Diathesis in the Semitic Languages: A Comparative Morphological Study 9004088180, 9789004088184

Book by Retso, Jan

224 54 13MB

English Pages 254 [272] Year 1997

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Diathesis in the Semitic Languages: A Comparative Morphological Study
 9004088180, 9789004088184

Table of contents :
Title Page
Copyright Page
Table of Contents
Abbreviations of terms
Preface
Introduction
Problem and method
Terminology
Transcription
I. THE APOPHONIC PASSIVE IN ARABIC
Introduction
The passive construction
Morphology
The inner or apophonic passive in Arabic
The drag-chain argument
The push-chain argument
The simplification argument
Conclusions
II. THE APOPHONIC PASSIVE MARKER IN SEMITIC
General survey
The original extension of the apophonic passive: arguments
Origin of the apophonic passive: arguments
The auxiliary argument
The qutal argument
The intransitive argument
Evaluation of the arguments
Summary
III. THE YUQTAL AS PASSIVE MARKER
Introduction
The yuqtal: general considerations
yuqtal-forms in Biblical Hebrew
Supposed yuqtal-forms in Ugaritic
The Amarna letters
The yuqtal-forms in BH, Ugarit and Byblos: Conclusions
IV. THE SEMITIC CAUSATIVE CONJUGATION
Introduction
The Semitic causative conjugation: Preliminary
The non-augmented imperfect of the CCj
The elision hypothesis: Hebrew
The elision hypothesis: Aramaic
The elision hypothesis: Geʿez
The elision hypothesis: Arabic
The haplology argument
The non-elision argument
yaqtil in Amoritic
yaqtil and the CCj: Preliminary conclusion
The maqtil participle
The differentiation of the CCj imperfect: the Hebrew hifʿil
Differentiation of the CCj: attentuation
I w/y verbs
Northwest Semitic
Geʿez
The yqtl/ʾqtl in Ugaritic
The CCj imperfect: summary
The morphological origin of the CCj perfect in Semitic
The unitary origin hypothesis
The separate origin hypothesis I: š/s versus h/ʾ
The separate origin hypothesis II: š- versus h- versus ʾ
The ʾ-affix
The š/s-affix
The h-affix
Conclusions on the augmented CCj forms
The CCj in Semitic: a diachronic synthesis
Excursus I: the CCj in South Arabian
Excursus II: Iw/y verbs in South Arabian
Excursus III: the CCj in Phoenician
V. THE CAUSATIVE CONJUGATION IN ARABIC
Introduction
The CCj in Vernacular Arabic
Thematic CCj in VA
The s-augment in VA
The North Arabian A-B dialects
Irregular distribution of the a-augment
Excursus: Iw/y imperfects in VA
The CCj in VA/HA: explanations
The CCj in Middle Arabic
Pre-MA evidence
The function of the qatal/ʾaqtal in the ʿArabiyya
The yaqtil/yuqtil form in Modern Standard Arabic
The CCj and the connection HA-VA
The Arabic CCj in a Semitic perspective
The st-stem
Excursus: the verbs 'give' and 'come' in VA
VI. THE IMPERFECT PASSIVE MARKER OF THE G-STEM IN SEMITIC
Introduction
The reemployment of yuqtal
The connection yuqtal-yaqtil
The morphology of the imperfect PM in Semitic in general
The Gt stem as PM
The nG and the yuqtal PM
The yuqtal and the CCj
The tG PM
The ttG stem as PM
The distribution of the nG and tG/ttG and its background
The t-forms
The n-forms
The passive construction and the subjectless construction
The ultimate origin of the yuqtal/yiqtal imperfect
The i and u prefix vowels
The yuqtal/yiqtal as PM in VA: solution
Transformations of the yuqtal/yiqtal forms
Conclusions
Excursus: Frequencies of stems
VII. THE OTHER PASSIVE MARKERS IN SEMITIC
The PM in the perfect
The qatīl versus qutal/qutil
qutal/quttal
qutal conjugations in other Semitic languages: fʿāl
The qutal as PM: summary
The qutil/huqtil as PM: survey
The origin of the qutil/huqtil pattern
The treatment of the u-u- sequence in Semitic
qīl
qūl
qitil
qitil forms in Hebrew
yqattal/mqattal
The origin of the yqattal imperfect
VIII. SUMMARY AND CONCLUSIONS
Summary
Concluding remarks
Appendix I
Appendix II
Bibliography
General index
Index of forms

Citation preview

DIATHESIS IN THE SEMITIC LANGVAGES

STUDIES IN SEMITIC LANGVAGES AND LINGUISTICS EDITED BY

J. H . HOSPERS Professor of Semitic Languages and Literature and Archeology of the Near East in the University of Groningen

XIV ]AN RETSÖ

DIATHESIS IN THE SEMITIC LANGUAGES

DIATHESIS IN THE SEMITIC LANGVAGES A COMPARATIVE MORPHOLOGICAL STUDY

BY

JAN RETSÖ

E.J. BRILL LEIDEN • NEW YORK • K0BENHAVN • KÖLN 1989

Printed with grants-in-aid from Swedish Council for Research in the Humanities and Social Seiences Kungl. och H vitfeldtska Stipendieinrättningen

Library of Congress Cataloging-in-Publication Data Retsö, Jan. Diathesis in the Semitic languages: a comparative morphological study I by Jan Retsö. cm.-(Studies in Semitic languages and linguistics, ISSN p. 0081-8461; 14) Bibliography: p. Includes index. ISBN 90-04-08818-0 1. Semitic languages-Voice. 2. Arabic language-Voice. I. Tide. II. Series. PJ3041.R48 1988 88-39786 492-dc19 CIP

ISSN 0081-8461 ISBN 90 04 08818 0

© Copyright 1989 by E. J. Brill, Leiden, The Netherlands All rights reserved. No part of this book may be reproduced or translated in any form, by print, photoprint, microfilm, microjiche or any other means without written permission from the publisher PRINTED IN THE NETHERLANDS BY E.

J.

BRILL

CONTENTS Abbreviations of terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preface ........................................... .............................. lntroduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Problem and method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Terminology ......... ................. .................................... Transcription . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I.

XI XIII XIII

xv XVI

THE APOPHONIC pASSIVE IN ARABIC

lntroduction The passive construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The inner or apophonic passive in Arabic . . . . . . . . . . . . . . . . . . . . . . . . . . . . The drag-chain argument . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The push-chain argument .. . . . .. . . . . . .. . . . . . .. . . . . . . . . .. . . . . . . . . . . . . . . . The simplification argument . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

II.

IX

1 1 4 6 9 15 17 18

THE APOPHONIC PASSIVE MARKER IN SEMITIC

General survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The original extension of the apophonic passive: arguments . . . . . Origin of the apophonic passive: arguments . . . . . . . . . . . . . . . . . . . . . . . . . The auxiliary argument . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The qutal argument . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The intransitive argument . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Evaluation of the arguments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary .....................................................................

111.

20 20 23 23 24 24 26 28

THE YUQTAL AS PASSIVE MARKER

lntroduction The yuqtal: general considerations ...................................... . yuqtal-forms in Biblical Hebrew ......................................... . Supposed yuqtal-forms in U garitic ...................................... . The Amarna letters ............................................ The yuqtal-forms in BH, Ugarit and Byblos: Conclusi ns ....... . ••••••••••••

32 32 33 38 42

46

VI

CONTENTS IV. THE SEMITIC CAUSATIVE CONJUGATION

Introduction The Semitic causative conjugation: Preliminary .................... . The non-augmented imperfect of the CCj ........................... . The elision hypothesis: Hebrew ..................................... . The elision hypothesis: Aramaie ..................................... . The elision hypothesis: Gecez ........................................ . The elision hypothesis: Arabic ....................................... . The haplology argument .............................................. . The non-elision argument ............................................. . yaqtil in Amoritic ........................................................... . yaqtil and the CCj: Preliminary conclusion .......................... . The maqtil participle ....................................................... . The differentiation of the CCj imperfect: the Hebrew hi:fil ..... . Differentiation of the CCj: attentuation .............................. . I wly verbs ................................................................... . Northwest Semitic ...................................................... . Gecez ....................................................................... . The yqtl/ 'qtl in U garitic ................................................... . The CCj imperfect: summary .......................................... . The morphological origin of the CCj perfect in Semitic .......... . The unitary origin hypothesis ........................................ . The separate origin hypothesis I: f/s versus hl' .................. . The separate origin hypothesis II: s- versus h- versus ' ....... . The '-affix ................................................................ . The .SIs-affix .............................................................. . The h-affix ................................................................ . Conclusions on the augmented CCj forms ........................... . The CCj in Semitic: a diachronic synthesis ......................... . Excursus I: the CCj in South Arabian ............................... . Excursus II: Iw6' verbs in South Arabian ........................... . Excursus III: the CCj in Phoenician .................................. .

49 52 53 53 56 57 59

60 60 62 66 68 71 71 74 74 76 78

79 80 80 81

83 83

84 85

86 86 90

92 93

V. THE CAUSATIVE CONJUGATION IN ARABIC Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The CCj in Vernacular Arabic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thematic CCj in VA ........ .............. ... ................ ..... ..... The s-augment in VA . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . The North Arabian A-B dialects . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . Irregular distribution of the a-augment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Excursus: Iwly imperfects in VA . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . .. . . .

95 95 105 106 106 108 109

CONTENTS

The CCj in VA/HA: explanations ....... ............... ................ The CCj in Middle Arabic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pre-MA evidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The function of the qatall'aqtal in the 'Arabiyya . . . . .. . . . . . . . . . . . . . . . The yaqtillyuqtil form in Modern Standard Arabic . . . . . . . . . . . . . . . . . . The CCj and the connection HA-VA.................................. The Arabic CCj in a Semitic perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The st-stem . .. . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . .. . . . . . . . . . . . . . . . . Excursus: the verbs 'give' and 'come' in VA . . . . . . . . . . . . . . . . . . . . . . . . VI.

VII

110 114 119 122 124 127 130 133 135

THE lMPERFECT pASSIVE MARKER OF THE G-STEM IN SEMITIC

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The reemployment of yuqtal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The connection yuqtal-jaqtil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The morphology of the imperfect PM in Semitic in general . . . . . The Gt stem as PM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The nG and the yuqtal PM ........ .................................... The yuqtal and the CCj . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The tG PM ................................................................ The ttG stem as PM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . The distribution of the nG and tG/ttG and its background . . . . . . . The t-forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The n-forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The passive construction and the subjectless construction . . . . . . The ultimate origin of the yuqtallyiqtal imperfect . . . . . . . . . . . . . . . . . . . . The i and u prefix vowels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The yuqtallyiqtal as PM in VA: solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transformations of the yuqtallyiqtal forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Excursus: Frequencies of stems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

VII.

139 139 140 141 142 143 145 148 150 153 153 154 155 157 159 160 160 163 163

THE ÜTHER pASSIVE MARKERS IN SEMITIC

The PM in the perfect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The qatzl versus qutallqutil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . qutal/ quttal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . qutal conjugations in other Semitic Ianguages: J'äl . . . . . . . . . . . . . . . . . . The qutal as PM: summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The qutillhuqtil as PM: survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The origin of the qutil/huqtil pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The treatment of the u-u- sequence in Semitic . . . . . . . . . . . . . . . . . . . . . . .

165 165 166 170 172 172 1 74 175

VIII

CONTENTS

qi:l . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . qül . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . qitil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . qitil forms in Hebrew . . . . . . . . . . . .. . . . . . . . . . . . . . . .. . . . .. . . . . . . . . . . . . . . . . . . . . yqattall mqattal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The origin of the yqattal imperfect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

VIII.

178 180 180 184 18 7 193

SuMMARY AND CoNCLUSIONS

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

195 196

Appendix I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . General index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Index of forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

206 223 236 250 253

ABBREVIATIONS OF TERMS AC AM ASA BAr BH CA CCj EC ES HA I Ar IE IH ]PAr

K MA MH MSA OAr PC PM aPM dPM

Q SA SC TAr VA WS

Active construction Active marker Ancient South Arabian Biblical Aramaie Biblical Hebrew Classical Arabic Causative conjugation Ergative construction East Syriac '' Hocharabisch'' Imperial Aramaie ("Reichsaramäisch") Indo-European Israeli Hebrew J ewish Palestinian Aramaie Ktiv Middle Arabic Mishnaic Hebrew Modern South Arabian Old Aramaie Passive construction Passive marker Apophonic passive marker Derived passive marker Qre Standard Arabic Subjectless construction Talmudic Aramaie V ernacular Arabic West Syriac

PREFACE This is the conclusion of a study of the diathesis in the Semitic languages which was initiated in 1981, when I started working on a thesis on the passive conjugations in modern Arabic dialects. In that work, The Finite Passive Voice in Modern Arabic Dialects (Göteborg 1983), which was basically a synchronic study, a detailed study of the diachronic aspects of this category was promised (p. 11 ). The present volume constitutes the fulfilment ofthat promise. Several persans besides the author have been involved in the production of this book. I first want to thank Professor Ebbe Egede Knudsen, University of Oslo, who kindly undertook to check the quotations from the Amarna letters and make the transliteration more up-to-date. I also want to thank Professor Tore J anson, Gothenburg U niversity, who read the two general introductions to the "passive" and "causative" respectively and affered valuable suggestions and improvements. My thanks are further extended to Daniel Ridings, Ferenc Tafferner and Kjell Andersson, all of the Classical Department in Gothenburg University, for their generosity in putting their computer skill at my disposal du ring many critical moments of the production. The staff of the Gothenburg University Library should also be included in my gratitude, having provided me with all the exotic titles that I needed during the work. I thank once more Jon van Leuven for making the English of the book more readable. Last but not least I thank Karin Hult, not only for her unlimited patience in typing out a preliminary version of the book, but also for her comments and advice both on the contents and the Iayout. Without her, this book would not have come into being. December 30, 1987

J.R.

INTRODUCTION Problem and method

The aim of this study is to sketch the history of two morphological categories of the verb in the Semitic languages, the "passive" and the "causative". It is thus diachronic and comparative. The reason for treating these two grammatical categories tagether will become evident in the course of the study. Since we are concerned with the history of certain language-specific morphemic categories, this study will be of interest mainly for Semitists. The intention is not to present or discuss any new theoretical approach to the categories in question. This is an activity pursued by theoretical linguistics, where these two categories are often discussed. The method chosen is not deductive, i.e. we do not start from a theory about the nature of the categories in question which we try to apply to the data. Instead we follow an inductive method. There are many good reasons for this, the most important being that a corpus of data is necessary before any patterns are discernible, and that such a corpus has not been available. The data have been gathered from descriptions of the various Semitic languages, in the form of monographs or articles as well as from dictionaries. With a few exceptions data from texts have not been used. This is a weakness, especially in the field of Arabic dialects, where some areas are documented only by texts. There are, however, good reasons to believe that this does not have any impact on the results arrived at. The inductive discovery procedure implies that we first try to survey the extension and use of certain morpho-syntactic categories from a more or less synchronic viewpoint. With this as a basis, we set up hypotheses about diachronic relations to other categories which are close to them either by use or by some similarity in form. By making a similar survey of these we can eventually verify the hypotheses, which may then serve as a foundation for new hypotheses. The sum of verified hypotheses will present a picture of the diachronic relationship-the historical development of the categories concerned. The disposition of this work reflects to a large extent the course of the investigation process. The results that emerged during the course of the work were not expected from the outset. It has been considered worthwhile to let the reader follow the path of argumentation, rather than present the results with supporting arguments according to a more schematic plan.

XIV

INTRODUCTION

It turnsout that the diachronic study of languages still basically has to follow traditional comparative methods established during the last century. Some of the fundamental concepts used by them, such as analogy and sound change, are still the basic tools for comparative linguistics. This of course does not imply that one has to adhere to the neogrammatic views on language in general, or that one should not adopt the results of structural and generative linguistics. But the traditional comparative methods have proved immensely fruitful and the groundwork in comparative linguistics still has to lean heavily on them. The method of survey and verification of hypotheses demands a certain breadth, and we have chosen to adduce }arge parts of the material used for this. The concentration is on lexical and paradigmatic systems, since it turns out that most of what has been written about the problems in question in the Semitic field is, with some exceptions, of limited value. We have judged it necessary to weary the reader with collections of material from dictionaries etc. These lists should, however, be closely studied, since they are the foundation on which the whole building of conclusions rests. An example which has been followed in this respect is J. Aro's study ofthe vocalisation ofthe G-stem ofthe Semitic verb, which combines gathering and accounting of basic data with non-technical but common-sense judgement. A rather non-technical apparatus of terminology has also been adopted here, since it will make the work more easily accessible to Semitists in general and at the same time readable to non-Semitists, including theoreticallinguists. The terminology is at the same time eclectic: there is no strict adherence to any special school of linguistic thought. Terms coined by different schools have been chosen rather freely with the sole purpose of making the phenomenon described or the concept used as clear as possible. The author is firmly convinced that a lot can be said about language by use of common sense without necessarily any equipment of deterrent terminology. The purpose is not to construct or verify a universal theory about the grammatical categories in question. This study thus does not fulfill Hjelmslev's criteria for a scientific treatment oflanguage. Yet it is hoped that it will join all the others which have increased our knowledge about human language(s) without fulfilling such criteria. In spite of what has been said, two theoretical sketches of what is understood by "passive" and "causative" are presented. They are necessary models for handling the data but do not claim to be either original or universal. Anyone who has worked with similar problems knows that all of them cannot be solved. Especially when investigating systems, as in this work,

INTRODUCTION

XV

doors are continuously being opened in many directions due to the systematic character of language: everything is connected. One of the exciting results of such an investigation is in fact the wealth of new questions arising. There are several of them here, and we have tried to single them out.

Terminology It is more difficult than usually realized to find proper designations for the different forms of Arabic. In accordance with modern German terminology we have chosen the term High Arabic (Hocharabisch), abbreviated HA, as a general term for the language of the Gähiliyya poets (the 'Arabiyya proper) and alllater developments of it which have preserved its basic characteristics. We thus discern (a) the 'Arabiyya of the old poetry and the Qur)än; (b) Classical Arabic (CA)= HA as standardized by the grammarians from al-Xalil onwards, as weil as the literary language of the Middle Ages acknowledging this norm; (c) Modern Standard Arabic (SA), the idiom employed today as a literary medium in the Arab world. Beside these, we also discern the form of HA with interference of different kinds from the vernaculars called Middle Arabic (MA) documented during most of the Islamic Middle Ages. For the spoken forms of Arabic today the term Neo-Arabic has been introduced and used mainly by German scholars. It will become apparent during the course of this study why we have not applied it here. We preliminarily suggest the term Vernacular Arabic (VA) as a working term, although it is not altogether adequate. As far as the different forms of Aramaie are concerned we distinguish Old Aramaie (OAr), Biblical Aramaie (BAr), Imperial Aramaie (IAr), andjewish Palestinian Aramaie QPAr). By Talmudic Aramaie (TAr) is meant the Aramaie of the Babylonian Talmud. The other Aramaie dialects and languages have their own special designations. Wehave used the term Turaic for the dialects of Tür 'Abdin, usually called Türöyo, this in analogy with the correspondence armoyo - Aramaic. 1 In Hebrew we distinguish three different forms: Biblical Hebrew (BH), Mishnaic Hebrew (MH) and Israeli Hebrew (IH). 1 Of the aneient Semitie languages the doeumentation of Aramaie is the one most diffieult to use. Most deseriptions tend to eonfuse different stages and dialeets. To this eomes that the deseriptions of the J ewish dialeets are mostly based on bad manuseripts from periods when the writers were no Ionger aware of the differenees e.g. between Palestinian and Babylonian dialeets. Some ofthese deficiencies arestill to a eertain extent found in Segert's reeent work on OAr/BAr/IAr. This explains why the Jewish Aramaie dialeets are not extensively referred to here. The Aramaie material is basieally drawn from OAr/IAr/BAr, Syriae and the doeumentation of the Neo-Aramaie languages.

XVI

INTRODUCTION

For the languages of South Arabia the abbreviations ASA for Ancient South Arabian and MSA for Modern South Arabian are used.

Transcription Due to continuing technical progress, it is no Ionger possible to have a book printed with different scripts, at least not with costs which can be managed by usual fund-raising. Such pieces of art as Vivian Ridler's pages in Drower-Macuch's Manclean dictionary are not yet within the grasp of computer technology. Therefore, when dealing with languages like Semitic one has to use transcriptions. In this study the common systems used for Semitic, as found in Moscati's lntroduction, will be followed. The principles followed for transcription of modern Arabic dialects have been outlined in Retsö, Passive p. 20. For the vowels of Ethiopic/Ge'ez we follow the system in Moscati, Introduction p. 53, i.e. without quantity marking. For Syriac we use a pure transliteration of the vocalized Serto script, i.e. we do not mark (hypothetical) quantity of vowels: zqopo is (o), pto~o is (a), ~bo~o is (i), 'e~o~o is (u) and rbo~o is (e). We thus do not mark swa or gemination, since they are not marked in the script. W e have also in general abstained from distinguishing between qusoyo and rukoko, since the begadkefat shift is floating in the Syriac reading tradition and is not marked consistently in the script. For the plene writings see the following paragraph on Hebrew. For Akkadian we use the system found in von Soden's Grundriss, i.e. a reconstruction of the phonological system of the language. When quoting texts, viz. the Amarna letters, we give a mixed transliteration/transcription including the determinatives of the original, based on Knudtzon's editon but brought up to date by checking the copies. In some cases the forms have been reconstructed from Knudtzon's transcription. The stumbling-block of all transcription systems is Hebrew. As far as BH is concerned we give a transliteration of the Tiberian vowel system. Opinions are, however, divided on how this should be clone. Since the pronunciation reflected in the Tiberian system can no Ionger be heard and analyzed, one should abstain from marking vowel length, distinguishing swa mobile and quiescens etc., since none of these is marked in the Tiberian script. A strict transliteration should also ignore the ( reconstructed) difference between dages forte and dages lene. Ideally, every vowel sign should have its Latin equivalent. A consonant with dagef should have one simple marking, and plene and defective writings should be distinguished (as in Syriac). One should also distinguish two )alefs, the

INTRODUCTION

XVII

pronounced one and the unpronounced one. Swa is not marked. Wehave thus marked dagef with a point after the consonant which has dagef in the Tiberian script, following the system found in Rosen's Contemporary Hebrew. Plene writing with he is marked with a line over the vowel; plene writings with yod or waw are marked with circumflex. The result may seem somewhat complicated at first glance, but is clear and practical for our purposes here. 2 A similar system is used for Syriac as well asfor BAr. For IH we ignore the Tiberian vowels and render the five phonemes of the Sefardic pronuciation as i, e, a, o, u indicating plene writing as for BH. The same system is followed for MH although the vowel system here was certainly different from the Sefardic tradition of BH. 2 This transliteration deviates on some points from the principles advocated by Richter (57 ff.). Thus, it is most doubtful whether 'alef without fwa should be seen as "Vokalanzeiger" in the consonantal text. It rather represents an etymological spelling. Even if it were a mater lectionis this function should be marked differently from its consonant value (op. cit., 29). From the point of view of pure transliteration which demands correspondence between two script systems only, there is no reason to have different markings of dagef (op. cit., 30-31). Swa is retrievable from the transliteration if it is not marked at all: it will then appear with all consonants followed by another consonant in the transliteration. As Richter points out, a transliteration system should be reversible (62). It may be added that it should also be as simple as possible.

CHAPTER ONE

THE APOPHONIC PASSIVE IN ARABIC lntroduction

The passive construction The passive construction (PC) has played a central role in modern linguistics since Chomsky's Syntactic structures in 1957, where it was adduced as a paradigmatic example of the explanatory power of transformational grammar. This kind of theoretical approach has, however, turned out to be difficult to adapt to the variegated group of constructions usually labelled "passive". A survey of these, as clone recently by Siewierska, shows that a clearcut, universal definition that combines all of them is problematical and perhaps impossible. 1 The constructions that will be dealt with here are basically two, corresponding to those traditionally called "personal" and "impersonal" passive. Both are often labelled "passive", but there are good reasons to see them as syntactically different constructions altogether, even if there are often morphological similarities between them. Both belang to a type of construction involving the relations between the verbal kernel of a sentence and its nominal elements. These relations may be analyzed both in semantic and morpho-syntactic terms. A covering term for the semantic relations and their morpho-syntactic representation is diathesis. The two constructions dealt with here are thus two types of diathesis. Later a third type, the causative, will be introduced (Chapter IV). The ''personal passive'' in the languages investigated here has the following characteristics: (a) The morpho-syntactic subject, i.e. the noun that governs the agreement of the verb, is its semantic object. This implies that the semantic features usually found with objects are found with the subject in a PC. (b) The "abnormal" semantic content of the morpho-syntactic subject is marked in the verb in different ways. 2 Siewierska, 255 ff. Siewierska, 75 ff. Fora discussion of the semantic content of the morpho-syntactic subject, see Saad, 16 ff. 1

2

2

THE APOPHONIC PASSIVE IN ARABIC

These two characteristics have one important consequence, viz. that a "personal passive" or passive construction (PC) should have an active counterpart with the subject of the PC appearing as object and another noun phrase as subject. The PC is mostly seen as a derivation from the active construction (AC). There are, however, divided opinions on the nature of this derivation. One problern is whether all PC are convertible to AC. Another is the role of the subject in the corresponding AC, which in the PC obviously does not appear in its usual position, but appears either as a prepositional or adverbial phrase or is absent altogether. A common explanation for this is that the appearance of the object in an unexpected position is the result of focusing or topicalization, i.e. the object is moved into focus position. 3 Another view is that the subject is deleted for different reasons, and the object is moved in to fill its place. 4 As far as Semitic is concerned, both these explanations raise problems. The focus/topic model implies that the PC is merely a reordering of constituents of a normal AC, which makes the absence of the agent ( = the subject in the AC) difficult to handle. All Semitic languages, including Arabic, allow mentioning of the agent in the PC, even if this is oflow frequency in some languages. A paradoxisthat the most ardent adherents ofthe topicalization explanation (mostly Arabists) at the same time go to lengths to deny the existence of agent extension. 5 On the other hand, the deletion explanation should have difficulties in accounting for the existence of agent extension: why should a phrase that is deleted in one part of a sentence reemerge in another? This notwithstanding, the deletion hypothesis has some advantages over the topicalization explanation, since the majority of the PC in Semitic do not have agent extension. 6 It also establishes a link with the "impersonal passive''. The "personal passive" is dependent on the existence of a semantic object. 7 The "impersonal passive" is not, and may be formed from all verbs. We have argued elsewhere that a more proper name for at least one type of "im personal" passive is "subjectless sentence". 8 We will here use the term subjectless construction (SC). The basic characteristic of this construction is the absence of a semantic subject in normal 3 Siewierska, 7 f. Cf. ibid. 41. This is a popular explanation among Arabists, cf. Retsö, Passive, 33-37; Talmoudi, Notes, 51 ff. 4 Siewierska, 6 f. 5 Cf. Talmoudi, Syntax, 50-51; Retsö, Passive, 25-28. 6 Cf. Retsö, Passive, 183 f.; Cf. Meillet, Caractere, 196. It is, of course, not correct to say that if the PC is nothing but a reversion of the AC it is superfluous. 7 In fact the possibility of passivization seems to be the proper definition of an object, cf. Siewierska, 15. 8 Thus Retsö, Sentences, 71-73.

THE APOPHONIC PASSIVE IN ARABIC

3

position as morphosyntactic subject. Same languages have a "dummy'' with no semantic content as morphosyntactic subject, others have no noun phrase at all. The verb morphology in an SC is often more variegated than in a PC. The absence of subject may be marked or unmarked in the verb. In Semitic, SC without dummy and with marking of the verb is rather common. 9 It may occur with both transitive and intransitive verbs, although with the former the PC is far more common. lt thus appears that with transitive verbs Semitic has two options, PC or SC. This is one of the main reasons that can be advanced for preferring the deletion explanation to topicalization as the trigger of the PC . 10 Both constructions are a device for handling the same problem: absence or non-specification of semantic subject in a sentence. The deletion argument does not, however, hold for all cases. In all classical Semitic languages the PC is employed not only when the semantic subject ( = the "agent") is unknown or unspecified. Contrary to what is usually claimed, it has a much wider use and, as was pointed out above, also allows of the introduction of so-called agent extension. 11 The Semitic PC thus resembles the one found in Western European languages. It may be argued that PC with explicit agent extension should be seen as a topicalization of the object. The PC, originally perhaps a device for handling the absence of agent, thus gives the languages a means for grammaticalization oftopicalization. Thus the PC in Semitic should probably be seen as two different constructions, ( 1) deletion or absence of the semantic subject ( = the "agent") for different reasons, (2) topicalization of the object. Both are diachronically related, but an the synchronic level they are of different types. They are easily discernible by the absence or presence of agent phrase. Their diachronic relationship is indicated by the common morphology of the verb in both. Both are also, for the same reason, diachronically related to the SC. Since this study is mainly concerned with the morphology of the verbs in these constructions we use the term PC for all three of them although the differentiation between them is important for the study of their historical origin and development. That agent extension is rather infrequent in Semitic must be due to the fact that many of these languages have other means for topicalization. One language Iacks most of these means, viz. Aramaic. Here, consequently, we encounter a far more extensive employment of PC with

9 1

Brockelmann, Grundriss II, 119 f.; Retsö, op. cit., passim. cf. Retsö, Passive, 22-24, 183-186. Brockelmann, Grundriss I, 537 f. This contrary to Meillet, op. cit.

° For Arabic,

11

4

THE APOPHONIC PASSIVE IN ARABIC

agent extension. 12 A possible development from such a stage is the rise of an ergative construction (EC) which, from one viewpoint, may be described as the obligatoriness of the PC with agent extension with transitive verbs. The pre-stage of an EC is also found in at least one modern Aramaie dialect, Turaic. 13 It seems that the emergence of EC takes place under certain circumstances, mainly adstratum influence. 14 Morphology The main concern of this study is the morphology of the finite verb in the PC, and implicitly also the SC. Of the different types of morphological marking the periphrastic one, found in the modern Western European languages, will not concern us here. Such a construction is found regularly in some modern Aramaie dialects only and represents a recent development. 15 Basically the morphological marking of the verbin a Semitic PC is of a synthetic type, being opposed to a similar synthetic form in the AC. The verb shows both directed derivative and non-directed correlative oppositions. The former implies that the verbin the PC is derived from the one in the AC by addition of an affix, in Semitic usually an n- or a t-prefix. In some cases the t may occur as an infix. Both morphologically and syntactically, such a verb is best described as a derivation from a more basic form. We have elsewhere introduced the term "augmented forms'' for these derivations in Semitic. 16 The non-directed correlative indicates that the opposition is constituted by variants of the same root, i.e. no morphological material is added but some elements are replaced by others. In Semitic as weil as in Indoeuropean it is mostly vowels which may alternate in this type of opposttlon. This alternation is usually called apophony, Ablaut. Rundgren has claimed the necessity for a distinction between apophony, i.e the morphological exploitation of phonetic processes in the past (like

12 This is an impression based on extensive reading of Syriac texts. Cf. e.g. the following passages in the story of the martyrs of Bet Slok (Brockelmann Grammatik) • 50:4, *52:19, *58:14-15, *60:14; 4 Ezra 5:10, 5:45, 6:45,46, 7:20,45,62 etc. Cf. also Nöldeke, Syrische Grammatik, 183-84. The matter deserves a thorough investigation. Fora historical interpretation of the l-construction as a possessive construction see Kutscher, Passive, (4)-(17). 13 Jastrow, Midin, 60-61, 72. 14 Turaic is spoken in an area where ergative languages of different families are found during millennia. 15 Cf. Jacobi, 158-162; Nöldeke, Grammatik, 288-90; Maclean, 89-90 (Urmian). Cf. Macuch, Handbook, 283 (Modern Mandaic); Sara, 72 ("C.haldean"). 16 For the terminology used here see Nedyalkov/Silnitzky.

THE APOPHONIC PASSIVE IN ARABIC

5

the Ablaut in Indoeuropean), and apothematism i.e. a similar morphological alternation without this background (like most cases of vowel alternation in Semitic)Y Since from a synchronic viewpoint the result of the two is identical, we still use the term apophony for the non-directed correlative opposition qutil/qatal etc., being well aware that diachronically this may be a reflex of other processes than those reflected in sing-sangsung etc. For synchronic description, the distinction is mostly unimportant since the alternation works in the same fashion regardless of its origin. We will later return to the term apothematism, which becomes helpful when dealing with certain diachronic aspects. It may, however, be stated already here that we agree with Rundgren that the Semitic "inner passive" yuqtallqutil etc. is basically an apothematic formation in his terms and not the result of phonological processes in the past. W e will thus distinguish between a term in a correlative opposition of an apophonic type used in PC, called apophonic passive marker ( aPM) and a term in a derivative opposition in the same function, called derived passive marker (dPM). The verbs functioning as aPM are thus derived syntactically but not morphologically: they are different forms of the same root employed in different contexts and constructions, but it cannot be said that the one is morphologically primary to the other. It should be remarked here that conversive paradigmatic oppositions in PC do not seem to occur in Semitic: there is no set of finite verb suffixes that mark active/passive and person at the same time, like Latin vocat-vocatur. But there are cases of syntactically conversive oppositions, i.e. when the context only is marking whether there is a PC or not. 18 There are, in other words, instances where there is no morphological marking in the verb, and such cases are often ambiguous. This leads to one of the central and, for the diachronic study, most important facts: all morphological devices in Semitic occurring in PC have other functions as well. This is in accordance with most languages possessing the PC, where there is a continuous overlapping between the "passive" verb (PM) and other categories. Since, as has been sketched above, the PC may have emerged as a means of handling absent subjects/agents, it may be assumed that the languages use, to this end, means already at hand. If the PC is basically a means for handling absent or unspecified subjects/agents, it doubtless had and still has an improvisatory character. This would explain the continuous remodeHing and innovation of the 17 18

Rundgren, Ablaut, 49, 51, 57. Cf. Retsö, Sentences, 75 (type 2b2).

6

THE APOPHONIC PASSIVE IN ARABIC

morphology of the PM which can be observed during the history of the languages where it occurs. In Indoeuropean one can observe how the general medium conjugation (i.e. a paradigmatic conversive category) is used both for intransitive "reflexive" and in PC (e.g. Classical Greek). In Latin this conjugation seems either to have been put aside or perhaps never existed, and instead an ancient means for designating unspecified subjects/agents by a suffix rwas extended to the verbin PC. 19 We can thus see here that two devices, intransitive marking and subjectless marking, were extended to the PC. In Semitic at least intransitive marking is widely spread as a means for marking passive subjects (derivative verbs with taffix). The reason for this is not difficult to understand: one characteristic of the verb in a PC is intransitivization (since the object is transformed into subject). 20 As will be shown later (pp. 154 ff.) there are some reasons to believe that one category of passive marking is in fact an old device for su bjectlessness. In Semitic there is another category that plays a crucial role in the PC, viz. the stative. The stative as a semantic category is widely represented in Semitic by finite verbs with correlative root oppositions to non-stative verbs. This morphological category is also employed in PC. I t is clear that the reason for this must be a semantic feature which the subject of a stative verb shares with the subject in a PC: non-agentivity. 21 Our purposein outlining the history ofthe PC in Semitic is to establish the varying relationships between these morphological categories during the observable (and hopefully also partly the non-observable) history of these languages. Since this study has grown out of a study of the PC in modern Arabic dialects, and since the morphology of the PM in Arabic as a whole can be seen as an analogy to the situation in Semitic as a whole, we will start with a survey of the Arabic evidence and try from this evidence to arrive at a proper formulation of the problem. From this formulation the further pursuit of the study will become apparent. The inner or apophonic passive in Arabic

One of the more conspicuous differences between HA and most forms of VA is the finite verb used in PC. While the form er marks a PC by means 19 Leumann/Hofmann/Szantyr I, 515-17; cf. Brugmann 11:3:2, 657-60. For recent explanations of the r-forms, see Szemerenyi, 224-225, 305-311. 2 ° Cf. Siewierska, 162 f., by her called "reflexive passive". 21 Siewierska, 75. By the termstativein this study is included not only verbs meaning 'to be something' but also 'to become something'. Both these meanings may be seen as two different aspects (in the linguistic sense) of a basic sememe with 'stative' content. Cf. Rundgren, Intensiv, 35.

THE APOPHONIC PASSIVE IN ARABIC

7

of a correlative apophonic opposition between two variants of the verb root, achieved by different vowels, the dialects instead use forms augmented with a prefixed n or t opposed to the active. The latter may also occur as an infix. A closer survey of available material has shown that the correlative type or the apophonic opposition is not as rare in spoken Arabic as has often been supposed. In certain areas, mostly with dialects of the badawi-type, the vowel opposition seems to be the normal form employed in PC. 22 Unfortunately, evidence from these areas is still scanty as weil as contradictory. Nevertheless the fact remains that the finite verb in a PC in most forms of VA morphologically differs considerably from the one found in the corresponding construction in HA. Even the dialects having the apophonic opposition do not have the same form in the PC as HA. 23 Already in the first modern description of an Arabic dialect, Spitta's grammar of Cairene (1880), this difference is pointed out and the author's view ofthe reasons for it is clear from his choice ofwords: " ... das Passivum [i.e. the classical apophonic forms ofthe finite verb] ist bis auf wenigen Spuren verschwunden . . . Die Reste des alten Passivs sind folgende Formen ... ". 24 Spitta' s standpoint is thus that the ancestor of the dialect spoken in Cairo did possess a passive system similar to or identical with the one found in HA, a system which has disappeared and been replaced by the system with augmented verbs. Spitta compared this to "den anderen semitischen Dialecten" where a similar situation obtains. According to Spitta it is doubtful if there are any survivals at all from the ancient system in Cairene. 25 In 1883 Nöldeke made a remark, rather en passant, which is worth quoting: in a discussion of the IV stem in Arabic and its absence in the modern dialects, he explains this as due to an "unbestimmte, trübe Aussprache der Vokale". And then, in a note, he continues: "Diese dürfte auch die Hauptursache sein, dass das alte, durch innere Vokalveränderungen gebildete Passiv beim Verb. fin. fast eben so verloren gegangen ist wie IV. Der Unterschied der Aussprache der Aktiv- und Passivform blieb eben nicht mehr deutlich genug. In alter Zeit war die Vokalisation wohl klarer ins Ohr gefallen. " 26 These two Statements, viz. that VA once had a passive system similar or identical to the one found in HA and that it has disappeared due to 22 Retsö, Passive, 151-152 (Hadramaut), 153-154 (Oman), 158, 162-163 (Northern Arabia), 130-130 (Southern Tunisia). 23 Retsö, Passive, 159-63. 24 Spitta, 193; my italics. 25 Spitta, loc. cit.; Retsö, Passive, 91 f. Tomiehe gives a couple ofpossible aPM forms from Cairo: ye~ram (96-97 =SC), golobt 110 • 26 Nöldeke, Beiträge, 36, note 4.

8

THE APOPHONIC PASSIVE IN ARABIC

phonological development, constitute a standard explanation of the difference between the two language forms which is repeated by most Arabists. There is almost universal consensus about the first statement among those Arabists, mostly dialectologists, who have commented upon the question. 27 The only scholar expressing doubt about the original general existence of the inner passive in Arabic is Vollers, who claims that it is not found as such in any other Semitic language and that its onetime supremacy in Arabic is doubtful.2 8 Even with him, however, a characteristic contradiction is seen in the wording. The "passive" (i.e. the apophonic form) is "eingeengt" by "reflexive" t- and n-stems; it has been "replaced" by intransitive Ja'ila and Ja'ala; he uses terms like "Verdrängung" and "Ersatz" of the inner passive; it is "im Schwinden"; the evidence ofthe early grammarians shows that "man im gewöhnlichen Leben diese Passive nicht mehr gebrauchte''. Nöldeke's phonological explanation of the reduction of the aPM is repeated and somewhat expanded by some modern scholars. In his study of the dialect in I:Iöran from 1946, J. Cantineau parallels the development of the intransitive/stative qatila > qetel with the PM qutila > qetel, i.e. coincidence between the intransitive perfects and the PM forms due to the reduction of short vowels in unstressed open syllables tagether with regressive assimilation of a in verbs of the qatila-type. 29 This coincidence would thus have been the reason for the introduction of other forms to replace the old aPM forms. This is what is called a drag-chain development: the rise of a vacuum in a system is filled by material being close to it structurally, in this case n- and t-augmented stems. This argument is repeated by Blau ( 1966) and Palva ( 1978). 30 A somewhat different view is hinted at in Cantineau's earlier study of the dialect of Palmyra ( 1934), where the elimination of the "passif' is seen as due to intrusion (concurrence) of the "reflexives" with n- or taugment which often have ''passive meaning' '. 31 This is obviously a different view from the one exposed fifteen years later in his I:Iöran study. The change in the passive system would here be the result of 27 Grand'Henry, Mziib, 86; Ph. Mar~ais, Djidjelli, 145-146; W. Mar~ais, Tlemcen, 89; Feghali, 198; Sasse, Mballam!ye, 258;Jastrow, Q:ltu, 187; id., Tikrlt, note 19; id., Struktur, 136; Borg, 75; Palva, ljesbiin, 35, note 60; Cantineau, Palmyre, 120; id., ljöran, 207; lngham, North East, 40; Blau, Christian Arabic, 150-51; id., Emergence, 105; Handbuch, 61, 262; Fischer, Zeugnisse, 84; Diem, t-Stiimme, 62, note 58. 28 Vollers, Volkssprache, 122 ff.; followed by Porath, 22. One should also mention those who only cite the absence of the aPM in VA: thus Blanc, Dialects, 97; Cowell, 234; Tomiche, 110. 29 Cantineau, ljöran, 207. 30 Blau, Christian Arabic, 64; Palva, ljesbiin, 35 note 60. 31 Cantineau, Palmyre, 120-21.

THE APOPHONIC PASSIVE IN ARABIC

9

semantic intrusion and not primarily phonological factors. This is in fact a push-chain development: one category disappears because another category close to it in the system begins to contend with it. The place of the first category is "overcrowded" and it is reduced or disappears altogether. A third explanation is formulated by Czapkiewicz. According to him, the verbal system in Arabic shows a pulsatory evolutional process consisting of reduction of morphological distinctions together with their regeneration. The difference between HA and VA, as far as the PM system is concerned, is a result of the reductional process, thus a simplification of the original system. The reductional process is governed by the increased amount of information, which is countered by the tendency to decreasing "costs" for transmitting it. The increase of distinctions leads to a reduction of the means for transmitting themY This explanation is accepted by J astrow. It has lately been repeated by Versteegh in connection with his adaptation of the pidgin/creole hypothesis for the origin of VA. 33 A few scholars have chosen not to make any diachronic Statements, and it is worth noting that the difference between the two formations of verbs in passive constructions are not listed by Ferguson as features of his Arabic Koine, even though he classifies the neutralization of vowel contrasts as belanging to the drift of Arabic and points to its consequences for the participal forms in the Maghrebinian dialects. 34 Since it appears that most scholars agree that the passive system in the dialects represents a development from a system similar to the one found in HA, we will start with a more detailed discussion of the explanations suggested for the change from one sy-stem to another. The drag-chain argument Nöldeke's explanation seems to imply that the disappearance of the "inner passive" created a gap which had to be filled. It is also based on the idea, typical for its time, that the "passive idea" had something to do with the "dark" u-sound. We will return to this argument later (p. 26). It is, however, obvious that the i- and a-phonemes have an equally weighty role in the morphology ofthe PM, and that neither in Arabic nor in Semitic is the passive characterized by "trübe" vowels. Further, if the "passive" vowel were unfit for functioning in the

32 33 34

Czapkiewicz, Accumulation, 15, 22-23 point c, 29. Versteegh, 83 ff., especially 89-90; Jastrow, Q]ltu, 143 f. Ferguson, 619.

10

THE APOPHONIC PASSIVE IN ARABIC

active-passive opposition in the finite verb, why was it preserved in the participles of the derived stems? In most Arabic dialects, the opposition is marked by different vowels in the participles of the derived stems. The exception is the sedentary dialects in the Maghreb, which will be returned to in due course. This is a remarkable fact which must be drawn into the discussion, and the conspicuous difference between the G-stem and the derived stems must somehow be accounted for: many dialects employ "inner" vowel contrasts as a means of marking passive constructions in some verbs, but only a few on the Peninsula may do so even in the derived stems (mostly the L- and D-stem). But the vowel opposition is regularly found in participles of the derived stems in most dialects ( even in those which do not use the vowel opposition in the Gstem as a passive marker), whereas this morphemic device is never used in the finite verb. A survey of the morphology of the participles of the D-, L-, tD- and tL-stems shows that the dialects may be divided into three groups: 35 A) In this dialectal group, the opposition ila in the last syllable of the participle is a means of distinguishing between "active" and "passive" participles of all these derived stems, i.e. a phonematic distinction is employed for morphological purposes. This system is found in the SyroPalestinian dialects (including I:Iöran). 36 It also seems to exist in Mekka, and in the Muslim dialect of Baghdad. 37 In the Mesopotamian qJltu dialects, there are some cases of a-forms of the participle of the L-stem: mbärak 'blessed', msewa 'made straight', m.telab 'taken out', m'ewan 'helped'. 38 The opposition is, on the whole, far less developed in the Lstem than in the DQR-stems, and is in some cases limited to IIIw,Y stems. The vowel opposition is maintained in the augmented forms (tD ... tL) in at least some of the Syro-Palestinian dialects, and incidentally in some lexemes in Baghdad M of tD. 39 B) In this group, the vowel opposition is maintained in the DQR-stems but not in the others (Land t-augmented), which usually have i (*mqattil) as theme vowel. Here belong some qJ!tu dialects (M}:lallamiye), Cairene, the Arabic in the Sä.Q.el area described by Roth (Abbeche) and Reichmut (Sukriyya), further Kfar 'Abida, Palmyra and in the Maghreb, 35 For the vocalization of the D-stem, see the more detailed survey in Chapter VII, pp. 187 ff. 36 Sabuni, 92-94; Cowell, 134-35; El-Haije, 73-74, 83, 84; Cantineau, /Joran, 248, 253, 255. 37 Schreiber, 37; Erwin, 221, 223-224, 230; Qafisheh, 136 ff., 145. 38 Vocke/Waldner, s.v. 39 Erwin, 226; Cowell, 134-135.

THE APOPHONIC PASSIVE IN ARABIC

11

Benghazi, Süsa, Takrouna. 40 The dialect of Gabes also belongs here, judging from the wordlist compiled by Mar~ais and Guiga, e.g. mbaddid 'fixant un Iimite', mbaddad 'Iimite'; mba$$3! 'obtenant', mba$$al 'obtenu'; mbayyir 'inquietant', mbayyar 'inquiete' etc. The opposition is weakly developed in this dialect since it is neutralized when a suffix is attached: mbayyrtn, mbaddidah etc., a trait which is widely spread. C) In the last group, there is no vowel opposition in the participle with morphemic function. Here belong most ofthe North African dialects like sedentary Moroccan (Rabat, Ouargha), Tunis, Maltese, Cherchell and the beduin dialects of Arba-' and Bou Saada in Algeria. 41 There are thus, in these dialects, several D-participles with double meaning, e.g. mxer;lr;ler 'having garnished' /' having been garnished', mteriem 'having translated' I 'having been translated' etc. 42 A closer look at the documented forms seems to show that the i-forms are underlying the present ones. Thus, participles from IIIw/y roots usually end in i notwithstanding their meaning in Moroccan, while other roots may have had a. In Cyprus the documented D-participles tend to have "passive meaning" which, however, according to the investigator, is the result of fusion of the i-a contrast. 43 It is evident from this that vowel opposition is, in fact, widely employed in the Arabic dialects in order to make distinctions in certain types of adjectives/participles similar to the opposition AM-PM in the finite verb. The occurrence of this distinction is especially noteworthy in Cairene and qaltu which in general do not employ this kind of morphological marking at all in the finite verb in passive constructions. As far as Cantineau' s drag-chain explanation is concerned, certain objections can be made. The phonological process that is said to have neutralized the cantrast between the "passive" and the intransitive stative should, in fact, have preserved the "active-passive" opposition in several verb types. Classical weak verbs like qlla/yuqiilu,yubnii and derived stems like kütiba etc. would not have been affected at all by the phonological development which is supposed to affect short vowels in unstressed syllables. Even if we suppose the existence of original forms like quttil, tuwu.ffiy, ustuqtil it is difficult to see why the resulting forms

40 Sasse, M~allamiye, 183 f.; Spitta, 211 f.; Tomiche, 130 f.; Roth, Abbecht, 24, 31; Reichmut, 252, 257; Cantineau, Palmyre, 148, 150; Feghali, 168, 199; Panetta, 190, 195; Owens, 119, 121; Talmoudi, Süsa, 98-100. " Harre!!, 57 ff.; Ph. Marc;:ais, Esquisse, 82; D. Cohen, Tunis, 115, 119; Grand'Henry, Cherchell, 57 f.; Singer, 410 f.; Levi-Provenc;:al, 27-28; Dhina, 327-28; Ph. Marc;:ais, Bou Sadda, 66, 69, 74. 42 Harreil, 58. 43 Harrell, loc. cit.; Borg, 104; cf. also Maltese: Schabert, 146 ff., especially 148.

12

THE APOPHONIC PASSIVE IN ARABIC

qittil, (z)twijfi (z)stiqtil etc. would not have been able to continue to carry the functionalload of marking a passive subject, and the more so since, in these derived stems, there is no coincidence with intransitive-stative verbs. This is also what they do in some dialects on the Peninsula. If we compare this with the cases of inner passive actually documented in the dialects outside the Peninsula, we get the following picture based on the material in Retsö, Passive (FPV), which can now be supplemented by Reichmut from the Sudan: Table 1 Root

Meaning as PM

Area

Type

Reference

,xd

'be taken'

)kl

'be 'be 'be 'be 'be 'be 'be 'be 'be 'be 'be 'be 'be 'be

eaten' drowned' discharged' left over' afflicted' worn out' completed' suspected' laid waste' gnawed'· put, made' cheated' informed' destroyed'

'be (by 'be 'be 'be 'be

possessed spiri ts )' hidden' emptied' robbed' suffocated'

Maräzlg, Gabes, Takrouna Gabes, Takrouna Takrouna Takrouna Gabes Syria Syria Takrouna Marazig Gabes Alg. Sahara Syria Alg. Sahara Gabes Syria (Tvakrouna) Sudan (Sukriyya) Alg. Sahara

bed. sed. bed.,sed. sed. sed. bed. sed. sed. sed. bed. bed. bed. sed. bed. bed. sed. bed. bed.

FPV 131 FPV 105 FPV 133, 105 FPV 103 FPV 103 FPV 132 FPV69 FPV69 FPV 103 FPV 131 FPV 131 FPV 135 FPV69 FPV 135 FPV 132 FPV 69 (103) Reichmut 21 7 FPV 135

Takrouna Takrouna, Tunis Maräzlg Alg. Sa~ara Sudan (Sukriyya) Syria Syria, Maräzlg, Alg. Sahara Chad Syria, Takrouna Djidjelli, Fez, Wädi Sous Sudan (Sukriyya) Alg. Sahara

sed. sed. bed. bed. bed. sed. sed. bed.

FPV 103 FPV 103, 104 FPV 131 FPV 135 Reichmut 217 FPV69 FPV 69, 131, 135

bed. sed. bed. bed.

FPV97 FPV 69, 103, 123, 118 139 Reichmut 217 FPV 135

bed. sed. bed. sed. bed. sed. bed.

Reichmut 217 FPV 69, 131 FPV 103 FPV 135 FPV70 FPV 132

b~r

bry bgy bly tif tmm thm idb ird g'l xd' xbr xrb

xtf xjy xly xnb xnq xzy xlJ xlb xlq

~r.J. ~rq ~rm

~Jl ~km ~lq

dny

'be humiliated' 'be finished, be paid' 'be exhausted' 'be created, born' 'be worked' (earth) 'be burned' 'be 'be 'be 'be 'be

forbidden' received' arrested' shaved' infected'

Sudan (Sukriyya) Syria, Maräzlg Takrouna Alg. Sahara Syria Gabes

13

THE APOPHONIC PASSIVE IN ARABIC

dbb dhb

rf

rwy sbt s~r

s~q

srq slm

'be 'be 'be 'be 'be 'be 'be 'be 'be

killed' wasted' lifted up' irrigated' proved' bewitched' eliminated' stolen' saved'

ify 'dm '.tb 'mr glq glb gwy jdy fqd

'be cured' 'be spread' 'be chosen (as head of family)' 'be cru cifi ed' 'be hit, beaten' 'be lighted' 'be hit by a glance' 'be extinguished' 'be bereft or 'be hit by evil' 'be filled' 'be closed' 'be defeated' 'be seduced' 'be made free' 'be lost'

jhm qbrj

'be understood' 'be arrested'

q.t' qtl qdy qlb ksr

'be 'be 'be 'be 'be

ktb krm km!

'be written' 'be honored' 'be finished'

knz

'be hidden' 'be stung' 'be hit by eye-disease' 'be sharpened' 'be helped' 'be ruined' 'be pierced' 'be worn out, exhausted' 'be exhausted'

fjy fn' ~fy

#b rjrb rjwy .trf

ldg

l.tm mrjy ngd hby htk hfy hlk

mutilated' killed' destroyed' defeated' broken'

Syria Syria Syria Syria Syria Maräzlg Maräzlg Chad Syria Cairo Syria Zaer Gabes

bed. sed. sed. sed. sed. bed. bed. bed. sed. sed. sed. bed. bed.

FPV FPV69 FPV69 FPV69 FPV69 FPV 131 FPV 131 FPV97 Bergsträsser Eider 70:8 FPV69 FPV 139 FPV 132

Syria Syria Susa Gabes

sed. sed. sed. bed.

FPV69 FPV69 FPV 104 FPV 131

Tunis Syria Alg. Sahara Fez Tunis Syria Syria Gabes Syria Sudan (Sukriyya)

sed. sed. bed. sed. sed. sed. sed. bed. sed.

FPV 104 FPV69 FPV 135 FPV 112 Stumme 32:30 FPV69 FPV69 FPV 132 FPV69

bed. bed. sed.

Reichmut 21 7 FPV 135 FPV 118, 135

Alg. Sahara Ouargha Alg. Sahara Syria Syria Alg. Sahara Alg. Sahara Gabes Alg. Sahara Maräzlg Syria Takrofma Morocco Tripoli (Libya) Wadi Sous Syria Maräzlg Gabes

sed. sed. sed. (?) sed. (?) bed.

FPV69 FPV69 FPV 135 FPV 135 FPV 132, 135

bed. sed. sed. sed. sed.? bed. sed. bed. bed.

FPV 131 FPV69 FPV 103 112 112 139 FPV69 FPV 131 FPV 132

Gabes Maräzlg Maräzlg Maräzlg Gabes

bed. bed. bed. bed. bed.

FPV FPV FPV FPV FPV

Syria

sed.

FPV69

132 131 131 131 132

14

THE APOPHONIC PASSIVE IN ARABIC

Of the 75 items listed, 55 are strong roots, and the correlative opposition between AM and PM is represented by short vowels of different quality. No cases with aPM of D or other derived or augmented stems are found in these dialects outside the Peninsula. Even if the opposition in a few cases is not definitely documented (the same form probably being used both as AM and PM), the picture ernerging from the areas in question does not show the aPM as phonetic survivals. It is clear that the different vowel systems in the spoken forms compared to HA does not present any obstacle to using the apophonic opposition for morphosyntactic purposes. Especially noteworthy are the cases from North African sedentary dialects with their "ruined" vocalism as compared to both HA and other dialects. This survey of "survivals" in the main modern dialects shows that their distribution does not coincide with what would be expected from the explanation suggested by Cantineau. In the G-stem where the neutralization should have taken place and where the preconditions for the replacement of the old form by new ones would have been much more favorable than with the derived stems, we find numerous cases of employment of the intransitive-stative yiqtal-qitil in PC. In the derived stems, on the contrary, where according to Cantineau no such neutralization would have happened and where we would thus expect to find the old system more or less intact, no "survivals" are found at all. The situation is thus as follows: most modern forms of VA have a morphological opposition in the G-stem between two basic forms (in some weak roots even three), one of which is mostly found with fientic verbs of action, transitive or intransitive, and the other with stative and mostly intransitive verbs, thus yiqtullyiqtil-qatal versus yiqtal-qitil. The latter category may occasionally be employed in PC. In certain dialects on the Peninsula, this use seems tobemorefrequent than elsewhere. In these a similar cantrast is found with the derived stems. All dialects except the ones in the Maghreb have an opposition of this kind in the participles of the derived stems. There is another serious objection to Cantineau's explanation: the coincidence of PM with the old intransitive would not necessarily have created a gap at all; the existence of a general intransitive form (=medium) employed both in PC and other intransitive constructions, like "reflexives" of different kinds etc., is weil known from many languages. Suffice it to refer to the medium in Classical Greek or the "reflexive" in Slavonic. More conspicuous is that this is also what is found with derived stems in most VA. The t-augmented forms of these stems are used both in PC and as intransitives in general. If this system works with derived stems, why could it not have worked with the G-stem?

THE APOPHONIC PASSIVE IN ARABIC

15

The push-chain argument I t is still possible that the coincidence between the PM and the general intransitive-stative could have triggered off a push-chain development, even if it remains to be explained why this did not happen in the derived stems. This would mean that a new form was introduced in order to mark a passive subject which was no Ionger distinguishable, as it had been when the yaqtullyaqtillyaqtal-yuqtal/ Iqatal/qatil-qutil opposition was working. This may be called a "concurrence" in Cantineau's terms, the intrusion of a new and more distinct form into a new domain. Such a process has been described as a mending of distinctions that have become blurred by phonological or other types of development, or have disappeared altogether. 44 In this case, the question is which form would be employed as the new PM. N ew forms are not created ex nihilo; morphological material already existing is reemployed for new functions. Aprerequisite is, however, that their old functions are close to the ones they are to replace and that they are morphologically transparent, i.e. they must be analyzable by the speakers so that they can become productive categories. 45 A good example is, in fact, the passive systems of the Western European languages. In both Romance and Germanic, the ancient PM have been replaced by new periphrastic constructions. In these cases one could well say that the old forms have disappeared due to "concurrence" from the new analytic constructions. This mechanism of "concurrence", i.e. innovation of new morphological categories, has been studied by Meillet in a dassie essay. But there is one fact that stands out clearly: the new forms are originally separate words which are grammaticalized. Because of the need for more expressivity and distinctiveness, innovations tend to be analytic. Apart from the passive in Western European languages, one could adduce the future tense in Romance and the Greek perfect as other examples of this. 46 As far as Arabic is concerned, there are indications of the existence of an analytical PM at least from northern Egypt, consisting of an auxiliary rii/:t or baqa + the passive participle: rii/:t marmijiPari(i = he was thrown to the ground. 47 This is synonymaus with the traditional synthetic irtama jiPari(i. But this example brings out the central issue: the cantrast Bynon, 43-44; Czapkiewicz, Verb, 213-217; id., Accumulation, 15. Bynon, 17, 35. 46 Meillet, Evolution, 144-46; Browning, 30, 32 f. 47 Woidich, 93. For other such "modern" constructions in Arabic, see Czapkiewicz, Verb, 129-54. Cf. ibid., 210. This is the basic objection to Versteegh, who seems to believe that )inqatal is analytical whereas qutil is synthetical (90, point 3). Cf. Corriente's remarks 30-32. 44

45

16

THE APOPHONIC PASSIVE IN ARABIC

between HA and VA is not generally synthetic versus analytic as in this example. Both language forms have PM of a synthetic type. This is contrary to what would have been expected. It may be objected that there are cases of one synthetic form being replaced by another. A brilliant example is provided by Latin. The medium form in IE, serving as PM in Greek, Sanskrit and Gothic, is in Latin replaced by finite forms with an r-suffix. The r-suffix is probably from the beginning a marker of unspecified subject or even SC. 48 It is thus a category rather close to the PM. But this Latin example is instructive: the r-conjugation in Latin is basically a PM and its extension as an intransitive marker (so called deponens) is secondary. All evidence indicates that there is no IE "passive" conjugation; the medium was basically an intransitive conjugation, which as such might be employed in PC as weil, according to a well-known pattern. The differentiation of a special "passive" conjugation is secondary. In Greek such a differentation was made only in certain tenses; in Latin it was carried out consistently. In IE we thus from the beginning have a general intransitive conjugation. Special passive conjugations are innovative and the result of reemployment of productive, morphologically transparent categories with functions and meanings close to those of the PC. In Arabic, on the contrary, HA has a distinctive passive conjugation which is fully apt to carry this function even after considerable phonological changes. This is shown by the functionality of this kind of opposition with the participles of derived stems as weil as with the finite verb in some Peninsular dialects. In most forms of VA there are no traces of a system like the one in HA in the derived stems. If we suppose that the HA system has developed as sketched by Cantineau, we could imagine a new analytic passive conjugation in the G-stem. U nfortunately the forms actually employed in VA, the nG, tG and Gtstems, do not fulfill the demands for being reemployed in this way. At least the nG and the Gt-stems give the impression of being lexicalized survivals already in HA. The attempts by the classical Arab grammarians to ascribe to them a weil defined semantic content are not very convincing and the categories suggested are admittedly non-existent with many of them. 49 They arealso improductive in HA. From a general viewpoint their reemployment as a regular PM is thus not very likely. The cantrast with the situation in Latin deserves to be underlined: there 48 Cf. note 19. That the "passive" meaning is a secondary development of the medium is underlined by Benveniste, 168. 49 Cf. Fleisch, Traitill, 14-15; seealso Cowell, 48 ff. for the problern with derivation.

THE APOPHONIC PASSIVE IN ARABIC

17

a multifunctional form ( = the IE medium) has been replaced by an unambiguous one as PM, while in Arabic the unambiguous yuqtal-qutil is said to have been replaced by a multifunctional one. From the general nature of the PC, which tends to be continually remodelled, this development is unlikely. 50 The possibility that such a development has occurred in spite of the objections presented cannot be excluded, but the problern needs a closer investigation. Since the development traditionally supposed is unlikely, claims that it has actually happened must be supported with solid evidence, not ad hoc explanations. But before this is made, one should check if it is possible to prove that Proto-Arabic actually did have a passive system like HA. This must be shown before the reemployment explanation is considered further. The simplification argument A basic weakness in Czapkiewicz's reductional argument is the lack of documentation of the actual process. He supposes the HA or a language close to it as the ancestor of VA, but admits that it is undocumented and simply puts HA in its place. The argument for this is that HA shows close parallels to U garitic, which is said to be an indication of its antiquity. 51 The problern is that there is no evidence in Arabic for intermediate stages corresponding to e.g. Middle High German between Old High German and Modern High Germ an, or Old French between V ulgar Latin and Modern French. In these languages we can observe the changes fairly weil and define several intermediate systems between, let us say, the case-marking system of Latin and- that of the modern Romance languages. Such intermediate stages may of course have existed also in Arabic, but they arenot observable and cannot be taken for granted. In fact Czapkiewicz takes what should have been his results as his starting point. He gives a very neat description of the differences between the verbal systems in HA and VA, but cannot adduce any evidence for the genetic relations he claims to exist between them. Czapkiewicz also supposes a process of regeneration which in fact is similar to what has been stated by Meillet. 52 Such innovations adduced by Czapkiewicz are the formation of verbal stems like the sa-causatives in Mauretania, the tG PM and the "fCäl" themes in North Africa, etc. But even here no decisive arguments for these being innovations are

50 51

52

On this remodeHing of the PM, see Retsö, Passive, 196-197. Czapkiewicz, Verb, 18 ff. Czapkiewicz, Verb, 129, 198; Meillet, loc. cit.

18

THE APOPHONIC PASSIVE IN ARABIC

presented. It should be underlined once more that none ofthese supposed innovations are analytic. They are as synthetic as those they are supposed to have replaced. Of Czapkiewicz' other parallels, the one with Neo-Aramaic is important since, in cantrast to his simplification process, the development from OAr to Neo-Aramaic is very likely and also partly observable. 53 I.e. the regenerating process, as far as analytic constructions are concerned, is very likely in both VA and Neo-Aramaic. But his derivation ofOAr from Amoritic is as unprovable as his derivation of VA from HA. 54 It should be stressed that, at this stage, we do not claim that VA (and OAr) cannot have developed as Czapkiewicz supposes. We also agree that the pulsatory evolution of language in general as sketched by him and earlier by Meillet is probably correct. But this does not imply that all differences between related linguistic systems should be explained in this way. Since Czapkiewicz has not been able to produce evidence that such a simplification process has actually happened in the Arabic verbal system, the simplification hypothesis cannot yet be accepted for the passive system before such evidence is presented.

Conclusions It seems that none of the usual explanations presented for the different passive systems in Arabic are tenable. The distribution of "survivals" and the type of material supposed to be reemployed as PM give no immediate confirmation of any of the explanations put forward, and serious objections can be made against all of them. . All explanations referred to here presuppose the development from a system similar to the one found in HA. But as has emerged during this survey, a question mark must be put against this, at least for the derived stems: there are no indications whatsoever that the ancestors of most forms of VA did possess a system similar to the HA system. As far as the G-stem is concerned, the existence of several cases that can be interpreted as '' survivals'' makes such a presupposition somewhat more likely. But, as we have seen, even here there are serious difficulties in explaining the diachronic relationship between HA and VA. It is clear that the G-stem deserves a closer Iook. The main task will then be to show whether or not the predecessor(s) of VA did have a passive conjugation identical with or similar to the HA system. The problern is that, since these predecessors are not accessible in documents, we must try to reconstruct from available evidence. 53 54

Czapkiewicz, op. cit., 211. Czapkiewicz, op. cit., 213-17.

THE APOPHONIC PASSIVE IN ARABIC

19

When the problern is formulated in this way, it becomes obvious that one has to look beyond the confines of Arabic, since the passive system in Arabic shows close affinities to the ones found in other Semitic languages, and the internal Arabic diachronic evidence gives no solution. The question whether the HA passive system represents a starting point for the dialects is connected with the history and development of the passive system in the whole of Semitic. This means that not only the finite verb should be discussed, but also the nominal forms (participles) which show close morphological affinity with the verb. This has not been clone either in Arabic or in Semitic in general. A further obstacle is the lack of studies on the functions of the participles, both the '' active'' and the ''passive''. It is to be hoped that the following will point out some of the directions for future study of the subject.

CHAPTER TWO

THE APOPHONIC PASSIVE MARKER IN SEMITIC General survey

In Semitic the correlative oppos1t10n with finite verbs in active and passive constructions or the apophonic PM is found, apart from HA and some Arabic dialects, in Hebrew at all stages of its history, sporadically in OAr, BAr, IAr and finally in MSA. In Phoenician, U garitic, Amoritic and the language of Ebla, there are isolated cases which may be interpreted as aPM. This type of verb does not exist in Akkadian, the Ethiopian languages or possibly Ancient South Arabian. 1 In all languages which possess the aPM system, augmented stems ( n or t) may also be used in PC; they tend to dominate even in Hebrew which is the language most similar to HA, and some dialects on the Peninsula which have the most extensive use of these forms. As to Afroasiatic, the aPM system is absent in all branches except Ancient Egyptian, where the sdmj conjugation may be employed in PC and perhaps had a different vocalism than in AC. 2 Since the verbal forms in Egyptian have a nominal origin and are not connected with the common Afroasiatic prefix conjugation, the hypothetical aPM in Egyptian are probably not directly related to the ones discussed here. 3 The use of correlative opposition in active-passive constructions is thus a characteristic found in a group of languages within Semitic. It is also necessary to underline that all the Semitic languages having this device belong to the larger group which uses a suffix conjugation, the traditional "perfect", as opposed to the prefix forms, the "imperfect". Even within this group, traditionally called West Semitic, not alllanguages have the aPM system. Apart from several modern Arabic dialects the entire Ethiopic group should be mentioned, as well as Aramaic. The original extension of the apophonic passive: arguments

More attention has been paid to the problern of the ongm and development of the aPM system in Semitic as a whole than in Arabic. 1 For possible instances in ASA see Häfner, 82; Beeston, Sabaic, 14; for Geel 692 yaCazi-y· ahu 693 yda(-yahu 654 cf >et-yMa.c 582 ya~aJi- >et

ya~azi- >et

yho-yakin 600 yakin 670 ykO.n-yO.hU 672

yamlek 678 yho-natan 606 ntan- >et 978-80 c(l?ar-yahu 105 7 Cazar->et 1054 yho-yaqim 601 yho-yarib 602 yarib 730 yarab-cam 722 yrub-ba cal 721 yho-ram 612 yirm-yahu 739 ram-ya 126 yipt·a~ 710 p·ta~-yä 1185 yiim(y)->et 743 skan-yahU 1334 yifmaUe[ 766 yismac-yahu 767 sma (-yahu 13 71

The obvious conclusion from this Iist is that, even if the exact identification of every root is not secure, all the documented Amoritic yaqtil forms must be G-stems since the corresponding names in the other languages contain such verbs only. All other imperfects occurring in proper names in Amoritic are of the yaqtul or yaqtal type, thus definitely G-stems. There is thus no indication whatsoever that any of the yaqtil verbs are anything but G-stems in Amoritic. Further, it can be supposed

64

THE SEMITIC CAUSATIVE CONJUGATION

that Amoritic had a CCj of the same type as the one in Akkadian, i.e. with the I-affix. 51 The opinion commonly held, that some of the Amoritic yaqtil are "hi:fil" or ")a:fel" forms, must be considered unlikely, and the more so since there are no tenable arguments for elision of h or ) in Amoritic either. 52 This is a clear indication that the languages in Syria in the third and second millennia B. C. had a system of derived verbal stems similar to the one in Akkadian. The next step must be to compare all the documented Amoritic yaqtil forms with BH verbs in order to study the relationship between the old G-stem and the later development of these verbs. They are given in the following table: Table 4

Amoritic 1. ia-bi-sum me-bi-sum 2.

ia-bini-im

3.

ia-ab-ni

4.

ia-gi-ib

5. ia-di-nim 6.

za-wz

7.

ia-ab-bi

8.

ia-a!J,-wi

9.

ia-ah-1i

ia-di-ibl ia-da-ab/ i-da(ia-dab) 11. ia-#I ia-aw-si/ i-zi me-1i-tu-um 12. ia-aw-fi-bu 10.

Corresponding Hebrew verb (BH) yebäf b·as 'be ashamed' yabif höbis 'act shamefully, mebif 'appear being ashamed', 'put to shame' yabin bin/ b· an 'understand, hebin 'be able to discern, mebin 'understand' yibnle b·an%. 'build' (- CCj) yagi"~ - 'burst forth' megi"b yadin d·an 'plead o's cause, execute judgement' - (- CCj) yihyle haya 'be, become' + CCj (yahwrE) ya~b· i' hrEbb· i' 'hide' ~b"bä-yJ (Noth 178, 197) nG yi!tyle ~ay%. 'be alive' hrE!tyi 'preserve alive, revive' yrE~rezle ~az%. 'see' ma~"zle 'vision' yeda' yada' 'know' yödi"' hödi"' 'tell' ye1e' y~a' 'go out' yö1i' 'bring out' yefeb yäfab 'sit, dwell' yöfib höfib 'put, make dwell'

51 Cf. Gelb, 3.3. 7. 7 .1. Perhaps also in Eblaitic, see Müller, Verba/system, 230; id., Ebla, 164. The consequences of Rabin, Vocalisation, must be that the names are G-stems. 52 Thus against Gelb, 3.3.7.6.1.; Huffmon, 66-73.

THE SEMITIC CAUSATIVE CONJUGATION

13. ji-ti-irlia-tar 'surpass' me-tar 14. ia-ki-inlia-ku-un me-ki-nu-um 15. ia-al-e 'be able, strong' 16. ia-am-lik 'rule, possess, counsel' 17. ia-am-silji-im-si 'wash, purify'

yotir hotir 'leave over' nG 'be left' yakin (ken) hekin nG yil'ii! ta>ä 'be weary' yal'ii! hat>ä 'make weary' yimlok/ matak 'reign' yamlik himlik 'make king' yamsiE himsiw 'dissolve, melt,

overflow' (intr.) 18. ia-am-,i 'find' 19. ia-an-nu-ub meli-ni-bu-um

yim,O.' mO.,O.' 'find' yam,i' him,i' 'cause to find' yamJah na~ 'repose' yania~ henia~

'cause to rest'

hin· ;a~

'cause to rest' 'pay, put' yif. 0.' nafa' 'Iift up' yaf.i' hif.i' 'in die Höhe heben' ya'· ib hi,. ib 'set u p' nG 'be set up' yit-en natan 'give' menia~

mania~

20.

ia-as-si 'raise up'

21.

ia-an-,i-iblia-,i-ib

22.

ia-an-ti-in/ya-ti-in

23. ia-ab-zi-ib 'rescue' 24.

ia-ab-zi-ir 'help'

25.

ia-ab-li 'be high, exalted'

26. me-eh-ni-yu

27.

28. 29.

ia-pi-i!J/ia-a-pa-a!J 'shine' me-pi-!Ju-um ia-ap-ti-i!Jiya-ap-ta-!Ju

'open'

30.

ia-aq-qim-im 'vindicate' or me-qi-im 'rise' ia-aq-ri 'call'? 'happen'?

31.

ia-ar-bi 'be great'

32. ia-ar-ip 'recompensate' 33.

za-n-zm 'be high' or

'Iove'? 34.

za-sz-zm 'put' ('be pretty'?)

35.

ia-si-it 'put, place'

ya"'zob 'dzab 'leave; set free' ya"'zor 'O.zar 'help' ma'zir (2 Chron. 28,28) K: l'zyr (_2 Sam. 18,3) yaCafiif 'o.ta 'ascend' ha '. 95 Speiser' s claims crumble when confronted with the following facts: ( 1) The elision hypothesis is unprovable and improbable. (2) As has been shown here, U garitic has the CCj with I but, at the same time, th~e hforms of the personal pronouns. Of the MSA languages, some have-CCj with ' but pronouns with I. 96 The distribution of these elements in fact gives no support for the idea that they would be diachronically identical. Nevertheless, Speiser's ideas have found their way into the latest handbooks on comparative Semitics. 97 This led Bravmann to defend the old theory of the basic unity of the three prefixes, pointing to U garitic which possesses both sibilant and Iaryngal elements and also the Itaugmented verbs which are found in other languages and usually considered "reflexives" from the CCj with hl'-augments. Bravmann refers to the shift s > h in Indoeuropean as a parallel. 98 Unfortunately none of this is tenable either. To start with the last argument it suffices to remark that the shift s > h is not the same as I> h which must be the situation in Semitic. Secondly, the very fact that U garitic has CC j with f tagether with h-pronouns, as well as the occurrence of It-forms in languages with Iaryngal augmented CCj, must not be ignored ( cf. pp. 78-79). Finally the coexistence of CCj with I and h in Mehri, and I and ' in the other languages, cannot be explained away with the one being a loan from outside, as has been clone by e.g. Rabin. 99 Even if some of Rabin's examples from Arabic are possible loans from another dialect where the characteristic South Semitic change I > s has not occurred, there is nothing to support the idea that all the I-verbs in Hebrew and Aramaie must be loans as well. This argument is based on the fact that these roots are found with the I-augment only and not as unaugmented roots. This may, however, be explained by the fact that they are old denaminative forms with I-augment. If the phonetic shifts would have taken place in Semitic as supposed by Bravmann and the other "unitarians", one must also explain all the cases in which they have not occurred. All Semitic languages abound with words with initial I, s, h and '· That the shift in the CCj could be a morphophonematic one affecting certain morphemes only cannot be ruled out, but the rule for this must first be established before the 95 96 97 98

99

Garbini, Alternanza; cf. Greenstein; Loprieno, 142-144; Leemhuis, 4. Johnstone, South Arabian, 25-26. Moscati, lntroduction, 104 f. Bravmann, passim. Rabin, Safe!, 157.

THE SEMITIC CAUSATIVE CONJUGATION

83

hypotheses of the CCj prefixes and the personal pronouns can be accepted. The separate origin hypothesis II: f versus h versus ) As a consequence of this, there are scholars who have advocated a separate origin for all the three affixes concerned. This was suggested already by Brockelmann. The most thorough investigation was undertaken by Nyberg, who already in 1921 clarified many aspects ofthe problem. Judging from the studies mentioned above, this has not had the influence it deserves. 100 From his study Nyberg drew the following conclusions pertaining to our subject:

( 1) The three causative augments are identical with the ones found as nominal prefixes in Semitic, with functions similar to the m-prefixes. 101 (2) The nominal forms are in principle older than the verbal forms, i.e. the formation of nouns with these prefixes is a prerequisite for the formation of verbs. 102 (3) The three affixes are of different phonetic ongm. They cannot be directly derived from each other. According to Nyberg they are old demonstrative elements with a rather similar function, which explains their interchangeability . 103 The survey carried out here makes it difficult to reject Nyberg's results. The distribution of these three elements in Semitic as a whole makes it impossible to uphold the view that they are basically the same morpheme from the beginning. A rejection of that view entails the necessity of explaining all the documentable cases where they occur under similar circumstances in the same language as loans. It is clear that this is a circular proof. All evidence points to the fact that the observable interchange between these elements is due to one replacing another and not phonetical transformation oj one element to another. The )-affix Nyberg quite correctly compares the )-prefix with the same element found in some common inherited words in most Semitic languages, e.g. 100 Brockelmann, Grundriss I, 521; Nyberg, Wortbildungen, 243, 261. See also Rundgren, Bildungen, 179. 101 Nyberg, op. cit., 262-263. 102 Op. cit. 263; cf. Speiser, Studies, 30-32. These explanations need tobe corrected but the idea of the nominal origin of the perfect is basically right. 103 Nyberg, op. cit., 196, 243; cf. Speiser, Studies, 27-28.

84

THE SEMITIC CAUSATIVE CONJUGATION

Arabic 'afba< 'finger' (cf. Syriac feb
'1j

CO

0

!'-:)

103 104 yamfa,t 'take away' tG 'be taken away' 105 106 yan'as 'diminish (tr.) tG 'be considered small' 107 108 109 yanday 'make poor' tG 'be made poor' 110 yandaf 'hurt' tG 'be hurt' 111 112 yanxal 'destroy' tG 'be destroyed' 113 114 115 116 yanwar 'vituperate' tG 'be vituperated' 117 118 yanfos 'expire' tG 'breathe'

97 98 yam'az 'diffuse odor' tG 'be perfused with odor' 99 100 101 102

napeqlnepuq 'spend'

namsar/nemsur 'despise' tG 'be despised'

namse 'cause to melt' tG 'be melted'

nam,tarlnem.tar 'rain' tG 'to be rained upon'

96

94 yamtaq 'Iift up' tG 'be lifted up' 95

'weaken, become weak' tG 'become weak'

nam~el/nem~al

nam'enlnem 'an 'irk, bore'

93

88 89 90 91 92 yamhaw 'melt'(tr.) tG 'be melted'

'Iead'

hin~a

ya_t·a:lyi,t-a: 'stretch out' nG 'be stretched out' yania'fyanua< 'toss' nG 'be tossed about' yanip:nap 'sprinkle'

yan~a::na~a

yag·ia'fyig·a' 'tauch' yad· ia~ 'scatter' nG 'be scattered'

yamfa::mlifa 'draw out' himfa

yamiq 'cause to rot' nG 'rot, putrefy' yamra::mara 'be rebellious' himra yamriq:maraq 'polish, scour' himriq [hemas]lyimas 'melt' (tr.,intr.) nG 'be melted'

yam,tir 'rain' nG 'be rained upon' yamir 'change' nG 'be changed' yamli,t 'deliver' nG 'save o's'

[ma'mir] 'painful' nG 'hurt'

ylili$ 'be spokesman; deride' laf 'talk big' yal'iglyil'ag 'mock, deride' nG 'stammer'

\.0

0

)'.:J

0

x

t%1

"'"' z

>

139 140 141 142 143 144 145 146 147 148

129 130 131 132 133 134 135 136 137 138

128

ya e and the Treatment of lamed-he formations in Biblical Hebrew", JNES 41 (1982), pp. 35-43. Knudtzon, J. A., Die El-Amama Tafeln mit Einleitung und Erläuterungen. I: Die Texte, li: Anmerkungen und Register, Leipzig 1915. (VAB 2.) Kofler, H.,"Reste altarabischer Dialekte B: Verbum", WZKM 48 (1941), pp. 52-88. Kropat, A., Die Syntax des Autors des Chronik verglichen mit der seiner Quellen. Ein Beitrag zur historischen Syntax des Hebräischen, Giessen 1909 (BZA W 16). Kurylowicz, J., L'apophonie en semitique, Wrodaw/Warszawa/Krak6w 1961. (Prace jezykoznawcze 24.) - - , Studies in Semitic Grammar and Metrics, Wrodaw/Warszawa/Krak6w 1972. (Prace jezykoznawcze 63.) Kutscher, Y., The language and linguistic background of the Isaiah Scroll (I Qisa), Leiden 1974 (Studies in the Textes of the Desert of Judah, Vol. VI). - - , "Mishnaic Hebrew", Encyclopaedia judaica, J erusalem 1971, vol. 16, col. 1590-1607.

244

BIBLIOGRAPHY

''Two Passive Constructions in Aramaie in Light of Persian' ', Proceedings of the International Congress of Semitic Studies held in Jerusalem 19-23 july 1965, Jerusalem 1969, pp. 132-151. Köhler, L. Baumgartner, W., Lexicon in Veteris Testamenti libros, Leiden 1958. König, F. E., Historisch-kritisches Lehrgebäude der hebräischen Sprache I: Lehre von der Schrift, der Aussprache, dem Pronomen und dem Verbum, Leipzig 1881. Labat, R., "Le pre,fixe n- en akkadien", GLECS 2 (1937), pp. 47-48. de Landberg, C., Etude sur les verbes fi'il, publie parK. V. Zettersteen, Uppsala 1939. (UUÄ I:6) Lane, E. W., An Arabic-English Dictionary. Book I:1-8, London/Edinburgh 1863-1893, repr. 1955. Leemhuis, F., The D and H Sterns in Koranic Arabic. A comparative study of the function and meaning oj the faccla and ,afala forms in Koranic usage, Leiden 1977. Lefebvre, G., Grammaire de l'egyptien classique, 2me ed. avec Ia collaboration de G. Sauneron, Le Caire 1955. Leslau, W., "Le passif interne en soqotri", GLECS 2 (1937), pp. 91-92. - - , "Sur Je prefixe n- en soqotri", GLECS (1937), pp. 45-47. - - , "Le rapportentre s et h en semitique", Annuaire de l'institut de philologie et d'histoire orientale et slave, Bruxelles 7 (1939-1940), pp. 265-272. Leumann, M./Hofmann, I./Szantyr, I., Lateinische Grammatik I: Laut- und Formenlehre, von M. Leumann, Neuausg. München 1977. Levi-Proven~al, E., Textes arabes de l'Ouargha. Dialecte des Jbala (Maroc septentrional), Paris 1922. Lisän al-'Arab guz, 1-20. al-tab'a -~-,ülä, Buläq 1300. Loprieno, A., Das Verbalsystem im Agyptischen. Zur Grundlegung einer Aspekttheorie, Wiesbaden 1986 (Göttinger Orientforschungen IV Reihe: Ägypten Bd. 17). MacLean, A.J., Grammar of the Dialects of Vernacular Syriac, Garnbridge 1895, repr. Amsterdam 1971. Macuch, R., Grammatik des samaritanischen Hebräisch, Berlin 1969 (Studia Samaritana Bd. I). - - , Handbook of Classical and Modern Mandaic, Berlin 1965. Malaika, N., Grundzüge der Grammatik des arabischen Dialektes von Baghdad, Wiesbaden 1963. Mar~ais, Ph., "Contribution a l'etude du parler arabe de Bou-Saada", BIFAO 44 (1945). - - , Esquisse grammaticale de l'arabe maghribin, Paris 1977. - - , Le parler arabe de Djidjelli (Nord-Constantinois, Algerie), Paris 1956. Mar~ais, W., Le dialecte arabe parle a Tlemcen. Grammaire, textes et glossaire, Paris 1902. Mar~ais, W./Fares, J., "Trois textes arabes de l'el-l;lämma de Gabes", JA 218 (1931 ), pp. 193-247; 221 (1932), pp. 193-270; 223 (1933), pp. 1-88. Mar~ais, W./Gulga, A., Textesarabesde Takrounall: Glossaire. tome 1-8, Paris 1958-1963. Marcus, D., "The qal Passive in Ugaritic",JANES 36 (1970-1971), pp. 103-111. Matous, L., "Die semantische Analyse des VII und VIII Verbalstammes im modernen Schriftarabisch", Semitistischer Cercle III, ArOr 35 (1967), pp. 293-297. Margoulis, M., Lehrbuch der aramäischen Sprache des babylonischen Talmuds. Grammatik, Chresthomatie und Wörterbuch, München 1910. · Meillet, A., "L'evolution des formes grammaticales", Linguistique historique et linguistique generale, Paris 1975, pp. 130-148. - - , "Sur !es caracteres du verbe", Linguistique historique et linguistique generale, Paris 1975, pp. 175-198. Merrill, E. H., "The aphel Causative: Does it Exist in U garitic?", JNWSL 3 ( 1974), pp. 40-49. Meyer, R., Hebräische Grammatik I-IV, Berlin 1966-1972. Mez, A., "Über einige sekundäre Verba im Arabischen", Orientalische Studien Theodor Nöldeke zum siebzigsten Geburtstag ( 2 März 1906) gewidmet von Freunden und Schülern, hrsg. von Carl Bezold, Bd I, Giessen 1906, pp. 249-254. Mirkin, R., "Misqal mefU'al", Lefonenu 32 (1967-1968), pp. 140-152.

BIBLIOGRAPHY

245

Mitchell, T. F., An Introduction to Egyptian Colloquial Arabic, Oxford 1956, repr. 1978. Mittwoch, E., "Die traditionelle Aussprache des Äthiopischen", MSOS 28 (1925), pp. 126-148. Moran, W. L., "The Hebrew Language in its Northwest Semitic Background", The Bible and the Near East. Essays in Honor of William Foxwelt Albright, London 1961, pp. 54-72. Moscati, S./Spitaler, A./Ullendorff, E. /von Soden, W., An Introduction to the Comparative Grammar of the Semitic Languages: Phonology and morphology, Wiesbaden 1969. Mozayni, H. Q., Vowel Alternation in a Bedouin Hi:fazi Arabic Dialect: Abstractness and Stress, Diss., Austin Texas 1981. Müller, H. P., "Das eblaitische Verbalsystem nach den bisher veröffentlichten Personennamen", LE pp. 211-233. - - , "Ebla und das althebräische Verbalsystem", Biblica 65 (1984), pp. 145-167. Nedyalkov, V.P./Silnitsky, G.G., "The Typology of Morphological Causatives", Trends in Soviet Theoretical Linguistics, ed. F. Kiefer, Dordrecht 1973, pp. 1-32. .. Newman, P., "The Formation of the Imperfective Verb Stern in Chadic", AuU 60 (1977), pp. 178-192. Noth, M., Die israelitischen Personennamen im Rahmen der gemeinsemitischen Namengebung, Stuttgart 1928, repr. Hildesheim 1980. Nyberg, H. S., Hebreisk Grammatik, 2 uppl. Stockholm 1972. --,"Wortbildungen mit Präfixen in den semitischen Sprachen", MO 14 (1920), pp. 177-291. Nöldeke, Th., Beiträge und Neue Beiträge zur semitischen Sprachwissenschaft, Strassburg 1904-1910, repr. Amsterdam 1982. - - , Kurzgefasste syrische Grammatik, 2 Aufl. Leipzig 1898. - - , Grammatik der neusyrischen Sprache am Urmia-See und in Kurdistan, Leipzig 1868, repr. Hildesheim 1974. - - , Mandäische Grammatik, Halle an der Saale 1875, repr. Darmstadt 1964. - - , Zur Grammatik des klassischen Arabisch, Wien 1897, reprint Darmstadt 1963. Odeberg, H., The Aramaie Portions of Bereshit Rabba //: Short Grammar of Galilaean Aramaic, Lund/Leipzig 1939. 0' Leary, L. , Comparative Grammar oj the Semitic Languages, London 1923, rep. Amsterdam 1969. Owens, J., A Short Reference Grammar of Eastem Libyan Arabic, Wiesbaden 1984. Palva, H., "The Arabic Dialect of the I:Iwetät Tribe", OrSUI!c 33-35 (=On the Dignity of Man. Orientaland Classical Studies in Honour of Frithioj Rundgren, ed. T. Kronholm and E. Riad, Stockholm 1986), pp. 295-312. - - , ''Characteristics of the Arabic Dialect of the Bani Saxar Tribe'', OrSuec 29 (1980), pp. 112-139. - - , Narratives and Poems from lfesbän. Arabic Texts recorded among the Semi-Nomadic el-'Agärma Tribe (al-Balqä' District, Jordan), Göteborg 1978. (OG 2.) - - , Studies in the Arabic Dialect oj the Semi-Nomad el- 'Agarma Tribe (al-Balqä' District, Jordan), Gothenburg 1976 (OG 2). Panetta, E., L'arabo parlato a Bengasi, Vol //: Grammatica, Roma 1943. Parsons, F. W., "The Vebal System in Hausa", AuÜ 44 (1960/61), pp. 1-36. Petnicek, K., "Die innere Flexion in den semitischen Sprachen IV: Entstehung und Entwicklung des Systems", ArOr 31 (1963), pp. 577-624. Plazykowsky-Branner, H., "Der Kausativ in den sogenannten kuschitischen Sprachen", Anthropos 54 (1959), pp. 129-140. Porath, E., Die Passivbildung des Grundstammes im Semitischen. Eine sprachgeschichtliche Untersuchung, Breslau 1926. Porges, N., "Über die Verbalstammbildung in den semitischen Sprachen", SAWW 79 (1875), pp. 281-354. Prasse, K.-G., Manuel de grammaire Touaregue (tahaggart) IV-V: Nom, Copenhague 1974, VI-VII: Verbe, Copenhague 1973. Praetorius, F., "Zur Kausativbildung im Semitischen", ZS 5 (1929), pp. 39-42.

246

BIBLIOGRAPHY

- - , "Über die aramäischen und arabischen Passivperfekta", ZS 2 (1924), pp. 134-141. - - , Athiopische Grammatik, repr. New York 1955. Prochazka, Th., "The Shi