310 26 10MB
English Pages 312 [316] Year 1979
Persian Grammar
Trends in Linguistics State-of-the-Art Reports 12
Editor
Werner Winter
Mouton Publishers The Hague · Paris · New York
Persian Grammar History and State of its Study
Gemot L. Windfuhr
Mouton Publishers The Hague · Paris · New York
ISBN 90 279 7774 2 © Copyright 1979 by Mouton Publishers, The Hague. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced in any form - by photoprint, microfilm, or any other means - nor transmitted nor translated into a machine language without written permission from the publisher. Printing: Karl Gerike, Berlin. - Binding: Lüderitz & Bauer Buchgewerbe GmbH, Berlin. Printed in Germany
To my wife
G UDR UN
TABLE OF CONTENTS
Acknowledgments
1
Foreword 1. 2.
Previous surveys The present survey 2.1 Scope 2.2 Approach to this volume Transcription
3 3 4 4 5 7
1.
The study of Persian grammar 1.0 Introduction 1.1 Studies by Europeans 1.1.1 Theoretical approaches 1.2 Studies by Iranians 1.3 Appendix: Grammars by Europeans
9 9 12 17 22 23
2.
Syntax 2.0 Introduction 2.1 The noun and noun phrase 2.1.1 Number and definiteness 2.1.1.1 hä and plural 2.1.1.2 / and definiteness 2.1.1.3 Colloquial -e 2.1.2 Cases 2.1.2.0 Introduction 2.1.2.1 Prepositional cases and adverbs . . . . 2.1.2.2 rä and definiteness 2.1.2.2.1 rä as a case 2.1.2.2.1.1 Personal suffixes . . . 2.1.2.2.2 rä as a referent 2.1.3 Noun-phrase embedding
27 27 29 31 31 34 40 41 41 46 47 49 52 53 57
3.
viii
2.2
2.3 2.4 2.5
2.6 2.7
3.
2.1.3.1 Ezäfe 2.1.3.2 Subordinate clauses 2.1.3.2.1 Relative clause 2.1.3.2.2 Other subordinate clauses . . 2.1.3.3 Topicalization 2.1.3.3.1 Emphatic ke and anacoluthoη 2.1.3.3.2 Subject reduplication . . . 2.1.3.3.3 Emphatic and reflexive xod . 2.1.3.3.4 mar 2.1.3.4 Participial clauses 2.1.3.5 Nominal word classes 2.1.3.6 Nominal compounds 2.1.3.7 Derivative nouns 2.1.3.8 Word groups 2.1.4 Arabic grammar in Persian 2.1.4.1 Gender The verb and verb phrase 2.2.1 The contemporary verbal paradigm 2.2.1.1 Tense, aspect, mood 2.2.1.1.1 Tense and aspect 2.2.1.1.2 Mood 2.2.2 The verbal paradigm in the lOth-12th centuries . . 2.2.2.1 bi 2.2.2.2 hame me and e 'Tobe' Modals 2.4.1 bäyad Aspectual auxiliaries 2.5.1 där and the progressive 2.5.2 gir 'take' 2.5.3 xäh 'want' sav and 'passive' Compound verbs 2.7.1 Prefixal preverbs 2.7.2 Nominal preverbs 2.7.3 Indirect (middle) verbs
Phonetics and phonology 3.0 Introduction 3.1 Survey of scholarship 3.2 Vowels
57 62 62 68 70 70 72 73 74 74 75 76 80 80 80 82 83 83 85 86 92 92 94 96 97 99 100 102 102 103 104 105 113 114 116 126 129 129 129 134
ix 3.3 3.4
3.5 3.6 3.7 3.8
Diphthongs Consonants 3.4.1 q 3.4.2 Glottal stop 3.4.3 Conditioned variation Phoneme distribution Notes on diachrony Stress and other suprasegmental features 3.7.1 Diachronic observations The writing system
137 138 138 139 140 143 144 144 149 149
4.
Lexis 4.0 Introduction 4.1 Lexicology 4.1.1 Lexis in grammar 4.1.2 Word structure 4.1.3 Innovations 4.1.4 Etymology 4.1.5 Borrowings 4.2 Lexicography 4.3 Concordances 4.4 Historical dictionaries 4.5 Phraseology
151 151 152 152 152 153 154 155 158 162 163 163
5.
Stages of modern Persian 5.0 Introduction 5.1 Studies of periods and diachrony 5.2 Emergence and formative period 5.3 Oldest preserved originals 5.4 Manuscripts from the 10th century onward
165 165 166 169 170 171
6.
Appendix: Topical bibliography 6.0 Introduction 6.1 References 6.2 Some grammars and textbooks (after 1900) 6.3 Some dictionaries
173 173 176 191 192
7.
Bibliography 7.1 Introduction 7.2 Alphabetical list of abbreviations of journals and collective works including festschriften 7.3 Alphabetical bibliography I: items without specified author. 7.4 Alphabetical bibliography II: by author
193 193 199 209 211
ACKNOWLEDGMENTS
When the editor of the Series critica first asked me to contribute a monograph on Persian of some 80—160 pages I did not know to what extent I was going to run over that page limit and, I should add, the time limit then set. I thank him for his going along with my changing the plans. I greatly appreciate the assistance of many colleagues who were kind enough to send me more or less detailed bibliographical information or gave me copies of their books, articles and papers. I hope my colleagues forgive me for not mentioning them all here by name. During the course of writing this monograph I enjoyed the mental and manual help of a number of people, among them, first, Iraj Bashiri (now Assistant Professor of Iranian and Turkish Studies at the University of Minnesota), succeded by David Peterson (now director of the American Institute of Iranian Studies, Tehran); together we had many mutually enlightening discussions on practical and theoretical points; and, finally, the indefatigable John Workman (doctoral student at the University of Michigan). I thank them very much for their help. In compiling the bibliography I enjoyed the help of many colleagues and fellow-researchers. First of all, I should like to thank Iraj Afsär, the bibliographer supreme of Iranian Studies, who so generously provided me with copies of the linguistic sections of his as yet unpublished new volume of the Index Iranicus. I thank Dr. Hormoz Miläniän and Dr. Xosrow FarSidvard for checking, with me, the details of the Persian entries of this bibliography. They also kindly provided me with additional bibliographical data and many of their offprints as did their colleagues in the Department of Linguistics at the University of Tehran, Dr. Ahmad Tafazzoli, Dr. Ali ASraf Sädeqi, Dr. Mohammad Rezä Bäteni, Dr. Mohsen Abolqäsemi. I am grateful to Dr. 2äle ÄmuzegärYegäne, the keeper of the library of that department, as I am to Dr. Jamal Rezä'i, the departmental chairman, and to Dr. Ma'sume Qarib and Dr. Vosuqi and Mr. Aziz 2iyän. Dr. Käzemi helped me in my work at the Central Library of the University. I thank my colleagues at the Iranian Language Academy, Dr. Sädeq Kiä, its eminent president, further Dr. Feridun Badre'i and Dr. Yaddolläh Samare as well as Dr. HaqSenäs, Dr. Iran Kalbäsi and Mr. Majidi. The Secretary General of the Iranian Culture Foundation, Dr. Parviz Nätel Xänlari, kindly offered his advice, and repeatedly provided me with copies of the Foundation's publications on Persian.
2 Dr. Mohammad Este'lämi, chairman of the Department of Literature at the Pedagogical University, helped me in many ways. In Tabriz, my thanks go to Dr. Manucehr Mortazavi, Dean of the Faculty of Letters, and to Dr. Baqäi and Dr. Bahman Sarkaräti for their information. I would also like to express my gratitude for their kindness in sending me their journal. Outside Iran, many colleagues have been helpful. Dr. Ehsan Yarshater supplied many copies of his work and bibliographical data; so has Dr. Richard N. Frye; Dr. Don L. Nilsen sent his bibliography on Afghan linguistics and other data. Among many others who sent their material I would like to mention Dr. Iraj Bashiri, Mr. Donald Croll, Mr. Amin Hasanpur, Dr. Mohammad Ali Jazayery, Dr. Wolfgang Lentz, Dr. John A. Moyne, Dr. Herbert H. Paper, Dr. Hasan Sharifi, Dr. Donald L. Stilo. I further thank Mr. Frank Shulman for his data on recent French dissertations. To all of these named I am deeply grateful, as I am to a large group of others, too numerous to mention, who have aided me in my work over the years. The bibliography would not have come about had it not been for the untiring help of research assistants, a selsele of four of my students (in temporal sequence): Dr. Karl Krahnke, Dr. Iraj Bashiri, Dr. David Peterson and Mr. John Workman. These four were joined by their fellow-student Ms. Judith Cederblom who not only typed the draft of this bibliography but also assisted me in correcting it. I hope they will find equally devoted fellow-researchers for their own work. Last but not least, I acknowledge with gratitude the support of this project by the John S. Guggenheim Memorial Foundation, the Joint Committee on the Near and Middle East of the American Council of Learned Societies and the Social Science Research Council, the continuous assistance by the Center for Near Eastern and North African Studies at the University of Michigan, and grants from the Rackham School of Graduate Studies of the same university. While so many people were kind enough to help, the responsibility for the many shortcomings of this attempt at a survey-bibliography lies with its author alone. G. L. W. Ann Arbor, Michigan 1975
FOREWORD
1. PREVIOUS S U R V E Y S
The history of Iranic studies has been surveyed repeatedly. The most comprehensive collection of research centers and scholars was without doubt compiled by §afa (1969). Among shorter surveys of scholarship one may cite Dresden (1968: general). In terms of countries the Russian/Soviet scholarship has been most extensively reviewed (e.g. by Bertel's [1927]: the years 1914-20; B. Miller [1946]; Radovil'skij [1958]: history of teaching; Abaev [1959]: comparative-historical studies; Oranskij [1960]; Rastorgueva, [ed.] [1962]; Giunasvili [1971]: Georgia). The history of Western scholarship is much less extensively covered (e.g. Morgenstierne [1941]: Scandinavia; Arberry [1942]: British contributions to Persian studies; Spuler [1962, 1963]: German work, quite general; cf. also the various numbers of the Rähnemä-ye Ketäb, the new Acta Iranica, and Studia Iranica where similar brief surveys are included). Iranian scholarship was surveyed by Yarshater (1970a) and by Afsar (1971a, b). The politics of Iranic studies and its future were discussed among others by Frejman (1946) in the Soviet context and by Vahman (1969), Tabätabä'i (1972) and Enäyat (1972) in the Iranian context. There have appeared but a few surveys of Iranic linguistics proper. As an introduction to the field the best surveys to date are undoubtedly those by Oranskij (1960, 1963b) which ought to be translated into a western language. More specialized and/or detailed surveys are rare. The first (and latest) comprehensive survey is now eighty years old, namely the Grundriss der Iranischen Philologie (1895-1904) which includes Horn's contribution on Persian and contributions mostly by Geiger on West Iranic dialectology. The state of the study of Iranic dialectology and Persian was quite briefly summarized some forty years later by Bailey (1936b). More recently, the volume on Iranic linguistics of the Handbuch der Orientalistik (1958) includes: 1) a survey of Iranic dialectology by Morgenstierne; this, except for a few pages of discussion of general problems, is a brief geographically arranged description of dialects and some of their major features with notes on comparative-dialectal problems; 2) a survey of Modern (Standard) Persian by Lentz; this is a fairly detailed presentation of many problems of Persian grammar with suggestions
4 for their solution from the point of view of descriptivist-structuralist grammar. As such it is not only quite original - and idiosyncratic - in approach, but also very stimulating. The most recent surveys are included in volume 6 of Current Trends in Linguistics (1970) : 1) by Lazard on the study of Persian and Tajiki, 2) by Redard on Iranic dialectology, and 3) by Yarshater, the first concise survey of the study of Iranic linguistics in Iran and Afghanistan. (It is worth noting that Kurdish was totally excluded since the author of the scheduled article did not submit it in time for publication.) These three articles are essentially annotated bibliographies which focus on the scholarship since 1950 providing, however, also important background data (especially Yarshater). They cover a wide range of topics with the notable exclusion of etymology in Lazard's and Yarshater's contributions (similar to Lentz [1958]). The study of etymology has (again) been recognized as a linguistic study separate from grammar (see also Reichelt [1924-1927] and C. F. and F. M. Voegelin [1965]).
2 . THE PRESENT SURVEY
2.1 Scope The objective of the volumes in the state-of-art reports subseries of Trends in Linguistics is to survey the history and the state of research in the various disciplines of linguistics and language areas. Iranic languages and linguistics have been allotted several volumes. The first volume to be delivered (still within the context of the Series critica) was that by Oranskij on Iranic languages in the Soviet Union for which he could consult a uniquely active tradition of surveys of scholarship in that country (a tradition of which he is a prominent representative). The area of Oranskij's contribution to the Series critica is essentially politically/geographically defined; linguistically, it includes members of all major subgroups of Iranic. Similarly, the present contribution is basically politically/geographically defined; it mainly treats the languages in the country of Iran. Dialectologically speaking, it covers most of the West Iranic dialects, foremost Persian, the Hochsprache and lingua franca of the entire Iranic Kulturbund for now more than one thousand years. Certain adjustments have been made; largely excluded are: a) Kurdish since it is to be covered in a separate volume of the present series; b) Tajiki-Persian since it was covered in Oranskij's monograph; included are, however: a) major eastern Persian dialects spoken in Iran and Afghanistan and, formerly, in India; b) TatPersian spoken in the East Caucasus; c) Kumzari spoken in Oman; d) Baluchi spoken in Iran, Pakistan, Afghanistan and Central Asia; e) Talisi spoken in East
5 Azarbayjan on both sides of the Iran-Soviet border; f) Gurani spoken on both sides of the Iran-Iraq border; g) Zaza spoken in the Dersim district in East Anatolia. Given the amount of research in the specific areas of West Iranic the present survey will consist of two parts: volume 1: Persian grammar, and volume 11: West Iranic dialectology, including the dialectology of its major dialect, Persian (but excluding Tajiki-Persian and Kurdish, as indicated). These two volumes are accompanied by an alphabetically arranged bibliography. In the present volume, the topical bibliographies on Persian grammar appear together in the appendix.
2.2 Approach to this volume The objective of this survey is to describe the history and the present state of scholarship in the field of Persian grammar, not so much in terms of chronologically arranged annotated biographical lists, but rather in terms of what has been called scientific paradigms; not to present an all-inclusive Handbuch, but a critical, quite selective, though substantial, survey of the major achievements in the analysis, description, and interpretation of the linguistic data. The brief historical surveys of the major topics are followed by relatively detailed presentations of particular problems and recent suggestions proposed for their solution. These, in turn, are critically reviewed and suggestions of my own are added, not as final solutions but as points for discussions. On the one hand, such an approach may help provide the non-specialist with the basic linguistic facts so as to allow him to follow the main practical and theoretical arguments. On the other hand, it permits the brief presentation of the various theoretical approaches — and their advantages as well as disadvantages — to the Persianists and Iranists who, similar to scholars of other language areas, are considerably divided among themselves as to their linguistic convictions. In fact, some modern theoretically, or at least school oriented, groups of scholars of Persian are little concerned with, or aware of, Iranistic scholarship outside their own approaches. Likewise, an admittedly small group of 'older' scholars have doubts about the necessity of even considering the theoretical ambitions and premises of such approaches as generative grammar. Much of this can be attributed to the misunderstanding, by these^groups of scholars, of each others' objectives. This problem is by no means a recent one ; it was pointed out in the lucid, evidently forgotten theoretical portions of Lumsden (1810) who pleaded for the objectives of 'universal, philosophical grammar' as being totally separate from those of etymology (i.e., comparativehistorical grammar which was just gaining ground at that time).
6 The overall approach of presentation here is not to defend any one of the theories or beliefs, but to list and discuss, in rather pragmatic fashion, what a particular scholar has done to contribute to the elucidation — or even solution — of particular grammatical problems. The often pervasive and sometimes amusing animosities were not reported on. The attempt here was to evaluate and appreciate, not to join in fashionable, and easy, derision. Given the fact that generative-transformational grammar is the most recent approach in Iranic linguistics, there is a certain amount of generative formalism, including generative 'trees'. Such diagrams are convenient means for stating explicitly the implicit relationships; it is simply not true that they only restate 'known facts', which is easy to claim in hindsight; many of those interrelationships were not known, but at best intuitively recognized or felt. There is no doubt that, in terms of new discoveries, some such insights are perhaps not, or not entirely, new; some are evidently wrong or banalities - yet one must admit that such relationships as that between 'underlying' and 'surface' cases were never systematically investigated, and could not be formulated. More importantly, the descriptive mechanism available today permits the systematic testing of the correctness, or falseness, of an analysis. Even today, many studies are taxonomic descriptions and interpretations without an investigation of their validity by a systematic analysis of counter-examples, and/or attempt to define classes, rules, and terms. The selection of topics discussed will be biased. While etymology cannot properly be considered a part of grammar (although an important field of linguistics), the following topics are rather minimally discussed because they have so far defied any systematization due to the grammatical complexity involved: derivation by affixation, phraseology, stylistics, rhetoric, discourse and text-structure, sociolinguistics, and contrastive grammar. The dialectology of Persian and related problems will be discussed in volume 2. The main attention is paid to nominal and verbal syntax and morphology and, to some extent, to phonology. In preparing this survey I have tried to get hold of as many studies as possible. Yet only about half or perhaps two thirds of the western and Iranian and about one third of the Soviet publications were accessible to me. Access to many is extremely difficult, especially in the Soviet Union, even for Soviet scholars, as is evident, e.g., from the considerable number of incomplete citations in Rastorgueva et al. (1962). This is the more regrettable since more than half of the scholarly work on Iranic linguistics is carried out in the Soviet Union. In spite of these difficulties I hope to have been able to draw the main outlines, sometimes even more detailed lines, of the progress in this fairly comprehensive survey of the state and history of Persian since its beginning
7 in the 17th century. Finally a note on the kind of Persian discussed. A definition is impossible; the contemporary 'standard' dialect may be circumscribed as follows: it is the socially most prestigious dialect of Tehran which is not yet, but is fast becoming, the standard dialect of Iran. As such it is the geographically least marked dialect as opposed to the local dialect (Mundart) of Tehran. It has a written and a spoken, informal variant, here called contemporary colloquial, i.e. the style, which according to Hodge (1957: 366) is used between persons very well acquainted, though they may not be on familiar terms as regards politeness level.
3. TRANSCRIPTION
The transcription follows the Persian phonological system, which may be listed as follows: p t c k b d j g
f s s x vzzq
mn l r
yh'
ieaey uoäow
Ezäfe, rä and pronominal suffixes are connected by hyphen; the suffix / after vowels is preceded by e.g. xane'i. The following diacritics are used to distinguish Arabic letters in the chapter on phonology and in Arabic titles: te
tä
/
ze
qdf
'qeyn
/
'alef
zä $eyn
zad
zäl
/
sin
/
he
sad
se
he(-yejimi)
1. THE STUDY OF PERSIAN GRAMMAR
1.0 INTRODUCTION
'Persian is a child's game by comparison with Arabic.' This remark by Friedrich Engels (Werke, Berlin 1963, vol. 28: 260-61) reflects the reaction of many who were confronted with the seeming simplicity of Persian grammar, in particular with its rudimentary inflectional system. In fact, it may be possible to read and speak Persian in a short time with the help of a dictionary. The barrier of inflection is almost absent; only the barrier of the writing system has to be overcome. The relatively easy access to content is a phenomenon of many 'simple' languages. Many morphological distinctions in other languages such as German are easily collapsed into more general, overtly less marked, rules of grammar. The learner whose native language is morphologically more complex intuitively recognizes in Persian the similarity of certain general rules in his own language and of those of Persian, and equates both. More often than not this leads to misinterpretation not always only of subtleties. While it is true that with increased exposure to Persian most learners acquire, intuitively, a certain working knowledge of Persian, the intricacies of Persian grammar have been consciously mastered by only a few, if any, foreigners. In terms of scholarly activity the grammar and lexicon of Modern Persian were first studied in the 11th century by Iranians, who were later joined by Turkish and then by Indian and other Muslim scholars, and finally, since the 17th century, by western scholars. The two main paradigms with which Persian grammar has been studied and described are the 'Muslim/Near Eastern' and the 'Western' paradigm. (A third paradigm, that of Indian (Sanskrit) grammar, is said to have been applied in some grammars of Persian written in India during the time of Akbar.) The 'Western' paradigm reflected in the study of Persian grammar is the sequence which may be described as follows: 1) descriptive-prescriptive 'Latin' grammar, 2) comparative-diachronic, 3) descriptive-synchronic and structuralist grammar, 4) generative-transformational grammar. In overview, from the grammatical philosophy of Avicenna to some of today's approaches to Persian grammar, one observes a dialectic progression from logical grammar to behavioristic
10 'surface' grammar (bringing about the separation between diachronic and synchronic grammar) back to natural-logical grammar. The study of grammar in Iran is of great age. Probably even during Sasanian times (224-651 A.D.) phonetics, morphology, and lexicon were being studied, mainly in connection with the task of preserving the old sacred (Zoroastrian) texts (cf. Reichelt [1924], Windfuhr [1971 ], Sädeqi [1971J). In Islamic times Iranians were the first to write grammars of Arabic. The first known discussion of Persian is by the famous philosopher Avicenna (Ebn-e Sinä, 980-1037 A.D.) in his Dänesnäme-ye Alä'i. Avicenna's approach to grammar was essentially philosophical and abstract; following the classical writers, he attempted to demonstrate the universality of language and how it functions to express the logical (theoretical) hierarchy of the 'essential' (cf. the notes in Bashiri [1972]). The point of view of Avicenna was thus as remote from the pragmatic descriptive/prescriptive view with which Arabic grammar was investigated — and codified — as are, e.g., some schools of generative semantics from descriptivestructural linguistics, or Saumjan's applicative linguistics from the work of most of his contemporary linguistic colleagues in the U.S.S.R., including Iranists. Sprachphilosophie never outlasted the challenges of natural languages, but it can have an extraordinary indirect impact on grammars since they are often ideologically conditioned and oriented. Avicenna belongs to the time of the reassertion of the Iranian Middle East, and thus of Persian as its main language. The attempts of Avicenna and others (like Biruni and Näser Xosrow) , to create Persian technical terms in lieu of Arabic ones (cf. Mo' in Dehxodä [1959: 63-70] or Ferdowsi's 'reduction' of his Arabic vocabulary (cf. Mo'infar [1970]) have repeatedly been discussed (cf. discussions of lexis). This ideological movement resulted in the gradual, initially only 'literary', Persianization of Western Iran, overlaying the 'Pahlavi'-dialects (cf. the discussion of the early stages of Persian). Dictionaries were among the first needs of Iranian non-Persian writers. Asadi's brief Lu'qat al-Furs (about 1050 A.D.), the oldest one preserved, was written at least partially to answer these needs (cf. Nafisi [1959 b: *69-70]'). Basic grammars as such seem, however, not to have been in much demand by litterateurs. Rather, Persian grammar gained attention first by way of the next higher liberal art, rhetoric. The first scholarly work to discuss a good number of rules of Persian morphology appeared in the 13th century, the famous book on Persian rhetoric by §ams-e Qeys of Rey. It already shows a well-developed terminology and most probably is based on earlier works. Little is known about other early grammars. One grammar entitled Mantiq al-xurs fflisän al-furs 'The eloquence of the mute with regard to the Persian tongue' written in Arabic by a certain Abu Hayän Nahvi (1256-1344 A.D.) is
11 known only by name. There exists one copy at least of a later grammar, also written in Arabic, entitled Kitäb hilyat al-' insän fi halbat al-lisan 'The book of the ornament of men on the race-course of the tongue' written by a certain Ibn al-Muhannä during the Ilkhan period (end of 13th or beginning of 14th century). According to Homä'i (1959: 124; cf. also Telegdi [1973]), who learned about this grammar from the famous scholar Mohammad Xän Qazvini, it is a description of three languages — Arabic, Mongolian Turkish, and Persian. The author remarked that in his time the knowledge of all three was a necessity. A few other grammars of similar age are listed by Afsar (1953) who was assisted by information from the scholars Minovi and Homä'i. The most important subsequent advance in both grammar and lexicography was due to the 'literary' Persianization outside Iran proper, i.e., of the Ottoman empire in the West and of India in the East, where Turkish and Indian scholars compiled the first extensive dictionaries, provided grammatical prefaces, and wrote other treatises; an endeavor for which not a few native Persian scholars were employed. Western scholarship, preoccupied with Arabic grammar, has done little to study Persian grammatical treatises and their theoretical background. Not only Avicenna, but many other classical scholars are likely to offer important insights for comparative linguistic theory. Those non-philosophical grammatical treatises and 'prefaces' did not, in their intent, differ much from grammars of European languages. They were descriptiveprescriptive, identifying the proper morphological, semantic, and sometimes symbolic use of the 'letters'. But unlike European grammars the basic order of presentation in the grammatical prefaces to dictionaries (e.g. Borhan-e Qäte*) was the alphabet, such that suffixes are listed by their final letter and prefixes by their first. Those letters constituted what may be called a 'littereme'; e.g. the functions of t are said to be two: 2nd singular pronominal suffix and 3rd singular copula; i.e., this t represented the morphemes -at and -ast. This was quite a practical approach with regard to the custom of arrangement of dictionary entries. Today Behruz (1972) is an illustrative example for this approach; e.g. under the heading 'M' he lists: (1) 'M' is a letter connecting left and right, (2) = ma- 'prohibitive prefix', (3) = -om 'ordinal number suffix', (4) = -am Ί am', (5) = am 'personal suffix 1 sg.' (6) = ma- 1 sg. for man in ma-rä 'me', (7) = -om, a rare 'feminine' (? ) marker asinxcrtom 'lady'. The modern reader with a different attitude towards 'letters', accustomed to grammar being separate from dictionary, has to work hard to collect and rearrange the various grammatical statements. Some of the few attempts at evaluation were made by Homä'i (1959) in very general terms. 3tolbova (1967) described some characteristic features of the linguistic introduction to the Borhan-e Qate' (17th century) and pointed out the problems of terminology (and the forms referred
12 to), the incompleteness and the often ad-hoc character of rules, and the repetition of morphemes in different contexts. As in the classical(ized) West, phonetics was considered a subscience not of grammar but of metrics. Together with traditional Near Eastern grammar, it had its highlights before the 15th cent. Thus, among others, Avicenna wrote Asbäb hudüt al-hurüf (also called maxarif al-hurüf 'articulation of sounds/letters', i.e. 'phonetics', and Näser Tusi wrote Mfyär al-ascär 'Measures of poetry', which contains a section on that topic (cf. Yarshater [1970 a: 671 fn. 9]). In 1958, Xanlari included an insightful discussion on sound in his innovative study of Persian metrics. The science of tafvid, i.e., the correct pronunciation of the Qorän, declined with the other sciences. Letters and pronunciation were, however, discussed in books on rhetoric and in the prefaces to the major traditional dictionaries (cf. discussion of lexicography). Nevertheless, such traditional statements have been helpful to earlier Western scholars of Persian such as Lumsden (1810), Splieth (1846), Vullers, (1840-50, 1870), Platts (1894), and Phillott (1919), who observed the 'hair splitting' attempts by native grammarians to list the nuances of words and affixes in various semantic contexts. As for theory, more recently Bashiri (1972) has tried to utilize some theoretical 'generative' notions of Avicenna in his own quite original theory of the generative grammar of Persian. Some notions of classical Arabic grammar such as lafz and ma'ni have been investigated by Peterson (1972) as to their relation to concepts of modern theory ; and Moyne (1971: 100 fn. 16) refers to some classical concepts in his study of emphatic xod.
1.1. STUDIES BY EUROPEANS
Europe was initially interested in Persian for practical, i.e., missionary and economic, and later, for literary and linguistic reasons. There is a relatively long list of manuscripts of Persian translations of parts of the Bible, the New Testament, and related literature translated between the 13th-l 7th centuries from Syriac, Greek, Latin, and later, in the 16th century, from French and Portuguese (cf. Lagarde [1884: 6]). The lack of translations of other texts during the earlier period is evidently due to the fact that only the Bible was considered worthwhile literature for translation. Translations of commercial and political communications of that period may be found, but they were unimportant for the course of the European study of Persian. Lagarde (1884) gives an annotated list of most of the printed Persian texts available to European scholars until 1700: a (Judeo-) Persian translation by Jacob Tävus of the Pentateuch, printed 1546 in Constantinople, which was reprinted in Walton's Polyglotta (1657); de Dieu's (1639)
13 reprint of the Persian translation of Xavier's Portuguese Historia Christi (translated 1602 for Akbar by a certain Abd al-Sattär) and, in his grammar (1639), a reprint of the first two chapters of the Genesis from the translation by Jacob Tävus mentioned above; the some hundred sentences in Warner (1644); the first European print of a Persian classic, of Sa'di's Golestän, by Gentius (1651); the Gospels edited by Pierson (1657); Hyde's (1665) translation of Ulug Beg's' Longitudes and latitudes of the stars'; further the many hundred sentences and phrases (with mini-grammar) by Angelus (1684). The 17th century, the time of Shah Akbar I, is thus the starting period of Persian studies in Europe. Side by side with direct 'field-research' in Persia, India, and Turkey, manuscripts in European libraries — be they Biblical translations or original manuscripts of literary works or works on grammar and, more importantly, lexica — began to be utilized for bi- and multilingual dictionaries and for grammatical sketches. As mentioned above, Western grammatical studies of Persian were initially based on direct observation in Iran. As one of the first preserved or known studies one may mention the scattered grammatical notes in the Codex Cumanicus (Bodroligeti [1971]) compiled in the 13th century (and thus contemporary with Sams-eQeys of Rey). The list of European grammars begins with Raymundus (d. 1614) of Cremona, who had been in the Orient. The existence of his grammar can be inferred from various manuscript catalogues. No copy has been found yet. Based on the available information, Xromov (1970) concluded that the grammar was probably first compiled by Raymundus in 1601 and then rewritten by one Ilaminio Clementino Amerino in 1614. This was followed by the 95 pages of de Dieu (1639), published in Leiden, a grammar which influenced several subsequent studies. (De Dieu is probably the pseudonym for Johann Elichmann, cf. Hammer [1833: 49, 67].) De Dieu's grammar is the result of his study of Xavier's Dästän-e Masih 'Historia Christi' and the Historia Petri as well as Tävus' translation of the Pentateuch (for both, cf. Lagarde [1884: 3-4] ).He wrote it upon the request of the editors of the two Historiae. It was meant to be a tentative introduction only, and he welcomed the cooperation of others for improvement and additions. One of the interesting features is that he derives the present stem from the past stem (pp. 28-35), offering the first of many attempts to describe systematically the irregular verbs (cf. the attempt by Barb [1860], who proudly considered himself to have found the best solution, or the most recent attempt by Extiär [1972], who applied the formalism of transformational grammar). Another interesting fact is that he offered lists of those verbs which most frequently occur in compounds (pp. 16-22), distinguishing between such verbs which form active and those which form passive compounds.
14 Gravius (1649) was published soon thereafter in London. Again only a few years later followed the brief treatises by Ignatius (1661) in Rome, twenty years later Meninski's (1680) introductory grammatical sketch in Vienna. One outstanding contribution to the knowledge of Persian, even compared with many modern textbooks, was published by Joseph Labrosse, later Pater Angelus a Sancto Josepho, (1648) in Amsterdam. It was written as an 'exercitatio in controversiam maximam cum Persiis, Turciis, de Divinitate Christi' for missionaries, language professors, merchants, and others. It is a voluminous practical textbook with a 'mini-grammar' of 13 folio pages and 474 pages of words, phrases and sentences in Latin, Italian, French, and Persian with separate glossaries. It is based on the living language. He wrote it after he had studied for 14 years in Esfahan with Balthazar the Portuguese (Lusitano) whom he praises greatly. Most of the major morphological features are noted including such observations as the identification of the 'neuter' verb form (i.e. intransitive compounds with transitives such as ruze dästan 'to fast') and the distinction between karde hast: 1. factus est, 2. fecit. It is interesting to note that his derivations of forms are done in the fashion of traditional Arabic grammar; thus the present stems are derived from the past stems, and endings are conceived of as 'substitutes' for infinitive ending; e.g. xord-am ' I ate' is derived from xordan 'eating' by dropping an and adding am. (Jones [1771: 41] still followed this analysis; so did Vullers [1870: 107]). The same approach can still be found in Qarib et al. [1950? : 119]). Given the fact that the Portuguese were active in the Persian Gulf during Safavid times, Portuguese dictionaries and grammars, or at least word lists, may be found yet. After these and other early grammars based on Persian as spoken by Persians begins what might be called the period of the study of 'colonial', i.e., Indian Persian, coinciding with the declining interest in Persia and the ever increasing attraction to Indian riches. Indian scholars and literati produced a large number of grammars and textbooks of Persian both in Persian and in English (cf. Afsär [1953]). It was Indian grammars, in particular the prefaces of the Farhang-e Raadi (compiled 1654), the Borhän-e Qäte' (compiled 1651) and the Haft Qolzom (compiled 1538-1545) which were utilized henceforth by Western grammarians such as Lumsden (1810) and Splieth (1846). The British adventure in India gave the study of Persian an extraordinary momentum. The first influential grammar, practically rather than linguistically oriented, was compiled by Jones (1771) at a high tide of the British interest in India. Jones claimed that with his grammar and a living instructor the learner would be able in less than a year's time to translate a letter from an Indian prince and to converse. That he used de Dieu (1639) can only be deduced from a rather unobtrusive note (bottom of p. 31). Some valuable insights of Jones, apparently based on his knowledge of
15 traditional Perian grammar, are the recognition of the collective, i.e., generic, function of the simple noun vs. the 'particular or limited' noun marked by i and ra\ the oblique function of the personal suffixes for the genitive, dative and accusative; the difference between the emphatic and the reflexive xod. Jones' grammar is by no means free of apparent mistakes, some of which — like bude sodam said to be the 'preter-pluperfect' o f ' t o be' (pp. 37-38) — might be due to his Indian informants and thus to lingua franca rules developed by non-native speakers. (One should mention here a passive paradigm porside sodam Ί was asked' by the Persian Ibrahim in Ibrahim-Fleischer [1875: 64-65 J, which does also not agree with modern Persian.) Nevertheless, this work had a great effect and was republished many times perhaps largely because of the commercial need for such a practical work. More than a century ago Chodzko (1852: 187-88) stated briskly:'Jones, Gladwin et Lumsden, dont les ouvrages ont servi la base ä toutes les grammaires (footnote: 'excepte celle de Mirza Ibrahim'), avaient appris le Persan aux Indes.' later: 'c'est un dialecte importe par les Mongols et adopte plus tard comme une langue officielle, dont jusqu'ä present se servent les musulmans indigenes et la noblesse lettree des Hindous.' Giving some example of the difference between Indian and 'Persian' Persian he stated: 'Tout celä est intolerable dans une grammaire persane (his italics) du XIX e siecle.' The appearance of the grammar by Chodzko, a diplomat, again coincides with a major political shift, the first phase of Iran's self-reassertion under the Qajars. He was the first European to describe the Persian spoken in North Iran, and to include many observations on the colloquial dialects, without, however, always identifying them as such. Eleven years earlier, the Persian Ibrahim (1841) had written a very useful practical textbook, utilizing a sequence of conversations between a Westerner and a Persian to exemplify the points made in the brief grammatical sections. In these sections he attempted to discuss the problems of Persian without the distortion of the preconceived European grammatical model. It was soon translated into German and somewhat rearranged according to European models by Fleischer (1847, new edition 1875). At the same time Persian and other Iranic languages began to be recorded from the north, from Russia. After early reports such as the 15th century travelogue by Afanasij Nikitin, who included Persian and Turkish notes (see A. Zaj^czkowski [1951-52] and Ligeti [1962j), the records began to increase in the middle of the 18th century. Commissaries and emissaries of the Russian imperial court were sent out to explore the empire. Their records provided valuable data for the awakening interest in Europe, including Russia, in the languages of the world and their interrelationship. In the second half of the 18th century, Bachmeister urged their collection and devised questionnaires,
16 the Bachmeister'sche Sprachproben (utilized e.g. by Zukovskij [1888J). Within the Russian context, Persian was first recorded in the form of its Buxärä-dialect, i.e., Tajiki (e.g. by one G.F. Miller in 1735). In 1786 appeared the monumental dictionary by Pallas comparing 200 words in 149 Asian and 57 European languages, among them Buxärian Tajiki, Kurdish, Pasto, and Ossetic. Gmelin (1788) included glossaries of Persian, Gilaki, Turkish. Falk (1796) published a glossary including Tajiki besides the Turkish languages Kirgiz, Kalmuk, Kazax. In 1832, in Tiflis, appeared a Persian grammar by Abbas Qoli Bakixanov written in Persian, which was translated into Russian ten years later as Kratkaja grammatika persidskago jazyka. It is probably the first Persian grammar published in Russia. The first Russian to publish a textbook of Persian was Berezin (1853b) with the assistance of the Persian Mirzä Käzem Beyk. Similar to Ibrahim in England, Iranians were highly influential in the course of Persian studies in Russia. In 1833 appeared the conversational textbook and in 1884 the grammar book by Mirzä Dzafar, Persian teacher at the Lazarevskij Institute of Oriental languages. This was followed by the publication in 1890 of one of the outstanding grammars of Persian, written by two leading Iranists, Salemann and 2ukovskij, the latter being the first, and one of the great, dialectologists of Iranian. The excellence of this grammar is probably largely due to Zukovskij who had years of field experience in Iran. Thus, besides classical Persian, it describes the contemporary pronunciation, includes idiomatic phraseology, not only in Arabic-Persian script but in Russian transcription. In Russia as in Western Europe there appeared a steady stream of guides, textbooks and practical grammars. Among them, special mention must be made of Barb (1886a,b), which deserves more notice than it has received. It is the first Persian grammar which does not use Arabic-Persian script at all but instead a transcription using modified phonetic symbols of the time (which makes it somewhat difficult to read). Another interesting feature is that Barb used what may be called phonemic transcription. The rules for conditioned changes are listed in the chapter on Lautlehre (those rules seem to reflect the features of affected speech). The unsurpassed, most inclusive, though unorganized, grammar is certainly that by Phillott (1919) which also includes notes on the carefully identified eastern, i.e. Afghani and Indian, variants of Persian, as well as on colloquial Persian. The stream of textbooks and pedagogical grammars, etc., continues today. Among the many methods used one may mention the structuralistically oriented textbooks which have appeared in North America since the post-Sputnik enthusiasm for international studies; among these the most used were Hodge (1960) and Obolensky etal (1963). As their basic technique they used the
17 so-called build-up (i.e., listing the words first, followed by the complete sentence, as a part of the conversation) and various forms of drills combined with language laboratory usage. It should be noted that they emphasized the (educated) colloquial style of Persian. 1.1.1. Theoretical approaches While most of the aforementioned Western grammars are more practically oriented as guides for the traveller or the reader of Persian literature, there is a relatively small number of grammars which consciously applied linguistic theory. Lumsden (1810.1: III) like others also applauded Jones for his 'popular performance' because it increased the 'number of European votaries to that department of Oriental Studies', but he is quite sarcastic and by no means favorable toward the 'execution' of the grammar. Lumsden's two huge volumes constitute indeed one of the most remarkable achievements of his time, a fact which has been undeservedly overlooked for a long time. It is a combined theoretical and descriptive-interpretative grammar which partially reads like a discussion of presentday linguistic theory. Thus he strongly made a case for describing Persian in its own terms, not in terms of the grammatical canons of the East or the West which do not fit Persian. Instead, features are to be described as to their agreement with, and difference from, universal philosophical grammar. Only the latter is the domain of the grammarian; etymology and philosophical grammar are entirely different and unrelated. Lumsden's extensive discussions on this topic reflect the beginning disputations between the adherents of universal grammar and those of comparativehistorical grammar, the latter of which were to dominate for some 125 years before the resurgence of 'philosophical' grammar in the last twenty years. While he remarked that grammatical rules are no help for the beginner (p. XXVI), he rejected comparative-contrastive English-Persian tables for his grammatical description as providing little insight into Persian (1810.2: 320). Some of the insights which appear to come out of modern discussions but were obviously derived from Arabic grammatical theory were: a)
the recognition of the semantic equivalence of utterances like zeyd bakr-rä dänä dänest zeyd dänest ke bakr dänä-st zeyd dänä'i-ye bakr dänest Zeyd knew Bakr to be learned Zeyd knew that Bakr is wise Zeyd knew the wisdom of Bakr (1810.2: 359);
b)
the assumption of ellipsis of (underlying) verbs like Ί swear', Ί call' in
18 vocatives and oaths (the vocative is considered a kind of maf'ul 'object' in Arabic grammar [1810.1: 209,2: 57]); c)
the analysis of inteijections like guya 'say, as if and pendari 'you may think, may be' as (underlying) conditional clauses (1810.2: 366-67).
Finally one may mention his reference to a linguistic article in the contemporary Encyclopedia Britannica where in almost generative fashion verbs are identified as a species of attributes and where reference is made to Persian lexicographers who identify the connective e of the ezäfe- construction as originally being the predicative 'is' (this, however, Lumsden rejected [1810.2: 294]). After Lumsden, two major turning points can be identified: the 40-50's of the 19th century with their emphasis on comparative-historical grammar, and the 40-50's of this century when structural grammar was applied to modern Persian. Both approaches aimed at better insights into the workings of modern Persian. The objectives and expected results of the two approaches, however, differed greatly as to their basic linguistic models or paradigms. Comparative-historical grammar researched the diachrony and diatopy of Persian viewing its modern system as the final result of a series of changes from the depth of history to the present surface. In this view modern Persian appeared as the result of a continuous sequence of 'rule-deletions' and 'rule-additions' to use two terms now en vogue. 'Quam vero necessaria sit et harum (i.e. Old and Middle Iranian) dialectorum et linguae sanscritae comparatio ad persicam linguam critica ratione examinandam recteque perspiciendam (sic! ) ex ipsa, quae inter illas est, cognatione intelligitur, tum inprimis eo potest probari, quod plurimae verborum persicorum formae et immutationes ex illis sunt deducendae atque explicandae.' (Vullers [1870: VII]; Lagarde [1884] deplored Vullers' 'horrible' Latin). The turn by the middle of this century was caused by a belated recognition of the major views of structural grammar. In a dialectical turn-about, the diachrony and diatopy of Persian were placed in the background, and the main emphasis was put on the synchrony; with the pragmatic attitude of 'first things first' the contemporary linguistic system at any one point in the development was focussed on. In practice this meant mainly educated standard Persian of today. Insights into Persian grammar were sought from the living informant who was to be skillfully questioned, rather than from dictionaries and old texts. This approach ended once and for all the assumption implicit in earlier studies that Persian was somehow a monolithic unchanging entity the purest example of which was Ferdowsi's Sähnäme (early 11th century); this was expressed by Darmesteter (1883: 3) as follows: 'La langue de la Perse moderne, dont le type le plus pur se trouve dans le Libre des Rois de Firdousi (X e siecle) et qui n'a point subi d'alteration sensible depuis cette epoque jusqu'ä nos jours'. Darmesteter's
19 view persisted side by side with the better insights in the many changes in the millennium of Persian by Iranists like Zukovskij (cf. preface Salemann-Shukovski [1889] ). Once the stability of Modern Persian was recognized as a myth, the various periods of its development began to be recognized. Most important was perhaps the recognition of the socio-linguistic layers of Persian, the difference between contemporary literary standard Persian and the layers of'colloquials', although the latter had been referred to before. The first detailed comparative grammar was written by Vullers (1840-50, expanded and revised 1870). He relied on Lumsden (1810) and native grammars, but also included notes on colloquial, i.e., modern standard, from Chodzko (1852) and Ibrahim-Fleischer (1847). Vullers' main contributions are his comparative notes and a rather concise description of disperse material on the morphology of Persian. Subsequently, the most highly esteemed study was that of Darmesteter (1883), an historical grammar of Persian with reference to its earlier stages, Old and Middle Persian (it is a reduced version of a manuscript that had been honored by the Prix Volney in 1881; Vullers had been awarded half of that prize in 1867). In it he places emphasis on the facts on 'syntax' (= morphology), as it is intricately interrelated with morphophonology and its changes 'parce qu'au fond du changement des formes persanes se trouve un changement de construction' (p. VII). The first part of his discussion consists of a brief survey of the difference between (Old) Persian and Avestan, for which he refers to Spiegel (1851) and Schleicher's Compendium. He begins his study of modern Persian with a note on the 'decomposition' of the Persian forms which happened essentially during the transition from Old to Middle Persian. He saw the grammatical structure of modern Persian as already established in Middle Persian. Modern Persian had become, he said, one of the most analytical and one of the structurally simplest members of the Indo-European family. He presented a skillful compilation of the knowledge accumulated by his time, better arranged than those of his predecessors and supplying more explicit references to Old and Middle Persian. (Like Jones but entirely unlike Vullers, he prefers to reduce acknowledgements to a minimum. Thus the 'feminine' ending u in bänu 'lady' and yarn 'friend' [p. 134] or the relative construction man ke pedaram [p. 179] are evidently literal copies from Vullers [ 1870] without any mention of him. Vullers is mentioned in a footnote as the single source for Darmesteter's four pages on syntax.) The scholarly, historical-comparative attitude is reflected in the first grand review of Iranistics, the Grundriss der Iranischen Philologie (1895-1904), where Horn in the part entitled 'Neupersische Schriftsprache', devoted some 80·% of the description to etymological and diachronic observations and discarded syntax, i.e. clause syntax, altogether. It is still the most detailed and concise description of that approach.
20 One of the outstanding, still partly historically oriented, descriptive grammars of the first part of this century was written by Jensen (1931), who not only includes historical comparison but likewise compares pre-classical, classical, and modern including colloquial Persian, though not consistently. Particularly noteworthy is a long chapter on syntax, which he based on Phillott (1919) and on his own research on published texts in colloquial Persian. His grammar represents the transitional stages between the two main approaches to Persian. One of the major contributions to the study of Persian syntax is Arends (1941). With a clarity and straightforward descriptive approach similar to Jones' (1771) he listed and discussed many of the basic issues of Persian syntax. Against the background of the work: of his predecessors the relatively refined simplicity of Arends' contribution stands out as a land-mark. At about the same time the phonology of Persian began to be reconsidered and defined, initially in passing by non-Iranists (by Trubetskoy [1936] and a few others, cf. Kramsky [19391, bibliography), then, since 1939, by Krämsky in a series of studies following the methodologies and theories of the Prague school, and by other scholars. Phonetics, phonology, morphology, and syntax suddenly came to bloom in the fifties, most prominently in the U.S.S.R. That was the time when Stalin's statements on language found the latter less directly conditioned by the rules of historical materialism than other superstructures (cf. Frejman [1951] ). Sokolova and Zav'jalova conducted extensive phonetic experiments on Persian and other Iranian languages and dialects. Rastorgueva and Pejsikov and later Rubincik approached questions of Persian morphology, phrase syntax, and sentence syntax with the by now typical combination of comprehensive review of the contemporary linguistic 'corpus', or data, with taxonomic arrangement accompanied by carefully phrased tentative interpretations, which, being focussed on the particular taxonomic group under consideration, do not infrequently contrast with, if not contradict, conclusions about other groups (cf. Hincha [1961: 182]). In the West, the Prague school approach had merged with the structuralism of American linguist-anthropologists. Nye's (1955) dissertation on standard and colloquial Persian is the outstanding — and rather lonely — work of its time. In Western Europe, Lazard (1957) represents the pragmatic, in-between approach, describing the phonology, morphology, and syntax of contemporary literary Persian but including extensive notes on the colloquial. It is structuralist to the extent that many statements are based on criteria of substitution, testing, and classification, but it is basically taxonomic-descriptive, arranged according to established grammatical categories. This pragmatic attitude has made it the outstanding recent descriptive-structuralist grammar of contemporary Persian. As such it is not unlike Darmesteter (1883) which also provided a comprehensive, clarified, and prag-
21 matic summary (of the historical grammar of Persian). Prior to the major shift discussed above, the study of Persian was marked by the fact that the vast majority of articles focussed on etymology and/or Iranic dialectology, whereas articles on special problems of Persian grammar were few. Such problems of Persian were generally discussed collectively in grammars and textbooks. But since the late forties there has been an ever increasing stream of special studies on grammar, while the number of etymological studies has drastically declined. This phenomenon is clearly the result of new methodologies and theories: following the Soviet descriptive pragmatism and structuralist grammar of the fifties, there were applied such theories/methodologies as Neo-Firthian grammar (Bäteni [1967b] ), Geneva and Martinet School of linguistics (Miläniän [1965], Sädeqi [1967]), transformational and generative grammar (Moyne [1970], Bashiri [1972] Peterson [1974]) as well as contrastive and pedagogic methodologies (Yarmohammadi [1965], Marashi [1970]), to name only some of the more comprehensive studies (all of those mentioned are doctoral dissertations). One may add here that, except for Bashiri and Peterson, these authors are unaware of — or do not cite — the considerable number of special studies which have appeared mainly since the sixties, nor do they consider earlier grammatical studies with the general exception of Phillott (1919), Lambton (1953), Lazard (1957). This oversight is probably the result not only of the enthusiasm for the new methodologies and theories but also of the lack of communication in all its aspects such as the problem of accessibility of the literature, lack until recently of comprehensive bibliographical surveys of Persian linguistics (although the Bibliographie linguistique recorded most of the relevant literature) and most importantly, it seems, problems of communication between the various linguistic schools. The significance of studies such as the dissertations mentioned above for the study of Persian lies in the fact that most of them were primarily linguistic and only secondarily orientalistic dissertations. They evidence perhaps the most crucial revolution in Persian studies and herald the establishment of the theoretically oriented 'linguist' as a viable alternative, if not substitute, for the area specialist 'orientalist'; we have witnessed an epistemological shift towards the priority of the disciplinary approach over the area-specialist approach. Thus articles increasingly discuss general linguistic problems with Persian as the main language investigated, but in an attempt to contribute to general linguistic theory (e.g.Bäteni [1970a]on relativity in language; Moyne [1971]on reflexivity and emphasis in language; or Bashiri's 1972 version of deep semantics). Scholars of Persian studies like scholars in other area-specific fields have become aware of a built-in dichotomy between language-area specialization and disciplinary specialization. One can only hope that scholars will find a way of avoiding either extreme and learn to acknowledge the necessity of combining both approaches.
22 1.2. STUDIES BY IRANIANS
The dissertations cited above (except for Peterson [1974]) were written by young Iranian scholars. They are evidence for the active, increasing role of Iranians in the study of Iranian linguistics. They continue a tradition which —after a long lapse — began anew toward the end of the 19th century. According to Homä'i (1959: 125-26), the first work to include syntax appears to have been written by Abdolkarim Iraväni (lithographed 1845 in Tabriz). However, in Homä'i's opinion, the first modern grammar by an Iranian was written by Mirzä Habib Esfahäni (1890) in Constantinople and widely used there although unknown to the Iranian public. It was instigated by a 'German named Vamberg' who studied with him. The book included several Persian-German comparative tables (Afsar [1953: 31 fn. 3]).Esfahani was probably also the first to use the Persian word dastur for grammar. The first grammar of Persian for use in schools, later reedited more than twenty times, was a small booklet written by one of Iran's most prominent scholars, Mirzä Qarib (1911), who utilized European, mainly French and English, arrangements. (According to Sadiq [1966], Persian grammar was introduced as a subject in the prestigious first college, the Därolfonun, only in 1914). The best-known sequel of the latter was written (1953) by Qarib and four other eminent professors of Tehran University, Bahär, Homä'i, Foruzänfar, and Yäsemi; it is much more extensive than the above and includes notes on syntax. Like many other grammars it is essentially oriented towards classical or correct Persian. Two outstanding contributors to Persian linguistics in the thirties whose fundamental impact is only now beginning to be recognized were Fo'ädi, who followed the Russian approach to linguistics, and the reformer Kasravi. Their linguistic work and its importance has been little assessed so far (cf. Aliev [1961], Yarshater [1970a: 672]). In overview, it can be said that there were a few Iranian scholars highly competent in various disciplines of the humanities who contributed to the study of Persian in Iran. They collected, sifted, and began to analyze the linguistic heritage of Persia; thus Mo'in, to whom we owe the most complete surveys of traditional grammatical analyses, covering a large range of morpho-syntactic topics; thus Bahär, whose three volumes contain the first and, as of now, the only systematic study of the stages in the development of Modern Persian from its beginnings to the present, a study which covers and combines the grammatical as well as rhetorical/poetical aspects. They and their colleagues fulfilled a function in the history of the study of Persian which is almost uniquely that of Iranian scholars: their work makes it evident that they recognized their role of re-asserting the indigenous scholarship and of preparing it — so to speak — for the scrutiny of
23 the subsequent Iranian and non-Iranian linguists. In doing so they made 'giant leaps' in specific areas, e.g., Mo'in's ( 1 9 6 3 ) innovating approach to lexicography. At present, one of the most active instigating and innovating 'mediators' in Persian synchronic and diachronic grammar and lexicography is undoubtedly Xanlari. The list of grammars of Persian published in Iran is long and has continued to grow, especially since the Second World War. It is true, as has been remarked repeatedly, that although they have incorporated certain Western notions o f analysis and description, the Persian grammarians largely follow the traditional Muslim approach and are preoccupied with prescriptive statements and stylistic advice, generally failing to distinguish between contemporary Persian and the monolithic learned language called Classical Persian. However, similar observations can be made with regard to most grammars written in the West which adhere to a modified model of classical 'Latin' grammar. Further, while the observation o f the (educated) spoken language began quite early (e.g. Angelus [ 1 6 8 4 ] , Chodzko [ 1 8 5 2 ] ) , it is only since the fifties of this century that a clear separation is seen in most grammars, between the description of 'classical' Persian and contemporary Persian (Boyle [1966] still combines both, although he generally, often obliquely, identifies the two stages). As indicated earlier, both in Persian-speaking countries and elsewhere the traditional approach to grammar is increasingly countered by approaches based on structuralist premises and their more recent derivations. There is a marked difference o f 'didactics', so to speak, between linguistic studies published in the West and those published in Iran. In the West scholars presuppose a knowledge of current linguistic theories, thus invariably stunning their more conservative colleagues, if not actually antagonizing them. In contrast, scholars in Iran spend a large part of their efforts in popularizing and explaining their theoretical premises (as striking examples one may cite Bäteni [ 1 9 6 8 ] , Miläniän [ 1 9 7 0 ] on neutralization; Badre'i [1967] on semantics). The present state of linguistic studies in Iran and Afghanistan has been described for the first time in a refreshingly clear and concise fashion by Yarshater (1970a). Avoiding the chronological, annotated-bibliography approach, he surveyed grammar, dialectology, lexicography, modern linguistics, and language teaching, adding numerous details about the scholarly and intellectual background as well as providing some highly significant grammatical observations o f his own.
1.3. APPENDIX: GRAMMARS BY EUROPEANS
There are probably more than a few scholars who have been surprised at the early beginning and proliferation of Persian grammatical studies in Europe. After
24 the appearance of a moderate number of Persian grammars between 1600 and 1799 there has been a virtually never-ending 'deluge' of grammars since 1800. They have appeared in such rapid succession that, if one included the re-editions of many of them, there would be long stretches during which almost annually a grammar of Persian was published. This is demonstrated by the following fairly complete list of grammars of Persian (including grammatical prefaces) published in Europe or by Europeans between 1600 and 1900 (only first editions are cited): 1600-1699: Raymundus (Rome 1601 or 1614? ), de Dieu (Leiden 1639), Warner (Leiden 1644; example sentences), Gravius (London 1649), Ignatius (Rome 1661), Meninski (Vienna 1680), Angelus (Amsterdam 1684), Podesta (Vienna 1686); 1700-1799: Otho (Frankfurt/Main 1717), Zanolini (Padua 1747), Jones (London 1771), Hadley (Bath 1774), Bieyra (Dublin 1789), Moises (Newcastle 1792), Ouseley (London 1795); Gladwin (London-Calcutta 1796). 1800-1899: Dombay (Vienna 1804), Wilken (Leipzig 1805), Rousseau (London 1805), Lumsden (Calcutta 1810), Gilchrist (London 1821), Salih-Price (London 1823), Forbes (London 1828), Andrew (London 1830), Possart (Leipzig 1831), Vullers (Giessen 1840, vol. II 1850), Ibrahim (London 1841), G. Rosen (Berlin 1843), Ballantyne (London 1843), Geitlin (Helsingfors 1845), Splieth (Halle 1846), Ibrahim-Fleischer (Leipzig 1847), Falconer (London 1848), Berezin (Kazan 1853b), Nicolas (Paris 1857), Bleeck (London 1857), Schultze (Elbing 1863), Terent'ev (St. Petersburg 1875), Wahrmund (Giessen 1875), Clarke (London 1878), Guyard (Paris-Leiden 1880), Kazimirski (Paris 1883), Le Strange (London 1883), Pizzi (Leipzig 1883), Darmesteter (Paris 1883), Finn (London 1884), Mirza Dzafar (Kazan-Paris 1884), Barb (Vienna 1886), Salemann-Shukovski (Berlin 1889; Moscow 1890), Seidel (Vienna 1890), F. Rosen (Leipzig 1890), Platts (London 1894), Huart (Paris 1899), Horn (Strassburg 1895-1901). The 20th century saw the publication of an ever-increasing number of grammars and textbooks, in Iran and elsewhere, with a notable slowdown during the two world wars and a resurgence thereafter. Those grammars which for one reason or another stand out are specially mentioned elsewhere in the appropriate contexts. It is not the task here to provide a complete and up-to-date bibliography of grammars or to evaluate them as to their relative merits. Linguistic progress is no longer achieved mainly in the context of full-scale grammars but rather in focussed studies. Grammars and textbooks are more or less indirectly evaluated by their audience and users, their prominence depending on a combination of such factors as coverage, linguistic scrutiny, lay-out and general appeal and, last but not least, marketing and availability. Thus Jones (1771) was republished many times, as were Salemann-Shukovski (1889) and Lambton (1953).
25 The evaluation of grammars and textbooks thus resembles to some degree the evaluation of works of literature. It would be improper, in view of the present dominance of'linguists' in the study of Persian, to make linguistic excellence the only criterion of judgement. With these points in mind, one would include the following in a list of some of the most prominent grammatical discussions and textbooks: Platts-Ranking (Oxford 1911), Phillott (Calcutta 1919), Jensen (Heidelberg 1931), Rossi (Rome 1947), Lambton (London 1953), Rastorgueva (Moscov 1953), Lazard (Paris 1957), Hodge (Washington 1960), Rubincik (Moscow I960), Obolensky et al. (Washington 1963), Alavi-Lorentz (Berlin 1967).
2. SYNTAX
2.0. INTRODUCTION
General observations on clause ('sentence') syntax constitute the basis of classical Muslim and European linguistic approaches to Persian. A three-step progression can be observed from 'logical' syntax to taxonomic-structuralist syntax to the 'logical' syntax of several modern linguistic approaches. The basic notions of the functional parts of the sentence have however little changed; The essential; basically logical, pair was subject and predicate (mosnad-eleyh and mosnad). The functional sequence was agent=subject, object, adverb and verb (fa'el, maful, qeyd, fe'l). Muslim and thus Persian grammar distinguished further'actional' sentences with a fe'l, from 'stative' (hät) sentences consisting of an initial statement=subject ( m o b t a d ä a n d the 'affirmation/predication' by a substantive (verb) phrase (.xabar). The latter pair of terms has a much wider application than the 'stative' sentence. In fact it implies the notions of (main) topic and comment. It thus reflects the recognition by Muslim grammarians of topicalization as a decisive factor in grammar which has only fairly recently been focussed on by contemporary mainstream theories. Whatever the basic assumptions about syntax and the sentence, the important issues are the particular structure, the rules and their hierarchy in which the parts of speech are combined to produce grammatical sentences, i.e., the 'natural logic' of a language like Persian. Implicit in this view is the question of what is to be considered a basic sentence or the set of basic sentences in Persian from which more complex sentence structures are derived. Thus, Vosuqi (1972) attempted to show that there are four 'kernel sentences' in Persian (question marks are mine): 1. a basic (? ) sentence: NP-VP (e.g., Xsod ' X became', Xgoft 'X said'); 2. a transitive (? ) sentence: NP- (NP-ra) -VP; 3. a stative (? ) sentence: NP-copula; 4. an agentless, impersonal sentence: (NP -rä)-WΡ (e.g. (u-ra) nemisavad did 'one cannot see him, he can't be seen'). These are of course not all 'kernel sentences'. As an example for a strictly prag-
28 matic, uninterpreted list of Persian sentence types one may cite the one by Jazayery-Paper (1961: 322-44). After the excellent, partially analytical, detailed survey of syntax in Jensen (1931) which is still the most comprehensive study and has to be read side-byside with Lazard (1957), it was Arends (1941) who presented basic facts of Persian syntax, and one must mention here the survey by Lentz (1958) which, based on structuralistic principles, includes many highly suggestive notes on the structure and functions of modern literary standard Persian. Since the fifties scholars writing in East European countries above all began a thorough survey of virtually all important aspects of Persian syntax. As some of the most prominent among them one may cite Telegdi, Pejsikov, Rubincik, Shafa'i and Shaki. Their approach was essentially descriptive-taxonomic, emphasizing not so much 'native intuition' but extensive data collation as a basis for syntactic classification, avoiding the tendency to 'reduce' the variety of the evidence. Some of those studies appear to have been originally presented as doctoral dissertations or equivalents. Similarly, in Iran and the West, a considerable percentage of studies of Persian morphology and syntax is due to doctoral dissertations, most of them by Iranians. One may cite Nye (1955), Hincha (1966), both structuralist; Milaniän (1965), Sädeqi (1967), both Martinet school; Bäteni (1967), basic neo-Firthian; Moyne (1970) generative, Palmer (1971) case-grammarian; and Bashiri (1972) presenting his own linguistic model. The widely varying degree of application of the particular linguistic theory and of the contribution to Persian and/or linguistic theory may be briefly indicated by comparing Bäteni (1967), Moyne (1970), and Bashiri (1972). Bäteni 'probably the first structural grammar of Persian' (1968: II, my translation), presented basic syntagms and paradigms of the Persian sentence, clause, phrase and word (fomle, band, goruh, kaleme). For each level the function slots and the categories which can fill them (detailing possible reduplication and morphological specifications) are described. In fact, Bäteni in 1968 presented the first rather comprehensive 'immediate constituent', and thus surface, grammar of Persian. There is little that addresses itself to more intricate problems of syntax than surface hierarchy. Moyne (1970) subscribed fully to the late-60s version of transformational grammar, the then-current analysis (of English) which he quite skillfully applied to Persian. He was able to contribute towards the solution of a good number of syntactic surface-deep structure problems of Persian such as the compound verb, 'passive', emphasis and subject reduplication (cf. the citations under the various topics in this survey). Bashiri (1972) presented an attempt to develop a new model of what may be
29 called generative-functional grammar; developing mainly the higher or 'deep' levels of its hierarchy, he discussed general linguistic problems rather than problems of Persian even though Persian is the language he drew upon.
2.1. THE NOUN AND THE NOUN PHRASE
The simple noun phrase in Persian stands out because of its minimal morphology in terms of number of morphemes. Besides the stem-form there are: the plural marker M, the (unstressed) 'indefinite article' /, several prepositions, and one post-position, the marker of the 'definite direct object' ra. In colloquial Persian there is further a 'determinative' clitic e with singular noun-phrases. To these the demonstratives änjin 'that/this' are to be added. (The terminology used here is traditional; new terms based on newer insights will be discussed in the appropriate sections.). Unlike the formal aspects of the noun phrase morphology, the syntactic-semantic aspects present problems many of which are yet unsolved. In this respect Persian does not differ from most other languages. Detailed observations can be found in traditional grammars where insights into the function of the various morphemes were sought from the observation of the individual morphemes in various syntactic and semantic contexts. The traditional Muslim studies and interpretations of the noun phrase morphology have been carefully collected by Mo'in (cf. bibliography). They considered exclusively the standardized, traditional literary Persian. European grammarians, directly basing their grammatical descriptions on native Persian grammar, increasingly reduced those often too sophisticated, if not casuistically overburdened descriptions, to a few manageable 'basic' meanings or functions of those morphemes. The three main arguments to do so seem to have been 1) didactic simplicity, then 2) etymological simplicity in the 19th century, and finally, 3) structural simplicity since the middle of our century. As examples for the respective approaches one may cite 1) Jones (1771), 2) Vullers (1870), and 3) Nye (1955). The traditional 'monolithic' approach to Persian noun phrase morphology and syntax began to be overcome since about the fifties. First to mention, Bahar (1947-49) in his monumental diachronic study included detailed notes on the functions and the changes of these morphemes. At the same time individual morphemes as part of the noun phrase syntax began to be researched in certain texts or periods (cf. topical discussions and topical bibliography). The structuralist approach began to be applied. It focussed on synchronic contemporary Persian (such as Nye [1955]). For the first time the combinatory patterns of the morphemes involved, which had been disregarded before, were systematically investigated (first by Nye [1955], e.g., tables on pp. 80, 84).
30 As of today, the linguistically most satisfactory and concise study of the simple, non-complex noun-phrase is still that by Hincha (1961). He was the first to clearly identify the basic noun phrase sequence: stem form
-
(i)-(rd)
and to determine the functions of these against each other. In addition, he identified their combinatory patterns with noun-phrase embedding, mainly eza/e-constructions, and pronominal sufflxation. His syntactic-grammatical scope and his suggestions for a functional analysis may not be all-inclusive; nevertheless his stiidy is as yet unsurpassed. Since, several aspects, especially those regarding grammatical levels higher than the clause and sentence involved by these morphemes, have been discussed. The three major syntactic-semantic problems involved are genericity (or 'class'-'selector' in Seller's [1960] terms), definiteness, and specificity/reference. These problems have invariably been touched upon by every grammatical description, but they have been little focussed on. Much of the traditional indecisiveness, even mutually contradictory statements of interpretation, can be related to the fact that the three main categories mentioned are not confined to any particular one of the morphemes involved but rather are expressed by their combinatory matrix. Thus 'definiteness' is said to be expressed by the 'indefinite article' i, and the 'definite direct object' marker rä. However, definiteness is not only circumstantial or contextual but may also be implicitly defining in certain 'stem-forms'. In fact, all modifications of a noun phrase, be they affixes or noun phrase embedding, seem to contribute to semantic-syntactic specifications and changes in genericity and definiteness, and ideally they should be investigated in toto with regard to these factors. Formally, the stem form is defined negatively by the absence of other markings. It has been generally recognized that the unmarked stem form identifies the noun as generic (similar to the singular in Old Persian and elsewhere). This is of course not always necessary, as was recognized already by Ibrahim (1841). Thus, there is no doubt that the discussion of the syntax of the noun phrase will be on firmer grounds only after the semantics of the noun itself has been included in the investigation to a sufficient degree. It is of particular interest to note that the new attention to these theoretical questions by generative-transformational grammar had its counterpart in ArabicPersian theory. Lumsden (1810.2: 130-45), with his knowledge of that theory, was one of the last grammarians to include detailed discussions of these points. Distinguishing between 'individual' and 'generic' (including 'collective') nouns, he identified as 'known'/generic (mo'arref): a) names of a species without reference to conditions: 'the plane triangle', b) names of a species as opposed to an individual: 'man never dies', c) nouns denoting individuals of a species collectively, or d)
31 denoting any one or more individuals as opposed to all other members of the species: 'give me the book'. But a noun is 'unknown', indefinite (monakker) if denoting, indifferently, any one or more individuals among many others of the same class: 'give me a wafer, that is to say any wafer.' Accordingly, all nouns are identified as definite except when marked with the indefinite i. The interesting point here is that there is only one 'unknown' case. This notion in Arabic grammar deserves detailed study. 2.1.1. Number and definiteness 2.1.1.1 hä and plural A large portion of the research on the morphology of the plural in Persian considered 1) the diachronic change from the earlier ending am to its general substitution later by hä, 2) the loan plurals from Arabic, plural endings as well as Arabic broken plural forms (cf. the discussion of Arabic grammar in Persian, below.). The implied question of stylistics has occasionally been touched upon, especially in works which consider the modern spoken language, or borrowing. The third problem often discussed was that of number agreement between subject and predicate. The most comprehensive study was done by Mo'in (1958) who carefully collected data from classical Persian grammarians and researched the texts, without offering, as far as I can see, more than a taxonomic classification of the traditionally known possibilities. The commonly agreed-upon insight is that only the implicitly or explicitly marked plural of subjects identified as [+ human] is obligatorily copied into the verb while with other subjects the copying of the plural implies emphasis on the number of individual items. A good survey-chart and discussion for contemporary Persian is included in Bäteni (1968: 102). Proving to be more difficult than the morphology as such and than concord was the identification of the function of the plural morpheme. Already the earlier grammars recognized two distinct functions of hä: that of a plural morpheme of nominals, such as ketäb-hä 'books' and of an optional adverbial marker, such asdir(-hä) 'longo tempore, diu' (e.g. Vullers [1870: 163J). (The older plural marker ση was recognized as having a similar 'adverbial' function, e.g. bämdädän 'at morning time', nägah-än 'untimely=suddenly' e.g. Vullers [1870: 163]. This 'adverbial' hä has been early identified as derived from Middle Persian adverbial i/ehä, while the origin of the plural hä is uncertain, but is supposed by most scholars to be identical in origin with the 'adverbial' hä, cf. Jensen [1931]). It was Lazard (1957: 60-61) who recognized and described the productivity of the adverbial hä in contemporary Persian.
32 Hincha (1961) could show that the function of hä in all environments is the same, at least in contemporary Persian. The diversity of functions attributed to hä is due either to semantic-contextual overdifferentiation (as introduced by native grammarians and their followers) or due to Western grammatical concepts. The basic function was identified by him as 'amplification', which appears as 'plural' with count nouns, as increase or extent with mass nouns, or as generalization with adverbial nouns, e.g.: count nouns mass nouns
ketäb-hä ab -hä äbfow-hä bälä-hä
adverbials
'books' 'waters, plenty of water' 'all kinds of beer/plenty of beer, the beers' 'up (stairs)'
Hincha's interpretation of the function of ha as 'amplification' appears to be the best explanation for the occurrence of hä with predicative noun phrases. In that position the simple noun phrase is usually not marked by hä even if the plural is implied, according to a well-known rule. However, if hä is present in that position it expresses emphasis (with emphatic stress), i.e. 'amplification', e.g.: count noun object
mä mehmän
mä mehmän-hä därim
därim
we guest-0 have 'we have guests' mass noun object
as opposed to
we guest - pi. have 'we have (all kinds of) guests'
man äbfow mixoram
man äbfow-hä mixoram
I beer-0 drink Ί drink beer'
I beer - pi. drink Ί drink all kinds of beer'
One should add that such differentiation is only possible for the object. Thus there is
änhä mo'allem-φ
hastand
they teacher(s)
are
but not
(*)änhä mo'allemhä hastand The latter is, however, acceptable if reference is implied. In that case, it means: 'they are the teachers' and is usually followed by an explanatory remark. This implicit reference is not to be confused with 'amplification'. Instead it
33 follows another rule: a counted or countable plural whether in the predicate, object or not, is obligatorily marked by ha in the presence of overt or covert modification. In the example above the modification is a covert reference which triggers the surfacing of ha. An example of this rule applied to the object is, e.g.: mä mehmänhä-ye ziyäd därim 'we have many guests' Thus, while hä itself is not a marker of reference it nevertheless can function as an indicator of reference, as in the clear case of the predicative noun. This function is even more conspicuous in syntactic contexts where the occurrence of hä is believed to be strictly blocked, i.e., after numbers larger than one. Such cases have been observed by Miläniän (1970a: 168) in sub-standard colloquial, e.g.: fun) setä mardfhjä raftand substandard English: 'them three men have left' Reference is probably implied also in all the so-called adverbial occurrences of M\ in such cases the adverbial 'plural' is usually specified by an ezäfe '=' construction and/or a demonstrative, e.g.: un bälähä-s cetowr bude 'up there, how was it? (asked of a cosmonaut in a cartoon). Statements like the following are questionable if not impossible: ? *bälähä raftand *'they went upstairs' The so-called compound verbs deserve particular interest since there the admissibility or not of hä could serve as one criterion to identify the many different syntactic structures underlying them (cf. discussion of compound verbs). Compare the occurrence of hä with compound verbs like komak kardan 'to (make) help': be u komak (hä) kardam Ί gave him (all sorts of) help/helped him (a lot)'
34 Here the 'amplification' by ha is quite acceptable. However, it is questionable with other compounds like: tarfome (ha?) kardam Ί made (all sorts of? ) translations'. ha is impossible if the compound verb governs a direct object, e.g.: dn-rd tarfome (*hä) kardam Ί translated it (*a lot)' This point was also made by Meyer—Ingwersen (1974) who further was the first to note that only ha can function as amplifier but not the Arabic broken plural. Thus there is fekr-hä kardam Ί thought a lot' but never *afkdr kardam However, the regular Arabic plural seems to be admissible, e.g.: tahqiq-ät-i kard 'he made certain investigations' The predicative position presents open problems with regard to ha. Nevertheless, Hincha's identification of ha as a marker of'amplification', even if tentative and generalizing, indicates the relation of hä to problems of genericity in a much clearer fashion than earlier identifications. 2.1.1.2. i and definiteness As mentioned above, definiteness in Persian has traditionally been discussed in the context of two morphemes, one being the 'definite direct object' marker rd, the other the (usually unstressed) 'indefinite article' i as it is called in most European grammars (e.g. Chodzko [1852: 72],Lazard [1957: 66]). Traditional Persian grammarians were well aware of a peculiar difference, i.e. the difference between the 7 of unit' (yd-ye vahdat) and the Ί of indefiniteness' (yd-ye tankir) (cf. the collation of traditional statements in Mo'in [1958: 232 sqq.]). In English,
35 a comparable ambiguity is found with 'a(n)', e.g., Ί am looking for an usher', where 'an' may imply 'one' particular usher or 'any, some' usher or, in terms of Lumsden (1810.2: 132, 422-24), who was the first European to discuss this crucial point of universal and Persian grammar in greater detail, following Arabic—Persian grammar, 'bring me a book (no matter what book)' and Ί slept one (his italics) night until morning'. The difference between specific and non-specific indefinite noun phrases has been recently brought to renewed attention by generative grammar (cf. Peterson [1974]; see below). Traditional Persian grammar distinguished yet another /' which occurs before relative clauses, i.e. the so-calledya-ye tarif/eMre 'referential/demonstrative f , e.g.: asb-i ke däram
'the horse that I have'
Vullers (1870: 254) iemarked in quite a complex fashion that i implies that the thing, which by the addition of the indefinite article is proposed as either a single thing or something special (praecipuum), is defined more accurately in the following relative clause. What is implied is, of course, the observation that i marks a restrictive relative clause. Any of these three kinds of / may be combined with the plural, e.g. the 'indefinite' ι: in veläyat [mardom-än-e xub ]-idärad 'this region has (some) nice people (pi.)' (Chodzko [1852: 78]; with the plural marker -άη and with i regularly following the nominal phrase.) Ibrahim (Ibrahim-Fleischer [1875: 33]) is one of the first to remark on the 7 of unit' after plural, e.g. gol-hä'i translated by the French 'des roses'; here i is said to identify the 'indefinite plurality' (unbestimmte Mehrheit), but cf. also Lumsden (1810.2 : 131). Moreover, the 'definite object' marker rä may follow i, e.g.: kas-i-ra ferestad
'he sent someone';
here the i implies 'a certain', not 'any' one. The three outstanding studies on i are Murav'eva-Rubincik (1959), Hincha (1961) and, in partial reply to the latter, Lazard (1966) who offered a good survey of scholarship. Hincha attempted to define the common function of all three kinds of i and found it to be a 'restriction'; i.e. the restrictive selec-
36 tion out of a generic unit or out of a plurality (cf. Rubincik's term 'selective'); e.g.: count: mass:
ketäb-hä-i ketäb-i äbfow-hä-i äbfow-i
'some books' 'some, a book' 'some kinds of beer' 'some, a beer'
Hincha's identification of i as 'restrictive' can help to explain the difference between ketäb-i: yek ketab : yek ketäb-i which were generally seen as having the same meaning. But'yek indicates a certain individual item which is not specified', and Ί indicates any item of a species'. Thüste/: ketäb is 'one book', ketäb-i is 'a/some book', and yek ketäb-i is 'one certain book' (for more detail esp. regarding yek, cf. Hincha [1961: 169-70]). More importantly, Hincha (1961) identified the formerly isolated 'demonstrative /' before restrictive relative clauses with the 'indefinite z" as 'restrictive /'; he was led to this conclusion by the notion of the 'restrictive' clause (cf. discussion of relative clauses below). Lazard (1966) basically agreed with Hincha's identification, but he pointed out that Hincha's structuralist, distributional arguments do not consider several important counterexamples, e.g. the fact that the 'restrictive i 'does not occur after demonstrative in non-relative constructions. This is true in cases like *än mard-i ämad 'that one man came' which is ungrammatical. But the argument is too general. Demonstrative and i co-occur e.g. in: an yek-i-ra did kodämkas-i-rä did 'he saw that one','which one did he see? ' motarjemin ... be conin dasturhä'i ehtiyäf däste and 'the translators... had need of such grammars' (Soxan 20.1:33).
37 There remain doubts. Lazard (1966), beginning his argument with a diachronic survey, recalled that at an earlier stage, the demonstratives än/in 'that, this' and the 'demonstrative' / were mutually exclusive before relative clauses (cf. Jensen [1931: 29], Lazard [1966]), i.e.:
an mard 0 ke ämad (p mard-i ke ämad
either: or:
'that/the man who came' This mutual exclusion and complementary distribution in earlier forms of Persian was recognized by native grammarians who, consequently, termed this i 'yä-ye esärat' 'demonstrative /'. (Eilers [1956: 206-11] suggests the etymology *aita for the 'demonstrative /', as opposed to the commonly recognized etymology aiva for the 'indefinite/restrictive' /). Thus the occurrence of i before relative clauses after the demonstratives is only a later development. For this and other reasons, in spite of the elegance of Hincha's solution, it is preferable for the time being to accept a'demonstrative' i with relative clauses which then would seem to be pleonastic after demonstratives (for further observations, cf. the discussion of the relative clause). Reviewing the evidence briefly, two functions of i can be distinguished: a) specific or referential: '(a certain) one' and b) non-referential '(any)one':
yek ketäb-ί xaridam yek ketäb-i-rä xaridam Ί bought a (some) book' Ί bought a (certain) book' Both are, as Hincha found, 'restrictive' in terms of identifying an item or a group of a larger set or class. He overlooked, however, that the restrictive function is not confined to /; also rä, e.g., 'restricts': ketäb-rä xaridam Ί bought the book'. Thus in terms of categories one has to distinguish at least between reference, definiteness and selectivity. Their combinatory pattern may be illustrated as follows (ex: gorg 'wolf in the direct case): ref.
1. gorg-rä 2. gorg-i-rä 3. gorg-i 4. gorg-rä 5. gorg
+ +
def. +
sei + + +
+
'the wolf (talked about) ' a (certain) wolf, "this" wolf 'a (some) wolf 'the wolf (-species)' '(a) wolf/wolves' (generic unspecified)
38 The distinguishing function of i can thus be defined as the combination of (-definite) (+selective). Applying the traditional notions and terms, while both 2 and 3 would imply ya-ye vahdat 'the i of unit', the i of 3 would be subcategorized more specifically asya-ye tankir 'the non-referential, unknown (not: 'indefinite'! ) /'; and the i in 2 would be subcategorized as ya-ye ta'rif/esärat 'the known/referential z". It must be emphasized that the presence of the 'referential' i implies, and is usually accompanied by, 'reference', mostly in form of a paratactic or a relative clause, e.g.: (yek) gorg-i rä didam; (yek) gorg-i-rä didam
(än) ajib bozorg bud. ke ajib bozorg bud.
Ί saw a/some wolf; he was really big' Ί saw a/some wolf that was really big.' This is not the place to go into more detailed discussion of this problem of reference and specificity and the problem of genericity vs. the generic use of nouns and their interrelationship with definiteness. The simple fact is that these problems have been little and inadequately discussed. As to the non-referential /, there is a recent study by Peterson (1974) which approaches the problem on the basis of generative semantics and which may be considered one of the most suggestive so far. Peterson proposes that the function of this i might best be regarded as marking the difference between count and mass. He cites the above mentioned fact that this i, when attached to a mass noun, converts it into a count noun, as in the following pair: dbejow 'beer
xord he drank'
äbefow-i xord 'a (glass of) beer he drank' He also notes that if a noun which is semantically a count noun appears without i, it becomes mass. Yet such a noun, at least when in object position, differs significantly from a true mass noun, as is shown by the evidence of the pro-form as. In a sentence with a mass noun like äbefow mixore ammä dust-as nadäre beer he drinks but like for it he has not. 'he drinks beer but he does not like it'
39 the as may refer either 1) to the beer or 2) to the act of drinking it whereas in a similar sentence with what is semantically a count noun, the pro-form has only one interpretation, e.g.: name minevise amma dust-as nadäre letter he is writing but like for it he has not. Here, as necessarily refers only to the act, never to the noun only: 'he is writing a letter but does not like doing so'. From these facts he concludes that such a noun should be regarded as part of the underlying predicate. Stated another way, a count noun loses its count-noun status when it is underlying part of a predicate, and being no longer a count noun, it cannot be marked with i. This argument is supported by the fact that any simple, unmodified noun, count or mass, when used predicatively appears without the suffix i, as in: un ketäb-e 'that's (a) book'
un äb-e 'that's water'
These arguments and conclusions are presented, not as a refutation of earlier analyses, e.g. those of Hincha (1961) and Rubincik (1959), but rather as an attempt to clarify these by relating them both to the more traditional terminology and to certain more recent theoretical proposals. In particular, Peterson suggests that just as a predicate noun is represented in an underlying form as a logical predicate, so any surface noun which is interpreted as part of the verb be represented as underlying part of the predicate. Thus, just as un that
ketäb-e book — is
is represented by the logical formula BOOK (that), so name minevise letter he writes should be represented as WRITE LETTER (he). On the other hand name-i minevise a letter he writes would be represented in such a way that the noun is not part of the predicate,
40 i.e. WRITE (he, a letter). Finally, he observed that since mass nouns can be converted to count nouns, and vice versa, it might seem that all nouns are either or both. Yet, as we have seen, the difference seems to be genuine. And so, in order to help preserve the distinction, he proposes that mass/count is a difference in the way the things named are perceived, but not a grammatical distinction, and that we borrow from logic another pair of categories, namely non-referencedividing/reference-dividing. This distinction, he suggests, is a difference in the way particular nouns are used and is therefore part of grammar. It is this distinction which is marked by the (non-relative) suffix i. Of course, these two pairs of categories will have a fairly high correlation, since a mass noun will usually be used in a non-reference-dividing way. But the correlation, as we have seen, is by no means absolute. Finally one should mention a distributive function of i. Horn (1895-1901: 107) identified the 'demonstrative' function of i in contexts other than the relative clause; e.g.jyek-i-ye ma 'one of us' (with stress on z). The same yä-ye mo.'ruf 'definite V has been noticed by Xoräsäni (1950b: 529), e.g.: folan do pesar darad: bozorg-ί(pesar-e bozorg, bozorgtar-e) panf säle να kucek-ί (pesar-e kucek, kucektar-έ) cahar säle ast 'X has two sons: the older (-i) one (i.e. the older boy, the older (-e)j is five and the younger f-i) one (i.e. the younger boy, the younger (-ej) is four years old' Nye (1958: 78) identified this as the 'distributing' function of i, which is unstressed in cases like metr-i panf tomdn 'five tomans per meter' but is stressed in other cases such as the above or as in kilo' i miforusi 'do you sell it by the kilo' andyek-iyek-i 'one by one'. Note also instances like se farsang-i-ye tehrän 'three farsangs from Tehran'. It appears that this morpheme ζ (or this function of ζ) is indeed 'distributive'. Here belongs the seemingly irregular example an yek-i-ra 'that one (obj.) mentioned' (p. 36) (the discussion of i is continued below in the context of the relative clause). 2.1.1.3. Colloquial -e In colloquial near-standard Persian there is a suffix -e (= hi after vowels, ye after z) which has been little discussed. The citation from Xoräsäni (1950b: 529) above includes his parentheses, which 'explain' the meaning of the definite /: bozorg-i (= bozorgtar-έ) 'the old one' (= 'the older one')
41 i.e., the definite i is equated with the definite suffix e. While both seem to share similar functions in this case, they are, however, not substitutable at all, but exhibit different functions, e.g. pesar-e 'the boy (there)' but not *pesar-i; yek-iye ma 'one of us' but not *yek-i-ye mä. e is found in many dialects of Iran. It was first recognized by Andreas (unpublished notes, 1875-82); Christensen called e 'demonstrative' (1930: 31; Ac Or 12 [1934]: 310, in Andreas 1939: 42 fn. 1). Xoräsäni noted that e can be attached to proper names, e.g. hoseyn-έ 'the Hoseyn mentioned'. Comments are also found in Ivanov's writings on dialects of Iran. Nye (1955: 79) called it definite, the 'one mentioned'; so did Jazayery-Paper (1961: 264-67); Lazard (1957: 69) called it 'determinative', e.g. pesar-e 'le gar?on en question' and observed that it is mutually exclusive with plural ha. This very fact led Hincha (1961) to conclude that the function of e is the opposite of the 'amplification' function of ha, thus 'punctualization'. However, the occurrence of e also after the attributive adjective as in käj-e bozorg-e 'the big pine (there)' indicates that there may not be a one-to-one opposition between ha and e as Hincha assumed. Of particular interest here is the above-mentioned observation by Miläniän (1970: 168) on substandard colloquial Persian that after numbers larger than one, where standard Persian requires the singular, the colloquial referential marker is (h)d, not έ\ e.g.: (un) se tä mard-ä ämadand 'them three men have come'
2.1.2. Cases 2.1.2.0. Introduction The notion of 'case' in most grammatical literature on Persian does not show a clear distinction between a) the outward form of the inflectional 'paradigmatic' case (such as inflectional or pre/postpositional nominative, accusative etc.), b) their function as 'parts of speech' within the sentence (such as subject and predicate), and c) the implicit 'underlying' relationship between the parts of the clause (such as the 'logical' agent and object). While the recognition of these levels is old and while many observations have been made as to their inter-relationship, they could only be described in often lengthy and even more often incomplete and contradictory statements, due to the lack of a more adequate descriptive linguistic methodology (on verbal rection and intransitive-transitive, cf. also the discussion of the passive).
42 Hakim (1974) privately published a pamphlet (apparently a term paper) attacking the terminological and theoretical laxness in prominent grammars published recently in Iran. He demonstrated that the various authors contradicted each other and themselves in the definition and description of such terms as maful. Differing from author to author, this term is used to describe the maful 'non-subject' vs. fä'el 'subject'; as such it may be used to refer to all formally oblique noun phrases (including adverbial noun phrases, qeyd ); or to their function as non-subject cases as opposed to the subject, fa el (again this includes adverbs); or, finally, maful is used to refer to the direct object, only. As a result of this confusion, some grammars describe subject or object complements (motammam) variously as maful or as qeyd 'adverb' as e.g. in: xoshäl happy
raft he went
u-rä him
xoshäl happy
didam I saw
While there are no case endings in Persian, the cases are expressed by means of affixation and/or order. Earlier grammarians had no difficulty in equating most diverse structures of Persian phrase and clause syntax with the standard paradigm of the cases of inflected (Latin) grammar. Thus, besides the nominative (=stem-form), the connective -e- of the ezäfe was identified as the genitive marker; the definite direct object marker-ra as the accusative marker ; the interjection ey or the postposition -a as the vocative marker; finally, as the dative marker were identified both -rä and the preposition be-. The Persian paradigm established in European grammars thus consisted of five cases (cf., e.g., Vullers [1870: 177]): Nom Gen Dat Acc Voc
Q -e-räjbe-rä ey-j-ä
Genitive, dative, accusative were also collectively identified as 'oblique' (cf. the discussion Jones [1771: 29]; maful in Arabic grammar). Prepositional cases did not fit well into this system (except for the dative be), or rather, they proved to be less amenable to a clear-cut equation with other case systems. Cases like instrumental, ablative, or locative could not be simply equated with a single preposition. Thus Vullers retained the five case system but treated the prepositional cases separately. So did others, e.g. Salemann-Shukovski (1889: 26-36) followed by Jensen (1931: 41-43) who had 1. nominative, 2. vocative,
43 3. oblique (dative/accusative), 4. ezäfe, while prepositions are discussed separately (for the notion of the oblique, cf. discussion of rä below). Jensen represents the transitional stage from 'universal' and comparative grammar towards structuralist grammar with its attention to morphemes and related structures which is much more evident in Lazard (1957). In the latter's grammar instead of 'nominative' there is a general discussion of the substantif; instead of 'genitive' the various forms of the ezäfe are listed; instead of the 'accusative' various functions of rä are discussed; the indirect object in be is discussed together with other prepositions, while the section on particules d'apostrophe lists the various 'vocative' markers. The notion of paradigmatic cases is gone, the distributional-functional aspects are foregrounded. It should, however, be noted that the actual discussion of these points does not basically differ much from earlier grammars. In those, the distributional-functional aspects were also discussed, generally in the chapters on 'syntax' (cf. Geitlin [1845]). Thus, in spite of the shift away from traditional patterns of description towards the structuralist attention to combinatory patterns, the basic arrangement and sequence of these cases has remained surprisingly constant. This shift, one of the main objectives of which was to describe the 'inherent' structure of Persian, was likely to obscure the fact that traditional Persian grammarians had equally deep insights into their language. After the dialectically important stage of structural grammar, with its overemphasis on 'one morpheme-one meaning' and on (surface) structure, transformational-generative grammar, especially 'case'-grammar, again investigates the underlying syntax of the cases. The point of reference is neither the 'surface case' nor the morpheme and its distribution, as in structural grammar, but the 'logical' and 'underlying' case as part of the 'case-frame' determining each verb. Fillmorean 'case-grammar' has apparently been applied only once so far, in an unpublished dissertation by Palmer (1970). Following insights of that theory he outlined a number of ways in which the 'underlying'cases of Persian can be 'surfaced' to appear as the subject, object, etc. case (cf. also discussion of rä below). He demonstrated how and by what rules deep cases like 'agent' and 'dative' can appear as either the subject or object on the surface, e.g. (in abbreviated form):
44 1. agent raised as subject:
Agent
pesar
hasan
hasan (subj) pesar-rä kost 'Hasan killed the boy' 2. dative (experiencing object) raised as subject resulting in a'passive' sentence: VP
Μ
past
(tavassot-e hasan)
pesar
kos"
(tavassot-e hasan)
pesar
koste sod
'The boy was killed (by Hasan)' Underlying case was also studied by Bashiri (1972) who developed an original theory of the deep syntax of Persian. The sentence hasan pesar-rä kost 'Hasan killed the boy' would be handled as follows (cf. Bashiri [1972: 172-73]; only the lower parts of his derivation are given in order to avoid lengthy theoretical explanations): Agent [+phys.] [+funct.]
. ALV (i.e. agent level of verb)
Experiencer [+phys.] [+funct.j
I
1 -ELV (i.e. experiencer [+funct.] level of verbs)
45 hasan pesar-rä Hasan boy 'Hasan killed the boy'
kost kill-past
The passive transformation is triggered by the switch of kos 'kill' into the 'object' (i.e. complement) position with the rule of identity between experiencer and object: Agent [+phys.] [+funct.]
AVL [- funct.]
Experiencer [+phys.] [+funct.]
ELV [- funct.]
Object OLV [+phys.] [- funct.1 [+funct.]
\
(tavassot-e pesar hasan) koste (by Hasan) boy 'The boy was killed (by Hasan)' killed
sod became
Due to the novelty of his approach and the formalism he had to develop, it may take a long time before his insights will be appreciated and scrutinized by a larger audience. In view of the lack of any other application of his intricate theory the reader is referred to his book (1973), since an abbreviated statement may do injustice to Bashiri. One of the most systematic and comprehensive studies of parts of speech is by Meyer-Ingwersen (1974) who explored the semantic- syntactic 'case-frame', so to speak, of the compound verb fekr kardan 'think (of)·' For the first time he discussed and defined the combinatory regulations of the direct object, the objectoid, the theme, the benefactive, further adverbs of gradation, of duration, and modal adverbs of quality and similarity (for detail see discussion of the compound verb below). With regard to particular cases, in reviewing the history and state of grammatical research in Persian, it becomes evident that there are 'major' cases, i.e. the five first cases of European grammarians of Persian, and 'minor' cases. In the
46 former class, most attention was paid to a) the direct object/ 'accusative' and b) the eza/e-construction/ 'genitive', whereas the subject/ 'nominative' and indirect object/ 'dative' rank third and fourth. For some deeper reason the degree of attention paid to these cases has been more or less equal in the grammatical study of Persian, irrespective of linguistic theory (as is the case with the grammatical study of any other language). The vocative is somehow aligned with the former, but never treated adequately, except for certain phonolical and phonetic studies (cf. phonology). 2.1.2.1. Prepositional cases and adverbs The 'minor' cases, and parts of speech, i.e. the 'prepositional' cases and 'adverbs' have so far proved to be at least as problematic as the 'basic' cases. The prepositions in individual texts have been studied, such as those of the Golestän by Shari'at-Rezavi (1963) and of the Tärix-e Sistän by Smirnova (1958); and prepositions have been discussed individually (cf. topical bibliography). The adverb as a part of speech, and with it prepositions, has been generally treated by means of listing simple or compound adverbs and prepositional adverbs according to their semantics, cf. e.g. the sequence of adverbs of 1) time, 2) location, 3) quantity, 4) quality, 5) comparison, 6) question, 7) affirmation and 8) negation, 9) doubt in Chodzko (1852: 133-138). This area of grammar is a dumping ground for most diverse structures, as is the case in grammars of other languages. Among the very few special essentially descriptive and classificatory studies are Rzaeva (1953), Xrisanov (1960) on derivative adverbs, and FarSidvard who has studied adverbials more than anyone else, beginning with his dissertation (1963; futher 1964a: indefinite i and adverb, 1964b: interrogative adverb, adverbs of quality and kind; 1965; comparative and instrumental; 1966a: adverbial phrases in Persian and Arabic; 1966b: prepositional adverbs; 1967 ; adverbs in an). His articles provide an excellent, mostly diachronically oriented survey of the various adverbs and adverbial phrases. Note the remark by Lazard (1957: 84) 'La classe des adverbes est assez mal delimitee; elle interfere d'un part avec la classe des adjectives, d'autre part et d'un autre maniere avec celle des substantives.' The basic reason for this state of affairs is undoubtedly the confusion between word-class, morphology, surface syntax and deep syntax. Also confusing were predicative adjectives like:
xoshdl raft
u-rä xoshäl did
happy — he went
him happy saw-he
47 Such cases, which had been mistaken for adverbs {qeyd) have attracted the attention of Iranian linguists, and have properly been identified as complements, as motammam-efä' eli 'subject-related complement' with intransitive verbs and as motammam-e mafuli 'object-related complement' with transitive verbs (most recently by So'är [1972b] who reviews the recent history of this discovery in Iran; cf. also discussion of the passive). Similar insights can, however, be found in Arabic-Persian grammar where such predicates were clearly distinguished and were considered an instance of häl 'the state of the agent' (cf. the detailed discussion in Lumsden [1810 vol II: 212-22]. Also cf. the differentiation between perfect-tense forms and predicative participles which traditionally would be identified as häl; e.g., morde ast 'he has died' and 'he is dead' (Lumsden [1810.2: 305]: see also Angelus' [1684] example karde ast 'fecit' and 'factus est', both of which present interesting problems of verbal semantics). Similarly, that theory distinguished the underlying syntactic relationships of such phrases as the so-called tamyiz (e.g. u bad ast az nazar-e seid: u bad-sekl ast 'he is ugly with the regard to shape': 'he is ugly-shaped'), the ta'kid (cf. discussion of emphatic xod below), and badal (cf. discussion of emphatic ke and anacoluthon below). 2.1.2.2. ra and definiteness There is hardly any other morpheme or particle in modern Persian which has been the subject of as many specialized studies as the enclitic rä. The reason for this attention is directly related to a) the elusive syntactic function(s) expressed by rä and the rules associated with it, and b) the difficulty of teaching or learning its correct usage. The basic commonly accepted statements are: 1) in contemporary Persian rä marks the direct object if the latter is 'definite' either implicitly by context or explicitly by the presence of any kind of pronoun or a name, e.g.: hasan-rä/u-rä/dust-as-rä/än-rä didam Ί saw Hasan/him/his friend/that (X)' There are many exceptions to the general rule of thumb. Thus certain classes of names do not always require rä, e.g.: farsi balad am Persian I know Ί know Persian'
48 Here farsi-rä would be used only in the contrastive context of questioning about other languages, e.g. 'yes, I know Persian" (as opposed to French, etc., cf. Hincha [1961: 182]). Also note the lack of rä in evidently definite cases where one would expect the marking by rä, like: pä-(J> tu-ye kaß karde foot in shoe put 'she put her foot into the shoe' bim-e άη-ψ däram ke... I have the fear ofthat that Ί am afraid that...' On the other hand, rä is obligatory after verbal nouns, 'infinitives' (masdar), like: harf zadan-rä dust nadäram talking I don't like esq-rä nemifahmid 'love he didn't understand' Moreover, rä is also allowed — and required - even after the 'indefinite article' -i if the object in question is (implicity) identified as a 'certain X' (cf. discussion of 0; e.g. zan-i-ra didam.... a woman I saw... Ί saw a (certain) woman...' (...ke) kaleme-i-rä nasenide nagozärand 'the women listened so that) they would not let pass unheard a single word (she was saying)'
While the examples cited so far concerned direct object noun-phrases, there are perfectly acceptable occurrences of ra where it marks non-objective, 'adverbial' phrases, mainly of location or time (cf. already Horn [1895-1901: 109]; e.g.: emsab-ra inja bäs tonight here be 'stay here (for) tonight'
49 Noteworthy are further referential cases like:
masin-rä ce kär komm car what do I do 'what do I do with the car? ' Besides, rä is seemingly interchangeable with certain prepositions; e.g. :
u-ra 1 amal kardand baru
Finally, in early and classical Persian, rä could also mark the indirect object, 'dative', a famous example for this being:
mard-i-rä do
pesar
man — a 2 sons 'a man had two sons'
budand were
Thus, the impression created by frequency that rä is only an object marker is misleading. The general uncertainty of scholars is reflected in statements like Chodzko's ( 1 8 5 2 : 165): 'Ou faut-il conserver ou supprimer la terminaison du regime direct, rä ? Tel est le probleme que se sont souvent propose les orientalistes, sans pouvoir trouver une solution satisfaisante.' And further: 'Un etranger peut aussi acquerir ä un certain degre ce sens exquis ä d'attention et d'usage.' This is reflected in our days by Lazard's statement (1970a: 175): 'L'emploi de la postposition rä n'est pas defini par des regies formelles rigoureuses. II depend de conditions semantiques et grammaticales complexes.' The major problem in discussing rä can be reduced to one simple observation: the function of rä as a case marker has been confounded with its function as a marker of reference. The main objective of both Ovcinnikova's ( 1 9 5 6 a , b ) and Hincha's ( 1 9 6 1 ) studies was to separate just those issues. They emphasized, however, the 'deflniteness' function while neglecting the case function o f rä (which was a dialectically necessary step).
2.1.2.2.1. rä as a case Contemporary grammars usually accept the function of rä only as a marker for
50 the (definite) direct object, and declare the function of rä as a marker for the indirect object to be obsolete. Lazard (1970) presented a statistical study of the change in the usage of rä from its indirect and direct object functions in early texts until its shift today to productive occurrence in direct objective position only (used by Gustav Herdan, Quantitative Linguistics [1964: 177-82]). There is, however, sufficient evidence that rä still functions as a marker of the indirect object. In classical Persian the 'dative' rä + the substantive verb expressed possession and alternated with dar 'have', e.g.:
u-ra do pesar budand
u do pesar däst
him two boy were
he two boy had
'he had two boys'
In contemporary Persian only the latter is admissible. The logical equivalence was noted implicitly early, but mentioned specifically first, it seems, by Homä'i (1959: 144). Palmer (1971: 65 et passim), following insights of case-grammar, demonstrated the interrelationship of the two constructions, showing that the dor -construction is derived by the subject-raising of the underlying dative-case, an obligatory rule for contemporary Persian:
S
->
hasan (subj.)
ketäb (obj.)
däst
'Hasan had a book'
In earlier forms of Persian the object could be raised as subject: hasan-rä (dat.) ketäb (subj.) bud (to) Hasan a book was 'Hasan had a book' Palmer demonstrated further that just such constructions underly the possessive
vzäfe, e.g.:
51
ketäb- e book- of 'his book'
u — he *-
ketäb-i ke u-rä fast) which him (is)
Grammarians of classical Persian generally distinguished between the dative marked byrä and the dative marked by be 'to' (cf. e.g., Vullers [1970: 172], Darmesteter [1882: 131]). In fact, the dative in be is distinguished from the directional be. The syntactic difference between these two kinds of dative has been shown by Moyne (1970). Thus, he found that pronouns can pronominalize only indirect objects (datives), but not directions, e.g.:
be hasan dädam => dädam-as Ί gave (it) to him.' but be Tehran raftam => *raftam-as Ί went to Tehran.' The dative function of rä is evidenced by the interchange between prepositions and rä (this exchange was first mentioned by de Dieu [1639: 48], more recently by Ovcinnikova [1956]), e.g.:
beugoftam u-rä goftam Ί told him.' bä u moläqät kard => u-rä moläqät kard 'met him' bar u amal kard =*u-rä amal kard 'operated (on) him' Further, the dative is found in indirect verbs. According to Xanlari (1970, 1973), here the surface object is in fact subject (nehäd); e.g.: classical: ma-rä xos ämad modern: xos-am ämad
'it came me well' - ' I liked it'
Here in this expanding-construction type of contemporary Persian, the underlying dative is surfaced as dative (not as subject) which obligatorily has to be a personal suffix (cf. also discussions of indirect verbs). Thus, the one major difference between classical and contemporary usage is the loss of the 'possessive' dative in rä + substantive verb. This is true for all synchronic registers of the language. In certain highly stylized contexts this construction is still available as a part of a 'diachronic' register; e.g.
52 Iräniän-e qadim-rä e'teqäd conin bud ke... 'the ancient Iranians had the belief that...' or more literally '(as) for the ancient Iranians, the belief was that...' (M. Honari, Ä 'inhä-ye nowruzi, Tehran 1974, p. 13) The alternative interpretation would make this a rare instance of the non-object referential function of rä (see below), and thus a rare register-marked variant of anacoluthon constructions (discussed below), such as: Iränihä-ye qadim, e'teqad-esän... 'as to the ancient Iranians, their belief...' 2.1.2.2.1.1. Personal suffixes. The syntactic behavior of the personal suffixes is of interest here; they substitute/ pronominalize in exactly the cases discussed, i.e. the (direct) object and the indirect object, and in the possessive (ezäfe) construction: direct object: dative: possessive:
u-rä didam —didam-as Ί saw him'. u-rä/be u goftarn —• goftam-as Ί told him.' ketab-i ke u-rä —ketäb-e u—*~-ketäb-as 'his book'
Jones (1771: 29) saw this relationship, stating that the personal suffixes substitute for the 'oblique cases, i.e. genitive, dative, accusative'. Similarly, SalemannShukovski (1889: 26-27) observed that the 'oblique case' in Persian implies functions equivalent to the German accusative, dative and genitive. This 'deep' observation, overlooked since, has been made again by case-grammatical analysis. Case-grammarians derive the dative/indirect object and the possessive surface cases from an underlying 'dative' case, which was also identified as the'experiencer' case. The very same deep case is seen also as the source for (animate) direct objects on the surface. There appears thus to be a surprising (at least partial) coincidence between those cases which in earlier forms of Persian were marked by rä and the deep dative/experiencer case of case-grammarians. These approaches towards the identification of rä as a case marker have contributed considerably to both our understanding of rä and, more importantly, to our understanding of cases in Persian. It remains to be seen whether there exists any relationship between the deep dative case and those adverbial cases which may be marked by rä such as:
53 emsab-rä injä bäs 'tonight here be/stay' Also, it is unclear whether there is a syntactic relationship between the deep dative case and occurrences of the 3rd sg. suffix -as in referential instances like: man haqq-asin-e ke hie madrese naraftam 'as to me, the truth is that I never went to school' or άη bälä-hä-s cetowr bude that up (pi) - its how was it 'how was it up there? ' (a question asked by Khrushchev upon the return of the Russian cosmonauts, reported in the satirical magazine Towfiq). In such instances the suffix appears to refer back to entire clauses, if not paragraphs (cf. Lazard [1957: 104-08]). 2.1.2.2.2. rä as a referent The very fact that rä has proved to be elusive to grammatical analysis shows that it is conditioned by at least some rules which transgress the boundaries of the single clause or sentence. This means, in particular, rules of reference and specificity. Jones (1771) opposed gol cidam and gol-rä cidam and identified the function of rä as 'restrictivity': Ί gathered a/any flower' as opposed to Ί gathered the (given or particular) flower'. Vullers (1870: 173) opposed a. käqaz nevestam b. käqaz-rä nevestam and translated a. scripsi epistolam quameumque b. scripsi epistolam, de qua sermo est and thus identified the function of rä as referential. The most conspicuous seeming contradictions are cases where the 'definite' rä occurs with the 'indefinite' /', e.g.: kas-i-rä didam 'someone I saw' Chodzko (1852: 78) explains such instances as an avoidance of obscurity. Jensen
54 (1931: 228) recognized the function of rä by translating 'einen gewissen' (a certain). Some frequent terms suggested to identify that special function of rä are 'determined' (Jensen [1931: 216]), 'definite' (Nye [1955: 87]), 'concrete, definite' (Rastorgueva [1964: 5 2 ] , Ovcinnikova [1956a,b: 390-91]). Hincha criticized the basic inadequacy of these terms: they are hardly ever defined in a non-contradictory fashion. Hincha's (1961) main conclusions are: rä marks the object of reference for which the predication is true or valid. He suggests the term polarization; which is to say, that the noun-phrase marked by rä is correlated with the predicate as its object of reference; it is thus distinct from other modifications of the predicate, such as formal modifications (most prominently by personal suffixes) or semantic modifications (most prominently with compound verbs). Hincha thus detailed an observation by Telegdi (1951: 326) who had argued that rä identifies not definite vs. indefinite, but that it 'actualizes' the nounphrase (utilizing a Prague school concept also known as 'foregrounding'). The characteristics of rä can be explained as derived from this function. Thus rä is most prominent with direct objects because they are the most likely to be polarized; it explains that it is obligatory with inherently specific objects such as pronouns and personal names; also, this polarizing function explains why rä can occur only once in a clause (except for reference 'piling'). However, the question why rä is obligatory after verbal nouns (masdar) is less easily answered, and remains to be investigated. One may refer to the assumption of Arabic-Persian grammar (Lumsden [1810.2: 130-45], mentioned earlier) that such verbal nouns are implicitly 'known' (mo'arref) similar to such generic nouns as mard 'man(kind)' in: xodä mard-rä äfarid 'God created man' where rä is obligatory. As indicated, the impression gained from most grammars notwithstanding, rä is not confined to use with direct objects. Polarization is observed in temporal and locational adverbs; e.g.: emsab-rä infä bas tonight here be 'be here for tonight' ru-yeyax-rä ab rixt-and on ice water poured they 'they poured water on the ice' (Hincha [1961: 183]). More conspicuous is a small class of verbs which take an
55 indirect object, marked either by rä or the preposition 'to' be, e.g.: ma-rä j dad be man
'He gave (to) me.'
ma-rä be man
komak kard
'He helped me!'
ma-rä ] be man
goft
'He said to me.'
Here, rä as opposed to the prepositions clearly marks polarization (ma-rä goft may also mean 'He means/said me.') On the other hand, there appears to be a referential hierarchy: if both indirect and specific direct object co-occur, the direct object is obligatory marked with rä, e.g.: be-man ketäb-rä däd, but not *ma-rä ketab(-rä) däd to me book he gave 'He gave the book to me.' Lentz (1958: 702) cites an interesting complex case of two subsequent rä: dosman-rä dar har lebäs ke bäsad, foz darb-e dädgäh-e melli, hame darbhä-rä be ru-ye u mibandim. 'As to the enemy (-rä), in whatever garb he may be, we will close all doors (-rä) on him except the door of the national court of justice.' Lentz translated the first noun-phrase as referential and not as a direct or indirect object: 'as to the enemy' (was den Feind angeht). In a brief'squib' Browne (1970) compared Persian rä with its Turkish and Macedonian equivalents. His conclusion as to rä is that it marks specificity, not definiteness. Thus, the object 'a/some book' in ketäb-(i)-rä didam booka/some I saw Ί saw a book' although (—definite) because of i, is nevertheless (+ specific) because of rä. He concluded: while all (+ definite) objects are (+ specific) and take rä, not all (+ specific) objects are (+ definite). (An example from English cited by Browne
56 is Ί am looking for an usher', which is ambiguous: 'an usher' may be 'a known, particular usher' or 'just any usher, identity not vital'). That reference to the context is distinctive was hinted at by Isenberg (1971: 156) who sees parallels in the object-markation of Albanian, Modern Persian, and Bengali. He cites it as a feature determined by the larger text-structure similar to anaphora, pronominalization, etc. Neither Browne nor Isenberg knew about the 'dative' function of rä. Since rä does not occur with all cases (e.g. never with the subject) it is evident that it is a marker of oblique cases only, where it has the functions of specificity and reference. These two notions are promising for the further investigation of rä, as well as the notion of topicalization (and commentation) which together were studied in more detail than before in the dissertation of Peterson (1974). Primarily on the basis of the facts of its distribution, Peterson argues first that rä is a marker of topicalization (i.e. theme) and second that the necessary and sufficient condition for its use is not that the noun phrase be definite but rather that it be specific, i.e., that it have, from the speaker's point of view, a unique referent. In support of the first claim, he points out that the single most salient facts about the distribution of ra are that it can occur in any noun phrase except the subject but only once in a simplex sentence. These facts can most easily be explained if we adopt the Praguean analysis (seconded by generative 'case'-grammar) in which the subject of a sentence is regarded as the primary topic. Given this, it is possible to formulate an extremely straightforward and reasonable rule: rä, a marker of (secondary) topicalization, cannot be attached to any noun phrase which has already been topicalized by being made subject. This analysis is supported, of course, by certain well-known examples (mentioned above) in which rä must be interpreted as a topic marker. Peterson's second claim rests on the following definitions of definite and specific. First, the notion of definiteness: a noun phrase is definite if the speaker 1) believes that it designates a unique individual, and 2) if he believes that the addressee shares this belief. Second, specificity: indefinite noun phrases are divided between specific and non-specific. A noun-phrase is a specific indefinite if the speaker 1) believes that it designates a unique individual, but 2) believes that the addressee does not share this believe. If the speaker himself does not believe in the existence of a unique referent, the noun phrase is non-specific. On the basis of these definitions, which are fairly well defended and explicated, Peterson notes that definites and specific indefinites have in common that, from the speaker's point of view, they designate unique individuals, and he argues that it is this property, the existence of a unique referent, that is the necessary and sufficient condition for the use of rä. He then shows that environments in which an indefinite is likely to be interpreted as specific, e.g. sentences with past tense verbs, favor the use of -rä. Thus,
57 xäne-i-rä ätes zadand is preferred over
'a house they burned' xäne-i ätes zadand
Conversely, if an environment makes a specific interpretation of a noun phrase difficult, as in komak-i 'some help
talab mikard he was seeking'
the version without rä is preferred. He notes also that while a specific indefinite may take rä, it does not necessarily do so, which is not surprising since even certain definite direct objects do not require rä (see example on p. 46 above). Peterson's study of rä, though largely based on well-known facts and incorporating a number of earlier observations, is an attempt to go beyond the traditional method of cataloguing various aspects of usage and thus to provide a unified analysis. Finally, mention should be made of as yet undiscussed instances where rä follows a relative clause, e.g.: Towzih änke Mahmud Haqiqi, nevisande-ye ketäb-e 'dar läbelä-ye äsär-e Jamälzäde', \yek nosxe az ketäb-e xod va boride-ye bahshä'i [ke dar bäre-ye Jamälzäde dar Keyhän cap sode bud] ] ra hamräh bä name Ί baräye ostäd ersäl däst. (Keyhän, 12 Amordäd 1354, p. 12) 'The explanation being the fact that Mahmud Haqiqi, the author of the book "The inner folds of Jamalzade's works" sent [a copy of his book and excerpts of discussions [which had been published about Jamalzade in Keyhan together with a letter ] ] {-ra) to the master.' (The direct object phrase and the relative clause embedded therein are marked by brackets.)
2.1.3. Noun-phrase
embedding
2.1.3.1. Ezäfe Noun-phrase embedding marked by the connective -e- has been subjected to analysis since the earliest grammars. In recent years, it became obvious that this construction lends itself to a 'generative' analysis more easily than any other construction. Traditional Arabic-Persian theory distinguished between a) the 'e of addition'
58 consisting of the mozäf eleyh and the mozäf'the attributed to and the attribute (noun)' and b) the 'e of description' consisting of the mowsuf and the sefat 'the described (noun) and the description (= adjectival phrase)'. Such an analysis was based on the recognition of the underlying logical relationships, which are neutralized in ezäfe constructions, i.e., between the attributed and the attributing part of the construction. The formal means of demonstrating the deep syntax was the paraphrase. This traditional 'logical' approach is further evidenced by the fact that attributions without the connective -e- are also included in discussions of the ezäfe such as the fakk-e ezäfe 'the sealed ezäfe' (e.g. , säheb-xäne 'house-owner') and the ezäfe-e maqlub 'the inverted ezäfe\ e.g. irän-dust 'Iranophil'. Thus, the ezäfe was seen as the neutralization of prepositions like az 'from, o f , dar 'in', bar' on', cun 'like', baräye 'for'. The most extensive survey of the traditional analysis and interpretation of the ezäfe can be found in Mo'in (1952). In terms of descriptive detail and syntactic-semantic insights into the ezäfe in contemporary Persian, Pejsikov (1959: 41-108) stands out. Unfortunately his descriptive approach did not provide him with the more sophisticated algorithm which is needed to handle the syntactic/ semantic regulations in a less circumlocutory manner. The pattern of the combinatory distribution of ezäfe constructions was concisely described first by the structuralist Hincha (1961). In the following only a very few of the many ezäfe constructions are discussed. Two main classes of the ezäfe were analyzed and defined by Palmer (1970), who, in order to systematize and prove older insights, utilized the methodologies of case-grammar. The first type is shown to originate in an underlying relative clause with a) budan 'be' (adjectival ezäfe) and b) dästan 'have' (possessive ezäfe) in their predicates, e.g.: 'be':
ketäb-i ke gerän ast => ketäb-e gerän 'the expensive book'
'have':
ketäb-i ke hasan därad => ketäb-e hasan 'the book of Hasan'
(cf. the traditional insight that e originally is the predicative 'is', Lumsden [1810 2: 294]). With regard to possession, Palmer is the first to recognize and to discuss the difference between alienable and inalienable possession; thus ketäb-e hasan 'the book of Hasan' can be derived from the restrictive relative clause
59 ketäb-i ke hasan därad 'the book that Hasan has' but
dast-e hasan 'the hand of Hasan'
cannot be thus derived, *dast-i ke hasan därad '*the hand that Hasan has' unless the intent is a separated hand held by Hasan. Similarly, a sentence like: dast mal-e hasan ast 'the hand is Hasan's' can only refer to this possibly macabre situation, but not to Hasan's attached hand. Palmer's case-grammar solution to the problem of inalienable possession is to refrain from deriving such constructions as dast-e hasan from an underlying restrictive relative clause, but instead to mark inalienably possessed nouns in the lexicon as nouns requiring, in their deep structure, the 'dative' of the possessor. This procedure, however, eva:des the problem rather than solves it. Thus, the adjectival and alienable possessive eza/e-nominalizations are generated by the ellipsis of the verbs 'be' and 'have' (which two are considered dummy verbs by case-grammarians since they are predictable from the underlying casestructure and add no semantic meaning; cf. pp. 56 abpve). Secondly Palmer discussed the infinitival ezäfe (actually, gerund-eza/e); it is generated by the nominalization of the verb with which the cases are adjoined by the ezäfe: e.g.
hasan raft => raftan-e hasan 'Hasan's going'
or:
(hasan) be tehrän raft => raftan-(e hasan) be tehrän 'going of (Hasan) to Tehran'
The infinitivization has to operate on the entire verb phrase as a complex unit, i.e. includes foreverbs, as in:
60 väred sodan-e... pä'in ämadan-e..
'the entering of... 'the coming down of..'
Infinitivization is a very late, 'near-surface' operation in terms of generativetransformational grammar. It presupposes the generation of the 'passive' construction, i.e. there is an infinitive of an active (non-passive) distinct from an infinitive of a passive, e.g.: sir koste sod => koste sodan-e sir 'the killing (being killed) of the lion (subj.)," vs. sir-ra kost => kostan-e Sir 'the killing of the lion (obj.)' It is not so much the basic analysis of these types as it is the specific conditions and restrictions identified by the generative approach, which represents a definite progress in our understanding Persian (cf. also the discussion of compound verbs). It is not possible in this context to refer exhaustively to Palmer's and similar studies, or to include critical remarks on a good number of false or incomplete assumptions and conclusions. One of the more intricate problems of the ezäfe involves constructions with participles (only hinted at by Palmer [1970: 90-92]in the context of the adjectival ezäfe). Similar to infinitivization, such participial constructions nominalize, i.e., participialize, the entire verb phrase; they include foreverbs and apply only after the passive transformation (cf. discussions of nominal compounds below): kafs'(-i ke) väks zade (sode) ast 'the shoe (which) was waxed' kaß-e väks zade => shoe — polished 'the polished shoe' However, I may add, if the subject of the relative clause is different from the head-noun, the participial ezäfe transformation is restricted to inalienable possession; thus it is grammatical and quite usual to have: kot-i the coat
ke which
rang-as its color
paride ast faded
61 => kot-e rang paride 'coat (with) faded color' or: pesar-e boy (with) 'a cursed boy'
pedar father
suxte burned
but it is ungrammatical to derive the following: pesar-i ke masin -as suxte ast ' a boy whose car burnt out' => * pesar-e mäsin-suxte * 'a boy with a burnt car' This observation as well as that made by Palmer (mentioned above) indicates that the semantic differentiation between alienable and inalienable possession is an important one in Persian, which needs to be investigated. Participial eza/e-constructions, especially those with purely passive connotations, are quite rare in Persian, if derived from Persian transitives, e.g.: mard-e koste
'the killed man'
This was observed by Pejsikov (1959: 81). On the other hand, he found that participial constructions with borrowed Arabic active and passive participles are quite common without exhibiting any particular restrictions, e.g.: mard-e äqel mard-e maqtul
'the knowledgeable man' 'the killed man'.
This observation provides evidence for the fact that the Arabic loan component · in Persian is of far greater importance than is generally realized (for further observations see the discussion of nominal compounds below). One of the unsolved problems with regard to the ezäfe is its relationship to appositional constructs and the problem of restrictiveness (i.e. the traditional 'partitive' construction). It has been very briefly touched by Seiler (1960: 11819) in his study of the relative clause, attribution, and apposition in general and in Avestan in particular. He tentatively suggests that the basic differentiation of the syntactic-semantic relationships of modern Persian may be identical with those of Avestan. Referring to Lentz' survey (1958: 209 ssq.), he found: the ezäfe with the e (derived from an earlier relative pronoun) identifies the attributive relationship; the absence of e, however, marks apposition, e.g.:
62 appositive: yek qäsoq cay 'a spoon (full of) tea' attributive:
yek qäsoq-e cäy 'tea-spoon' or 'a spoonful of this tea' The latter shows surface ambiguity, it can be read either as 'a tea-spoon' or 'a spoon of the tea (talked about)'; i.e., in the first instance the opposition is between 'tea-spoon' as opposed to other kinds of spoons, in the second reading 'this tea' is selected as opposed to another kind of tea (or 'coffee', then with selective stress). 2.1.3.2. Subordinate clauses The discussion of the syntax of subordinate clauses in a more than cursory manner was confined to a few outstanding grammars such as Lumsden (1810 2), Phillott (1919), and, most of all, Jensen (1931) and Arends (1941). Since then, it has been mainly Soviet scholars who have devoted monographs and articles to this problem. The discussion of subordinate clauses is organized according to the 'case' and/or semantics of the noun-phrase into which the subordinate clause is embedded. As an example, Lazard (1957) after a brief summary on subordination discusses the following clause types: 1) completive, 2) relative, 3) temporal, 4) final, 5) consecutive, 6) causal, 7) concessive, 8) comparative, 9) conditional clauses. This list is more or less identical with others such as Jensen's (1931), who, however, classified them as 1) subject, 2) predicative, 3) object (relative and conjunctional), 4) attributive, 5) adverbial clauses. 2.1.3.2.1. Relative clauses Some features of the internal structure of relative clauses are among the first syntactic observations made by early scholarship: there are no relative pronouns, or only rudiments thereof; instead, the related identical noun occurs as a personal pronoun, separate (I) or suffixed (II). Thus the following paradigm was created (only the first two persons cited): I
II
gen. man ke pesar-e man / man ke pesar-am to ke pesar-e to / to ke pesar-et acc. man ke ma-rä / man ke-m to ke to-rä / to ke-t (cf. Vullers [1870: 205]).
Ί whose son..' 'you whose son.. Ί whom..' 'you whom..'
63 Only ke and ce in har (an) ke 'each one who, whoever', and har (än) ce 'whatever' have preserved their pronominal characteristics although their status in contemporary Persian is uncertain. However, the related pronoun is not always surfaced. In contemporary Persian, except for emphatic instances, the pronoun is elided if it functions as the subject or direct object of the relative clause: mard-i ke mard amad => mard-i ke (u)ämad 'the man that (he) came' mard-i ke mard-rä didi => mard-i ke (u-rä) didi 'the man that (him) you saw' Ellipsis is also usually found under as yet undetermined conditions with adverbial cases, e.g.: fd-i ke be fa residi => jä-i ke (be an) residi 'the place that (to it) you came to/arrived at' (An excellent survey of the relative clause is still Jensen [1931: 288-97], cf. also the details in Lumsden [1810: 2: 89-106, on the ellipsis of adverbials p. 98].) Scholars have early recognized a peculiar syntactic rule of relative-clause embedding: the 'prolepsis' of a preposition and, more frequently, of the postposition ra (e.g. Lumsden [1810 2: 201]). This construction, criticized by modern purists as a sign of the bad grammar of many journalists, is quite old, e.g.: [to-rä] ke dast belarzad, gowhar ce dani soft 'you (dat.) whose hand trembles, how can you pierce a jewel' (cf. Jensen [1931: 294]) [be har zarf-i] ke sar-etan begonjad,... 'into each pot which fits your head...' (iibidem) A convincing syntactic explanation for this prolepsis has not been offered. A most crucial problem with Persian relative clauses is their 'types' and
64 their relationship to the head noun. The relative clause is said to be either restrictive (or determinative) or non-restrictive (cf. Rubincik [1955]). The head noun-phrase of restrictive clauses is marked as [+ specific], expressed by a demonstrative pronoun and/or the demonstrative suffix i: än mard(-i) ke diruz raft => mard-i ke diruz raft 'that/the man that (he) went yesterday' The occurrence of the demonstrative i with restrictive relative clauses (cf. discussion of ι) has hampered, it seems, the research on this subject. Throughout, the issues are not separately discussed. Most importantly, a distinction must be made between the head-noun on the one hand and the relative clause on the other hand: thus, the relative clause may be restrictive or non-restrictive; but also, a head-noun of a relative clause may be specific or not, definite or not. The question to be asked is: what are the combinatory patterns of these constitutents. The issues may best be illustrated by an ambiguous example: mard-i ke diruz ämad... This has three readings: a) än mard-i ke 'that man who'; in this case the head noun-phrase is [+specific] [+ definite] ,the i is therefore the demonstrative /; b) and c)yek mard-i ke 'a/one certain man who' in this case, the relative clause would be restrictive, too; however, the head noun-phrase is ambiguous: it may be read as [-definite] but [+ specific], i.e., 'a certain man who', or it may be read as [—definite], [—specific], i.e., 'a (any one) man who'. Irrespective of the three readings of the head noun-phrase, the relative clause here is restrictive with all three readings. The syntactic difference between restrictive and non-restrictive relative clauses remains an open question. It has been suggested by Palmer (1970: 75) that the non-restrictive clause is to be derived from an underlying coordinate clause; e.g., he would analyze a non-restrictive clause from a coordinate clause as follows: coordinate: mard-e mosenn-i v&red sod [va kif dast-as bud] relative: mard-e mosenn-i vared sod [ke kif dast-as bud] embedded: mard-e mosenn-i [ke kif dast-as bud]väred sod 'an older man entered [ (and) a briefcase was in his hand'] 'an older man entered [who had a briefcase in his hand'] 'an older man [who had a briefcase in his hand] entered'
65
S mard-e mosenn-i väred sod 'an older man entered
co'nj^ va and
" S kifdast-e mard bud a briefcase was in his hand'
To this then apply the relative clause embedding rules. The essential difference between non-restrictive and restrictive relative clauses may be illustrated as follows, each with a. [+ specific] [+ definite] head noun and b. [+ specific] [—definite] head noun (a non-specific head noun seems to exclude a non-restrictive relative clause):
1. non-restrictive relative clause a. (an) mard-e mosenn 'the older man
ke kif dast-as bud who had a briefcase'
b. (yek) mard-e mosenn-i 'an older man
ke kif dast-as bud who had a briefcase'
2. restrictive relative clause a. fan) mard-e mosenn 'the older man
-i ke kif dast-as bud who had a briefcase'
b. (yek) mard-e mosenn-i 'an older man
-i ke kif dast-as bud who had a briefcase
(where -i + i
/).
Thus, only restrictive relative clauses are introduced by what is called 'demonstrative' i. Consequently the function of the 'demonstrative i is to identify restrictive relative clauses only. Its presence or absence is the morphological criterion for distinguishing between the two types of relative clauses. The syntactic criterion for the distinction between the two types has been observed by many; it is the fact that restrictive relative clauses cannot be separated from their antecedent noun-phrase, while non-restrictive clauses can. — This function of the 'demonstrative' i is reminiscent of, if not identical with, the function of i
66 as 'restrictive', better 'reference-dividing' vs. 'non-reference-dividing' (cf. discussion of ι above). In conclusion one may tentatively state that restrictive relative clauses are obligatorily introduced by i but that non-restrictive relative clauses are obligatorily unmarked, yet their head-noun phrase may be 'restricted' or 'reference-divided' as shown above. This statement implies that i has two different functions which somehow overlap: a) as [+reference-dividing] of the head noun-phrase of relative clauses and b) as [+demonstrative], identifying the embedded attribute-relative clause as [+restrictive]. This 'double' function is not confined to relative clauses. In fact it may facilitate the understanding of an ezäfe construction found in elevated style, e.g. mard-i-st bozorgvar 'he is a man, magnanimous' Here, quite similar to the option with non-restrictive relative clauses, the adjectival attribute appears separated from its head noun by the intervening verb. Mard-i is evidently reference-divided/restricted. In fact, all cases like this obligatorily have /. But the attribute is not included/embedded. In non-elevated style all attributes are obligatorily embedded, so that there are but two surface constructions: a. mard-e bozorgvar ast b. mard-e bozorgvar-i-st Here a. is specific restricted: 'he is (a) magnanimous man', but b. is ambivalent; it implies either that the entire noun-phrase is restricted, i.e. 1. that he is one of the set of magnanimous men: 'he is a [magnanimous man]' or 2. that the head noun only is restricted, i.e. that he is a man who is magnanimous: 'he is [a man [magnanimous] '. It is in these cases that the elevated register admits non-embedding of the attributive adjective. There is in fact no double function at all; rather there is a 'double ' application of one and the same function: restriction/reference-dividing can apply to either the entire noun-phrase or to the head noun only. While the latter is found in all styles of language with nonrestrictive relative clause attributes, it is confined to the elevated register with other attributes. Constructions such as mard-i bozorgvar have been little discussed. Chodzko (1883: 113) describes this construction as a 'deviation from the general rule', but admits that contemporary authors use it just as did Sa'di. Similarly Phillott (1919: 133) mentioned it as an
67 alternative possibility in 'old Persian', but added 'classical and modern' in parentheses. Lazard (1957: 200) described this construction as a rare possibility in the presence of 'qualificative' adjectives. Jensen (1931: 56-57) was much more specific. In the older language the 7 of unit' was added to the head-noun only. Thus it was possible to distinguish asb-i digar 'another horse' and asb-e digar-i 'the horse of somebody else'. In the presence of more than one adjective, however, the i is affixed to the final adjective, i.e. to the entire noun-phrase. In turn, if the head-noun is in the plural the i was invariably affixed to the plural, e.g. bandegän-i sadeq 'loyal servants'. In contrast to these regulations for the of unity', the 'demonstrative' i before relative clauses has to be phrase final, i.e., affixed to the final attribute. One of the better discussions of this construction is Xoräsäni (1950b: 527-28). He pleaded for following the example of the earlier masters and for affixing the i to the noun, not to the adjective. The reason is to avoid ambiguity. In an appealing, didactic fashion he described the problem as follows: when once told that someone äseq-e zan-e ruzexan-i sode he interpreted this as someone having fallen in love with the 'wife of a preacher'. Only when told that this happened because of her nice voice did the ambiguity become apparent; she was a 'woman-preacher'. Thus she should have been identified as zan-i ruzexän 'a woman (who is a) preacher', which eliminates the ambiguity. Similarly one should say qadah-i ab 'a cup of water' rather than the ambiguous qadah-e ab-i 'a cup of (a) water'. As noted, the general non-elevated rule today is, however, the ambiguous construction where the i is affixed to the attribute, i.e., the entire complex noun-phrase. The earlier rule still applies to indefinite pronouns followed by digar, e.g. kas-i digar, yek-i digar and ciz-i digar (ciz-e digar-i 'someone else, another one, something else'.) Most types of relative clauses are obligatorily introduced b y ke. This is n o t the case with a particular sub-class o f generalized relative clauses introduced b y har 'each, every', where ke is 'optional', e.g. harkas(i)
(ke) sarbäz
bude...
whoever was a soldier once...' To these may belong temporal clauses introduced b y (har) vaqt(-i) e.g.: (har) vaqt(-i)
'each time',
residi...
'when (ever) y o u arrive...' as well as clauses introduced by conjunctional phrases like az bas(ke) ally 'enough'), e.g.: az bas (ke) xaste sod... 'he became so tired that...'
(cf. Rubincik [ 1 9 6 3 a ] ) .
'so' (liter-
68 The 'optionality' of ke (besides that of f ) here is evidently semantically conditioned by the fact that the head-noun is [-definite] [—specific]. This in turn would indicate that ke is not a dummy but implies semantic differentiation yet to be determined.
2.1.3.2.2. Other subordinate clauses The general approach is to list, for each clause type, the conjunctions or conjunctional phrases used to express the basic meaning and its nuances and their combinatory possibilities; compare, e.g., the detailed list of causal conjunctions by Lazard (1957: 238): zi-rä, zirake 'parce que', cun, cegune, cunke 'comme, puisque', ce, ceke 'car, car que', baräy e inke 'parce que', az in fehat ke 'pour la raison que, parce que', be ellat-e inke 'ä cause (de ceci) que', az änfä-ike 'du fait que' (literally, 'de ce lieu que'), to which is later added simple ke, and az (bas(ke) 'ä force de (litt, 'que'), tant' ('if the cause is conjoined with intensity'); and finally, a paratactic alternative is offered for the latter: in qadr 'tellement, tant'. With other clause types, such as temporal, conditional, and final clauses, notes are added on the possible omission of the conjunction, or the sequence of tenses. In Persian, as in other languages, the problems of the 'cases' and of the 'tenses' are compounded with the problem of noun-phrase embedding, and there is little likelihood that the intricacies of the syntax of clause subordination will be solved before the problems of the simple single clause. More recent grammatical theories have offered little help so far to approach these levels. One of the more basic problems is that of the general properties of clause subordination in Persian. It has been recognized for a long time that 'subordination is indicated mostly by subordinating words: conjunctions, conjunctive phrases, pronouns, pronominal phrases' (as phrased by Lazard [1957: 211]). The conjunctional phrases deserve more attention than just listing them and than just recognizing the fact that the demonstratives an/in occur in a significant number of such phrases, as does the demonstrative i\ see, e.g., the causal conjunctional phrases: be ellat-e in ke be ellat-i ke be an ellat ke
'for the cause of this that' 'for the cause that' 'for that cause that'
The abstract 'it' character of the demonstratives in/än becomes evident after
69 prepositions such as ba in ke 'with this that' ('because') (cf. German dafür, dass, etc.). The same injän can be shown to underlie other subordinate clause types such as predicative, subject, and object clauses; the emphatic-topicalized inversions of such clauses obligatorily insert in (or άη)\ e.g.: xubast (ke) diruz naraftim => inke diruz naraftim xub ast this that yesterday we did not go good is 'it is good that we did not go yesterday' fahmidam (ke) to narafti => inke to narafti fahmidam this that you did not go I understand Ί understand that you did not go' It would be important to investigate the possibility that all subordinate clauses in Persian are to be derived from underlying relative clauses, and if so, to identify the types of subordinate clauses in terms of their specific syntactic-semantic conditions. One puzzling question in relation to subordination is the particle ke. Already Lumsden (1810 2: 92) had recognized that ke and ce are particles and not pronouns, an insight which was not heeded in many subsequent grammars. Ke is the most used subordinate 'mot-outil' as Lazard (1957: 211-14) puts it, who lists the following usages of ke : 1) introduces relative clauses 2) introduces subordinate completive clauses 3) functions as expletive after other conjunctions (e.g., agar (ke) '(even) i f ) 4) combines with other elements to form numerous conjunctive phrases 5) serves occasionally to introduce the apodose in conditional sentences and 6) is used in idiomatic expressions (examples for point 1-5 can be found in the list of causal conjunctions cited above) The apparent multitude of the functions of ke is quite easily explained by the observation that ke in contemporary Persian is a 'dummy' which is inserted in the surface string in case of subordinate clause embedding. The rules for its obligatory occurrence or omission have yet to be specified (cf. Rubincik [ 1959a] Windfuhr [1971]).
70 2.1.3.3. Topicalization The notion of topicalization in Persian is fairly recent; Peterson (1974) is one of the first to touch on it in the context of rä (cf. above). Constructions involving topicalization, however, have been observed and described since the earliest grammars, usually described as emphasis (in Persian ta'kid). (This is not unlike the case of the perfective aspect in the verbal system which similarly has usually been described as emphasis, ta'kid.) 2.1.3.3.1. Emphatic ke and anacoluthon The observation that ke may function as an emphatic, most prominently in the colloquial near-standard dialect, is not new; thus Chodzko (1852: 141) suggested the translation-meaning to be 'voilä, mais, eh bien', e.g.: rrn ke raftim 'nous voilä partis' u ke mord 'mais il est mort' This usage is old, e.g.: hec amsäsfand ke an qovvat nadarand ke... T)ut no amesa spenta has the power to...' found by Amuzegär-Yegäne (1967) in early Zoroastrian-Persian texts. Jensen (1931: 225-28), who had utilized Phillott (1919) extensively, saw an idiomatic 'Hervorhebung', and syntactically identified such cases as a special type of relative clause peculiar to colloquial Persian: man ke änfä nabudam 'Ich war ja garnicht dort', 'but I wasn't (even) there' In fact, on first sight this seems to be true; similar to relative clauses, the rule of pronominalization applies in such cases; e.g.: hasan ke howsele-e hasan sar rafte => hasan ke howsele-as sar rafte 'Hasan ke his patience ran out.' Firstly, the pronominalization rule is not bound to relative clauses but applies to any sequence of identical noun-phrases. Further, ke is not bound to the first
71 noun phrase; e.g.: man ke pul nadäram man pul ke nadäram man pul nadäram ke Ί have no money' Rather, ke appears to be able to 'emphasize' any noun-phrase or the entire clause. These are evident cases of topicalization, and thus probably related to what has been called nominativus pendens or anacoluthon (well known already in Old Persian); e.g.: mahmud ketdb-e mahmud gom sod => mahmud ketab-as gom sod 'Mahmud, his book got lost' Such topicalization has been called the 'preposition of the psychological subject' o f a clause (Jensen [1931: 191]; cf. also Homä'i [1959: 145]). This construction has been discussed in most detail by Pejsikov (1959: 18796, 380-87). His results are not quite conclusive. He basically accepted Arends' (1941: 32-34) interpretation that in a construction like amu zan-ds... 'uncle, his wife...' amu is the 'emphasized subject' while zan-as constitutes a referential part of the predicate in the clause, either as subject or in any other case-function. While refuting Rastorgueva's (1953) assertion that such constructs constitute a unitsubject, equivalent to the ezäfe — construction zan-e amu 'uncle's wife' (which she evidently based on Tajiki-Persian evidence), he did not provide sufficient syntactic-semantic criteria to separate the two constructions and, in fact, argued that in certain cases they may indeed constitute a unit. In terms of syntax the anacoluthon is not to be confounded with the ezäfe (as Pejsikov seems to do); rather, it is a case of obligatory pronominalization of the second of two identical noun-phrases: mahmud ketdb-e mahmud... => mahmud ketab-as 'Mahmud, his book'
72 In fact, traditional grammar considered such constructs a species of badal; the first member is introduced in the sentence 'merely as a preliminary in order to prepare and fix the attention of the bearer on the badal by which it may happen to be followed' (see Lumsden [1810.2: 228-31]). In order to account syntactically for these constructions it would thus be possible to formulate a topicalization rule by which a particular constituent part of a clause is reduplicated and preposed to the clause, creating an e-vocative phrase. To this then would apply the rule which pronominalizes the second of two identical noun-phrases: ketäb-e hasan gom sod => hasan ketäb-e hasan gom sod => hasan ketab-asgom sod 'Hasan, his book got lost' The possibility of A:e-insertion suggests the presence of a subordinate clause in these constructions; and, moreover, the fact that ke is not obligatory suggests that this subordinate clause is not a relative clause, but rather a complement clause. In this context the so-called indirect or middle verbs should be mentioned (cf. below) which imply, obligatorily, an underlying dative pronoun, e.g.: man howsele-am sar rafte I my patience has run out Ί have lost my patience.' qahr-emän sod quarrel-our/to us happened 'we began to quarrel'
2.1.3.3.2. Subject reduplication Moyne (1971: 158, expanded 1974 together with Carden) in his discussion of the syntactic-semantic properties of the emphatic vs. the reflexive marker xod refers to 'subject reduplication'; 'the derived subject of a sentence can be reduplicated for emphasis (optional)', e.g.: husang, husang dide sod => husang, u dide sod 'Husang, he was seen'
73 (with obligatory pronominalization of the second noun-phrase). This — passive — example demonstrates, as Moyne convincingly argues, that subject reduplication applies only to the derived subject (here: passive subject = underlying object); i.e., only after the logical/underlying object such as hus'ang has become the subject of the clause can it be reduplicated. Reduplication of a non-subject is blocked; e.g. there is no: *husang zdle-rd, u-ra, zad' *Husang hit Zale, her.' This is not quite true; note the reduplication of the object in: zale, husang urd zad 'Zale, HuSang hit her.' Evidently, reduplication is a case of anacoluthon and thus topicalization. 2Λ 3.3.3. Emphatic and reflexive xod The Persian reflexive xod has presented considerable problems which were, however, not recognized as such. It was seen that xod can fulfill the same functions as the personal pronouns, i.e., object, prepositional complement, determinant of a noun. It can further be substituted for by the simple personal suffix, if the identity with the subject is not emphasized. It can be used as a non-reflexive to underline identity, in which case it may precede or follow the respective noun or pronoun (Lazard [1957: 110-11J; cf. Chodzko [1852: 115-20]), e.g.: reflexive: xod-(as )-rd kost non-reflexive: hasanxod xod-e hasan
'killed himself 'Hasan himself; (lui meme).
Moyne (1970; especially 1971) clarified the issue considerably. Applying the methodology of generative grammar, he came to the following main conclusions (rearranged and somewhat streamlined here): 1) reflexive The direct and indirect object (and, one should add, other cases) are obligatorily reflexivized if identical with the subject of the same clause (German sich), e.g.: hasan hasan Hasan 'Hasan did
az (hasan) az (xod-(as)) from self not like himself
xos-as naydmad => xos-as naydmad well —him came—not
2) emphatic Any noun-phrase can be emphasized by preposing xod (cf. German selbst); e.g.:
74 subject: (xod-e) hasan dmad 'Hasan himself came' object: (xod-e) zale-rä did 'he saw Zale herself 3) combination of emphatic and reflexive hasan ketäb-e (xod-e) hasan-rä gereft => hasan ketäb-e (xod-e) xod-rä gereft => hasan ketäb-e (xod-(as)) -rä gereft 'Hasan took his own book' Moyne himself (1971: 160) referred to similar observations on emphasis by classical grammarians. It would be interesting to investigate other means of emphasis/reduplication discussed by Lumsden (1810.2: 220-28) and Telegdi (1955: 133), who refer to the classical notion of ta'kid 'emphasis' expressed by Arabic equivalents for xod such as nafs 'self, ayn 'identical, self, and semantically more complex emphatics as hame, fam'iyat 'altogether', e.g.: mä hame (=män) hastim 'we, all of us are here' 2.1.3.3.4. mar In this context one may cite mar, a particle which is confined to earlier forms of Persian. It occurred with the subject (only with demonstratives), as well as direct and indirect objects (with or without rä). Its exact function is unknown, but it may have been to mark 'restriction', or 'Hervorhebung' (Jensen [1931: 46]). Gray (1937) saw one of the functions of mar to be emphasis of the subject if combined with demonstratives, or generally as a 'mere intensive particle'. Lazard (1963: 451) suggests its function to have been 'mettre en relief le terme qu'elle precede: 'precisement, justement'. — It does not seem impossible that mar was an 'emphatic' colloquialism similar to the 'emphatic' ke in modern colloquial Persian. 2.1.3.4. Participial clauses Participial clauses are generally confined to the literary dialect. There are two types, subordinate participial clauses (fe'l-e vasfi 'participial verb form') and coordinate participial clauses (hezf-e fe'l 'ellipsis of (finite) verb marking'). Coordinate participial clauses imply a literary rule by which all but the last
75 verb form of coordinate clauses are surfaced as a (perfect) participle; e.g.: vazir bar gaste, να esterähat xähand nemud 'the vezir returned, and will relax' 'the vazir will return and relax' Subordinate participial clauses participialize the verb of the subordinate clause; e.g.: vazir bar gaste esterähat xähand nemud 'The vezir having returned will relax.' Both constructions generally require identical subjects, but not obligatorily so. Both constructions are known in early Persian already, where the finite form was not obligatory the final one (for their synchrony and diachrony cf. Jensen [1931: 140-42]; Bahär [1947 I: 352; and passim]). As can be seen from the two examples above, which on the surface differ only by the conjunction va 'and', the differentiation between the two constructions is not always self-evident. It is worth noting that it appears to be a stylistic feature of pseudo-literature to insert wrongly a va 'and' after a subordinate participle; this habit has been critized repeatedly by literati on stylistic and partially linguistic grounds (e.g., Xänlari [1961: 330-33]), but again, is found in earlier forms of Persian. These participial clauses raise several important questions. One of them is the possibility of interference from Turkish (which has several participial constructions roughly equivalent to the Latin participium coniunctum). This, in turn, raises the question of the development of participialization in Tajiki— Persian as opposed to participialization in Iranian Persian with its syntactic and 'stylistic' restrictions (cf. discussion of compound nouns below). In terms of general linguistic theory, the important question raised is the question of the place, the order, and the conditions under which not only tense, mood, and aspect are mapped into the transformational chain, but also the subject marker, i.e., the personal ending. To put it differently, the question is whether participial clauses are nominalizations of underlying finite clauses, or whether, conversely, the markings of finiteness are not surfaced in participial clauses. 2.1.3.5. Nominal word classes In view of the fact that nouns in Persian are not inflected, the general opinion held by scholars was that the nominal word classes are badly defined (as late
76 as Rastorgueva [1953] = [1964: 16]). This view was expressed in particular with regard to the distinction between nouns and adjectives. However, Ovcinnikova (1961) offered proof for exactly that distinction. Arguing that the difference is not simply and merely one of semantics, she considers morphological and syntactic criteria. Thus, a noun is distinguished by the fact that 1) it occurs in all parts of speech, including the predicate, that 2) it takes prepositions, and that it admits 3) plural, 4) indefinite i, 5) numbers, 6) attributes, 7) pronominal (personal) suffixes, 8) demonstrative pronouns. The distinctive criteria of the adjective are that it admits comparison and adverbials like xeyli 'much', kämelan 'completely'. Nominal affixation with adjectives is found only where the noun of a noun-adjective phrase is elliptic. Such a possibility of 'transposition' by ellipsis of the noun does not blur the distinction between the two classes, but on the contrary, presupposes it. This is exactly the conclusion arrived at by Telegdi (1962) who proved the distinction (using early transformational methodology) by contrasting seemingly identical pairs like del-tang and tang-del (cf. the discussion of nominal compounds). 2.1.3.6. Nominal compounds From the earliest grammars of Persian onward one can observe a cyclic movement in the approach to derivation and compounding: first, morphological derivation by pre- and suffixation gradually was separated from derivation by compounding. Then, more recently, the tendency has again been to describe both processes together, with more grammatical sophistication after the dialectically necessary phase of separation. The 'affixal' phase was decisively influenced by the morpho-etymological interest of comparative-historical grammarians. Horn's description in the Grundriss der Iranischen Philologie represents the last grand summary of those efforts. Jensen (1931) is the last great grammar to include some explanatory comparative-historical notes on productive and unproductive (fossilized) affixation. Another source of the attention paid to affixation was the traditional Persian approach where homophonous/homographic units were lumped together according to the (last or first) letter of the graphic unit, and their various meanings and functions listed. This tradition is still found preserved in Vullers (1870: 250, 252), who, e.g., under the heading -/, lists the stressed abstract derivative i, the denominalizing i, together with the unstressed enclitic i. After some promising beginnings (Nye [1955], Lazard [1957]), the major breakthrough is connected with the works of Pejsikov (1959a), Shaki (1963, 1964a, b), and Telegdi (1962, 1964). Being descriptivist-structuralists at a time when structuralist linguistics was already highly advanced, they skipped the
77
morpheme-oriented phase and immediately approached the structural-syntactic types of word-derivation and compounding - termed types of 'transposition' (the term is adopted from Bally by Pejsikov, who restricted it to non-suffixal exocentric constructs; Lazard [1957: 256] used the term in the general sense). The most convenient summary of these, especially of his own efforts on the analysis of nominal compounds, was presented by Shaki ([1963], cf. also [1964 b: 112-13]). He identified five basic exo- and endocentric types with productive and unproductive subtypes; they are (in abbreviated form): 1.copulative (gav-mis 'water buffalo' lit. 'cow-sheep'); 2. reduplicative (sar-ä-sar 'all over' lit. 'head-to-head'); 3. determinative: a. dependent, i.e., with case-relationship between the components (dänes-ju 'student' lit. 'knowledge-seek(er)', b. descriptive, attributive, numerical, adverbial (now-favän 'youth' lit. 'newyoung'), c. bahuvrihi: (gol-rox 'flower-faced'); 4. verbal phrase: (negäh-dast 'caring' lit. 'observation-hold'); 5. anomalous: (aftäb-zard(i) 'sun-set' lit. 'sunshine-yellow(ness)'). This list is preceded by a list of such features of compounds as syntactic modifications, morphophonological modifications, juncture features, internal structures, 'sense', and the distinction between endocentric and exocentric constructs. The recognition that many types of compounds derive from an underlying £za/