > frz. fait, lassen die Sub
stratwirkung in einem weiteren Zusammenhang, neben den nicht umstrittenen lexika
lischen Entlehnungen des Galloromanischen aus dem Keltischen, deutlicher hervortreten
und ihr scheinbar so mysteriöses Wirken empirisch doch greifbarer werden: der Ausgang
für den Lautwandel ü > ü müßte so doch eine recht frühen Disposition des keltisch be
einflußten Lateins der Galloromania zur Palatalisierung zugeschrieben werden. Damit wä
ren wir aber wieder am Anfang unserer kritischen Sichtung der Schwachstellen der
Keltenthese angelangt, bei der alten Wartburg-These.
Andere Erklärungen, frühstrukturalistische, bieten sich an: bloß krankt die von
Haudricourt und Juilland, Schüler von Martinet, vertretene systemimmanente, phonolo-
gische These von der Überlastung der hinteren, velaren Vokalreihe an dem Faktum, daß
das Französische nichts Eiligeres zu tun hatte, als spätestens zum 13. Jahrhundert die
durch die Palatalisierung von lat.  erreichte Räumung der alten w-Steile schnellstens
wieder durch ein neues  aufzufüllen. Bei aller explikativen Stringenz, die wir solchen
phonologischen Erklärungen zugestehen, klammern sie notwendigerweise - von ihrem
Ansatz her allerdings programmatisch gerechtfertigt - das von uns diskutierte sprachge
schichtliche Problem völlig aus. Sie sind ebenfalls einseitig und können nur einen
Teilaspekt des Problems erörtern.
Kommen wir zum Schluß. Eine moderat konzipierte Substratwirkung wird man nicht
grundsätzlich in Frage stellen: die "Ausgliederung" der romanischen Sprachen, ihre Ver
schiedenheit ist - neben anderen wichtigen Faktoren wie beispielsweise die Chronologie
der Romanisierung, die diatopischen und diastratischen Verschiedenheiten usw. - zu
einem großen Teil eben nur durch die Wirkung der verschiedenen vorrömischen
Sprachen zu erklären. Darüber kann es - nicht nur bei den Romanisten - keinen Zweifel
geben. Neben dem eigentlichen Nachweis des ursächlichen Zusammenhangs, eben der
Substratwirkung auf die weiterlebende Sprache, bereitet vor allem die Frage größte
Schwierigkeiten, wie die zeitliche Kluft überbrückt werden kann, die sich zwischen dem
z.T. schon frühen Aussterben der Substratsprachen und den oft erst spät in den einzelnen
romanischen Sprachen beobachtbaren substratbedingten Erscheinungen auftut.
Wir müssen in Anbetracht solcher zeitlicher Überlieferungslücken aber nicht not
wendigerweise jedes Mal von einer geheimnisvollen, dunklen Wirksamkeit des längst
nicht mehr lebendigen und daher oft kaum noch faßbaren Substrats ausgehen. Der einmal
durch das Substrat eingeleitete Wandel vermag sich auch ohne Notierung in den ältesten
Quellen durchzusetzen. Man wird also - sind gerade die älteren Etappen eines auf Sub
stratwirkung zurückgeführten Wandels nicht belegbar - heute nicht ernstlich mehr etwa zu
einer anatomischen Erbanlage der Sprechorgane Zuflucht nehmen wollen, die die in der
Substratsprache einmal geprägten Bewegungen zur Rettung der alten Aussprachege
wohnheiten weitervermittelt oder zu einer Art latenter - erst nach längerer Unterbrechung
wieder unerwartet hervortretenden Aussprache oder ähnlichen - in der Tat - mysthischen
Vorstellungen über das Wirken des Substrats. Andererseits erscheint auch die unaus
gesprochene Forderung, eine vermittelnde Haltung zwischen Substratomanie und Sub-
stratophobie, also eine Art "goldene Mitte" einzunehmen, zwar als selbstverständlich,
aber doch als zu simpel. Davon abgesehen, hat uns Rohlfs, dem wir die Verwendung
dieses antonymen Paares zur Titelformulierung verdanken sowie die Beherzigung zur
Einhaltung des rechten Weges, in der Keltenfrage eher eine allzu kritische, vielleicht
sogar voreingenommene, d.h. negative Beurteilung der Erklärungsmöglichkeiten durch
die Substratthese geliefert. Wir zweifeln, ob die Einhaltung der Mitte überhaupt eine
Rolle spielt: gefragt ist eine Entscheidung für oder gegen die Substratthese - erst die Ent
scheidung zugunsten der Substratthese sollte dann bei der Demonstration ihrer Anwen
dungsmöglichkeiten von jener Mäßigung geprägt sein.
350 RUDOLF WINDISCH
Sollten also die Zweifel an einer keltischen Substratwirkung in dem von uns dis
kutierten Fall überwiegen, wäre die Substraterklärung in den Handbüchern zu streichen,
zumindest aber ihr Erklärungsdefizit deutlich hervorzukehren. Dazu bedürfte es aber alle
mal eines gewissen Maßes an Respektlosigkeit gegenüber den großen Meistern wie
Ascoli oder v. Wartburg, die diese These aufgestellt und mit mehr oder weniger plau
siblen Argumenten verteidigt haben.
LITERATURVERZEICHNIS
Alessio, G. 1949. Le origini del francese. Firenze.

Bolelli, T. 1940. Contributo allo studio dell'elemento celtico nella fonetica romanza. Arch. Rom,
24.188 ff.
Bolelli, T. 1943. L'u latino e i prestiti latini nelle lingue britanniche. L'Italia Dialettale 19.31-49.
Brauße, M. 1981. Zur Vorgeschichte der historischen Sprachwissenschaft. Die Auseinandersetzungen im
Frankreich des 18. Jahrhunderts über den keltischen Anteil an der französischen Sprachentwicklung.
BRPh 20.79-94.
Dauzat, A./G. Deslaudes/C. Rostaing. 1978. Dictionnaire étymologique des noms de rivières et de
montagnes en France. Paris.
Duval, P.-M. 1969/70. Chronique gallo-romaine. REA 71/3-4.402-439.
Duval, P.-M. 1971. La Gaule jusqu'au milieu du V e siècle. Paris.
Evans, D. Ellis. 1977. The Contribution of (non - Celtiberian) Continental Celtic to the Reconstruction
of the Celtic "Grundsprache". Indogermanisch und Keltisch. Koll. der Idg. Gesellschaft, 1976, Bonn.
Hrsg. R. Ködderitzsch/K. H. Schmidt. 66-88. Wiesbaden.
Falc'hun, F. 1950-51. Histoire de la langue bretonne d'après la géographie linguistique. Rennes.
. 1963. Histoire de la langue bretonne. Paris.
. 1981. Perspective nouvelles sur l'histoire de la langue bretonne.
Gamillscheg, E. 1919. Beiträge zur frz. Lautgeschichte. I. Zur u-ü-Frage. ZFSL 45.341-349.
Gebhardt, K. 1968. A propos du changement ū > ü. Mise au point des principales hypothèses. BJR
15.44-51.
Goidánich, P. G. 1907. L'Origine e le Forme della Dittongazione Romanza. Halle.
Gröber, G. 1886. Vulgärlateinische Substrate romanischer Wörter. ALLGr. 3.507 ff.
Gurridge, M. 1966. Romance linguistic influence on Middle Welsh. A review of some problems. Studia
Celtica 1.63-92.
Haarmann, H. 1973. Der lateinische Lehn Wortschatz im Bretanischen. Hamburg.
Hatt, J.-J. 1970. Kelten und Galloromanen. Genf.
Jackson, K. 1953. Language and History in Early Britain. A Chronological Survey of the Brittonic
Languages First to Twelfth Century A. D. Edinburgh.
Jacoby, E. 1916. Zur Geschichte des Wandels von lat.  zu [] im Galloromanischen. Berlin.
Lewy, E. 1942. Der Bau der Europäischen Sprachen. Proceedings of the Royal Irish Academy
XLVIII,C,2.
Merlo, . 1940. La Francia linguistica odierna e la Gallia di Giulio Cesare. Atti R. Acad. d'Italia, Serie
7, Vol. 2.
Meyer-Lübke, W. 1913/1917/1919/1927. Zur Frage u > ü. ZFSL 41.1-7, 44.75-84, 45.350-357, 49.272-
289.
Meyer-Liibke, W. 1907/1910. Lit.blatt f. G. Roman. Phil. 28.330-334, 31.280-286.
Montis, J. J. 1913. A Welsh Grammar. Historical and Comparative. Phonology and Accidence. London.
(Reprinted 1955.)
Pilch, H. 1958. Das kymrische Lautsystem. KZ 75.24-57.
. 1975. Advanced Welsh Phonemics. Zeitschrift für Celtische Philologie 34.60-102.
Pokorny, J. 1953. Keltologie (Wissenschaftliche Forschungsberichte, Geisteswissensch. Reihe. Bd. 2: V.
Pisani/J. Pokorny, Allgem. und Vergi. Sprachwissensch. Indogermanistik/Keltologie. S. 95 ff.)
Bern.
Rivet, A. L. F./C. Smith, 1979. The place-names of Britain. London.
Rohlfs, G. 1957. Zur Methodologie der romanischen Substratforschung (Substratomanie und Sub-
stratophobie). Syntactica und Stilistica, Festschrift E. Gamillscheg. 495-509. Tübingen.
Schmidt, K. H. 1977. Die festlandkeltischen Sprachen. Innsbruck.
Schuchardt, H. 1880 Rez. E. Windisch, Kurzgefaßte Irische Grammatik, 1879. ZRPh 4.124-155.
Schürr, F. 1954. Zum Wandel u > ü im Französischen. Estudios Menéndez Pidal. V:133-140. Madrid.
Sofer, J. 1941-43. Mehrere Berichte. ZCP.
Ternes, E. 1978. Zur inneren Gliederung der keltischen Sprachen. KZ 92.195-217.
Terracini, B.A. 1937. Sostrato. Scritti in onore di A. Trombetti. 321-364. Milano.
Thurneysen, R. 1884. Keltoromanisches. Die keltischen Etymologien im Etymologischen Wörterbuch
der Romanischen Sprache von F. Diez. Halle.
Tovar, A. 1961. The Ancient Languages of Spain and Portugal. New York.
Vendryes, J. 1902. De Hibernicis Vocabulis quae a Latina Lingua Originem duxerunt. Lutetia
Parisiorum.
. 1925. Celtique et Roman. Revue de Ling. Romane 1.262-277.
Weisgerber, L. 1931. Die Sprache der Festlandkelten. Bericht der Römisch-Germanischen Kommission
20.147-226. Mainz.
Whatmough, J. 1970. EA1: Being Prolegomena to A Study of the Dialects of Ancient Gaul. The
Dialects of Ancient Gaul. Prolegomena and Records of the Dialects. 1-76. Cambridge, Mass.
Windisch, E. 1904-1906. Keltische Sprache. Gröbers Grundriss der romanischen Philologie. 2. Aufl.
1:371-404. Straßburg.
SUMMARY
It goes without saying that direct contact between two or more languages can lead to interactive
influence and change. What ist difficult to prove is how this idea is tenable on the basis of specific
examples. How can one attribute the sound shift from latin  > french [ü], lat. MURUM > f. mur, to a
Celtic-Gaulish substratum, when the surviving Celtic dialects - with the exception of Irish - usually
352 RUDOLF WINDISCH
demonstracte i rather than [ü], as in Welsh dir < lat. DURUS, cib < CUPPA, etc? Is this not a case of a
natural physiological process taking place in the two languages independently of each other, at the same
time? Is this not simply a case of palatalisation of  through [ü] to i? Can we not say that the
surprisingly inadequate collecting of facts available to Romance scolars of the positivist school would
rule out a Celtic influence? How can we maintain that continental Gaulish, which died out at the latest in
the fourth or fifth centuries caused the shift from L.  to F. [ü], if we only find evidence in the ninth or
tenth centuries, in Gallo-roman times?
The logical conclusion would seem to be that we must reject the "substratum" principle. Perhaps we
should call this phenomenon "Substratophobia"? This 'conclusion' would, of course, be too simplistic;
the fact that this sound shift can be found allover the areas settled by the Celts, and the internal
congruency of this shift in the two respective areas still supports I.G. Ascoli's ethnic-linguistic
substratum theory. In this case, at least, it would be unreasonable to talk of a "fixation with substratum",
substratomania.
VII.
ROMAN HISTORY
ÜBER DAS RÖMISCHE KÖNIGTUM
ENDRE FERENCZY
Universität Budapest
Die bahnbrechende wissenschaftliche Tätigkeit des hochgeehrten Jubilars hat auf die
verwandten Disziplinen, so auch auf die Alte Geschichte, befruchtend gewirkt. Mein
Beitrag möchte die Verehrung seitens des ehemahgen Kollegen und die innige Zuneigung
des Jugendfreundes zum Ausdruck bringen.
***
Das römische Königtum ist eine der meistdiskutierten Epochen der römischen
Geschichte, die trotz der umfangreichen Literatur über sie noch immer als weitgehend
unaufgeklärt gilt.
Die Gründe für die Schwierigkeiten einer Rekonstruktion dieses Zeitalters können
einerseits durch das völlige Fehlen der zeitgenössischen bzw. primären Quellen, anderer
seits durch die späte Entstehung der römischen Geschichtsschreibung erklärt werden,
deren Authentizität schon von den Römern selbst bzweifelt wurde.1 Obwohl schon früher
Bedenken gegen die Glaubwürdigkeit der literarischen Quellen erhoben wurden, so
gegen die Werke von T. Livius und Dionysius Halicarnassensis (die sog. Tradition), die
aus dem augusteischen Zeitalter stammen, blieben sie die Hauptquellen des römischen
Königtums bis zum 19. Jahrhundert. Seit dem Beginn des 19. Jahrhunderts trat die von
B. G. Niebuhr begründete, sog. philologisch-kritische Richtung der Althistoriker auf,
der sich ausgezeichnete Historiker wie A. Schwegler, B. Niese und auch der größte, Th.
Mommsen, angeschlossen haben. Diese Gelehrten trachteten mittels einer auf dem Ver-
356 ENDRE FERENCZY
gleich der Texte der antiken Autoren beruhenden kritischen Analyse nach der Rekon
struktion der älteren Epochen Roms, vor allem natürlich des römischen Königtums.2 Die
philologisch-historische Richtung verwandelte sich in den Werken des Italieners E. Pais
in eine radikale Hyperkritik3, die die Authentizität der älteren Geschichte Roms bis zur
gallischen Katastrophe - mit Ausnahme zweier Momente4 - in Abrede stellte. Nach
Meinung Pais' zeigt diese Periode der römischen Geschichte vielfach Antizipationen und
Duplikationen aus der späteren Geschichte und ist voller ätiologischer sowie etymolo
gischer Mythen und ahistorischer Elemente5. Diese von Pais initiierte Richtung hat nur
wenige Anhänger gefunden, weil fast gleichzeitig mit der Erscheinung der Werke Pais'
neue Tendenzen in der älteren römischen Geschichte auftraten, die die Glaubwürdigkeit
der von Pais angegriffenen Tradition bestätigten. Die Ausgrabungen von Boni auf dem
Forum Romanum, die die archaische Inschrift des Lapis niger zu Tage förderten (1899),
der Fund Vaglieris auf dem Palatin (1907) usw. haben bewiesen, daß die Angaben der
Tradition der Glaubwürdigkeit nicht entbehren und deshalb nicht außer Acht gelassen
werden dürfen. Die konservative Richtung wurde besonders durch das monumentale
Werk G. De Sanctis bekräftigt6, der unter Benutzung der Ergebnisse der historischen
Hilfswissenschaften auf die Hinfälligkeit der hyperkritischen Feststellungen von Pais
hinwies.7
Die Verteidigung der radikalen Kritik an der Tradition hat der große Althistoriker K.
J. Beloch in seine monumentale Römische Geschichte übernommen.8 Er beschränkte
jedoch die Behandlung der Königszeit auf einen geringen Raum und stellte maßvoller als
Pais die Argumente der philologisch-historischen Richtung gegen die Tradition über die
Königszeit dar.
Trotz seiner großen Autorität konnte Beloch die Positionen der konservativen
Richtung nicht ins Wanken bringen.
In Folge der Ausgrabungen wurde das Bild des römischen Königtums fortwährend
klarer: J. Binder9, dann L. Wenger10 haben überzeugend darauf hingewiesen, daß die
Herrschaft der Etrusker über Rom historische Realität war und die Vorstellung
Mommsens über das einheitliche Königtum nicht mehr haltbar sei.11 Statt einer muß das
römische Königtum in zwei Perioden geteilt werden, in eine voretruskische und eine
etruskische Periode.12
ÜBER DAS RÖMISCHE KÖNIGTUM 357
Die Rehabilitierung der Tradition wurde nach dem erfolglosen Angriff von Beloch
entscheidend durch die Arbeiten von F. Altheim herbeigeführt. Sie haben eine völlig neue
Betrachtungsweise vor allem durch die Einbeziehung der vergleichenden Religionsge
schichte ins Leben gerufen,13 sie erweiterten und vertieften die schon von G. De Sanctis
initiierte Benutzung der historischen Hilfwissenschaften (vor allem der Sprachwissen
schaft). Von besonderer Bedeutung war Altheims Versuch, die Geschichte Italiens und
Roms in die des Mediterraneums einzugliedern, womit er den Spuren Ed. Meyers
folgte.14
Die Anhänger der konservativen Richtung mehrten sich durch den Einfluß der Werke
von De Sanctis und Altheim. So hat L. Pareti seine Synthese hauptsächlich auf der Basis
der literarischen Quellen geschaffen.15 Sein Versuch ging dahin, in der Tradition alles
nur irgend Mögliche für glaubwürdig zu erkennen. So erklärt er allein den Namen
Romulus für fiktiv, betrachtet jedoch die anderen Könige als historische Personen. Die
Legende über die von der Wölfin ernährten Zwillinge sei etruskischer Herkunft und
ebenso gingen die Organisationen der 3 Tribus und der 30 Kurien auf das Zeitalter der
etruskischen Herrschaft zurück16. König Numa könne eine "reale Person" sein und auch
die Geschichte des Tullus Hostilius habe einen historischen Kern. Er akzeptiert den
Feldzug des Tullus Hostilius gegen Alba Longa, die Zerstörung dieser Stadt und die
Übersiedelung ihrer Bewohner nach Rom als glaubwürdig, hält jedoch die Legende, die
die Gründung der Curia Hostilia Tullus Hostilius zuschreibt, nicht für authentisch.
Entgegen den Ausgrabungsergebnissen hält er es nicht für unmöglich, daß eine Siedlung
am Ort der späteren Stadt Ostia schon in der Zeit des Ancus Marcius entstand. Pareti
billigt den antiken Beleg über die Identität des Servius Tullius mit Mastarna, den er als
einen Würdenamen betrachtet. Servius Tullius war nach seiner Ansicht ein latinischer
magister, der nach den Tarquiniern die Macht übernahm und so die Sympathie der
römischen Annalistik erworben hat. Pareti betrachtet die Gesellschafts- und Verfassungs-
reform von Servius Tullius als Antizipation aus der Zeit der Republik, weist diesem
König, in dem er einen Vorläufer der Republik vermutet, jedoch ein demokratisches
Regime zu. Tarquinius Superbus war laut Pareti eine historische Person, jedoch kein
Tyrann griechischen Typs, wie ihn die Tradition darstellt, sondern ein etruskischer
lucumo, der seine Macht auf Latium ausgedehnt hat. Die Eroberung Latiums war aber
keine römische, sondern eine etruskische Expansion.
358 ENDRE FERENCZY
In der gleichzeitig verfaßten Storia die Roma17 nimmt R. Paribeni, ähnlich Pareti,
einen konservativen Standpunkt gegenüber der Tradition ein. Auch er hält mit Ausnahme
des Romulus die anderen drei Könige der voretruskischen Zeit für historische Personen,
obwohl er die Skepsis gegen die historische Realität des Ancus als rechtmäßig anerkennt.
Paribeni betont die Bedeutung der etruskischen Könige in Rom, die die italienischen
Städte in Bünden organisierten, und betrachtet sie deshalb als die Vorläufer der Ver
einigung Italiens.18
Auch der englische Historiker H. Last zeigt sich ähnlich den erwähnten Zeitgenossen
als ein Fortsetzer der konservativen philologisch-historischen Richtung und übernimmt
die Verteidigung der Tradition.19 Nach Last enthält die Tradition in ihrem Kern histo
rische Realität - mit Ausnahme der Romulus-Legende.
Die Liste der Könige entstand nach Last vor dem 5. Jahrhundert und es gibt keinen
Grund, sie als Fiktion im Interesse der damaligen römischen Würdenträger darzustellen.
Er hält die sabinische Herkunft Numas für authentisch und nimmt die Anwesenheit
sabinischer Ansiedler schon im älteren Rom an. Nach seiner Meinung war Rom eine
latinische Stadt von Numa bis Ancus Marcius, und darauf folgten die 3 Könige der
etruskischen Periode.
Fast gleichzeitig mit den oben angeführten Werken erschienen die Arbeiten des
hervorragenden schwedischen Archäologen E. Gjerstad, die eine andere Richtung in der
Forschung über die römische Königszeit zeigten.20 Der schwedische Gelehrte hat eine
völlig neue Konzeption über die älteste römische Geschichte erarbeitet. Er ging von der
methodologischen These aus, daß die Rekonstruktion der ältesten Geschichte Roms statt
auf dem unglaubwürdigen literarischen Quellenmaterial auf den archäologischen Funden
aufgebaut werden müsse und die literarischen Quellen dabei nur als ergänzendes
Quellenmaterial beigesteuert werden dürfen.21 Eine große mit Gjerstads Konzeption
verbundene Neuerung war vor allem die völlige Umgestaltung der altherkömmlichen
Chronologie der Königszeit. Er setzte die Gründung der Stadt, die er den Etruskern
zuschrieb, nicht mehr auf das traditionelle (varronische) Datum 753, sondern auf 575 v.
Chr. und vereinte sie mit der des Staates. Damit würde die Dauer der Königszeit, die sich
in der Überlieferung auf etwa 244 Jahre belief, soweit verkürzt, daß sich die Ereignisse
bis zur Unwahrscheinlichkeit verdichtet hätten. So fühlte sich Gjerstad gezwungen,
einerseits eine präurbane Periode als Vorgeschichte dem Zeitalter des Königtums einzu-
gliedern und andererseits, um die Häufung der Ereignisse zu vermeiden, die letzte
Periode der Königszeit bis zum Dezemvirat zu verlängern.
Schon daraus geht klar hervor, daß der Versuch Gjerstads, die Entwicklung seiner
Konzeption gemäß darzustellen, auf große Schwierigkeiten stieß. 22 Die Kritiker der
Konzeption von Gjerstad haben mit Recht darauf hingewiesen, daß die neuen
Grundlagen, auf denen der schwedische Gelehrte seine Rekonstruktion der Königszeit
aufgebaut hat, der Authentizität keineswegs näher liegen als jene, die durch die
Überlieferung geboten werden. Die Versuche Gjerstads, die Ereignisse der
Ausgrabungen mit den Angaben der literarischen Quellen in Einklang zu bringen, sind
völlig mißlungen bzw. lieferten den Nachweis, daß sich zwischen den Ergebnissen der
Ausgrabungen und den Belegen der Tradition keine unmittelbaren Beziehungen herstellen
lassen.
Viele von den interessanten, originellen Feststellungen des schwedischen Gelehrten
bedürfen einer näheren Erklärung. Die meisten seiner zahlreichen Hypothesen scheinen
nicht zwingender zu sein als die jener Historiker, die sich auf die Tradition stützen. So
wirkt jene Hypothese sehr kühn, nach der der Senat und die Kurien schon in der
präurbanen Epoche bestanden hätten, da die Ausgestaltung dieser Institutionen einen
schon entwickelten Staat voraussetzt. Ebenso kühn ist es, die Entstehung des Volks-
tribunats in der Zeit der Königsherrschaft anzusetzen. Die Ernennung des Volkstribuns
durch den König konnte keine historische Realität sein, da Königtum und Volksrechte in
einem antiken Staat sicherlich nicht nebeneinander bestehen konnten.
Die Forschungsergebnisse Gjerstads haben unter den Althistorikern großes Aufsehen
und bei den meisten Widerstand erregt,23 doch waren die Bedeutung seiner Methode und
die Wirkung seiner Ergebnisse für die zeitgenössische und spätere Forschung ent
scheidend. Das von Gjerstad betonte Primat der Archäologie in der Erforschung des alten
Rom wurde zwar von der Kritik angefochten, es konnte jedoch nicht mehr bezweifelt
werden, daß die Bedeutung der Archäologie derjenigen der schriftlichen Quellen
zumindest gleichzustellen ist.24
Die italienischen Archäologen verfolgen trotz der Tatsache, daß sie die Resultate
Gjerstads verworfen haben, gegenwärtig auch die Methode, die die Ergebnisse der
Ausgrabungen mit den Belegen der literarischen Quellen zu vereinbaren sucht. Der
Unterschied zwischen dem Verfahren Gjerstads und den italienischen Archäologen ist
360 ENDRE FERENCZY
jedoch bedeutend. Während Gjerstad neben dem archäologischen Quellenmaterial das

literarische nur als Hilfsquelle benützte, benutzen die italienischen Forscher im Gegenteil
die Ergebnisse der Ausgrabungen zum Beweis der Glaubwürdigkeit der literarischen
Quellen. Diese Bemühungen der italienischen althistorichen Forschung haben sich trotz
aller Erfolge auch unvermeidbar in Konflikte verwickelt, da unterschiedliches Quellen
material manchmal nicht denselben Tatbestand darstellt.25
Die Erfolge, die die italienischen Archäologen unter der Führung ihres
hochverdienten Meisters M. Pallottino erzielt haben, sind zweifellos sehr bedeutend.
Pallottino hat besonders in der Ausarbeitung einer neuen Disziplin, der Etruskologie,
außerordentliche Verdienste erworben,26 seine Tätigkeit wird jedoch auch auf anderen
Gebieten der altitalischen und altrömischen Forschung als erfolgreich anerkannt.27 Seine
Schüler (G. Colonna, M. Coarelli, M. Cristofani, L. Quilici, M. Torelli, A. Sommella
Mura usw.) haben die Anregungen ihres Meisters mit ausgezeichneten Ergebnissen
befolgt. Ihre Ausgrabungen erweiterten sich über Rom hinaus und führten zur
Entdeckung der antiken Siedlungen Latiums und Etruriens.28 Andererseits haben die
Forschungen in Süditalien, in der ehemaligen Magna Graeca, neues Licht auf die
Beziehungen zwischen Hellenentum und Latium im königlichen Zeitalter geworfen und
die Behauptungen Altheims und seiner Anhänger bestätigt, die seit den 30er Jahren ein
unmittelbares Eindringen der griechischen Kultur in Latium und Rom zu beweisen
versuchten.29
Gleichzeitig mit der Forschungstätigkeit von Gjerstad und Pallottino eröffnete das
Werk A. Alföldis eine völlig originelle Richtung in der Erforschung der älteren römischen
Geschichte.30 Mit vielseitiger tiefer Gelehrsamkeit vor allem auf dem Gebiet der
Archäologie, Kunstgeschichte, hauptsächlich jedoch der Numismatik, hatte Alföldi einen
großen Vorteil gegenüber den zur philologisch-historischen Schule gehörenden Alt
historikern, von denen A. Momigliano, der hervorragende italienische Gelehrte, als
ewiger Rivale stets die Resultate Alföldis bestritt.31 Die wissenschaftliche Kontroverse
dieser zwei großen Gelehrten wurde im letzten Vierteljahrhundert zur ständigen Er
scheinung der Fachliteratur des Altertums, die bedauerlicherweise (von Momiglianos
Seite) auch Anzüglichkeiten nicht entbehrte. In seinen einschlägigen Untersuchungen hat
Alföldi vor allem die Frühgeschichte der Römer bzw. die Gesellschaft der Königszeit auf
komparativer Grundlage untersucht, indem er sich in seinen Resultaten auf die Gesell-
Schaftsstruktur der Nomadenvölker und der Griechen stützte.32 Die uralte Dreiteilung bei
den Römern wurde von Alföldi als ein indoeuropäisches Charakteristikum erwiesen,
deren Entsprechung auch bei den Doriern festgestellt werden kann.33 Die Entwicklungs
stufe der uralten Dreiteilung in Rom wurde durch die Institutionen der 3 Stammtribus und
der 30 curiae repräsentiert und stand auf der gesellschaftlichen Basis des Matriarchats.
Neben der Dreiteilung existierte auch eine Zweiteilung auf patriarchalischer Basis in
Rom, die durch die Doppelinstitutionen (Doppelmonarchie, Doppelmagistratur usw.) und
die Exogamie charakterisiert werden kann.34
Nach bedeutenden Vorstudien35 hat Alföldi in seinem vieldiskutierten Meisterwerk
Early Rome and the Latins seine grundlegend neue Konzeption über das römische
Königtum dargelegt.36 Die Verneinung der römischen Hegemonie über Latium in der
Königszeit bildete die hauptsächliche Neuigkeit dieses Buches, das auch noch andere
neue Forschungsergebnisse enthielt. Alföldi hat die römische Eroberung Latiums durch
die Könige als Fiktion erwiesen und dem ersten römischen Annalisten, Fabius Pictor,
zugeschrieben, dessen Glaubwürdigkeit er (gegen Mommsen) in Abrede stellt.37 Seine
Feststellung, daß Rom, abgesehen von der Herrschaft der Tarquinii, auch von anderen
etruskischen Städten (Caere, Vulci, Veii, Clusium) erobert wurde38, haben in ihrer Mehr
heit seine Kritiker ebensowenig akzeptiert wie seine Hypothese über eine kymäische
Lokalchronik, aus der Dionysius Halicarnassensis geschöpft habe.39 Nach Meinung
Alföldis hat diese Chronik den Verfall der Etruskerherrschaft in Latium bzw. den
Freiheitskrieg der Latiner im Bündnis mit dem Tyrann von Kyme, Aristodemo s dem
Malakos, geschildert. Die Grundlagen dieser Hypothesen von Alföldi verdienen, obwohl
sie tatsächlich auf nicht zwingenden Quellenbelegen fußen, doch eine weitere Prüfung, da
sie der Wahrscheinlichkeit nahe zu stehen scheinen.40 Weniger überzeugend ist die
Theorie über die Ausgestaltung des Patriziats, das sich nach Alföldi aus der ehemaligen
Leibgarde der Könige bildete.41 Entsprechend fehlt jeder ernsthafte Beweis für die
Theorie, daß die Authentizität des foedus Cassianum in Abrede zu stellen sei.42
Das Konzept der uralten Dreiteilung bei den Indogermanen (die auch Alföldi akzep
tierte) wurde von dem ausgezeichneten französischen Religionshistoriker G. Dumèzil auf
breiter vergleichenden Grundlage ausgearbeitet und weiterentwickelt. Seine
Bemühungen, die indogermanischen Wurzeln der Religion, Gesellschaftsstruktur usw.
bei den Völkern des Altertums festzustellen, betrafen unsere Kenntnisse über die Römer
362 ENDRE FERENCZY
im Kern.43 Dumèzil hat die 3 römischen Stammtribus mit der uralten Dreiteilung (Kasten)
der Indoeuropäer gleichgesetzt. Nach seiner Ansicht repräsentierten die Ramnes die
Priester, die Tities die Produzenten (Bauern) und die Luceres die Krieger. In der Religion
entsprachen dieser Dreiteilung in Rom Jupiter, Quirinus und Mars, die "himmlischen
Aufseher" der dreigeteilten Gesellschaft.44 Später modifizierte Dumèzil diese Theorie, er
hieltdie.drei römischen Urtribus nicht mehr für Kasten und wandte seine Ideologie über
die Dreiteilung nur auf dem Gebiet der Religion und der Mythen an.45
Unter den Kritikern Dumèzils befand sich der früher schon erwähnte Antagonist
Alföldis, Momigliano.46 Ähnlich seinem großen ungarisch-amerikanischen Rivalen hat
auch er mit seiner umfangreichen wissenschaftlichen Tätigkeit fast alle Gebiete der alten
Geschichte umgepflügt, doch fällt der Schwerpunkt seiner Forschungen auf den Bereich
der älteren römischen Geschichte.47 Die Mehrheit seiner Studien auf diesem Gebiet steht
in Beziehung mit den Feststellungen Alföldis, die Momigliano stets zu widerlegen suchte.
So bestreitet Momigliano z. B. die These Alföldis über die Vorherrschaft der Reiterei
nach dem Fall der Könige in Rom.48 Auf Grund der Unterordnung des magister equitum
unter den dictator stellt Momigliano fest, daß die größere gesellschaftliche Bedeutung im
ältesten Rom dem Fußvolk zukam.49 "Natürlich" tritt Momigliano für die These
Pasqualis über die "grande Roma dei Tarquinii" ein,50 weil AlföIdi diese Theorie zur
Fiktion des Annalisten Fabius Pictor gestempelt hatte.51 Statt auf die weiteren
unfruchtbaren Kontroversen zwischen den großen Althistorikern einzugehen, erwähnen
wir einen geistvollen, leider nicht beweisbaren Einfall Momiglianos, den er zur Erklärung
der ältesten Zusammensetzung des Senats vorbrachte. Bei der Anrede der Senatoren
"patres conscripti" sieht Momigliano in dem zweiten Wort "conscripti" nicht ein Adjektiv
zu "patres" und auch nicht einen Ausdruck für die plebejischen Senatoren, sondern für
die Repräsentanten einer gesellschaftlichen Schicht zwischen patricii und plebei. Nach der
Meinung Momiglianos vermischte sich diese Schicht später mit der Plebs.52 Ebenso wie
andere italienische Archäologen und Historiker hat auch Momigliano in seinen Werken
die Glaubwürdigkeit der Tradition verfochten.53
Angesichts seines weiten Gesichtskreises und seiner tiefen Gelehrsamkeit verdient G.
Giannelli eine eingehende Besprechung. Der Trattato di storia romana54 des italienischen
Historikers erkennt die Authentizität der Tradition über das Königtum außer im Hinblick
auf die zwei ersten Könige an. Er stellt die Struktur der Gesellschaft des Königtums
anders dar, als es der communis opinio entspricht. Nach Giannelli sind die ältesten
Organe der Gesellschaft die 30 Kurien, die schon vor der Staatsgründung entstanden.
Aus diesen wurden die drei Tribus künstlich gebildet und entwickelten sich die gentes
erst nach der Ausgestaltung der Tribus. Die Klassen der Gesellschaft, und zwar patricii,
plebs, clientes, entstanden nach Giannelli schon in der ersten Periode des Königtums;
unter der etruskischen Herrschaft begann eine Demokratisierung der Gesellschaft, die mit
dem Fall der Monarchie abbrach.
Von den deutschen Althistorikern stellt A. Heuss in seiner hervorragenden
Römischen Geschichte55 die sagenhafte Vorgeschichte Roms ebenso in Abrede wie auch
die erste (sog. latinisch-sabinische) Periode. Nach dem Göttinger Gelehrten beginnt die
Geschichte Roms in der Tat mit der etruskischen Herrschaft, und er spricht sich für "das
große Rom der Tarquinier" aus. Nach seiner Meinung "war diese 'etruskische'
Königszeit, mit der Rom seinen Einzug in die Geschichte vollzog, in jeder Hinsicht eine
glänzende und ruhmvolle Periode".56
Zu den großen Historikerpersönlichkeiten aus der Mitte des 20. Jahrhunderts gehört
auch J. Heurgon.57 Er gliederte die Geschichte des älteren Rom organisch in die des
Mediterraneums ein und stellte sie so im welthistorischen Rahmen dar. Heurgon übt eine
maßvolle Kritik an der römischen Tradition: die erste (latinisch-sabinische) Periode des
Königtums ist von sagenhafter Art und weist zum Großteil ahistorische Züge auf. Er
stellt detailliert die etruskische Periode dar und behandelt besonders eingehend die
Epoche des Servius Tullius, dessen gesellschaftliche und militärische Reformen er (gegen
die Meinung Alföldis, aufgrund der Theorie von Fraccaro) als glaubhaft beurteilt.58
Andererseits folgt er der Theorie Alföldis bezüglich der etruskischen Eroberungen Roms
und vermutet außer Tarquiniis Herrschaft über die Stadt auch die anderer etruskischer
Städte.59 Heurgon erkennt aufgrund der Ergebnisse der neuesten Ausgrabungen in Rom,
daß die etruskische Herrschaft tiefere Spuren in der Stadt zurückgelassen hat, als früher
angenommen wurde.60 Es zeigte sich jedoch aus den Ereignissen nach dem Fall des
Königtums, daß die Maßnahmen der etruskischen Herrscher die Opposition der lati
nischen Urbevölkerung nicht beseitigen konnten. Dies geht einerseits aus der Vorherr
schaft der patrizischen Geschlechter nach dem Untergang des Königtums hervor,
andererseits aus der Tatsache, daß die Mehrheit der Neuerungen der etruskischen
Könige, wenn auch nur vorübergehend, am Anfang der Republik aufgehoben wurde.
364 ENDRE FERENCZY
Von den neuesten Zusammenfassungen der römischen Geschichte, die auch die
Geschichte des Königtums eingehend behandeln, verdient das Werk des hervorragenden
Althistorikers H. Bengtson große Beachtung.61 Er akzeptiert die Konzeption Gjerstads
nicht, lehnt jedoch auch die Authentizität der Tradition für die Königszeit ab, weil sie der
Kriterien der historischen Realitäten entbehrt. Eine maßhaltende Übersicht und Wertung
der modernen Rekonstruktionen finden sich auch in den Werken des ausgezeichneten
Marburger Historikers K. Christ.62 Er bietet in seinen Schriften nüchterne Meinungen
und ausführliche Quellen- und Literaturnachweise zum großen Nutzen auch der Forscher
der älteren römischen Geschichte.63
Außer den schon angeführten historischen Werken haben die Arbeiten mehrerer
ausgezeichneter italienischer Philologen und Historiker zur Rehabilitation der literarischen
Tradition beigetragen. Der große Gelehrte von Pavia PI. Fraccaro64 und sein bester
Schüler E. Gabba beleuchteten die schwierigsten Probleme der römischen Vergangenheit
in einer fast unübersehbaren Reihe von gründlichen, nicht selten bahnbrechenden Stu
dien, und die Ergebnisse ihrer Untersuchungen fanden bei den Historikern großes Echo.
Bezüglich der Servianischen Verfassung weckte eine Studie von Fraccaro außer
ordentliches Interesse.65 Er entwickelte hier eine geistreiche Theorie, die mit Hilfe der in
der Servianischen Verfassung überlieferten Zahl der Centurien der Juniores der ersten 3
Klassen eine Erklärung über die Entstehung der römischen Legion bietet und andererseits
den Zeitpunkt und die Authentizität der Centurienordnung im Sinne der Tradition zu
bekräftigen versucht.66
Auch aus der das ganze Gebiet der römischen Geschichte umfassenden wissenschaft
lichen Tätigkeit von E. Gabba behandeln zahlreiche Studien die ältere Geschichte Roms.
Unter anderen gilt als besonders bedeutende Feststellung Gabbas, daß die griechischen
Schriftsteller als Quellen der römischen Annalistik für die Gründung und älteste Ge
schichte Roms dienten.67 Nicht minder wichtig sind seine Untersuchungen über die
Schriftsteller, deren Werke die römische Tradition gebildet haben.68 Unter den anderen
Altertumsforschern, die sich mit der älteren Geschichte Roms beschäftigen, ist besondern
beachtenswert die kontinuierliche wissenschaftliche Tätigkeit des Historikers von Padua
A. Bernardi.69
Das gediegene Werk von E. Peruzzi, Le origini di Roma10, gehört der ultrakon
servativen Richtung an. Der italienische Historiker erkannte in seiner Arbeit rückhaltlos
die Glaubwürdigkeit der Tradition an, die er noch mit neuen Beweisen zu verstärken
versuchte. Die Beweise, die er im Interesse seiner Zielsetzungen benutzt, sind haupt
sächlich von philologischer (d. h. linguistischer) Art, während die Ergebnisse anderer
Wissenschaften (Archäologie, Rechtsgeschichte usw.) nur in dem Maß zur Anwendung
gebracht wurden, wie sie die philologischen Beweise stärken konnten. Peruzzi hält auf
Grund linguistischer Analysen bzw. Voraussetzungen eine auf das 13./12. Jahrhundert
zurückgehende mykenische Siedlung auf dem Palatin für bewiesen und akzeptiert die
archäologischen Angaben über das frühe Eindringen der hellenischen Kultur in Latium,
als dessen Ausgangszentrum er die Stadt Gabii vermutet.71 Daneben betont er die
Bedeutung des sabinischen Einflusses in Rom sowohl in religiöser als auch kultureller
Hinsicht. Er glaubt auch ohne Einschränkung an die Realität aller römischen Könige, die
durch die Tradition überliefert wurden. Besonders unterstreicht Peruzzi die Wichtigkeit
des (nach seiner Version) sabinischen Königs Numa und bezweifelt auch nicht die
Authentizität der im Jahr 181 v. Chr. gefundenen, diesem König zugeschriebenen
Bücher. Er hält sie für glaubwürdige Dokumente der schon in der zweiten Hälfte des 8.
Jahrhunderts v. Chr. entwickelten sabinisch-römischen Kultur.72
Als ein Antipode des von Peruzzi betonten frühen hellenischen und sabinischen
Einflusses trat J. Poucet mit wichtigen, beachtenswerten Argumenten auf.73 Er hält die
linguistischen und archäologischen Beweise Peruzzis über die frühen hellenischen Ein
flüsse in Latium und Rom für völlig unbegründet bzw. unsicher und stellt auch die
Historizität der sabinisch-latinischen Periode des Königtums in Abrede. Er lehnt die
diesbezüglichen Ansichten von Gjerstad, Alföldi, Pallottino und anderer Forscher ab und
behauptet, daß die Tradition über die ersten 4 Könige von indoeuropäischen Elementen
durchwoben ist.74 Nach Poucets Meinung kann die der Tradition zustimmende Hypo
these der Archäologen, die Gründung Roms sei Jahrhunderte später als die von Alba
Longa und Lavinium anzusetzen, nicht für glaubwürdig gehalten werden. Die chrono
logische Ordnung der Städtegründungen stamme aus einer späteren Konzeption, die mit
der Legende von Aeneas in Verbindung steht und den Trojanischen Ursprung Roms zu
beweisen versucht.75 Auf Grund der Ergebnisse der Archäologie erkennt Poucet
entgegen der Tradition, daß nicht Rom, sondern Lavinium die führende Stellung im alten
Latium innehatte.76 Als Resultat seiner Untersuchungen stellt Poucet fest, daß einerseits
die Tradition bis zur Regierung der Tarquinier keine historische Authentizität habe77 und
366 ENDRE FERENCZY
daß andererseits die Angaben der Tradition über die Rolle der Sabiner am Anfang der
Geschichte Roms bzw. des Königtums als Erfindungen (Antizipationen) betrachtet wer
den müssen. Historische Beziehungen zwischen den Römern und S abinern kamen nach
Poucet vor der Mitte des 5. Jahrhunderts nicht zustande.78
***
Die Absonderung der rechtshistorischen Forschung von der historischen ist gegen
wärtig nicht gerechtfertigt, da im letzten halben Jahrhundert die Römische Rechtsge
schichte eine katexochen historische Disziplin wurde. Die Mehrheit der Rechtshistoriker
hat sich vom Zauber des Staatsrechts Mommsens gelöst und erkannt, "daß Mommsens
Systematik, wie fruchtbar sie auch für das Verständnis der römischen Institutionen war,
doch stark überspitzt und eine Quelle wesentlicher Schwächen des Werkes ist".79 Der
historische Faktor, der in Mommsens Werk im Hintergrund blieb, rückte in den neueren
rechtshistorischen Werken schrittweise in den Vordergrund, und zwar in solchem Maße,
daß das Gleichgewicht zwischen dem Juridischen und dem Historischen sich zum Vorteil
des Historischen veränderte.
Die niveauvolle Übersicht, die der große deutsche Romanist W. Kunkel über die
Mehrheit der bedeutendsten rechtshistorischen Werke der letzten Jahrzehnte verfaßte,80
enthebt mich der Verpflichtung, diese Arbeiten eingehend zu behandeln. Von diesen
werden hier nur jene erwähnt bzw. behandelt, die sich auch mit dem römischen
Königtum befaßt haben. Da eine umfangreiche Literatur sich mit dem Übergang von der
Königszeit zur Republik beschäftigt, werden wir diese Periode wegen des begrenzten
Rahmens dieser Studie nicht berücksichtigen.81
Die unten zur Besprechung kommenden Werke haben - trotz ihrer großen Unter
schiede bezüglich Auffassung und Methode - eine gemeinsame Eigenart: alle halten mehr
oder minder an der Authentizität der literarischen Tradition fest. (Daneben benützt die
Mehrheit dieser Rechtshistoriker auch andere Quellen, d. h. die Ergebnisse der hi
storischen Hilfswissenschaften.) Der historische Faktor steht bei diesen Gelehrten im
Gleichgewicht mit dem juristischen, ja ist sogar öfters überwiegend. Die Wirkung
Mommsens ist bei den meisten offenbar, so halten z. B. Siber82, Frezza83, De Martino84
und Guarino85 ähnlich Mommsen den König für einen Magistrat, während De
Francisci 86 und Coli87 diese Würden dezidiert unterscheiden. Die erwähnten
Rechtshistoriker (mit einer Ausnahme) akzeptieren die Zweiteilung des Königtums in eine
laziale (latinisch-sabinische) und eine etruskische Periode. Die erste Periode wird von der
Mehrheit dieser Forscher als verwandt mit der posthomerischen und als eine
patriarchalische Epoche des Königtums betrachtet - mit Ausnahme Colis, der für das
ganze königliche Zeitalter die unbeschränkte königliche Macht als charakteristisch
ansieht.88.
Ernste Meinungsunterschiede bestehen über die Königswahl bzw. die Thronfolge. De
Francisci,89 Bonfante 90 und v. Lübtow91 vertreten den Standpunkt, daß der König von
seinem Vorgänger ernannt wurde, während andere (Siber92, De Martino93, Frezza94, E.
Meyer95) der Ansicht sind, daß der König durch den Interrex designiert wurde.96 Nach
Coli 97 , Kunkel 98 und Catalano " waren bei der Designierung (Wahl) des Königs die
religiösen Akte (captio, augurium) entscheidend.
Die Meinungen der Rechtshistoriker über das römische Königtum wurden natürlich
durch die Forschungsergebnisse der Historiker wesentlich beeinflußt bzw. gefördert100,
doch haben fast alle bedeutsamen Rechtshistoriker selbständige Konzeptionen ausge
arbeitet, mit denen sie zur historischen Rekonstruktion beitrugen. E. Meyer behandelt in
seinem meisterhaften Werk Römischer Staat und Staatsgedanke101 das römische Staats
recht in seiner historischen Ausgestaltung. Der Schwerpunkt seiner Synthese liegt in der
Behandlung der Republik, er bezieht jedoch sowohl das Königtum als auch die Kaiserzeit
in seine Betrachtung mit ein. Er akzeptiert im allgemeinen die Auffassung Altheims und
seiner Anhänger, übernimmt jedoch nicht einfach kritiklos die Berichte der antiken
Schriftsteller oder die Feststellungen der modernen Forscher. Nach E. Meyers Ansicht
hatte der römische König eine einheitliche allumfassende und ausschließlich religiös ge
bundene Macht, die sich auf sämtliche Gebiete des Staatslebens erstreckte. Diese Er
kenntnis wird aus der Kompetenz der Rechtsnachfolger der Könige, der Konsuln, her
geleitet. Der despotische Charakter des Königtums geht nach E. Meyer auf die Etrusker
zurück und erhält Bestätigung durch den Fund von Vetulonia, wo in einem Grab ein
Doppelbeil, das Abzeichen der Machtvollkommenheit bei den Römern, entdeckt wurde.
Die Storia della costituzione romana von F. De Martino ist nicht nur die größte
staatsrechtliche Synthese seit Mommsens monumentalem Römischen Staatsrecht,
sondern ist wirklich auch eine bahnbrechende Leistung, die die Entwicklung der
römischen staatlichen Institutionen auf völlig neuen Grundlagen rekonstruierte.102 Der
368 ENDRE FERENCZY
Versuch des Verfassers, die marxistische Theorie in der Rekonstruktion durchzusetzen,

läßt seine Darstellung nicht einseitig werden, da er - obwohl er die wirtschaftlichen
Aspekte betont - jedem Dogmatismus fernsteht und seine Geschichtsanschauung nicht
überspitzt.103
Als Gegenpol De Martinos hat P. De Francisci in seinen Synthesen die Entwicklung
der römischen Institutionen in ultrakonservativem Geist dargestellt.104 Es gibt kein
rechtshistorisches Werk, das auf so breitem und in die Tiefe gehendem Fundament die
ältere Geschichte Roms rekonstruiert hätte wie die Primordia civitatis De Franciscis.105 Er
hat hier die Ergebnisse auch ferner historischer Hilfswissenschaften benützt und die
Anfänge der römischen Rechtsgeschichte tief in die Vorgeschichte verlegt. Er sieht in der
gens ein Aggregat von Familien mit gemeinsamen Bräuchen; das römische Gemeinwesen
sei aus einer Zusammenfassung autonomer gentes entstanden. Diese Zusammenfassung
wurde durch einen ductor von charismatischer Stellung ausgeführt. Wie De Francisci
glaubt, entwickelte sich der Staat nicht durch eine Föderation der gentes, sondern durch
die Verschmelzung zur Gefolgschaft des ductor (?). Der italienische Gelehrte nennt die
auf den ductor folgende Phase des Königtums die Periode des rex inauguratus, die sich
in der späten Königszeit durch die Einführung der Hoplitenphalanx zu einer
militarisierten Monarchie entwickelte. Es ist leicht einzusehen, daß dieser von mir stark
verkürzte Gedankengang De Franciscis großteils auf Hypothesen basiert, deren
Fundierung schlechthin fraglich erscheint. Der große Reichtum von Einzelheiten in
diesem Werk macht es jedoch zu einem unentbehrlichen Handbuch der Rechtsgeschichte
des römischen Königtums.106
Die Storia del diritto romano107des bedeutenden Romanisten von Neapel A. Guarino
besticht durch ihre Originalität und Gründlichkeit. Der sehr produktive und vielseitige
Gelehrte stellt die Geschichte der Institutionen des Königtums frei von jedweden
Schablonen dar. Seine Periodisierung der Geschichte des alten Rom beruht nicht auf
äußerlichen Merkmalen, wie z. B. den Arten der Regierungsform, sondern auf der
Eigenart der Gesellschaft. Entsprechend dieser historischen Anschauung nennt Guarino
die ersten vier Jahrhunderte Roms "civitas Quiritaria", weil die Vollkommenheit der
Bürgerrechte im Zeitalter vom 8. bis zur Mitte des 4. Jahrhundert allein den Quintes, d.
h. den Mitgliedern der gentes patriciae, zugestanden war. Das Zeitalter der civitas
Quiritaria verteilte sich nach Guarino auf zwei Phasen: die latinisch-sabinische und die
etruskisch-latinische. Die Regierungsform von den Anfängen bis zum traditionellen

Datum 510 v. Chr. war das Königtum, und obwohl er die dem König Servius Tullius
zugeschriebene politische Reform, die Einführung der comitia centuriata, dezidiert in
Abrede stellt, hebt er die große Bedeutung der militärischen Reform der etruskischen
Könige hervor. Die Aufstellung des exercitus centuriatus (der Hoplitenphalanx) war eine
äußerst bedeutende Neuerung der etruskischen Könige, die mit Hilfe dieser Armee die
Eroberung Latiums durchführten. Diese militärische Reform blieb nach Guarino nicht
ohne Einfluß auf die Lebensart der Römer, jedoch hält er an der Ansicht fest, daß diese
Neuerung, obwohl sie den Plebejern den Eintritt in den exercitus centuriatus ermöglichte,
die gesellschaftlich-politische Lage der plebejischen Bevölkerung unverändert ließ.108
Der Fall des Königtums geschah nach Guarino mit Gewalt, und die Plebs spielte eine
Rolle beim Sturz der etruskischen Monarchie.
Von den neuesten italienischen Rechtshistorikern vertrittt L. Capogrossi Colognesi
die Ansicht, daß die Kurien vor der Tribusorganisation entstanden.109 Die Vorläufer der
30 Kurien waren nach Capogrossi die pagi. Die 24 pagi, die in den sacra Argeorum
erschienen, wurden bei der Gründung des Staates auf 30 ergänzt, während die Tribus
später zustande kamen. Seiner Meinung nach haben die Latiner den Palatin und die
Sabiner den Quirinal besiedelt, und der Staat entwickelte sich aus der Vereinigung dieser
beiden Völker.
Das Königtum wurde nach Capogrossi durch drei grundlegende Organe charak
terisiert. Diese sind: der rex, die Versammlung der Alten und der ganzen Gemeinde. Im
Laufe der Entwicklung wuchs die Macht des Königs durch die Unterstützung der
Bevölkerung, die dem König die Krieger stellte, so daß sie die bedeutendste im Staat
wurde. Auch der Senat, der Repräsentant der Geschlechter, vermehrte seinen Einfluß110,
aber der König als Führer der Armee und Richter der Gemeinde bewahrte seine Autorität
nicht nur, sondern erhöhte sich noch durch seine priesterlichen Funktionen, d. h. durch
seine Beziehungen zu den Göttern. Diese religiöse Position war nach Capogrossi von
zentraler Bedeutung besonders bei der Wahl des neuen Königs, und er stellt die
Wichtigkeit der priesterlichen Kollegien im alten Rom richtig dar. Capogrossi hält an der
Tradition fest und folgt im allgemeinen den Ansichten jener Forscher, die die Glaub
würdigkeit der Überlieferung verteidigen. So hält er auf der Grundlage der bekannten
Theorie von Fraccaro die Reformen des Servius Tullius für historisch.
370 ENDRE FERENCZY
Diese kurze Übersicht der neueren Untersuchungen auf dem Gebiet der Geschichte
(und Rechtsgeschichte) des römischen Königtums konnte nicht einmal ein annähernd
vollkommenes Bild über den heutigen Stand der Forschungstätigkeit in diesem Bereich
der Alten Geschichte bieten. Wir hoffen jedoch, daß wir mit der Darlegung des Erreichten
und des Angestrebten den Wissenschaftlern eine Anregung gegeben haben, die Bemü
hungen der Vorgänger zur Erlangung der gemeinsamen Ziele weiterzuführen.
ANMERKUNGEN
1 Vgl. G. Perl, Der Anfang der römischen Geschichtsschreibung, Forschungen und Fortschritte 38,
1964, 185 ff., M. Geizer, Der Anfang der röm. Geschichtsschreibung, Kl. Schriften, 3, Wiesbaden
1964, 93 ff., Römische Geschichtsschreibung, hrsg. von V. Pöschl, Darmstadt 1969.
2 Über die Entwicklung der modernen Forschung siehe außer der in Anm. 1 angeführten Literatur: H.
Bengtson, Grundriß der röm. Geschichte2 I, 1970, 4 ff., K. Christ, Römische Geschichte und
Wissenschaftsgeschichte, Bd. 3, Darmstadt 1983.
3 Vgl. C. Barbagallo, Il problema delle origini di Roma, 1970 [1. ed. Milano 1926].
4 Diese sind: die etruskische Herrschaft über Rom und die darauffolgende sabinische Invasion
5 Storia di Roma, Torino 1898 - 1899, Vol. 1-2., Storia critica di Roma antica, Roma 1913-1920,
Vol 1-5., Storia di Roma, Torino 1926, Vol 1-6.
6 Storia dei Romani, Torino 1907, Vol 1. [2.a ed. 1956.], Vol 2. Torino 1907 [2.a ed. 1960].
7 Er hat schon unter anderem die Ergebnisse des Linguisten W. Schulze, Zur Geschichte der
lateinischen Eigennamen, Abh. Gött. Ges. d. Wiss. V, 5, 1904 [Neudruck 1933] 62 ff. benutzt.
8 Berlin-Leipzig 1926.
9 Die Plebs, Studien zur röm. Rechtsgeschichte, Leipzig 1909.
10 Hausgewalt und Staatsgewalt im röm. Altertum, in: Misc Fr. Ehrle, 2, 1924.
11 Vgl. F. Leifer, Studien zum antiken Ämterwesen, 1: Zur Vorgeschichte des röm. Führeramts,
Leipzig 1931, (Klio 23. Beih.) 77 ff.
12 Vgl. P. De Francisci, Storia del diritto romano, 1, Milano 1925, 122. [2.a ed 1943, 166 ff.],
Arcana imperii III.l. Milano 1948, 30 ff.
13 Röm. Religionsgeschichte, Bd. 1-3, 1931-1933, Berlin. Ders., Röm. Religionsgeschichte, Baden-
Baden 1951-1953, Bd. 1-2., Altheim-Mattingly, A history of Roman Religion, London 1938.
14 Epochen der röm. Geschichte, Bd. 1-2, Frankfurt a.M. 1934-1935, Italien und Rom, Amsterdam
1941, Römische Geschichte, Bd. 1-2, Frankfurt a.M. 1951-1953, 2. verb. Aufl. 1956. Vgl. Ed.
Meyer, Geschichte des Altertums, Bd. 1-3, Bern 1953-1954.
15 Storia di Roma, 1, Torino 1952.
16 Vgl. Α. Magdelain, Recherches sur l'imperium: la loi curiate et les auspices d'investiture, Paris
1968.
17 Bologna, 1, 1954.
18 A.a.O., 64.
19  VII, 1954. Vgl. JRS 35, 1945, 30 ff.
20 Early Rome, Lund, 1, 1953, 2, 1956, 3, 1960, 4,1-2, 1966, Discussions concerning Early Rome,
Opuscula Romana, 3, Lund 1961, Legends and Facts in Early Roman History, Scripta minora R.
Societatis Humaniorum Litterarum Lundensis, 1960-1961, 2, Lund 1962, Discussions concerning
Early Rome, 2, Opuscula Romana, 5, Lund 1965,1 f f . , h e Origins of the Roman Republic, Les
origines de la République romaine, Fondation Hardt XIII., Vandoeuvres-Genève 1967, 3 ff.,
Discussions concerning Early Rome, 3, Historia 16, 1967, 257 ff.
21 Vgl. Discussions concerning Early Rome, 3, 257 ff.
22 Über die Kritik an Gjerstads Werken: G. Poma, Gli studi recenti nell'origine della repubblica
romana, Bologna 1974, 99 ff. u. passim.
23 Vgl. H. Riemann, GGA 213, 1960, 166 ff., 1962, 214, 16 ff., 222, 1970, 25 ff., 223, 1971, 33
ff., M. Pallottino, Archeologia classica 12, 1960, 1 ff., A. Momigliano, JRS 53, 1963, 958 ff. =
Terzo contributo degli studi classici, 2, Roma 1966, 661 ff., A. Alfòldi, Röm. Frühgeschichte,
Heidelberg 1976,20 ff.
24 Zustimmend: R. Bloch, Les origines de Rome, Paris 1959, Tite-Live et les premiers siècles de
Roma, Paris 1965.
25 Vgl. Lapis Satricanus, Archeological, epigraphical, linguistic and historical aspects of the new
inscription from Satricum, by C.M. Stibbe, G. Colonna, . de Simone and H.S. Versnel with an
introduction by M. Pallottino, Archeologische Studien van het Nederlands Instituut te Rome,
Scripta Minora, 5, S'Gravenhage 1980. Vgl. E. Ferenczy, Gymnasium 94, 1987, 97 ff.
26 Etruscologia6, Milano 1975.
27 Eine Gesamtbibliographie von Pallottinos Werken findet sich in: M. Pallottino, Saggi di
Antichità, 1, Roma 1979, XIX-XLVI. Über seine wissenschaftliche Bedeutung siehe: Gli Etruschi e
Roma, Incontro di studio in onore di M. Pallottino, Roma 1981.
28 Vgl. F. Castagnoli, Lavinium, 1, Roma 1972, 2, 1975, Lazio arcaico e mondo greco, La Parola del
Passato (PP) 32, 1977 mit den Beiträgen von A.M. Colini, P. Virgili usw., Lazio arcaico e mondo
greco, Il Convegno di Roma, PP 36, 1981. Die Synthese der Vor- und Frühgeschichte Italiens
wurde neuestens unternommen von ausgezeichneten italienischen Forschern in dem monumentalen
Werk: Popoli e civiltà dell'Italia antica (Coordinamento generale: M. Pallottino, G. Mansueti, A.
Prosdomici, O. Parlangeli). Im 2. Bd. dieser Arbeit wurde die Vor- und Frühgeschichte Italiens von
G. Colonna behandelt. Vgl. noch L. Quilici, Roma primitiva e le origini della civiltà laziale,
Roma 1979.
29 Über die Literatur der griechischen Kolonisation Süditaliens J. Heurgon, Rome et la Méditerranée
occidentale, 2,1980, 33 ff. Vgl. noch: F. Sartori, Problemi di storia costituzionale italiota, Padova
1953, G. Pugliese Carratelli, Lazio, Roma e Magna Grecia prima del secolo quarto a.C, in: Atti
dell' VIII Convegno Intern, di Studi sulla Magna Grecia, Taranto 198 = Scritti sul mondo antico,
Napoli 1976, 320 ff., D. Musti, Città di Magna Grecia, 1, RFIC 111, 1983, 265 ff., 2.114, 1986,
286 ff.
30 Über das wissenschaftliche Wirken A. Alföldis siehe: H. Bögli, Oeuvre litteraire d'André Alfòldi,
Sonderdruck aus dem Jahrbuch für Numismatik und Geldgeschichte 10, 1959/60, Andrew Alfòldi
1895-1981, The Institute for Advanced Studies, Princeton 1982.
31 Contributo alls storia degli studi classici e del mondo antico (I), 1955 - VII, 1984.
32 Vgl. Die Struktur des voretruskischen Römerstaates, Heidelberg 1974, passim.
372 ENDRE FERENCZY
33 Vgl. Die Struktur d. voretr. Römerstaates, 54 ff. Die große Literatur zur Indogermanenfrage, die in
E. Meyers Werk (Röm. Staat u. Staatsgedanke4, Zürich u. München 1964, 456 ff.) angeführt
wurde, sei hier nur durch die folgenden Arbeiten ergänzt: E. Benveniste, Le vocabulaire des
institutions indoeuropéennes, 1-2, Paris 1969. O. Szemerényi, Studies in the kinship terminology
of the Indo-European languages, Textes et Memoires, Vol. VII, Varia 1977, Acta Iranica 16,1978.
34 Vgl. A. Alföldi, La struttura politica di Roma nei suoi primordi, Rend. Cl. di scienze morali,
storiche e filologiche (Accad. dei Lincei), Serie VIII, vol. XXVII, fasc. 5-6, 1972 [1973]. Eine
Auswahl diesbezüglicher Studien: Struktur d. voretr. Römerstaates, passim. Der von Alföldi
vorausgesetzten Struktur der altr. Gesellschaft steht die Konzeption von E.A.R. Palmer entgegen,
die die Grundlage der Struktur der altr. Gesellschaft in den Kurien erkennt (The Archaic Community
of the Romans, Cambridge 1970). Vgl. die Kritik über Palmers Werk von F. Coarelli, RFIC 102,
1974, 233 ff. u. von A. Alföldi, Gnomon 44, 1972, 787 ff.
35 Die bedeutendsten davon: Der frührömische Reiteradel und seine Ehrenabzeichen, Baden-Baden 1952
[Repr. mit neuer Einl. Roma 1979], Die Trojanischen Urahnen der Römer, Basel 1957 [Repr.
Roma 1979], Rom und der Latinerbund im 500 v. Chr., Gymnasium 67, 1960, 193 ff., Ager
Romanus antiquus, Hermes 90, 1962, 187 ff.
36 Ann Arbor 1965 (im Folgenden: ERL). Deutsche Übersetzung: Das frühe Rom und die Latiner,
Darmstadt 1977.
37 ERL 123 ff., Derselbe, Römische Frühgeschichte, Heidelberg 1976, 48 ff. Anders: E. Gabba,
Considerazioni sulla tradizione letteraria sulle origini della Repubblica, in: Les origines de la
République romaine, Fondation Hardt XIII, 1967,135 ff.
38 ERL 206 ff.
39 ERL 56 ff. Die Anführung der Literatur über diese Frage bei G. Poma (o. Anm. 21) 34 ff.
40 Die Kritiken zu diesem Werk und die Gegenantwort des Verfassers: Römische Frühgeschichte.
Kritik und Forschung seit 1964, Heidelberg 1976. Zum Problem des "ager Romanus antiquus" vgl.
neuerdings: M. Humbert, Municipium et civitas sine suffragio, Rome, 1978.
41 Der frührömische Reiteradel (o. Anm. 34) 102 ff., Ders., Die Herrschaft der Reiterei in
Griechenland und in Rom nach dem Sturz der Könige, Festschrift für K. Schefold, Antike Kunst,
Beiheft 4, 1967, 13 ff. Gegen diese Theorie von Alföldi: F. Altheim, Römische Geschichte, 2,
Baden-Baden 1953, 429 ff., P.-Chr. Ranouil, Recherches sur le Patriciat, Paris 1975, E. Stuart
Staveley, The Nature and Aims of the Patriciate, Historia 32, 1983, 24 ff., E. Ferenczy,
Bemerkungen zu den neueren Theorien über den Ursprung des Patriziats, Festschrift K. Christ,
Darmstadt 1988,158 ff.
42 ERL 13 ff. = Das frühe Rom u. die Latiner, 330 f., Röm. Frühgeschichte 198. Vgl. E. Ferenczy,
RIDA 3e Serie 22, 1975, 223 ff., Mélanges J. Ellul, Paris 1983, 233 ff.
43 Eine kritische Darstellung von Dumèzils Werken bei: J. Heurgon, Rome et la Méditerranée
occidentale2, Paris 1980,41, A. Momigliano, Terzo contributo..., 2, 1966, 545 ff.
44 Vgl. Jupiter, Mars, Quirinus, essai sur la conception indoeuropéenne de la société et sur les
origines de Roma, 1941, Horace et les Curiaces, 1942, Tarpeia, cinq essais de philologie
comparative indo-européenne, 1947, L'idéologie des Indo-Européens, 1958.
45 Vgl. La religion romaine archaique, Paris 1966, Mythe et épopée, Histoires romaines, Paris 1973,
203 f.
46 Vgl. o. Anm. 42. Weitere Kritiken: A Alföldi, Die Struktur des voretruskischen Römerstaates,
Heidelberg 1974, 167, Röm. Frühgeschichte, 40 ff., E. Meyer, Röm. Staat und Staatsgedanke4,
Zürich München 1964,47757.
47 Sein konservativer Standpunkt der Tradition gegenüber wird unter anderem geäußert in JRS 47,
1957, 104 ff. = Wege der Forschung 90, 1969, 312 ff., Terzo contributo..., Roma 1966, 2, 545 ff.
48 Siehe o. Anm. 40. und seine Studie, [Centuria] procum patricium, Historia 17, 1968, 444 ff.
49 JRS 56, 1966, 16 ff. = Quarto Contributo..., Roma 1969, 377 ff.
50 Vgl. JRS 1967, 211 ff., RSI 81, 1969, 10 ff., Settimo Contributo..., Roma 1984, 406.
51 ERL 123 ff., Röm. Frühg., 52 ff.
52 Osservazioni sulla distinzione fra patrizi e plebe, in: Les Origines de la République romaine (.
Anm. 19), 199 ff.
53 Vgl. JRS 53, 1963, 95 ff. = Terzo Contributo..., 2, 1966, 545 ff., Settimo Contributo..., 1984,
379 ff.
54 Roma3 1965.
55 Braunschweig2 1964.
56 A.a.O., 12.
57 Rome et la Méditerranée2, Paris 1980.
58 Op. cit., 247 ff. Vgl. M. Pallottino, Servius Tullius, à la lumière des nouvelles découvertes
archéologiques et épigraphiques, CRAI 1977,216 ff.
59 Heurgon, op. cit. 240 ff. Vgl. Alföldi, ERL, 209 ff.
60 Op. cit., 257 ff.
61 Grundriß der römischen Geschichte mit Quellenkunde2, Bd. 1, München 1970.
62 Die Römer, Eine Einführung in ihre Geschichte und Zivilisation, München 1979, Römische
Geschichte und Wissenschaftsgeschichte, Bd. 3: Wissenschaftsgeschichte, Darmstadt 1983.
63 Im Werk "Die Römer" finden sich gründliche bibliographische Hinweise auch auf das Gebiet der
älteren römischen Geschichte (255 ff.). Entsprechend instruktiv sind die vortrefflichen Biographien
und Charakterisierungen großer Altertumsforscher im 3. Bd. d. Röm. Gesch. u. Wissenschafts
geschichte (B.G. Niebuhr, Th. Mommsen, E. Meyer u.a.).
64 Die Anführung seiner wissenschaftlichen Werke in: Opuscula, 1, XV ff., 3, 295 ff.
65 Vgl. Ancora nell'età dell'ordinamento centuriato, in: Atti del II. Congresso Nazionale di Studi
Romani, Roma, III, 1931, 41 ff. = Opuscula, 2, 287 ff., La storia dell'antichissimo esercito
romano e l'età dell'ordinamento centuriato, Athen 22, 1934, 57 ff. = Opuscula, 2, 293 ff.
66 Über das Echo dieser Studien vgl. J. Heurgon (o. Anm. 55) 254. Vgl. E.S. Staveley, The
Constitution of the Roman Republic, l, The origin and early development of the centuriate
organisation, Historia 5, 1956, 74 ff.
67 Besonders interessant und bedeutend: Considerazioni sulla tradizione letteraria sulle origini della
repubblica, Les origines de la République romaine (o. Anm. 19) 135 ff. Vgl. noch: Miscellanea
Rostagni, 192 f., Synteleia Arangio-Ruiz, 1, 486 f., Rend. Ist. Lomb. 112, 1978, 250 ff., Athen,
N.S., 63, 1985, 265 ff., 65, 1987, 203 f.
68 Vgl. Studi su Dionigi da Alicarnasso 1, Athen, N.S. 38, 1960, 175 ff., 2, 39, 1961, 98 ff., 3, 40,
1964. 29 ff., La "Storia di Roma arcaica" di Dionigi di Alicarnasso, in ANRW II, 30, 1, 1982, 794
ff.
69 Vgl. Dagli ausiliari del "rex" ai magistrati della "respublica", Estr. dall'Athen, N.S., 30, 1952,
L'Italia antichissima e le origini di Roma, Nuove Questioni di Storia Antica, Milano 1968, 241 ff.,
Nomen Latinum, Pavia 1973.
70 Bologna, 1, 1970, 2, 1973. Vgl. Ders. Aspetti Culturali del Lazio primitivo, Firenze 1978,
Mycenaeans in Early Latium, Roma 1980.
71 Vgl. ausführlich o. Anm. 70.
374 ENDRE FERENCZY
72 Origini di Roma, 2, 107 ff.

73 Recherches sur la legende sabine des origines de Rome, Louvain 1967, Ders., Les Sabins aux
origines de Rome. Orientations et problèmes, in: ANRW I, 1, 1972, 48 ff., Ders., Les origines de
Rome, Tradition et histoire, Bruxelles 1985.
74 Op. cit., 176 ff.
75 Über dieses Problem: F. Bömer, Rom und Troia. Untersuchungen zur Frühgeschichte Roms,
Baden-Baden 1951, G.K. Galinsky, Aeneas, Sicily and Rome, Princeton 1981, G. Dury-Moyders,
Enée et Lavinium, Collection Latomus 174, Bruxelles 1981.
76 Poucet, Les origines de Rome, 306.
77 Poucet, op. cit., 164.
78 Vgl. Poucet, op. cit., 290 f. In der Sabinerfrage entwickelt Momigliano (Settimo Contributo...
407) keinen definitiven Standpunkt. Heurgon (o. Anm. 55), 91, tritt für die frühe Anwesenheit der
S abiner in Rom ein.
79 W. Kunkel, Bericht über neuere Arbeiten zur römischen Verfassungsgeschichte, Kleine Schriften,
Weimar 1974, 442. Die Meinung von E. Meyer (o. Anm. 32) 443: "Allerdings darf auch nicht
verschwiegen werden, daß Mommsen in dieser logischen Rekonstruktion des römischen Staatsrechts
gelegentlich zu weit gegangen ist und der Jurist verschiedentlich über den Historiker gesiegt hat."
80 W. Kunkel (o. Anm. 76), 441 ff.
81 Vgl. E. Ferenczy, Zur Verfassungsgeschichte der Frührepublik, Festschrift F. Altheim, 1, Berlin
1969, 13 ff., Ders., Lo stato romano fra la monarchia e la repubblica, Studi Biscardi, 3, Milano
1982, 101 ff.
82 H. Siber, Römisches Verfassungsrecht in geschichtlicher Entwicklung, Lahr 1952.
83 P. Frezza, Corso di storia del diritto romano 2 , Roma 1968.
84 Storia della costituzione romana2 1, Napoli 95, 118 f., 129 f.
85 Storia del diritto romano 6 , Napoli 1987, 69.
86 Arcana imperii, Roma, 3, 1948, 30 f., Primordia civitatis, Roma 1959, 511 ff.
87 U. Coli, Regnum, SDHI 17, 1951, 1 ff. = Scritti di diritto romano, 1, 1973, 321 ff.
88 Regnum, Scritti di diritto romano, 1, 462 ff.
89 Primordia civitatis, 580.
90 Storia del diritto romano, 1, Milano 1959, 76 ff.
91 Das römische Volk, Sein Staat u. sein Recht, Frankfurt a.M. 1955, 190.
92 Röm. Verfassungsrecht (o. Anm. 79), 24.
93 Storia della cost, rom, 1, (o. Anm. 84), 103.
94 Corso di storia del dir. rom. (o. Anm. 80, 80), 45.
95 Röm. Staat u. Staatsgedanke (o. Anm. 32), 21.
96 Vgl. A. Magdelain, Cinq jours épagomènes à Rom? REL 40, 1962, 201 ff., Ders., Auspicia ad
patres redeunt, Hommages J. Bayet, 1964, 449 ff., U.v. Lübtow, Die lex curiata de imperio, SZ
69, 1952, 154 ff., E. Ferenczy, Über das Interregnum, Festg. f. U.v. Lübtow, Berlin 1980,45 ff.
97 U. Coli, Scritti 1, (o. Anm. 83), 354 f.
98 W. Kunkel (o. Anm. 76), 465 f., Ders., Zum röm. Königtum, Kl. Schr., 352 f.
99 P. Catalano, Contributi allo studio del diritto augurale, 1, Torino 1960, 391 ff.
100 So schließt sich z. B. v. Lübtow (o. Anm. 87) in der Charakterisierung des Königtums dem
Standpunkt Altheims an. De Francisci folgt (Studi Arangio-Ruiz, 1, Napoli 1953, 1 ff., Primordia
civitatis [o. Anm. 82.] 668 ff.) im Fall des Problems der Tribus- und Zenturienordnung der früher
(o. Anm. 63) erwähnten Theorie des PI. Fraccaro, die für die Authentizität der Angaben der
Tradition eintritt. Vgl. E. Ferenczy, From the Patrician State to the Patricio-Plebeian State,
Amsterdam-Budapest 1976,15ff.
101 Zürich und München4 1975.
102 I. Napoli 1972, vgl. Ders., Storia economica di Roma antica, 1-2, Firenze 1979, Ders., Nuovi
studi di economia e diritto romano, Roma 1988.
103 Vgl. W. Kunkel, Kl. Schr. 450 ff. Meine Auffassung über den 1. Bd. von De Martinos Werk
weicht wesentlich von der Frezzas (SDHI 18,1952, 279 ff.) ab.
104 Storia del diritto romano 3 , 1-3 vol., 1943-1944, Arcana imperii 1-3, vol., 1947-1948 [Neudr.
1970], Sintesi storica del diritto romano 4 , Roma 1969.
105 Primordia civitatis, Roma 1959.
106 W. Kunkel, Kl. Schr., 463 ff., E. Meyer (o. Anm. 45) 465, 470 ff.
107 Napoli7 1987.
108 Vgl. J.-Cl. Richard, Les origines de la plèbe romaine. Essai sur la formation du dualisme patricio-
plébéien, Rome 1978.
109 Lineamenti di storia del diritto romano, sotto la direzione di Mario Talamanca, Milano 1979, 3 ff.
110 Vgl. A. Biscardi, "Auctoritas patrum", Napoli 1987.
SUMMARY
Up to the beginning of the 20th century research into the Roman kingship was characterised by the
philological-historical method, which was displaced at the turn of the century by a radically hypercritical
approach. Exponents of this line have called into question the credibility of the Roman tradition as a
witness to the oldest periods. At the beginning of the present century archaeological investigation
acheived important successes which promoted the emergence of a conservative approach. Adherents of this
trend have pointed out the weaknesses of the hypercritical method and, with the help of adjacent
disciplines such as comparative historical linguistics and comparative religion, have successfully defended
the tradition's credibility. Since the fifties the renowned Swedish archaeologist E. Gjerstad has brought
about a significant change in the evaluation of the Roman monarchy. He was able to reconstruct Rome's
early history on the basis of archaeological discoveries, while attributing to the literary tradition the role
of a mere supplementary source. Even though Gjerstad's findings are rejected by many scholars, his
appeal for the recognition of archaeology as a primary historical source seems to be substantiated by the
tendency of the conservative Italian school of Roman history to see a confirmation of the tradition's
credibility in the results of excavation.
INDEX OF AUTHORS
INDEX OF AUTHORS
This index is a selective one, containing only the authors treated or cited in the text of the
individual essays, names in the footnotes and references, however, are not included.
A.
Alföldi, A. 360-363, 365
Altheim, F. 357
Ascoli, G. I. 35, 38, 240, 340-341, 350
.
Badía  Margarit, A. 329
Bastardas Parera, J. 329
Battaglia, 1.108-109
Beloch, . J. 356-357
Bengston, H. 364
Benveniste, E. 329
Binder, J. 356
Blumenthal, A. von 250
Boehmer, E. 119
Boisacq, É. 118
Bonfante, G. 238, 239
Bonnet, M. 322
Bossong, G. 301
380 INDEX OF AUTHORS
Bottiglioni, G. 257
Bréal, M. 213, 217, 223
Brugmann, K. 67, 79-80
Buck, . D. 89
Budaeus, G. 118-119
Budé see Budaeus, G.
Bundy, M. W. 116
Burkert,W. 188
.
Calvo, A. G. 211-215, 217-219, 221-222
Campanile, E. 238
Capogrossi Colognesi, L. 369
Catalano, P. 367
Chadwick, J. 87
Chantraine, P. 79-80, 100-101, 107, 114-115, 120
Christ, K. 364
Coarelli, M. 360
Cole, T. 158
Coli, U. 366-367
Colonna, . 360
Comrie, B. 303
Condillac, E. . de 299
Contini, G. 117
Conway, R. S. 242
Corssen, W. 269
Cowgill, W. 68
Cristofani, M. 239, 360
Curtius, G. 114
D.
Debrunner, A. 47, 51
De Francisci, P. 366, 368
Della Corte, F. 230, 235
De Martino, F. 366-368
De Sanctis, G. 356-357
Devoto, G. 119, 229-231, 235, 237, 268, 282-283
Diehl,E. 319
Diez, F. 118
Domseiff, F. 103
Dumézil, G. 220-221, 361-362
Durante, M. 229,282
E.
Einstein, A. 120
Eisenhut, W. 221
Ernout, A. 229, 245, 247-248, 252, 270, 280, 285
F.
Fraccaro, PL 364, 369
Fränkel, E. 62, 79, 114
Frezza, P. 366-367
Frisk, H. 79, 100, 110
G.
Gabba, E. 364
Gabrici, E. 100
Gelsomino, R. 235
Giannelli, G. 362-363
Givón, T. 300
Gjerstad, E. 358-360, 354-365
Goeman, A. C. M. 74
Gordon, A. E. 211, 213-215, 218
Greenberg, J. H. 303, 305
Gressmann, H. 198
Grienberger, von 222
Guarino, A. 366, 368-369
H.
Hadzidakis, G. 29-32, 38
Haudricourt, A. 348
, . 188
Heger, . 309
Hermann, E. 81
Hesseling, D. . 118
Heubeck, . 138
Heurgon, J. 363
Heuss, A. 363
Hoenigswald, H. 68
Hoffmann, . 29-30, 32-33
Hofmann, J. . 280, 314
Husserl, E. 300
J.
Jakobson, R. 155
Juilland, A. J. 348
K.
Kalleris, J. 29-30, 33
Knös, B. 118
Kölver, U. 331
Kretschmer, P. 28, 30, 100, 240
Kühner, R. (Kühner-Gerth) 47-49, 52
Kunkel, W. 366-367
Kuryłowicz, J. 68
L.
Labbé, P. 118-119
Lascaris 118-119
Last, H. 358
Lee, K. H. 53
Lehmann, W. P. 303
Leopardi, G. 24
Lepschy, G. 115
Lindsay, W. M. 217
Lobel, E. 86
Lübtow, U. von 367
M.
McKenzie, R. 100
Manni Paraino, M. T. 101
Marinetti, A. 239
Martinet, A. 36, 348
Meillett, A. 67, 270, 280, 285
Menaggio, E. 111
Meneghello, L. 115
Mentz, F. 217, 219, 222
Meringer, R. 273
Meyer, E. 357, 367
Meyer-Lübke, W. 82, 117-119, 329, 340-341
Molinari, M. V. 279
Momigliano, A. 360, 362
Mommsen, Th. 355-356, 361, 366
Montesinos, J. F. 119
Morpurgo Davies, A. 85-88, 90-91
N.
Niebuhr, B. G. 355
Niese, . 355
Norden, E. 215, 217, 219-223
.
Olivier, J.-P. 127-130, 132
P.
Pais, E. 356
Pallottino, M. 360-365
Palmer, L. R. 221, 229, 244
Pareti, L. 357-358
Paribeni, R. 358
Pariente, E. 240-241
Parry, M. 175
Peruzzi, E. 233, 235, 247-248, 256, 280, 285, 364-365
Petronius, A. 117
Pighi, G. B. 219
Pisani, V. 35, 100-101, 214, 222, 232, 240-241, 268-280, 284
Planta, R. von 233-234, 245
Poccetti, P. 269
Pokorny, J. 114
Pope, M. 135
Porzio Gernia, M. L. 281
Poucet, J. 365-366
Prosdocimi, A. L. 239, 249, 281-283, 286
Q.
Quilici, L. 360
R.
Radke, G. 214, 218, 220, 221
Ramat, P. 284, 319
Rasmussen, J. E. 68
Reenen, P. Th. van 74
Reichenbach, H. 300
Rheinfelder, H. 342
Risch, E. 77
Ritschi, Fr. 211, 213, 215
Rix, H. 78
Rohlfs, G. 342, 349
S.
Sapir, E. 300
Schadewaldt, W. 140, 141
Schlerath, B. 68
Schlicher, J.J. 319
Schuchardt, H. 340
Schulze,W. 239,281
Schwegler, A. 355
Schwyzer, E. 41, 47, 51, 68, 79-81, 100-101
Seiler, H. 329
Siber, H. 366-367
Skeat, T. 99
Solmsen, F. 67, 80
Sommella Mura, A. 360
Specht, F. 114
Spitzer, L. 117-119
Strunk, K. 41-42
Stuart Jones, H. 100
Sylvius Ambianus, I. 119
Szantyr, A. 314
Szemerényi, O. 15, 35, 38, 47, 85, 113, 331
T.
Tagliavini, . 343
Terracini, . 282
Tesnière, L. 300, 303, 305
Thorn, R. 300,305,331
Tigay, J. H. 188
Torelli, M. 360
Trendelenburg, F. A. 60-61
Tsopakis, A. 29
V.
Väänänen, V. 222
Vaillant, A. 67
Valdés, J.de 119
Varchi, . 118
Vennemann, Th. 303
Verdenius, W. 50
Viano, M. 299-271, 273, 278, 281-282, 284
Vine, . 154
W.
Wackernagel, J. 64, 88-89
Walde, A. 280
Wartburg, W. von 82, 117, 341-342, 347, 350
Watkins, C. 68
Wenger, L. 356
West, 50, 188
Wilamowitz, U. 50
Windisch, E. 339

(Current Issues in Linguistic Theory 87) Bela Brogyanyi, Reiner Lipp - Historical Philology_ Greek, Latin, and Romance. Papers in honor of Oswald Szemerényi II-John Benjamins Publishing Company (1992)

Hochgeladen von

Dokumentinformationen

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

(Current Issues in Linguistic Theory 87) Bela Brogyanyi, Reiner Lipp - Historical Philology_ Greek, Latin, and Romance. Papers in honor of Oswald Szemerényi II-John Benjamins Publishing Company (1992)

Hochgeladen von

Copyright:

Verfügbare Formate

HISTORICAL PHILOLOGY: GREEK, LATIN, AND ROMANCE

AM001STERDAM STUDIES IN THE THEORY AND

Series IV - CURRENT ISSUES IN LINGUISTIC THEORY

Advisory Editorial Board

Henning Andersen (Los Angeles) ; Raimo Anttila (Los Angeles)

Bela Brogyanyi and Reiner Lipp (eds)

Historical Philology: Greek, Latin, and Romance

BELA BROGYANYI and REINER LIPP

JOHN BENJAMINS PUBLISHING COMPANY

Historical philology : Greek, Latin, and romance : papers in honor of Oswald

SEPTVAGINTA QVINQVE ANNOS NATO

Freiburg i. Br., September 21, 1991

Bela Brogyanyi and Reiner Lipp

II. GREEK LEXICOLOGY

IV. GREEK PHILOLOGY

V. ITALIC AND LATIN PHILOLOGY

VI. LATIN AND ROMANCE LANGUAGES

VIL ROMAN HISTORY

INDEX OF AUTHORS 377

exposiciones de este tipo encontramos colecciones de datos que, convenientemente cla­

Claro que añadimos ocasionalmente referencias a otros "dialectos", sobre todo a la

1. Tendencia a generalizar una norma en todo el Griego

G. (según los verbos) del todo y de la parte, la especialización, la norma, es secundaria,

2. Norma que surge en varios "dialectos"

Nótese que el uso asintáctico, tanto representativo como impresivo-expresivo, del

Otros "rebrotes" no son exactamente idénticos, pero sí comparables. Uno es el del G.

3. Norma que es propia de solo un dialecto

1. In both our German example and in the syntactically comparable phrase

complementary) we have just mentioned, are able to describe but no to explain. In

which may be translated "they ordered him to leave because,

. Here, ως introduces an absolute accusative which coincides neither with the

1 Greek syntax would have required here

1. The split theory,

Thus, the need to explain the appearance in Macedonian of word-forms like

nicht stimmlose harte, sondern stimmlose weiche Explosivlaute mit nachklingelndem

2. The theory of successive evolution

(i) Law of devoicing

i. e. we would have the form instead of the attested

3. A review of the two theories

(i) change to TA (devoicing)

"aucun auteur ou grammairien antérieur à lui [meaning Plutarch] n' en a fait la

Our reasons for supporting this position are the following:

The acceptance of such an assumption would lead to a disorder in the arrangement

spirantisation), a general phonological phenomenon in the evolution of Greek, Italic and

Therefore the change of TA >SS>VS may be seen as a twofold/threefold process of

What we have here is the creation of affricates (a combination of occlusive and

Therefore, the ancient Macedonian dialect is by no means a deviation from the

Buck, C D . 1955. The Greek Dialects. Chicago.

Schwyzer (1938:800) unterscheidet zwischen griechischen Imperativformen für 2.

Bartholomae, Christian. 1904. Altiranisches Wörterbuch. Berlin.

Having recently helped to edit a selection of Oswald Szemerényi's Scripta Minora, in

Odysseus contrasts Antinous' handsome appearance with his lack of sagacity:

A negative version of this expression is found in the speech of Heracles, contrasting

The moment recalled by the words is that at which Heracles went to

Then, a little later,

(Pylades suddenly appreciates Orestes' dangerous situation) (Or. 755)

(The messenger begins his pert remarks with an exclamation of suprise at

(The Chorus realize the true character of Medea) (Med. 1279-1280)5

(Bdelycleon asks who is clinging to the donkey) (Wasps 183-184)

(Dicaepolis asks who the devil this is)

We may compare the following comments on this observation:

fundamental importance in his metaphysics: This phrase, which is not

Existence is not simply essence but essence considered qualitatively and

creation, Trendelenburg draws up the following scheme:

exposiciones de este tipo encontramos colecciones de datos que, convenientemente cla

l'apertura a considerare una sorta di sovradeterminazione etimologica. Il termine incurio