The foundations of Arabic linguistics. IV, The Evolution of Theory 9789004389687, 9004389687

Citation preview

The Foundations of Arabic Linguistics IV

Studies in Semitic Languages and Linguistics Editorial Board Aaron D. Rubin and Ahmad Al-Jallad

volume 97

The titles published in this series are listed at brill.com/ssl

The Foundations of Arabic Linguistics IV The Evolution of Theory

Edited by

Manuela E.B. Giolfo Kees Versteegh

LEIDEN | BOSTON

The Library of Congress Cataloging-in-Publication Data is available online at http://catalog.loc.gov LC record available at http://lccn.loc.gov/

Typeface for the Latin, Greek, and Cyrillic scripts: “Brill”. See and download: brill.com/brill‑typeface. ISSN 0081-8461 ISBN 978-90-04-38968-7 (hardback) ISBN 978-90-04-38969-4 (e-book) Copyright 2019 by Koninklijke Brill NV, Leiden, The Netherlands. Koninklijke Brill NV incorporates the imprints Brill, Brill Hes & De Graaf, Brill Nijhoff, Brill Rodopi, Brill Sense, Hotei Publishing, mentis Verlag, Verlag Ferdinand Schöningh and Wilhelm Fink Verlag. All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher. Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill NV provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, MA 01923, USA. Fees are subject to change. This book is printed on acid-free paper and produced in a sustainable manner.

Contents Notes on Contributors

vii

Introduction: the Evolution of Theory in the Arabic Linguistic Tradition Manuela E.B. Giolfo and Kees Versteegh Contribution to a Modern Reading of Sībawayhi Hassina Aliane

1

10

Pronouns in Sībawayhi’s Kitāb and Related Concepts: ḍamīr, ʾiḍmār, muḍmar 30 Georgine Ayoub Grammar for Beginners and Ibn Hišām’s Approach to Issues of ʾiʿrāb Ramzi Baalbaki Sallaṭa/tasallaṭa, a Possible Parallel for ‘Govern’? Michael G. Carter

61

89

The Notion of taqdīm wa-taʾḫīr in al-Kitāb and Its Development in the Arabic Grammatical Tradition until the 4th/10th Century 106 Hanadi Dayyeh The Intriguing Issue of Dictionary Arrangement in Medieval Arabic Lexicography 123 Joseph Dichy Can Ambrosiana X Sup 56 Improve Our Understanding of Sībawayhi’s Grammar? 133 Jean Druel Conditionality: Syntax and Meaning in al-Sīrāfī and Ibn Sīnā Manuela E.B. Giolfo and Wilfrid Hodges

157

The Technical Terms taqdīr and tahfīf in Persian Classical Sources Éva M. Jeremiás

182

vi

contents

How to Parse Effective Objects according to Arab Grammarians? A Dissenting Opinion on al-mafʿūl al-muṭlaq 198 Almog Kasher The Phenomenon of ittisāʿ al-kalām in Old Arabic Aryeh Levin

212

Which Verbal Nouns Can Function as Adverbial Accusatives of State or Condition (ḥāl) according to Sībawayhi and Later Grammarians? 225 Arik Sadan What is Definiteness in Arabic? Focusing on Proper Nouns for Genera and ʾasmāʾ mubhama ‘Ambiguous Nouns’ 233 Haruko Sakaedani Definition and Determination in Medieval Arabic Grammatical Thought 253 Manuel Sartori The Concept of tawṭiʾa in the Medieval Arabic Grammatical Tradition Beata Sheyhatovitch Malay Grammar between Arab and Western Model Kees Versteegh Index

319

295

274

Notes on Contributors Hassina Aliane holds a Ph.D. in Computational Linguistics. She is director of the Digital Humanities research division and head of the Natural Language Processing (NLP) and Digital Content team at the Algerian Research Center on Scientific and Technical Information. She is working on developing tools and resources for Arabic NLP but her main research interest is understanding Sībawayhi’s thought and methodology to get new insights for Arabic (computational) linguistics and (computational) linguistics more generally. Georgine Ayoub is professor of Arabic linguistics at the Institut national des langues et civilisations orientales (INALCO), Paris, France, and a researcher at Cermom in the same university. Her fields of research include theoretical linguistics, the history of the Arabic language, Arabic linguistic thought, and ancient Arabic poetry. Her books include Prédicat, figures, catégories: La question de la phrase nominale en arabe littéraire (Lille, 1996). She has published widely on Sībawayhi’s Kitāb and on syntax and semantics in Arabic linguistic theory. Ramzi Baalbaki is the Margaret Weyerhaeuser Jewett Chair of Arabic at the American University of Beirut and the Head of the Academic Council of the Doha Historical Dictionary of the Arabic Language. He has published extensively on the Arabic grammatical theory and Arabic lexicography. His books include The legacy of the Kitāb: Sībawayhi’s analytical methods within the context of the Arabic grammatical theory (Leiden, 2008) and The Arabic lexicographical tradition from the 2nd/8th to the 12th/18th century (Leiden, 2014). Michael G. Carter after a D.Phil. (Oxon) taught at Sydney University (1968–1985), then Duke (1985–1986), New York University (1986–1996) and Oslo University (1996–2004) until retirement. His research interests are Sībawayhi and early Arabic grammatical theory, and the relationship between grammar, law and philosophy in Medieval Islam. His 1968 doctoral thesis has recently been published under the title Sībawayhi’s principles: Arabic grammar and law in early Islamic thought.

viii

notes on contributors

Hanadi Dayyeh (Ph.D., American University of Beirut), is a researcher in the field of Arabic historical linguistics. Her research concentrates on Sībawayhi’s linguistic theory and its impact on the evolution of the Arabic linguistic tradition. Her research in the field of Arabic linguistics also focuses on language acquisition and its implications for the teaching of Arabic. Her work experience in the field of teaching and researching teaching methods in Arabic language, both to native and non-native speakers, spans a period of more than fifteen years. Joseph Dichy born 1951 in Beirut, is Professor of Arabic Linguistics in Lyon (France). He is the author of a reference thesis on the writing system of Arabic (Lyon, 1990), of many works on Arabic descriptive and computational linguistics (DIINAR lexical db), and on Medieval Arabic rhetoric and argumentation. He has coordinated the DIINAR-MBC Euro-Mediterranean project (EU, DG XIII, 1999–2001). He is also a recognized expert in the teaching of Arabic to speakers of other languages (TASOL), and in translation studies involving Arabic, English and French. Jean N. Druel obtained a Master’s degree in teaching Arabic as a foreign language (American University in Cairo, 2006), and in 2012 he obtained his Ph.D. at the University of Nijmegen with a thesis on the Arabic grammarians’ theories about the syntax of numerals. He is a researcher in the history of Arabic grammar; since October 2014, he has been the director of IDEO (Dominican Institute for Oriental Studies) in Cairo. His current research focuses on the manuscript tradition of Sībawayhi’s Kitāb. Manuela E.B. Giolfo holds an M.A. in philosophy from Milan University, and a Ph.D. in Arabic linguistics from Aix-Marseille Université. She was lecturer in Arabic at Exeter University (2008–2013). In 2013 she moved to the University of Genoa, where she is lecturer in Arabic language and philology. From 2014 she is also chercheuse associée at IREMAM—CNRS—Aix-Marseille Université. She edited Arab and Arabic linguistics (Oxford, 2014) and, with Manuel Sartori and Philippe Cassuto, Approaches to the history and dialectology of Arabic in honor of Pierre Larcher (Leiden, 2016). She is the author of Les systèmes hypothétiques de l’ arabe classique (Rome, 2017).

notes on contributors

ix

Wilfrid Hodges FBA is Emeritus Professor of Mathematics at Queen Mary, University of London, specializing in mathematical logic and logical semantics. Since his retirement he has been working on Medieval Arabic logic, in particular that of Ibn Sīnā. Books on the logics of Ibn Sīnā and al-Fārābī are in preparation (one joint with Saloua Chatti). He also has a project with Manuela E.B. Giolfo to compare the views of al-Sīrāfī and Ibn Sīnā in areas where linguistics and logic overlap. Éva Jeremiás graduated in Iranian languages and Ancient philology (Latin and Greek) and completed her doctoral studies under the guidance of the late Professor Zsigmond Telegdi (General Linguistics and Iranian Studies). She is former founder and head of the Department of Iranian Studies, Eötvös Loránd University (Budapest), and currently professor emeritus and head of the Ph.D. program in Iranian Studies. Her main fields of research are New Iranian philology, Classical and Modern Persian language (descriptive and historical problems), history of grammar, history of linguistic ideas (European and Oriental traditions), Classical Persian literature: poetics, lexicography etc. Almog Kasher has a Ph.D. degree (2007) in Arabic; he is lecturer in Bar-Ilan University. His main field of study is the Medieval Arabic grammatical tradition, with the emphasis on its early history, Sībawayhi’s commentaries, and pedagogical grammars. Aryeh Levin was born in Israel in 1937. He is professor emeritus of Arabic at The Hebrew University of Jerusalem. He wrote his Ph.D. thesis on The ʾimāla in the Arabic dialects (The Hebrew University of Jerusalem, 1971). His main fields of research are: Arabic Medieval grammatical thought and terminology, history of the Arabic language, and modern Arabic dialects. He was the Head of the Department of Arabic Language and Literature, 1987–1992, and the Head of the Institute of Asian and African Studies of The Hebrew University of Jerusalem, 1992–1998. In 2010 he was awarded the prestigious “Israel Prize in General Linguistics” for his achievements in the field of Arabic linguistics. Arik Sadan holds a Ph.D. (2010) in Arabic language and literature from the Hebrew University of Jerusalem. His research fields are Arabic grammatical thought, Arab grammarians, Classical, Modern and Colloquial Arabic linguistics, manuscripts

x

notes on contributors

in Arabic grammar and other fields. He teaches at various academic institutions in these fields. Among his publications are A critical edition of the grammatical treatise Taḏkirat jawāmiʿ al-ʾadawāt by Muḥammad b. Aḥmad b. Maḥmūd (Wiesbaden, 2012); The subjunctive mood in Arabic grammatical thought (Leiden, 2012); and, together with Almog Kasher, A critical edition of the grammatical treatise Mīzān al-ʿarabiyya by Ibn al-ʾAnbārī (d. 577/1181) (Wiesbaden, 2018). Haruko Sakaedani is part-time lecturer in Arabic at Keio University, the University of Tokyo, Tokai University and Waseda University. She holds an M.A. in Teaching Arabic as a Foreign Language from the American University in Cairo and a Ph.D. in Arabic linguistics from Tokyo University of Foreign Studies. Manuel Sartori after graduating in Comparative Politics at the Institute of Political Studies (IEP, Aix-en-Provence, 1999) and in Arabic studies at Aix-Marseille Université (AMU, 2004), became senior teacher (professeur agrégé) in Arabic (2009) and completed a Ph.D. in Arabic language and linguistics at AMU (2012). First having been Lecturer at IEP and Researcher at IREMAM, he is now Professor at AMU. His research interests include Arabic grammar and linguistics (diachronic and synchronic, Medieval and contemporary) and the history of the Arabic language. Beata Sheyhatovitch holds a Ph.D. in Arabic and is Lecturer in Arabic and Islamic Studies at Tel Aviv University. Her main field of study is the Medieval Arabic grammatical tradition, with emphasis on its terminology and contacts with other Islamic disciplines. She is the author of The distinctive terminology in Šarḥ al-Kāfiya by Raḍī al-Dīn al-ʾAstarābāḏī (Leiden, 2018). Kees Versteegh is emeritus professor of Arabic and Islam at the University of Nijmegen (The Netherlands). He specializes in historical linguistics and the history of linguistics, focusing on processes of language change, language contact, and pidgin and creole languages. His books include The Arabic linguistic tradition (London, 1997), and The Arabic language (Edinburgh, 1997, revised ed. 2014). He was the editor-in-chief of the Encyclopedia of Arabic language and linguistics (Leiden, 2006–2009).

introduction

The Evolution of Theory in the Arabic Linguistic Tradition Manuela E.B. Giolfo and Kees Versteegh

The first three Conferences on the Foundations of Arabic Linguistics took place in 2010 (Cambridge), 2012 (Cambridge), and 2014 (Paris). The original aim of Amal Marogy, the organizer of the first two conferences, was to bring together a group of scholars whose research focused on the development of the Arabic grammatical and lexicographical tradition, in particular during the foundational period of Arabic linguistics in the first four centuries of Islam. After the upsurge of interest in the history of Arabic grammar during the 1980s and 1990s, the pace seems to have slackened somewhat during the second half of the 1990s. Thanks to the conferences, a new group of scholars has now been attracted to the field. The proceedings of these conferences have proved to be an important tool in promoting the study of the Arabic grammatical tradition. In 2012 the first volume, edited by Amal Marogy, appeared, in 2015 the second volume, edited by Amal Marogy and Kees Versteegh, and in 2018 the third volume, edited by Georgine Ayoub and Kees Versteegh. Originally, the conferences concentrated on the theories of the first grammarian of Arabic, Sībawayhi. Since then, the scope of the studies published in the proceedings has been expanded. In the first volume, seven out of eleven papers dealt with Sībawayhi directly. The second volume also contained eleven contributions, six of which dealt exclusively with Sibawayhi, while the remaining papers were concerned with the reception of the Kitāb. In the third volume, there were twelve contributions, only three of which mentioned the name of Sībawayhi in their title. Yet, in this third volume, too, most contributions dealt with grammarians who in some way were indebted to the legacy of the Kitāb, without however following him slavishly: in fact, they tended “to quote Sībawayhi as if he agreed with them, while developing their own analysis which contrasted with his” (Marogy and Versteegh 2015:4). The present volume contains sixteen papers presented at the 4th Conference on the Foundations of Arabic Linguistics, which was organized by Manuela Giolfo at the University of Genoa in September 2016. It brings together signifi-

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_002

2

giolfo and versteegh

cant contributions to the field, both by older and by younger scholars, all having to do with the development of the Arabic grammatical tradition.1 Two papers deal with Sībawayhi exclusively. Jean Druel’s study touches on what is arguably a central topic of the field, the famous Ambrosiana manuscript, which has long been hailed (see Humbert 1995) as one of the most interesting testimonies of the textual history of the Kitāb Sībawayhi, because it is not represented in the current editions. Druel has edited three chapters of the Kitāb on numerals and geminate verbs, using the Ambrosiana manuscript as his basis. He found a large number of differences between the manuscript and the text of the Kitāb as it was published in the Derenbourg and Bulaq editions. Not all of these are equally important, but in some cases the manuscript provides interesting, perhaps even better, readings than the published text. If we extrapolate this to the entire manuscript, it is clear that a complete edition of the Ambrosiana manuscript is bound to bring significant changes to our understanding of the text. The second paper in this category is of a more general nature. Hassina Aliane proposes a new reading of the underlying theoretical and methodological concepts of the Kitāb (and of the Arabic grammatical tradition in general), based on a comparison with modern linguistic models. She compares the Arabic grammarians’ method of distributional analysis with that of structuralist linguistics, as it was first developed by Zellig Harris in 1946, and then concludes that Mathematical Category Theory provides an even better model for the analysis of the system of hierarchically organized levels in Arabic grammatical theory. Although she does not focus specifically on Sībawayhi’s Kitāb, much of her analysis aims at a better understanding of the underlying theoretical presuppositions of its theories. An interesting contrast with Aliane’s paper is that by Ramzi Baalbaki on the development of pedagogical grammar, a topic that was also addressed by Kasher in the 3rd conference (Kasher 2018). Baalbaki draws the attention to the fact that, even in the Classical period, many students found the ‘official’ treatises on grammar too difficult to understand. This is why at a relatively early stage textbooks for beginners began to be composed. After presenting a gen1 In the present volume, we have followed more or less the same editorial guidelines as in the previous volumes. The transcription of Arabic follows the system of the Encyclopedia of Arabic Language and Linguistics (Leiden: E.J. Brill, 2006–2009), with one major difference, ḫ instead of x. Initial hamza is transcribed when it is morphological, but not when it is merely phonetic (thus: wa-ktub ‘and write!’, but wa-ʾaktib ‘and make write!’). Declensional and inflectional endings are represented fully in Qurʾānic and poetic quotations and in grammatical examples; in other quotations and book titles we have opted mostly for a simplified system, in which pausal rather than contextual forms are used. Yet, in some papers, authors preferred to use full representation throughout.

the evolution of theory in the arabic linguistic tradition

3

eral characterization of this genre and its relationship to the formal teaching of grammar, Baalbaki turns to one treatise in particular, Ibn Hišām’s (d. 761/1360) Muġnī l-labīb and analyzes its special approach to the didactics of teaching grammar. Just like Baalbaki’s paper, most of the other papers in the present volume deal with developments in linguistic theory after Sībawayhi’s Kitāb. Two papers deal with the reception and development of Sībawayhi’s grammatical theory in the first few centuries after him, especially by al-Mubarrad (d. 285/898), Ibn alSarrāj (d. 316/928), and al-Zajjājī (d. 337/949). Aryeh Levin addresses the issue of ittisāʿ al-kalām, which had been discussed by Hanadi Dayyeh at the 2nd Conference on the Foundations of Arabic Linguistics (Dayyeh 2015). Levin traces the development of this notion, used by the grammarians to explain phenomena of flexibility in speech, i.e. expressions that go beyond the ordinary constructions, and shows that it is fundamentally different from the grammarians’ taqdīr, because it refers to an action by the speakers. The speaker also has a central role in Hanadi Dayyeh’s paper on the notion of taqdīm wa-taʾḫīr. She discusses the way Sībawayhi treats the speaker as an arbiter of correct speech, who is free to choose between alternative word orders (even if one particular order is more frequent or normal). Later on, in al-Mubarrad’s and Ibn al-Sarrāj’s approach, the speaker is assigned by the grammarian the position of a learner being told which deviations in canonical word order are permitted and which are not. Thus, the speaker’s freedom has disappeared and made way for a more rule-dictated form of speech, in which it is no longer the speaker who makes the rules. Four papers go still further forward in time and involve the later tradition after the 4th/10th century in their discussion, in particular the views of grammarians such as Ibn Hišām and al-ʾAstarābāḏī (d. ca. 700/1300). Arik Sadan deals with the construction of the verbal noun (maṣdar) as ḥāl in constructions like ʾataytuhu mašyan ‘I came to him walking’. Interestingly, there are two different views on this construction. According to Sībawayhi, the verbal noun can be used when it is actually attested, in which case it is analogous to a participle (as in ʾataytuhu māšiyan). Otherwise, it is not permitted. Al-Mubarrad on the other hand states that by analogy the verbal noun may be used as the cognate object of an implied verb (as in ʾataytuhu*ʾamšī mašyan). Sadan follows this issue up to the 7th/14th century and finds that both views remained available as alternative explanations throughout the entire tradition. Haruko Sakaedani comes back to a topic that continues to fascinate researchers in this tradition, that of (in)definiteness. She deals with the shifts in the lists of ʾasmāʾ mubhama, including demonstratives, pronouns and proper names, throughout the tradition. Demonstratives were also dealt with by Arik Sadan in his contribution to the 3rd Conference on the Foundations of Ara-

4

giolfo and versteegh

bic Linguistics (Sadan 2018), and the topic of (in)definiteness is an important focus in Georgine Ayoub’s contribution to the present volume. In Sakaedani’s paper, the definiteness hierarchy set up by the grammarians is compared with the Givenness Hierarchy in modern cognitive linguistics as proposed by Gundel et al. (1993). Sakaedani concludes that the order in which the grammarians list the members of the mubham class is compatible with the order of linguistic entities in the Givenness Hierarchy. In the present volume Manuel Sartori continues his study of the terms that are used in the Arabic linguistic tradition for the categories of definiteness and determination. He regards these categories as distinct, even though they are often confused. In his contribution to the 3rd Conference on the Foundations of Arabic Linguistics, Sartori (2018) dealt primarily with taḫṣīṣ ‘particularization’. In the present contribution he focuses on the complementary term of taʿrīf. Underlying his contribution is the idea that even though the Arabic grammarians did not have specific terms to distiguish between definiteness and determination, they were aware of the difference between the two and developed their own terminology to deal with this. A related topic is that of Georgine Ayoub’s contribution on the analysis of pronouns by the Arabic grammarians. In her contribution to the 3rd Conference on the Foundations of Arabic Linguistic (Ayoub 2018) she had studied the relationship between case and reference in the case of proper names. In the present contribution she looks at another category of referring words, that of pronouns, as indicated by the three interconnected notions of ḍamīr, ʾiḍmār, and muḍmar. She points out that the Arabic terms for the three persons (mutakallim, muḫāṭab and ġāʾib) were praised by the French linguist Émile Benveniste (1946: 228) as an adequate reflection of the relation between the participants in a speech seting. In the papers mentioned thus far, while the chronological focus is on the later tradition, the grammarians whose work is reviewed more or less explicitly build on Sībawayhi’s ideas. In three other papers the focus is on terms that were never even used by Sībawayhi. In this sense the grammatical teachings investigated represent innovations: while the linguistic phenomena remained the same, their theoretical framework shifted. Almog Kasher deals with the notion of mafʿūl muṭlaq, the absolute object, for which Sībawayhi uses the term ism al-ḥadaṯ. Kasher traces the use of the term mafʿūl muṭlaq, which first appeared in Ibn al-Sarrāj’s ʾUṣūl al-naḥw, in the works of two later grammarians, Ibn Hišām and al-Jurjānī (d. 471/1078). He shows how they based themselves primarily on the literal sense of terms used for the phenomenon of the absolute object. According to Kasher, the modern Western term of ‘effective object’ may be helpful in disentangling the analysis of the various objects, as it is pursued by the Arabic grammarians. Focusing

the evolution of theory in the arabic linguistic tradition

5

on the literal sense of the technical terms involved, Ibn Hišām and al-Jurjānī managed to arrive at a new definition of the mafʿūl muṭlaq. Beata Sheyhatovitch’s paper, too, discusses a technical term, tawṭiʾa ‘preparation’ (also muwaṭṭiʾ ‘preparatory’) that does not occur in Sībawayhi. In fact, the term does not seem to appear until the 10th/4th century. Its function is not easy to delineate. Sheyhatovitch describes how it is used as a general tool for constructions or forms that prepare the way for another construction or form, noting that this demonstrates how speakers move around in the linguistic system, having at their disposal the complete structure of the Arabic language. Interestingly, with respect to the focus of the present volume, all constructions for which this concept is used by the later grammarians are derived ultimately from the Kitāb. In some cases, the use of the term is even framed as an act of adherence to the doctrine of Sībawayhi. The fact remains that this is a clear example of a term newly coined for an existing phenomenon in the grammatical tradition, in other words, we may justifiably speak here of a theoretical evolution. The third paper in this category is Michael Carter’s study of the term sallaṭa, which likewise is unknown in the Kitāb, but occurs, though infrequently, in later grammarians. In earlier publications (e.g. Carter 1989), Carter had stated his objections to the view that the term ʿamila fī (from which the term ʿāmil is derived) was used with the connotation of ‘government’ in syntax. According to him, ʿamila denotes a linear and horizontal relationship between constituents, and should be translated with ‘operating’. At first sight, the term sallaṭa might have just that connotation of ‘government’, but after a thorough analysis of the occurrences, Carter concludes that it cannot be interpreted in this sense and should be taken to mean ‘affecting’. Two papers deal with the place of grammar within the Islamic sciences and its connections with other disciplines. Joseph Dichy studies the science of Arabic lexicography, which is closely linked to Arabic grammar. He distinguishes between the heuristic approach to dictionary making, as in al-Ḫalīl’s attempt to present an exhaustive account of the Arabic lexicon, and a consultationoriented approach, which looked at dictionaries as tools for the user, who wished to have at their disposal a convenient method of looking things up. Through this comparison Dichy establishes a link between ordering principles and the function of dictionaries in the Arabic tradition. Manuela Giolfo and Wilfrid Hodges continue their comparison of the views of the grammarian al-Sīrāfī (d. 368/979) and the logician Ibn Sīnā (d. 428/1037) on issues that are of interest to both linguists and logicians. In their contribution to the 3rd Conference on the Foundations of Arabic Linguistics (Giolfo and Hodges 2018), they stated their reasons for choosing these two scholars as the primary authors to compare (though Sībawayhi, al-Ḫalīl and al-Fārābī all

6

giolfo and versteegh

play a role), pointing out that al-Sīrāfī was more directly interested in questions of logic than other grammarians, while Ibn Sīnā had a deeper insight in the relation between language and logic than other logicians. Their contribution to the present volume deals with conditionality. A sample question is: Why is it that al-Ḫalīl considers ʾin to be the canonical conditional particle, whereas Ibn Sīnā in his conditional logic consistently avoids ʾin and prefers kullamā? Al-Sīrāfī’s notion of the ‘verb of the condition’ ( fiʿl al-šarṭ), which has to be recognized even if it is only implicit, serves as a key to underlying assumptions about conditionals apparently shared by al-Ḫalīl and Ibn Sīnā. These underlying assumptions give a framework for clarifying al-Ḫalīl’s notion of mubham, al-Sīrāfī’s classification of conditional particles, and Ibn Sīnā’s quantification over times or events. (They also closely match the late 20th century LewisKratzer theory of conditionals.) Another kind of question is how we can tell that a statement carries an implied but unspoken antecedent; from different starting-points al-Sīrāfī and Ibn Sīnā reach a similar answer to this question. Finally, two papers go beyond the Arabic-speaking world and deal with the influence of the model of Arabic grammar in the description and analysis of other languages. The comparative study of grammatical descriptions used for the analysis of other languages than the ones for which they were designed, has recently been accepted as a research project entitled Grammaires étendues at the Université de Paris—Diderot. This project is coordinated by Émilie Aussant; the Arabic part of the project is coordinated by Jean-Patrick Guillaume. Two papers in the present volume fit right into this research question. Éva Jeremiás continues her study of how Arabic grammatical terms were adapted to the special requirements of the study of Persian. In earlier publications (e.g. Jeremiás 2000) she discussed terms like zāʾid and ʾaṣl, which acquired their own meaning in the Persian tradition. In the present contribution she focuses on two well-known Arabic terms, taqdīr and taḫfīf, and explains how these acquired a new use in the study of Persian compounds: taqdīr is used to explain the meaning of compounds from the underlying building elements, and taḫfīf ‘lightening’ explains the phonological processes that operate in the formation of compounds. Both terms retain some of their original meaning in Arabic grammar, but are adapted to the needs of the Persian language, in which compounds are very frequent. The second paper, by Kees Versteegh, deals with the short-lived impact of Arabic grammar at the end of the 19th century in descriptions of Malay. The study of Arabic was a central component of the curriculum in Southeast Asia, while Malay served as an auxiliary language in instruction, which did not need an analysis of its own. An exception was the work of Raja Ali Haji (probably d. 1873), who described Malay grammar and lexicon, using the only framework

the evolution of theory in the arabic linguistic tradition

7

known to him, that of Arabic grammar. He even went so far as to introduce the concept of ʾiʿrāb in his treatment of Malay, a language without case marking. The influence of this project did not last long: contemporary grammars of Malay/Indonesian work within a Western framework, with only a few Arabic technical terms. The papers in this volume show that, while Sībawayhi remains the solid basis for all studies on the Arabic grammatical tradition, there has been a marked shift in focus from the 1st to the 4th conference. The editors have chosen as subtitle for the present volume The evolution of theory, because this corresponds to the approach of the majority of the contributions that were presented in Genoa. These contributions illustrate the fact that, in the course of time, the teaching of the Kitāb became a venerable heritage, known and cited by all professionals, but not authoritative to the point of being followed blindly. The two key words in the title of this volume are ‘evolution’ and ‘theory’. Later grammarians express their respect and admiration for Sībawayhi’s work, but often go beyond his teachings and theory, if they feel their discipline has to evolve. The linguistic facts as reported in the Kitāb Sībawayhi are hardly ever contested, but the theoretical explanation of these facts is refined and deepened. After four conferences on the foundations of Arabic linguistics, it is time to reflect on the results. We feel that these conferences have contributed to at least four issues. In the first place, the discussion about a supposed dichotomy between Arab and Arabic linguistics has been hotly debated for some time, but may perhaps be put to rest now. While being aware that not everybody will agree with this, we firmly believe that a distinction should be made between ‘Arab’ as a marker of ethnicity, and ‘Arabic’ as a linguistic marker. Accordingly, we use ‘Arabic’ when the focus is on the language, and ‘Arab’ when ethnic identity is at issue. Anything related to the tradition of linguistic exploration, which has to do with the language rather than with any ethnicity, should therefore be referred to as Arabic linguistics. Arab linguistics could only mean linguistic studies by Arabs, which in the Classical period would be a rather meaningless notion since grammarians did not belong to a single ethnic group and could hardly be called Arabs collectively. In the contemporary period, one might use Arab linguistics to linguistic studies carried out by scholars from Arabicspeaking countries. The question is, however, what the advantage would be of creating a dichotomy between the study of the Arabic language in the Arab world and that in other countries. A second result has to do with the exploration of intuitions that are common to the Arabic tradition and modern/Western linguistics, or even beyond linguistics, if we follow Aliane’s paper. Comparisons between different linguistic frameworks have long been regarded as hazardous because of the risk of

8

giolfo and versteegh

presentism. But some comparisons, as Owens (1988) has shown, can be illuminating. In the contributions to the four conferences, interesting suggestions have been put forward concerning the intellectual commonalities between Arabic linguistic theories and modern theories in the fields of cognitive theory, categorial grammar, speech act theory, category theory, semantics, theories of referentiality, distributional analysis, and structuralist linguistics. Needless to say, there are many more aspects of linguistic research that could profit from a comparative analysis of the underlying postulates. Thirdly, at least some of the papers have shown that within the sciences in Islam there are close links between grammar and other disciplines. The links between grammar and other language-related disciplines, such as lexicography, exegesis, and the ʾuṣūl al-fiqh, are obvious, but there are also links with disciplines outside the domain of the core Islamic sciences, in particular with logic (ʿilmā l-ḥadd wa-l-istidlāl). The latter was presented as a systemic link in al-Sakkākī’s (d. 626/1229) ‘Key to the sciences [of the Arabic language]’ (Miftāḥ al-ʿulūm), but has remained largely unstudied. A final point is that of the central place of the Arabic language and its grammatical description within the Islamic world. The need to study Arabic grammar in order to understand the heritage of texts written in Arabic affected the linguistic perception of languages other than the Arabic language in ways that are still largely unexplored. In the proceedings that have appeared thus far Hebrew, Syriac, Persian, and Malay grammar were discussed as tributary to the Arabic tradition, but as the Grammaires étendues project has shown, there are still many more cases to be studied. We do not know yet what future conferences on the foundations of Arabic linguistics will bring, but we may be sure that the project of studying the foundations of Arabic linguistics that was initiated eight years ago by Amal Marogy will bring more results in, hopefully, unexpected ways. The first new development will be the 5th Conference on the Foundations of Arabic Linguistics, to be held at Cambridge in September 2018, hosted again by Amal Marogy. The theme announced for this conference is a focus on the challenges to the Kitāb’s status during the formative and Medieval periods of Arabic grammatical activities. The contributions to be presented in Cambridge will doubtlessly continue to highlight the many connections within and without the tradition. The editors wish to express their gratitude to the University of Genoa for its support during the organization of the conference that was at the basis of the present volume. They thank the editorial staff of Brill, in particular Maarten Frieswijk and Wilma de Weert, for their help in producing this volume. Genoa and Batenburg, June 2018

the evolution of theory in the arabic linguistic tradition

9

Bibliographical References Ayoub, Georgine. 2018. “Case and reference: The theory of mā yanṣarif wa-mā lā yanṣarif in Sībawayhi’s Kitāb”. Ayoub and Versteegh (2018:11–49). Ayoub, Georgine, and Kees Versteegh, eds. 2018. The foundations of Arabic linguistics. III. The development of a tradition: Continuity and change. Leiden: E.J. Brill. Benveniste, Émile. 1946. “La nature des pronoms”. (Repr., Problèmes de linguistique générale, I, 251–257. Paris: Gallimard, 1966.) Carter, Michael G. 1981. Arab linguistics: An introductory classical text with translation and notes. Amsterdam: J. Benjamins. Carter, Michael G. 1989. “The Arabic and Medieval Latin terms for ‘governing’ ”. Speculum historiographiae linguisticae, ed. by Klaus D. Dutz, 29–36. Münster: Nodus. Dayyeh, Hanadi. 2015. “Ittisāʿ in Sībawayhi’s Kitāb: A semantic ʿilla for disorders in meaning and form”. Marogy and Versteegh (2015:66–80). Giolfo, Manuela E.B. and Wilfrid Hodges. 2018. “Syntax, semantics, and pragmatics in al-Sīrāfī and Ibn Sīnā”. Ayoub and Versteegh (2018:115–145). Gundel, Jeanette K., Nancy Hedberg, and Ron Zacharski. 1993. “Cognitive status and the form of referring expressions in discourse”. Language 69.274–307. Harris, Zellig. 1946. “From morpheme to utterance”. Language 22:3.161–183. Humbert, Geneviève. 1995. Les voies de la transmission du Kitāb de Sībawayhi. Leiden: E.J. Brill. Jeremiás, Éva. 2000. “Arabic influence on Persian linguistics”. History of the language sciences, ed. by Sylvain Auroux, Konrad Koerner, Hans-Josef Niederehe and Kees Versteegh, I, 329–333. Berlin and New York: Mouton de Gruyter. Kasher, Almog. 2018. “Early pedagogical grammars of Arabic”. Ayoub and Versteegh (2018:146–166). Marogy, Amal Elesha, ed. 2012. The foundations of Arabic linguistics: Sībawayhi and early Arabic grammatical theory. Leiden: E.J. Brill. Marogy, Amal Elesha and Kees Versteegh, eds. 2015. The foundations of Arabic linguistics. II. Kitāb Sībawayhi: Interpretation and transmission. Leiden: E. Brill. Owens, Jonathan. 1988. The foundations of grammar: An introduction to Medieval Arabic grammatical theory. Amsterdam and Philadelphia: J. Benjamins. Sadan, Arik. 2018. “Demonstratives in Sībawayhi’s Kitāb”. Ayoub and Versteegh (2018: 178–189). Sartori, Manuel. 2018. “Origin and conceptual evolution of the term taḫṣīṣ in Arabic grammar”. Ayoub and Versteegh (2018:203–228).

Contribution to a Modern Reading of Sībawayhi Hassina Aliane

1

Introduction

The first objective of this work was to build an automatic syntactic analyser for Arabic ‘respecting specificities of Arabic language’. In fact, this latter statement usually appears in Arabic Natural Language Processing literature. However, whatever the task and the language, computer engineers work with the tools and models they have at their disposal. Regarding language modeling, it always rapidly turns out that no available computational model accounts for Arabic language. But isn’t this true for all natural languages? Indeed, computational existing systems work only for specific tasks or implement very little parts of linguistic theories. If it is true that modern linguistic theories are occidental, they are, nevertheless thought to deal with language as a human faculty or a human production in general even if linguists take as examples for explaining their artefacts some specific language examples. The grail for the linguist is to find the theory which embraces the whole language phenomenon and thus the grail for mathematicians and computer scientists is to make the theory computable. In this sense, considering what Arabic linguistics may bring to the general picture is worth undertaking. Hence, ‘considering the specificities of Arabic language’ leads to seeking insights from Arabic linguistics which inevitably leads to Sībawayhi and the Arabic grammatical tradition (AGT). The problem is that the Arabic grammatical tradition is not a theory: “… AGT has not made fully explicit its underlying theoretical premises. The existence of these premises cannot however be in doubt …” (Suleiman 1999:30). The present contribution is a reading of Sībawayhi’s approach which has been developed within the corpus of the Arabic grammatical tradition, and an attempt to explicit its theoretical premises. As linguistics (or any other science) often makes use of the language of mathematics and logic to build the language of its theories, it would not be surprising, when seeking again in these disciplines, to find some new artefacts since the old ones are not satisfying. Precisely, we propose modern Mathematical Category Theory (MCT) as a theoretical foundation for the Arabic grammatical tradition. Today, Mathematical Category Theory is the theory of structures par excellence: it is the theory of objects and their transformations, which is not only founded on paradigmatic classification, but it is the only one which makes

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_003

contribution to a modern reading of sībawayhi

11

a synthesis of class and order by abstracting universal properties of objects through their interactions in order to give objects positions in structures. This way of doing seems to correspond to the way Sībawayhi and his followers deal with language structures: every language item has a position in an abstract structure in Sībawayhi’s approach and the ultimate goal of the procedures of analysis is precisely to assign a position to linguistic objects in miṯāls. For the Arabic grammarians speech is a system in equilibrium, whether it is the result of a revelation from Allah, or of an agreement between men. Each and every letter, word, category, has its own place and its own rights (Versteegh 1978:261). The present paper describes an attempt to understand this system in order to propose formal foundations. We do not aim to build here the technical apparatus of the grammatical tradition by means of the tools of Mathematical Category Theory, which is a whole proposed research program, but we shall try to explain guiding intuitions of the proposal. Some of the terminology discussed is not found in Sībawayhi’s Kitāb, but was developed by later grammarians. We shall not distinguish between the terms introduced by Sībawayhi or by other grammarians from this tradition, since our purpose is to analyze the grammatical theory of the first grammarians before Arabic grammar was influenced by logic.

2

The Arabic Grammatical Tradition: the Heritage

When we look at the landscape of Arabic linguistics, Sībawayhi’s Kitāb rises like a star around which all (linguistic) constellations orbit. As pointed out by Yasir Suleiman (1999b:30) on his behalf and on behalf of other scholars, modern studies of the Arabic grammatical tradition have as one of their primary concerns the reconstruction of the methodological, theoretical and epistemological foundations that are thought to have informed the Arab grammarians in their treatment of the language. In one of his papers, Carter (1973:146) points out: “The book of Sībawayhi presents a type of structuralist analysis unknown to the West until the 20th century”. In the same paper, Carter remarks that “Sībawayhi’s method is remarkably similar to Immediate Constituent Analysis, with which it shares both common techniques and inadequacies”. The Arabic grammatical tradition constitutes for Arabic linguistics a—perhaps sometimes heavy and unbearable but very precious—unavoidable heritage. As Suleiman (1999b:31) says, contemporary researchers who tried to

12

aliane

understand or demystify the tradition fall into two groups: those who try to fit components of the tradition to some modern (Western) linguistic theory and those who take modern linguistics as “an interpretative instrument by means of which AGT can be interrogated” (Bohas et al. 1990). The present paper falls into the second category, maybe by expanding the interrogation of modern linguistics to modern science in general, since existing theoretical frameworks are unsatisfactory. Nevertheless, we consider Carter’s paper cited above to be foundational, as it highlights two very important intuitions regarding the Arabic grammatical tradition, the distributional nature of Sībawayhi’s analysis, and the fact that “Sībawayhi is treating language as a form of social behavior” (Carter 1973:146). The two points are linked, the latter marking the pecularity of the distributional approach in the Arabic grammatical tradition. Arabic linguists admit as a fact that the Arabic grammatical tradition is not universal, not only because its approach fails to be explained in the light of modern Western linguistic frameworks, but also because it is commonly considered to have been elaborated for the ultimate aim of preserving the Qurʾān and the religion of Islam. Arabs consider their language to be the best language since it is the one of the divine message. They divided languages into al-ʿArabiyya and ʾaʿjamī, which encompassed all other languages. Being data-oriented at inception, the Arabic grammatical tradition was, not unnaturally, more concerned with description than with theory-building. This orientation gained extra potency because of the non-universal character of the tradition, to which may be added its functionality as a tool in the pedagogic enterprise (Suleiman 1999b:30). Surprisingly however, the approach of the Arabic grammarians, on the contrary, is universal because it is the only linguistic approach that is founded on universal methodological principles instead of theoretical linguistic ones. In fact, all relies on how to understand de Saussure’s adagium: “The linguist must take the study of linguistic structures as his primary concern and relate all other manifestations of language to it” (Saussure 1983:20, quoted after Suleiman 1999b:33). Sībawayhi’s aim was not to develop a linguistic theory but to describe the naḥw or way of speaking of the Arab, and to do so, he makes use of intuitive methodological tools, such as reccurence, economy and pause. The most interesting connection made between the Arabic grammatical tradition and modern linguistics is, indeed, the one with Bloomfieldian linguistics, hence distributionalism (Hassan 1979; Carter 1973). The notion of mawḍiʿ in the analysis of the Arabic grammarians corresponds to the notion of distribution as a criterion for classification. The parallel with Harris’ Immediate Constituent Analysis has been investigated in particular by Carter (1973). Thus, the next sec-

contribution to a modern reading of sībawayhi

13

tion is dedicated to distributional analysis or, more precisely, its limits in order to show how Sībawayhi’s approach goes beyond those limits.

3

Contemporary Distributional Analysis

The distributional approach in contemporary linguistics was proposed by Leonard Bloomfield (1933), and then formalized and popularized by Zellig Harris (1951). This approach to language study is characterized by its dismissal of meaning and observation of distributions of linguistic elements as a means of language description. The distribution of an element is the set of the environments in which the element occurs. Distributional analysis techniques have been originally proposed by Harris for structural linguists to use as an aid in uncovering the structure of a language. The procedures Harris describes are intended to uncover regularities in the distributional relations among the features of speech. To use distributional analysis to determine empirically whether boy and girl are in the same word class, the linguist would need to determine whether the two words are licenced to occur in the same environments. The same principle is used to determine constituents and utterances (Harris, 1946). But Harris (1951:6) also says: “The particular methods described in this work are not essential. They’re offered as general procedures of distributional analysis applied to linguistic material”. According to Nevin (1993:373), Harris himself never claimed that his procedures allow to generate a grammar. He simply observes that “the parts of a language do not occur arbitrarily relative to each other: each element occurs in certain positions relative to certain other” (Harris 1954: 146). In Harris, two elements sharing the same distributions are considered to belong to the same class. This sounds like the quest for naẓāʾir in the Arabic grammatical tradition. Nevertheless, if the notion of distribution is central to linguistic analysis for the discovery of syntagmatic and paradigmatic relations between language units, this notion is also problematic. Indeed, formally, the method does not tell anything regarding what precisely a distribution is and what precisely is the size of an environment that will enable one to determine formal classes of language units and how to build and hierarchize these classes. In order to do so, a formal theory of language is needed, which the distributional method is not. We may also cite the practical impossibility of the method, which has been pointed out by Mahmoudian (1981): On constate qu’une analyse distributionnelle au sens strict du terme n’a jamais été effectuée pour une langue. Les applications que l’ on con-

14

aliane

naît sont des descriptions où, guidé par l’ intuition sémantique, le linguiste opère des segmentations et des classements; mais les arguments qu’il avance en faveur de ces opérations sont de nature distributionnelle. Or, les phénomènes distributionnels sont nombreux d’ une part et d’autrepart, ils ne sont pas tous pris en compte de façon systématique. Il s’en suit que dans l’ensemble des faits de distribution, il y en a qui étaieraient une description mais on trouve aussi qui iraient à l’ encontre de cette même description. L’analyse distributionnelle dans l’ acception stricte du terme c’est à dire sans critère sémantique est une utopie. This is a simple and relevant criticism against the distributional method, but Harris (1951:1) himself says: These procedures also do not constitute a necessary laboratory schedule in the sense that each procedure should be completed before the next is entered upon. In practice, linguists take unnumbered shortcuts and heuristic guesses and keep many problems about a particular language before them at the same time.

4

The Distributional Method of the Arabic Grammatical Tradition, Occam’s Razor

The Arabic grammatical tradition is called by its authors ʿilm al-ʿarabiyya, which means the science of the Arabic language, and which has been defined as “rules induced from the speech of Arab people” (maqāyīs mustanbaṭa min kalām al-ʿArab). This definition tells us that the building of Arabic grammar was corpus-based and linguists are unanimous about the fact that Arabic grammarians built the grammatical tradition on the basis of the observable corpus of Arabic fuṣḥā or of unaffected Arabic: the one of the Bedouin and the Qurʾān. Moreover, the establishment of the rules (maqāyīs) was the result of applying a sophisticated technical apparatus of analysis called qiyās, which is founded on the concepts of bāb, ʾaṣl, farʿ, naẓīr and mawḍiʿ. Actually, we have chosen to highlight these concepts as key concepts in Sībawayhi’s analysis and his followers in order to bring the reader to understand the outcomes of our analysis. Indeed, we are aware of the fact that besides this terminology, there are many other terms in Sībawayhi’s and his followers’ terminology that are of importance, but, as Baalbaki (1979:22) points out, key terms used by Sībawayhi are linked together as they play a role in the complete system of analysis:

contribution to a modern reading of sībawayhi

15

It is to Sībawayhi’s credit that if one sets out to examine one of the concepts of his grammatical analysis, one will eventually run across the other concepts, since there are several threads which tie all these concepts together to form one coherent system. A bāb roughly corresponds to the concept of class; ʾaṣl (pl. ʾuṣūl) is a polyvalent word: its first meaning is one of first in the sense of premise, hypothesis, something from which other things can be inferred or derived; the thing that is first with respect to others, thus also origin, prototype. Farʿ (pl. furūʿ) refers to the derived or inferred things; it also means a branch, an instance of a prototype. Naẓīr (pl. naẓāʾir) means equivalent, regarding some feature or comparison criterion, and mawdiʿ means position, place, distribution. The machinery of the Arabic grammatical tradition, which uses these tools, is called qiyās, which has been translated by ‘syllogism’ and sometimes by ‘analogy’. Qiyās is an approach of exploratory data analysis, which seeks to classify the elements of speech by observing their mawḍiʿ or position/distribution, and how they behave in the corpus, hence it looks for naẓāʾir or elements that behave the same way in order to abstract the miṯāl which describes or represents this behavior. In order to study language facts and establish ʾaḥkām and ḥudūd, Sībawayhi always begins with the minimum hypothesis or premises that best describe the reality: these are the ʾuṣūl. A minimum hypothesis is also the most economic one (ʾaḫaff ). When these ʾuṣūl recur in other, larger contexts or forms, he establishes the furūʿ and the bābs, using some other formal principles and regarding some linguistic point of view the grammarian is concerned with. Sībawayhi notably makes use of methodological principles in order to refine the classification, which are: economy, autonomy of realization, and recurrence. To synthesize things and to bring this idea closer to the reader, this goes in the spirit of the principle of Occam’s Razor, which is a principle of economy: the hypothesis (utterance) which is ʾaḫaff is always the one to be taken as ʾaṣl, whatever the problem at hand or the level of analysis. Language units are first determined on the basis of the autonomy of realization or the (possibility of) pause, waqf, sukūt; rules relating to the behavior of the language units (through their distributions) are established regarding recurrence of this behavior over the corpus. The Arabic grammarians use the principle of autonomy of realization at each level of language description. This point has probably received less attention than other concepts of the grammatical tradition, compared to its importance in Sībawayhi’s analysis. Studies of the notion of ‘linguistic pause’ (waqf ) in the Arabic grammatical tradition have been summarized by Al-Ani (2007:247). Nevertheless, we believe that the concept of pause being connected with the concepts of ibtidāʾ and infiṣāl is essential in formally establishing ʾuṣūl

16

aliane

and thus qiyās. This is true at each level of linguistic analysis: kitā is not actualizable, while kitāb or kitābu zaydin or jāʾa zaydun are. An actualizable language unit in this sense may be an element meaningful in the sens of fāʾida in itself or by participating via entering into relations with other elements to produce a larger unit which brings fāʾida. This is a formal and objective criterion to delimit linguistic structures and has the advantage of avoiding the problem of what accounts for an environment in modern distributional analysis. In fact, reading Sībawayhi’s Kitāb may be difficult at first glance for someone who is more familiar with classical logical classification. But it turns out that the classification of the Kitāb denotes in itself the principles of Sībawayhi’s approach. For Sībawayhi, each bāb has been conceived by means of miṯāl/ tamṯīl and naẓīr, and then, other terminology is linked to these principles, such as mawḍiʿ, jarā majrā, bi-manzilat, … What is interesting here is that Sībawayhi uses what was to become later the technical lexicon of his analysis, in a natural way without even defining the words. This supposes that these words are understood by his interlocutors and that the underlying kind of reasoning used to be natural in the everyday life of the Arabs. To come back to Sībawayhi’s formal analysis, he observes as something evident that the single noun (ism mufrad) is the most mutamakkin (Kitāb I, 20f.). After that, he always compares the behavior of language elements to the behavior of the ism, for instance to describe the verbs (Kitāb I, 20), or to describe language constructions “that exhibit the same behavior as the single noun” (allatī tajrī majrā l-ism al-wāḥid). Sībawayhi’s method of looking for naẓāʾir and setting up analogies has been investigated by contemporary scholars. Baalbaki (1979:18) formulates it as follows: “His belief in the existence of a norm, the force makes him regard certain forms, constructions, etc. being basic or central, while others become peripheral.” He adds: “One of the grammarians’ tasks, for him, seems to be to find an aspect of agreement or similarity between the two to justify the process of analogical extension (qiyās) and either to generalize it or to limit it to the exemplars”. Another important tool used by Sībawayhi in his process of abstraction by naẓāʾir and tamṯīl is taqdīr, which consists in positing an underlying level of an analyzed utterance. This concept has been investigated by Baalbaki (2007:3, 2004:7), Versteegh (1994:280), and Hadj Salah (1979). Qiyās consists in looking for language units and abstract miṯāls, which describe the structural behavior of language units. The process of abstracting miṯāls proceeds by putting utterances into correspondences with respect to some occurring kind of tašābuh, and decomposing the unit to be analyzed into its constituents, so that each miṯāl describing a given unit integrates (blends with) the miṯāls of its con-

contribution to a modern reading of sībawayhi

17

stituents in a compositional way. Besides similarity analysis at the form level of linguistic realizations, Sībawayhi also makes use of qiyās in studying all other language phenomena: he is always looking for correpondences in operations by relating one thing to another (ḥaml) in order to abstract the most economical miṯāl which allows the integration of facts that seem a priori heterogeneous, as in the example of taḥqīr and jamʿ taksīr below. This makes qiyās the only economical and unifying approach of linguistic analysis that abstracts models of language realization. The lexical meaning of the word qiyās is ‘measure’. The process of qiyās always begins with a classificatory operation that essentially consists in finding for every entity to be analyzed its bāb and then reconstructing it, using the miṯāls available for this bāb. The kind of abstraction performed with qiyās is called by Arabic grammarians iʿtibār, which concretely consists in testing how to relate an element to another via a bāb, thus assigning to each element a position in a miṯāl (Kitāb I, 104) fa-ʿtabir mā ʾaškala ʿalayka hāḏā bi-hāḏā, fa-ʿtabir hāḏā bi-hāḏā ṯumma jʿal tafsīra kulla wāḥidin jiʾta bihi tafsīra mā huwa miṯluhu; al-qiyāsu iʿtibāru šayʾin bi-šayʾin bi-jāmiʿin (Ibn al-ʾAnbārī, Lumaʿ 93). Hence, every entity under a bāb is either an ʾaṣl or a farʿ. A farʿ might be an ʾaṣl for other furūʿ. Hadj Salah (1979) has highlighted the power of qiyās as an abstraction process through the example of the realization of the diminutive form (taḥqīr) and the broken plural ( jamʿ taksīr). We see in these examples qiyās going beyond comparison of realizations to comparison of processes. According to Sībawayhi, the operations involved in taḥqīr, whose miṯāl is fuʿayl, are the same that are involved in jamʿ taksīr, as in mafāʿil. This means that these operations may be abstracted by muwāfaqa and munāsaba in the same operational miṯāl consisting in modification of the first segments, adjunction of an ḥarf madd at the third position, and adjunction of the ḥaraka i: fuʿayl and mafāʿil are thus instances of this operational miṯāl, which is also called jāmiʿ as it abstracts the two others. Another interesting example is that of the mabnī li-l-majhūl. The two realizations ḍaraba zaydun ʿamran ‘Zayd hit ʿAmr’ and ḍuriba ʿamrun ‘ʿAmr was hit’ have a relation of ʾisnād. The niyya of the speakers to show or hide the subject determines their choice of one of these miṯāls of realizations. In order to explain this, the Arabic grammarians posit, on the basis of the operations involved in the construction of each realization either from its basic immediate constituents or going from one to another, that there exists a virtual (muqaddar), abstract miṯāl, which may be the jāmiʿ for these two miṯāls of realizations. We have to note here that in qiyās all operations are reversible (radd al-šayʾ ʾila ʾaṣlihi), whatever the level of analysis we are engaging in. In fact, Sībawayhi’s underlying concern in his description and explication of language facts is to

18

aliane

render in a formal way the operations the speakers use on linguistic material to produce utterances fitting their niyya. Carter (2004:56–57) observes that: one of the most striking features of Sībawayhi’s analysis is that it concerns itself almost exclusively with language as behavior: speech is a set of actions, each named according to its intention, e.g. istifhām ‘asking a question’, taṯniya ‘making something dual’, tanbīh ‘drawing attention to something’ … Hence, the analysis of qiyās in its quest to unveil the miṯāls of language realizations is nothing else than unveiling the movements where the speaker puts available language structures. A simple and significant example is the example of ḥaraka, which is never a vowel, since the vowel is an integrated part of the Western word, while in Arabic the realization of a ḥaraka is always subject to the will of the speaker. In the example of passive and active voice constructions, each construction, active or passive, may be built from basic components or from one another. This is possible because the operations involved in such constructions are compositional. From a mathematical perspective, the group of operations involved in the construction of the active form from its basic constituents {ḍaraba, zayd, ʿamr} is isomorphic to the group of operations involved in the construction of the passive voice from these constituents, and also to the group of transformations involved in the construction of the passive voice from the active one. The existence of an underlying operational miṯāl which abstracts those groups of transformations is peculiar to qiyās and reveals an approach to language which is different from the deep/surface levels analysis in modern linguistics. Something that has probably not explicitly been highlighted regarding Sībawayhi’s grammatical analysis is the notion of ‘preservation’ or ‘transport’ of structure. Indeed, in the kind of abstraction described above, whatever the utterance and the criterion of correspondence at hand, the structure that has been recognized as an ʾaṣl is preserved in each miṯāl or scheme of transformations. This is true at each level of language analysis: the structure of ḥurūf is preserved when they form lexical items, the structure of the lexical items is preserved in the utterances they form, and the structure of processes is preserved when abstracted in a larger operational structure. To form an utterance, the miṯāls of ism and fiʿl are involved in a bināʾ: the individual structures of the constituents miṯāls are preserved and are just transported (without losing their properties) into the new utterance’s miṯāl, for instance when entering in an active or passive voice realization or in other processes like jamʿ taksīr and taḥqīr: not only the form of the structures is preserved but also

contribution to a modern reading of sībawayhi

19

all the possible operations they support. Just as naẓāʾir are generally viewed as equivalent classes in the mathematical sense, this concept of structure preservation and transport entailing equivalence, is also a mathematical concept. We conclude this section, whose aim was to summarize key elements in the qiyās approach, by citing Baalbaki (1999:86), who refers to this integration of heterogeneous facts of language in the Arabic grammatical tradition as follows: Coalescence or fusion in the sense of merging two linguistic elements, particularly two morphemes, features in several disparate parts of Sībawayh’s Kitāb. Although Sībawayh does not devote a special heading to the description of the rules governing coalescence or the nature of the resulting blend, he does not fail to see how the very concept of coalescence can be used as an extremely helpful tool of grammatical analysis. Rediscussing all concepts making up the sophisticated apparatus of Sībawayhi’s analysis lies beyond the scope of this paper, but we shall retain the following pre-eminent elements of his analysis to motivate our proposal: i. the quest for regularities through the concept of mawḍiʿ ii. the search for correspondences between elements through the concept of naẓīr: each time a similarity is found between two elements, they are put into correspondence in order to abstract some new miṯāls or just to confirm and explain exemplars; this similarity (tašābuh) may be in structure, in processes of modification of language elements, or in the process of building larger utterances iii. decomposition and radd al-šayʾ ʾilā ʾaṣlihi in order to explain miṯāls of realizations and thus to elaborate qiyās iv. the process of analysis results in the bābs Sībawayhi depicts in his book, which are nothing but the miṯāls abstracted from Arabic speech on the basis of some similarity which makes it possible to compare them. We can say that for Sībawayhi each utterance of attested language is a miṯāl from which he immediately seeks to find some other abstractions, resulting in new miṯāls: the present miṯāl being an exemplar for similar constructions or possibly entering in building some larger abstract ones.

5

The qiyās is Hierarchized

All elements in the Arabic grammatical tradition are defined and explained in an operational way and never in simple logical class membership. Carter (1973)

20

aliane

and Mosel (1980) point out the distributional nature of the Arabic grammarians’ analysis: Sībawayh seems to be aware of the notion of distribution and environment for he often says that a word, a phrase, …, is an ism (noun) because it occurs in the same position (manzila) as a common noun or any other parts of speech that have already been classified as ism (noun). Mosel 1980: 30

This, indeed, renders to a certain extent the distributional approach of the Arabic grammatical tradition. Nevertheless, unlike contemporary distributional analysis, qiyās with its economical set of concepts and principles succeeds in inducing all language (structure) behavior or maqāyīs al-luġa and can thus be seen not only as a theory of language, but as the only holistic one. The key words of a distributional analysis are regularities, similarity, and classification. Indeed, whether in Harris or in the Arabic grammatical tradition, distributional analysis seeks to unveil regularities or structural similarities in the corpus in order to classify similar elements, in distributional classes for Harris, and in bābs for the Arabic grammatical tradition. Nevetheless, Harris’ distributionalism fails to elaborate rules of language, because the notion of context is not formally defined, while qiyās relates language units via a bāb through the relation ʾaṣl ↔ farʿ, which is never arbitrary since it abstracts admissible operations (miṯāls), which permit going from one to the other. Indeed, the ultimate objective of qiyās as a distributional process is to assign to each language unit a position in a miṯāl. Each element under a bāb is either an ʾaṣl or a farʿ, while a farʿ may be an ʾaṣl for other furūʿ. Furthermore, the relation ʾaṣl ↔ farʿ determines equivalence classes in the mathematical sense and not only in the class membership sense. Thus, by assigning a position to language units in a miṯāl, which is a dynamic generative structure, qiyās is the only linguistic approach which combines syntagmatic and paradigmatic analysis. But, whenever we find regularities or similarities, this means there are invariants. It is only because there are invariants that we can tell about regularities: it is the observation of a similarity in behavior that always leads the linguist to find the jāmiʿ that abstracts the units sharing this similarity. Invariants may be in form like in words, lexis, syntactic structures, or in behavior. Hence, as we have seen, qiyās is hierarchized in: – correspondence of similar level structures: qāma, kataba, jalasa …; kitāb, alkitāb, bi-kitāb … – correspondence of different level structures: for instance the abstraction of the miṯāl describing the construction of a syntactic unit from lexical items:

contribution to a modern reading of sībawayhi

21

in such a miṯāl the structures of the lexical items are preserved, and their behavior in this new language unit determines their properties at this level – correspondence of (sets of) transformations: for instance in the examples of taḥqīr and jamʿ taksīr realizations. To understand qiyās, the idea is to consider invariants and their distributional contexts from an operational point of view. Instead of considering form invariants or their environments in themselves, we consider the transformations on these invariants. At the structural level, in modification in form or environments of an element, operations of adjunction or deletion may be considered rather than the resulting form itself. Obviously, such an enterprise cannot be undertaken without the help of mathematical tools. In fact, the relation between language and mathematics has been investigated early on. In the structuralist tradition, Harris himself proposed an operator grammar based on mathematical principles. Despite the fact that he pointed out some interesting breakpoints in the Arabic grammatical tradition, Hadj Salah (1979) proposed a syntax model on similar principles. His grammar differs from Harris in that he considered each operator as defining a structure model. On the other hand, it is clear that the foundational concepts of the Arabic grammatical tradition that we have described above are of a mathematical nature. For instance, qāma, kataba, jalasa have as miṯāl faʿala, which is an equivalence class in the mathematical sense; faʿila and faʿula are also miṯāl or equivalence classes for the bāb of ṯulāṯī. Furthermore, whatever the objects of analysis, the relation ʾaṣl ↔ farʿ always defines equivalence classes. In this spirit, and pointing out the importance of radd al-šayʾ ʾilā ʾaṣlihi, Hadj Salah also proposed to consider linguistic operations in the Arabic grammatical tradition at each level of analysis as a mathematical group structure à la Piaget. In his genetic epistemology, Piaget was the first to use mathematical structures, precisely for the analysis of group structure in human sciences. Louis Massignon (1954) was the first to point out the analogy between reversible operations on linguistic objects and reversible operations in the mathematical group structure which leaves an object invariant: À travers toutes ses variétés, la langue arabe maintient l’ usage de types de structures, de ‘groupes’ au sens où le ‘groupe’ signifie: une famille de modifications des termes, telles qu’elles maintiennent intacte, dans tous les cas, l’exactitude de leur disposition initiale. Nevertheless, group structure is still insufficient since it defines a closed static structure which does not allow inference and creation of new knowledge. The

22

aliane

coalescence process of qiyās, at any level it proceeds, always leads to the discovery and abstraction of new miṯāls, and thus to new knowledge. We have found out that Mathematical Category Theory is the best and only framework allowing us to understand qiyās as a holistic approach to linguistic analysis. It is amazing to see that the same intuition of considering operations rather than objects is found in the two approaches and that both make use of similar conceptual tools as well.

6

Category Theory

In modern mathematics, Category Theory studies mathematical structures. It was invented by Eilenberg and Mac Lane in the early 1940s. Today, it is considered to be the theory of structures par excellence. In Category Theory, objects are never considered in themselves, but only studied through their interactions with other objects. These relations are called morphisms or arrows. Category Theory rapidly witnessed great success and showed a great power in generalizing problems or moving from one problem to another. Its economical language and its unifying concepts led to its adoption in several scientifc domains. The abstractive method of Category Theory is founded on the concept of transformation, first between objects where transformations are called morphisms or arrows, but also between categories where transformations are called functors. The abstraction process is extended to functors where the arrows are called natural transformations. Properties of objects are uncovered only through these transformations. A category is a collection of objects and arrows which satisfies some elementary rules of composition between arrows (transformations, operations); an arrow has a domain or a source of the arrow and a co-domain or a target of the arrow. The rules ensure that the compound of two consecutive arrows always exists, that any object is related to itself by an identity arrow, and that composition of arrows is associative. We notice that this definition is more general than its Aristotelian equivalent definition of a category. Category Theory gives a minimalist definition of a category only in terms of objects and arrows. The notion of a category corresponds to any possible universe constituted of entities called objects. Constructions over objects and categories allow us to determine these objects through universal participation in a category, rather than through the notions of membership and inclusion as in set theory, which are used in classical classification and categorization methods in linguistics or elsewhere. According to Ehresmann and Vanbremeersch (2007:2), Category Theory provides tools to study concepts relevant to complexity, such as:

contribution to a modern reading of sībawayhi

23

– the binding problem: how do simple objects bind together to form a ‘whole that is greater than the sum of its parts’? – the emergence problem: how do the properties of a complex object relate to the properties of the more elementary objects that it binds? – the hierarchy problem: how may we explain the formation of increasingly complex objects, beginning with elementary particles? Like qiyās, Mathematical Category Theory proceeds by mapping stuctures in order to abstract which would be their common behavior. To do so, a basic tool of Mathematical Category Theory is the one of ‘co-limit’, which abstracts properties of morphisms to show how specific (elements, categories or functors) participate in the abstracted structure by sharing properties over the morphisms. Ellerman (2014:1) explains the notion of universal participation as follows: Given all the entities that have a certain property, there is one entity among them that exemplifies the property in an absolutely perfect and universal way. It is called the ‘concrete universal’. There is a relationship of ‘participation’ or ‘resemblance’ so that all the other entities that have the property ‘participate in’ or ‘resemble’ that perfect example, the ‘concrete universal’. And conversely, every entity that participates in or resembles the universal also has the property. The concrete universal represents the ‘essence’ of the property. We believe that this is the very idea of the abstraction process of Sībawayhi’s method, which is exemplified by the most famous of verbal miṯāls, f-ʿ-l mentioned above, which is at work in all the processes of qiyās. Category Theory may serve as a formal framework for understanding the Arabic grammatical tradition. Goldblatt (1984:25) gives a definition of the process of category construction which is surprisingly close to the process of construction of bāb in qiyās. Moreover, he even uses the word ‘measure’. The following excerpt where he designs Category Theory as a pathology of abstraction may help to understand that the intuition underlying the two approaches is the same: The process of identifying the notion of a category is one of the basic modi operandi of (pure mathematics) it is called abstraction. It begins with the recognition through experience and examination of specific situation that certain phenomena occur repeatedly, that there are a number of formal analogies in the behavior of different entities. Then comes the actual process of abstraction, wherein these common features are presented in

24

aliane

isolation […]. Having obtained our abstract concept, we then develop its general theory and seek for further instances of it. These instances are called examples of the concept or models of the axioms that define the concept. Any statement that belongs to the general theory of the concept will hold true in all models. The search for new models is a process of specialization, the reverse of abstraction. Progress in understanding comes as much from the recognition that a particular new structure is an instance of a more general phenomenon as from the recognition that several different structures have a common core. Our knowledge of (mathematical) reality advances through movement from the particular to the general and back again […]. These are propositions to the effect that any model of the axioms for a certain abstract structure must be (equivalent to) one of a particular list of concrete models. They ‘measure’ the extent to which the original motivating examples encompass the possible models of the general notion. We believe that the concept of bāb in the Arabic grammatical tradition finds its equivalent in the mathematical notion of a category. The objects of our bāb/category are then all the instances (miṯāls/models) of this structure: there are the miṯāls; the arrows are the operations wich allow to go from one miṯāl to another. More precisely, the miṯāl means rather the operations which lead to the models and loosely means also the resulting model or structure: “[…] the key lies, not in the particular nature of objects or arrows but in the way the arrows behave” (Goldblatt 1984:23). As for Category Theory, we have seen that qiyās is hierarchized into correspondence of similar forms, correspondence of structures, and correspondence of transformations. The use of the same economical set of conceptual tools to analyse and determine structures from heterogeneous point of views gives Category Theory and qiyās their power as a unifying framework for mathematical and linguistic analysis, respectively. We believe that this is the answer to Harris’ request: D’une façon générale ce que Z. Harris demande aujourd’hui aux mathématiques, c’ est une aide en vue de dégager les plus profondes de ces ‘régularités inévitables’ que présente le langage. Lentin 1992:16

We cannot end this reflexion on Category Theory, qiyās and structuralism without addressing the implications of those forms of structural analysis which are clearly different from the usual class/membership-based analysis. Indeed, Cat-

contribution to a modern reading of sībawayhi

25

egory Theory constitutes an important epistemological turn regarding structuralism. By stating that language is a system of structures, structuralism certainly aimed at studying the objects of language in their interrelationships, but it failed because it remained wihin the logic of essence and the relation of inclusion, while considering language to be a system of structures involved in a sterile membership/inclusion relation: Le structuralisme en mathématique comme en linguistique, s’ il propose d’organiser le corpus, ne dit rien sur les mécanismes d’ apprentissage et de création des savoirs. La science dont il rend compte est une science figée. Patras, 2005:403

Sībawayhi has succeeded where 20th century structuralism failed, because his approach considers language to be a system of structures in movement subordinate to the freedom of the speaker. Qiyās assigns to each linguistic object a mawḍiʿ in the language system through the miṯāl, which is an abstraction of the dynamic ʾaṣl ↔ farʿ. In such a perspective, the examples ʾakala, ḍaraba, ḏahaba, kataba do not belong to the class of the verb faʿala, but each of these units participates in a structure (miṯāl/bāb) it represents by the operations it may likely support by the action of the speaker. This is called in Category Theory ‘participation by universality’, and it is determined by the properties of the admissible interactions that relate an object to a category. The conceptual tool that correponds to jāmiʿ (coalescence) in Category Theory is co-limit, which allows integrating structures in higher abstract level ones. For instance, for language, a word is the co-limit of the (ordered) family of its letters; in the same way, a sentence is the co-limit of the (ordered) family of its words (Ehresmann 2007:97). Hence, the proposal of the present paper is to conceive of Category Theory as the formal foundations for the approach of the Arabic grammatical tradition, although it would be more accurate to regard Category Theory as a framework to express the approach of the Arabic grammatical tradition in a modern formal language. Indeed, the universality of the conceptual tool of qiyās makes it generalizable to other natural languages and thus, it may be seen as a foundations for linguistics.

7

Conclusion

The convergence between distributional analysis and Category Theory may seem natural if we consider that, just like the distributional method in Har-

26

aliane

ris and the Arabic grammatical tradition, Category Theory proceeds from the observation of regularities in the behavior of objects to build categories. The originality and power of Mathematical Category Theory lies in that it is the only known theoretical framework in which an object of study is not considered in itself, but through the possible operations it undergoes. The object structure is abstracted through the notion of arrow and we have seen that this is precisely what qiyās does by abstracting the structure of objects through the notion of miṯāl. We have also seen that both qiyās and Category Theory are founded on an economical set of conceptual universal tools and that they are hierarchized in a similar way. What was lacking in contemporary distributionalism to stand as a theory of language is to consider operations resulting in form changes, rather than considering the forms themselves (distributions and environments), which are just objects in different states. What is of importance is how the forms change, i.e. the transformations, rather than the new forms they become by means of transformations. Moreover, modeling linguistic structures in terms of objects and morphisms may bring new insights to linguistics, as it has done in other scientifc domains, such as biology or physics. Indeed, the interesting thing which would constitute an epistemological turn in linguistic structuralism is studying linguistic objects through universal properties: to determine objects through their participation in a linguistic structure, and not by their membership in that structure. We appropriate this theory for linguistics by the following citation about mathematics (Resnik 1981:53): In mathematics, I claim, we do not have objects with an internal composition arranged in structures, we have only structures. The objects of mathematics, that is, the entities which our mathematical constants and quantifiers denote are structureless points or positions in structures. As positions in structures, they have no identity outside of a structure. Qiyās is a mathematical method which succeeds in describing language in a holistic approach, and it is the only method to do this for langage. The strength of qiyās, we believe, lies in the fact that it is founded on the use of intuitive, natural and universal principles, rather than logical principles, in order to discover patterns in the language corpus: economy, autonomy of realization, and recurrence in the corpus. In such a perspective, all language miṯāls are just available structures to the speakers who arrange language stuff at their convenience or according to their niyya, and one can finally say that what Sībawayhi formalized was the way Arabic speakers manipulate the code of their language. This is very important since

contribution to a modern reading of sībawayhi

27

in such a vision objects are not considered as members of a structure, but as participating in a structure by means of which they share some of the properties of the structure. Secondly, and this principle is a consequence of the first one, objects participate in structures rather than being members in a class inclusion relationship. For instance, f-ʿ-l is the abstract scheme (class) for triradical (ṯulāṯī) words in Arabic, in an operational view focusing on operations (here ḥarakāt), thus, kataba will not be viewed as a member of the class f -ʿ-l, but as participating in the structure of f-ʿ-l, which may have as instances all the possible words in the combination of the three radicals. The epistemological consequence of such a view is that language structures are just available for the speakers who make them in movements/relationships in order to realize the structures serving their (communicative) intentions. In fact, as we have seen above, the idea of considering language objects as participating in structures rather than considering them as members of linguistic classes is not a new one. Harris himself alluded to such an idea (Lentin 1990:89). What we suggest here is that such ideas were explicit in Sībawayhi’s approach and that of the Arabic grammatical tradition regarding other linguistic approaches over time and that Mathematical Category Theory may be the adequate framework to express the formal premises of Arabic grammatical tradition.

Bibliographical References A

Primary Sources

Sībawayhi, Kitāb = ʾAbū Bišr ʿAmr ibn ʿUṯmān Sībawayhi, al-Kitāb. Ed. by Muḥammad ʿAbd al- Salām Hārūn. 5 vols. Cairo: Maktabat al-Ḫānjī, 1988. Ibn al-ʾAnbārī, Lumaʿ = ʾAbū l-Barakāt ʿAbd al-Raḥmān ibn Muḥammad Ibn al-ʾAnbārī, Kitāb lumaʿ al-ʾadilla fī ʾuṣūl al-naḥw. Ed by Saʿīd al-ʾAfġānī. Dār al-Fikr, 1957.

B

Secondary Sources

Al-Ani, Salman H. 2007. “The linguistic analysis and rules of pause in Arabic”. Approaches to Arabic linguistics, ed. by Everhard Ditters and Harald Motzki, 247–254. Leiden: E.J. Brill. Baalbaki, Ramzi. 1979. “Some aspects of harmony and hierarchy in Sībawayhi’s grammatical analysis”. Zeitschrift für arabische Linguistik 2.7–22. Baalbaki, Ramzi. 1999. “Coalescence as a grammatical tool in Sībawayhi’s Kitāb”. Arabic grammar and linguistics, ed. by Yasir Suleiman, 86–106. Richmond: Curzon Press. Baalbaki, Ramzi. 2007. “Inside the speaker’s mind: Speaker’s awareness as arbiter of

28

aliane

usage in Arab grammatical theory”. Approaches to Arabic linguistics, ed. by Everhard Ditters and Harald Motzki, 3–23. Leiden: E.J. Brill. Benoist, Jocelyn. 2007. “Mettre les structures en mouvements: La phénoménologie et la dynamique de l’intuition conceptuelle, sur la pertinence phénoménologique de la théorie des catégories”. Rediscovering phenomenology: Phenomenological essays on mathematical beings, physical reality, perception and consciousness, ed. by Lucian0 Boi, Pierre Kerszberg, and Frédéric Patras, 339–355. Dordrecht: Springer. Bloomfield, Leonard. 1933. Language. New York: Henry Holt. Bohas, Georges, Jean-Patrick Guillaume and Djamel Eddine Kouloughli. 1990. The Arabic linguistic tradition. London and New York: Routledge. Carter, Michael G. 1973. “An Arab grammarian of the eighth century: A contribution to the history of linguistics”. Journal of the American Oriental Society 93:2.146–157. Carter, Michael G. 2004. Sibawayhi. Oxford: Oxford University Press. Chomsky, Noam. 1969. La linguistique cartésienne. French transl. by Nelcya Delanoë and Dan Sperber. Paris: Editions du Seuil. Ehresmann, Andrée C. and Jean-Paul Vanbremeersch. 2007. Memory Evolutive Systems: Hierarchy, emergence, cognition. (= Studies in Multidisciplinarity, IV.) Amsterdam: Elsevier Science. Eilenberg, Samuel and Saunders Mac Lane. 1942. “Natural isomorphisms in group theory”. Proceedings of the National Academy of Sciences 28:12.537–543. Eilenberg, Samuel and Saunders Mac Lane. 1945. “General theory of natural equivalences”. Transactions of the American Mathematical Society 58:2.231–294. Ellerman, David. 2014. “On concrete universals: A modern treatment using Category Theory”. Available at: SSRN: https://ssrn.com/abstract=2435439 or http://dx.doi.org/ 10.2139/ssrn.2435439 Goldblatt, Robert. 2006. Topoi, the categorial analysis of logic. 2nd rev. ed. Mineola, NY: Dover Publications. Hadj Salah, Abderahmane. 1979. Linguistique arabe et linguistique générale. Thèse de doctorat, Université de la Sorbonne. Harris, Zellig. 1946. “From morpheme to utterance”. Language 22:3.161–183. Harris, Zellig. 1951. Structural linguistics. Chicago: The University of Chicago Press. Harris, Zellig. 1954. “Distributional structure”. Word 10:2–3.146–162. Lentin, André. 1990. “Quelques réflexions sur les références mathématiques dans l’ oeuvre de Zellig Harris”. Langages 99.85–91. Lentin, André. 1992. “Mots”. Intellectica 13–14.325–336. Mac Lane, Saunders. 1998. Categories for the working mathematician. 2nd ed. Berlin: Springer-Verlag. Mahmoudian, Mortéza. 1981. La linguistique. Paris: Seghers. Massignon, Louis. 1954. “Réflexions sur la structure primitive de l’ analyse grammaticale en arabe”. Arabica 1:1.3–16.

contribution to a modern reading of sībawayhi

29

Mosel, Ulrike. 1980. “Syntactic categories in Sībawayhi’s Kitāb”. Histoire, Epistémologie, Langage 2:1.27–37. Nevin, Bruce E. 1993. “A minimalist program for linguistics: A perspective on the work of Zellig Harris”. Historiographia Linguistica 20:2–3.355–398. Patras, Frédéric. 2003. “L’horizon sémantique et catégorial de la méthode axiomatique”. Formes et crises de la rationalité au XXième siècle. II. Epistémologie, ed. by Jean-Luc Gautero and Antoine Miquel, 9–29. (= Noesis, no. 5.). Paris, Vrin. Patras, Frédéric. 2005. “Phénoménologie et théorie des catégories”. Geometries of nature, living systems and human cognition, ed. by Luciano Boi. World Scientific 401–419. Piaget, Jean. 1938. “La réversibilité des opérations et l’ importance de la notion de ‘groupe’ pour la psychologie de la pensée”. Rapports et compte-rendus du XIe Congrès International de Psychologie, ed. by Henri Piéron and Ignace Meyerson, 433–435. Agen: Imprimerie moderne. Piaget, Jean. 1970. Genetic epistemology. Transl. by Eleanor Duckworth. New York: The Norton Library. Resnik, Michael D. 1981. “Mathematics as a science of patterns: Ontology and reference”. Noûs 15.529–550. Suleiman, Yasir (ed.). 1999a. Arabic grammar and linguistics. Richmond: Curzon Press. Suleiman, Yasir. 1999b. “Autonomy versus non-autonomy in the Arabic grammatical tradition”. Arabic grammar and linguistics, ed. by Yasir Suleiman, 30–49. Richmond: Curzon Press. Versteegh, Kees. 1978. “The Arabic terminology of syntactic position”. Arabica 25.261– 281. Versteegh, Kees. 1997. The Arabic language. New York: Columbia University Press.

Pronouns in Sībawayhi’s Kitāb and Related Concepts: ḍamīr, ʾiḍmār, muḍmar Georgine Ayoub

1

Introduction

As early as 1946, the denominations of personal pronouns by the Arab grammatical tradition as mutakallim ‘he who speaks’, muḫāṭab ‘he who is spoken to’, ġāʾib ‘he who is absent’, were praised by a great linguist, Emile Benveniste, in his major essay about “The structure of relations of persons in the verb”. The reason why he praised these denominations was that in them “is involved a just notion of the relation between persons”. And, relating subjectivity to persons in language, Benveniste stressed in his “Subjectivity in language” the role of “the linguistic status of person” in founding self-consciousness, as “ ‘Ego’ is he who says ‘ego’” (Benveniste 1966:I, 224). He pointed out that “it is only in and through language that man constitutes himself as a subject”, i.e. as a “psychic unity that transcends the actual experiences it assembles and that makes the permanence of the consciousness”. In this context Benveniste (1966:I, 228) also pointed out that the Arabic tradition assigns indeed a different status to pronouns than the Indian and Greek traditions and, later on, the European grammars that inherited it according to their role in the speech situation, i.e. according to a pragmatic criterion: the first and second person are speech participants, while the third one is ‘absent’ in enunciation. Benveniste, too, distinguishes between pronouns according to their discourse function: the first and second person are speech act participants, the third one “is not a person”. It is an unspecified actor or an abbreviation of a noun that precedes, i.e. an anaphora. ‘Personal pronouns’, or pronouns referring to a person are only the first and the second ones; they constitute a class of very peculiar words. They “escape the status of all the other signs of language”, as the reality to which they refer is “the reality of the discourse”.1 For instance “I refers to the act of individual discourse where it is pronounced, and by this, it designates the speaker […]”.2

1 Benveniste (1966: I, 252). 2 Benveniste (1966: I, 252).

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_004

pronouns in sībawayhi’s kitāb and related concepts

31

Although many recent studies have examined some aspects of the analysis of pronouns in the Arab grammatical tradition, the study of persons involved therein has not attracted the attention of researchers so much. The present paper, devoted to pronouns in Sībawayhi’s Kitāb, intends to explore the notions involved, as well as some aspects of the multiple links they have with other concepts and the analyses to which they have given rise. In the last part of the paper, we will try to sketch something of their subsequent history.

2

Mutakallim, muḫāṭab, and ġāʾib in the Kitāb and Some of Their Functions

The notions of mutakallim and muḫāṭab, understood as the two terms of the speech situation, are central in the theory of language developed in Sībawayhi’s Kitāb. Their presence is constitutive of every utterance. The Kitāb makes this point strongly in the chapter on the vocative (al-nidāʾ): As the beginning of every enunciation is always the vocative, unless you can do without it because the addressee is approaching you. Thus, the vocative constitutes the beginning of every utterance you enunciate. By it, you draw the attention of the one you address (li-ʾanna ʾawwala l-kalāmi ʾabadan al-nidāʾu, ʾillā ʾan tadaʿa-hu stiġnāʾan bi-ʾiqbāli l-muḫāṭabi ʿalayka, fa-huwa ʾawwalu kulli kalāmin la-ka, bi-hi taʿṭifu l-mukallama ʿalayka). Kitāb I, 274

Every utterance, note the ʾabadan, begins by a vocative. In other words, it postulates an addressee, the one spoken to (al-muḫāṭab, al-mukallam), with whom you establish a connection (taʿṭifu l-mukallama ʿalay-ka). This first statement is very close to Benveniste’s statement (1976:II, 82) about enunciation: “Any utterance is explicitly or implicitly an address. It postulates an addressee” (Toute énonciation est, explicite ou implicite, une allocution. Elle postule un allocutaire). Yet, the two statements are different with respect to their place in grammar. In both Sībawayhi’s Kitāb and in modern studies about enunciation, the necessary addressee has empirical traces in the utterance. The specificity of the analysis in Sībawayhi’s Kitāb, is that the relation between the speaker and the addressee determines in many ways the structure of utterances, i.e. their form and their syntax: the study of the notion of wājib and of the complex notions used throughout the Kitāb to describe the verb, such as maḍā, inqaṭaʿa, waqaʿa,

32

ayoub

illustrates this perfectly.3 It is the communicative intention of the speaker towards his addressee, whether he is informing him or commanding him or questioning him—in other words, the modality of the enunciation—which determines in the Kitāb the status wājib/ġayr wājib of the utterance, and consequently, the word order of the sentence and the case of the initial noun in some contexts. In other terms, as we have argued elsewhere,4 the notions of mutakallim and muḫāṭab are central not only as pragmatic notions, but as articulating enunciation to the form of utterances and their syntax. The notion of ġāʾib is also firmly present in Sībawayhi’s Kitāb in relation to the speech situation, precisely as absent to the speech situation. It is mentioned for the first time when Sībawayhi evokes two major contexts of the verb, which constitute two major discourse functions, ‘to speak with an addressee’ (al-muḫāṭaba) and ‘to inform/to convey information about an absent person’ (al-iḫbār ʿan ġāʾib), as in the following quotation from chapter 20 of the Kitāb (I, 22), in which mā and lāta are distinguished from laysa: “… since it [lāta] is not like laysa with respect to speaking with an addressee and informing about an absent” (li-ʾanna-hā [lāta] laysa ka-laysa fī l-muḫāṭabati wa-fī l-ʾiḫbāri ʿan ġāʾib). Two major properties seem to characterize the ġāʾib. First, as far as we know, this notion is always mentioned while speaking of the third person pronoun. Secondly, it is defined implicitly with another major discourse function, predication. The term al-ġāʾib is always associated with a performative notion designating the logical subject of the utterance, ‘the one you are talking about’ (al-muḥaddaṯ ʿanhu), e.g. ‘the concealed one you are talking about and who is absent’ (al-muḍmar al-muḥaddaṯ ʿanhu l-ġāʾib; Kitāb I, 337) and, in another context, ‘you are talking about an absent’ (tuḥaddiṯu ʿan ġāʾib). Referring to al-ġāʾib, i.e. the pronoun of the third person, Sībawayhi says al-muḍmar al-muḥaddaṯ ʿanhu (Kitāb I, 330). The notion of muḥaddaṯ ʿanhu and that of muḥaddaṯ bihi, which forms a pair with it, are performative notions designating the two terms of predication. Sometimes they replace the more dynamic and semantic terms mubtadaʾ and fāʿil in the analysis of nouns, and fiʿl and ḥadaṯ in the analysis of verbs, respectively, which are much more common in the Kitāb.5 In the example ḏahaba zaydun ‘Zayd has gone’, the noun zayd is al-muḥaddaṯ ʿanhu, while al-ḏahāb ‘the fact of leaving’ is al-muḥaddaṯ bihi ʿan al-ʾasmāʾ. Note that al-muḥaddaṯ ʿanhu could be present in the speech situation, for instance as the addressee or the speaker. In the following quotation, al-muḥaddaṯ ʿanhu is the addressee: 3 Cf. Ayoub (1991, 2010, 2015). For the notion of wājib see also Carter (2006). 4 Cf. the references cited above, n. 3. 5 See Ayoub (2010: 11).

pronouns in sībawayhi’s kitāb and related concepts

33

You only want to inform about the state in which is the one you speak about in the situation of speech, so you say: ‘You are now like this’ (ʾinnamā turīdu ʾan tuḫbira bi-l-ḥāli llatī fīhā l-muḥaddaṯu ʿanhu fī ḥāli ḥadīṯika fa-qulta ʾanta l-ʾāna ka-ḏālika). Kitāb I, 128

There is a crucial difference between al-mutakallim and al-muḫāṭab on the one hand, and al-ġāʾib on the other: the ġāʾib is only described as the logical subject of the utterance, the one who does not speak but about whom others are speaking, while the mutakallim or the muḫāṭab are those who make the utterance and determine partly its syntax, in addition to possibly being the logical subjects of the utterance. In other words, the notion of ġāʾib has none of the properties and functions of the mutakallim and the muḫāṭab mentioned above, namely a role in the form and the syntax of utterances. Thus, there is an asymmetry between the notions of mutakallim and muḫāṭab on the one hand, and ġāʾib, on the other, with respect to the speech situation and their function in language.

3

The Analysis of Personal Pronouns: First Remarks

What about these notions in the analysis of pronouns? Do we find the same asymmetry in their analysis? What about the asymmetry between the dialogue pronouns, i.e. the first and the second person, which are essential for the appropriation of the language by the speaker by their special referential properties, as we have seen above,6 and the third person which is a ‘non-person’? How does Sībawayhi approach the analysis of pronouns? 3.1 The Organization of the Chapters Eighteen chapters of the Kitāb are devoted to the analysis of personal pronouns (chapters 203 to 221 in the Derenbourg edition). The approach to pronouns in these chapters is pragmatic as well as morpho-syntactic, two dimensions that are found articulated throughout the Kitāb and that define its approach. Their aim is to present systematically the whole pronominal system, to identify all the members of the class, determining what is and what is not a pronoun (ʿalāmat muḍmar), in the genitive, accusative and nominative case, thus presenting all strong and bound, i.e. separate and attached pronouns (muttaṣil/munfaṣil).

6 See also Jakobson (1957).

34

ayoub

The question of case is central in Sībawayhi’s approach. The eighteen chapters are organized according to two important criteria, as shown by the designation of the pronoun and by the titles of the chapters. The first criterion is a pragmatic one, which gives the pronouns their names (al-muḍmar al-mutakallim; al-muḍmar al-muḫāṭab; al-muḍmar al-ġāʾib). The second one is the ‘case’ of the pronoun, i.e. the morphology of the pronoun according to its function. This leads to the following rough organization: – ch. 203: a general chapter announcing all the subsequent chapters (hāḏā bābu majrā ʿalāmāti l-muḍmarīn) – ch. 204–205: the nominative pronouns (hāḏā bābu ʿalāmāti l-muḍmarīn almarfūʿīn) – ch. 206–209: the accusative pronouns (hāḏā bābu ʿalāmati l-muḍmarīn almanṣūbīn) – ch. 210: the genitive pronouns (hāḏā bābu ʿalāmati ʾiḍmāri l-majrūrin) The other chapters deal with: – ch. 211: double cliticization of the accusative pronoun and hierarchy – ch. 212: reflexive pronouns – ch. 213–215: morphological issues (nūn al-wiqāya, the marker li- that becomes la- with a bound pronoun) – ch. 216: coordination between a lexical element and a pronominal one. – ch. 217: markers that cannot accept a bound pronoun – ch. 218–221: semantico-syntactic usages of the pronoun (ṣifa—badal— faṣl—mubtadaʾ) For the question addressed in the present paper, three major observations seem to be essential for understanding Sībawayhi’s analysis: i. The first one concerns the designation of pronouns and the network of concepts to which they are linked. It appears that the concepts used to designate pronouns, in their entirety, are not specific for them, but are also used to designate other grammatical phenomena. Pronouns have no real specific and stable name in the Kitāb. ii. The second observation is that, throughout the Kitāb, it is stressed that pronouns constitute a homogeneous class. The definition of this class is syntactic, referential, and pragmatic, and involves the two terms of the speech situation, the speaker and the addressee. iii. Although the pronouns of the first and second person are not explicitly distinguished with respect to enunciation and to their peculiar way of referring according to modern linguistics (as the reality to which they refer is ‘the reality of the discourse’), a notion of precedence and hierarchy is established between pronouns with respect to enunciation.

pronouns in sībawayhi’s kitāb and related concepts

35

3.2 The Designation of Pronouns All designations of the personal pronoun in the Kitāb are constructed on the root ḍ-m-r, which, according to Ibn Fāris’ Maqāyīs al-luġa, designates two different notions: what is thin, and what is hidden or concealed.7 It is this second notion from which the designation of the personal pronoun derives. 3.2.1 Ḍamīr In the Kitāb, we only find six occurrences of ḍamīr, the common metalinguistic term for the personal pronoun in late Classical grammatical treatises. The term is translated by Troupeau (1976:132) as “dissimulé”, “implicite”. In fact, five out of the six occurrences unequivocally designate personal pronouns. We present here two of them. They occur in chapter 23, which discusses what will later be known as ištiġāl: If you say ḍarabūnī wa-ḍarabtuhum qawmaka you put qawmaka in apposition to hum, because the verb must necessarily have a subject. The subject8 here is a plural, and the ḍamīr of the plural is wāw (wa-ʾiḏā qulta: ḍarabūnī wa-ḍarabtuhum qawmaka, jaʿalta qawmaka badalan min hum li-ʾanna l-fiʿla lā budda lahu min fāʿilin, wa-l-fāʿilu hāhunā jamāʿatun waḍamīru l-jamāʿati l-wāwu). Kitāb I, 30

And in the same way you say ḍarabūnī wa-ḍarabtu qawmaka, if you make the second verb operate on the noun that follows it as its object. The first one must then necessarily include a ḍamīr of the subject, for the verb cannot lack a subject (wa-kaḏālika taqūlu ḍarabūnī wa-ḍarabtu qawmaka, ʾiḏā ʾaʿmalta l-ʾāḫira fa-lā budda fī l-ʾawwali min ḍamīri l-fāʿili li-ʾallā yaḫluwa min fāʿilin). Kitāb I, 30

Only one of the six occurrences of ḍaṃīṛ does not indicate the pronoun and refers to the mind of the speaker. Sībawayhi analyses a verse in which the poet, praising himself, describes his journey and the journey of his companions as

7 ʾaḥaduhumā yadullu ʿalā diqqatin fī l-šayʾi wa-l-ʾāḫaru yadullu ʿalā ġaybatin wa-tasatturin (Ibn Fāris, Maqāyīs s.v. ḍamara http://www.baheth.info). 8 Sībawayhi refers, of course, to the subject of the first verb, the second verb being in the first person.

36

ayoub

very long and speedy: “He made wajīfa l-maṭāyā9 a tawkīd of ʾawjaftu,10 which is in his ḍamīr” ( fa-jaʿala wajīfa l-maṭāyā tawkīdan li-ʾawjaftu llaḏī huwa fī ḍamīrihi; Kitāb I, 161). 3.2.2 Muḍmar Apart from the rare occurrences of ḍamīr, the personal pronoun in the Kitāb is commonly designated by the term muḍmar, which occurs opposed to muẓhar: The verb cannot lack an implicit (muḍmar) or explicit (muẓhar) noun in the nominative [which functions as its subject]/ The verb necessarily has a subject, lexical or pronominal, marked for the nominative11 (lā yaḫlū lfiʿlu min muḍmarin ʾaw muẓharin marfūʿin min al-ʾasmāʾ). Kitāb I, 31

3.2.3 ʾIḍmār The personal pronoun is also designated by the term ʾiḍmār as in the following example, which lists all clitic and strong forms of personal pronouns, and in which iḍmār without any ambiguity means pronoun: As for ʾiḍmār it is like huwa, ʾiyyāhu, ʾanta, ʾanā, naḥnu, ʾantum, ʾantunna, hunna, hum, hiya, the -t(u, a, i) of faʿal-tu, faʿal-ta, faʿal-ti, and all augments after the -t, such as faʿal-tumā, faʿal-tum, faʿal-tunna, the -ū of faʿal-ū, the -nā of faʿal-nā for the dual and the plural, the -n(a) of faʿal-na, and the ʾiḍmār that does not have a phonetic [lit. explicit] marker, such as qad faʿala ḏālika, the -ā of faʿal-ā, the -k and -h in raʾaytu-ka and raʾaytu-hu and the augments after them, such as raʾaytu-kumā, raʾaytu-kum, raʾaytuhumā, raʾaytu-hum, raʾaytu-kunna, raʾaytu-hunna, -ī (-iy) in raʾaytu-nī, the ʾalif and nūn [-nā] in raʾaytu-nā and ġulāmu-nā, the -k(a) and -h(i) as in bi-ka, bi-hi, and all augments after them, such as bi-kumā, bi-kum, bi-ka, bi-kunna, bi-himā, bi-him, bi-hinna, -iy in ġulām-ī and b-ī. Kitāb I, 188

3.2.4 ʿAlāmat ʾiḍmār Pronouns are also designated by the term ʿalāmat ʾiḍmār, as in the following quotation, which is the title of a chapter in which Sībawayhi distinguishes 9 10 11

This expression may be translated as follows: “Their beasts, on which they rode, were swift”. “I spur the animal on”. See below for a comment on these two translations.

pronouns in sībawayhi’s kitāb and related concepts

37

between the strong nominative pronouns (e.g. ʾantum ḏāhibūn) and the nominative pronouns attached to the verb (e.g. faʿal-tum): “On their use of ʿalāmat al-ʾiḍmār which cannot have the same position as the one which is on the verb, when it does not have this position” (hāḏā bābu stiʿmālihim ʿalāmata al-ʾiḍmāri llaḏī lā yaqaʿu mawqiʿa mā yuḍmaru fī l-fīʿli ʾiḏā lam yaqaʿ mawqiʿahu; Kitāb I, 330). In this chapter heading as well as in the whole chapter, ʿalāmat al-ʾiḍmār refers to both the strong pronoun and the suffixed pronoun. In the quotation ʿalāmat al-ʾiḍmāri llaḏī lā yaqaʿu mawqiʿa mā yuḍmaru fī l-fīʿli designates the strong nominative pronoun, while mā yuḍmaru fī l-fīʿli is the suffixed nominative pronoun on the verb or the implicit one. 3.2.5 ʿAlāmat muḍmar Finally, the pronoun is also designated by the term ʿalāmat muḍmar: “The -t is the ʿalāma of the muḍmar of the addressee [of the pronoun of the second person] in the nominative” ( fa-l-tāʾu ʿalāmatu l-muḍmari l-muḫāṭabi l-marfūʿi; Kitāb I, 104). The title of chapter 212, which treats of the reflexive pronouns, is: On the contexts in which it is not possible to use the ʿalāma of the muḍmar addressee, nor that of the speaker, nor that of the one we are talking about and who is absent [/ the pronoun of the second person, nor that of the first or that of the third] (bābun lā tajūzu fīhi ʿalāmatu l-muḍmari l-muḫāṭabi wa-lā ʿalāmatu l-muḍmari l-mutakallimi wa-lā ʿalāmatu l-muḍmari l-muḥaddaṯi ʿanhu l-ġāʾibi). Kitāb I, 337

3.3 Conclusion In sum, there are at least five different designations of the pronoun in the Kitāb: ḍamīr, muḍmar, ʾiḍmār, ʿalāmat ʾiḍmār, ʿalāmat muḍmar, ḍamīr being probably the rarest one. This multiplicity of designations, as opposed to the notion of ʿamal, for instance, which is firmly established in the Kitāb, suggests that the metalanguage concerning pronouns had not yet been stabilized, and that these names are a description (of the nature) of the element, rather than a metalinguistic designation. However, the term ʾiḍmār, though used in a technical way referring to the pronoun in the contexts we have presented, seems to retain its full lexical meaning. The literal meaning of ‘the process of hiding, of concealing something’ is never far removed, as with the verb tuḍmiru in the following quotation: “The pronoun (ʾiḍmār) is definite because you conceal a noun (tuḍmiru sman)” (ṣāra l-ʾiḍmāru maʿrifatan li-ʾannaka tuḍmiru sman; Kitāb I, 188).

38

ayoub

This statement is confirmed by a crucial observation: neither the notion of ʾiḍmār, nor the notions of muḍmar and ʿalāma are specific for personal pronouns, as we will see in the following section. So what is ʾiḍmār and what is ʿalāma?

4

ʾIḍmār, muḍmar and ʿalāma: the Non-observable Entities in the Utterance

4.1 ʾIḍmār in Exegesis and Grammar The notion of ʾiḍmār seems to be rather ancient. Versteegh (1997:16f.) points out that ʾiḍmār occurs twelve times in the Tafsīr of Muqātil (d. 150/767) “to refer to something in the message that remains implicit”. The implicit text restored by exegesis is, for instance, a prepositional phrase, as in Q. 33/50 “Those who emigrated with you [to Medina deleted, Tafsīr III, 501.2]”, or the implicit consequence of an action, as in Q. 12/5 “Do not tell your vision to your brothers [or else they will be jealous of you deleted, Tafsīr II, 318.3]”. As Versteegh points out, these examples show that ʾiḍmār is a semantic device used for exegetical purposes. The important point for our present argument is that in Muqātil’s Tafsīr this device seems to have no rules, i.e. no predictable constraints limiting the nature of the hidden element and its occurrence, nor grammatical evidence that proves its existence.12 In Sībawayhi, ʾiḍmār has a syntactic and semantic function. In the first place, it is a procedure specific to the speaker. We know the distinction between tamṯīl (lā yutakallamu bihi) and ʾiḍmār in the Kitāb.13 The former, tamṯīl lā yutakallamu bihi is a metalinguistic activity, which consists of giving an example “that is not spoken” in order to make explicit the relations between some terms of the utterance that are otherwise difficult to grasp. It is a procedure of the grammarian, essential to his approach, as his aim is to explain kalām alʿArab. By contrast, ʾiḍmār lit. ‘hiding’, which consists of hiding or concealing an element, is supposed to be a procedure specific to the speaker, who chooses to retain an element of the utterance in his mind. The agentive form of the verb (the causative meaning of ʾaḍmara) shows explicitly that this operation implies an agent, here the speaker. The notion of ʾiḍmār forms a dyadic and opposite pair with ʾiẓhār ‘making a term explicit, apparent’. If we put aside the use of 12

13

Carter (2009) points out that “in later exegetical literature, the meaning of ʾiḍmār shifts to those instances of suppression where the suppressed element is necessary for the explanation of the surface structure”. Cf. Ayoub (1990:3f.).

pronouns in sībawayhi’s kitāb and related concepts

39

the term to designate pronouns, ʾiḍmār refers both to the operation of hiding and, by metonymy, to the result of this operation, i.e. to the hidden element. In its resultative meaning, when it designates the hidden element, ʾiḍmār may be defined in the same terms that were used in the case of Muqātil, as referring “to something in the message that remains implicit”, but at the same time it also refers to a mental representation, as it is hidden in the niyya (al-muḍmar fī lniyya). The ʾiḍmār concerns usually one single term, rather than an expression. This principle is explicit in al-Farrāʾ’s Maʿānī. The hidden element is also usually called muḍmar; it forms, like ʾiḍmār, an opposition pair with muẓhar, as we will see in the next sections. 4.2 Fiʿl muḍmar The element muḍmar is usually a part of speech. It transcends the division of parts of speech in the sense that it can be a noun (ism muḍmar), as in the quotation in 3.2.1, which illustrates perfectly the opposition between muẓhar/muḍmar, or it can be a verb ( fiʿl muḍmar) as in the following quotation: Know, from what I mentioned above, that the verb has three different ways with respect to the noun: a verb that is used explicitly and that it would be incorrect to keep implicit; a verb that is implicit, but may be used explicitly; and an implicit verb that is never used explicitly ( faʿraf fīmā ḏakartu laka ʾanna l-fiʿla yajrī fī l-ʾasmāʾi ʿalā ṯalāṯati majārin: fiʿlun muẓharun lā yaḥsunu ʾiḍmāruhu, wa-fiʿlun muḍmarun mustaʿmalun ʾiẓhāruhu, wa-fiʿlun muḍmarun matrūkun ʾiẓhāruhu). Kitāb I, 125

In another context, in utterances like ʾītinī bi-dābbatin wa-law ḥimāran ‘bring me a beast that is ridden, even if it were a donkey’ or ʾa-lā ṭaʿāmun wa-law tamrun ‘bring us to eat, even if it were dates’, there is always after law and ʾin an implicit verb that is the predicate on which the nouns are built14 ( fiʿl muḍmar tubnā ʿalayhi l-ʾasmāʾu; Kitāb I, 114). 4.3 Ism muḍmar Analyzing the accusative case of the two masdars in the elliptic expression suqyan wa-raʿyan ‘[We ask of Thee] a sending down of rain and protection’, which functions as a wish or a prayer of prosperity, Sībawayhi argues that this

14

Wa-law bi-manzilati ʾin lā yakūnu baʿdahā ʾillā l-ʾafʿālu fa-ʾin saqaṭa baʿdahā smun fa-fīhi fiʿlun muḍmarun fī hāḏā l-mawḍiʿi tubnā ʿalayhi l-ʾasmāʾ (Kitāb I, 114).

40

ayoub

accusative case is assigned by an implicit optative verb retained in the mind of the speaker, as the meaning of the expression implies a prayer. He argues that these masdars are not mentioned as the logical subject of a predication, nor as the predicate (‘to be built’) of an implicit noun in the mind of the speaker (ism muḍmar fī niyyatika): What also indicates to you that this is an accusative assigned by a verb is that you did not mention these masdars to build an utterance on them, as when you build [an utterance] on ʿabdullāhi when you start by it (/when it is a mubtadaʾ), and you did not make it built on an implicit noun in your mind, but on your prayer in his favor or disfavor (wa-mimmā yadulluka ʾayḍan ʿalā ʾannahu ʿalā l-fiʿli naṣbun ʾannaka lam taḏkur šayʾan min hāḏihi l-maṣādiri li-tabniya ʿalayhi kalāman kamā yubnā ʿalā ʿabdillāhi ʾiḏā btadaʾtahu, wa-ʾannaka lam tajʿalhu mabniyyan ʿalā smin muḍmarin fī niyyatika wa-lākinnahu ʿalā duʿāʾikā lahu ʾaw ʿalayhi) Kitāb I, 131

4.4 Mubtadaʾ muḍmar, fāʿil muḍmar The muḍmar may also be designated by its function in the sentence mubtadaʾ muḍmar; fāʿil muḍmar, as in the following quotation: It is good and right to say ʿabdullāhi fa-ḍribhu ‘ʿAbdallāh—hit him!’, if it is built on an explicit or implicit mubtadaʾ (mubtadaʾin muẓharin ʾaw muḍmarin). As for the explicit (muẓhar) one, this is like hāḏā zaydun fa-ḍribhu ‘this is Zayd—hit him!’ and if you want, you can keep hāḏā implicit (muḍmar), and it will operate in the same way as when it is explicit (muẓhar). This is like al-hilālu wallāhi fa-nẓur ʾilayhi ‘the new moon, by God—look at it!’, as if you said hāḏā l-hilālu, and then you give your order (wa-qad yaḥsunu wa-yastaqīmu ʾan taqūla ʿabdullāhi faḍribhu, ʾiḏā kāna mabniyyan ʿalā mubtadaʾin muẓharin ʾaw muḍmarin faʾammā fī l-muẓhari fa-qawlu-ka hāḏā zaydun fa-ḍribhu, wa-ʾin šiʾta lam tuẓhir hāḏā wa-yaʿmalu ka-ʿamali-hi ʾiḏā ʾaẓharta-hu, wa-ḏālika qawluka al-hilālu wallāhi fa-nẓur ʾilayhi, kaʾannaka qulta hāḏā l-hilālu ṯumma jiʾta bi-l-ʾamri) Kitāb I, 58

In none of these examples is there any ʾiḍmār in the mind of the speaker, if the utterance itself does not make it necessary, syntactically and/or semantically or pragmatically. If we reconsider the case of suqy-an wa-raʿy-an, and consider the speech situation in which this expression is used, we see that the speaker

pronouns in sībawayhi’s kitāb and related concepts

41

says these two masdars in the accusative when someone is mentioned in the discourse and the speaker wants to make a prayer in his favor or disfavor.15 It remains to explain at a syntactic level the accusative case of the nouns, as well as the semantic interpretation of this expression composed of two nouns as a prayer. These considerations lead Sībawayhi to postulate a verb meaning ‘I pray’, which is mentally hidden by the speaker and which produces this effect, i.e. which assigns this case, “as if you say ‘may God send down rain and pasture and may God offer you protection’” (kaʾannaka qulta saqāka llāhu suqyan wa-raʿāka llāhu raʿyan; Kitāb I, 133). In other words, Sībawayhi postulates that the mental representation of the utterance is different from the actual utterance, and that there are non-observable elements in the mind of the speaker that are necessary to the utterance. Note, however, that access to the mind of the speaker is strictly constrained by the actual utterance. It is only because the actual utterance contains a case ending that cannot be explained by the constituents of the utterance that the hypothesis of an implicit verb is set up. This hypothesis is reinforced by a grammatical argumentation, in this case by the fact that the meaning of the utterance does not allow any interpretation in which the two masdars are the logical subject or the predicate of a statement. In sum, we can say that the notion of ʾiḍmār, as it functions in the Kitāb, implies an important theoretical postulate assumed by this grammar, namely that there are non-observable elements implied by the terms of an utterance that have an empirical effect on the terms of the utterance. These nonobservable elements are also supposed to be mental representations of the speaker. In other words, the mental representation of the utterance may be different from its actual—and empirical—form. Note, however, that even though the hidden element is designated as muḍmar, as far as we know, the terms ʿalāma or ʿalāmat muḍmar are not used in the Kitāb to designate the unexplained case. The term ʿalāmat muḍmar always designates the pronoun. The postulate of non-observable elements operating on the actual form of the utterance distinguishes this grammar from numerous other grammars in the history of grammatical thinking. Here, one thinks in particular of structural theories, which are incompatible with the postulate of non-observable elements in a structure. By contrast, generative grammar seems to be close to the grammar initiated by Sībawayhi. Like Sībawayhi’s grammar, generative

15

Wa-ʾinnamā yantaṣību hāḏā wa-mā ʾašbahahu ʾiḏā ḏukira maḏkūrun fa-daʿawta lahu ʾaw ʾalay-hi, ʿalā ʾiḍmāri l-fiʿli, ka-ʾannaka qulta saqāka llāhu suqyan wa-raʿāka raʿyan (Kitāb I, 133).

42

ayoub

grammar postulates non-observable elements that have an effect on the phonic sequence—without postulating, however, that these elements are mental representations.16 4.5 Muḍmar and ḍamīr in the Kitāb al-ʿayn Actually, the theoretical use of the notions of ʾiḍmār and muḍmar seems to have been well-established before the Kitāb. The term muḍmar is used twice with the same theoretical characteristics in the Kitāb al-ʿayn, the first dictionary in the tradition, contemporaneous to the Kitāb and conceived by Sībawayhi’s teacher, al-Ḫalīl (d. 170/786 or 175/791). Both contexts concern a fiʿl muḍmar which assigns the accusative to the explicit noun in the same utterance. The first one is pragmatically and syntactically very close to suqyan. This is the welcome expression marḥaban! The term ʾiḍmār is used here by the Kitāb alʿayn for the hiding of a verb that is missing in the message and that explains the accusative: marḥaban is assigned accusative case by a fiʿl muḍmar: inzil ʾaw ʾaqim marḥaban (ʿAyn III, 215). This last expression is from al-Layṯ, who comments on al-Ḫalīl’s statement that in marḥaban, there is a hidden verb.17 The second context is another non-assertive utterance in a line by the poet alʾAʿšā in which he describes the Persians in the famous battle of Ḏū Qār at the beginning of the 7th century: qālū l-baqiyyat-a wa-l-hindiyyu yaḥsuduhum/walā baqiyyata ʾillā l-ṯaʾru fa-nkašafū ‘They said: Mercy! Mercy! and the swords were reaping [exterminating] them/There was no mercy except revenge and they were badly defeated’. The implicit verb assigning the accusative to albaqiyyat-a is, according to the Kitāb l-ʿayn, ʾabqū/ʾalqaw (naṣaba l-baqiyyata bi-fiʿlin muḍmarin ʾay ʾabqū/ʾalqaw; ʿAyn III, 112).18 The reason for legitimizing this ʾiḍmār is pragmatic: since the implicit element is known by the addressee, it can be hidden (or ‘killed’: fa-lammā ʿurifa maʿnāhu l-murādu ʾumīta l-fiʿlu; ʿAyn III, 215).

16 17 18

See for instance Chomsky (1981). Qāla l-Layṯu: wa-suʾila l-Ḫalīlu ʿan naṣbihi fa-qāla: fīhi kamīnu l-fiʿli, ʾarāda inzil ʾaw ʾaqim fa-nuṣiba bi-fiʿlin muḍmarin fa-lammā ʿurifa maʿnāhu l-murādu ʾumīta l-fiʿlu (ʿAyn III, 215). The published text gives: ʾalqaw. But the meaning of the word makes this reading difficult and favors another hypothesis. Indeed, the expression al-baqiyya here is pronounced by the Arabs when they ask their enemy the preservation of their life during a battle. It has to be associated with the verb ʾabqā, and the expression means ‘we ask, or beg, the being spared or mercy’ (Lane I, 238). So ʾalqaw is a taṣḥīf of the imperative form of ʾabqā, i.e. ʾabqū. ʾabqū is exactly the term of the Šāriḥ of Dīwān al-ʾAʿšā: al-baqiyyata: Qālū ʾabqū ʿalaynā wa-ḥfaẓūnā (Dīwān I, 310). and he explains lexically al-baqiyya in these terms: albaqiyya min ʾabqayta ʿalayhi wa-stabqaytahu ʾiḏā raʿaytahu wa-raḥimtahu (Dīwān I, 311).

pronouns in sībawayhi’s kitāb and related concepts

43

The exact same reason for ellipsis or elision (ḥaḏf, ḫazl) in found in Sībawayhi’s Kitāb. In all the following examples, the reason legitimizing the elision is pragmatic. Analyzing expressions like ʿamraka llāha ‘I beg God to prolong thy life’, Sībawayhi considers that ʿamr is assigned the accusative case by the verb ʿammara llāha ‘to ask God to prolong thy life’. In the following line, Sībawayhi analyses the expression mā fī qawmi-hā as implying an elision (ḥaḏf ) of ʾaḥadun by the speaker: law qulta mā fī qawmihā lam tīṯami/ yafḍuluhā fī ḥasabin wa-mīsami If you said that no one in her tribe equals her in nobility of lineage and in beauty/you would not sin Kitāb I, 328

In the same way, he analyses law ʾanna zaydan hunā to mean law ʾanna zaydan hunā la-kāna kaḏā wa-kaḏā, and laysa ʾaḥadun is analyzed with an implicit hunā as laysa hunā ʾaḥadun. All these elisions are legitimized by the following statement: “All this has been elided for the sake of lightness, and because it is possible to do without it because the addressee knows what it is about” ( fakullu ḏālika ḥuḏifa taḫfīfan wa-stiġnāʾan bi-ʿilmi l-muḫāṭabi bimā yaʿnī; Kitāb I, 328).19 Finally, if the Kitāb al-ʿayn defines ḍamīr in its current usage in language as the thing you concealed in your mind (al-šayʾu llaḏī tuḍmiruhu fī ḍamīri qalbika under ḍ-m-r, ʿAyn VII, 41), the usage of ḍamīr as a grammatical concept in the ʿAyn is actually that of an implicit element that has an effect on the actual utterance, i.e. what we have called the unobservable elements, as in this example wa-yuqālu maḥlūfatan bi-llāhi mā qāla ḏāka ‘[I swear] an oath by God, he never said that’, where maḥlūfatan is assigned the accusative by an implicit verb yaḥlifu qualified as a ḍamīr ( yunṣabu ʿalā ḍamīri ‘yaḥlifu’ billāhi maḥlūfatan ʾay qasaman; ʿAyn III, 231); under ʾimmā lā, the ʿAyn argues that the implicit concealed element in this expression is ʾin lā tafʿalu ḏāka ‘if you do not do that’, qualified again as a ḍamīr (ʾimmā lā fīhā ḍamīru mā ḏakartu laka [i.e. ʾin lā tafʿalu ḏāka]; ʿAyn VIII, 351). In other words, ḍamīr and ʾiḍmār are synonymous in the ʿAyn and indicate an implicit element that has an effect on the actual

19

The concept of ḥāḏf is very frequent in the Kitāb and in some ways it recalls ʾiḍmār. For ḥaḏf and kaṯrat al-istiʿmāl in the Kitāb, see Carter (1991), Carter (2009), Dayyeh (2012:75– 100).

44

ayoub

utterance. The meaning of ḍamīr intersects with one of the meanings of ʾiḍmār and muḍmar in the Kitāb. 4.6 The Pronoun as a muḍmar: an Apparent Paradox We have established that the pronoun belongs to the class of muḍmars, but actually this designation raises two methodological issues. In the first place, if muḍmar means pronoun, the problem is that this muḍmar is in most cases explicit, and has a phonetic realization. This means that the terminology includes, as though we are dealing with the same phenomena, implicit elements which never appear in the structure (an implicit noun or an implicit verb), as in suqyan laka; implicit elements that may appear in the structure, as in alhilālu wallāhi fa-nẓur ʾilayhi, in which hāḏā is concealed but could be explicit (hāḏā al-hilālu); and pronouns with and without a phonetic realization. Thus, expressions like mubtadaʾ muḍmar ʾaw muẓhar, which we have met with above and which are repeated here in the Arabic version only, could be ambiguous, as mubtadaʾ muḍmar could designate a pronoun in those cases where Sībawayhi clearly intends an implicit noun without phonetic realization: wa-qad yaḥsunu wa-yastaqīmu ʾan taqūla ʿabdullāhi fa-ḍribhu ʾiḏā kāna mabniyyan ʿalā mubtadaʾin muẓharin ʾaw muḍmarin fa-ʾammā fī l-muẓhari fa-qawluka hāḏā zaydun fa-ḍribhu wa-ʾin šiʾta lam tuẓhir hāḏā wayaʿmalu ka-ʿamalihi ʾiḏā ʾaẓhartahu, ka-qawlika al-hilālu wallāhi fa-nẓur ʾilayhi. Kitāb I, 58

In the same way, in the quotation presented above and repeated below, the expression muẓharin ʾaw muḍmarin could be translated either by ‘explicit or implicit’, or by ‘lexical or pronominal’, which does not yield the same analysis. As a matter of fact, we have hesitated for a long time between the two interpretations before opting for the second one. It is true that in the example analyzed (ḍarabanī wa-ḍarabtu qawmaka), there is no phonetic realization of the pronoun. But the second interpretation is more convincing, as pronouns are never referred to as muẓhar in the Kitāb: “The verb cannot lack an implicit or explicit noun in the nominative [which functions as its subject]/ The verb has necessarily a subject, lexical or pronominal, marked for the nominative” (lā yaḫlū l-fiʿlu min muḍmarin ʾaw muẓharin marfūʿin min al-ʾasmāʾ; Kitāb I, 31).20

20

See also Levin (1989:52), who interprets this passage in the same way.

pronouns in sībawayhi’s kitāb and related concepts

45

In the second place, the ambiguity is based on an apparent paradox. How is it that the pronoun has, in most cases, a phonetic realization, being literally a muẓhar, an explicit element, although it is always called a muḍmar? The following expression seems very paradoxical: “The muḍmar in the genitive case that you mentioned to address your discourse”21 (al-muḍmar al-majrūr allaḏī ḏakartahu li-l-muḫāṭabati; Kitāb I, 106). In sum, there is something implicit (muḍmar), which is explicitly mentioned to the addressee without ceasing to be called a muḍmar. At a first level, the paradox has a simple solution. Actually, what is muḍmar is the noun.22 This seems to be the reason why the pronoun belongs to the class of muḍmars: “The pronoun (ʾiḍmār) is definite because you conceal a noun (tuḍmiru sman) when …” (ṣāra l-ʾiḍmāru maʿrifatan li-ʾannaka tuḍmiru sman …; Kitāb I, 188). This means that the pronoun is not really the muḍmar and this designation is an abbreviation, or, more precisely, a metonymic designation. The explicit pronoun is only a ʿalāma of the muḍmar. The fact that the expression ʿalāmat muḍmar is the most frequent term for pronouns seems to confirm this hypothesis. What then about this notion of ʿalāma, which seems to be the necessary link to understand the apparently paradoxical designation? 4.7 What Is a ʿalāma in Sībawayhi’s Kitāb? Markers and Zero Sign The term ʿalāma means literally ‘sign’, ‘mark’, ‘indication’ (al-ʿalāma al-sima; cf. Lisān s.v.).23 The linguistic notion of ʿalāma is an ancient notion, prior to the Kitāb. In the Kitāb al-ʿayn, we find notions like ʿalāmat al-tawkīd (ʿAyn II, 152), ʿalāmat al-taʾnīṯ (ʿAyn VI, 296), which are very similar in their characterization to those found in Sībawayhi. In the Kitāb, ʿalāma is a well-established notion, with around two hundred occurrences, including its plural ʿalāmāt (Troupeau 1976:146). It is a general notion, not specific for one grammatical question, and it is found in different chapters throughout the book: [al-tanwīn] ʿalāmatu l-mutamakkin (Kitāb II, 157); ʿalāmat al-tawkīd (Kitāb II, 157); al-ʿalāma almuḫtaṣṣa (i.e. the proper name) (Kitāb I, 187);24 ʿalāmat al-nudba (Kitāb I, 361); ʿalāmat al-taʾnīṯ (Kitāb I, 361); ʿalāmat al-jamʿ or ʿalāma li-l-jamʿ (Kitāb II, 7). Undoubtedly, the notion of ʿalāma is fundamentally used for pronouns, and, annexed to muḍmar or ʾiḍmār, it is the main term for pronouns in the Kitāb.

21 22 23 24

The example analyzed is ʿalaykum ʾanfusikum, and the muḍmar in question is -kum in ʿalaykum. We will see that this interpretation is the one retained by some of the later grammarians. Cf. also Lane 1863 s.v. ʿalāma. http://www.tyndalearchive.com/tabs/lane/. This notion of al-ʿalāma al-muḫtaṣṣa reminds us of Kūfan terminology where the proper name is al-ism al-ḫāṣṣ as against ism al-ʿalam in Baṣran terminology.

46

ayoub

A simple look at the titles of the eighteen chapters dedicated to pronouns, as presented above, suffices to show this. They all deal with ʿalāmāt al-muḍmarīn; both clitic and strong pronouns are called ʿalāmat muḍmar or ʿalāmat ʾiḍmār, as in the following quotation, which is relevant for the strong pronouns: Know that the ʿalāma of the nominative muḍmar, if he speaks of himself, is ʾanā … The ʿalāma of the muḍmar to be addressed, if he is alone, is ʾanta and if you are addressing two people, their ʿalāma is ʾantumā … As for the ʿalāma of the muḍmar of which one speaks, it is huwa if he is alone or hiya if the muḍmar is feminine (iʿlam ʾanna l-muḍmara l-marfūʿa ʾiḏā ḥaddaṯa ʿan nafsihi fa-ʿalāmatuhu ʾanā … wa-ʾammā l-muḍmaru lmuḫāṭabu fa-ʿalāmatuhu ʾin kāna wāḥidan ʾanta wa-ʾin ḫāṭabta ṯnayni fa-ʿalāmatuhumā ʾantumā … wa-ʾammā l-muḍmaru l-muḥaddaṯu ʿanhu fa-ʿalāmatuhu huwa wa-ʾin kāna muʾannaṯan fa-ʿalāmatuhu hiya …). Kitāb I, 329f.

The following quotation introduces the suffixation of the clitic pronouns to the verb and the noun, and of the morphological changes they entail on the word to which they are suffixed. The clitic pronoun is called ʿalāmat al-muḍmar: Know that the elision of the nūn [on the noun in the plural] and the tanwīn [on the noun] is necessary with the bound pronoun (ʿalāmati l-muḍmari ġayri l-munfaṣili), as the latter is not used in isolation, but only if it is suffixed [attached] to a preceding verb or to a noun to which a ḍamīr can be suffixed. Consequently, it became like the nūn and the tanwīn on the noun (wa-ʿlam ʾanna ḥaḏfa l-nūni wa-l-tanwīni lāzimun maʿa ʿalāmati l-muḍmari ġayri l-munfaṣili li-ʾannahu lā yutakallamu bihi mufradan ḥattā yakūna muttaṣilan bi-fiʿlin qablahu ʾaw bi-smin fīhi ḍamīrun fa-ṣāra ka-ʾannahu l-nūnu wa-l-tanwīnu fī l-ismi). Kitāb I, 79

This presence of ʿalāma in the designation of pronouns has to be linked to the main aim of these chapters, the identification of all the forms of pronouns. Actually, the same morpheme can have two different statuses and be the ʿalāma of different phenomena. For instance the -ū of ḍarabū can be a bound pronoun (ʿalāma li-l-ʾiḍṃār), or a simple plural marker (alāma li-l-jamʿ).25 The -t suffixed to the verb can be a pronoun (ʿalāmat muḍmar), as in qul-tu, or a simple fem-

25

See the extensive discussion in Levin’s (1989) study of ʾakalūnī l-barāġīṯ.

pronouns in sībawayhi’s kitāb and related concepts

47

inine marker (ʿalāmat taʾnīṯ), as in qāl-at. The following quotation, extracted from the chapters devoted to proper names, discusses what happens if we call someone muslimīn or ḍarabat. When muslimīn is a proper name, the final -n is considered by some speakers not to be a suffix, but a ḥarf ʾiʿrāb, so they say hāḏā muslimīnun. They add the tanwīn and keep the ī even in the nominative, as the initial status of the -ū of the final -ūna in muslim-ūna is not that of a marker of ʾiḍmār (ʿalāmat ʾiḍmār), but a marker of the plural (ʿalāma li-l-jamʿ). Likewise, the ending -at of ḍarab-at, as it is a marker of taʾnīṯ, becomes ḍarab-at with a hāʾ [i.e. a tāʾ marbūṭa] “because the word then belongs to the class of nouns” (li-ʾanna-hā daḫalat fī l-ʾasmāʾi; Kitāb II, 7): You did this to that [name, i.e. muslimīn] since it [i.e., the suffix -īn] was not a ʿalāmat ʾiḍmār but a ʿalāma li-l-jamʿ, in the same way you did it to ḍarabat since [the suffix -at] was only a ʿalāma li-l-taʾnīṯ (wa-ʾinnamā faʿalta hāḏā bi-hāḏā ḥīna lam yakun ʿalāmatan li-l-ʾiḍṃāri wa-kāna ʿalāmatan li-l-jamʿi, kamā faʿalta ḏālika bi-ḍarabat ḥīna kānat ʿalāmatan li-ltaʾnīṯi). Kitāb II, 7

This assigns a morphological meaning to the notion of ʿalāma used for pronouns. In an extensive study of the suffixes on the verb as pronouns or simple markers of feminine and plural in the dialectal variant of ʾakalūnī l-barāġīṯu, Levin (1989) translates ʿalāma as ‘marker’. A marker, as we know, in its wider usage is a free or bound morpheme indicating the grammatical function of the marked word, phrase, or sentence. Actually, in most of its usages, ʿalāma is a marker, which is adequate for pronouns. There is nevertheless a use of ʿalāma that must be underlined, as it is a theoretically innovative notion, since the ʿalāma is not always ẓāhira and the absence of the realization of a morpheme may by itself be a ʿalāma: The tanwīn is a ʿalāma of what is the most powerful (al-ʾamkan) and the lightest for them [i.e. the Arabs], and its omission is the ʿalāma of what they consider to be heavy ( fa-l-tanwīnu ʿalāmatun li-l-ʾamkani ʿindahum wa-l-ʾaḫaffi ʿalayhim, wa-tarkuhu ʿalāmatun li-mā yastaṯqilūna) Kitāb I, 6

The absence of signs is a sign, it has a grammatical value. This reminds us of the zero sign, a notion developed by Jakobson in 1939, emphasizing the importance in language of the opposition between ‘something and nothing’. Languages frequently signify semantic oppositions by contrasting something to nothing, i.e.

48

ayoub

an overt sign to a zero sign. If we follow Jakobson, the zero sign is very close to what we discuss in two points: i. it is always considered in a pair where one term displays a sign and the other one displays the absence of this sign; and ii. this absence has thus a grammatical status. These two elements are always found in the discussion about the absence of the tanwīn in Sībawayhi: it is a zero sign designating what the speakers consider to be ‘heavy’, the overt sign (the tanwīn) being the sign of lightness and powerfulness. Consequently, even though the notion of ‘marker’ is often adequate, in some contexts it is appropriate to translate ʿalāma by ‘sign’, linking it to the theory of markedness, in order to include both overt signs, i.e. markers that have by definition a phonetic realization, and the zero sign, where the absence of the phonetic realization is a ‘marker’. Just like unmarked forms (e.g. nominative case in many languages) tend to be less likely to have markers, the light semantic categories in Sībawayhi’s Kitāb (masculine, singular, indefinite) tend to have no markers (a zero sign), contrasting with the heavy ones that do have a marker (feminine, plural, definite).26 The mere expression ʿalāma ẓāhira means that alāma is not always a morpheme and is something wider than a marker with a phonetic realization. Yet, it seems that the notion of ʿalāma ẓāhira in the Kitāb does not define a pronoun as there are pronouns without any ʿalāma ẓāhira, such as the subject pronoun of the third person, and both of them have the same manzila: “hiya cannot have the same function as the pronoun (ʾiḍmār) in faʿala because the latter has the same linguistic status as the pronoun (ʾiḍmār) that has a ʿalāma” (hiya lā taqaʿu mawḍiʿa l-ʾiḍmāri llaḏī fī faʿalat li-ʾanna ḏālika l-ʾiḍmāra bi-manzilati l-ʾiḍmāri llaḏī lahu ʿalāmatun; Kitāb I, 330). In this context, ʿalāma could be correctly translated as ‘a phonetic realization’. However, with respect to pronouns, the absence of any marker on the verb is not a ʿalāma in the Kitāb, unlike the absence of tanwīn. It is easy to understand why this is so: the absence could be an absence of pronoun in the case of a lexical subject, or it could be a pronoun of the third person singular, where there is no lexical subject. So, in his study of pronouns, Sībawayhi only considers two cases: the case of pronouns with a phonetic realization (ʿalāma ẓāhira), and those without one, as in this pair of opposite elements: al-ʾiḍmār allaḏī lahu ʿalāma vs al-ʾiḍmār allaḏī laysa lahu ʿalāma ẓāhira. From this point of view, the notion of ʿalāma, applied to pronouns, seems to have a more morphological significance than above. The expressions ʿalāmat muḍmar or ʿalāmat ʾiḍmār refer exclusively and specifically to the explicit pronoun, never to an implicit one.

26

See Ibn Yaʿīš, who justifies the absence of a morpheme in the third person in terms very similar to those of the theory of markedness.

pronouns in sībawayhi’s kitāb and related concepts

49

It is used in particular when Sībawayhi presents the forms of pronouns (as in the titles of the chapters cited above). At the same time, it seems to us that almuḍmar fī l-niyya (Kitāb I, 131) is never used for pronouns, but exclusively for what we have called non-observable entities. Later on, the absence of a specific morphology for the third person will be treated as a ʿalāma, exactly like the absence of the tanwīn. It will therefore be included in what we call the zero sign. This passage of Ibn al-Sarrāj for example is very significant: “The sign of the absent is that it has no sign” ( fa-ṣāra ʿalāmatu al-ġāʾibi ʾanna lā ʿalāmata lahu; ʾUṣūl II, 115). Once again, this statement is thought of in an opposing pair: it is because the first and second person pronouns have a marker that the third person may not have a marker, or have a zero marker or sign (ibid. and see above, n. 24). 4.8 What Is Hidden Is the Reference As a conclusion to this characterization of muḍmar and ʿalāma, the following question may be asked: what brings together the class of muḍmars or the two meanings of the term muḍmar, which designates at the same time pronouns with and without phonetic realization, as well as non-observable elements implied by the structure and having an effect on the form of the structure? Two answers are possible. If we believe that the designation of pronouns as muḍmar is a metonymy, the real muḍmar being a noun concealed in the mind, the class of muḍmars denotes a class of implicit elements without phonetic realization. But this answer is somewhat unsatisfactory, since pronouns with a phonetic realization are actually called muḍmar. It looks like the true opposition of muḍmar/muẓhar is at the reference level, the phonetic one being irrelevant: the lexical element, called muẓhar by Sībawayhi, has an explicit reference. What is muḍmar, whether it is an explicit or implicit element, keeps its reference implicit or concealed: its form does not indicate its specified reference or it has no form. If we are right, this means that it is not always adequate to translate muḍmar in the Kitāb with ‘implicit element’. The notion used by Muqātil has seen its extension expanded.

5

Semantic and Referential Characteristics of Pronouns

What does characterize the type of muḍmars that we call pronouns? The following quotation gives us the most important characteristics of pronouns: The pronoun (ʾiḍmār) is definite because you conceal [the reference of] a noun only after you know that the one you have spoken to knows whom

50

ayoub

you intend or what you intend, and that you intend [to refer to] something specified [to the exclusion of the others]/that he knows (wa-ʾinnamā ṣāra l-ʾiḍmāru maʿrifatan li-ʾannaka ʾinnamā tuḍmiru sman baʿdamā taʿlamu ʾanna man tuḥaddiṯu qad ʿarafa man taʿnī wa-mā taʿnī wa-ʾanna-ka turīdu šayʾan bi-ʿayni-hi/ yaʿlamu-hu27)28 Kitāb I, 188 Derenbourg; II, 6 Hārūn

So pronouns are only nouns: “The pronoun (ʾiḍmār) is definite because you conceal a noun (tuḍmiru sman) when …” (ṣāra l-ʾiḍmāru maʿrifatan li-ʾannaka tuḍmiru sman …; Kitāb I, 188). Even if the interpretation of this quotation could be that the element concealed is a noun, it is certain that the pronoun is a noun, unambiguously, for Sībawayhi, because it has the distribution of a noun, as ʾanta in the example ʾanta l-rajulu kullu l-rajuli ‘You are the perfect man’ (Kitāb I, 190), and we know that the distribution of an element is essential to determining its syntactic category in the Kitāb.29 Pronouns are a homogeneous class because of two additional properties that constitute a necessary and sufficient condition for pronominalization or for hiding (the reference of) a noun: – referential property: by using a pronoun the speaker refers to a particular, i.e. an individual, not to a predicate (turīdu šayʾan). Moreover, one’s reference to this particular is to the exclusion of all other individuals (šayʾan bi-ʿayni-hi). This feature, a particular referred to, to the exclusion of all other individuals, is one of the features distinguishing this muḍmar from other muḍmar and implicit elements fī l-niyya: the implicit noun in al-hilālu wallāhi could be hāḏā or huwa. Here, one has to identify an operator with a meaning shared by more than one synonym; in the case of a pronoun, one identifies the exact reference of an individual. – pragmatic property: the knowledge of the addressee of what or whom the speaker means and the speaker’s knowledge that the listener knows; the

27

28 29

We agree with Derenbourg’s reading turīdu šayʾan bi-ʿayni-hi (Kitāb I, 188) as against Hārūn’s reading turīdu šayʾan yaʿlamuhu (Kitāb II, 6), since bi-ʿayni-hi is a reading consistent with all definitions of definite elements in the same chapter. In each definition, Sībawayhi introduces the formula “you intend to designate something [specific], exclusively” (turīdu šayʾʾan bi-ʿayni-hi). Moreover, yaʿlamu-hu would be redundant with the first part of the quotation (man tuḥaddiṯu qad ʿarafa man taʿnī). For a translation and discussion of this passage, see also Marogy (2010:102, 107 f.). See Kitāb I, 2, where it is argued that the verb (bināʾ yafʿalu) is not a noun as it does not have the mawḍiʿ of nouns: wa-yubayyinu laka ʾannahā laysat bi-ʾasmāʾ ʾannaka law waḍaʿtahā mawāḍiʿa l-ʾasmāʾ lam yajuz laka ʾa-lā tarā ʾannaka law qulta ʾinna yaḍriba yaʾtīnā waʾašbāha hāḏā lam yakun kalāman.

pronouns in sībawayhi’s kitāb and related concepts

51

speaker pronominalizes a noun when he knows that his addressee has identified it. In other words, the reference of a pronoun is what is called in general linguistics a specific reference: the individual referred to is identified by the speaker and the addressee. Sībawayhi adds another precision: not only does the addressee know exactly which individual is referred to, but pronominalization is triggered only when the speaker knows that the addressee knows. The fact that the pronoun is still designated as a muḍmar signifies one of its important characteristics: its reference, though specific, is not presented by the form of the utterance. According to Sībawayhi, this kind of reference is precisely the reason why the pronoun cannot be qualified: “Know that the pronoun (muḍmar) cannot be qualified, as you only pronominalize/hide the reference (tuḍmiru) when you see that the one you speak to knows who you mean” (wa-ʿlam ʾanna lmuḍmara lā yakūnu mawṣūfan min qibali ʾannaka ʾinnamā tuḍmiru ḥīna tarā ʾanna l-muḥaddaṯa qad ʿarafa man taʿnī; Kitāb I, 190). The validity of this reason might be called into question, as a proper name can be qualified, even when it is identified with absolute certainty by the listener. However, the fact that kull in marartu bihim kullihim ‘I passed all of them’ (Kitāb I, 190) indicates a set of individuals and not a predicate, may explain why Sībawayhi does not regard it as ṣifa: the gloss of this sentence ʾay lam ʾadaʿ minhum ʾaḥadan ‘i.e. I haven’t left out anyone’, clarifies this point. The term ʾaḥadan refers to an individual, and in Sībawayhi’s analysis kull does not attribute a quality (taḥliya), such as ṭawīl ‘tall’, a relationship (qarāba), such as ʾaḫīka or ṣāḥibika, therefore, it is not a ṣifa, but generalizes, by emphasizing that the statement applies to all the individuals of a class. The naḥwiyyūn, seeing that it has the same case as the noun (tajrī majrāhu), take it for a ṣifa.30 The three persons are in fact treated alike in this approach. They constitute a homogeneous class according to syntactic, referential and pragmatic criteria. In conclusion, it may be pointed out that the fact that the knowledge of the addressee is sufficient for the pronominalization of a noun, even in the case of the ġāʾib, implies that the definition of pronominalization does not require

30

The exact formulation is as follows: lākinnahā maʿṭūfatun ʿalā l-ismi tajrī majrāhu fa-liḏālika qāla l-naḥwiyyūna ṣifatun (Kitāb I, 190). As this formulation shows, it is not because these pronouns generalize and corroborate that they are analyzed as ṣifa by the naḥwiyyūn, but because they have the same case. The analysis in terms of generalization and corroboration is Sībawayhi’s analysis and it demonstrates that pronouns are not ṣifa.

52

ayoub

an antecedent—before or after—in the structure. Nothing is mentioned in the Kitāb about this.31 Another important point made clear by this discussion is that according to Sībawayhi, it is the kind of reference that determines the characterization of the pronoun, whatever the interpretation we give to its denomination as muḍmar.

6

Asymmetry and Hierarchy: Double Cliticization

In the treatment of many referential or syntactic properties of the noun, such as cliticization, reflexivity, qualification by a pronoun, tawkīd of a pronoun, it is clear that no distinction is made between the first and the second person and the third person, in the Kitāb. They form a homogeneous class. There are, however, some contexts where there is asymmetry and hierarchy between the three persons. In double cliticization on the verb, for instance, the first person takes precedence over the other two, followed by the addressee pronoun and then the pronoun of ‘the absent’. We say ʾaʿṭānīhi lit. ‘he gave-me-it’, i.e. ‘he gave it to me’; ʾaʿṭānīka lit. ‘he gave-me-you’, i.e. ‘he gave you to me’. About the forms *ʾaʿṭāhūnī ‘he gave-him-me’, i.e. ‘he gave me to him’, *ʾaʿṭākanī lit. ‘he gave-you-me’, i.e. ‘he gave me to you’ he says that “they are bad; they are not used by the Arabs, but formed analogically by the naḥwiyyūn” ( fa-huwa qabīḥun lā takallamu bihi l-ʿArabu wa-lākinna l-naḥwiyyīna qāsūhu; Kitāb I, 335).32 The reason for this hierarchy is clearly a pragmatic one: the proximity with respect to the situation of enunciation: “This is considered bad by the Arabs, since they greatly dislike that the speaker begins, in this context, by the furthest,

31

32

Levin (1998: 41, n. 4) refers to the passage Kitāb II, 331.20 f., as showing that “a pronoun of the 3rd person must be preceded by its antecedent”. For us, this passage does not speak of the antecedent. It is rather a justification of why the personal pronoun, unlike nouns, can consist of only one letter or consonant. This is because it is then bound. The context strongly suggests that by the expression lā taṣarrafu wa-lā yuḏkaru ʾillā fīmā qablahā means that the bound pronoun has only one form and is never mentioned except bound to the term preceding it: fīmā qablahā. Another passage is given by Levin to show the necessity of the antecedent. It is I, 188.8–10 translated below. If this passage asserts there is no pronominalization without the addressee’s knowledge of the referent of the pronoun, it does not say that this referent must be an antecedent in the text. It can be known by the speech situation. Another passage is given by Levin to show the necessity of the antecedent. It can be known by the speech situation. However, the question raised by Levin, whether a pronoun of the third person should be preceded by its antecedent, seems to be legitimate for the later tradition. See also for a discussion Peled (2009).

pronouns in sībawayhi’s kitāb and related concepts

53

before the nearest” (wa-ʾinnamā qabuḥa ʿinda l-ʿArabi karāhiyata ʾan yabdaʾa lmutakallimu fī hāḏā l-mawḍiʿi bi-l-ʾabʿadi qabla l-ʾaqrabi; Kitāb I, 335 f.). The same hierarchy applies to the second and third person in cliticization on the verb. One starts by cliticizing the pronoun of the addressee, before the pronoun of the absent, as in ʾaʿṭaytu-ka-hu lit. ‘I gave-you-it’, i.e. ‘I gave it to you’; qad ʾaʿṭā-ka-hu lit. ‘he gave-you-it’, i.e. ‘he gave it to you’. Otherwise, the pronoun of the addressee is not cliticized and the strong accusative pronoun is used, as in qad ʾaʿṭā-hu ʾiyyā-ka lit. ‘he gave-him you’, i.e. ‘he gave you to him’ (Kitāb I, 336). The order of cliticization is justified by the properties of the speech situation: the speaker has priority, followed by the addressee, and then by the absent, whatever it is. The discourse-pragmatic rule governing this is summarized by Sībawayhi in the following way: The addressee must be given priority because he is closer to the speaker than the absent. Just as the speaker must quote himself first before the addressee, the addressee must be quoted first because he is closer than the absent (wa-ʾinnamā kāna l-muḫāṭabu ʾawlā bi-ʾan yubdaʾa bihi min qibali ʾanna l-muḫāṭaba ʾaqrabu ʾilā l-mutakallimi min al-ġāʾibi fa-kamā kāna l-mutakallimu ʾawlā bi-ʾan yabdaʾa bi-nafsihi qabla l-muḫāṭabi, kāna l-muḫāṭabu llaḏī huwa ʾaqrabu min al-ġāʾibi ʾawlā bi-ʾan yubdaʾa bihi min al-ġāʾibi). Kitāb I, 336

This passage shows us that, undeniably, priorities regarding the pronouns are organized in the Kitāb around the speaker. We know that this is the case in discourse in general, as noticed above.33 Sībawayhi’s examples clearly show that the speaker is distinguished regarding his link to enunciation:34 the Kitāb

33

34

In this case, too, the hierarchy is efficient in language in general, not only for pronouns. Thus, some contexts exclude the interpretation of the ġāʾib, the interpretation being exclusively for the addressee. This is the case when you say ‘Zaydan!’ to someone, where there is no pronoun. But this utterance cannot be interpreted as telling your addressee to inform someone else to hit Zayd. The only possible interpretation is determined by the pragmatic situation: you are in the presence of an addressee whom you tell to hit Zayd (but see Marogy 2010:88 for another interpretation). Sībawayhi believes there is no difference between the three persons regarding the obligation to use nafs + pronoun in order to obtain the reflexive interpretation, but al-Sīrāfī (Šarḥ III, 130) reports the arguments of some grammarians to justify the presence of reflexives, as in ḍaraba nafsahu and not ḍarabahu, by focusing on the third person as this is the only case where there might be confusion; in ḍarabtunī and ḍarabtaka there is no confusion.

54

ayoub

makes a distinction between the class of those spatial and temporal expressions that take the speaker as origin of the location, and the class of those spatial and temporal expressions that do not take the speaker as origin.35 Finally, one might ask why, as shown by Sadan (2018), the third person is sometimes included by Sībawayhi in ‘the vague nouns’ (ʾasmāʾ mubhama), i.e. the demonstratives, and sometimes not. This may be either because the third person has more than one reference, or because it can replace a demonstrative (ism mubham) in indicating someone or something. The second hypothesis is probably more plausible. Huwa can have a deictic dimension, but, even in this case, it remains the one about whom you speak and who does not speak, almuḥaddaṯ ʿanhu.

7

The Development of the Terminology

The ambiguity of the reference of the term muḍmar, which refers to two different kinds of entities, an implicit element and a pronoun, and the paradoxical terminology in which a muḍmar has a phonetic form (ʿalāma ẓāhira), without ceasing to be called a muḍmar, seems to have favored the only designation specific for pronouns, ḍamīr. This term probably has a passive meaning, both in the Kitāb al-ʿayn and in Sībawayhi’s Kitāb—the past participle faʿīl, as we know, can have a passive meaning in the language (ex: jarīḥ meaning majrūḥ). So, in both cases ḍamīr has the same meaning as muḍmar, although its theoretical meaning in Sībawayhi is different from that in the Kitāb al-ʿayn. Its specific use in the Kitāb was widely adopted later as the metalinguistic designation of pronoun, without abandoning ʿalāmat ʾiḍmār, ʿalāmat muḍmar, or muḍmar. However, in al-Mubarrad’s (d. 285/898) Muqtaḍab, this terminology is firmly established: ḍamīr has a firm metalinguistic meaning, and is used exclusively in all of its around fifty occurrences for what we mean presently by it, a pronoun. Moreover, al-Mubarrad adopts taqdīr instead of—and along with—ʾiḍmār for the hiding of an element missing in the message. But taqdīr in some of its occurrences is still used in the sense of tamṯīl in the Kitāb, i.e. as a representation by other elements of the abstract relations in the utterance. Thus, for instance, the taqdīr of sarranī qiyāmu ʾaḫīka ‘the standing up of your brother makes me happy’ is sarranī ʾan qāma ʾaḫūka ‘it makes me happy that your brother stood up’ (Muqtaḍab I, 14). The term muḍmar, in most of its occurrences in the Muqtaḍab, indicates an implicit element having an effect on the form of the

35

Cf. Ayoub (2010:13).

pronouns in sībawayhi’s kitāb and related concepts

55

utterance. Yet, a few of its occurrences designate a pronoun, for instance, the bound pronoun in al-mālu la-ka; al-mālu la-nā; al-darāhimu la-hum is called muḍmar (Muqtaḍab I, 254). More important is the conceptual development. Sībawayhi’s allusion to his disagreement with the naḥwiyyūn, who do not respect the speaker’s precedence, the multiplicity of designations of the pronoun in the Kitāb, both of these signs mean that the theory is still not stabilized. Two interpretations will exist later regarding the notion of ʾiḍmār and ʿalāmat ʾiḍmār. The first one is alMāzinī’s (d. 248/862) interpretation according to which the noun muḍmar is fī l-niyya and the pronoun is only a ʿalāma, the second one is that the pronoun is the ism itself and this is the muḍmar. The latter interpretation is that of al-Sīrāfī (d. 368/978): The -tu in qul-tu is the noun indicating the speaker36 (ism al-mutakallim), while the -at in qāl-at is a sign [a marker] indicating that the verb is about a feminine [subject]. ʾAbū ʿUṯmān [al-Māzinī] and other grammarians assert that the -ā in qām-ā, and the -ū in qām-ū are only two particles (ḥarfān). [According to them], these do not indicate the dual or plural implicit subject. The subject is only in the mind of the speaker [ fī l-niyya]. This is like when you say zaydun qāma ʿZayd stood up’, there is in qāma a pronoun implicit in the mind without a phonetic sign (ʿalāma ẓāhira). So, if we put it in the dual or the plural, the pronoun is also in the mind, but there is a phonetic sign that indicates it. ʾAbū Saʿīd [al-Sīrāfī] says: My opinion is the one of Sībawayhi; they are unanimous in saying that -tu in qum-tu is the noun indicating the speaker and its pronoun (ismu lmutakallimi wa-ḍamīruhu) ( fa-hāḏihi l-ḥurūfu ʿinda Sībawayhi fī wuqūʿihā ʾasmāʾa marratan wa-ḥurūfan marratan bi-manzilati l-tāʾi fī qawli-ka qultu wa-qāl-at fa-l-tāʾu fī qul-tu ismu l-mutakallimi, wa-l-tāʾu fī qālat ʿalāmatun tuʾḏinu bi-ʾanna l-fiʿla li-l-muʾannaṯi wa-qad qāla ʾAbū ʾUṯmāna waġayruhu min al-naḥwiyyīna ʾinna l-ʾalifa fī qāmā wa-l-wāwa fī qāmū ḥarfāni lā yadullāni ʿalā l-fāʿilayni wa-l-fāʿilīna al-muḍmarīna wa-ʾinna l-fāʿila fī l-niyyati, kamā ʾanna-ka ʾiḏā qulta zaydun qāma fa-fī qāma ḍamīrun fī l-niyyati wa-laysat lahu ʿalāmatun ẓāhiratun fa-ʾiḏā ṯannā wa-jamaʿa fa-l-

36

Actually, we hesitate in the translation of ismu l-mutakallimi between: ‘the [proper] name of the speaker’ or ‘the noun of the speaker’, i.e. a noun indicating the speaker. The context of this passage indicates unambiguously that what al-Sīrāfī intends here by ism is the syntactic category of nouns, as shown by this formulation where he tries to prove that the -ā in qām-ā is a noun, since it has the distribution of the nouns: fa-lammā ḥalla maḥalla mā lā yakūnu ʾillā sman wajaba ʾan yakūna sman (Šarḥ I, 150).

56

ayoub

ḍamīru ʾayḍan fī l-niyyati ġayra ʾanna lahu ʿalāmatun; qāla ʾAbū Saʿīdin alqawlu fīhi ʿindī mā qālahu Sībawayhi, wa-ḏālika ʾanna lā ḫilāfa baynahum ʾanna l-tāʾa fī qum-tu hiya smu l-mutakallimi wa-ḍamīruhu) Šarḥ I, 149f.

Finally, the precedence of the first and second persons over the third as regards the speech situation seems to be unanimously recognized. It is based on the proximity of the actants to the speech. The speaker is closer to it than the addressee, and the latter is closer than the absent. But in the 3rd/9th and 4th/10th centuries, even if it is was asserted that the ʾaṣl was this precedence in the order of cliticization, the statement of Sībawayhi excluding the other orders was questioned by al-Mubarrad and Ibn al-Sarrāj, who do not believe it. Suffice it here to quote Ibn al-Sarrāj: If you mention a ditransitive verb and you want to attach pronouns [to it], the most legitimate is to start with the nearest before the farthest. I mean by the nearest: the speaker before the addressee and the addressee before the absent […]; so you say ʾāʿṭā-nī-hi and ʾāʿṭā-nī-ka. And you can say: ʾāʿṭā-ka-nī, and if he [the speaker] begins by the absent, ʾaʿṭā-hū-nī. Sībawayhi says these last two are bad; the Arabs do not say them. But ʾAbū l-ʿAbbās [al-Mubarrad] says: These utterances are correct. They are not bad ( fa-ʾin ḏakarta al-fiʿla llaḏi yataʿaddā ʾilā mafʿūlayni fa-ḥaqqu hāḏā lbābi ʾiḏā jiʾta bi-l-muttaṣili ʾan tabdaʾa bi-l-ʾaqrabi qabla al-ʾabʿadi wa-ʾaʿnī bi-l-ʾaqrabi al-mutakallimu qabla l-muḫāṭabi wa-l-muḫāṭabu qabla l-ġāʾibi […] wa-taqūlu ʾaʿṭā-nī-hi wa-ʾāʿṭā-nī-ka wa-yajūzu ʾāʿṭā-ka-nī wa-ʾāʿṭā-hū-nī wa-qāla Sībawayhi huwa qabīḥun lā takallamu bihi l-ʿArabu; wa-qāla ʾAbū l-ʿAbbās hāḏā kalāmun jayyidun laysa bi-qabīḥin). ʾUṣūl I, 120

8

Concluding Remarks

We will conclude with a few provisional remarks, relating both to the history of concepts and terminology, and to the history of grammar: i. The first observation is relative to terminology: the usage of the term ḍamīr exclusively for ‘pronoun’ is an innovation by Sībawayhi. It is this term that will be current in the later usage to designate the pronoun. ii. Sībawayhi, and the entire tradition after him, is very much aware of the asymmetry between pronouns of the first and the second person (almutakallim and al-muḫāṭab), on the one hand, and the third person

pronouns in sībawayhi’s kitāb and related concepts

iii.

iv.

57

pronoun (al-ġāʾib), on the other, regarding their discursive function. His approach of personal pronouns may be in terms of a homogeneous class having syntactic and pragmatic properties, but the fact remains that the precedence of the speaker and the hierarchical relations between pronouns according to the speech situation are firmly established. This hierarchy determines some morphological facts, namely the order of suffixation of pronouns. It seems that there was a debate about these facts before Sībawayhi, and after him. The approach of the Kitāb operates with strong criteria for the definition of pronouns, but it is not yet stabilized in its concepts and terminology, permitting divergent interpretations in the century following the Kitāb, particularly in connection with the status of agreement markers on the verb. Nowhere in the Kitāb, do we find any statement about the noun hidden in the case of the first and second persons. If the pronoun is a ʿalāma, what are the first and the second person markers of? Are they the markers of the implicit [proper name] of the addressee and the speaker? As far as we know, Sībawayhi nowhere settles this. These tentative conclusions undoubtedly raise many questions. We aim to examine these in a future study.

Bibliographical References A

Primary Sources

ʾAstarābāḏī, Šarḥ = Raḍī l-Dīn Muḥammad ibn al-Ḥasan al-ʾAstarābāḏī, Šarḥ Kāfiyat Ibn al-Ḥājib fī l-naḥw. Istanbul, 1893 (Repr., Beirut: Dār al-Kutub al-ʿIlmiyya, 1976). Farrāʾ, Maʿānī = ʾAbū Zakariyyāʾ Yaḥyā ibn Ziyād al-Farrāʾ, Maʿānī l-Qurʾān. Ed. by Muḥammad ʿAlī al-Najjār. 3 vols. Cairo: al-Dār al-Miṣriyya, 1966–1972. Ḫalīl, ʿAyn = ʾAbū ʿAbd al-Raḥmān al-Ḫalīl ibn ʾAḥmad al-Farāhīdī, Kitāb al-ʿayn. Ed. by Mahdī al-Maḫzūmī and ʾIbrāhīm al-Sāmarrāʾī. Baghdad: Dār al-Rašīd. Ibn al-ʾAnbārī, ʾInṣāf = ʾAbū l-Barakāt ʿAbd al-Raḥmān ibn Muḥammad Ibn al-ʾAnbārī, Kitāb al-ʾinṣāf fī masāʾil al-ḫilāf bayna l-naḥwiyyīn al-baṣriyyīn wa-l-kūfiyyīn. Ed. by Muḥammad ʿAbd al-Ḥamīd. Beirut: Dār al-Fikr, n.d. Ibn Fāris, Maqāyīs = ʾAbū l-Ḥusayn ʾAḥmad Ibn Fāris, Muʿjam maqāyīs al-luġa. Ed. by Muḥammad Murʿib and Fāṭima ʾAṣlān. Beirut: Dār ʾIḥyāʾ al-Turāṯ al-ʿArabī, 2001. Ibn Manẓūr, Lisān = Jamāl al-Dīn ʾAbū l-Faḍl Muḥammad ibn Mukarram Ibn Manẓūr, Lisān al-ʿArab. Cairo: Dār al-Maʿārif, n.d. Ibn al-Sarrāj, ʾUṣūl = ʾAbū Bakr ibn al-Sarī Ibn al-Sarrāj, Kitāb al-ʾuṣūl fī l-naḥw. Ed. by ʿAbd al-Ḥusayn al-Fatlī. 3 vols. Beirut: Muʾassassat al-Risāla, 1985.

58

ayoub

Ibn Yaʿīš, Šarḥ = Muwaffaq al-Dīn ʾAbū l-Baqāʾ Yaʿīš ibn ʿAlī Ibn Yaʿīš, Šarḥ al-Mufaṣṣal. Beirut: ʿĀlam al-Kutub and Cairo: Maktabat al-Mutanabbī, n.d. Jurjānī, Taʿrīfāt = ʿAlī ibn Muḥammad al-Sayyid al-Šarīf al-Jurjānī, al-Taʿrīfāt. Cited from www.alwaraq.net Mubarrad, Muqtaḍab = ʾAbū l-ʿAbbās Muḥammad ibn Yazīd al-Mubarrad, Kitāb almuqtaḍab. Ed. by Muḥammad ʿUḍayma. Cairo: Lajnat ʾIḥyāʾ al-Turāṯ al-ʾIslāmī, 1968. Mubarrad, Kāmil = ʾAbū l-ʿAbbās Muḥammad ibn Yazīd al-Mubarrad, al-Kāmil fī l-luġa wa-l-ʾadab. Beirut: Muʾassasat al-Maʿārif, n.d. Sībawayhi, Kitāb = ʾAbū Bišr ʿAmr ibn ʿUṯmān ibn Qanbar Sībawayhi, al-Kitāb. Ed. by Hartwig Derenbourg. Paris: Imprimerie Nationale, 1881. (Repr. Hildesheim and New York: G. Olms, 1970.)/Ed. Būlāq. 2 vols. 1316H. (Repr., Baghdad: Librairie al-Muṯannā, n.d.)/Ed. by ʿAbd al-Salām Muḥammad Hārūn. 5 vols. Cairo: al-Hayʾa al-Miṣriyya alʿĀmma li-l-Kitāb, 1966–1977. Sīrāfi, Šarḥ = ʾAbū Saʿīd al-Ḥasan ibn ʿAbdallāh al-Sīrāfī, Šarḥ Kitāb Sībawayhi. Vol. I. Ed. by Ramaḍān ʿAbd al-Tawwāb, Maḥmūd Fahmī Ḥijāzī and Muḥammad Hāšim ʿAbd al-Dāyim. Vol. II. Ed. by Ramaḍān ʿAbd al-Tawwāb. Cairo: al-Hayʾa al-Miṣriyya al-ʿĀmma li-l-Kitāb, 1986–1990 [Šarḥ = Ed. by ʾAḥmad Mahdalī and ʿAlī Sayyid ʿAlī. Dār al-Kutub al-ʿIlmiyya. 2008. Shamela]. Suyūṭī, Muzhir = Jalāl al-Dīn ʿAbd al-Raḥmān ibn ʾAbī Bakr al-Suyūṭī, al-Muzhir fī ʿulūm al-luġa wa-ʾanwāʿi-hā. Ed. by Muḥammad ʾAbū l-Faḍl ʾIbrāhīm et al. Cairo: Dār ʾIḥyāʾ al-Kutub al-ʿArabiyya, n.d. Zajjājī, ʾĪḍāḥ = ʾAbū l-Qāsim ʿAbd al-Raḥmān ibn ʾIsḥāq al-Zajjājī, al-ʾĪḍāḥ fī ʿilal al-naḥw. Ed. by Māzin al-Mubārak. Cairo: Dār al-ʿUrūba, 1959. Zamaḫšarī, Mufaṣṣal = ʾAbū l-Qāsim Maḥmūd ibn ʿUmar al-Zamaḫšarī, al-Mufaṣṣal fī ʿilm al-ʿarabiyya. Beirut: Dār al-Jīl, n.d. Zubaydī, Ṭabaqāt = ʾAbū Bakr Muḥammad ibn al-Ḥasan al-Zubaydī, Ṭabaqāt al-naḥwiyyīn wa-l-luġawiyyīn. Ed. by Muḥammad ʾAbū l-Faḍl ʾIbrāhīm. Cairo: Dār al-Maʿārif, 1973.

B

Secondary sources

Ayoub, Georgine. 1991. “La forme du sens: Le cas du nom et le mode du verbe”. The Arabist, Budapest Studies in Arabic 3–4.37–87. Ayoub, Georgine. 2010. “al-fiʿl wa-l-hadaṯ: La description sémantique du verbe dans le Kitāb de Sībawayhi”. Jerusalem Studies in Arabic and Islam 37.1–51. Ayoub, Georgine. 2015. “Some aspects of the relations between enunciation and utterance in Sībawayhi’s Kitāb. A modal category: wājib, ġayr wājib”, The foundations of Arabic linguistics. II. The Kitāb Sībawayhi: Transmission and interpretation, ed. by Amal Marogy and Kees Versteegh, 6–35. Leiden: E.J. Brill. Baalbaki, Ramzi. 1979. “Some aspects of harmony and hierarchy in Sībawayhi’s grammatical analysis”. Zeitschrift für arabische Linguistik 2.7–22.

pronouns in sībawayhi’s kitāb and related concepts

59

Baalbaki, Ramzi. 2008. The legacy of the Kitāb: Sībawayhi’s analytical methods within the context of the Arabic grammatical theory. Leiden: E.J. Brill. Benveniste, Émile. 1946 “La nature des pronoms”. (Repr., Problèmes de linguistique générale, I, 251–257. Paris: Gallimard, 1966.) Benveniste, Émile. 1958. “De la subjectivité dans le langage”. (Repr., Problèmes de linguistique générale, I, 258–266. Paris: Gallimard, 1966.) Blachère, Régis. “al-Farrāʾ”. Encyclopaedia of Islam. 2nd ed., II, 806–808. Leiden: E.J. Brill. Carter, Michael G. 1991 “Elision”. Proceedings on the Colloquium of Arabic Grammar, ed. by Kinga Dévényi and Tamas Iványi. The Arabist Budapest Studies in Arabic and Islam 3–4.121–141. Carter, Michael G. 2004. Sībawayhi. New York: I.B. Tauris. Carter, Michael G. and Kees Versteegh. 2009. “ʾIḍmār”. Encyclopedia of Arabic language and linguistics, ed. by Mushira Eid, Alaa Elgibali, Kees Versteegh, Manfred Woidich, and Andrzej Zaborski, II, 300–302. Leiden: E.J. Brill. Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht: Foris. Dayyeh, Hanadi. 2015. “Ittisāʿ in Sībawayhi’s Kitāb: A semantic ʿilla to disorders in form and/or meaning”. The foundations of Arabic linguistics. II. Kitāb Sībawayhi: Interpretation and transmission, ed. by Amal Marogy and Kees Versteegh, 66–80. Leiden: E.J. Brill. Dévényi, Kinga. 1990. “On Farrāʾ’s linguistic method in his work Maʿānī al-Qurʾān”. Studies in the history of Arabic grammar. II. Proceedings of the 2nd Symposium on the History of Arabic Grammar, Nijmegen, 27 April–1 May 1987, ed. by Michael G. Carter and Kees Versteegh, 101–110. Amsterdam: J. Benjamins. Dévényi, Kinga. 2007. “ʾiḍmār in the Maʿānī of al-Farrāʾ: A grammatical approach between description and explanation”. Approaches to Arabic linguistics presented to Kees Versteegh on the occasion of his sixtieth birthday, ed. by Everhard Ditters and Harald Motzki, 45–65. Leiden: E.J. Brill. Jakobson, Roman. 1957. “Shifters, verbal categories, and the Russian verb”. Cambridge, Mass.: Harvard University. (French transl. by Nicolas Ruwet, “Les embrayeurs, les catégories verbales et le verbe russe”, Roman Jakobson, Problèmes de linguistique générale, I, 176–196. Paris: Minuit, 1963.) Kinberg, Naphtali. 1996. A lexicon of al-Farrāʾ’s terminology in his Qurʾān commentary. Leiden: E.J. Brill. Lane, Edward William. 1863. Arabic-English Lexicon. London: Willams and Norgate. Levin, Aryeh. 1989. “What is meant by ʾakalūnī l-barāġīṯu?”. Jerusalem Studies in Arabic and Islam 12.40–65. (Repr., Aryeh Levin, Arabic linguistic thought and dialectology. Jerusalem: The Hebrew University, 1998.) Marogy, Amal. 2010. Kitāb Sībawayhi: Syntax and pragmatics. Leiden: E.J. Brill. Owens, Jonathan. 1988. The foundations of grammar. Amsterdam: J. Benjamins. Owens, Jonathan. 1990. Early Arabic grammatical theory: Heterogeneity and standardization. Amsterdam: J. Benjamins.

60

ayoub

Peled, Yishai. 2006. “Ḍamīr”. Encyclopedia of Arabic language and linguistics, ed. by Mushira Eid, Alaa Elgibali, Kees Versteegh, Manfred Woidich, and Andrzej Zaborski, I, 555–559. Leiden: E.J. Brill. Sadan, Arik. 2018. “Demonstratives in Sībawayhi’s Kitāb”. The foundations of Arabic linguistics. III. The development of a tradition: Continuity and change, ed. by Georgine Ayoub and Kees Versteegh, 178–189. Leiden: E.J. Brill. Troupeau, Gérard. 1976. Lexique-index du Kitāb de Sībawayhi. Paris: Klincksieck. Versteegh, Kees. 1993. Arabic grammar and Qurʾānic exegesis in early Islam. Leiden: E.J. Brill. Versteegh, Kees. 1997. Landmarks in linguistic thought. III. The Arabic linguistic tradition. London and New York: Routledge.

Grammar for Beginners and Ibn Hišām’s Approach to Issues of ʾiʿrāb Ramzi Baalbaki

1

Early Pedagogical Grammars

The difficulties faced by students in learning grammar were recognized quite early in the history of the grammatical tradition. Studying the earliest and most influential grammar work, the Kitāb of Sībawayhi (d. 180/796), was likened by al-Mubarrad (d. 285/898) in its difficulty to travelling by sea.1 One student is reported to have spent a long time (mudda ṭawīla) reading the Kitāb with al-Māzinī (d. 249/863), only to declare upon its completion that he understood nothing of what he had read ( fa-lam ʾafham minhu ḥarfan).2 Possibly during Sībawayhi’s lifetime, ʾAbū Muslim, a contemporary of Muʿāḏ al-Harrāʾ’s (d. 187/803), was so confused by the grammarians’ approach to morphology that he likened their incomprehensible jargon to the speech of Blacks and Greeks (kalām al-Zanj wa-l-Rūm).3 Similarly, and very shortly after Sībawayhi, Rufayʿ ibn Salama (also known as Damāḏ), who was a contemporary of ʾAbū ʿUbayd Maʿmar ibn al-Muṯannā’s (d. 209/824), sharply criticized the grammarians in a satirical poem, admitting his utter failure to grasp why they insisted that the subjunctive in the verb after fāʾ, wāw and ʾaw is attributed to an elided ʾan.4 The realization of the need for grammar curricula that are neither too detailed nor too complicated and that exclude material of little practical value was also expressed at a relatively early stage. In his Kitāb al-Mutaʿallimīn, alJāḥiẓ (d. 255/869) expresses his conviction that the study of grammar would be

1 Al-Mubarrad is reported to have addressed any student who wished to read the Kitāb under his supervision by the expression hal rakibta l-baḥr; see Sīrāfī, ʾAḫbār 50; Baġdādī, Ḫizāna I, 371. 2 ʾAbū l-Ṭayyib, Marātib 126. 3 Zubaydī, Ṭabaqāt 125–126; Qifṭī, ʾInbāh III, 292. Note that Muʿāḏ is said to have reached a ripe, though unspecified, age and thus it is highly probable that his reported response to ʾAbu Muslim’s ridicule of the grammarians actually took place before Sībawayhi’s death. 4 Ibn Qutayba, ʿUyūn II, 156; Sīrāfī, ʾAḫbār 77–78; Tanūḫī, Tārīẖ 66–67; Qifṭī, ʾInbāh II, 5–6; cf. Baalbaki (2008:264f.); Jabbārīn (1999:332); van Gelder (2011:250–252).

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_005

62

baalbaki

beneficial only if it did not extend beyond certain prescribed limits.5 According to him, the teaching of grammar should be aimed at guarding the pupil against blatant solecism ( fāḥiš al-laḥn) and should exclude what he describes as complex grammar (ʿawīṣ al-naḥw). Al-Jāḥiẓ’s praise, as quoted by al-Qifṭī,6 of al-Kisāʾī (d. 189/805) for having written comprehensible and well-elucidated works (kutub mafhūma ḥasanat al-šarḥ) is a clear indication that some other works were hardly comprehensible and hence useless for teaching purposes. His encounter with ʾAbū l-Ḥasan al-ʾAḫfaš (d. 215/830) also points in the same direction.7 Attempts at authoring works addressed to beginners thus come as no surprise. With the growing need for teaching grammar, it must have become obvious to scholars and teachers of grammar that a text such as the Kitāb is hardly useful for beginners. In fact, the Kitāb is most unlikely to have been intended by Sībawayhi as a textbook for teaching grammar. Rather, it gives a profound and detailed analysis of the syntactical and morphological features of the speech of the Arabs, and the arguments that characterize its author’s interpretation and justification of usage are obviously too complex to suit the needs of beginners. What is surprising, however, is the large number of relatively early works (i.e. not later than the 4th/10th century) that are intended for pedagogical purposes.8 In this respect, Ibn al-Nadīm’s (d. 380/990) Kitāb al-Fihrist is very useful since it is the most detailed bibliographical reference for books written up to its author’s time. By examining the Fihrist’s second part, which is devoted to the naḥwiyyīn and luġawiyyīn, one readily concludes that grammar teaching manuals appeared considerably earlier and were much more widespread than has been hitherto recognized. Carter’s conclusion that the first purely pedagogical grammars for beginners emerged in Ibn Jinnī’s (d. 392/1002) period,9 5 Jāḥiẓ, Mutaʿallimīn 38ff.; see also the English translation by Pellat (1969:113). 6 Qifṭī, ʾInbāh II, 271–272. 7 By asking al-ʾAḫfaš why his books are only partially comprehensible and why he gives precedence to what is complex (ʿawīṣ) over what is comprehensible (mafhūm), al-Jāḥiẓ indirectly makes the point that grammar books would be fully accessible to the learner if their authors were to simplify their approach and abandon their deliberately complicated techniques. AlʾAḫfaš’s response is equally telling: he defends his method of authorship which ensures that people would need his expertise, and admits that the comprehensible parts of his books are merely a snare intended to make people experience a sweetness (ḥalāwa) that would urge them to seek it again, but based on his own elucidation of what was previously incomprehensible to them. See Jāḥiẓ, Ḥayawān I, 91f. and cf. Baalbaki (2009:105). 8 This paper will not deal with works of the genre maʿānī l-Qurʾān, for whose role in ‘the pedagogization of Arabic grammar, the encapsulation and packing of Arabic grammar for didactic purposes’, see Aljassar and Owens (2015:32). 9 Carter (1990:131).

grammar for beginners

63

should be amended in the light of titles cited in the Fihrist. The most frequently used word in the titles of such grammars is muḫtaṣar, but whether these were abridgments of other works (and hence simplified versions of texts that are not suitable for beginners) or originally written for beginners is difficult to determine. In some cases, particularly when the books are referred to as Muḫtaṣar naḥw or Muḫtaṣar fī l-naḥw, it is possible that Ibn al-Nadīm gives a description of the work, rather than its title. That at least some of the works that include muḫtaṣar in their titles were indeed pedagogical grammars can be clearly demonstrated by titles (or descriptions) such as Muḫtaṣar naḥw li-l-mutaʿallimīn by ʾAbū ʿUmar al-Jarmī (d. 225/ 840) and al-Muḫtaṣar li-l-mutaʿallimīn by al-ʿAjlānī, who, Ibn al-Nadīm says, was close to his own time (qarīb al-ʿahd).10 It is quite significant that Ibn alNadīm also refers to al-Jarmī’s book as Muḫtaṣar al-naḥw11 since this strengthens the possibility that books with a similar title were indeed meant for beginners. Based on the precise wording of Ibn al-Nadīm, titles that include the word muḫtaṣar are of five types: – al-Muḫtaṣar, by Hišām ibn Muʿāwiya al-Ḍarīr (d. 209/824).12 – Muḫtaṣar naḥw, by Yaḥyā ibn al-Mubārak al-Yazīdī (d. 202/818), ʿAbdallāh ibn Yaḥyā ibn al-Mubārak al-Yazīdī (d. 237/851), Muḥammad ibn Qādim (d. after 251/866), ʾAbū Mūsā al-Ḥāmiḍ (d. 305/918), Muḥammad ibn alʿAbbās al-Yazīdī (d. 310/922), al-Zajjāj (d. 311/923), Ibn Šuqayr (d. 317/929), and Muḥammad ibn ʿUṯmān al-Jaʿd (d. after 320/932).13 – Muḫtaṣar al-naḥw, by al-Kisāʾī (d. 189/805), Ibn Kaysān (d. 320/932;14 see also below), and Muḥammad ibn ʿAlī al-Marāġī (a student of al-Zajjāj’s).15 – Muḫtaṣar fī l-naḥw, by al-Waššāʾ (d. 325/937) and Muḥammad ibn ʾAbī Ġassān al-Bakrī (d.?).16 – al-Muḫtaṣar fī l-naḥw, by Muḥammad ibn Saʿdān al-Ḍarīr (d. 231/846), Luġda al-ʾAṣbahānī (d. 310/922; see also below), and ʾAḥmad ibn Muḥammad alMuhallabī (d.?).17

10 11 12 13 14 15 16 17

Ibn al-Nadīm, Fihrist I, 162, 260. Ibn al-Nadīm, Fihrist I, 267. Ibn al-Nadīm, Fihrist I, 211. Ibn al-Nadīm, Fihrist I, 140, 141, 202, 240, 141, 177, 255, 252 respectively. Another date for Ibn Kaysān’s death in the sources is 299/911, but is considered by Yāqūt (Muʿjam V, 2309) to be wrong; cf. Suyūṭī, Buġya I, 19. Ibn al-Nadīm, Fihrist I, 196, 248, 265. Ibn al-Nadīm, Fihrist I, 263, 265. Ibn al-Nadīm, Fihrist I, 242, 249, 261. Note that the book by Ibn Saʿdān is also referred to as Muḫtaṣar al-naḥw; Ibn al- Nadīm, Fihrist I, 210.

64

baalbaki

Other titles mentioned by Ibn al-Nadīm are also very likely to have been pedagogical grammars for beginners. These include, in chronological order, alMubarrad’s (d. 285/898) al-Madḫal fī l-naḥw; al-Mufaḍdal ibn Salama’s (d. 290/ 903 or 300/913) al-Madḫal ʾilā ʿilm al-naḥw; Ṯaʿlab’s (d. 291/904) al-Muwaffaqī, described by Ibn al-Nadīm as muḫtaṣar; Ibn al-Sarrāj’s (d. 316/929) al-Mūjaz; Ibn al-Ḫayyāṭ’s (d. after 320/932) al-Mūjaz; ʾAbū Bakr ibn al-ʾAnbārī’s (d. 328/ 940) al-Wāḍiḥ fī l-naḥw; al-Karmānī’s (d. 329/941) al-Mūjaz fī l-naḥw; Ibn Ḫālawayhi’s (d. 370/980) al-Mubtadaʾ; and al-Rummānī’s (d. 384/994) al-ʾĪjāz fī l-naḥw and al-Mubtadaʾ fī l-naḥw.18 Only few of the titles cited above have reached us, and these strongly support the view that the works cited above were indeed pedagogical manuals. One of the muḫtaṣarāt we possess is the one by Luġda, and it is obviously a pedagogical manual. The same is true of Ibn Kaysān’s book (if indeed it is the text published under the title al-Muwaffaqī fī l-naḥw) and of Ibn al-Sarrāj’s al-Mūjaz, both of which have survived. Other pedagogical grammars of the same period that have been published include (i) Muqaddima fī l-naḥw attributed to Ḫalaf al-ʾAḥmar (d. 180/796), although its text contains clues which suggest that it was authored in a period considerably later than Sībawayhi’s Kitāb;19 (ii) al-Jumal fī l-naḥw, authored by Ibn Šuqayr (d. 317/929) and known as al-Muḥallā or Wujūh al-naṣb, but erroneously attributed to al-Ḫalīl ibn ʾAḥmad (d. 175/791); (iii) Talqīn almutaʿallim min al-naḥw erroneously attributed to Ibn Qutayba (d. 276/889), but of a much later date, possibly the 10th/16th century;20 (iv) al-Jumal by al-Zajjājī (d. 337/ 949); (v) al-Tuffāḥa fī l-naḥw by ʾAbū Jaʿfar al-Naḥḥās (d. 338/950); (vi) al-Wāḍiḥ by al-Zubaydī (d. 379/989); and (vii) al-Lumaʿ fī l-ʿArabiyya by Ibn Jinnī (d. 392/1002). The common feature among these books is that they are simplified presentations of the entire fields of syntax and morphology (and in some cases phonology), seldom quote šawāhid (attested material) from poetry or other linguistic sources, and almost fully exclude grammatical controversies. One would expect the other works which were mentioned earlier and which did not reach us to have been of a similar nature in content and method.

18

19

20

Ibn al-Nadīm, Fihrist I, 171, 224, 226, 182, 249, 230, 243, 259, 188 respectively. As noted in the next paragraph above, the text published under the title al-Muwaffaqī fī l-naḥw is by Ibn Kaysān. Cf. Baalbaki (2008:29). Note that Talmon (1990:155f.) argues that although the book is not Ḫalaf’s, its author is a contemporary of Sībawayhi (d. 180/796), al-Farrāʾ (d. 207/822) and ʾAbū ʿUbayda’s (d. 209/824). al-Muqaddima fī l-naḥw is also the title of a book by the famous lexicographer al-Jawharī (d. c. 400/1010), as we learn from Yāqūt, Muʿjam II, 657 and other sources mentioned in the editor’s footnotes. Cf. Carter (1979:267–273).

grammar for beginners

65

Yet caution should be exercized against generalization. A case in point is ʾAbū ʿAlī al-Fārisī’s (d. 377/987) al-ʾĪḍāḥ. The book is described by Ibn al-Ṭarāwa (d. 528/1134) as one of the muḫtaṣarāt,21 and ʿAḍud al-Dawla (d. 372/983), to whom al-Fārisī dedicated it, characterizes it as suitable for beginners ( yaṣluḥ li-l-ṣibyān).22 It is also widely held in modern scholarship that the book is a pedagogical grammar for beginners.23 In all likelihood, the book is indeed designed for beginners and there are indications that al-Fārisī authored it especially for the benefit of ʿAḍud al-Dawla’s nephews.24 This notwithstanding, a closer examination of the book reveals that it differs significantly from the other pedagogical manuals in our possession, and thus, in comparison, its appropriateness for beginners who have little or no prior knowledge of grammar is questionable. Most notable is that it untypically contains about 130 Qurʾānic šawāhid and eight from poetry, as well as references to partisan views.25 Furthermore, one of the main arguments adduced by Ibn al-Ṭarāwa in his critique of the book is that it is too complicated for beginners.26 He even goes as far as contrasting the difficulty that a beginner would encounter in al-Fārisī’s approach—which places complex material before the essentials—to the suitability of the Kitāb to the needs of the beginner given that it proceeds gradually from the simple to the more complex.27 Although this comparison is exaggerated, the fact remains that Ibn al-Ṭarāwa wanted to highlight the unsuitability of the ʾĪḍāḥ for beginners—unlike al-Zajjājī’s al-Jumal, which he repeatedly praises from this perspective.28

21 22 23 24 25 26

27 28

Ibn al-Ṭarāwa, ʾIfṣāḥ 16. Ibn al-ʾAnbārī, Nuzha 233; Yāqūt, Muʿjam II, 813; Suyūṭī, Buġya I, 496. Cf. Šalabī (1958:516); Carter (1990:131); Baḥr al-Marjān’s introduction to Fārisī, ʾĪḍāḥ 22 f. Zubaydī, Ṭabaqāt 120; Šalabī (1958:517); Baḥr al-Marjān’s introduction to Fārisī, ʾĪḍāḥ 22. As in references to the Baṣriyyūn; cf. Fārisī, ʾĪḍāḥ 117, 185, 252. Cf. statements such as yanbū ḏihn al-mubtadiʾ ʿanhu ‘The intellect of the beginner cannot grasp it’, istiṭālatan ʿalā l-mubtadiʾīn ‘in order to exalt himself above the beginners’, and fa-ʾiḏā ṭālaʿahu l-mubtadiʾ ʾaḏhalahu ḏālika wa-hālahu wa-ʾin ḥāwala tafahhum šayʾ minhu ʾaʿjazahu wa-ṭālahu ‘When the beginner reads it, he will be confused and horrified, and if he tries to understand some of it, it will render him helpless and incapable’; Ibn al-Ṭarāwa, ʾIfṣāḥ 37, 69, 91. On one occasion he says that were men and jinn to band together and back up one another, they would not understand the author’s expression (ʾIfṣāḥ 32). See, in particular, Ibn al-Ṭarāwa, ʾIfṣāḥ 37; also 31, 34. Ibn al-Ṭarāwa, ʾIfṣāḥ 35, 75, 81.

66 2

baalbaki

Ibn Hišām on Teaching Grammar to Beginners and the ṣināʿa of ʾiʿrāb

Many of the pedagogical grammars that were mentioned above and that were authored before the end of the 4th/10th century continued to be used as teaching manuals for several centuries, as were the numerous commentaries based on some of them. A cursory look, for example, at the long lists of commentaries on and abridgments of al-Zajjājī’s al-Jumal and Ibn Jinnī’s al-Lumaʿ readily demonstrates that many of these works (abridgments in particular) were designed for beginners.29 Following the wide circulation of Ibn Mālik’s (d. 672/1274) ʾAlfiyya, which obviously was meant to be memorized by students, the vast number of commentaries based on it—in particular, Ibn ʿAqīl’s (d. 769/1367)—became the most widespread manuals in grammar teaching. In certain traditional circles, the ʾAlfiyya and its commentaries are still taught at a very early stage of education. Since what primarily concerns us in the rest of this paper is the issue of ʾiʿrāb, it is appropriate here to mention that the text of the ʾAlfiyya was fully subjected to ʾiʿrāb and used for training students in this skill, whereby the case endings are justified and the grammatical functions of words (mufradāt) and sentences ( jumal), as well as the underlying structure, are elucidated. In the ʾAlfiyya tradition, this can be exemplified by Ḫālid alʾAzharī’s (d. 905/1499) Tamrīn al-ṭullāb fī ṣināʿat al-ʾiʿrāb, a word for word ʾiʿrāb of the whole ʾAlfiyya. The centrality of ʾiʿrāb to the grammatical tradition is very well documented by the fact that the term itself became synonymous with naḥw at a relatively early stage of the tradition.30 This is visible in the titles of several works that deal with grammar in general, and not specifically with ʾiʿrāb, but whose titles contain the latter term, rather than naḥw. One of the earliest works of this type is Sirr ṣināʿat al-ʾiʿrāb, in each chapter of which Ibn Jinnī (d. 392/1002) discusses the syntactic characteristics of one of the letters of the alphabet, but only following the discussion of its phonetic and morphological aspects, which obviously do not belong to the realm of ʾiʿrāb. An example from the 5th/11th century is Šarḥ ʿUyūn al-ʾiʿrāb, a commentary on al-Fazārī’s (d. c. 350/961) ʿUyūn alʾiʿrāb, authored by al-Mujāšiʿī (d. 479/1086) and intended for use as a manual for beginners. From later centuries, we can cite al-Zamaḫšarī’s (d. 538/1144) famous al-Mufaṣṣal, whose full title, as given by its author, is al-Mufaṣṣal fī ṣanʿat al29

30

Sezgin (1984:88–94, 174–176). Note that some commentaries are too lengthy and complicated to be of any use for beginners; e.g. Ibn Ḫarūf’s (d. 605–610/1209–1213) Šarḥ Jumal al-Zajjājī and Ibn ʿUṣfūr’s (d. 669/1271) book with the same title. Cf. Zajjājī, ʾIḍāḥ 91: wa-yusammā l-naḥw ʾiʿrāb.

grammar for beginners

67

ʾiʿrāb31 and al-ʾAsfarāyīnī’s (d. 684/1285) al-Lubāb fī ʿilm al-ʾiʿrāb—both of which are general grammars in which ʾiʿrāb is one among several components. At the level of poetry material, many later authors provide detailed discussion of the ʾiʿrāb of its šawāhid, as does al-ʿAynī (d. 855/1451), whose al-Maqāṣid alnaḥwiyya is based on the šawāhid cited in a number of commentaries on the ʾAlfiyya. Turning to Ibn Hišām’s (d. 761/1360) Muġnī l-labīb, its first part, i.e. Chapter One, belongs to the rich tradition of works that deal with ḥurūf al-maʿānī (i.e. letters that have a meaning; particles), a term that covers a wide variety of operants and particles and is differentiated from ḥurūf al-hijāʾ (letters of the alphabet), which have no syntactical significance.32 Some of these works are devoted to a specific letter, most notably hamza and lām, whereas others deal with a variety of ḥurūf al-maʿānī and straddle the phonetic, morphological and syntactical levels.33 Compared with the latter group of sources, the first part of the Muġnī is more detailed than most and deals exclusively with the syntactical aspects of the particles under discussion. The lemmata have clear subdivisions and abound with šawāhid. The arrangement of the particles in alphabetical order (although by first letter only) provides easy access to the material, and this part of the book thus belongs to the onomasiological (or mubawwab) lexical tradition. In his introduction, Ibn Hišām says that his book is for the use of students (ṭullāb),34 but by these he obviously has in mind those who have attained a certain level of sophistication that would allow them to follow some of the more intricate issues which he tackles. This notwithstanding, the introductory parts of most lemmata, which do not normally include the difficult šawāhid and complex argumentation related to controversial issues, may well be useful for beginners. Similarly useful for them is another related work by Ibn Hišām, al-ʾIʿrāb ʿan qawāʿid al-ʾiʿrāb, which he authored prior to al-Muġnī, and whose sections that are devoted to particles are simple and concise enough to be used by beginners.35

31 32

33 34 35

Zamaḫšarī, Mufaṣṣal 5. Note that the book was published under the title al-Mufaṣṣal fī ʿilm al-ʿArabiyya. The meanings of ḥurūf al-hijāʾ (i.e. the words ʾalif, bāʾ, tāʾ, ṯāʾ, jīm, dāl, etc.) are listed in Kitāb al-ḥurūf, which is attributed to al-Ḫalīl ibn ʾAḥmad, but whose real author is unknown. For example, ʾalif means ‘a wretched, weak man’, ḏāl ‘a rooster’s crest’, ḍād ‘a hoopoe’ etc. (pp. 24, 37, 40). For a detailed study of both types, see Baalbaki (2014:213–225). Ibn Hišām, Muġnī I, 9–10. For example, the lemma ʾan occupies 25 lines in ʾIʿrāb 79–81, whereas it is almost ten times that length in Muġnī I, 27–36. For a detailed study of the ʾIʿrāb, see Nīl (1985:15–46).

68

baalbaki

Unlike the first part of the Muġnī, which belongs to the established tradition of works dealing with ḥurūf al-maʿānī, some of the material in its second part seems to be unprecedented in the grammatical tradition as a whole. Due to the apparent overlap between the Muġnī and the ʾIʿrāb, it may be useful to identify those parts of the latter that correspond to the second part of the former and then examine them together. The longest of the chapters of the ʾIʿrāb, i.e. the third, consists of eight types of particles exemplified by twenty of the ḥurūf al-maʿānī. This chapter thus corresponds to the first part of the Muġnī, but represents no more than a fraction of its content. In the other three chapters, Ibn Hišām focuses on (i) the sentence ( jumla) and specifies the sentences that have or do not have a maḥall lit. ‘position’ in ʾiʿrāb, that is, as specific grammatical function (Chapter One); (ii) prepositions and their genitives (Chapter Two); and (iii) certain rules to be observed in ʾiʿrāb (Chapter Four). All three chapters have parallels in the Muġnī, most of whose material is manifestly of a more advanced level and a wider scope. In fact, Ibn Hišām himself in the introduction of the Muġnī mentions the fact that what prompted him to author this work was that his earlier work, the ʾIʿrāb, was very well received among students of grammar ( jamāʿat al-ṭullāb) for its usefulness. Furthermore, he compares the concise material of his earlier book and the vastness of the Muġnī to a mere drop in an ocean.36 In the second part of the Muġnī, the second and third chapters expand on the material in Chapters One and Two of the ʾIʿrāb respectively. The fourth chapter deals with frequently encountered syntactical rules (e.g. the distinction between subject and predicate and the types of the circumstantial accusative), whereas the eighth lists a number of general rules that embrace a countless number of subsidiary rules (mā lā yanḥaṣir min al-ṣuwar al-juzʾiyya), e.g. the extension of a rule that pertains to a certain element of the construction to what resembles that element or to a neighboring element. The three remaining chapters, i.e. the fifth, sixth and seventh, deal exclusively with ʾiʿrāb from various perspectives. Given that the second chapter, which discusses the jumla, is essentially a study of ʾiʿrāb as applied to sentences, and that other chapters often discuss matters related to ʾiʿrāb (indeed, the bulk of the first chapter, i.e. the lexical part of the book, abounds with syntactical material related to ʾiʿrāb), the Muġnī, as a whole, can be characterized as a work that primarily studies ʾiʿrāb. Yet, what sets the book apart from previous works are those parts which deal with ʾiʿrāb as a technique that should be mastered by students and

36

Ibn Hišām, Muġnī I, 9.

grammar for beginners

69

that is subject to clear-cut criteria which are expressed in set formulae. It is specifically in these parts that Ibn Hišām’s contribution can be best appreciated as they are most likely to be unprecedented in the history of the tradition as far as our knowledge goes. With respect to the second chapter, which applies ʾiʿrāb to the jumla, its material occurs in earlier sources and is in many cases traceable to the general rules described, say, in the Kitāb. What is innovative, however, is the organization of these rules into two distinct categories—i.e. the jumal that have a maḥall and those that do not—each of which comprises seven types.37 The discussion under a single heading of these fourteen types, as well as the division of the jumal according to various criteria (e.g. ismiyya, fiʿliyya and ẓarfiyya based on whether its initial element is a noun, a verb or an adverb, and kubrā and ṣuġrā based on whether its predicate is a jumla or not) makes this chapter one of the earliest of its kind in the grammatical tradition.38 The contents of this chapter and its organization, both in the Muġnī and the ʾIʿrāb, were largely adopted by subsequent authors, including the numerous commentaries on both works.39 On a contemporary note, students who have been trained in the techniques of ʾiʿrāb at high school level will readily realize if they read Ibn Hišām’s chapter in either of his two books that it is almost identical to what they were taught. Accordingly, it can be safely stated that Ibn Hišām’s study of the jumla has left its mark on grammar teaching for several centuries. Equally unparalleled in the tradition are the fifth, sixth and seventh chapters of the Muġnī, all of which discuss matters related to ʾiʿrāb. But whereas the material of the fifth chapter is for the benefit of those who are well accomplished in the knowledge of ʾiʿrāb, the material of the other two chapters, as Ibn Hišām himself notes, is related to widely spread errors of ʾiʿrāb committed by the mubtadiʾūn (beginners), also referred to as mutadarribūn (trainees) and mutaʿallimūn (apprentices).40 This notwithstanding, the three chapters are best analyzed together since they reveal much about the basics and techniques of ʾiʿrāb and how they were practiced and taught in Ibn Hišām’s time. We propose to highlight three issues in the discussion that follows, namely, the

37 38

39 40

Ibn Hišām, Muġnī II, 382–428. Ibn Hišām, Muġnī II, 376–382. Among Ibn Hišām’s contemporaries who discuss these two types of jumal are ʾAbū Ḥayyān (d. 745/1344) in Manhaj 217–220 and Taḏyīl IX, 202–204, and al-Murādī (d. 749/1348), who wrote a short treatise on the subject, a description of which is found in Tawḍīḥ, Introduction, 111–113. I am grateful to Almog Kasher for kindly drawing my attention to these works. See, for example, the long list of commentaries on the ʾIʿrāb prepared by Nīl (1985:26–43). Ibn Hišām, Muġnī II, 664, 651, 653 respectively.

70

baalbaki

concept of the ṣināʿa in ʾiʿrāb; the level of proficiency among students and masters alike; and the comparison between some contemporary ʾiʿrāb practices and views expressed by Ibn Hišām. 2.1 The Concept of the ṣināʿa in ʾiʿrāb Ibn Hišām goes to great length in order to demonstrate that proper ʾiʿrāb requires mastery of a craft (ṣināʿa)41 that embraces distinct rules (qawāʿid) which should be expressed by conventional phrases. It is quite telling that he uses the term qawāʿid al-ʾiʿrāb in the title of his book which shares most of its material with the Muġnī, i.e. al-ʾIʿrāb ʿan qawāʿid al-ʾiʿrāb.42 The rules pertaining to ʾiʿrāb are spelled out in various parts of the ʾIʿrāb and of chapters five, six and seven in the Muġnī and can be divided into two major types, namely, those which relate to the expressions that should be used in ʾiʿrāb, and those which identify the objections that can be raised against certain ʾiʿrāb practices. Although both types are obviously conducive to a sināʿa of ʾiʿrāb and thus complement each other, we shall discuss them separately for the sake of clarity. 2.1.1 Rules Relating to the Expressions That Should Be Used in ʾiʿrāb Ibn Hišām frequently refers to the expression (ʿibāra) that should be used by the one who practices ʾiʿrāb (i.e. the muʿrib) in each case. For example, the expression used in the ʾiʿrāb of the subject of a passive verb should be nāʾib ʿan al-fāʿil and not mafʿūl mā lam yusamma fāʿiluhu, and the correct expressions in the ʾiʿrāb of lan and fāʾ are determined as ḥarf naṣb wa-nafy wa-stiqbāl and ḥarf ʿaṭf li-l-tartīb wa-l-taʿqīb respectively.43 Many of these ʾiʿrāb expressions are still taught to students throughout the Arab world almost exactly as reported by Ibn Hišām. This includes several expressions which he does not approve of, as we shall see in 3 below. Ibn Hišām’s insistence on what he considers to be the exact wording of ʾiʿrāb expressions is understandable given that such fixed expressions are the essential building blocks in the establishment of a solid ṣināʿa that is widely taught to beginners. Furthermore, the perfection of the ṣināʿa necessitates the establishment of certain conditions in formulating ʾiʿrāb expressions. Accordingly, Ibn Hišām highlights the importance of two such conditions that 41 42

43

See the term in Muġnī II, 527, 539, 649; ʾIʿrāb 107. See also the introduction of the ʾIʿrāb 31, where Ibn Hišām specifically mentions that his book deals with a number of issues related to qawāʿid al-ʾiʿrāb. It is noteworthy that, throughout the Arab world today, students often refer to naḥw itself as qawāʿid and reduce the notion of naḥw to training in ʾiʿrāb. This shift in terms might have originated from the expression qawāʿid al-ʾiʿrāb and is yet another proof of the centrality of ʾiʿrāb in the traditional study of grammar. Ibn Hišām, Muġnī I, 284; II, 664; ʾIʿrāb 105f.

grammar for beginners

71

he considers to be essential for any ʾiʿrāb expression to become an accepted formula, namely that it should be concise (mūjaza) and inclusive of the main points that deserve mention (mustawfāt).44 As far as conciseness (ʾījāz) is concerned, Ibn Hišām proposes brief substitutes to certain expressions that seem to have been in circulation. Examples include his preference for ẓarf mustaqbal to ẓarf li-mā yustaqbal min al-zamān and for fiʿl māḍī lam yusamma fāʿiluhu to fiʿl māḍī mabnī li-mā lam yusamma fāʿiluhu in reference respectively to ʾiḏā and the passive perfect verb.45 The pedagogical reason he cites for justifying the condition of conciseness is the frequent use of the ʾiʿrāb expression, which the mutadarribūn should find easy to utter (taḫiff ʿalā l-ʾalsina).46 The concise expressions which take the needs of the mutadarribūn into account become fixed formulae that are uniformly used by the whole community of the ṣināʿa (or what we can refer to as ʾahl al-ṣināʿa), beginners and scholars alike. Conciseness, however, does not by itself ensure the correctness of the ʾiʿrāb expression, which also has to include the essential elements that describe the function of the word or sentence at hand. Many examples of the expressions that are considered to fulfill this condition (hence described as mustawfāt) occur in the case of particles, such as lam (ḥarf jazm li-nafy al-muḍāriʿ waqalbihi māḍiyan), ʾammā (ḥarf šarṭ wa-tafṣīl wa-tawkīd) and ṯumma (ḥarf ʿaṭf li-l-tartīb wa-l-muhla), and are consistently used by Ibn Hišām.47 The concept of istīfāʾ also applies to grammatical functions that are necessarily linked to other parts of the utterance, even if they are elided. In the case of the verb, for example, the muʿrib has to specify its subject, and if that subject is elided, it has to be mentioned in order for the ʾiʿrāb expression to be complete.48 Similarly, the predicate has to be mentioned, even when elided. Other required elements that should be mentioned include the part of the utterance upon which an adverb or a preposition is dependent (mutaʿalliq), and the person that refers to a preceding relative pronoun (ʿāʾid).49 A related type of ʾiʿrāb expressions that lack a required element is that in which is mentioned the category of the word, but not its grammatical function or maḥall. This would be true if in the ʾiʿrāb of the relative pronoun allaḏī the muʿrib settles for ism mawṣūl without specifying its maḥall, e.g. fāʿil (agent).50 Accordingly, and in order to 44 45 46 47 48 49 50

Note that both terms are used in the title of the fourth chapter of Ibn Hišām’s ʾIʿrāb 103. Ibn Hišām, Muġnī II, 651, 664; cf. ʾIʿrāb 105. Ibn Hišām, Muġnī II, 651. Ibn Hišām, ʾIʿrāb 105f.; cf. Muġnī I, 227, 56, 117 respectively. Ibn Hišām, ʾIʿrāb 107; Muġnī II, 672. The expression uniformly used in this case is wafāʿiluhu ḍamīr mustatir taqdīruhu … Ibn Hišām, ʾIʿrāb 107; Muġnī II, 667, 672. Ibn Hišām ʾIʿrāb 107; cf. Muġnī II, 666.

72

baalbaki

avoid the omission of the required elements of ʾiʿrāb, Ibn Ḥišām specifies what the beginner (mubtadiʾ) should be asked to include in his ʾiʿrāb. His examples appropriately deal with some of the most basic types of ʾiʿrāb expressions. In the case of verbs, for example, the beginner is required to specify (i) whether the verb under consideration is māḍī (past), muḍāriʿ (present) or ʾamr (imperative); (ii) whether it is mabnī (indeclinable) or muʿrab (declinable); (iii) its characteristic final vowel or lack thereof; and (iv) in the case of the muḍāriʿ, the reason for its rafʿ (indicative), naṣb (subjunctive) and jazm (jussive).51 Precision in formulating ʾiʿrāb expressions is shown to be necessary not only in analyzing difficult and often controversial cases of ʾiʿrāb (see next paragraph), but also as a necessary condition for the establishment of universal norms that apply to the most elementary stages of ʾiʿrāb. These norms obviously ought to be mastered by beginners and upheld by scholars, in adherence to a ṣināʿa that has set rules of expression, let alone its own axioms and analytical tools. 2.1.2

Rules Identifying the Objections That Can Be Raised against Certain ʾiʿrāb Practices The second aspect of the ṣināʿa of ʾiʿrāb that can be discerned in Ibn Hišām’s work relates to several types of error into which the incautious muʿrib might fall. In a very lengthy chapter of the Muġnī,52 Ibn Hišām identifies ten types of objection for which the muʿrib may have to answer. Each type is explained by the analysis of a large number of constructions (e.g. twenty-two in the first type, thirteen in the second, etc.). Unlike the material discussed in (i) above, the argumentation and methods of analysis in this aspect of the ṣināʿa can be beneficial to scholars or students who have attained a highly advanced level of mastery of ʾiʿrāb, and are certainly not addressed to beginners. The first two objections are probably the most essential since they clearly demonstrate the intricate relationship between the ṣināʿa and meaning (maʿnā)—an issue that is frequently faced in deciding the most appropriate ʾiʿrāb, particularly in those cases where the construction is tricky or requires analysis that goes beyond what may seem to be apparent or commonplace. The first objection is raised when the muʿrib takes into account what Ibn Hišām calls ẓāhir al-ṣināʿa,53 that is, the apparent or conventional. For example, in the verse Allāhu ʾaʿlamu ḥayṯu yajʿalu risālatahu ‘Allah knows best where to place His Message’ (Q. 6/124), the muʿrib may well be tempted to consider ḥayṯu to be an adverb as stipulated by the conventions of ʾiʿrāb. Yet a closer look at the verse 51 52 53

Ibn Hišām, Muġnī II, 667f. Chapter Five (II, 527–650). Chapter Five II, 527.

grammar for beginners

73

reveals that ḥayṯu is a direct object since the intended meaning (al-murād) is that God knows the place or locality that is worthy of His Message and not that His knowledge is in that place ( yaʿlam al-makān al-mustaḥiqq li-l-risāla lā ʾanna ʿilmahu fī l-makān).54 Conversely, an ʾiʿrāb can be objectionable if it only takes the meaning into account and does not comply with certain axioms of the ṣināʿa. This applies, for example, to those who consider Ṯamūdan in waʾannahu ʾahlaka ʿAdān l-ʾūlā wa-Ṯamūdan fa-mā ʾabqā ‘And that He destroyed the first ʿĀd and Ṯamūd, leaving nothing behind’ (Q. 53/50–51) to be a direct object of the transitive verb ʾabqā. Although this ʾiʿrāb is appropriate for the meaning of the construction, it fails to take into account the syntactical rule related to mā and other particles of negation, namely, precedence (li-ʾanna limā ḥaqq al-ṣadr). Hence, the verb after mā cannot govern a noun that precedes mā. Ibn Hišām concludes that the correct ʾiʿrāb—i.e. that which satisfies both maʿnā and ṣināʿa—is that Ṯamūd is either conjoined to ʿĀd (maʿṭūf ) or the direct object of an elided verb, assumed to be ʾahlaka, as in the preceding sentence.55 The other types are equally, if not more complex, than the first two and embrace a variety of syntactical issues and conventions of ʾiʿrāb that need to be assessed by the muʿrib in order to perfect his ʾiʿrāb. Accordingly, the muʿrib should neither adopt an ʾiʿrāb that is not well proven in Arabic (lam yaṯbut fī l-ʿArabiyya)—as in ʾAbū ʿUbayda’s (d. 209/824) claim that kāf is a jurative particle—nor opt for a farfetched ʾiʿrāb when a more straightforward and widely attested one is available—as in considering ʾahla l-bayti in ʾinnamā yurīdu l-Lāhu li-yuḏhiba ʿankum al-rijsa ʾahla l-bayti ‘Allah only seeks to remove abomination from you and purify you fully, O people of the House’ (Q. 33/33) to be an accusative of specification (iẖtiṣāṣ) and not vocative (munādā).56 The muʿrib is also expected to consider several possibilities of ʾiʿrāb and assess their strengths and weaknesses—as in the three different ways of interpreting the maḥall of the independent pronoun in ʾinnaka ʾanta l-samīʿu l-ʿalīmu ‘You are, indeed, the All-Hearing, the Omniscient’ (Q. 2/127).57 He also has to make sure that his ʾiʿrāb, which may seem correct in itself, is not contradicted by other occurrences that may prove it to be deficient—as in considering fīhi in ḏālika l-kitābu lā rayba fīhi hudan li-l-muttaqīna ‘This is the Book which is not to be doubted and is a guide to the God-fearing’ (Q. 2/2) to form a sentence with what follows it and not to belong to the preceding words. Such an ʾiʿrāb is con54 55 56 57

Chapter Five II, 531. Chapter Five II, 539. Chapter Five II, 546, 551. Chapter Five II, 556.

74

baalbaki

tradicted by another verse, tanzīlu l-kitābi lā rayba fīhi min rabbi l-ʿālamīna ‘The revelation of the Book from the Lord of the Worlds, wherein there is no doubt’ (Q. 32/2) and is therefore refutable.58 By devoting this lengthy chapter to the objections with which the muʿrib can be challenged and to the host of examples—mainly from Qurʾān and poetry— that justify what he considers to be the correct approach to ʾiʿrāb, Ibn Hišām demonstrates his profound knowledge of the many issues that need to be mastered in order to achieve perfection in ʾiʿrāb. In fact, his skills of syntactical analysis in the study of ʾiʿrāb and ability to analyze structure from several perspectives—both in this chapter and elsewhere—are hardly matched in the tradition and must have contributed to his wide reputation as a most skillful grammarian—witness Ibn Ḫaldūn’s (d. 808/1406) reported view that daringly describes him as having been more proficient in grammar (ʾanḥā) than Sībawayhi.59 Similar to the concept of poetry criticism (naqd al-šiʿr), the fifth chapter of the Muġnī is probably the most serious attempt we know of, which is aimed at the establishment of general principles that relate to a subdiscipline which we can call naqd al-ʾiʿrāb. Just like poetry and prose were referred to as al-ṣināʿatān—cf. ʾAbū Hilāl al-ʿAskarī’s (d. after 395/1005) Kitāb al-Ṣināʿatayn— ʾiʿrāb as a major component of naḥw can be safely referred to as a ṣināʿa given Ibn Hišām’s efforts to demonstrate that it is subject to specific rules of expression and universal norms of correctness and criticism. 2.2 The Level of Proficiency among Students and Masters Alike Much can be learnt from Ibn Hišām’s discussion of ʾiʿrāb issues about the level of linguistic proficiency in his time, not only among those whom he refers to as beginners (mubtadiʾūn) or students (ṭalaba), but also among those whom one would assume to have reached a more advanced level in mastering Arabic. Ibn Hišām clearly states in the outset of the seventh chapter of the Muġnī that it addresses the beginners, although it soon becomes clear that the term mubtadiʾūn can refer to students who begin their study with little or no knowledge of ʾiʿrāb as well as to students who are acquainted with its jargon and basic rules. In certain cases, mistakes cited by Ibn Hišām are most probably attributable to true beginners, as in the wrong assumption that ʾakramtu and taʿallamtu are imperfect verbs because they begin with ʾalif and tāʾ (i.e. used in the prefixed conjugation); that the wāw and fāʾ in waʿaẓa and fasaḫa are conjunctions; and that the bāʾ in bayt and the lām in laʿiba are prepositions.60 The same can also 58 59 60

Chapter Five II, 593. Ibn Ḥajar, Durar II, 309; Suyūṭī, Buġya II, 69. Ibn Hišām, Muġnī II, 668.

grammar for beginners

75

apply to the inability to differentiate between the -in that indicates the genitive and the -in (i.e. tanwīn) in defective nouns such as qāḍin and zānin, and to confusing the dual for plural (as in al-muṣṭafayn) and the genitive pronominal suffix for the accusative or vice versa (as in ġulāmī and ʾakramanī).61 There are, however, cases in which the reported mistakes seem to have been made by students of a more advanced level, given the context specified by Ibn Hišām. He reports, for example, that to illustrate the unusual occurrence of the nominal sentence without the circumstantial wāw, he quoted the verse tarā llaḏīna kaḏabū ʿalā l-Lāhi wujūhuhum muswaddatun ‘You will see those who uttered falsehood about Allah with their faces blackened’ (Q. 39/60), at which point one of the students objected by saying that the wāw is present in this example, obviously mistaking the wāw in wujūhuhum for a circumstantial wāw.62 Ibn Hišām also reports how some of those who are expected to be quite proficient in grammar do not master certain basic grammatical matters. A most telling example is that in which a senior jurist (rajul kabīr min al-fuqahāʾ) wondered why in a line of poetry that has ʾa-tabītu … wa-ʾabīta ‘Would you and would I spend the night’, the first verb ends with -u and the second with -a although the former is in the second person and the latter is in the first. Ibn Hišām had to explain to the one who reported this to him that the two vowels indicate the indicative and subjunctive respectively, whereas the prefix in each of the two verbs indicates person.63 It is noteworthy that although Ibn Hišām reports this anecdote on the authority of another person, most of his examples are obviously drawn from his own experience with students.64 It is well known that teaching grammar was among Ibn Hišām’s priorities,65 and it seems that this included the teaching of beginners who had little or no prior training in the discipline. Although Ibn Hišām does not cite any reason for what seems to be widespread weakness among students of grammar, it is clear that he was not satisfied with the incompetence of some of their teachers, including exegetes, as his reference to ḍuʿafāʾ al-muʿribīn wa-l-mufassirīn indicates.66 In certain cases, he

61 62 63 64

65 66

Ibn Hišām, Muġnī II, 670f. Ibn Hišām, Muġnī II, 669. Ibn Hišām, Muġnī II, 668. Note, for instance, that he refers to comments and questions raised by his students (e.g. fa-qāla baʿḍ man ḥaḍara; wa-qad saʾalanī baʿḍuhum ʿan ḏālika; Muġnī II, 669 f.) and to his own questions to them (e.g. wa-saʾaltu kaṯīran min al-ṭalaba ʿan ʾiʿrāb … and wa-saʾaltu ṭāliban; Mugnī II, 673f.). Ibn Ḥajar, Durar II, 308; Suyūṭī, Buġya II, 68. Ibn Hišām, Muġnī II, 653. In their introduction to Suyūṭī’s Tuḥfa, Malḫ and Naʿja (pp. 18 f.) note that Ibn Hišām, in spite of his status as a great scholar, did not disdain to supervise teachers who taught Arabic to beginners.

76

baalbaki

seeks excuses for mistakes made by teachers. One such excuse is taqrīb (simplification) and applies to their ʾiʿrāb of the apodosis ( jawāb al-šarṭ) in a sentence like iʾtinī ʾukrimka—which he interprets as conditional, i.e. ʾin taʾtinī ʾukrimka— as the complement of the imperative ( jawāb al-ʾamr; cf. 2.3(v) below). Similarly, he cites taqrīb as the reason for their ascription of the indicative (rafʿ) in the imperfect to the absence of an operant that governs the subjunctive or jussive (li-ḫuluwwihi min nāṣib wa-jāzim), instead of citing the correct reason, namely, the occurrence of the verb in the position (maḥall) of a noun (cf. 2.3(vi) below).67 In other cases, however, Ibn Hišām is less tolerant of those who resort to simplification without recourse to correctness. He is, for example, harshly critical of the practice of some teachers (muʿallimūn) who refer to the suffixed pronoun in ḍarabtu as ta, rather than al-tāʾ, in saying ta fāʿil.68 Since al-tāʾ in the construction al-tāʾ fāʿil is subject, and hence a noun, Ibn Hišām argues that it cannot be replaced by ta—a reference to the rules of noun morphology, which do not admit the occurrence of monoliterals as nouns. More grave errors occur when a teacher fails to come up with proper ʾiʿrāb because he misunderstands the meaning of the construction. One of the masters (mašāyiḫ) is reported to have explained to a pupil of his that naʿam in a line of poetry ending with ʾiḏ qāla l-ḫamīsu naʿam ‘when the army said naʿam’, in which naʿam ends with quiescence, is a particle for reply (ḥarf jawāb), whereas the intended meaning is ‘cattle’ (singular of ʾanʿām) and the correct ʾiʿrāb is that it is the predicate of an elided subject (ḫabar li-maḥḏūf ), hence hāḏihi naʿam(un).69 Another šayḫ is reported to have said to his pupil that qayyiman in al-ḥamdu li-l-Lāhi llaḏī ʾanzala ʿalā ʿabdihi l-kitāba wa-lam yajʿal lahu ʿiwajan / qayyiman ‘Praise be to Allah who revealed the Book to His servant and did not leave in it any crookedness’ (Q. 18/1–2) is an adjective of ʿiwajan, unaware of the fact that qayyim ‘straight’ has the exact opposite meaning of ʿiwaj ‘crookedness’70 It is noteworthy that Ibn Hišām’s complaint about the poor level of Arabic among students and teachers is reminiscent of the views of many of his predecessors. Other than dissatisfaction—expressed by some lexicographers, e.g. Ibn Durayd (d. 321/933), and authors of books on solecism (laḥn), e.g. alZubaydī (d. 379/989)71—with the general ignorance of their contemporaries,

67 68 69 70

71

Ibn Hišām, Muġnī II, 653. Ibn Hišām, Muġnī II, 665. Ibn Hišām, Muġnī II, 528. Ibn Hišām, Muġnī II, 534. Ibn Hišām further explains that qayyiman in this construction is a circumstantial accusative, which, among other possibilities, refers to Allāh, hence ʾanzalahu qayyiman, or to the Holy Book, hence ʾanzala l-kitāba qayyiman. Ibn Durayd, Jamhara I, 39–40; Zubaydī, Laḥn 8.

grammar for beginners

77

their lack of interest in seeking knowledge, and the need of authors to simplify their material and its arrangement in response to linguistic incompetence, several authors sharply attacked their predecessors and contemporaries and accused them of being unreliable and incompetent.72 Ibn al-ʾAṯīr (d. 606/1210) more specifically complained of the inability of most students of ġarīb al-Ḥadīṯ (whom one would suppose to be highly competent in linguistic sciences) to differentiate between radicals that are part of the root and radicals that are affixed to it.73 For his part, Ibn Manẓūr (d. 711/1311) lamented the fact that linguistic errors had become prevalent and that his contemporaries regarded speaking Arabic as disgraceful (min al-maʿāyib) and took pride in using other languages.74 Yet whereas most earlier authors expressed merely in brief comments their discontent with the poor level of general linguistic proficiency among students, Ibn Hišām focused on the issue of weakness in ʾiʿrāb and dealt at length with specific errors that were widespread among students. By drawing on his own teaching experience, he accurately identified errors of ʾiʿrāb and set clear criteria that assist the muʿrib not only in understanding why a certain ʿiʿrāb is flawed but also in applying the necessary criteria (e.g. soundness of maʿnā, adhering to the ṣināʿa, comparison between similar constructions, etc.) that ensure the correct ʾiʿrāb of words and sentences. 2.3

The Comparison between Some Contemporary ʾiʿrāb Practices and Views Expressed by Ibn Hišām Reference was made in 2.1.1 above to the expressions that are used by the muʿrib and which ibn Hišām regards as set formulae that should be closely adhered to. As pointed out previously, many of these expressions are still used verbatim in the teaching of grammar throughout the Arab world. This notwithstanding, Ibn Hišām quotes several widely used ʾiʿrāb expressions which he deems to be incorrect. In fact, he devotes the sixth chapter of the Muġnī to cautioning the reader against generally accepted formulae which he considers to be erroneous.75 He notes that mistakes of this type are abundant (kaṯīra) and cites twenty cases by way of illustration. Some of these illustrations, however, pertain to grammatical issues that are related to ʾiʿrāb but are not, strictly speaking,

72

73 74 75

See, for example, the harsh criticism by both al-ʾAzharī (d. 370/981) and Ibn Sīda (d. 458/ 1066) of earlier and contemporary scholars in the introductions of Tahḏīb al-luġa (I, 30–31, 40) and al-Muḥkam (I, 38–41, 49) respectively. Ibn al-ʾAṯīr, Nihāya I, 11. Ibn Manẓūr, Lisān I, 8. The title of this chapter is Fī l-taḥḏīr min ʾumūr uštuhirat bayna l-muʿribīn wa-l-ṣawāb ḫilāfuhā; Muġnī II, 650.

78

baalbaki

formulae used in ʿiʿrāb. In the third example in this chapter, for example, Ibn Hišām cites the expression “The attribute follows the substantive in four out of ten [things]” (al-naʿt yatbaʿ al-manʿūt fī ʾarbaʿa min ʿašara).76 Although the muʿrib is usually asked when he encounters a naʿt to say that it follows the manʿūt, the expression cited by Ibn Hišām is not used in ʾiʿrāb, but appears in the chapter on naʿt in grammar textbooks.77 Similarly, the fourth example is the ʾiʿrāb of raġadan in fa-kulā minhā raġadan ‘and eat from it to your heart’s content’ (Q. 2/35), which some claim is an adjective whose substantive is an elided verbal noun (naʿt maṣdar maḥḏūf ) but which Ibn Hišām determines is a circumstantial accusative.78 Accordingly, he does not object to the expression itself, but to its use in the ʾiʿrāb of this specific word within this context. The same is true of the sixth and the tenth to the eighteenth examples. The remaining eight examples are quite interesting because they contain expressions that are still widely used in the same wording which Ibn Hišām finds fault with. This conclusion is not based solely on my personal experience during my training in ʾiʿrāb in secondary school as well as my undergraduate study as a student majoring in Arabic language, let alone my teaching experience of ʾiʿrāb for about four decades, but more objectively on a representative sample of some of the most well-known contemporary reference works in ʾiʿrāb. Arranged on the basis of their date of publication, these are: ʾAsmar (1969), ʾAltōnjī (1974), Qabāwa (1978), Rājiḥī (1979), Ḥusayn (1981), ʿĀṣī and Yaʿqūb (1987), Labadī (1988), Yaʿqūb (1988), ʿAbd al-Masīḥ and Tābirī (1990), ʾAltōnjī and ʾAsmar (1993), and Sinnū et al. (2010). We shall briefly discuss each of the eight cases79 that we have identified in the Muġnī and demonstrate that the expressions criticized by Ibn Hišām (and quoted in the beginning of each of the next eight paragraphs) are still employed in our eleven reference works.80 (i) That law is ḥarf imtināʿ li-mtināʿ (a particle denoting prevention [of the apodosis] due to prevention [of the protasis]). In the alphabetically

76 77 78 79 80

Ibn Hišām, Muġnī II, 651. See, for example, commentaries on the ʾAlfiyya in the chapter on naʿt; e.g. Ibn al-Nāẓim, Šarḥ 490–493; Ibn ʿAqīl, Šarḥ 399–401. Ibn Hišām, Muġnī II, 652. All eight cases occur in Chapter six of the Muġnī II, 650–664, in the following order: first, second, fifth, seventh, eighth, ninth, nineteenth and twentieth. It should be noted that some of the eleven works cited above, in particular Sinnū et al. (2010), quote several grammatical sources and often cite conflicting ʾiʿrāb expressions and views regarding the same particle or phenomenon, including those adopted by Ibn Hišām. Reference will be made to the occurrence of expressions with which Ibn Hisām disagrees. For the relevance of the Arabic grammatical tradition to modern language teaching—a subject beyond the scope of the present paper—cf. Owens (2005) and Peled (2010).

grammar for beginners

79

arranged part of the Muġnī, under the lemma law, Ibn Hišām describes this expression as prevalent among the muʿribūn and as having been stipulated (naṣṣa ʿalayhi) by a group of grammarians, i.e. as the correct ʾiʿrāb. In a lengthy discussion,81 he adduces various arguments to prove that this expression is wrong and contradicts the correct meaning of several attested constructions. He also proposes alternatives to it based on statements by Sībawayhi (d. 180/796) and Ibn Mālik (d. 672/1274). Yet the expression ḥarf imtināʿ li-mtināʿ is standard in contemporary teaching and in most of our eleven reference works.82 (ii) That ʾiḏā is a ẓarf li-mā yustaqbal min al-zamān wa-fīhā maʿnā l-šarṭ ġāliban (an adverbial denoting anticipated time and mostly indicates the conditional). Ibn Hišām finds fault with this expression from several perspectives. For example, he objects to its use irrespective of whether the construction at hand does indicate the conditional or not. Based on the principle of conciseness discussed in 2.1.1 above, he also objects to its wordiness since li-mā yustaqbal min al-zamān could be easily replaced by one word, mustaqbal. As in (i) above, the reference works consulted preserve the expression criticized by Ibn Hišām verbatim or almost so.83 (iii) That the conditional fāʾ, as in our example ʾin zurtanī fa-ʾanā ʾukrimuka, is jawāb al-šarṭ (apodosis), whereas the correct expression is rābiṭat jawāb al-sarṭ (binder of the apodosis). In other words, the apodosis is what follows the fāʾ, whose function is to bind the apodosis to the protasis.84 A similar expression to the one which Ibn Hišām finds fault with, but which he does not mention, is wāqiʿa fī jawāb al-šarṭ, which effectively means that the fāʾ is part of the apodosis. This latter expression is used in some of our contemporary works,85 although rābiṭat jawāb al-šarṭ is prevalent.

81 82

83

84

85

Ibn Hišām, Muġnī I, 257–260; cf. ʾIʿrāb 83–85. ʾAsmar (1969:93); ʾAltōnjī (1974:138); Rājiḥī (1979:66); Ḥusayn (1981:171); ʿĀṣī and Yaʿqūb (1987: II, 1087); Yaʿqūb (1988:409); ʿAbd al-Masīḥ and Tābirī (1990:349); ʾAltōnjī and ʾAsmar (1993: I, 532). ʾAsmar (1969:9); ʾAltōnjī (1974:18); Qabāwa (1978:57, 81, 82, etc.); Rājiḥī (1979:247); Ḥusayn (1981:16); ʿĀṣī and Yaʿqūb (1987: I, 69); Yaʿqūb (1988:78); ʿAbd al-Masīḥ and Tābirī (1990:42); cf. ʾAltōnjī and ʾAsmar (1993: I, 25); Sinnū et al. (2010: I, 309). Note that Ibn Hišām’s use of the word rābiṭa (Muġnī II, 652) and his view that the sentence ( jumla) is the apodosis strongly indicate that the fāʾ is not part of the apodosis. In the ʾIʿrāb (p. 105), however, he determines the apodosis to be the whole sentence and not only the fāʾ and thus apparently considers the fāʾ to be part of the apodosis although he still refers to it as rābiṭa. Cf. the use of taqaʿ in ʾAltōnjī (1974:101) and wāqiʿa in Rājiḥī (1979:348, 440).

80

baalbaki

(iv) That bal is a ḥarf ʾiḍrāb (a particle denoting digression or retraction). Ibn Hišām considers this to be defective and insists that the correct ʾiʿrāb is ḥarf istidrāk wa-ʾiḍrāb, which takes into account its function of istidrāk (rectification) since it is equivalent to lākin after negation and prohibition. Most of the reference works use ḥarf ʾiḍrāb in the ʾiʿrāb of bal to the exclusion of the notion of istidrāk.86 (v) That the jussive in the verb which follows the imperative is due to its being the complement of that imperative ( jawāb al-ʾamr), as in iʾtinī ʾukrimka. As noted in 2.2 above, Ibn Hisām argues that the underlying structure of this construction is ʾin taʾtinī ʾukrimka and thus ʾukrimka is in the jussive because it is the apodosis ( jawāb al-šarṭ). The terms jawāb al-ʾamr and jawāb al-ṭalab are still widely used;87 they are interchangeable and often appear side by side.88 As Ibn Hišām himself notes, those who substitute the correct expression jawāb al-šarṭ by jawāb al-ṭalab probably do so for the sake of simplifying the matter to students (taqrīb al-masāfa ʿalā l-mutaʿallimīn). This would explain why the two expressions are interchangeable. (vi) That the rafʿ (indicative) in the imperfect is due to the absence of an operant that causes the subjunctive or jussive, hence the expression fiʿl muḍāriʿ marfūʿ li-ḫuluwwihi min nāṣib wa-jāzim, which is used for the purpose of simplification (ʾirādat al-taqrīb). In line with the Basran view— which, according to Ibn Hišām is accepted even by those who use this flawed expression—the correct wording should be li-ḥulūlihi maḥall alism i.e. the indicative verb syntactically replaces the noun.89 (vii) That each of the future particles sīn and sawfa is described in ʾiʿrāb as ḥarf tanfīs (a particle denoting amplification), rather than ḥarf istiqbāl (a particle denoting futurity). For Ibn Hišām, the latter expression is better (ʾaḥsan) and clearer (ʾawḍaḥ). In this case, he does not go as far as saying that tanfīs is wrong, but explains that the term is synonymous with tawsīʿ (expansion) and that it indicates the broadening of the present (ḥāl), which is the confined tense (al-zaman al-ḍayyiq), to denote the future, which is the broad tense (al-zaman al-wāsiʿ).90 In contemporary grammar 86 87 88 89 90

ʾAsmar (1969:29); ʾAltōnjī (1974:64); Labadī (1988:130); Yaʿqūb (1988:190); ʿAbd al-Masīḥ and Tābirī (1990:128); ʾAltōnjī and ʾAsmar (1993: I, 130); Sinnū et al. (2010: I, 1537). Cf. ʿĀṣī and Yaʿqūb (1987: II, 943) and Yaʿqūb (1988:254) for the types of ṭalab (requisition) to which the term jawāb al-ṭalab is applicable. Cf. Qabāwa (1978:530). Ḥusayn (1981:116); ʿĀṣī and Yaʿqūb (1987: I, 345; II, 942); ʿAbd al-Masīḥ and Tābirī (1990:139); ʾAltōnjī and ʾAsmar (1993: I, 152); cf. Labadī (1988:42); Sinnū et al. (2010: II, 3627). Other than Muġnī II, 663, see I, 138.

grammar for beginners

81

books and reference works, tanfīs and istiqbāl are used interchangeably (a third synonym is taswīf ), and the two terms are often used together in the ʾiʿrāb of sīn and sawfa.91 (viii) That the adverb causes the genitive in the noun that follows it. In jalastu ʾamāma zaydin, for example, it would be incorrect, according to Ibn Hišām, to attribute the genitive in zaydin to the preceding adverb since it is nowhere stipulated that the word that precedes the construct should be an adverb. Instead, the correct expression is not maḫfūḍ bi-l-ẓarf, but maḫfūḍ bi-l-ʾiḍāfa (genitive by annexation).92 It might well be that the attribution of the genitive to the adverb was made on the analogy between adverbs and prepositions. This analogy was probably facilitated by the term šibh jumla, which refers both to an adverb and its genitive and to a preposition and its genitive. It is often stated in contemporary grammars that the adverb causes the genitive, and no distinction is made between adverbs and prepositions in the definition of šibh jumla.93

3

Conclusion

Ibn Hišām’s al-Muġnī differs significantly in arrangement and content from other grammatical works in the whole tradition. In fact, it is the only work we know of that consists of a lexicon of ḥurūf al-maʿānī as well as several sections that deal with various, mainly syntactical issues. Unlike works discussed in section 1 above, the Muġnī is not a survey of morphology and syntax that the student can refer to on the basis of the traditional division of the material into nouns, verbs and particles or, in the case of nouns, for example, into those that are in the nominative, accusative or genitive. The lexicon, which occupies about half of the book, is certainly useful for students, yet given the intricate details it includes, it can be best described as an encyclopedic reference that addresses complex and controversial issues. This is also true of some of the book’s other chapters that were detailed above. But the most distinctive feature of the Muġnī within the grammatical tradition is its focus on ʿiʿrāb. In addition to its lexicon, which contains ample material on ʾiʿrāb, and to the lists detailing the ʾiʿrāb of sentences, several of its chapters are devoted to issues of ʾiʿrāb. Ibn 91 92 93

ʾAltōnjī (1974:90); Rājiḥī (1979:163, 355); Ḥusayn (1981:118); ʿĀṣī and Yaʿqūb (1987: II, 702); Labadī (1988:227); Yaʿqūb (1988:215); ʿAbd al-Masīḥ and Tābirī (1990:237). Ibn Hišām, Muġnī II, 664; cf. ʾIʿrāb 106. Rājiḥī (1979:98); ʿĀṣī and Yaʿqūb (1987: II, 729); Labadī (1988:111); ʿAbd al-Masīḥ and Tābirī (1990:170, 240); ʾAltōnjī and ʾAsmar (1993: I, 222).

82

baalbaki

Hišām’s other book referred to above, al-ʾIʿrāb ʿan qawāʿid al-ʾiʿrāb, also demonstrates his profound interest in investigating ʾiʿrāb. The method in which Ibn Hišām presents his material in both works is most probably unprecedented in the tradition and rests on the discussion of general issues that pertain to the terminology, techniques and errors of ʾiʿrāb. Rather than arranging his material—which practically covers the whole range of syntactical issues that make up the traditional grammars—on the basis of the ʾiʿrāb of specific particles, constructions, etc., ample illustrations, including examples drawn from Ibn Hišām’s own experience with students, are introduced to support the issues under discussion. Accordingly, one can talk of kulliyyāt, or universals, which embrace a host of examples and apply to numerous šawāhid (attested material), and which represent Ibn Hišām’s global approach to the notion of ʾiʿrāb. In this respect, the Muġnī, and to a lesser extent the concise book entitled the ʾIʿrāb, are analogous to Ibn Jinnī’s (d. 392/1002) al-Ḫaṣāʾiṣ, which addresses global questions on phonology, morphology and syntax, but within which is embedded a huge number of examples that cover a large portion of the material that makes up the works which follow the prevalent traditional arrangement of the ʾabwāb of grammar. The previous discussion has shown how keen Ibn Hišām was to demonstrate that ʾiʿrāb has to be considered a ṣināʿa, among whose most essential foundations are formulaic expressions which should be strictly adhered to and which must be concise but contain specific information in each case (cf. the notions of ʾījāz and istīfāʾ in 2.1.1 above). In the ʾiʿrāb of ʾinna, for example, the fixed expression ḥarf tawkīd yanṣib al-ism wa-yarfaʿ al-ḫabar is frequently cited by Ibn Hišām,94 and it states, in the fewest words possible, the main features of ʾinna, namely, that it is a particle, that it indicates emphasis, and that it governs its noun and predicate. The same expression is used with ʾanna with one additional term, maṣdarī,95 to indicate that it is paraphrased with its noun as an infinitive. Similarly, in the ʾiʿrāb of yatarabbaṣna, the standard expression fiʿl muḍāriʿ mabnī ʿalā l-sukūn li-ttiṣālihi bi-nūn al-ʾināṯ96 contains the essential elements that must be mentioned in a specific order. Numerous other formulaic expressions are spread over the lexical part of the Muġnī and several chapters therein, and many have parallels in the ʾIʿrāb. Most of these expressions cited by Ibn Hišām have survived almost verbatim and are familiar to students studying grammar at high school or college throughout the Arab world. The most intriguing question in this regard is: 94 95 96

Ibn Hišām, Muġnī I, 37; II, 668; ʾIʿrāb 106. Ibn Hišām, ʾIʿrāb 106. Ibn Hišām, Muġnī II, 667.

grammar for beginners

83

When did these expressions become rigid or fossilized formulae? To be sure, the information and technical terms contained in these expressions are derived from grammarians as early as Sībawayhi, but they do not feature in grammar books as standard formulae that should be emulated. Even the pedagogical works which have survived, among which are those mentioned in section 1 above, do not include these expressions in their fixed forms. Obviously, these expressions were in wide circulation in Ibn Hišām’s time as he often mentions how teachers and students alike dealt with them. It is most likely therefore that they belong to a long tradition of teaching, but there was no attempt before Ibn Hišām—as far as we are able to determine—to record them and subject them to scrutiny. But even the material presented in the Muġnī and the ʾIʿrāb is not exhaustive, as noted, for example, in the absence of expressions used in the ʾiʿrāb of the munādā (vocative), the mamnūʿ min al-ṣarf (diptote), the noun after a ḥarf jarr zāʾid (otiose preposition), etc. The survival till the present of a large number of ʾiʿrāb expressions in their fixed forms demonstrates the power and authority which the grammar teaching tradition represents. Ironically, the authoritative nature of the tradition meant that certain widespread ʾiʿrāb expressions which are arguably erroneous were very difficult to replace. Accordingly, Ibn Hišām’s tireless effort at proving the faulty nature of some of the standard expressions (cf. the eight examples in 2.3 above) was hardly successful, as these expressions continued to be used almost seven centuries after Ibn Hišām’s time. Furthermore, Ibn Hišām’s appeal for adherence to conciseness in ʾiʿrāb expressions does not seem to have had a tangible impact. For example, the expression ẓarf li-mā yustaqbal min alzamān, instead of simply ẓarf mustaqbal is still frequently used,97 and mabnī li-, which Ibn Hišām says is redundant (cf. 2.1.1 above) is still used in fiʿl mabnī lil-maʿlūm/li-l-majhūl instead of fiʿl maʿlūm/majhūl.98 It is quite interesting that Ibn Hišām himself used some of the expressions which he criticized. On one occasion, he determines the ʾiʿrāb of ʾiḏ to be ẓarf li-mā maḍā min al-zamān,99 contrary to his own objection to such wordy expressions. Similarly, he inadvertently used the term ḥarf tanfīs with reference to sīn, although he argues elsewhere in favor of its alternative term ḥarf istiqbāl.100 It thus seems that some of the ʾiʿrāb expressions criticized by Ibn Hišām were so prevalent that even he was accustomed to using them occasionally. 97 98 99 100

ʾAsmar (1969:9); Rājiḥī (1979:247); ʿĀṣī and Yaʿqūb (1987: I, 69); Yaʿqūb (1988:78); ʾAltōnjī and ʾAsmar (1993: I, 25). Qabāwa (1978:411, 420); ʿĀṣī and Yaʿqūb (1987: II, 937f.). Ibn Hišām, ʾIʿrāb 69; but see Muġnī I, 80, where ism li-l-zaman al-māḍī is used. Ibn Hišām, Muġnī I, 81.

84

baalbaki

Bibliographical References A

Primary Sources

ʾAbū l-Ṭayyib, Marātib = ʾAbū l-Ṭayyib ʿAbd al-Raḥmān ibn ʿAlī al-Luġawī, Marātib alnaḥwiyyīn. Ed. by Muḥammad ʾAbū l-Faḍl ʾIbrāhīm. 2nd ed. Cairo: Dār Nahḍat Miṣr, 1974. ʾAsfarāyīnī, Lubāb = Tāj al-Dīn Muḥammad ibn Muḥammad al-ʾAsfarāyīnī, al-Lubāb fī ʿilm al-ʾiʿrāb. Ed. by Šawqī al-Maʿarrī. Beirut: Maktabat Lubnān, 1996. ʿAskarī, Ṣināʿatayn = ʾAbū Hilāl al-Ḥasan ibn ʿAbdallāh al-ʿAskarī, Kitāb al-Ṣināʿatayn: al-kitāba wa-l-šiʿr. Ed. by ʿAlī Muḥammad al-Bijāwī and Muḥammad ʾAbū l-Faḍl ʾIbrāhīm. Cairo: Dār ʾIḥyāʾ al-Kutub al-ʿArabiyya, 1952. ʿAynī, Maqāṣid = ʾAbū Muḥammad Maḥmūd ibn ʾAḥmad al-ʿAynī, al-Maqāṣid al-naḥwiyya fī šarḥ šawāhid al-ʾAlfiyya. [in the margin of Baġdādī’s Ḫizāna, 4 vols. Būlāq: 1299 A.H.] ʾAzharī, Tahḏīb = ʾAbū Manṣūr Muḥammad ibn ʾAḥmad al-ʾAzharī, Tahḏīb al-luġa. Ed. by ʿAbd al-Salām Hārūn et. al. 15 vols. Cairo: al-Muʾassasa al-Miṣriyya al-ʿĀmma li-l-Taʾlīf wa-l-ʾAnbāʾ wa-l-Našr and al-Dār al-Miṣriyya li-l-Taʾlīf wa-l-Tarjama, 1964– 1967. ʾAzharī, Tamrīn = Zayn al-Dīn Ḫālid ibn ʿAbdallāh al-ʾAzharī, Tamrīn al-ṭullāb fī ṣināʿat al-ʾiʿrāb. Ed. by ʿAzīz ʾĪġzīr. Sidon: al-Maktaba al-ʿAṣriyya, 2009. Baġdādī, Ḫizāna = ʿAbd al-Qādir ibn ʿUmar al-Baġdādī, Ḫizānat al-ʾadab wa-lubb lubāb lisān al-ʿArab. Ed. by ʿAbd al-Salām Muḥammad Hārūn. 13 vols. Cairo: Dār al-Kātib al-ʿArabī, 1967–1986. Fārisī, ʾĪḍāḥ = ʾAbū ʿAlī al-Ḥasan ibn ʾAḥmad al-Fārisī, al-ʾĪḍāḥ. Ed. by Kāẓim Baḥr alMarjān. Beirut: ʿĀlam al-Kutub, 1996. Ḫalaf, Muqaddima = ʾAbū Muḥriz Ḫalaf ibn Ḥayyān al-ʾAḥmar al-Baṣrī, Muqaddima fī l-naḥw. Ed. by ʿIzz al-Dīn al-Tanūḫī. Damascus: Wizārat al-Ṯaqāfa wa-l-ʾIršād alQawmī, 1961. Ḫalīl, Ḥurūf = ʾAbū ʿAbd al-Raḥmān al-Ḫalīl ibn ʾAḥmad al-Farāhīdī, al-Ḥurūf. In Ṯalāṯat kutub fī l-ḥurūf, ed. by Ramaḍān ʿAbd al-Tawwāb. Cairo: Maktabat al-Ḫānjī and Riyad: Dār al-Rifāʿī, 1982. Ḫalīl, Jumal = ʾAbū ʿAbd al-Raḥmān al-Ḫalīl ibn ʾAḥmad al-Farāhīdī, al-Jumal fī l-naḥw. Ed. by Faḫr al-Dīn Qabāwa. 2nd ed. Beirut: Muʾassasat al-Risāla, 1987. Ibn al-ʾAnbārī, Nuzha = ʾAbū l-Barakāt ʿAbd al-Raḥmān ibn Muḥammad al-ʾAnbārī, Nuzhat al-ʾalibbāʾ fī ṭabaqāt al-ʾudabāʾ. Ed. by ʾIbrāhīm al-Sāmarrāʾī. Baghdad: Maktabat al-ʾAndalus, 1970. Ibn ʿAqīl, Šarḥ = Bahāʾ al-Dīn ʿAbdallāh ibn ʿAbd al-Raḥmān Ibn ʿAqīl, Šarḥ Ibn ʿAqīl ʿalā ʾAlfiyyat Ibn Mālik. Ed. by Ramzī Munīr Baʿalbakī. Beirut: Dār al-ʿIlm li-l-Malāyīn, 1992. Ibn al-ʾAṯīr, Nihāya = Majd al-Dīn ʾAbū l-Saʿādāt al-Mubārak ibn Muḥammad Ibn al-

grammar for beginners

85

ʾAṯīr, al-Nihāya fī ġarīb al-Ḥadīṯ wa-l-ʾAṯar. Ed. by Ṭāhir ʾAḥmad al-Zāwī and Maḥmūd Muḥammad al-Ṭanāḥī. 2nd ed. 5 vols. Beirut: Dār al-Fikr, 1979. Ibn Durayd, Jamhara = ʾAbū Bakr Muḥammad ibn al-Ḥasan Ibn Durayd, Jamharat alluġa. Ed. by Ramzī Munīr Baʿalbakī. 3 vols. Beirut: Dār al-ʿIlm li-l-Malāyīn, 1987–1988. Ibn Ḥajar, Durar = Šihāb al-Dīn ʾAbū l-Faḍl ʾAḥmad ibn ʿAlī Ibn Ḥajar al-ʿAsqalānī, alDurar al-kāmina fī ʾaʿyān al-miʾa al-ṯāmina. Ed. by Fritz Krenkow. 4 vols. Hyderabad: Dāʾirat al-Maʿārif al-ʿUṯmāniyya, 1350A.H. Ibn Ḫarūf, Šarḥ = ʾAbū l-Ḥasan ʿAlī ibn Muḥammad Ibn Ḫarūf al-ʾIšbīlī, Šarḥ Jumal alZajjājī. Ed. by Salwā Muḥammad ʿUmar ʿArab. 2 vols. Mecca: Jāmiʿat ʾUmm al-Qurā, 1998. Ibn Hišām, ʾIʿrāb = Jamāl al-Dīn ʾAbū Muḥammad ʿAbdallāh ibn Yūsuf Ibn Hišām, alʾIʿrāb ʿan qawāʿid al-ʾiʿrāb. Ed. by ʿAlī Fūda Nīl. Riyad: Jāmiʿat al-Riyāḍ, 1981. Ibn Hišām, Muġnī = Jamāl al-Dīn ʾAbū Muḥammad ʿAbdallāh ibn Yūsuf Ibn Hišām, Muġnī l-labīb ʿan kutub al-ʾAʿārīb. Ed. by Muḥammad Muḥyī l-Dīn ʿAbd al-Ḥamīd. 2 vols. Cairo: al-Maktaba al-Tijāriyya, 1959. Ibn Jinnī, Lumaʿ = ʾAbū l-Fatḥ ʿUṯmān Ibn Jinnī, al-Lumaʿ fī l-ʿArabiyya. Ed. by Ḥāmid al-Muʾmin. 2nd ed. Beirut: ʿĀlam al-Kutub, 1985. Ibn Jinnī, Sirr = ʾAbū l-Fatḥ ʿUṯmān Ibn Jinnī, Sirr ṣināʿat al-ʾiʿrāb. Ed. by Ḥasan Hindāwī. 2 vols. Damascus: Dār al-Qalam, 1985. Ibn Kaysān, Muwaffaqī = ʾAbū l-Ḥasan Muḥammad ibn ʾAḥmad ibn Muḥammad Ibn Kaysān, al-Muwaffaqī fī l-naḥw. Ed. by ʿAbd al-Ḥusayn al-Fatlī and Hāšim Ṭāhā Šalāš. al-Mawrid 4:2 (1975) 103–124. Ibn Manẓūr, Lisān = Jamāl al-Dīn ʾAbū l-Faḍl Muḥammad ibn Mukarram Ibn Manẓūr, Lisān al-ʿArab. 15 vols. Beirut: Dār Ṣādir, 1968. Ibn al-Nadīm, Fihrist = ʾAbū l-Faraj Muḥammad ibn ʾAbī Yaʿqūb Ibn al-Nadīm, al-Fihrist. Ed. by ʾAyman Fuʾād Sayyid. 4 vols. London: Muʾassasat al-Furqān li-l-Turāṯ al-ʾIslāmī, 2014. Ibn al-Nāẓim, Šarḥ = Badr al-Dīn ʾAbū ʿAbdallāh Muḥammad ibn Muḥammad Ibn Mālik, known as Ibn al-Nāẓim, Šarḥ ʾAlfiyyat Ibn Mālik. Ed. by ʿAbd al-Ḥamīd alSayyid Muḥammad ʿAbd al-Ḥamīd. Beirut: Dār al-Jīl, n.d. Ibn Qutayba, Talqīn = ʾAbū Muḥammad ʿAbdallāh ibn Muslim Ibn Qutayba al-Dīnawarī, Talqīn al-mutaʿallim min al-naḥw. Ed. by Jamāl ʿAbd al-ʿĀṭī Muḫaymar. Cairo: Maṭbaʿat ʾAbnāʾ Wahba Ḥassān, 1989. Ibn Qutayba, ʿUyūn = ʾAbū Muḥammad ʿAbdallāh ibn Muslim Ibn Qutayba al-Dīnawarī, ʿUyūn al-ʾaḫbār. 4 vols. Repr. from the Cairo edition, Beirut: Dār al-Kitāb al-ʿArabī, 1973. Ibn al-Sarrāj, Mūjaz = ʾAbū Bakr Muḥammad ibn Sahl Ibn al-Sarrāj, al-Mūjaz fī l-naḥw. Ed. by Muṣṭafā al-Šuwaymī and Bin Sālim Dāmirjī. Beirut: Muʾassasat Badrān, 1965. Ibn Sīda, Muḥkam = ʾAbū l-Ḥasan ʿAlī ibn ʾIsmāʿīl Ibn Sīda, al-Muḥkam wa-l-muḥīṭ alʾaʿẓam. Ed. by ʿAbd al-Ḥamīd Hindāwī. 11 vols. Beirut: Dār al-Kutub al-ʿIlmiyya, 2000.

86

baalbaki

Ibn Šuqayr, Muḥallā = ʾAbū Bakr ʾAḥmad ibn al-Ḥasan Ibn Šuqayr, al-Muḥallā, wujūh al-naṣb. Ed. by Fāʾiz Fāris. Beirut: Muʾassasat al-Risāla and Irbid: Dār al-ʾAmal, 1987. Ibn al-Ṭarāwa, ʾIfṣāḥ = ʾAbū l-Ḥusayn Sulaymān ibn Muḥammad Ibn al-Ṭarāwa, Risālat al-ʾIfṣāḥ bi-baʿḍ mā jāʾa min al-ḫaṭaʾ fī l-ʾĪḍāḥ. Ed. by Kāẓim Baḥr al-Marjān. Beirut: ʿĀlam al-Kutub, 2011. Ibn ʿUṣfūr, Šarḥ = ʾAbū l-Ḥasan ʿAlī ibn Muʾmin Ibn ʿUṣfūr, Šarḥ Jumal al-Zajjājī. Ed. by Ṣāḥib ʾAbū Janāḥ. 2 vols. Beirut: ʿĀlam al-Kutub, 1999. Jāḥiẓ, Ḥayawān = ʾAbū ʿUṯmān ʿAmr ibn Baḥr al-Jāḥiẓ, Kitāb al-Ḥayawān. Ed. by ʿAbd al-Salām Muḥammad Hārūn. 8 vols. Repr. from the Cairo edition, Beirut: Dār al-Jīl, 1992. Jāḥiẓ, Muʿallimīn = ʾAbū ʿUṯmān ʿAmr ibn Baḥr al-Jāḥiẓ, Kitāb al-Muʿallimīn, in Rasāʾil al-Jāḥiẓ, ed. by ʿAbd al-Salām Muḥammad Hārūn, III, 27–51. 4 vols. Cairo: Maktabat al-Ḫānjī, 1979. Luġda, Naḥw = ʾAbū ʿAlī al-Ḥasan ibn ʿAbdallāh al-maʿrūf bi-Luġda al-ʾAṣbahānī, Muḫtaṣar fī l-naḥw. Ed. by ʿAbd al-Ḥusayn al-Fatlī. al-Mawrid 3:3 (1974) 221–246. Mujāšiʿī, ʿUyūn = ʾAbū l-Ḥasan ʿAlī ibn Faḍḍāl al-Mujāšiʿī, Šarḥ ʿUyūn al-ʾiʿrāb. Ed. by Ḥasnāʾ ʿAbd al-ʿAzīz al-Qunayʿīr. Riyad: Jāmiʿat al-Malik Saʿūd, 1993. Naḥḥās, Tuffāḥa = ʾAbū Jaʿfar ʾAḥmad ibn Muḥammad al-Naḥḥās, al-Tuffāḥa fī l-naḥw. Ed. by Kōrkīs ʿAwwād. Baghdad: Maṭbaʿat al-ʿĀnī, 1965. Qifṭī, ʾInbāh = Jamāl al-Dīn ʾAbū l-Ḥasan ʿAlī ibn Yūsuf al-Qifṭī, ʾInbāh al-ruwāt ʿalā ʾanbāh al-nuḥāt. Ed. by Muḥammad ʾAbū l-Faḍl ʾIbrāhīm. 4 vols. Cairo: Dār al-Fikr al-ʿArabī and Beirut: Muʾassasat al-Kutub al-Ṯaqāfiyya, 1986. Sīrāfī, ʾAḫbār = ʾAbū Saʿīd al-Ḥasan ibn ʿAbdallāh al-Sīrāfī, ʾAḫbār al-naḥwiyyīn al-Baṣriyyīn. Ed. by Fritz Krenkow. Beirut: al-Maṭbaʿa al-Kāṯūlīkiyya, 1936. Suyūṭī, Buġya = Jalāl al-Dīn ʾAbū l-Faḍl ʿAbd al-Raḥmān ibn ʾAbī Bakr al-Suyūṭī, Buġyat al-wuʿāt fī ṭabaqāt al-luġawiyyīn wa-l-nuḥāt. Ed. by Muḥammad ʾAbū l-Faḍl ʾIbrāhīm. 2nd ed. 2 vols. Beirut: Dār al-Fikr, 1979. Suyūṭī, Tuḥfa = Jalāl al-Dīn ʾAbū l-Faḍl ʿAbd al-Raḥmān ibn ʾAbī Bakr al-Suyūṭī, Tuḥfat al-ʾadīb fī nuḥāt Muġnī l-labīb. Ed. by Ḥasan Malḫ and Suhā Naʿja. Irbid: ʿĀlam alKutub al-Ḥadīṯ, 2008. Tanūḫī, Tārīḫ = ʾAbū l-Maḥāsin al-Mufaḍḍal ibn Muḥammad ibn Misʿar al-Tanūḫī, Tārīḫ al-ʿulamāʾ al-naḥwiyyīn. Ed. by ʿAbd al-Fattāḥ Muḥammad al-Ḥulw. Riyad: Maṭābiʿ Dār al-Hilāl, 1981. Yāqūt, Muʿjam = Šihāb al-Dīn ʾAbū ʿAbdallāh Yāqūt ibn ʿAbdallāh al-Rūmī al-Ḥamawī, Muʿjam al-ʾudabāʾ. Ed. by ʾIḥsān ʿAbbās. 7 vols. Beirut: Dār al-Ġarb al-ʾIslāmī, 1993. Zajjājī, ʾĪḍāḥ = ʾAbū l-Qāsim ʿAbd al-Raḥmān ibn ʾIsḥāq al-Zajjājī, al-ʾĪḍāḥ fī ʿilal al-naḥw. Ed. by Māzin al-Mubārak. Cairo: Dār al-ʿUrūba, 1959. Zajjājī, Jumal = ʾAbū l-Qāsim ʿAbd al-Raḥmān ibn ʾIsḥāq al-Zajjājī, al-Jumal. Ed. by Muḥammad ibn ʾAbī Šanab. Paris: Klincksieck, 1957. Zamaḫšarī, Mufaṣṣal = ʾAbū l-Qāsim Maḥmūd ibn ʿUmar al-Zamaḫšarī, al-Mufaṣṣal fī ʿilm al-ʿArabiyya. Cairo: 1323 A.H.

grammar for beginners

87

Zubaydī, Laḥn = ʾAbū Bakr Muḥammad ibn al-Ḥasan al-Zubaydī al-ʾAndalusī, Laḥn alʿawāmm. Ed. by Ramaḍān ʿAbd al-Tawwāb. Cairo: al-Maṭbaʿa al-Kamāliyya, 1964. Zubaydī, Ṭabaqāt = ʾAbū Bakr Muḥammad ibn al-Ḥasan al-Zubaydī al-ʾAndalusī, Ṭabaqāt al-naḥwiyyīn wa-l-luġawiyyīn. Ed. by Muḥammad ʾAbū l-Faḍl ʾIbrāhīm. 2nd ed. Cairo: Dār al-Maʿārif, 1973. Zubaydī, Wāḍiḥ = ʾAbū Bakr Muḥammad ibn al-Ḥasan al-Zubaydī al-ʾAndalusī, Kitāb al-Wāḍiḥ. Ed. by ʿAbd al-Karīm Ḫalīfa. Amman: 1976.

B

Secondary Sources

ʿAbd al-Masīḥ, George Mitrī and Hānī George Tābirī. 1990. al-Ḫalīl: Muʿjam muṣṭalaḥāt al-naḥw al-ʿArabī. Beirut: Maktabat Lubnān. Aljassar, Talal and Jonathan Owens. 2015. “Variation, pedagogization, and the early Maʿānī al-Qurʾān tradition”. Zeitschrift für arabische Linguistik 62.5–37. ʾAltōnjī, Muḥammad. 1974. Muʿjam al-ʾadawāt al-naḥwiyya. 5th ed. Benghazi: Maktabat Qūrīnā. ʾAltōnjī, Muḥammad and Rājī ʾAsmar. 1993. al-Muʿjam al-mufaṣṣal fī ʿulūm al-luġa (alʾalsuniyyāt). 2 vols. Beirut: Dār al-Kutub al-ʿIlmiyya. ʿĀṣī, Michel and ʾImīl Badīʿ Yaʿqūb. 1987. al-Muʿjam al-mufaṣṣal fī l-luġa wa-l-ʾadab. 2 vols. Beirut: Dār al-ʿIlm li-l-Malāyīn. ʾAsmar, Jirjis ʿĪsā. 1969. Qāmūs al-ʾiʿrāb. Beirut: Dār al-ʿIlm li-l-Malāyīn. Baalbaki, Ramzi. 2008. The legacy of the Kitāb: Sībawayhi’s analytical methods within the context of the Arabic grammatical theory. Leiden: E.J. Brill. Baalbaki, Ramzi. 2009. “The place of al-Jāḥiẓ in the Arabic philological tradition”. AlJāḥiẓ: A Muslim humanist for our time, ed. by Arnim Heinemann, John L. Meloy, Tarif Khalidi and Manfred Kropp, 91–110. Beirut: Ergon Verlag Würzburg. Baalbaki, Ramzi. 2014. The Arabic lexicographical tradition from the 2nd/8th to the 12th/18th century. Leiden: E.J. Brill. Carter, Michael G. 1979. “A 16th century grammatical experiment that failed”. Arabica 26.267–273. Carter, Michael G. 1990. “Arabic grammar”. The Cambridge history of Arabic literature: Religion, learning and science in the ʿAbbasid period, ed. by Michael J.L. Young, J. Derek Latham and Robert B. Serjeant, 118–138. Cambridge: Cambridge University Press. Gelder, Geert Jan van. 2011. “Against the Arabic grammarians: Some poems”. In the shadow of Arabic: The centrality of language to Arabic culture. Studies presented to Ramzi Baalbaki on the occasion of his sixtieth birthday, ed. by Bilal Orfali, 249–263. Leiden: E.J. Brill. Gully, Adrian. 1995. Grammar and semantics in Medieval Arabic: A study of Ibn Hisham’s ‘Mughni l-Labib’. Richmond: Curzon. Ḥusayn, Zayn al-ʿĀbidīn. 1981. al-Muʿjam fī l-naḥw wa-l-ṣarf. Tunis: al-Dār al-ʿArabiyya li-l-Kitāb.

88

baalbaki

Jabbārīn, Muḥammad. 1999. “ʾAḫbār al-nuḥāt bayna l-wāqiʿ wa-l-tanaddur”. Israel Oriental Studies 19.287–341. (= Compilation and creation in adab and luġa: Studies in memory of Naphtali Kinberg (1948–1997), ed. by Albert Arazi et al.). Winona Lake, Ind.: Eisenbrauns. Labadī, Muḥammad Samīr Najīb. 1988. Muʿjam al-muṣṭalaḥāt al-naḥwiyya wa-l-ṣarfiyya. 3rd ed. Amman: Dār al-Furqān and Beirut: Muʾassasat al-Risāla. Nīl, ʿAlī Fūda. 1985. Ibn Hišām al-ʾAnṣārī: ʾĀṯāruhu wa-maḏhabuhu l-naḥwī. Riyad: Jāmiʿat al-Malik Saʿūd. Owens, Jonathan. 2005. “The grammatical tradition and Arabic language teaching: A view from here”. Investigating Arabic: Current parameters in analysis and learning, ed. by Alaa Elgibali, 103–116. Leiden: E.J. Brill. Peled, Yishai. 2010. “Sībawayhi’s Kitāb and the teaching of Arabic grammar”. Jerusalem Studies in Arabic and Islam 37.163–188. Pellat, Charles. 1969. The life and works of Jāḥiẓ: Translations of selected texts. Transl. from the French by David Martin Hawke. Berkeley and Los Angeles: University of California Press. Qabāwa, Faḫr al-Dīn. 1978. al-Mawrid al-kabīr: Namāḏij taṭbīqiyya fī l-ʾiʿrāb wa-l-ʾadawāt wa-l-ṣarf. 2nd ed. Beirut: Dār al-ʾĀfāq al-Jadīda. Rājiḥī, ʿAbdū. 1979. al-Taṭbīq al-naḥwī. Beirut: Dār al-Naḥda al-ʿArabiyya. Šalabī, ʿAbd al-Fattāḥ ʾIsmāʿīl. 1958. ʾAbū ʿAlī al-Fārisī: Ḥayātuhu wa-makānatuhu bayna ʾaʾimmat al-ʿArabiyya wa-ʾāṯāruhu fī l-qirāʾāt wa-l-naḥw. Cairo: Maktabat Nahḍat Miṣr. Sezgin, Fuat. 1984. Geschichte des arabischen Schrifttums. IX. Grammatik. Leiden: E.J. Brill. Sinnū, ʾAhyaf, Jīrār Jihāmī and Hiba Šibārū Sinnū. 2010. Mawsūʿat muṣṭalaḥāt al-ʿulūm al-naḥwiyya. 3 vols. Beirut: Maktabat Lubnān. Talmon, Rafael. 1990. “Kitāb Muqaddima fī l-naḥw al-mansūb ʾilā Ḫalaf al-ʾAḥmar: Dirāsa wa-fihris muṣṭalaḥāt”. al-Karmil 11.129–199. Yaʿqūb, ʾImīl Badīʿ. 1988. Mawsūʿat al-ḥurūf fī l-luġa al-ʿArabiyya. Beirut: Dār al-Jīl.

Sallaṭa/tasallaṭa, a Possible Parallel for ‘Govern’? Michael G. Carter

It is more than a century since Josef Weiss published an article which demonstrated that the equation of the grammatical concepts of Arabic ʿamal and Latin regere ‘to govern’, was lexically impossible, because the verb ʿamila in the sense of ‘to govern’ requires the preposition ʿalā, i.e. ‘to govern over’, while in grammar ʿamila occurs exclusively with the preposition fī, i.e. ‘to have an effect upon, operate on’.1 Half a century later I used this article in my doctoral thesis to support an interpretation of Sībawayhi’s grammatical theory which eliminates the hierarchical and vertical notion of ‘government’ superimposed on Arabic grammar in Western scholarship, and restores the original sense of ʿamila fī in Sībawayhi’s Kitāb, that words ‘operate’ on each other in a linear and horizontal sequence. There is, however, a term occurring sporadically in Arabic grammatical texts which at first sight could be taken as a calque of Latin regere, namely sallaṭa/tasallaṭa, lit. ‘to give or be given power or authority’, a denominative verb from the loan word sulṭān ‘power, authority’, in which meaning that word occurs in the Qurʾān as a borrowing from Aramaic or Syriac, according to your point of view.2 In the light of the subsequent evolution of sulṭān to denote the person of a ruler, sallaṭa might seem appropriate for expressing the idea of grammatical governing in the manner of its Latin analogue regere, cognate with rex. In this paper a number of occurrences of the term sallaṭa will be presented. They are grouped broadly by topic and arranged chronologically, including a couple of dubious provenance, the aim being to provide the context in which sallaṭa is used. In the case of Ibn Hišām only one specimen from each category is selected, with further examples indicated by their page numbers. To avoid preempting the conclusions, sallaṭa/tasallaṭa will be translated literally throughout, as ‘to give or be given power’.

1 See Weiss (1910). 2 See Jeffery (1938 [2007]:176f.).

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_006

90 1

carter

Sallaṭa Not Explicitly Associated with ʿamal

Item 1.1 *Ibn Qutayba (d. 276/889), (attrib.), Talqīn al-mutaʿallim. fol. 32b; Hidāyat Allāh 159 ‫فسل ّطوا الفتحة على الياء فقلبوها ألف ًا كما سل ّطوها على غزا ومضا ورما وما أشبهه‬ The author, whoever that may be,3 has just pointed out that weak 3rd radical verbs are virtually uninflected when the medial vowel is a fatḥa, so yabqā represents both *yabqayu and *yabqaya, because “they have given the fatḥa power over the yāʾ and converted it into an ʾalif, just as they give [ fatḥa] power [over the weak third radical] in the verbs ġazā, maḍā, ramā and the like”. It goes without saying that the Arabic spellings here,4 not to mention the linguistic argumentation, are highly suspect, which is one reason why the attribution to Ibn Qutayba can be safely rejected. Nevertheless it is an interesting example of sallaṭa in a morphological context, even if it cannot be securely ascribed or dated. Item 1.2 al-Rummānī (d. 384/994), Šarḥ Kitāb Sībawayhi 3, pt. 2, fol. 44r, Lines 4–6 from Bottom (§269 in Derenbourg 1, 419f./Būlāq 1, 471) ‫لم صارت قال مع إن بمنزلتهما مع ز يد اذا قلت قال ز يد عمرو خير الناس وهل ذلك لأن هذا‬ ‫الموقع الذي يقع فيه ز يد مبتدأ وتقع فيه إن وهو موقع واحد لا تسل ّط عليه قال لأنه موقع ابتداء‬ ‫الجملة فلا سبيل لهما اليه‬ Al-Rummānī’s Kitāb commentary first poses a series of questions about a chapter or group of chapters, then answers them one by one. The question here is easier to paraphrase than to translate in full: what al-Rummānī is asking is, why does qāla+ʾinna have the same status as qāla+zaydun in qāla zaydun ʿamrun ḫayru l-nāsi ‘Zayd said “ʿAmr is the best of people”’, and is this connected with the distributional fact that both qāla+zaydun and qāla+ʾinna precede equational sentences? Finally he asks whether this is the reason why “there is no way that qāla in either case (i.e. qāla+ʾinna and qāla+zaydun) can be given power over the equational sentence”.

3 Kāʾinan man kāna, in the words of the editor of the Talqīn, see Hidāyat Allāh 1986:51, though he argues elsewhere (39ff.) that there is no proof that Ibn Qutayba was not the author. See also Carter (1979) for arguments against Ibn Qutayba’s authorship. 4 In the Hidāyat Allāh edition the spellings of maḍā and ramā have been corrected.

sallaṭa/tasallaṭa, a possible parallel for ‘govern’?

91

When al-Rummānī eventually provides the answer to these questions a few folios later (without mentioning the term sallaṭa at all!), some of the obscurity evaporates: the issue is that ibtidāʾ is itself an autonomous operator (ʿāmil) and so it cannot be displaced by qāla. The problem arises because of the unique status of qāla, which in theory can only be followed by direct speech ḥikāya, hence, according to the rules for ḥikāya, the grammatical object (i.e. the words actually spoken) cannot be operated on by the verb qāla, unlike the ʾafʿāl al-qulūb which introduce equational sentences overtly marked as direct objects.5 Item 1.3 al-Rummānī, Šarḥ Kitāb Sībawayhi 4, pt. 1, fol. 116v, Lines 8–12 (§ 402 in Derenbourg 2, 155/Būlāq 2, 153) ‫ب ولم تدخل لتسل ّط على دخول النون وكذلك‬ ّ ‫لأن ما إنما دخلت لتسهل ذكر الفعل بعد ر‬ ‫حيثما تكون آتك يضعف لأن ما دخلت لتسل ّط حيث على الجزاء ولم تدخل لتصحح دخول‬ ‫النون فأما أينما تكونّن آتك فقوي حسن لأن ما قد خلصت لتسل ّط على دخول النون‬ This passage discusses two different functions of the suffix mā. The first is seen in the mā on rubbamā, “which is inserted merely to facilitate the mention of a verb after rubba and not to give the power for the energetic n to be suffixed [to that verb]”. Al-Rummānī compares this with the substandard conditional ḥayṯumā takūnu ʾātika ‘wherever you are I come to you’ (with takūnu for takun), where the mā “gives ḥayṯu the power to form a conditional sentence” (without an apocopated verb), and he then contrasts it with the well-formed construction ʾaynamā takūnanna ʾātika ‘wherever you may be I shall come to you’, illustrating the second function of mā, “which occurs exclusively to give the power for the energetic n to be suffixed.” In spite of some difficulties with the reading and interpretation of the above text, the point emerges clearly that sallaṭa here refers to the ‘empowering’ or ‘giving authority’ to certain morphosyntactic features rather than to grammatical operation per se. There is a suggestive parallelism between sallaṭa and ṣaḥḥaḥa ‘to validate, authenticate’ used alongside sallaṭa in this passage.

5 This is indeed Sībawayhi’s topic in §269 in the Kitāb, cf. Guillaume (1983).

92

carter

Item 1.4 al-Zamaḫšarī (d. 538/1144), Mufaṣṣal 132 (§ 505) ‫ومنها أن الفعل الذي تسل ّطه على الاسم يجب تأخره عنها وأنه يجيء محذوف ًا في الاكثر كما حذف‬ ‫مع الباء في بسم الل ّٰه‬ The topic here is rubba phrases qualified by verb phrases. Al-Zamaḫšarī’s actual example is from a poem, rubba rifdin haraqtahu ‘there is many a bowl which you have poured out’. Note that the intuitive English translation ‘many a bowl have you poured out’ falsely implies a conventional inverted verbal sentence: on the contrary, haraqtahu is parsed as an adjectival qualifier (ṣifa) in the Arab analysis, and the apparent inversion is explained by the fact that the verb to which “you give power” over the rubba phrase is assumed to have been elided, thus *haraqta rubba rifdin haraqtahu. In practice, as al-Zamaḫšarī points out, this hypothetical verb is nearly always elided, as it is in bi-smi llāhi for ʾuqsimu bi-smi llāhi. Our last example shows sallaṭa in a grammatical text but not used as a technical term at all: Item 1.5 Ibn Hišām (d. 761/1360), Muġnī 2, 171, l. 7 from Bottom ‫لانك إذا قلت أنت منهي عن أن تقوم إلا أن يشاء الل ّٰه فلست بمنهي فقد سل ّطته على أن يقوم‬ ‫و يقول شاء الل ّٰه ذلك‬ It is enough to translate this to see that sallaṭa here is used literally: “because if you say to someone ‘you are forbidden to stand up unless God wills it and then you are not forbidden’, what you have done is give that person the power to stand up anyway and say ‘Well, that is just what God willed’ ”.

2

Cases Where sallaṭa is Explicitly Associated with ʿamal

Item 2.1 *al-Zajjāj (d. 311/923) (attrib.), ʾIʿrāb al-Qurʾān 1, 314 (re Q. 2/48 wa-ttaqū yawman lā tajzī nafsun ʿan nafsin šayʾan) ‫أما القياس فإن الصفة تخصص الموصوف كما أن الصلة تخصص الموصول ولا تعمل في‬ ‫الموصوف ولا ٺتسل ّط عليهكما لا تعمل الصلة في الموصول ومرتبتها أن تكون بعد الموصوف‬ ‫كما أن مرتبة الصفةكذلك‬

sallaṭa/tasallaṭa, a possible parallel for ‘govern’?

93

Leaving aside the dubious attribution to al-Zajjāj,6 there is at least an unmistakable association of sallaṭa with grammatical operation, ʿamal. The issue here is the apparent paradox that the terms denoting the antecedents of adjectival and relative constructions are passive in form, mawṣūf and mawṣūl, implying that they are operated on by the qualifiers which follow them, ṣifa and ṣila, seemingly in breach of the axiom prohibiting retroactive operation. *Al-Zajjāj explains this misleading terminology away by pointing out that these elements merely specify some particular quality of their antecedent and therefore have a lower syntactic rank (martaba), for which reason, as he puts it, “they do not operate on or have any power over their antecedent”. Although *al-Zajjāj does not mention it, this case is covered by the general principle of the tawābiʿ, the ‘concordants’, whose terminology is distributional (Head + Qualifier in that order), but the case agreement of the Qualifier is accounted for by the operation of the Head (matbūʿ) upon the following element (tābiʿ) in the natural order of operators, ʿāmil and maʿmūl fīhi. Note the hierarchical term martaba ‘rank’ which looks suspiciously like a component of the sallaṭa metaphor, but in fact the principle was established already in Sībawayhi that some words do have a higher status (manzila) than others, which determines the range of words on which they may operate. Item 2.2 al-Zajjājī, (d. 337/949 or Later), Lāmāt 77 ‫فلم يجز لما قبل إن أن يعمل فيما بعدها واللام بينهما لأن لام الابتداء حاجز يمنع ما قبله من‬ ‫التخطي إلى ما بعده ألا ترى أنك تقول علمت لز يد منطلق ]وحلفت لأخوك قائم[ ولا يكون‬ ‫لـ”علمت“ تسلط على ما بعد اللام‬ In a discussion of the role of lām al-ibtidāʾ al-Zajjājī argues, “It is not allowed for what precedes ʾinna to operate ( yaʿmala) on what follows ʾinna while lām al-ibtidāʾ intervenes between them, because this lām is a barrier which prevents what precedes it from stepping over to what follows it. Do you not see that you say ʿalimtu la-zaydun munṭaliqun ‘I know indeed Zayd is departing’ […] and ʿalimtu has no power over what follows the lām?”. As a result zaydun munṭaliqun is prevented from taking the dependent form zaydan munṭaliqan as the sentential direct object of ʿalimtu.

6 The work is also attributed to al-Zajjājī, but in any case the authorship is marginal to the point under discussion.

94

carter

Item 2.3 al-Rummānī Ḥudūd, Jawād/Maskūnī 47; Nāṣir 39; Troupeau 127 ‫حروف التعدية هي التي تسلط العامل على ما بعدها حتى يتعلق بها كحرف الاستثناء في الايجاب‬ ‫وحروف الجر‬ There is no doubt that sallaṭa here denotes giving the power to be an operator (ʿāmil) on the following word, but the link with transitivity as we understand it is hard to grasp. Perhaps al-Rummānī sees zaydan in jāʾa l-qawmu ʾillā zaydan as somehow operated on by the preceding sentence, a well established syntactical principle.7 For the prepositions it may be that al-Rummānī was thinking of their function with indirectly transitive verbs, as in marartu bi-zaydin etc., but both these notions need further exploration. However, our present purpose is simply to document the connection between sallaṭa and ʿamal in the 4th/10th century. Troupeau’s translation of this passage is as follows: “Les particules de dépassement: ce sont celles qui imposent le régissant à ce qui est après elles, de sorte qu’ il est attaché à elles, comme les particules de l’exception dans l’affirmation et les particules de l’ étirement (de la finale)”. Item 2.4 ʾAbū Ḥayyān (d. 735/1344), Manhaj 28, l. 11 from Bottom; Gille 37, Item 40 (on ʾAlfiyya vs. 96) ‫إذا قلت ما ذا صنعت فإن كانت ذا موصولة لم يتسل ّط صنعت على ما قبله لأنه صلة و يكون‬ ‫الضمير محذوفا وهو معمول صنعت‬ The issue here is the difference between mā ḏā as two words and māḏā as one word. With mā ḏā the verb ṣanaʿta is a qualifier (relative), so in mā ḏā ṣanaʿta the verb ṣanaʿta cannot “be given power” over the preceding word (ḏā) as if it were a simple inverted sentence, and therefore we must assume an elided resumptive pronoun in the relative clause, *mā ḏā ṣanaʿtahu ‘what is that which you did’, where the suffix -hu is operated on (maʿmūl) by ṣanaʿta. All this in contrast with the purely interrogative māḏā ṣanaʿta ‘what did you do?’. Gille translates these lines as follows:

7 Another solution to the syntax of exceptive sentences is offered by al-Jawharī, who is credited with inventing the term mafʿūl dūnahu for this construction, see Qaṭr 201 = Goguyer 218.

sallaṭa/tasallaṭa, a possible parallel for ‘govern’?

95

so regiert, wenn ḏā relativisch ist, ṣanaʿta nicht über das, was vor ihm steht, weil es sich um einen syndetischen Relativsatz handelt.

3

Specific Grammatical Topics Involving sallaṭa

The first four items in this group all deal with the same topic, the hybrid sentence type zaydan ḍarabtuhu, which is neither an inverted verbal sentence zaydan ḍarabtu nor a compound nominal sentence with Zayd as topic, zaydun ḍarabtuhu. The dependent form of zaydan is accounted for by ištiġāl, that is, the verb is ‘distracted’ or ‘preoccupied’ by the suffixed pronoun in ḍarabtuhu from operating on zaydan as a preposed direct object, leaving zaydan syntactically stranded. It can only be explained as the object of a hypothetical preceding ḍarabtu which has been elided, *ḍarabtu zaydan ḍarabtuhu. Item 3.1 *Ibn Mālik (d. 672/1274), ʾAlfiyya, Goguyer, Index 286 sallaṭa ʿalā = “attribuer régime (à un mot) sur (un autre)” The Ibn Mālik quotation is not genuine data, since he does not, as far as I can see, use the term sallaṭa in the ʾAlfiyya, but it is listed in Goguyer’s glossary to his 1888 edition and translation of that work. It could have strayed into the ʾAlfiyya glossary from Goguyer’s translation of the Qaṭr al-nadā of Ibn Hišām, published the year before (1887), where sallaṭa appears several times. The link may be that in his Qaṭr translation Goguyer refers to commentaries on verse 255 of the ʾAlfiyya, where ištiġāl is dealt with, and one of those commentaries, by Ibn ʿAqīl, uses sallaṭa (see item 3.3 below). The next three quotations, however, contain genuine examples of sallaṭa and can be taken together: Item 3.2 ʾAbū Ḥayyān, Manhaj 119, l. 10 (on Alfiyya vs. 257) ‫فمثال الاول ز يد ًا ضر بته لأنك لو لم تشغل ضر بت بالضمير لتسل ّط على الاسم فنصبه فقلت‬ ‫ز يد ًا ضر بت‬

96

carter

Item 3.3 Ibn ʿAqīl (d. 769/1367), Šarḥ al-ʾAlfiyya 226; Dieterici 134 (on ʾAlfiyya vs. 255) ‫وكل من ضر بت ومررت لو لم يشغل بالضمير لتسل ّط على ز يد كما تسل ّط على الضمير فكنت‬ ‫تقول ز يد ًا ضر بت‬ Dieterici’s paraphrase: “Fände die Zurückhaltung nicht durch das Pronomen statt, so würde das Verbum zaidân [sic] regieren, wie es jetzt das Pronomen regiert.” Item 3.4 Ibn Hišām (d. 761/1360), Qaṭr 192 = Goguyer 201 ‫ل في ضميره و يكون ذلك الفعل بحيث‬ ٌ ‫ل عام‬ ٌ ‫ضابط هذا الباب أن يتقدم اسمٌ و يتأخر عنه فع‬ ‫سل ّط على الاسم الاول لنصبه‬ ُ ‫لو فرُ ّغ من ذلك المعمول و‬ See also Qaṭr 194, 196, 197 = Goguyer 204, 208, 209, and Muġnī 2, 159 (bis). All three are statements of the same principle, indeed it looks as though Ibn ʿAqīl is directly paraphrasing ʾAbū Ḥayyān, “an example of the first type is zaydan ḍarabtuhu ‘Zayd I hit him’, for if you had not distracted ḍarabtu by means of the pronoun it would have been given power over the preceding noun and made it dependent, and you would have said zaydan ḍarabtu”. Ibn Hišām’s formulation in item 3.4 is more verbose but still essentially an elaboration of the earlier definitions of Ibn ʿAqīl and/or ʾAbū Ḥayyān, and he may well have been inspired by one or both of them. Item 3.5 Jarmānūs Farḥāt (d. 1732), Baḥṯ al-maṭālib 3, 156 ‫ولهذا س ُم ّي الاشتغال؛ لأنه لولا الضمير لتسل ّط الفعل على الاسم ونصبه‬ This can be translated as “This is why it is called ištiġāl, because if it were not for the pronoun the verb would have had power over the preceding noun and made it dependent”. Note the similarity with the wording of ʾAbū Ḥayyān in item 3.2. With the next example we turn from ištiġāl to a related topic, tanāzuʿ, the conflict of operators in the sentence type ḍarabanī wa-ḍarabtu zaydun or zaydan.

sallaṭa/tasallaṭa, a possible parallel for ‘govern’?

97

Item 3.6 Jarmānūs Farḥāt (d. 1732), Baḥṯ al-maṭālib 3, 147 ‫لا يمكن تسليط العاملين على معمول واحد‬ This is a general statement about tanāzuʿ which Jarmānūs Farḥāt seems to have devised on his own initiative. It clearly echoes Ibn Hišām’s definitions of other dependent forms, but Ibn Hišām himself does not use the term sallaṭa in his treatment of tanāzuʿ (see further below, Concluding Remarks 6.3).

4

Sallaṭa and the mafʿūl muṭlaq

Item 4.1 Ibn Hišām, Qaṭr 224 = Goguyer 240 ‫ل من لفظهكضر بت ضر باً او من معناه‬ ٌ ‫والمفعول المطلق وهو المصدر الفضلة المسل ّط عليه عام‬ ‫سا‬ ً ‫كقعدت جلو‬ The term sallaṭa is very prominent in Ibn Hišām’s definition of the mafʿūl muṭlaq, and he repeats it several times, in Qaṭr 224, 225 = Goguyer 241, 242, and al-Jāmiʿ al-ṣaġīr fī l-naḥw 106. Since we do not find this in earlier treatments of the mafʿūl muṭlaq it may be an innovation of Ibn Hišām’s, which we may translate as follows: “The mafʿūl muṭlaq is the verbal noun, structurally redundant, over which some operator has been given power, and having the same radicals as the verb, for example ḍarabtu ḍarban, or the same meaning, for example qaʿadtu julūsan”. An important feature of this definition is that the ʿāmil is not specified, “some operator”, because it does not have to be a verb, or even formally expressed at all. Item 4.2 Jarmanūs Farḥāt, Baḥṯ al-maṭālib 3, 175 ‫المفعول المطلق هو المصدر المسل ّط عليه إما عامل من لفظه أو من معناه‬ Here Jarmānūs Farḥāt is simply echoing Ibn Hišām’s wording.

98

carter

Item 4.3 Anon., al-ʾAjwiba l-jaliyya fī l-ʾuṣūl al-naḥwiyya, Beirut 50 f. ‫ ما هو المفعول المطلق؟‬.‫س‬ ‫ إن المفعول المطلق هو المصدر المسل ّط عليه عامل إما من لفظه ومعناه مثل ضر بته ضر با‬.‫ج‬ ‫فضر با مصدر منصوب مسل ّط عليه عامل موافق له في لفظه ومعناه وهو ضرب وإما من معناه‬ ‫فقط نحو قعدت جلوسا فجلوسا مصدر منصوب مسل ّط عليه عامل من معناه لا من لفظه وهو‬ ‫قعد‬ Item 4.4 Anon., al-ʾAjwiba l-jaliyya fī l-ʾuṣūl al-naḥwiyya, Malta 103 ‫ ما هو المصدر؟‬.‫س‬ ‫ المصدر هو الاسم المنصوب الذي يجيء ثالثا في تصر يف فعله ]…[ تقول ضرب يضرب‬.‫ج‬ ‫ضر با‬ Both quotations in items 4.3 and 4.4 are taken from the same work, or rather two works with the same title. This is a conversion of the ʾĀjurrūmiyya into a catechism, whose ultimate authorship and publication history cannot be verified: the two editions quoted here, Beirut and Malta 1841, are the earliest we can be sure of.8 As the extracts show, there are striking textual differences between the two, both in wording and the arrangement of contents. It is obvious that the Beirut version in item 4.3 is a direct paraphrase of Ibn Hišām’s definition in item 4.1 and Jarmānūs Farḥāt in item 4.2, and equally obvious that the Maltese version in item 4.4 is word for word the same as that of the original ʾĀjurrūmiyya (Carter 1981:342 = §17.1). However, the anonymous preface to the Maltese edition claims that its contents have been extracted from the Baḥṯ al-maṭālib of Jarmānūs Farḥāt, even though the passage we have is clearly not from that work. This is a contradiction which will need some effort to resolve. Why the two different versions appeared at all we may never know, but it is significant that the more abstract Beirut formulation involving sallaṭa moves the work out of its elementary level (the ʾĀjurrūmiyya was written for infants) to something pedagogically more advanced. 8 Brockelmann, GAL S 2, 332, mentions Muḥammad Beg Talḥūq [al-Lubnānī] as the author of a Maltese edition of 1831, but no such work can be found. This author is indeed known, but was still a teenager in 1870, as he states himself in his similarly titled al-ʾAjwiba l-jaliyya fī l-ʾuṣūl al-ṣarfiyya, Beirut 1870, with which our ʾAjwiba has possibly been confused.

99

sallaṭa/tasallaṭa, a possible parallel for ‘govern’?

5

Sallaṭa and the mafʿūl fīhi

Item 5.1 Ibn Hišām, Qaṭr 229 = Goguyer 246 (and Again in Qaṭr 229 = Goguyer 247) ‫سل ّط عليه عامل على معنى في‬ ُ ‫ وهو ما‬،‫والمفعول فيه‬ Note the similarity with Ibn Hišām’s definition of the mafʿūl muṭlaq in 4.1, suggesting that he is using that formula as a template for defining the mafʿūl fīhi as “that [dependent noun] over which some operator has been given power, with the meaning of ‘in’”. Here, too, “operator” is left unspecified because it does not have to be formally a verb. Item 5.2 al-Širbīnī (d. 977/1570), Nūr al-sajiyya 352 (= 18.1 and n. 2) ‫ خرج بذلك بقية المفاعيل لأن تسل ّط‬.‫ظرف الزمان هو الاسم المنصوب بتقدير في الظرفية‬ ‫العامل ليس على معنى في‬ The relevant part of al-Širbīnī’s definition has been translated (Carter 1981: 352) as “thus excluding the rest of the objects, because the power exercised by their operators is not from the meaning of fī ‘in’ ”. This might well be a paraphrase of Ibn Hišām, perhaps directly, rather than through intermediate sources, as al-Širbīnī was certainly familiar with his Qaṭr al-nadā and Muġnī l-labīb.

6

Concluding Remarks

The data presented here were collected haphazardly over a long period, and cannot be considered as exhaustive, particularly because no attempt has been made to track this concept on the Internet (more than 700,000 hits for taslīṭ al-fiʿl, 600,000 for tasalluṭ al-ʿāmil, for example). The conclusion of this paper will therefore consist of half a dozen broad assertions based on the limited evidence presented. 6.1 The History of Its Appearance We start with a negative observation that sallaṭa is not found in the Kitāb of Sībawayhi, nor in the Maʿānī l-Qurʾān of al-Farrāʾ, and a hasty search of likely places in the Muqtaḍab of al-Mubarrad (d. 286/898) failed to turn up

100

carter

any specimens. We will borrow an idea here from Brunschvig (1956), “Simples remarques négatives sur le vocabulaire du Coran”, where a surprising number of words are listed, and indeed entire roots, which are absent from the Qurʾān. Among them are terms for logical classification and categories, such as jins, nawʿ, ṣinf, fann, ḍarb. Brunschvig speculates that the reason for this might be that Arab culture had not yet reached a sufficient level of abstraction, but this is not the only possible explanation: I would propose instead that when sallaṭa did eventually appear, it was not filling a vacuum (as Brunschvig’s approach implies), but simply extending an aspect of the system which Sībawayhi and al-Farrāʾ had developed as far as was necessary for their own purposes under the concept of ʿamal, where there was no need for sallaṭa. It is not until the 4th/10th century that sallaṭa is firmly attested, in al-Zajjājī (d. 337/949) and al-Rummānī (d. 384/994). With a curious hiatus in the 7th/12th century (unless we count Ibn Mālik in 3.1), sallaṭa then occurs with increasing frequency, reaching a peak in the works of Ibn Hišām (d. 761/1360). The examples from *Ibn Qutayba in 1.1 and *al-Zajjāj in 2.1 do not materially change the picture: although they cannot be authenticated, they still have relevance as an illustration of what sallaṭa did mean at some time or another. 6.2 How the Term Was Used A striking feature of sallaṭa, which stands out even in the small body of evidence we have, is that in its early stages it occurred indifferently with either the speaker or the speech element as its agent, possibly reflecting a linguistic continuum which is very marked in the oldest grammars. By the time of Ibn Hišām, however, there is a clear tendency for sallaṭa to appear impersonally, in the passive (sulliṭa, musallaṭ) and the equally impersonal verbal noun taṣlīṭ, or its Stem V correlate tasallaṭa, tasalluṭ. The term taslīṭ al-fiʿl, used several times by Ibn Hišām, might represent the promotion of the concept to a higher technical level, identifying a new syntactical relationship (examples in Qaṭr 196, 197 = Goguyer 208, 209, and cf. item 3.4 above). The more general expression tasalluṭ al-ʿāmil later made its appearance, for example in al-Širbīnī, item 5.2 above. 6.3 The Connection of sallaṭa with ʿamal If sallaṭa was indeed felt to be a technical term, it remains to identify its grammatical application. I had hoped to show that sallaṭa was brought into the grammatical vocabulary to replace the term ʾaʿmala ‘to cause a word to operate’, as in Sībawayhi and al-Farrāʾ. Unfortunately Ibn Hišām, the most prominent user of sallaṭa, himself

sallaṭa/tasallaṭa, a possible parallel for ‘govern’?

101

prefers ʾaʿmala in the context of tanāzuʿ, the conflict of operators seen in the sentence type ḍarabtu wa-ḍarabanī ʾaḫawāka. His analysis is constructed without invoking sallaṭa at all (unlike Jarmānūs Farḥāt above, item 3.6), which suggests that for him ʿamal presupposed tasalluṭ here, that is, the issue is not whether an element has the power to operate, but which element will exercise that power, hence this grammatical phenomenon, he says, is also labelled bāb al-ʾiʿmāl. As he puts it (Qaṭr 198f. = Goguyer 212 f.): “there is no disagreement that any operator you like can be made the operator: the disagreement is only about which one to choose”:9 ‫لا خلاف في جواز إعمال أي العاملين او العوامل شئت وإنما الخلاف في المختار‬ We therefore have to look for a meaningful theoretical application for sallaṭa which does not duplicate ʾaʿmala in the sense of “causing to operate”. Since the conferral of grammatical power, taslīṭ, is done by the languageuser, we might separate sallaṭa from ʾaʿmala by noting that even though it is the speaker who makes a word operate in ʾaʿmala, the operation itself, ʿamal, is determined by the rules of the language, and the speech elements must comply with them. By contrast sallaṭa can be seen as denoting the speaker’s ability to confer the general power to operate on a chosen class of words, which, when operation does occur as ʿamal, will follow the rules for that syntactic category. A rather crude analogy may help here: a driving licence confers the authority to drive, but the way a person drives is subject to the rules of the road. This would rule out a translation of tasalluṭ al-ʿāmil as “[the process of] governing by the operator”, which is a tautology, two different ways of expressing the same effect of one word upon another. A better translation is proposed at the end of this paper. All this assumes that sallaṭa was in fact a fully developed technical term, which is by no means certain: merely on the grounds of its rarity it must be considered marginal. It did not become one of the ʾuṣūl al-naḥw, and it is not admitted into the specialist dictionaries as a technical term, from as early as the Mafātīḥ al-ʿulūm of al-Ḫwārazmī (d. 387/997) to late works such as the Kulliyyāt of al-Kaffawī (d. 1094/1683) and the Kašf iṣṭilāḥāt al funūn of al-Tahānawī (active 1158/1745). Its marginality is reinforced by the fact that even Ibn Hišām, its most prolific user, applies sallaṭa to only three of the fifteen manṣūbāt, and it would be a real

9 As might be expected, this turns out to be one of the points of dispute between the Baṣrans and Kūfans, Qaṭr 199 = Goguyer 213.

102

carter

challenge to account for this selectivity on his part, especially as sallaṭa is not used for any other kinds of grammatical relationship, as it had been in some earlier sources such as al-Rummānī in items 1.3 and 2.3. 6.4 Western Translations These tend to confuse the issue by rendering sallaṭa arbitrarily as “govern”, synonymous with ʿamila, thus “regieren” in Dieterici’s translation of Ibn ʿAqīl (item 3.3), and Gille’s “regiert … nicht” = “does not govern” for ʾAbū Ḥayyān’s lam yatasallaṭ (item 2.4). Even when an attempt is made to distinguish the underlying notion of sallaṭa from that of ʿamal, the latter is invariably rendered “government” anyway, e.g. Troupeau’s “imposent le régissant” for Rummānī’s tusalliṭu l-ʿāmil in ex. 2.3, and Goguyer in his definition of sallaṭa in his glossary to Ibn Mālik (item 3.1), “attribuer régime (à un mot) sur (un autre)”. On the other hand Goguyer’s translations of taslīṭ [al-fiʿl] in the Qaṭr al-nadā (208, 209, in the context of ištiġāl, see item 3.4) do avoid the pitfall of equating it with “régime”, and are quite literal: “on donne action au verbe”, or “recevoir action sur le nom”. 6.5 External Connections The possibility of influence from Western culture cannot be dismissed out of hand, but it would be difficult to document. Mere terminological similarities are not convincing, and can be downright misleading in the case of ʿamila fī and regere. If anything there is a slender chance of the movement of ideas in the opposite direction, bearing in mind that a number of Latin MSS (among them the Toledo MS of Donatus’ Ars Grammatica) are embellished with Arabic glosses inserted by their Christian readers.10 Likewise the manner of entry of sallaṭa into grammatical discourse cannot be determined: it could conceivably have been introduced independently by more than one grammarian, under varying degrees of interaction with other scholars, not necessarily Arabs or even grammarians, but conclusive evidence is lacking. A motive for such an innovation can only be guessed at: perhaps the increasing dominance of the Madrasa and the growth of scholasticism led to a qualitative change in grammatical theory, a search for a principle of linguistic authority which would compensate for the loss of access to Bedouin

10

These ideas are based on work done some time ago, see Carter (1989:33f.) and references there to Holtz (1981:384–386) and van Koningsveld (1977:49). The general position of the article at that time was that borrowings into the Christian grammatical vocabulary were unlikely. However, it was not possible in preparing this paper to ascertain whether van Koningsveld’s suggestion that these glosses “deserve to be studied separately” has ever borne fruit.

sallaṭa/tasallaṭa, a possible parallel for ‘govern’?

103

informants, as a result of which correct usage had to be generated by abstract rules (qiyās) instead of by the direct imitation of natural language (samāʿ). 6.6 Final Remarks Given the marginality of sallaṭa, its obscure origins, and its limited range of applications, the question of substantial parallels between sallaṭa and regere can be answered in the negative. There is no evidence that the Arab grammarians deliberately, still less systematically, imposed the hierarchical metaphor of ‘governing’ generically upon the linear metaphor of ‘operating’. Even the notion of martaba ‘rank’ is an extension of the indigenous legal concept of manzila ‘status’ (cf. item 2.1). We would therefore not be justified in translating tasalluṭ al-ʿāmil as “the operator’s governing power”. Until further research uncovers new material the best we can do is to interpret sallaṭa literally, as we have throughout this paper, and translate tasalluṭ al-ʿāmil simply as “the operator’s having power”, that is, the power to operate, not to govern.

Bibliographical references A

Primary Sources

ʾAbū Ḥayyān, Manhaj = ʾAbū Ḥayyān Muḥammad ibn Yūsuf al-Ġarnāṭī, Manhaj al-sālik fī l-kalām ʿalā ʾAlfiyyat Ibn Mālik. Ed. by Sidney Glazer, Manhaj al-sālik: Abū Ḥayyān’s Commentary on the Alfiyya of Ibn Mālik. New Haven, Conn.: American Oriental Society, 1947. Anon. al-ʾAjwiba l-jaliyya fī l-ʾuṣūl al-naḥwiyya. Beirut, 1841./Malta, 1841. Ibn ʿAqīl, Šarḥ al-ʾAlfiyya = Bahāʾ al-Dīn ʿAbdallāh Ibn ʿAqīl, Šarḥ ʾAlfiyyat Ibn Mālik. Ed. by Ramzī Munīr Baʿalbakī. Beirut, 1992. Ibn Hišām, al-Jāmiʿ al-ṣaġīr = Jamāl al-Dīn ʾAbū Muḥammad ʿAbdallāh ibn Yūsuf Ibn Hišām, al-Jāmiʿ al-ṣaġīr fī l-naḥw. Ed. by ʾAḥmad Maḥmūd al-Harmīl. Cairo, 1980. Ibn Hišām, Muġnī = Jamāl al-Dīn ʾAbū Muḥammad ʿAbdallāh ibn Yūsuf Ibn Hišām, Muġnī l-labīb. 2 vols. Cairo, n.d. Ibn Hišām, Qaṭr = Jamāl al-Dīn ʾAbū Muḥammad ʿAbdallāh ibn Yūsuf Ibn Hišām, Qaṭr al-nadā wa-ball al-ṣadā. Ed. by Muḥyī l-Dīn ʿAbd al-Ḥamīd. Cairo, 1963. Ibn Mālik, ʾAlfiyya = Jamāl al-Dīn ʾAbū ʿAbdallāh Muḥammad ibn ʿAbdallāh Ibn Mālik, al-ʾAlfiyya. Ed. by Antoine Isaac Silvestre de Sacy, Alfiyya, ou la quintessence de la grammaire arabe. Paris, 1833. Ed. and transl. by Antoine Goguyer, La Alfiyyah d’IbnuMalik. Beirut, 1888. *Ibn Qutayba, Talqīn = ʿAbdallāh ibn Muslim Ibn Qutayba [attrib.], Talqīn al-mutaʿallim. MS Paris 4715./Ed. by Muḥammad Salāmat Allāh Muḥammad Hidāyat Allāh. Mecca, 1986, accessed on http://ia802606.us.archive.org/21/items/talkeen1/1.pdf.

104

carter

Jarmānūs Farḥāt, Baḥṯ al-maṭālib = Jarmānūs Farḥāt, Baḥṯ al-maṭālib fī ʿilm al-ʿarabiyya. Beirut, 1865. Rummānī, Ḥudūd = ʾAbū l-Ḥasan ʿAlī ibn ʿĪsā al-Rummānī, al-Ḥudūd fī l-naḥw. Ed. by Muṣṭafā Jawād and Yūsuf Yaʿqūb Maskūnī, Rasāʾil fī l-naḥw wa-l-luġa, 37–50. Baghdad, 1969./Ed. by Batūl Qāsim Nāṣir, al-Mawrid 23, n.d., 32–47. [Accessed on https:// archive.org/details/AlhododFiNaho.] Rummānī, Šarḥ al-Kitāb = ʾAbū l-Ḥasan ʿAlī ibn ʿĪsā al-Rummānī, Šarḥ Kitāb Sībawayhi, MS Fayzullah 1984–1987. [References are to the internal volume numbers of the MS.] Sībawayhi, Kitāb = ʾAbū Bišr ʿAmr ibn ʿUṯmān Sībawayhi, al-Kitāb. Ed. by Hartwig Derenbourg, Le livre de Sībawayhi. 2 vols. Paris: 1881, 1889. (Repr., Hildesheim and New York: G. Olms, 1970.)/Ed. Būlāq, Kitāb Sībawayhi, 2 vols. Būlāq, 1898, 1900. (Repr., Baghdad, n.d.) Širbīnī, Nūr al-sajiyya = Muḥammad al-Širbīnī al-Ḫaṭīb, Nūr al-sajiyya fī ḥall ʾalfāẓ alʾĀjurrūmiyya. Ed. and transl. by Michael G. Carter, Arab linguistics: An introductory Arabic text with translation and notes. Amsterdam: J. Benjamins, 1981. *Zajjāj, ʾIʿrāb al-Qurʾān = ʾAbū ʾIsḥāq ʾIbrāhīm ibn al-Sarī al-Zajjāj [attrib.], ʾIʿrāb alQurʾān. Ed. by ʾIbrāhīm al-ʾAbyārī. 3 vols. Cairo, 1963–1965. Zajjājī, Lāmāt = ʾAbū l-Qāsim ʿAbd al-Raḥmān ibn ʾIsḥāq al-Zajjājī, Kitāb al-lāmāt. Ed. by Māzin al-Mubārak. 2nd ed. Damascus, 1985. (Accessed on http://ia802502.us .archive.org/3/items/shamela_waqfeya2/48669.pdf.) Zamaḫšarī, Mufaṣṣal = ʾAbū l-Qāsim Maḥmūd ibn ʿUmar al-Zamaḫšarī, al-Mufaṣṣal. Ed. by Jens Peter Broch. Christianiae: Libraria P.T. Mallingii, 1879. (Repr., [Baghdad], n.d.)

B

Secondary Sources

Brunschvig, Robert. 1981. “Simples remarques négatives sur le vocabulaire du Coran.” Studia Islamica 5.19–32. Carter, Michael G. 1979. “A sixteenth century grammatical experiment that failed”. Arabica 26.267–273. Carter, Michael G. 1989. “The Arabic and Medieval Latin terms for ‘governing’ ”. Speculum historiographiae linguisticae, ed. by Klaus D. Dutz, 29–36. Münster: Nodus. Dieterici, Friedrich. 1852. Ibn ʿAḳîl’s Commentar zur Alfijja des Ibn Mâlik aus dem arabischen zum ersten Mal übersetzt. Berlin: F. Dümmler. Gille, Christiane. 1995. Das Kapitel al-Mauṣūl (“Das Relativum”) aus dem Manhaǧ alsālik des Grammatikers Abū Ḥayyān al-Ġarnāṭī (1256–1344). Hildesheim, Zurich and New York: G. Olms. Goguyer, Antoine. 1887. La pluie de rosée, étanchement de la soif. Traité de flexion et syntaxe par Ibnu Hijām. Leiden: E.J. Brill. Guillaume, Jean-Patrick. 1983. “Fragments d’une grammaire oubliée: Relations prédica-

sallaṭa/tasallaṭa, a possible parallel for ‘govern’?

105

tives non assertées, verbe déclaratif et verbes modaux d’ après Sībawayhi (première partie)”. Bulletin d’Etudes Orientales 35.19–35. Holtz, Louis. 1981. Donat et la tradition d’enseignement grammatical. Étude sur l’ Ars Donati et sa diffusion (IV–IX siècle) et édition critique. Paris: CNRS. Jeffery, Arthur. 1938. The foreign vocabulary of the Qurʾān. Baroda: Oriental Institute. (Repr., Leiden and New York: E.J. Brill, 2007.) Koningsveld, Pieter Sjoerd van. 1977. The Latin-Arabic glossary of the Leiden University Library: A contribution to the study of Mozarabic manuscripts and literature. Leiden: New Rhine Publishers. Troupeau, Gérard. 1983. “Le second chapitre du ‘Livre des définitions’ d’ al-Rummāni”. Abḥāth 31.121–138. (Repr., Études sur la grammaire et la lexicographie arabes: Recueil d’articles sélectionnés. Hommage à Gérard Troupeau. Damascus: IFEAD, 2002.) Weiss, Josef. 1910. “Die arabische Nationalgrammatik und die Lateiner”. Zeitschrift der deutschen morgenländischen Gesellschaft 64.349–390.

The Notion of taqdīm wa-taʾḫīr in al-Kitāb and Its Development in the Arabic Grammatical Tradition until the 4th/10th Century Hanadi Dayyeh

1

Introduction

Few researchers have studied the notion of taqdīm wa-taʾḫīr ‘hysteronproteron’ in Sībawayhi’s Kitāb. Bohas, Guillaume and Kouloughli (1990:39) use this notion to build the hypothesis that Sībawayhi follows an enunciative approach to treating linguistic issues. In fact, they use Sībawayhi’s explanation of zaydan ḏāhiban ʾaẓunnu ‘Zayd is leaving, I think’, where advancing ʾaẓunnu is linked to the speaker’s intent to show doubt after stating a fact, in order to prove this enunciative hypothesis (1990:38f.). They briefly explain that the case of taqdīm ‘anteposing’ ʾaẓunnu shows that Sībawayhi considers the utterances to be the result of specific mental strategies that the speaker follows, choosing between various possible operations to convey the intended meaning (1990:40). If speakers choose to antepose ʾaẓunnu, they intend to convey their doubts first; whereas, if they postpose ʾaẓunnu, they intend to convey first the fact that Zayd is leaving, and then to share their doubts about this fact. Bohas, Guillaume and Kouloughli conclude that Sībawayhi’s treatment of taqdīm wa-taʾḫīr—and other notions in al-Kitāb, such as the notion of muḥāl ‘absurd’—prove his enunciative approach. A hypothesis, they admit, that needs further investigation (1990:47). Baalbaki (2007:196–203), too, employs the notion of taqdīm wa-taʾḫīr in a study devoted to a comparison between Sībawayhi’s and al-Jurjānī’s (d. 471/ 1078) approach to linguistic analysis. He shows that the relation between taqdīm wa-taʾḫīr and the meaning of the utterance is evident in al-Jurjānī’s treatment (2007:197–200), whereas Sībawayhi’s position on this relation is not clear (2007:197). In fact, Baalbaki points out several instances in al-Kitāb where the effect of taqdīm wa-taʾḫīr on the form rather than the meaning is evident (2007:199f.).1

1 Ahmar, in her Master’s thesis on taqdīm wa-taʾḫīr between naḥw and balāġa, expands on Baal-

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_007

the notion of taqdīm wa-taʾḫīr in al-kitāb

107

Marogy (2010:95–149) also studies the notion of word order in Sībawayhi’s Kitāb. She uses it to show aspects of pragmatic identifiability in Sībawayhi’s definition and treatment of definiteness (2010:99–117). Marogy studies the relationship between definiteness, identifiability and word order in al-Kitāb (2010:125–136) and shows that word order in the nominal sentence takes into account the identifiability of the word. For this purpose, indefinite nouns may take the initial position in a sentence, a position that is originally assigned to definite nouns (2010:130–136). Just like Bohas et al., who use the notion of taqdīm wa-taʾḫīr to prove the enunciative hypothesis, and like Baalbaki who employs it to study the form-meaning dichotomy, the study of word order appears in Marogy as a case study to support her thesis on the complementarity of syntax and pragmatics in Sībawayhi’s Kitāb (2010:149). In the present paper, the study of the notion of taqdīm wa-taʾḫīr in al-Kitāb and its development in the Arabic grammatical tradition until the 4th/10th century will be employed to achieve two goals: first, to offer a study of the notion of taqdīm wa-taʾḫīr and some other related notions, such as ibtidāʾ ‘topicality’ and al-ḥaqq bi-l-ṣadāra ‘the right to initial position’, in al-Kitāb. And second, to study the development of the notion in the Arabic grammatical tradition and show that Sībawayhi’s approach to taqdīm wa-taʾḫīr in al-Kitāb invokes the role of the speaker as an originator and arbiter of the utterance, whereas his successors’ approach to the same linguistic phenomenon shows that the speaker is present there as a learner who is dictated what is permissible and not permissible.

2

Taqdīm wa-taʾḫīr in Sībawayhi’s Kitāb

In one of the introductory chapters of al-Kitāb, namely the chapter on the object (mafʿūl bihi), Sībawayhi deals with the notion of word order. He explains that the norm (al-ḥadd) is to place the subject before the object: “Thus, the norm is for the subject to be anteposed in the utterance, and this [i.e., the utterance] is Arabic that is good and frequent” ( fa-min ṯamma kāna ḥadd al-lafẓ fīhi ʾan yakūna l-fāʿil muqaddaman wa-huwa ʿarabiyyun jayyidun kaṯīr).2 However, Sībawayhi states that it is permissible to place the object before the subject, because the intended meaning will not be affected: baki’s theses and presents more evidence to show that al-Jurjānī’s treatment of the notion focuses on the meaning, whereas Sībawayhi’s analysis of the same notion focuses on the form of the utterances (Ahmar 2001:98). 2 Sībawayhi, Kitāb I, 15.

108

dayyeh

If you antepose the object and postpose the subject, the utterance will be analogous to the first [utterance before changing the word order]. An example of this is when you say ḍaraba zaydan ʿabdullāhi ‘he hit Zayd, ʿAbdallāh’, because by postposing it [the subject] you intended the same [meaning] you intended by anteposing it (wa-ʾin qaddamta l-mafʿūl waʾaḫḫarta l-fāʿil jarā l-lafẓ kamā jarā fī l-ʾawwal wa-ḏālika qawluka ḍaraba zaydan ʿabdullāhi li-ʾannaka ʾinnamā ʾaradta bihi muʾaḫḫarran mā ʾaradta bihi muqaddaman).3 He then points out that this change in word order is related to the speakers’ intent to advance what is of more interest and importance to them: As if they [the Arabs] antepose that which is of more importance to them to show and that which they are more interested in to show, even though both together are of importance and interest to them [the Arabs] (kaʾannahum ʾinnamā yuqaddimūn allaḏī bayānuhu ʾahammu lahum wa-hum bi-bayānihi ʾaʿnā wa-ʾin kānā jamīʿan yuhimmānihim wa-yaʿnīyānihim).4 Thus, the speakers play a main role in taqdīm wa-taʾḫīr. They decide on the order of the words in the utterance based on what is important to them to utter first, and what they want the listener to hear first. They advance a word in the utterance due to its importance to them and their intentions to communicate this word to the listener first. To look further into the role of the speaker in taqdīm wa-taʾḫīr, the various occurrences of the notion, which are scattered across al-Kitāb, will be examined. These occurrences will be grouped and classified according to the mode of communication they express. One sample utterance will be examined in detail in each mode and reference to the other examples will be made. This classification is meant not only to organize the research and facilitate the examination of the various occurrences of taqdīm wa-taʾḫīr across al-Kitāb, but also to further highlight the role of the speaker in taqdīm wa-taʾḫīr in its relation to certain modes of communication. 2.1 Taqdīm wa-taʾḫir in the Interrogative Mode (al-istifhām) In an attempt to explain why the noun comes before the verb in ʾa-zaydan laqīta ʾam bišran ‘is it Zayd [whom] you met or Bišr?’, while the norm is to

3 Sībawayhi, Kitāb I, 14. 4 Sībawayhi, Kitāb I, 15.

the notion of taqdīm wa-taʾḫīr in al-kitāb

109

place the verb first in questions, Sībawayhi states that the speaker anteposes the noun/name when asking about the person that the recipient met. The speaker knows that the meeting has happened, but is not sure if the recipient has met Zayd or Bišr, so the name is placed before the verb to communicate to the listener that this is the specific piece of information that is sought after: Know that if you intend this meaning, then anteposing the name is better because you are not asking about the [act of] meeting, but about one of the two names [of the two people], as you do not know which one of them is the one [whom the listener met], so you start [the utterance] with the name [of the person] since you intend the endeavor to fetch the listener to inform you which of the two names is with him, and you make the other name equivalent to the first (wa-ʿlam ʾannaka ʾiḏā ʾaradta hāḏā l-maʿnā fa-taqdīm al-ism ʾaḥsan li-ʾannaka lā tasʾaluhu ʿan al-luqā waʾinnamā tasʾaluhu ʿan ʾaḥad al-ismayn lā tadrī ʾayyahumā huwa fa-badaʾta bi-l-ism li-ʾannaka taqṣid qaṣd ʾan yubayyina laka ʾayyu l-ismayn ʿindahu wa-jaʿalta l-ʾāḫar ʿādilan li-l-ʾawwal).5 Sībawahi further explains that if a speaker is asking whether the act of meeting Bišr has happened, the verb is anteposed. The speaker’s intentions and interests in certain information affect the choice of word order in the utterance despite what the norm dictates. Sībawayhi clarifies that certain ḥurūf ‘particles’ are followed only by the verb: “And it is the case that some particles are particles after which only the verb is mentioned” (wa-ḏālika ʾanna min al-ḥurūf ḥurūfan lā yuḏkar baʿdahā ʾillā l-fiʿl),6 and he adds that ʾalif al-istifhām is one of these particles.7 Despite the norm which dictates that ʾalif al- istifhām is supposed to be followed by a verb, the speakers place a name after the ʾalif. Their interest in directing the lis-

5 Sībawayhi, Kitāb I, 483. For similar examples cf. Sībawayhi, Kitāb I, 487 f. 6 Sībawayhi, Kitāb I, 50. 7 Sībawayhi, Kitāb I, 51. It is worth noting in this context that Sībawayhi draws an analogy between the interrogative and the imperative modes of communication to justify why the speaker starts the interrogative with a verb. Sībawayhi clarifies that these modes are similar in the sense that they both communicate a need that the listener is expected to satisfy, stating (Kitāb I, 51): “They [the Arabs] did this [adding ʾalif al-istifhām to the verb] in the interrogative because it is similar to a command in the sense that it is not obligatory and is intended to ask the listener for something that has not yet been established for the person who asks” (waʾinnamā faʿalū hāḏā bi-l-istifhām li-ʾannahu ka-l-ʾamr fī ʾannahu ġayr wājib wa-ʾannahu yurīdu bihi min al-muḫāṭab ʾamran lam yastaqirra ʿinda l-sāʾil).

110

dayyeh

tener to the answer they are looking for overrides the norm, so they change the word order to better express their interests and intentions and ensure successful communication. 2.2

Taqdīm wa-taʾḫīr in the Command and Prohibiting Mode (al-ʾamr wa-l-nahy) Just like utterances in the interrogative mode, those in the imperative and prohibitive mode are supposed to start with the verb: “Primarily, the verb comes before the noun in it [interrogative utterances], and so is the case in commanding and prohibiting [utterances]” (wa-kāna l-ʾaṣl fīhā ʾan yubdaʾa bi-l-fiʿl qabla l-ism fa-kaḏā l-ʾamr wa-l-nahy).8 The speaker, however, may choose to start the utterance with the noun, saying ʿabdullāhi ḍribhu ‘hit ʿAbdallāh’ instead of iḍrib ʿabdallāhi.9 Sībawayhi clarifies that the speaker chooses to advance the noun to draw the listener’s attention to the name: “You start with [the name] ʿAbdallāh and put it in the nominative case by virtue of topicality (ibtidāʾ), and you call the listener’s attention to him [ʿAbdallāh], so that he [the listener] knows him by name” (ibtadaʾta ʿabdallāhi wa-rafaʿtahu bi-l-ibtidāʾ wa-nabbahta l-muḫāṭab lahu li-yaʿrifahu bi-smihi).10 Here, too, the speaker decides to change the order of the words despite what the norm dictates. This decision comes as a result of the speaker’s attempt to draw the listener’s attention to the name of the person to whom the commanding and prohibiting act is addressed. The speakers’ interest in successful communication overrides the norm. They originate a structure that does not follow the norm but ensures successful communication of their needs and interest. 2.3 Taqdīm wa-taʾḫīr in the Informing Mode (al-ʾiḫbār) The speaker, when informing in instances such as kāna zaydun ḥalīman ‘Zayd was gentle’, may choose to put zayd before ḥalīm or place ḥalīm before zayd. According to Sībawayhi, the two utterances are the same in terms of conveying the intended meaning. He adds that there is a slight difference, though: if the speaker chooses to advance zayd, the utterance starts with the name that is known to the listener. The listener, in response, identifies the name and waits to hear the news/information about Zayd: If you say kāna zaydun then you have started with what is known to him [the listener] and to you likewise, so he waits for the predicate, and when 8 9 10

Sībawayhi, Kitāb I, 69. Sībawayhi, Kitāb I, 69. Sībawayhi, Kitāb I, 69.

the notion of taqdīm wa-taʾḫīr in al-kitāb

111

you then say ḥalīman, you have conveyed to him what you know ( fa-ʾiḏā qulta kāna zaydun fa-qad ibtadaʾta bi-mā huwa maʿrūfun ʿindahu miṯlahu ʿindaka fa-ʾinnamā yantaẓiru l-ḫabar fa-ʾiḏā qulta ḥalīman fa-qad ʾaʿlamtahu miṯla mā ʿalimta).11 On the other hand, when the speaker chooses to advance ḥalīman, the intention is to communicate first the attribute of the person talked about. The listener in this case expects to hear the name of the person next, so the name is postposed, even though it is the topic in the utterance: But if you say kāna ḥalīman, he expects you to introduce him to the person with that trait. It [the name of the person] is indeed the topic, even if it is postposed in the utterance (wa-ʾiḏā qulta kāna ḥalīman fa-ʾinnamā yantaẓiru ʾan tuʿarrifahu ṣāḥib al-ṣifa fa-huwa mabdūʾun bihi fī l-fiʿl wa-ʾin kāna muʾaḫḫaran fī l-lafẓ).12 It is clear that Sībawayhi’s approach to the hysteron-proteron construction invokes the role of the speaker as the sole arbiter of the utterance. It is the speaker’s choice that determines the word order, and this choice is made according to the speaker’s intentions and interest in what to present first to the listener. The speakers thus create an utterance that does not necessarily conform to the norm. They enjoy, in this sense, the freedom to create the utterances in ways that meet their needs rather than comply with the standard. In this context, it is important to discuss two notions that might run counter to this freedom, the notion of topicality (ibtidāʾ) and the notion of the right to initial position (al-ḥaqq bi-l-ṣadāra). These two notions imply the existence of certain limitations on the speaker’s choice of taqdīm or taʾḫīr. They both seem, at a first glance, to mean that certain words need to be placed at the beginning of the utterance, so that the speaker has no choice but to conform to that. However, a close examination of these two notions as they are present in al-Kitāb refutes such an assumption, as sections 2.4 and 2.5 will show. 2.4 The Notion of ‘Topicality’ (ibtidāʾ) in al-Kitāb Sībawayhi uses the term ibtidāʾ in al-Kitāb to convey two meanings.13 The first refers to the relationship that is established between a noun—the topic— 11 12 13

Sībawayhi, Kitāb I, 22. See a similar example of hysteron-proteron with ʾinna in Sībawayhi, Kitāb I, 285. Sībawayhi, Kitāb I, 22. Marogy (2010:178–200) studies the notions of mubtadaʾ and ibtidāʾ in al-Kitāb in light of

112

dayyeh

called mubtadaʾ or musnad, and a predicate, called ḫabar or musnad ʾilayhi,14 which relays information about the topic: The topic is every noun that is initiated to build the speech on; [both] the topic and what is built on it are nominative. Topicality exists only when something is built on it [the topic], so the topic is first and what is built on it comes after: these are musnad and musnad ʾilayhi ( fa-l-mubtadaʾ kull ismin ubtudiʾa li-yubnā ʿalayhi kalāmun wa-l-mubtadaʾ wa-l-mabniyyu ʿalayhi rafʿun fa-l-ibtidāʾ lā yakūnu ʾillā bi-mabniyyin ʿalayhi fa-l-mubtadaʾ al-ʾawwal wa-l-mabniyyu mā baʿdahu ʿalayhi fa-huwa musnad wa-musnad ʾilayhi).15 As to the second meaning, Sībawayhi uses the term ibtidāʾ to indicate the nominative case. He says: Know that the primary state of the noun is the nominative (ibtidāʾ), however, accusative, nominatives—other than topicality—and genitives affect the topic (mubtadaʾ) (wa-ʿlam ʾanna l-ism ʾawwal ʾaḥwālihi l-ibtidāʾ wa-ʾinnamā yadḫulu l-nāṣib wa-l-rāfiʿ siwā l-ibtidāʾ wa-l-jārr ʿalā l-mubtadaʾ).16 Sībawayhi explains this as follows: This is when you say ʿabdullāhi munṭaliqun ‘ʿAbdallāh is leaving’, if you wish you insert raʾaytu ‘I saw’, so you say raʾaytu ʿabdallāhi munṭaliqan ‘I saw ʿAbdallāh leaving’, or you say kāna ʿabdullāhi munṭaliqan ‘ʿAbdallāh was leaving’, or marartu bi-ʿabdillāhi munṭaliqan ‘I passed by ʿAbdallāh while he was leaving’ (wa-ḏālika ʾannaka ʾiḏā qulta ʿabdullāhi munṭaliqun ʾin šiʾta ʾadḫalta raʾaytu fa-qulta raʾaytu ʿabdallāhi munṭaliqan ʾaw marartu bi-ʿabdillāhi munṭaliqan).17

14 15 16 17

the notions of ‘topicalization’ and ‘thematization’. She states that the notion of ibtidāʾ appears in al-Kitāb to coincide with the notion of ‘theme’ (2010:182), whereas mubtadaʾ in al-Kitāb conveys the meaning of ‘topic’ (2010:183). On the terms musnad, musnad ʾilayhi and ʾisnād see Levin (1981:145–165). Sībawahi, Kitāb I, 278. Sībawayhi, Kitāb I, 7. Sībawayhi, Kitāb I, 7.

the notion of taqdīm wa-taʾḫīr in al-kitāb

113

The two meanings are, in fact, interconnected by a cause-effect relationship. Ibtidāʾ in the first meaning refers to the relationship between a topic and a predicate. In this relationship the noun acquires the nominative case by virtue of being in its original state not affected by accusatives or genitives. The noun in this ibtidāʾ state is referred to as mubtadaʾ, and while identifying three types of predicates to this mubtadaʾ Sībawayhi alludes to the fact that the latter should be placed first: Know that the topic should have that which is built on it [the predicate] to be something that is identical to it or happening in a place or time, and each of these three [types of predicate] is mentioned after initiating [the topic] (wa-ʿlam ʾanna l-mubtadaʾ lā budda lahu ʾan yakūna l-mabniyyu ʿalayhi šayʾan huwa huwa ʾaw yakūna fī makānin ʾaw zamānin wa-hāḏihi l-ṯalāṯatu yuḏkaru kullu wāḥidin minhā baʿdamā yubtadaʾ).18 Such references to the mubtadaʾ being first (ʾawwal) suggest that it is the first word in the utterance. In which case, it should be placed first at all times and therefore the speaker is not free to move it to a different position in the utterance. However, an in-depth examination of Sībawayhi’s use of the term ʾawwal in his book reveals that this suggestion is not correct. Sībawayhi uses the term first to refer to a state rather than a position or a place. By using this term, he refers to an original state (ʾaṣl) that characterizes some words. He also utilizes such a characteristic to organize these words in what Baalbaki (1979:15–20) refers to as a hierarchical relationship to differentiate between the notions that these words signify, like lightness and heaviness (ṯiqal wa-ḫiffa), singular and plural (wāḥid wa-jamʿ), and definite and indefinite (maʿrifa wa-nakira). According to Sībawayhi, “verbs are heavier than nouns because nouns are [hierarchically] first” ( fa-l-ʾafʿāl ʾaṯqal min al-ʾasmāʾ li-ʾanna l ʾasmāʾ hiya l-ʾawwal).19 Similarly, “singular is more declinable than plural because it is [hierarchically] first” (wa-l-wāḥid ʾašadd tamakkunan min al-jamʿ li-ʾanna l-wāḥid al-ʾawwal)20 and “masculine is lighter for them [the speakers] than feminine because masculine is [hierarchically] first” (wa-l-muḏakkar ʾaḫaff ʿalayhim min al-muʾannaṯ li-ʾanna l-muḏakkar ʾawwal).21 In fact, Sībawayhi states clearly that the term ‘first’ used with mubtadaʾ conveys the same meaning that is implied in dealing with the notions of number and (in)definiteness: “The topic is first in the 18 19 20 21

Sībawayhi, Kitāb I, 278. Sībawayhi, Kitāb I, 6. Sībawayhi, Kitāb I, 7. Sībawayhi, Kitāb I, 7.

114

dayyeh

same sense that one is the first of all numbers and indefinite is before definite” ( fa-l-mubtadaʾ ʾawwal kamā kāna l-wāḥid ʾawwal al-ʿadad wa-l-nakira qabla lmaʿrifa).22 Thus, the mubtadaʾ is not necessarily the first word in the utterance. It is, however, the origin and the base upon which speech is built, regardless of where it is placed in the utterance. In this context, Sībawayhi explains that the norm is to position the mubtadaʾ at the beginning of the utterance, but this is not necessarily always the case. The speaker may say qāʾimun zaydun ‘Zayd is standing up’ instead of zaydun qāʾimun. Sībawayhi clarifies that anteposing the predicate is not the norm, yet it is permissible and considered good Arabic.23 Therefore, the topic is not identified by its place in the sentence, but rather by being the origin or the foundation upon which the predicate is constructed. To conclude this section, the notion of ibtidāʾ in al-Kitāb refers to a relationship between a topic and a predicate. In this relationship, the order of the words does not matter; and the speaker is free to change this order, provided that a topic is not uttered without a predicate or vice versa. Sībawayhi explains that in order to ensure successful communication the speaker needs both the musnad and the musnad ʾilayhi, as is clear in the chapter heading: “This is the chapter on musnad and musnad ʾilayhi, either of which cannot do without the other, and the speaker cannot find a way without it [uttering both of them]” (hāḏā bāb almusnad wa-l-musnad ʾilayhi wa-humā lā yastaġnī waḥid minhumā ʿan al-ʾāḥar wa-lā yajidu l-mutakallim minhu buddan).24 2.5 The Notion of the Right to Initial Position (al-ḥaqq bi-l-ṣadāra) In a chapter with the heading “This is the chapter of that which takes the place of a noun that is mubtadaʾ and substitutes for it” (hāḏā bāb mā yaqaʿu mawqiʿ alism al-mubtadaʿ wa-yasuddu masaddahu),25 Sībawayhi identifies some words that may be posed at the beginning of the utterance, regardless of the fact that these words are not the topic. Such a word is placed at the beginning “because it is a location and a position for what comes after” (li-ʾannahu mustaqarrun limā baʿdahu wa-mawḍiʿ),26 as in saying fīhā ʿabdullāhi ‘ʿAbdallāh is in it’, ṯamma zaydun ‘Zayd is there’, hāhunā ʿamrun ‘ʿAmr is here’. They may also be interrogative words, such as ʾayna zaydun ‘where is Zayd?’, kayfa ʿabdullāhi ‘how is ʿAbdallāh’, since, as Sībawayhi explains “the meaning of ʾayna is ‘in what 22 23 24 25 26

Sībawayhi, Kitāb I, 7. Sībawayhi, Kitāb I, 278. Sībawayhi, Kitāb I, 7. Sībawayhi, Kitāb I, 278. Sībawayhi, Kitāb I, 278.

the notion of taqdīm wa-taʾḫīr in al-kitāb

115

place’, and the meaning of kayfa is ‘in what state’. This [kind of words] can only be started with before the noun because they are interrogative particles” ( fa-maʿnā ʾayna fī ʾayyi makān wa-kayfa ʿalā ʾayyati ḥāl wa-hāḏā lā yakūnu ʾillā mabdūʾan bihi qabla l-ism li-ʾannahā min ḥurūf al-istifhām).27 As to the first type, the words referring to location replace the initial noun (mubtadaʾ) and are placed at the beginning of the utterance because together with the noun coming after them they form an utterance that communicates the intended meaning, which neither one of them alone can communicate: However, neither one of them is sufficient without its companion, so when they are combined they suffice to stop talking, so that they become similar with respect to sufficiency, as when you say hāḏā ʿabdullāhi ‘this is ʿAbdallāh’ (wa-lākinna kulla wāḥidin minhumā lā yustaġnā bihi ʿan ṣāhibihi fa-lammā jumiʿā staġnā ʿalayhimā l-sukūtu ḥattā ṣārā fī l-istiġnāʾ kaqawlika hāḏā ʿabdullāhi).28 As to the second type, interrogative words are also placed at the beginning of the utterance to serve a communicative purpose. The speaker starts the utterance with the question nouns or articles to communicate to the listener that a question is being asked. Therefore, the right of these words to be placed at the beginning is decided upon by the speaker, who chooses to do so to communicate a certain mode of speech to the listener. Thus, the notion of al ḥaqq bi-lṣadāra does not pose a restriction on the speaker’s choice to change the word order in the utterance. It is, in fact, a right that the speaker uses to serve certain communicative purposes. It is clear that Sībawayhi’s speaker is the sole arbiter who decides on the order of the words in an utterance. It is the speaker’s interest and intent that determine this word order. It is also clear that Sībawayhi’s approach to analyzing the notion of taqdīm wa-taʾḫīr invokes the role of the speaker as an arbiter. In what follows, we will trace the notion of taqdīm wa-taʾḫīr as it evolved after Sībawayhi in the writings of his successors in the 3rd/9th and 4th/10th centuries.

27 28

Sībawayhi, Kitāb I, 278. Sībawayhi, Kitāb I, 278.

116 3

dayyeh

Taqdīm wa-taʾḫīr after al-Kitāb

Two sources will be examined to trace the development of the notion of taqdīm wa-taʾḫīr after Sībawayhi. The first is al-Mubarrad’s (d. 285/898) al-Muqtaḍab. This book is considered key to studying the development of the Arabic linguistic tradition as it adopts Sībawayhi’s content almost in its entirety and exhibits a shift to a greater interest in the notions of analogy (qiyās), operant (ʿāmil) and cause (ʿilla) in approaching linguistic issues.29 The second source is alʾUṣūl fī l-naḥw by Ibn al-Sarrāj (d. 316/928). This book is the first source in the Arabic linguistic tradition that is dedicated to the foundations (al-ʾuṣūl). It is a landmark in the tradition as it represents a shift towards establishing the foundations of the rules of Arabic grammar.30 Our examination of the notion of taqdīm wa-taʾḫīr in these two sources will focus on finding out if Sībawayhi’s approach to the same notion as it relates to the role of the speaker continues in the writings of his successors. 3.1 Taqdīm wa-taʾḫīr in al-Muqtaḍab Although the notion of taqdīm wa-taʾḫīr appears in various places in al-Muqtaḍab, al-Mubarrad’s position on hysteron-proteron is explicitly presented in a chapter that deals with the transitive verbs of doubt and certainty like ʾaẓunnu ‘I believe’. In this chapter, al-Mubarrad explains that advancing one of the two objects of ʾaẓunnu affects the meaning of the utterance: Don’t you see that if you say ẓanantu zaydan ʾaḫāka ‘I believe Zayd [is] your brother’, the doubt falls upon the brotherhood, but if you say ẓanantu ʾaḫāka zaydan ‘I believe your brother [is] Zayd’, you cause the doubt to fall upon the name-giving (ʾa-lā tarā ʾiḏā qulta ẓanantu zaydan ʾaḫāka fa-ʾinnamā yaqaʿu l-šakk fī l-ʾuḫuwwa fa-ʾin qulta ẓanantu ʾaḫāka zaydan ʾawqaʿta l-šakk fī l-tasmiya).31 As a result of this impact on meaning, al-Mubarrad allows the anteposing or postposing of a word in an utterance only when the intended meaning is not affected by the change in word order: “Anteposing and postposing are appropriate when the utterance clarifies the meaning” (wa-ʾinnamā yaṣluḥu l-taqdīm

29 30 31

Cf Bohas et al. (1990:4f.) and Baalbaki (2008:236f.). Cf. Owens (1990:242f.) and Bohas et al. (1990:10f.). Mubarrad, Muqtaḍab III, 95.

the notion of taqdīm wa-taʾḫīr in al-kitāb

117

wa-l-taʾḫīr ʾiḏā kāna l-kalām muwaḍḍiḥan ʿan al-maʿnā).32 He clarifies that the meaning may not be affected by taqdīm wa-taʾḫīr when the case ending of the word signifies clearly the status of the word in the sentence as in saying ḍaraba zaydan ʿamrun ‘ʿAmr hit Zayd’ “because you know from the case ending [which is] the agent and [which is] the object” (li-ʾannaka taʿlamu bi-l-ʾiʿrāb al-fāʿil wa-lmafʿūl).33 Al-Mubarrad also explains that postposing or anteposing the object of the verb ʾaẓunnu may be allowed when there is no room for more than one interpretation for the utterance, as in ẓanantu fī l-dāri zaydan34 ‘I believe Zayd [is] in the house’. In this utterance, postposing the object does not affect the meaning. Al-Mubarrad thus draws the lines within which the speaker is allowed to change the order of the words in an utterance. Unlike Sībawayhi, who considers the speaker to be the main arbiter in choosing to advance or delay a word in an utterance, al-Mubarrad restricts the speaker’s choice to postposing or anteposing one of the two objects of ʾaẓunnu to those utterances where the intended meaning is not affected by the change in word order. In fact, he introduces the chapter on bi-transitive verbs by providing a list of the verbs of doubt and certainty: These verbs are verbs of doubt and certainty, such as ʿalimtu zaydan ʾaḫāka ‘I know Zayd [is] your brother’, ẓanantu zaydan ḏā mālin ‘I believe Zayd has money’, ḥasibtu zaydan dāḫilan ‘I assume Zayd [is] entering’ and ḫiltu bakran ʾabā ʿabdillāhi ‘I think Bakr [is] the father of ʿAbdallāh’ (wa-tilka l-ʾafʿāl hiya ʾafʿāl al-šakk wa-l-yaqīn naḥwa ʿalimtu zaydan ʾaḫāka wa-ẓanantu zaydan ḏā mālin wa-ḥasibtu zaydan dāḫilan wa-ḫiltu bakran ʾabā ʿabdillāhi).35 Then, al-Mubarrad explains that the speaker has to mention a second object due to the fact that the verbs of doubt and certainty need a second object in the same way a topic needs a predicate: As it is essential for the topic to have a predicate, likewise, the second object to it [i.e., one of the verbs of doubt and certainty] is unavoidable, because it [i.e., the second object] is the predicate of the topic and it is what you [the speaker] depend on in [conveying] doubt and certainty 32 33 34 35

Mubarrad, Muqtaḍab III, 95f. Mubarrad, Muqtaḍab III, 96. Mubarrad, Muqtaḍab III, 96. Mubarrad, Muqtaḍab III, 95.

118

dayyeh

( fa-kamā lā budda li-l-ibtidāʾ min ḫabar kaḏā lā budda min mafʿūlihā l-ṯānī li-ʾannahu ḫabar al-ibtidāʾ wa-huwa llaḏī taʿtamidu ʿalayhi bi-l-šakk wa-lyaqīn).36 After that, al-Mubarrad engages the speaker in an exercise where he creates a hypothetical dialogue between the speaker and another imaginary speaker. The latter asks the former for information and expects the speaker to reply using the correct form of the utterance (Muqtaḍab III, 95): If you say ẓanantu zaydan ʾaḫāka ‘I believe Zayd [is] your brother’ and he [a speaker] tells you to inform about yourself, you say al-ẓānnu zaydan ʾaḫāka nafsuka ‘the one who believes Zayd [is] your brother [is] yourself’; but when he [a speaker] tells you to inform about Zayd, you say al-ẓānnuhu ʾanā ʾaḫāka zaydan ‘I am the one who believes [that] your brother is Zayd’ ( fa-ʾiḏā qulta ẓanantu zaydan ʾaḫāka fa-qāla laka ʾaḫbir ʿan nafsika qulta al-ẓānnu zaydan ʾaḫāka nafsuka fa-ʾiḏā qāla ʾaḫbir ʿan zayd qulta al-ẓānnahu ʾanā ʾaḫāka zayd). The exercise continues when the speaker is challenged to construct more variations of the same utterance to convey different information. This exercise is used by al-Mubarrad to offer the speaker an exhaustive list of all possible structures of the utterance. The speakers appear in this situation as learners, who are engaged in an exercise intended to train them in constructing an utterance using the verbs of doubt and certainty. Sībawayhi’s speakers, on the other hand, appear as the sole arbiter who decide to antepose or postpose the verb of doubt depending on the meaning they want to convey to the listener. If they antepose ʾaẓunnu, they want to convey the doubt first, but if they postpose it, they intend to share the information with the listener first, and then express their doubts. It is worth mentioning in this context that the speaker consistently appears in al-Muqtaḍab as a learner. A study of the scheme that al-Mubarrad follows in constructing his chapters shows that he normally starts by drawing the speakers’ attention to the linguistic issue he wants to tackle. Then, he offers them an exhaustive list of all related concepts, along with examples focusing basically on the role of ʿāmil. The speaker thus appears to be a learner whom al-Mubarrad is adamant to teach about the linguistic issue in question. He makes sure to present a comprehensive explanation of the linguistic concept, to list all the possible examples, and to offer an explanation to each. Quite often, he engages

36

Mubarrad, Muqtaḍab III, 95.

the notion of taqdīm wa-taʾḫīr in al-kitāb

119

the speaker/learner in a training exercise in generating the various possible utterances related to the concept studied. In fact, he dedicates some chapters to assessing the speaker’s knowledge by presenting a complicated sentence that tests the speakers’ ability to find the operant in each word.37 It is beyond the scope of this paper to offer a study on the structure of alMuqtaḍab’s chapters in relation to the speaker appearing as a learner. It is clear, though, from the study of the notion of hysteron-proteron that there is a shift in the role of the speaker as it appears in al-Kitāb. Sībawayhi’s speaker is the arbiter and originator of the utterance, al-Mubarrad’s speaker is a learner who is didactically told how to form the utterance. This shift has impacted the development of the notion after Sībawayhi. Where the notion appears in al-Kitāb to show the speaker’s choice and freedom, it appears in al-Muqtaḍab to restrict the speaker’s choice. This shift to a more didactic approach where the speaker is treated as a learner rather than an originator of the speech will continue in the tradition to develop into a set of rules/foundations with Ibn al-Sarrāj’s alʾUṣūl, as section 3.2 will show. 3.2 Taqdīm wa-taʾḫīr in al-ʾUṣūl Ibn al Sarrāj dedicates a chapter in his ʾUṣūl to studying the notion of hysteronproteron. He starts this chapter by identifying thirteen cases where anteposing a certain word in the utterance is not allowed. Then he delves into explaining why anteposing is not allowed in each one of them. A close study of these thirteen cases shows that they all, with the exception of one case (see below), follow one basic principle, that the operant (ʿamil) cannot be placed after the operand (maʿmūl fīhi). Ibn al-Sarrāj’s detailed explanation of the thirteen cases, which could have been summed up in one basic principle, results in a list of instances that restrict the speakers and prohibit them from anteposing or postposing a word in the utterance. Most of these instances are hypothetical and not based on real utterances. The role of the speaker in these thirteen cases remains unmentioned, except in one case. In this specific case, Ibn al-Sarrāj does not allow anteposing or postposing the words in the utterance “when anteposing confuses the listener” (al-taqdīm ʾiḏā ʾulbisa ʿalā l-sāmiʿ),38 because he states that it is necessary for the speaker to form an utterance that is clear to the listener. In this context, he distinguishes between two types of utterances that might confuse the listener. The first type is when the case ending is absent in instances such as ḍaraba ʿīsā 37 38

Cf. the chapters in the Muqtaḍab (II, 62–64; IV, 59–71) entitled masāʾil ṭiwāl yumtaḥanu fīhā l-mutaʿallim ‘long [linguistic] issues for assessing the learner’. Ibn al-Sarrāj, ʾUṣūl II, 245.

120

dayyeh

mūsā39 ‘ʿĪsā hit Mūsā’. If ʿĪsā is the subject, the speaker cannot postpose it and advance Mūsā, because the listener will be confused, thinking that Mūsā is the subject. The second type of confusion results from the meaning of certain utterances such as ḍarabtu zaydan qāʾiman40 ‘I hit Zayd [while] standing up’. The listener will not be able to identify who is referred to as standing up, the speaker or Zayd. Therefore, the speaker cannot change the order of the words in similar instances. If the intention is to say that Zayd was standing up, zaydan should be placed before qāʾiman, whereas if the intention is to say that he was the one standing up, qāʾiman should be placed before zaydan. It is clear that in this chapter on hysteron-proteron Ibn al-Sarrāj attempts to exhaust all the cases that restrict the speakers and prevent them from changing the word order in the utterance. Ibn al-Sarrāj’s speaker, like al-Mubarrad’s, is addressed as a learner who is told what is permissible or not permissible in taqdīm wa-taʾḫīr. In fact, it is clear in the introduction of al-ʾUṣūl that Ibn alSarrāj intends to address this speaker/learner. In this introduction, he explicitly states that grammar (ʿilm al-naḥw) is intended to teach the speaker the way the Arabs speak: “By al-naḥw is meant that—when the speaker learns it—he would follow the example of the speech of the Arabs” (ʾinnamā ʾurida bihi ʾan yanḥuwa l-mutakallimu ʾiḏā taʿallamahu kalāma l-ʿArab).41 He explains that grammar belongs to a later stage, where advanced speakers examined the speech of the first speakers (the originators) and deduced the rules of grammar: “It is a discipline that the advanced [speakers] deduced by examining the speech of Arabs in order to grasp the purposes intended by the originators of the language” (wahuwa ʿilmun istaḫrajahu l-mutaqaddimūn fīhi min istiqrāʿ kalām al-ʿArab ḥattā waqafū minhu ʿalā l-ġaraḍ allaḏī qaṣadahu l-mubtadiʾūn bi-hāḏihi l-luġa).42 Thus, Ibn al-Sarrāj distinguishes between two speakers: a speaker who is the originator of the language, and a speaker who examines the language of the originator to learn it. In this context, he further distinguishes between two types of cause (ʿilla): a direct cause that is needed to acquire the language, like saying “every subject is nominative” (kullu fāʿilin marfūʿ),43 and a “cause of the cause” (ʿillat al-ʿilla), which is meant to reveal the wisdom of the Arabic language, like explaining why the subject is in the nominative. Ibn al-Sarrāj clarifies that in his book he will focus on the ʿilla that helps the speaker learn 39 40 41 42 43

Ibn al-Sarrāj, ʾUṣūl II, 245. Ibn al-Sarrāj, ʾUṣūl II, 245. Ibn al-Sarrāj, ʾUṣūl I, 35. Ibn al-Sarrāj, ʾUṣūl I, 35. Ibn al-Sarrāj, ʾUṣūl I, 35.

the notion of taqdīm wa-taʾḫīr in al-kitāb

121

the language: “My goal is to mention the ʿilla that, if it is consistent, he [the speaker/learner] can access their [the Arabs’] speech, and to mention what is foundational and common, because it is a book that is concise” (wa-ġaraḍī min hāḏā l-kitāb ḏikr al-ʿilla allatī ʾiḏā ṭṭaradat waṣala bihā kalāmahum fa-qaṭ waḏikr al-ʾuṣūl wa-l-šāʾiʿ li-ʾannahu kitāb ʾījāz).44 He also states clearly that he will use a style of writing that is comprehensible by the speaker/learner “since I did not create this book for the expert, but for the learner, I had to mention what is handy for the learner” (wa-lammā kuntu lam ʾaʿmal hāḏā l-kitāb li-l-ʿālim dūna l-mutaʿallim iḥtajtu ʾan ʾaḏkura mā yuqarribu ʿalā l-mutaʿallim).45 Thus, Ibn al-Sarrāj in his al-ʾUṣūl targets the speaker/learner. His book on the foundations of the language is intended to provide the learner with the required skills to acquire the language of the Arabs. This intent can be seen as a follow-up on the shift in the role of the speaker that happened after Sībawayhi, mainly in al-Mubarrad’s al-Muqtaḍab. As mentioned above, al-Mubarrad shows in his treatment of hysteron-proteron a shift in focus from the speaker as an originator of the language to the speaker as a learner of this language. This shift marks a change in the development of the notion after Sībawayhi. The speaker who is present in al-Kitāb as the sole arbiter of taqdīm wa-taʾḫīr is present in al-Mubarrad’s Muqtaḍab and later in Ibn al-Sarrāj’s al-ʾUṣūl as a learner who is didactically told what is permissible or not permissible in taqdīm wa-taʾḫīr.

Bibliographical References A

Primary Sources

Ibn al-Sarrāj, ʾUṣūl = ʾAbū Bakr Muḥammad ibn al-Sarī ibn Sahl Ibn al-Sarrāj, al-ʾUṣūl fī l-naḥw. Ed. by ʿAbd al-Ḥusayn al-Fatlī. 3 vols. Beirut: Muʾassasat al-Risāla, 2015. Mubarrad, Muqtaḍab = ʾAbū l-ʿAbbās Muḥammad ibn Yazīd al-Mubarrad, al-Muqtaḍab. 4 vols. Beirut: ʿĀlam al-Kutub, 2010. Sībawayhi, Kitāb = ʾAbū Bišr ʿAmr ibn ʿUṯmān Sībawayhi, al-Kitāb. 2 vols. Cairo: Maṭbaʿat Būlāq, 1316A.H.

44 45

Ibn al-Sarrāj, ʾUṣūl I, 36. Ibn al-Sarrāj, ʾUṣūl I, 37.

122 B

dayyeh

Secondary Sources

Ahmar, May. 2001. al-Taqdīm wa-l-taʾḫīr bayna al-naḥw wa-l-balāġa. M.A. thesis, American University of Beirut. Baalbaki, Ramzi. 1979. “Some aspects of harmony and hierarchy in Sībawayhi’s grammatical analysis”. Zeitschrift für arabische Linguistik 2.7–22. Baalbaki, Ramzi. 1983. “The relation between naḥw and balāġa: A comparative study of the methods of Sībawayhi and Ǧurgānī”. Zeitschrift für arabische Linguistik 11.7– 23. (Repr., Ramzi Baalbaki, The early Islamic grammatical tradition, 187–203. Britain: Ashgate Variorum.) Baalbaki, Ramzi. 2008. The legacy of the Kitāb: Sībawayhi’s analytical methods within the context of the Arabic grammatical tradition. Leiden: E.J. Brill. Bohas, Georges, Jean-Patrick Guillaume, and Djamel Eddine, Kouloughli. 1990. The Arabic Linguistic Tradition. London and New York: Routledge. Levin, Aryeh. 1981. “The grammatical terms al-musnad, al-musnad ilayhi and al-isnād”. Journal of the American Oriental Society 101.145–165. [Accessed 06/04/2014. http:// www.jstor.org/stable/601756 Marogy, Amal. 2010. Kitāb Sībawayhi: Syntax and pragmatics. Leiden: E.J. Brill. Owens, Jonathan. 1990. Early Arabic grammatical theory: Heterogeneity and standardization. Amsterdam and Philadelphia: J. Benjamins.

The Intriguing Issue of Dictionary Arrangement in Medieval Arabic Lexicography Joseph Dichy

1

Introduction: al-Ḫalīl’s Innovation

In an earlier study (Dichy 2014), I have highlighted the fact that the mathematical method elaborated by al-Ḫalīl ibn ʾAḥmad al-Farāhīdī (d. around 175/791) brought forth the first dictionary ever in the history of language study that aimed at covering the entire lexicon of a given language. This dictionary also paved the way for the subsequent development of Medieval Arabic comprehensive semasiological dictionaries dealing with the general vocabulary of the language. The central concept of the approach initiated by al-Ḫalīl is that of ordered sequences of two, three, four, or five letter-segments (ḥarf, pl. ḥurūf ),1 which he calls ‘constructs’ (bināʾ). He discovered the possibility of combining the letter-segments of his phonetic inventory into these ‘constructs’, thus designing a permutative matrix that covered the entire virtual vocabulary of the language, a subset of which was actually in use (Dichy 2014). The influence of alḪalīl’s method accounts for the fact that semasiological dictionaries are based on sequences of letter-segments corresponding mutatis mutandis to what we could call ‘formal roots’.2 A very intriguing question, though, seems to resist analysis: that of dictionary arrangement. In technical terms, this issue is that of the macrostructure of dictionaries, i.e. the constitution of word lists and the way in which they

1 For the analyses that led to the translation of ḥarf by ‘letter-segments’, see Dichy (1990a, b). Briefly, the basic idea is that Medieval Arabic language sciences, although they distinguish between lafẓ ‘phonic utterance’ and ḫaṭṭ ‘writing’, merge both substances into a unified notion, that of ḥarf. The word ‘segment’ refers to the phonic units that combine in lafẓ, and that of ‘letter’, to the inventory of their graphic counterparts and complements. 2 See Dichy (2003) for a strict definition of the term ‘formal root’, which does not correspond, needless to say, to the much more general and polysemic term of ʾaṣl ‘origin, principle’, ‘underlying morphological basis’. In this contribution, I may refer to ‘formal roots’ or, for short, to ‘roots’ in a somewhat anachronic way, owing to the fact that there is no name or term for that concept in Medieval Arabic language sciences.

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_008

124

dichy

are ordered and organized (as opposed to microstructure, which is concerned with the treatment of information associated with entries3). The question can be rephrased as follows: why did al-Ḫalīl and other authors of dictionaries go for a cumbersome ordering of what we call ‘formal roots’, instead of using the traditional ʾalif, bāʾ, tāʾ, ṯāʾ … alphabet and adopting an arrangement facilitating the looking up of words? Considering the order al-Ḫalīl devised for the Kitāb al-ʿayn, and that of later dictionaries elaborated under the influence of his approach, two corollary questions can be asked: i. Why resort to a phonetic order of letter-segments (ḥarf ) instead of using an alphabet that was most likely known to all learned readers? The traditional ʾalif, bāʾ, tāʾ, ṯāʾ … alphabet was certainly familiar to al-Ḫalīl himself, which can easily be inferred from the beginning of the introduction to the Kitāb al-ʿayn. (One remembers that al-Ḫalīl decided that a thorough inventory was needed, because the first letter-segment of the traditional alphabet is ʾalif, which he rightly considered as problematic, see Dichy 2014, section 3.) ii. Why did al-Ḫalīl and a number of authors of dictionaries choose such a difficult arrangement of formal roots, based on the permutation of lettersegments within sequences of two, three, or four radical consonants? John Haywood (1960:38) commented on what he called al-Ḫalīl’s ‘cumbersome’ arrangement in these terms: We may well ask ourselves what really can have induced al-Ḫalīl to invent a plan which his innate intelligence ought to have caused him to reject. […] Al-Ḫalīl could have been just as sure of including all roots using the normal alphabetic order, without anagrams, and without separating roots according to their length. But permutations are the plaything of the mathematician, and we may suspect that, once having got the idea, al-Ḫalīl could not get away from it. In spite of what the knowledge of Arabic lexicography owed for many years to Haywood’s pioneering work, these lines, as will be shown below, witness to a deep misunderstanding of al-Ḫalīl’s heuristic endeavour. The two questions above, though, remain to be answered.

3 Microstructural aspects will be analysed in a further contribution, although many points could be mentioned here.

the intriguing issue of dictionary arrangement

2

125

Phonetic Inventory vs. Traditional Alphabet

The familiar character of the ʾalif, bāʾ, tāʾ, ṯāʾ … alphabet is also shown by the fact that it was later referred to by Ibn Jinnī (d. 392/1002) in his Sirr ṣināʿat al-ʾiʿrāb in connection with the practice of educators (muʿallimūn) in their teaching of the alphabet (Sirr I, 43).4 It is to be remembered, in addition, that the alphabetic arrangement was used in the Kitāb al-jīm lexicon of rare and difficult (ġarīb) words by ʾAbū ʿAmr al-Šaybānī (d. around 206/821). This lexicon is ordered alphabetically on the basis of the first radical of formal roots or words. Baalbaki (2014: 333) considers that the Kitāb al-jīm could have been elaborated in a period very close to that of al-Ḫalīl’s Kitāb al-ʿayn. Of course, the scope of the two works consistently differs. Other dictionaries have been organized on the basis of a phonetic inventory of letter-segments. They usually feature, in addition, various types of ordering of what we call formal roots, based on permutative arrangements. The latter can also be based on the inventory of letter-segments provided by the traditional alphabet.

3

Permutative Arrangement Based on Either Phonetic or Alphabetic Inventories5

Al-Ḫalīl’s Kitāb al-ʿayn seems to have only reached Iraq and the milieu of linguists and lexicographers in the 3rd/9th century, around a century after it had

4 The point under discussion is worth mentioning, because it is related to the present-day teaching of the Arabic alphabet in schools. Ibn Jinnī recalls that the last three letters of the alphabet taught to children are wāw, lām-ʾalif, yāʾ, and that teachers were mistaken in considering that lām-ʾalif denotes the graphic symbol ‫لا‬, i.e. the writing of a double letter. His description comprises two steps. Firstly, he mentions the fact that the name of letters refer to their initial segment: thus ʾalif, the name of the first letter, does not refer to the long vowel ā, but to the phonic segment hamza (ʾ), as is the case with jīm or dāl, which denote, respectively j and d (in the present-day linguistics of writing, this is known as the acrophony principle). Secondly, Ibn Jinnī indicates that the latter principle does not apply to lām-ʾalif in the sequence closing the alphabet: lām-ʾalif refers, according to him, to the long vowel ā, which is only preceded by the consonant l because no word or syllable can begin with a vowel (Sirr I, 43). Until the late 1960s in Lebanon and other countries, the alphabet taught in schools ended with wāw, lām-ʾalif, yāʾ. Unfortunately, the second element of this sequence was later widely suppressed, in sheer ignorance of what was mentioned by Ibn Jinnī. As a result, the alphabet taught in schools in the Arab world includes 28 letters instead of 29, as was the case when lām-ʾalif was included. In a 28 letter alphabet, either hamza or the long vowel ā is omitted. 5 This section is partly indebted to Baalbaki (2014).

126

dichy

been completed by his pupils (mainly by al-Layṯ) in Ḫurāsān.6 The dictionaries that followed this ‘rediscovery’ did not adopt arrangements facilitating the looking-up of roots and words, until the end of the 4th/10th century. The first comprehensive dictionary, after the Kitāb al-ʿayn, was the Jamharat al-luġa by Ibn Durayd (d. 321/933). The work departs from al-Ḫalīl’s approach, in that it is not based on a phonetic inventory of letter-segments and features an alphabetic ʾalif-bāʾ-tāʾ-ṯāʾ … arrangement of mādda ‘dictionary super-entries’. On the other hand, Ibn Durayd retains al-Ḫalīl’s method of permutative variation of roots included under each mādda head, and orders his dictionary accordingly. The permutation is explicitly represented though the drawing of a circle on which two, three or more segment-letters are positioned ( Jamhara III, 513f.). Ibn Durayd adds to his dictionary morphological developments on augmented letters (ḥurūf al-ziyāda, Jamhara I, 10), and on a number of morphological patterns ( Jamhara III, 508–513). It is interesting to note that in the 19th century Emile Littré also included grammatical developments in his French dictionary. The following dictionary—which has not reached us in its entirety—is alBāriʿ fī l-luġa compiled by ʾAbū ʿAlī al-Qālī (d. 356/967). It is “heavily influenced by al-Ḫalīl’s system”, although it shows a number of modifications (Baalbaki 2014:303–311). The inventory of letter-segments is phonetic, and formal roots are considered though permutations. The arrangement of the dictionary is, in consequence, permutative. The 15 volumes of the Tahḏīb al-luġa, compiled by al-ʾAzharī (d. 370/981), as well as the 10 volumes of the Muḥīt fī l-luġa, compiled by al-Ṣāḥib ibn ʿAbbād (d. 385/995), also follow to quite some extent al-Ḫalīl’s method of permutation of letter-segments on the basis of a phonetic inventory, and are consequently arranged in a way that does not make consultation easy. The phonetic description of letter-segments due to al-Ḫalīl—which was criticized by Ibn Jinnī (Sirr I, 45–48)—is partly modified. Both dictionaries are, on the other hand, much more comprehensive than the Kitāb al-ʿayn when it comes to words listed under each root. The Maqāyīs al-luġa by Ibn Fāris (d. 395/1004), on the other hand, is based on an alphabetical (ʾalif-bāʾ-tāʾ-ṯāʾ …) inventory. Formal roots are dealt with according to their number of letter-segments, and are organized as comprising two, three, four or five radical consonants. They are arranged on the basis of a per6 See Talmon (1997); Schoeler (2000; 2002: 102–107); Dichy (2014) and Mahdī al-Maḫzūmī and ʾIbrāhīm al-Sāmarrāʾī’s introduction to their edition of the Kitab al-ʿayn, I, 17–27; all of these references include comments on discussions going back to the 3rd/9th and 4th/10th centuries.

the intriguing issue of dictionary arrangement

127

mutative order, the second letter being the one following the previous in the alphabet (one must imagine the letter-segments of the alphabet as positioned on a mental circle). Two-consonant formal roots are presented in the beginning of chapters. If one looks for instance, for what we consider as the formal root s-d-d, one needs to go to the chapter of formal roots whose initial lettersegment is /s/ (Maqāyīs III, 57ff.), which begins with a sub-section including bi-radicals with a doubled second radical, then follow up to /s-d/ (Maqāyīs III, 66), knowing that the chapter is ordered, not from ʾalif to yāʾ, but from /ʿ/ to /r/, the letter-segment following /s/ in this chapter being /ʿ/. The same principle applies to three-consonant formal roots: the corresponding sub-chapter of /s/ begins with s-ṭ-ʿ (Maqāyīs III, 70ff.); s-d-l, for example, only appears at Maqāyīs III, 149, the second radical /d/ being situated further down the ‘mental circle’ than /ṭ/, the second letter-segment of s-ṭ-ʿ, considering the fact that the starting point is /s/.

4

Later ‘Looking-up’ Arrangements, Based on the ʾalif-bāʾ-tāʾ … Alphabet

Not until the last part of the 4th/10th century do we find a dictionary aiming at a comprehensive coverage that is organized in such a way as to make it easy for learned users to look-up for roots and words. What could be called a ‘looking-up arrangement’ only appears in the Tāj al-luġa wa-ṣiḥāḥ (or ṣaḥāḥ) al-ʿarabiyya, due to ʾIsmāʿīl ibn Ḥammād al-Jawharī (d. 398/1007). This has been described as a ‘rhyme arrangement’, based on alphabetic order: formal roots are to be looked-up starting from their final letter-segment, then considering the alphabetic order of the first and second letter. If one looks, for example, for a word pertaining to the formal root ʿ-l-j, one has to go to the chapter of lettersegment /j/ (Tāj I, 297ff., roots ending with /j/), then to the sub-group of roots beginning with /ʿ/ (Tāj I, 327–332), and eventually follow the alphabetic order for the medial radical consonants (which happen to be, in this case t, ḏ, r, s, f ), until one comes to /l/, where the answer to the query is found (Tāj I, 330). In the 6th/12th century, the ʾAsās al-balāġa of al-Zamaḫšarī (d. 538/1143) is organized according to a purely alphabetic order, starting from the first lettersegment of the formal root and going on to the second, third and, if applicable, the fourth one. On the other hand, the same author’s Arabic-Farsi dictionary described in a subchapter of Haywood (1960), is organized according to morphological criteria (nouns and adjectives on one side, verbs on another), which renders the operation of finding a given root or word very difficult. The difference between the two works with regard to access to lexical information needs

128

dichy

to be explained. In the following two sections, we propose a few hypotheses to explain the existence, between the 2nd/8th and 4th/10th centuries, of dictionary arrangements that appear not to be concerned with easy consultation.

5

Heuristic vs. Looking-up Dictionary Macrostructures

Considering the arrangement of the Kitāb al-ʿayn, Haywood (1960:38), quoted above, went so far as to describe the permutation method as “the plaything of the mathematician”, adding that “we may suspect that, once having got the idea, al-Ḫalīl could not get away from it”. What escaped Haywood, in this case, is the fact that: (i) al-Ḫalīl’s approach was epistemological:7 he could not start with ʾalif, because it was a ḥarf muʿtall, i.e. a letter-segment subject to phonological transformation, and because he needed to build the inventory of letter-segments, which was to be the backbone of his dictionary, on solid grounds, through direct observation. Hence his phonetic-based inventory (see the introduction to the Kitāb al-ʿayn, and Dichy 2014, § 3.2). (ii) the arrangement of the Kitāb al-ʿayn can be described as heuristic,8 rather than ‘consultation-oriented’: it aimed at covering lexical data exhaustively though a combination of letter-segments into virtual ‘constructs’ (bināʾ), which were eventually to be confronted to extant lexical units. What Al-Ḫalīl’s endeavour primarily aimed at was the extension of the realm of knowledge in the crucial field of language rather than the elaboration of a dictionary provided with an easy and simple looking-up arrangement. Both remarks apply, mutatis mutandis, to Ibn Durayd’s Jamhara (first part of the 4th/10th century) and Ibn Fāris’s Maqāyīs and Mujmal (last third of the 4th/10th century). In these dictionaries, al-Ḫalīl’s phonetic-based ordering is replaced by other arrangements, which resort to the traditional alphabetic order. The result, though, remains altogether intricate from the strict standpoint of looking-up and consultation, as has been illustrated above in the case of the Maqāyīs. The fact that the three major dictionaries above are founded on heuristic grounds explains, in my view, why their authors came to such arrangements. But one must also reflect on how.

7 I.e., related to the structure of knowledge in one’s time. 8 Heuristics are related to discovery through procedures aiming at covering a given field of knowledge.

the intriguing issue of dictionary arrangement

6

129

Oral Transmission through Public Reading and Memory Access to Lexical Information

A complementary answer may well lie in the general conditions of knowledge transmission that were then prevalent, as opposed to our modern-times view of what a dictionary should be. A basic difference between what we call a general dictionary and Medieval Arabic lexica rests in the fact that modern dictionaries were mainly developed in the post-Gutenberg era, whereas Medieval lexica were diffused through a mixture of oral and manuscript transmission. Modern printed dictionaries are organized so as to allow quick and efficient looking-up and consultation procedures. The insistence of many authors (usually after the reference works of Naṣṣār 1956 or Haywood 1960) on dictionary arrangement constitutes, in this perspective, an anachronism in the context of the first three centuries of Arabic lexicography. Let us make a short comparison with other lexicographical contexts. In the case of French dictionaries, whose macrostructure features alphabetic ordering of words, Pruvost (2006) observes that their true development only occurred from the 18th century onwards, after the orthography of words became stable. Nowadays, ‘hit-and-go’ consultation of words has even become a widespread practice with the development of digital lexica on computers, tablets or smartphones, which allow almost instant consultation. The situation was of course completely different in Medieval Arabic culture. Knowledge was, in the first centuries of Islam, transmitted orally, often through teaching, and could eventually be written down in manuscripts. The latter were generally dictated, often to a group of scribes, and then copied (Déroche 2004). When a manuscript was finished, copies were sent to the ‘sponsors’ or ‘ordering parties’, who could afford paying for them. The received manuscript was subsequently read aloud in gatherings or sessions (majālis), to which the happy few with shared knowledge and culture were invited (for the early transmission of knowledge in Arab culture, see Schoeler 2002). One interesting point is that people were endowed then with auditory memories that were much more developed than we, in our urbanized Western 21st century life, can even imagine. A comparable level of auditory memory can still be observed to-day, for instance in Mauritania, or for some individual scholars or men of religion, in other parts of the Arab world. On the whole, literate memory today is higher in the Arab world than it is in Western countries; it was much greater in Medieval times. Of course, this does not mean that, after listening to a single reading, literate or learned listeners knew everything by heart. One must on the other hand insist on the fact that the transmission of knowledge relied on memory access

130

dichy

to quite a greater degree than we can think of, by comparison to our own contemporary memory skills. Manuscript dictionaries were read aloud in ‘reading sessions’ (majālis), and their contents were remembered by learned participants, at least to some extent. Direct memory access could thus make up for the looking-up of words in the dictionary. Significantly, Ibn Fāris, in the introduction of his Maqāyīs (I, 3f.), refers to the oral transmission (ʾaḫbaranā), of the Kitāb al-ʿayn through listening to a reading ( fīmā quriʾat ʿalayhi), and mentions a chain of learned transmitters. Manuscript transmission of knowledge and the development of memory among the literate may, at least partly, account for the fact that the use of the ʾalif, bāʾ, tāʾ, ṯāʾ … alphabet (often on a ‘rhyme’ basis), only appeared in general dictionaries from the end of the 4th/10th century onwards with alJawharī’s Ṣaḥāḥ, after the period of heuristic-oriented lexica had provided the basic word lists and methods of Arabic lexicography, between the second half of the 2nd/8th and the end of the 4th/10th centuries: dissemination thus followed heuristic gathering. True alphabetic ordering, though, only slowly became widespread during the following centuries, due to the persistence of the two causes above: (i) heuristic-based works, devoted to the discovery, extension and exhaustive coverage of lexical knowledge; and (ii) oral and manuscript transmission, associated with a high level of memory skills.

Bibliographical References A

Primary Sources

ʾAzharī, Tahḏīb = ʾAbū Manṣūr Muḥammad ibn ʾAḥmad al-ʾAzharī (d. 370/981), Tahḏīb al-luġa. Ed. by ʿAbd al-Salām Hārūn et al. 15 vols. Cairo: al-Muʾassasa al-Miṣriyya alʿĀmma li-l-Kitāb. Jawharī, Ṣiḥāḥ = ʾAbū Naṣr ʾIsmāʿīl ibn Ḥammād al-Jawharī (d. 398/1007), Tāj al-luġa wa-ṣiḥāḥ (ṣaḥāḥ) al-ʿarabiyya. Ed. by ʾAḥmad ʿAbd al-Ġafūr ʿAṭṭār. 6 vols. + 1. Cairo, 1956. Repr., Beirut: Dār al-ʿIlm li-l-Malāyīn, 1979. Ḫalīl, ʿAyn = ʾAbū ʿAbd al-Raḥmān al-Ḫalīl ibn ʾAḥmad al-Farāhīdī (d. around 175/791), Kitāb al-ʿayn. Ed. by Mahdī al-Maḫzūmī and ʾIbrāhīm al-Sāmarrāʾī. 8 vols. Baghdad, 1980–1985. Repr., Beirut, Muʾassasat al-ʾAʿlamī li-l-Maṭbūʿāt, 1988. Ibn Durayd, Jamhara = ʾAbū Bakr Muḥammad ibn al-Ḥusayn Ibn Durayd (d. 321/933), Kitāb jamharat al-luġa. Hyderabad, 1925. Repr., 3 vols. + 1 vol. indices. Beirut: Dār Ṣādir, n.d. [A better edition than the one used here is now available, edited by Ramzi Baalbaki, Beirut: Dār al-ʿIlm li-l-Malāyīn, 1987–1988.] Ibn Fāris, Maqāyīs = ʾAbū l-Ḥusayn ʾAḥmad Ibn Fāris (d. 395/1004), Muʿjam maqāyīs

the intriguing issue of dictionary arrangement

131

al-luġa. Ed. by ʿAbd al-Salām Muḥammad Hārūn. 6 vols. Cairo: Dār ʾIḥyāʾ al-Kutub al-ʿArabiyya, 1946–1952. Repr., Beirut: Dār al-Jīl, 1991. Ibn Fāris, Mujmal = ʾAbū l-Ḥusayn ʾAḥmad Ibn Fāris (d. 395/1004), Mujmal al-luġa. Ed. by Z.A. Sulṭān. 4 vols. in 2. Beirut: Muʾassasat al-Risāla, 1984. Ibn Jinnī, Sirr = ʾAbū l-Fatḥ ʿUṯmān Ibn Jinnī (d. 392/1002), Sirr ṣināʿat al-ʾiʿrāb. Ed. by Ḥasan Hindāwī. 2 vols. Damascus: Dār al-Qalam, 1985. Qālī, Bāriʿ = ʾAbū ʿAli ʾIsmāʿīl ibn al-Qāsim al-Qālī (d. 356/967), al-Bāriʿ fī l-luġa. Ed. by Hāšim al-Ṭaʿʿān. Baghdad: Maktabat al-Nahḍa and Beirut: Dār al-Ḥaḍāra alʿArabiyya. Ṣāḥib ibn ʿAbbād, Muḥīṭ = ʾAbū l-Qāsim ʾIsmāʿīl al-Ṣāḥib ibn ʿAbbād (d. 385/995), alMuḥīṭ fī l-luġa. Ed. by Muḥammad Ḥasan ʾĀl Yāsīn. 10 vols. + 1 vol. indices. Beirut: ʿĀlam al-Kutub, 1994. Šaybānī, Jīm = ʾAbū ʿAmr ʾIsḥāq ibn Mirār al-Šaybānī (d. around 206/821), Kitāb al-jīm. Ed. by ʾIbrāhīm al-ʾAbyārī et al. 4 vols. Cairo: al-Hayʾa al-ʿĀmma li-Šuʾūn al-Maṭābiʿ al-ʾAmīriyya, 1974–1983. Zamaḫšarī, ʾAsās = ʾAbū l-Qāsim Maḥmūd ibn ʿUmar al- Zamaḫšarī (d. 538/1143), ʾAsās al-balāġa. Ed. by A. Maḥmūd. Repr., Beirut: Dār al-Maʿrifa, 1982.

B

Secondary Sources

Baalbaki, Ramzi. 2014. The Arab lexicographical tradition from the 2nd/8th to the 12th/ 18th century. Leiden: E.J. Brill. Déroche, François. 2004. Le livre manuscrit arabe: Prélude à une histoire. Paris: BNF. Dichy, Joseph. 1990a. “Grammatologie de l’arabe. I. Les sens du mot ḥarf, ou le labyrinthe d’une évidence”. Studies in the history of Arabic grammar, II, ed. by Michael G. Carter and Kees Versteegh, 111–128. Amsterdam and Philadelphia: J. Benjamins. Dichy, Joseph. 1990b. L’écriture dans la représentation de la langue: La lettre et le mot en arabe. Thèse pour le doctorat d’Etat, Université Lyon 2. Dichy, Joseph. 2003. “Sens des schèmes et sens des racines en arabe: Le principe de figement lexical (PFL) et ses effets sur le lexique d’une langue sémitique”. La polysémie ou l’empire des sens: Lexique, discours, représentations, ed. by Sylvianne Rémi-Giraud and Louis Panier, 189–211. Lyon: Presses Universitaires de Lyon. Available at: www .concours‑arabe.paris4.sorbonne.fr/cours/dichy.doc Dichy, Joseph. 2014. “Al-Ḫalīl’s conjecture: How the first comprehensive dictionary in history was invented”. Arab and Arabic linguistics: Traditional and new theoretical approaches, ed. by Manuela B.M. Giolfo, 39–64. Oxford: Oxford University Press. Haywood, John A. 1960. Arabic lexicography: Its history and its place in the general history of lexicography. Leiden: E.J. Brill. Naṣṣār, Ḥusayn. 1956. al-Muʿjam al-ʿarabī: Našʾatuhu wa-taṭawwaruhu. Cairo: Dār Miṣr li-l-Ṭibāʿa. (2nd ed. 1968.)

132

dichy

Pruvost, Jean. 2006. Les dictionnaires français, outils d’une langue et d’une culture. Paris: Ophrys. Schoeler, Gregor. 2000. “Wer is der Verfasser des Kitāb al-ʿAyn?”. Zeitschift für arabische Linguistik 38.15–45. Schoeler, Gregor. 2002. Écrire et transmettre aux débuts de l’ Islam. Paris: PUF. Talmon, Rafael. 1997. Arabic grammar in its formative age: Kitāb al-ʿayn and its attribution to Ḫalīl b. Aḥmad. Leiden: E.J. Brill.

Can Ambrosiana X 56 Sup. Improve Our Understanding of Sībawayhi’s Grammar? Jean Druel

1

Ambrosiana X 56 Sup., an Exceptional Witness of the Kitāb

During her Ph.D. research (1992, published in 1995) on the transmission of Sībawayhi’s Kitāb, Geneviève Humbert discovered a fragment of the text in the Ambrosiana library in Milan. This manuscript, X 56 Sup., is entirely copied on parchment, which, according to her, is rare for a secular text. See Humbert (1995:199–203) for the complete codicological description. The manuscript is divided in ʾajzāʾ (probably around 12), and only the ninth and tenth juzʾ have reached us, in 115 folios. It contains chapters 327–435, according to Derenbourg’s numbering (Humbert 1995:170–186). Humbert believes that chances are good that the Milan manuscript has been copied in the region of Kairouan before the middle of the 5th/11th century (Humbert 1995:172). Al-Munajjid (1960: plate 17) published the reproduction of two folios from a microfilm copy of Ambrosiana X 56 Sup. held by the Manuscript Institute of the Arab League in Cairo. He dates the manuscript to the 4th/10th century. Forty-eight other folios of the same manuscript are found in the State Archives of the Republic of Tatarstan under the call number фонд 10, опись 5, дело 822 (Khalidov 2000:8f.). Khalidov knows a “very ancient” Milan copy, that he believes comes from Ṣanʿāʾ, but he does not identify both fragments as being membra disjecta of the same codex. He dates the Kazan folios to the 6th/12th century, or the beginning of the 7th/13th. Geneviève Humbert saw the Kazan folios in June 2009 and formally recognized them as part of the same codex as the Milan folios (personal communication, December 4, 2014). The text of the Kitāb that has reached us today is actually the result of an “authoritarian stranglehold” on the text by al-Mubarrad (Humbert 1995:92), which the Ambrosiana manuscript escaped, at least until it was “authoritatively corrected” around the year 715/1315, by a corrector who either put the variant readings between brackets, or struck them through, or even deleted them, based on a collation with copies containing the recensions of al-Naḥḥās (d. 338/949?) and al-Rabāḥī (d. 358/969) (Humbert 1995:189 f.). The main interest of the Ambrosiana manuscript, according to Humbert (1995:180) lies in two facts: its recension ignores the “canonical corpus of internal glosses” that are found in all other manuscripts, and its text seems to con-

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_009

134

druel

tain “less altered readings” which are “visibly more authentic”. What Humbert calls the canonical corpus of internal glosses is attributed to the three intermediaries between Sībawayhi and al-Mubarrad, namely ʾAbū l-Ḥasan al-ʾAḫfaš (d. 215/830), ʾAbū ʿUmar al-Jarmī (d. 225/839–840) and ʾAbū ʿUṯmān al-Māzinī (d. 248 or 249/863) (Humbert 1995:187), and that she believes have been added by al-Mubarrad to the text. Humbert says that she was unable to trace the exact origin of this Milan recension. She notes that the Ambrosiana fragment carries the name of ʾAbū l-Ḥasan ʾAḥmad ibn Naṣr, who is barely known to the grammatical tradition (Humbert 1995:189). According to al-Suyūṭī, in his Buġyat al-wuʿāt (I, 164), ʾAbū l-Ḥasan ʾAḥmad ibn Naṣr’s teachings where transmitted by one of Ṯaʿlab’s (d. 291/904) disciples, namely ʾAbū ʿUmar al-Zāhid. This scholar’s full name is Muḥammad ibn ʿAbd al-Wāḥid ibn ʾAbī Hišām ʾAbū ʿUmar al-Zāhid (d. 345/ 956–957), and he was called ġulām Ṯaʿlab “Ṯaʿlab’s young disciple”. The conclusion of Humbert (1993:138) is that the Ambrosiana recension of the Kitāb may have a link with Ṯaʿlab, al-Mubarrad’s Kufan main opponent. According to Humbert, the Ambrosiana recension of the Kitāb is a “fossil manuscript that challenges the edited text in a fundamental way” (Humbert 1995:186). Its text is less “worn out” and textual criticism proves that its readings are “more ancient and better” than that of the ‘Vulgate’ (Humbert 1995:189). She does not hesitate to write that this manuscript is “more valuable than all other manuscripts [of the Kitāb] together” (Humbert 1993:139). According to her, this manuscript may well support ʾAbū l-ʿAbbās Ibn Wallād’s (d. 332/943– 944) claim that al-Mubarrad’s copy of the Kitāb was of poor quality (Humbert 1995:190). Humbert (1995:183f.) provides an edition of one chapter of the Milan manuscript, chapter 332 according to Derenbourg’s edition, to support her claim. In her edition, this chapter covers 25 lines, 8 of which are not found in Derenbourg’s edition because of haplography. She explains that Derenbourg’s text is so mutilated that it is difficult to understand if one does not have the Milan manuscript at one’s disposal.

2

The Existing Editions of the Kitāb and Their Manuscript Basis

There are five main editions of the Kitāb: by Hartwig Derenbourg (Paris, 1881– 1889), by Kabīr al-Dīn ʾAḥmad (Kolkata, 1887), the Būlāq edition (Cairo, 1898), the edition by ʿAbd al-Salām Hārūn (1966–1977) and the edition by Muḥammad Kāẓim al-Bakkāʾ (Beirut, 2015). Commercial editions of the Kitāb regularly appear. See for example the edition by Émile Badīʿ Yaʿqūb (Beirut, 2009) or

ambrosiana x 56 sup.

135

Muḥammad Fawzī Ḥamza (Cairo, 2015). Yaʿqūb made an eclectic selection of either Derenbourg’s or Hārūn’s edition, with no justification and no manuscript collation. He has only filled the margins with lexicographical notes and identifications of the poetical verses. As for Ḥamza, he simply reproduced Būlāq’s edition and added in the margins a selection taken from two different commentaries. i. The edition of Hartwig Derenbourg (Paris, 1881–1889). This edition is based on four main manuscripts, which Derenbourg calls A, B, C, and L. Whenever possible, he chose A for the text and collated B, C and L in the margins. C covers only the first part and L was only available to Derenbourg when he was working on the second part of the Kitāb (Derenbourg 1881–1887:xiii). I will not mention C here, since the chapters I will study are in the second part of the Kitāb. Derenbourg used other manuscripts but they contain no variant reading in the three chapters I will study here. The manuscript that Derenbourg calls A (= Humbert: Ça) is an 18thcentury copy made on a descendant of an autograph of al-Zamaḫšarī (d. 538/1144). This descendant (= Humbert 2Ç), dated 647/1249 was probably discovered in Cairo in the 18th century and copied many times: 17 copies have reached us and A is one of them. A = Ça = Paris, Bibliothèque nationale, arabe 3887 (supplément arabe 1155). See Humbert (1995:297– 300). Manuscript B (= Humbert: 4G) is dated 1138/1725–1726. This late copy carries many mistakes. B = 4G = Saint Petersburg, Institute of Oriental Manuscripts (Akademija Nauk) C-272. See Humbert (1995:197). Manuscript L (= Humbert: 2O) is dated 629/1232. It contains the recension of ʾAbū ʿAbdallāh Muḥammad ibn Yaḥyā al-Rabāḥī (d. 358/969). L = 2O = Escorial, Biblioteca del Real Monasterio de San Lorenzo, ar. 1. See Humbert (1995:275–279). ii. The edition of Kabīr al-Dīn ʾAḥmad (Kolkata, 1887). The editor does not mention which manuscripts he worked on. This edition is partly princeps and completely independent from the other existing editions. The text contains the same corpus of inner glosses that is found in all known manuscripts, except that of Ambrosiana (Humbert 1994:9, n. 1). iii. The Būlāq edition (Cairo, 1898) has exactly the same text as that of Derenbourg, no manuscript was collated. It only adds some marginal glosses from al-Sīrāfī and al-Šantamarī (Humbert 1995:30). iv. Hārūn’s edition (Cairo, 1966–1977) is eclectic. He used Derenbourg’s edition as the basis of his edition and collated two different manuscripts that are late: Dār al-Kutub Naḥw Mīm 65 (probably from the 18th century =

136

druel

Humbert: V8), Dār al-Kutub Naḥw 141 (dated 1139/1726–1727 = Derenbourg: G = Humbert V6). Humbert has not consulted these two manuscripts. See Humbert (1995:196). v. The edition of al-Bakkāʾ (Beirut, 2015) is based on Baġdād ʾAwqāf 1351 (dated 1202 according to al-Bakkāʾ and 1204/1789–1790 according to Humbert = Humbert V10). Al-Bakkāʾ says that he has collated Mawṣil 6184 Ṣāʾiġ 14/11 as well as Hārūn’s and Būlāq’s editions. Humbert has not consulted Baġdād ʾAwqāf 1351 and she does not mention the existence of Mawṣil 6184 Ṣāʾiġ 14/11. She mentions Ṣāʾiġ 252 (= V13), which she believes to be the same manuscript (personal communication, June 8, 2017). See Humbert (1995:196). In the end, it seems that Derenbourg remains the best critical edition, based on an excellent manuscript (A = Ça). In my edition, I will collate only the editions of Derenbourg, Kolkata and al-Bakkāʾ, since they are completely independent from one another. However, as I will show in the footnotes to the chapters edited by me, Kolkata and al-Bakkāʾ are full of typos and they have either no critical apparatus (Kolkata) or only an extremely basic one (al-Bakkāʾ). In my study, I will only compare X 56 Sup. with what Derenbourg has collated from A, B and L.

3

Survey of Three Chapters: Numerals and Doubled Verbs

In order to check the quality of Ambrosiana X 56 Sup., I decided to begin with three chapters. The first one, 412 on numerals, was a natural choice, since I studied it in great detail for my Ph.D. dissertation (Druel 2012), the second and third ones, 408 and 409 on geminated verbs, were a suggestion from Michael Carter, who said that these two chapters were quite obscure and maybe the manuscript would shed a new light on them. I will use the following reference system: – A, B, L for the three main manuscripts used by Derenbourg (= Humbert: Ça, 4G and 2O). – Kolkata for the Kolkata edition by Kabīr al-Dīn ʾAḥmad in 1887. – Bakkāʾ will refer to the Beirut edition by al-Bakkāʾ in 2015. Any discrepancy mentioned by al-Bakkāʾ between Baġdād ʾAwqāf 1351 used as the base (= Humbert: V10) and Mawṣil 6184 Ṣāʾiġ 14/11 (= Humbert: V13?) will be mentioned. – I will follow Humbert and call M1 the first hand in Ambrosiana (5th/11th century?) and M2 the corrector (715/1315).

ambrosiana x 56 sup.

137

3.1 Results of the Collation There are 127 cases where either M1, M2, A, B and/or L has a different reading (although collated in the notes, the 66 additional cases where Kolkata and/or al-Bakkāʾ have a different reading from all other versions are not taken into consideration here because they are actually of poor quality and most of the cases seem to be typos, as will appear in the notes). In 78 cases (60 % of these 127 cases), M1 and M2 disagree. In the other 44 cases (40 %), M1 and M2 agree, but there are discrepancies between the other readings, A, B or L (between A and B in 28 cases; between A and L in 28 cases; between B and L in 5 cases). Here are the details of these cases. Cases where M1 is opposed to M2: 78 cases (60% of 127 cases) – M1 vs M2, A, B, L: 65 cases – M1, A vs M2, B, L: 4 cases – M1, A, B, L vs M2: 4 cases – M1, A, B vs M2, L: 1 case – M1 vs M2, B, L vs A: 2 cases – M1 vs M2, A vs B, L: 1 case – M1 vs M2 vs A, B, L: 1 case Cases where M1 and M2 have the same readings, different from the others: 44 cases (40% of 127 cases) – M1, M2 vs A, B, L: 22 cases – M1, M2, B, L vs A: 15 cases – M1, M2, A vs B, L: 3 cases – M1, M2, A, L vs B: 2 cases – M1, M2, A, B vs L: 1 case – M1, M2, B vs A vs L: 1 case There are 5 cases where it is difficult to decide and the Ambrosiana manuscript should be directly consulted. Conclusion: A carries some original readings (in 18 cases it differs from all the other versions), but most of these original readings are of poor quality. B and L are highly dependent on A, although they still differ from it in 28 cases out of 127. B and L are very dependent on one another, and only differ in 5 cases. In 92 cases, 72% of all cases, A, B and L agree. M1 and A contain the two furthest versions, they differ in 107 cases (84% of all 127 variant cases). M2, which is an attempt to align M1 on the Rabāḥī recension (represented here by L), still differs from L in 33 cases. It also differs from the Zamaḫšarī recension (represented here by A) in 50 cases.

138

druel

On the basis of these three chapters, we can say that M1 originally contained a very different version from both the Zamaḫšarī and the Rabāḥī recensions, confirming what Humbert had already noted. After correction by M2, the text partly keeps its originality against these two recensions. 3.2 Evaluation of the Variant Readings In this section an attempt will be made to sort and evaluate the variant readings of M1 in our three chapters, if compared with the other variants (A, B and L), not only M2. The number refers to the footnotes in the critical edition below, where the case is presented in detail. – 10 cases where M1 has a probably better reading: correct conjugation 14, 15, 148; clearer teaching: 29, 98, 191; better syntax: 20, 68, 166, 168. Out of these ten cases, only three cases are typical of M1 and were not known to us before: 15, 98 and 168. – 4 cases where M1 is easier to understand: gloss added: 53; example added: 62; clearer syntax: 172, 194. Three of these four cases (53, 62 and 172) are known to us only through M1. – 1 case where M1 had a different (but today unreadable) reading at a place where the ‘canonical’ reading is difficult: 145. All the versions collated here have the canonical reading. – 14 cases where M1 is more difficult to understand than the other versions: syntactic difficulties: 1, 35, 54, 93, 100, 157, 195; grammatical explanation or example lacking: 95, 97, 102, 149, 189, 193; anonymous poet: 125. – 16 cases where M1 contains a ‘mistake’: repetition of a sequence of words: 136, 164, 176; syntactic mistakes: 8, 16, 39, 67, 167, 178, 181; inaccurate glosses: 37, 108; spelling mistake: 177; undue negation: 77; conjugation mistakes: 9, 121. – 71 cases where M1 has a different reading than the other versions, with no impact on the meaning: different grammatical examples: 2, 4, 19, 26, 32, 33, 51, 59, 69, 79, 106, 113, 118, 130, 135, 141, 143, 147, 155, 158, 160, 169, 175, 188, 192; different verbal forms: 6, 24, 103, 114; different wording: 31, 42, 45, 56, 57, 61, 65, 71, 73, 75, 76, 80, 84, 85, 92, 104?, 109, 110, 111, 129, 138, 139, 140, 142, 146, 152, 153, 159, 162, 163, 165, 170, 171, 173, 174, 180, 183, 186, 187, 190, 196; different authorities quoted: 116. – 2 cases where it is difficult to decide without knowing what has been erased by M2: 122, 123. – 1 case where M2 has a better reading, different from A, B, L and M1: 81 – 1 case where M2 has suppressed a series of grammatical examples found in A, B, L and M1: 91 – 1 case where M2 has completed a verse that is incomplete in A, B, L and M1: 96.

ambrosiana x 56 sup.

139

In order to reach the total of 127 cases, one should add to these 121 cases the 5 doubtful cases (notes 22, 52, 64, 89 and 105) plus one case (note 124) where the text of A, B and L differ inside an addition that M2 has done to M1 (note 125). Altogether, our harvest of ‘better’ readings is really meagre: in 3 cases, the readings of M1 can probably be said to be better than the other versions and original to M1, and in 3 other cases, the readings are easier to understand than in the other versions. But in 14 cases, the readings in M1 are more difficult to understand, and in 16 cases, they are erroneous. In the large majority of cases (71 cases) the readings in M1 are different from at least one of the other versions, but these differences cannot be said to be better or worse. Interestingly, M2 brings a better reading, that was not known from the other versions. 3.3 Examples of Corrections by the 8th/15th Collator (M2) In 63r°.4, M2 has struck through six words (ʾillā fī l-ʾalif wa-l-lām wa-l-ʾalif alḫafīfa) that are found in all the versions collated here (see n. 81). The situation here is surprising, since M2 supposedly aims at correcting the text whenever it differs from the canonical version. Just like A, B and L, M1 actually had these six words but M2 struck them through. As commented upon in the note, the presence of these six words is not consistent with the grammatical teaching in this chapter. In 63r°.11, M2 added the words man qāla halummā wa-halummī that were not in the matn of M1 (see n. 98). All the other variants have them, however, they could well be an example of an internal gloss that has made its way into the text. Their implication in terms of grammatical teaching is not edge-cutting. They add a condition to a grammatical teaching that is otherwise more general. Based only on the evidence of the teaching, I believe it plausible that these words are a later addition to the text. In 71r°. 18, M2 changed the singular ḥālihi into the dual ḥālihimā, which is the reading of all other versions (see n. 168). The grammatical point at stake here is the modification of the status of iṯnāni in the compound numeral iṯnā– ʿašar. The formulation of the sentence lam tuġayyirū l-iṯnayni could lead the reader to think that al-iṯnayni refers to both iṯnay and ʿašar, not to iṯnay, the first part of the compound, and that the text comments the ḥāl of both iṯnay and ʿašar, which is not the case. It only deals with the ḥāl of iṯnāni, the first part of the compound, with does not change after coalescence. Establishing the text of the Kitāb is not an easy task because of the amount of glosses that have entered its matn and Derenbourg’s effort to ‘clean’ the text from these glosses has created new mistakes (Humbert 1995:41). In our three chapters, there are a few places where we can study the issue of the glosses (see n. 37, 91 and 108, as well as the case presented above with n. 98). In 37,

140

druel

a gloss has been added in the margin and then put between brackets. In 91, M2 has put between brackets a passage that is found in the matn of all the collated versions and that he probably considered to be an internal gloss providing additional examples. And in 108, a passage found only in the matn of M1 was put between brackets by M2, who probably considered it to be a gloss. In five different places (see n. 143, 147, 155, 169, 175), M1 systematically used the isolated forms of numerals when discussing them, i.e. the forms with an ending tāʾ marbūṭa between ‘three’ and ‘ten’, and the masculine form for eleven, whereas M2 changed them into the form actually used in the examples. In one place (see n. 145), the canonical version was doubted by many grammarians and M1 had a different reading, which is illegible on the photo of the manuscript at my disposal. I hope that a direct consultation of the manuscript can reveal what the original reading was. Lastly, it is really interesting to see how often M1 had different grammatical examples (25 cases, see the list above). In most cases, it is impossible to read the original readings in M1 because M2 has erased them and written above them. However, their mere existence is in itself a very eloquent testimony that, at some point in history, the text of the Kitāb was still flexible but that in 715/1315 this was not the case anymore.

4

Conclusion

M1 contains a version that greatly differs from A, the base of Derenbourg’s edition and later editions. And even after its correction by M2, the text still contains a good number of variant readings. This very old parchment (5th/11th century?) is an exciting witness of the reception of the Kitāb in the West of the Arab world, in particular because it is indirectly connected to Ṯaʿlab, the main ‘Kufan’ opponent of al-Mubarrad. The three chapters that I have edited in this paper have revealed interesting readings (6 out of 127), and there is little doubt that the remaining 106 chapters contained in the Milan manuscript (along with the Kazan folios) will bring more. My conclusion at this point, and based on only three of the 109 chapters is contains, is that the text of M1 and M2 cannot be said to be generally better than Derenbourg’s edition, yet, it is imperative to check the Milan manuscript whenever one studies a particular chapter of the Kitāb in case it may contain a rare and valuable reading. The quest for the ‘original’ text, that has led Derenbourg to create new mistakes by suppressing what he considered to be glosses, is comparable to the effort of M2 on the text of M1. What Ambrosiana X 56 Sup. points to is that

ambrosiana x 56 sup.

141

the text of the Kitāb has long been considered to be flexible, i.e., available for editing and adaptation at the hand of the grammarians who created their own working copies. When M2 corrects the text in 715/1315, it is clear that for him the text is no longer flexible and that he believes an ‘original’ version exists and can be reached. As Humbert (1994:10) puts it, the rich manuscript tradition of Sībawayh’s Kitāb has been overlooked by modern editors and scholars. Another exceptional manuscript, that has not received the attention it deserves is the autograph by the Andalusian grammarian ʾAbū l-Ḥasan ʿAlī Ibn Ḫarūf al-Ḥaḍramī (d. ca. 609/1212), kept in Paris, Bibliothèque nationale, arabe 6499, which contains a sophisticated critical recension. Even though our understanding of Sībawayh’s teachings is not fully renewed at this point, at least our vision of the richness of the textual tradition is.

Edition of Three Chapters The following guiding principles have been used in the present edition. The base text is that of M1, and the corrections of M2 are inserted directly into it, with the coding explained below and a note providing a brief commentary. Any other note, i.e. not following a correction mark in the text, indicates a discrepancy within the ‘Vulgate’ itself, represented here by the agreement of M2, A, B, L, Kolkata and al-Bakkāʾ. This means that one can easily see in the text the many cases where M2 has aligned M1 to the ‘Vulgate’, as well as any discrepancy inside this ‘Vulgate’. The following coding has been used for the corrections done by M2 within the text: words struck through: M2 simply struck them through. (words put between brackets): M2 added brackets instead of striking the words through. deletion marks in the text: In some cases, M2 added deletion marks above the words to delete. If one word, he added a single mark, similar to the head of a ṣād ‫ صـ‬above the word to be deleted. When more than one word, M2 added a mark similar to the head of a mīm ‫ مـ‬above the first word and a reversed head of a mīm on the last word. In both cases, I have underlined the words in my edition.

142

druel

[words erased by M2]: M2 erased them and eventually rewrote above them. [×××]: erased words that are not readable anymore. The number of ‘×s’ roughly corresponds to the number of characters erased. ⸂words added by M2⸃: M2 has either added these words in the matn, either above an erased passage, or between the lines or in the margins, and consistently uses signes-de-renvoi. [abc]→⸂def⸃ thus means that M2 has erased ‘abc’ and replaced it by ‘def’. ⸉margins added by M2⸊: M2 has added these words as commentaries in the margin, not as part of the matn. These margins are found only in the Milan manuscript. They are either glosses or collation marks. There are also cases where M2 has mentioned an inversion in the order of the words by adding the word muʾaḫḫar above the first word and the word muqaddam above the second one. In the edition, I have simply reproduced them as in: ‫ مؤخروهلَ ُ َمّ مقدملا ي َكسر‬which means that the expression in M1 was wa-halumma lā yaksiru and that M2 corrected it into wa-lā yaksiru halumma (see below, folio 63r°.11). Lastly, I have left in the text the paragraph markers (dāra): ʘ. Geminated verbs: Ambrosiana 61v°.2–63v°.20 = Derenbourg chapter 408–409, II, 162.1–165.7 = Kolkata 703–707 = Bakkāʾ V, 237–241 ‫خر الفعل‬ ِ ‫ف أن يكون آ‬ ُ ‫[ والتضعي‬3] ‫[ هذا باب مضاع َف الفعل واختلاف العرب فيه‬2] [‫ظ‬61] 3‫ت‬ ّ ُ ‫ت وا ِن ْق َد َْد‬ ُ ‫ وا ِجْترَ َْر‬2‫ت‬ ّ ُ ‫[ و]صـ؟[←⸃و⸂د َْد‬4] ‫ت‬ ّ ُ ‫ ر َد َْد‬1 ⸂‫ن من موضٍع واحد وذلك ⸃نحو‬ ِ ‫حرفا‬ ُ ‫خر‬ ِ ‫ف الآ‬ ُ ‫ت فإذا تحر ّك الحر‬ ُ ْ ‫ت وا ِْطم َ ْأن َن‬ ُ ‫[ وا ِْحمار َْر‬5] ‫ت‬ ُ ‫ وا ِْحمرَ َْر‬5‫ وترَ اد َْدنا‬4‫ت‬ ُ ‫ت وصار َْر‬ ّ ُ ‫وا ِْستعَ ْد َْد‬

1 M2 added naḥwa which was missing in M1 but found in all the other versions collated here. This naḥwa makes the text of M1 smoother. Maybe this could be a witness of a less ‘polished’ text. 2 M1 had a different example from all other collated versions before its correction. Maybe it was the verb ṣadda, as in al-Sīrāfī’s commentary (XIV, 57.9). 3 Kolkata has the erroneous inqaḏadtu. 4 A has wa-ṣārartu. B, L, Kolkata and Bakkāʾ have wa-ḍārartu. M1 has wa-ṣārartu and M2 did not correct it. Al-Sīrāfī’s commentary (Šarḥ XIV, 57.9) has the example ḍārra. 5 Kolkata has wa-tawādadnā.

ambrosiana x 56 sup.

143

‫ من موضٍع واحٍد‬9‫ لأن ّه لم ّا كان‬8⸂‫ ⸃ َأولى به‬7‫[ الإدغام وذلك فيما زعم الخليل‬6] ‫ على‬6‫ب م ُج ْم ِعون‬ ُ ‫فالعر‬ ّ‫خر فلما‬ ِ ‫[ الآ‬8] ‫ أن يرَ فعوا ألسنتهم من موضٍع ثم يعُ يدوها إلى ذلك الموضع للحرف‬10‫[ عليهم‬7] ‫ل‬ َ ُ َ‫ثق‬ ‫ وا ِْستعَ ِّدِي‬12‫ وا ِن ْق َُّدوا‬11‫[ وا ِجْت َرَ ّا‬9] ‫ثقل عليهم ذلك أرادوا أن يرَ فعوا ر َف ْعة ً واحدة ً وذلك قولهم ر ُدِّي‬ 16‫ف من هذا‬ ٌ ‫ فإذا كان حر‬ʘ 15‫[ ي َْطم َئ ُِّن‬10] ‫ وا ِْحمرَ َ ّ وا ِْحما َرّ وهو‬14‫ن‬ ِ ‫ ز َي ْد ًا وهما يرُ ادّا‬13‫وضارِّي‬ ‫خر فلم يكن‬ ِ ‫ن أهل الحجاز يضاعفون لأّنهم أسكنوا الآ‬ ّ ‫[ فإ‬11] ‫ فيه لام ُ الفعل‬17‫الحروف في موضٍع ت َسكن‬ ‫ وا ِجْترَ ِْر وإْن ت ُضارِْر ُأضارِْر‬18‫ن وذلك قولك ا ُْرد ُْد‬ ِ ‫[ قبله لأن ّه لا يلَتقي ساكنا‬12] ‫ب ُّد ٌ من تحر يك ال ّذي‬ َ ُ ‫[ ال َر ّج‬14] ِ‫[ وإْن ت َْستعَ ْدِْد َأْستعَ ْدِْد وكذلك جميع هذه الحروف و يقولون ا ُْرد ُد‬13] َ ‫ل وإْن ت َْستعَ ْدِدِ اليوَ ْم‬ 20‫ن هذا التحر يك ليس بلازٍم لها إن ّما حرّكوه‬ ّ ‫[ حاله ولا ي ُدِغمون لأ‬15] ‫ ي َد َع ُونه على‬19َ ‫َأْستعَ ْدِْد اليوَ ْم‬ ‫[ الفعل مبن ًي ّا عليهكالنون‬17] ‫ في‬21‫ن ال ّذي بعده‬ ُ ‫[ هذا الموضع لالتقاء الساكنينِ وليس الساك‬16] ‫في‬ ‫[كينِ لم ِا‬2]ّ ‫ن متحر‬ ِ ‫[ وأمّا بنو تميم في ُدِغمون المجزوم كما أدغموا إذ كان الحرفا‬1] [‫و‬62] ‫الثقيلة والخفيفة‬ ‫ن جميع ًا وهو قول غيرهم‬ ِ ‫ ي َسكـنا‬22‫[ لأّنهما لا‬3] ‫خر‬ ِ ‫ذكرنا من المتحر ّكين في ُسكنون الأّول و يحرِ ّكون الآ‬ ‫ت‬ َ ْ ‫[ فإذا كان الحرف ال ّذي قبل الحرف الأّول من الحرفينِ ساكناً ألقي‬4] ʘ ٌ ‫من العرب وهم كثير‬

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

A has the passive mujmaʿūn. B, L and Bakkāʾ have the active mujmiʿūn. M1 has mujmiʿūn and M2 did not correct it. Kolkata is not vocalized. According to Bakkāʾ, Mawṣil adds raḥimahu Allāh taʿālā. M2 added the two words ʾawlā bihi that were lacking in M1 and without which the text makes no sense. All the other versions collated here have them. A, B, L, Kolkata and Bakkāʾ have the more correct dual kānā. M1 has the singular kāna and M2 did not correct it. Bakkāʾ has the obviously erroneous ʿalayhi. Kolkata has wa-jtarrū. Kolkata has wa-nqaḏḏū. Kolkata has wa-staʿiddā wa-ḍārrā with a clear ending ʾalif maqṣūra and superscript ʾalif. A has the bizarre form yurāddanāni. Vocalised yuṭmaʾinnu by Derenbourg (according to A, B, L?). Bakkāʾ and M1 have the better vocalization yaṭmaʾinnu, which M2 did not change. Kolkata is not vocalized. A, B, L, Kolkata and Bakkāʾ have the more correct hāḏihi. M1 has hāḏā, and M2 did not correct this obvious grammatical mistake. Did it escape the attention of the corrector? Kolkata has yaskunu. Kolkata has urdudū. A, B, L, Kolkata and Bakkāʾ do not have this al-yawm. It is not found either in al-Sīrāfī’s commentary (Šarḥ XIV, 58.11). M1 has it and M2 did not correct it. A and Bakkāʾ have the less correct ḥarrakū. B, L and Kolkata have ḥarrakūhu. M1 has ḥarrakūhu and M2 did not modify it. Kolkata has only baʿda, without the suffix pronoun. Not clear. To be checked directly on the manuscript.

144

druel

‫ كان مفتوح ًا‬23‫ض َمّه وإن‬ ُ ‫[ ف‬6] ‫[ حركة الأّول عليه إن كان مكسور ًا فٱكسرْه وإن كان مضموم ًا‬5] ‫ عنها حيث‬25‫ل حذفت َها لأن ّه قد استغُ ني‬ ُ ‫[ أل‬7] ‫ عليه الحركة‬24‫فٱفتحْه وإن كان قبل ال ّذي يلُ ْقَى‬ ٍ ‫ف وص‬ ‫ت‬ َ ْ ‫ ألقي‬26ّ‫[ َأر ُ َد‬9] ّ‫ض وإْن ترَ ُ َد‬ ّ َ َ ‫ وذلك قولك ر ُ َدّ وف ِ َر ّ وع‬ʘ ‫[ إليها لسكون ما بعدها‬8] ‫ح ُرّك وإن ّما ا ُحتيج‬ ʘ ‫ت ذلك في غير الجزم‬ َ ‫ الألف كما فعل‬28‫ت‬ َ ‫[ف‬10]‫ على الساكن ال ّذي قبله وحذ‬27‫حركة الأّول منهما‬ ‫ت‬ َ ْ ‫ ألقي‬30ٌ ‫ حاجز‬29‫ن ال ّذي قبل الأّول بينه و بين الألف‬ ُ ‫[ وإن كان الساك‬11] ‫وذلك قولك ر ُ َدّا ور ُ ُدّوا‬ ‫[ عن الأصل‬13] ‫ في حال صاحبه‬31 ⸂‫ل واحٍد منهما ]يتحر ّك[←⸃ي َتحو ّل‬ ّ ‫نك‬ ّ ‫[ عليه حركة الأّول لأ‬12] ‫ن الحرف ال ّذي بعد‬ ّ ‫[ لأ‬14] ‫تحذف الألف‬ َ ‫ ولا‬32 ⸂‫ض‬ ّ َ َ ‫ت ذلك في ر ُ َدّ وف ِ َر ّ و]×××ا[←⸃ع‬ َ ‫كما فعل‬ 35 ⸂‫ت ⸃الألف‬ ْ ‫ فصار‬34ّ َِ‫ وإْن ت َْشم َئزَِ ّ َأْشم َئز‬33ّ ‫[ وا ِق ْشَع َ َر‬15] ‫ن‬ ّ َ ‫ن وذلك قولك ا ِْطم َ َأ‬ ٌ ‫ألف الوصل ساك‬ ّ ِ‫ وذلك قولك ا ِْطم َئنُِ ّوا وا ِْطم َئنَِ ّا ومثل )⸊ذلك ِإْن ت َْشم َئزِ ّ َأْشم َئز‬36‫[ في الخـبر‬16] ‫في الإدغام والجزم مثلها‬

23 24 25 26

27 28 29

30 31

32 33 34 35

36

Kolkata has fa-ʾin. A, B, L, Kolkata and Bakkāʾ have tulqī. M1 has yulqā and M2 did not correct it. The meaning is the same, only that it is in the passive voice. Kolkata clearly has the wrong form istaġnā, with an ending ʾalif maqṣūra and superscript ʾalif. A is the only one to have the ending vocalization wa-ʾin taruddu ʾaruddu. There are various possible ending vocalizations, as presented in the following chapter of the Kitāb, and it is not the point at stake here. What is at stake here is the middle vocalization. Kolkata has minhā, which is incorrect. Kolkata has ḥaḏafta without the connecting wāw. A is the only one to have al-ʾawwal instead of al-ʾalif. This formulation makes no sense, since al-ʾawwal refers already to the first of the two consonants that will be assimilated. What is intended is the presence of a phoneme between the first letter to be assimilated and the ʾalif waṣl. In this case, the ʾalif waṣl remains, and the first of the two consonants takes the vowel of the second identical consonant. Kolkata has ḥājizan. A has yataḥarraku. B, L and Bakkāʾ have yataḥawwalu. M1 had yataḥarraku and M2 corrected it to yataḥawwalu. The wording is different, but I am not sure whether one is better than the other. The same root in yataḥawwalu and ḥāl probably pleads in favor of this verb over yataḥarraku. M1 had a different example, apparently with an ending ʾalif. B is the only one to add wa-šmaʾazza after wa-qšaʿarra. Kolkata clearly has wa-ʾin tašmayʾizz ʾašmayʾizz, with a yāʾ, which is probably only a spelling issue of the hamza. Without the addition of al-ʾalif, the sentence is very unclear. The subject (the first al-ʾalif ) would be very far from this new verb. Does this addition amend a difficult ‘original’ reading in M1, as was the case above in note 1? All the other versions collated here have this al-ʾalif. Kolkata has al-jarr, which makes no sense in the context here.

ambrosiana x 56 sup.

145

‫ل‬ ُ ‫[ أل‬18] ‫ قبل الأّول متحر ّك ًا وكان في الحرف‬38‫[ ذلك ا ِْستعَ َِّد وإن كان ال ّذي‬17] 37(⸉‫ومثل‬ ٍ ‫ف وص‬ 40‫[ إلى تحر يكه ولا ت َذهب الألف‬19] ّ ‫ الحركة ُ عن حالـ]ـها[←⸃ـه⸂ لأن ّه لم يكن حرفا ي ُضطَر‬39⸂‫لم تغي ّر⸃ه‬ ‫ َأن ْق ََّد فصار في الإدغام‬43‫[ ا ِجْت َرَ ّ وا ِْحمرَ َ ّ وا ِن ْق ََّد وإْن ت َن ْق ََّد‬20] 42‫ وذلك نحو‬41‫ن الذي بعدها لم يح َر ّك‬ ّ ‫لأ‬ ‫ن‬ ّ ‫ لأ‬46‫[ لم تغي ّر‬1] [‫ظ‬62] ‫ف‬ ٌ ‫ كان قبل الأّول أل‬45‫[ الألف مثله في غير الجزم وإن‬21] 44‫وثبات‬ ‫ن‬ ّ ‫ الحرف لأ‬47‫[ الوصل في ذا‬2] ‫ن المدغ َم ُ في َحتمل ذلك وتكون ألف‬ ُ ‫الألف قد يكون بعدها الساك‬ ِ ِ ‫[ وإْن ت َْدها َمّ َأْدها َمّ فصار في الإدغام‬3] ‫ب‬ ّ َ ‫ اْحما َرّ واش ْها‬49‫ وذلك‬48‫الساكن ال ّذي بعدها لا يح َر ّك‬ ‫ف‬ ُ ‫ف ولم يكن في ذلك الحرف حر‬ ٌ ‫[ وإن كان قبل الأّول أل‬4] ‫وثبات الألف مثله في غير الجزم‬ [6] ّ‫تجا َر‬ ُ ‫ ولا‬51ّ‫[ بنائه وعن الإدغام في غير الجزم وذلك قولك ما َدّ ولا ت ُضا َر‬5] ‫ عن‬50‫ل لم يغ َي ّر‬ ٍ ‫وص‬ ʘ 53⸉‫ن والـكسر أجود‬ ٌ ‫ ألف ُه مقطوعة ً نحو َأم َِّد و َأع َِّد ⸊فتح الدال من أمّد وأعّد حس‬52‫ت‬ ْ ‫وكذلك ما كان‬ 37

38 39

40 41 42 43 44 45 46 47 48 49 50 51

52 53

It is not clear who added this marginal gloss and who put it between brackets. If we consider that M2 put it between brackets, then we probably have to consider that it is M1 who added it, as an auto-correction. It could also reflect more than two hands at work on this manuscript. This addition is of poor interest, it does not fit with the examples dealt with here (i.e. the cases where the ʾalif waṣl is maintained after assimilation of the repeated consonants). Kolkata has kāna l-ḥarf allaḏī, which only makes the expression clearer. The initial reading of M1 is not correct because al-ḥaraka cannot be the subject of the verb lam tuġayyar. What is at stake here is not a change in the vowel added on the doubled consonant, but the fact that adding this vowel will not change the morphology of the verb, i.e. the ʾalif waṣl will still be needed. All the versions collated here bear the corrected reading of M2. Kolkata does not have al-ʾalif. Kolkata has yataḥarrak. A, B and L do not have this naḥwa. Kolkata has it. Bakkāʾ has qawluka instead of naḥwa. M1 has naḥwa and M2 did not correct it. Kolkata has nanqadda. According to Bakkāʾ, Mawṣil has banāt instead of ṯabāt. A, B, L, Kolkata and Bakkāʾ have wa-ʾiḏā instead of this wa-ʾin. M1 has wa-ʾin and M2 did not correct it. Kolkata has yuġayyar. Kolkata has hāḏā. Kolkata has yataḥarraku. Kolkata has wa-ḏālika qawluka. Kolkata has tataġayyaru. Vocalised tuḍārru by Derenbourg (according to A, B, L?) and by Bakkāʾ. Kolkata is not vocalized. M1 has an ending fatḥa, which M2 did not change. This point is not what is at stake here and will be discussed at length in the next chapter, and lots of variation happens. So both forms can be regarded as equally possible here. The ʾalif is unclear. To be checked directly on the manuscript. The marginal addition is found only in M2, if it is really an addition by M2 and not an

146

druel

55‫[ أن ي َسكن هو والأّول‬8] ‫ لا يستقيم‬54⸂‫خر ⸃لأن ّه‬ ِ ‫[ هذا باب اختلاف العرب في تحر يك الآ‬7] ‫خر كتحر يك ما قبله فإن‬ ِ ‫ يحر ّك الآ‬57‫ن منهم من جعل‬ ّ ‫[ اعلم أ‬9] ‫ غير أهل الحجاز‬56⸂‫]في؟[←⸃من‬ ّ‫ ضم ّوه وإن كان مكسور ًا كسروه وذلك قولك ر ُ ُد‬58‫[ وإن كان مضموم ًا‬10] ‫كان مفتوح ًا فتحوه‬ ‫ فتحة ً وألف ًا‬60‫[ قبلها‬12] ‫ن‬ ّ ‫ واجْت َرَ ّ وا ِْحمرَ َ ّ وضا َرّ لأ‬59ِ‫ض وف ِرِ ّ يا فتى و َأم ِّدِ واْطم َا ئ ِ ّ ِن وا ِْستعَ ِّد‬ ّ َ َ ‫[ وع‬11] ‫ إليك ولا‬63‫ضنا وم ُُّدني‬ ّ َ َ ‫[ وع‬13] 62‫ضني‬ ّ ‫ ور ُ ُدّنا ولا ي ُِشلـ ِّكم الل ّه ُ وع‬61‫فهـ]ـي[←⸃ـو⸂ أجدر أن يفُ ت َح‬ َ ‫ ل ِم‬66‫ت الخليل‬ ُ ‫[ وسأل‬15] ʘ ‫[ فتحوا أبد ًا‬14] ‫ت الهاء ُ والألف‬ ْ ‫ جاء‬65‫ضك ُْم فإذا‬ ّ َ َ َ‫ الل ّه ُ ول ِيع‬64‫ي ُِشل ِّك‬ ‫ت‬ ْ ‫ فإذا كان‬69‫[ إذا قالوا ر ُ َدّها وغ َُل ّها‬16] 68‫ وغ َُل ّا‬67⸂‫ن الهاء خفي ّة ٌ فكأّنهم قالوا ر ُ َدّا و َأم َِّد⸃ا‬ ّ ‫ذاك فقال لأ‬

54 55 56 57

58 59

60 61

62 63 64 65 66 67 68 69

auto-correction by M1. This addition is compliant with the teaching of the next chapter, i.e. the vocalization of the Tamīm forms, where the fatḥa is far from being the best form (although not stated clearly). Without li-ʾannahu the text is less easy to read (less correct?) M1 is the only version not to have it. Kolkata has wa-huwa al-ʾawwal. M1 had a different particle than all the other versions, with no implication for the meaning or for the quality of the language. M1 is the only text to have jaʿala before the conjugated verb yuḥarrik, which is a possible construction meaning ‘to begin to’ (see Kitāb I, 364.20). However, this construction makes no sense here and M2 struck this word through. Bakkāʾ has an erroneous mā after maḍmūman. A: wa-qšaʿarra wa-ṭmaʾinna wa-staʿidda; B, L and Kolkata: wa-ṭmaʾinni wa-staʿiddi; Bakkāʾ: wa-qšaʿirri wa-ṭmaʾinni wa-staʿiddi; M1: wa-ʾamiddi wa-ṭmāʾinni wa-staʿiddi; M2: wa-ṭmāʾinni wa-staʿiddi. M1 has an additional example (wa-ʾamiddi) which M2 struck through, and a long ʾalif in wa-ṭmāʾinni. which M2 did not correct. The additional example ʾamiddi is already found above in 62v°.6=II, 163.9. Although it does not really add to the demonstration at this point, it is consistent with the teaching. Kolkata has li-ʾanna mā qablahā. A and Bakkāʾ: fa-hiya ʾajdaru ʾan tuftaḥa; B, L and Kolkata: fa-huwa ʾajdaru ʾan yuftaḥa; M1: fa-hiya ʾajdaru ʾan yuftaḥa; M2: fa-huwa ʾajdaru ʾan yuftaḥa. There is no difference in the use of the masculine or the feminine. M1 had one more example, that M2 rejected by adding a superscript mark above it. This additional example is found nowhere else. Kolkata has wa-maddanī with a fatḥa on the šadda as if the verb was in the past tense, not the imperative. The lām and its vocalization is unclear. To be checked on the manuscript. A, B, L and Bakkāʾ have fa-ʾin. Kolkata and al-Sīrāfī’s commentary (Šarḥ XIV, 61.1) have fa-ʾiḏā. M1 also has fa-ʾiḏā and M2 did not correct it. According to Bakkāʾ, Mawṣil adds raḥimahu Allāh taʿālā. Kolkata adds rḥ. The omission of ʾalif in M1 is clearly a mistake because it misses the very point of the demonstration, namely that the ending fatḥa is like a lightened ʾalif. A is the only one to have the incorrect ruddan wa-ʾamiddan wa-ġullan with tanwīn. L and Bakkāʾ are the only ones to add wa-ʾamiddahā after wa-ġullahā.

ambrosiana x 56 sup.

147

‫ت بالألف واللام‬ َ ‫ضه ُ فإن جئ‬ ّ ُ َ ‫ م ُُّده ُ وع‬70‫ضوا إذا قالوا‬ ّ ُ َ ‫[ م ُُّدوا وع‬17] ‫الهاء ُ مضمومة ً ضم ّوا كأّنهم قالوا‬ ‫[ إذا كان‬19] ‫ن الفعل‬ ّ ‫ت الأّول كل ّه لأن ّهكان في الأصل مجزوم ًا لأ‬ َ ‫ كسر‬72‫ الخفيفة‬71‫[لف‬18]‫والأ‬ ‫ل‬ َ ُ ‫[ ال َر ّج‬20] ‫ب‬ ِ ِ ‫كس ِر َ وذلك قولك ا ِض ْر‬ ُ ِ‫ لالتقاء الساكنين‬74⸉‫ ⸊صح⸉ ⸊فحرك‬73‫حر ّك‬ ُ ‫مجزوم ًا ]تـ[←⸃فـ⸂ـ‬ ‫ن أصله أن‬ ّ ‫ لأ‬75‫[ ردّدتهَ إلى الأصل‬21] ‫ف الخفيفة‬ ُ ‫ف واللام والأل‬ ُ ‫ت الأل‬ ْ ‫ك فلماّ جاء‬ َ َ ‫ب اب ْن‬ ِ ِ ‫وا ِض ْر‬ ‫ ذلك ج َر َى‬77‫ن نظا ئره من غير المضاع َف على غير‬ ّ ‫[ كما أ‬1] [‫و‬63] ‫ لغة أهل الحجاز‬76‫يكون مسّكناً على‬ ‫ن‬ ّ ‫[ على أ‬3] ‫ الميم‬78ِ‫[ فيمن َأسكن تقول م ُذ ُ اليوَ ِْم وذ َهبَ ْت ُم ُاليوَ ْم َ لأن ّك لم ت َبن‬2] ْ ‫ومثل ذلك م ُْذ وذ َهبَ ْت ُم‬ ‫[ إذا ]اجتمع[←⸃ التقى‬4] ‫ ونحوها ومنهم من يفَ تح‬79⸂‫ض‬ ٍ ⸃←[×]‫أصله السكون ولـكن ّه ح ُذف كياء قا‬ ‫ أّنهم شبّهوه‬82‫[ الخليل‬5] ‫ فزعم‬81‫ل حال إلّا في الألف واللام والألف الخفيفة‬ ّ ‫ على ك‬80⸂‫ن‬ ِ ‫ساكنا‬ ‫ جاءوا بالألف واللام والألف الخفيفة ما فعل‬83‫[ به إذ‬6] ‫ف وأشباه ذلك وفعلوا‬ َ ْ ‫سو‬ َ ‫فو‬ َ ْ ‫كي‬ َ ‫نو‬ َ ْ ‫ب َأ ي‬

70 71 72 73 74

75 76 77

78 79 80 81

82 83

Kolkata has qāla. A, B, L and Bakkāʾ have wa-bi-l-ʾalif. M1 has wa-l-ʾalif and M2 did not correct it. No incidence on meaning. Kolkata has bi-l-ʾalif wa-l-lām al-ḫafīfa which is a clear mistake. Kolkata has taḥarraka, which also was in M1 before correction by M2. The introduction of the apodosis by fāʾ and the passive form may reflect a higher language standard. Not only has M2 corrected the form in the ductus, but he also has added the corrected form in margin and added a collation mark. As above, in note 37, this could be the work of more than one corrector. A, B, L, Kolkata and Bakkāʾ have ʾaṣlihi. M1 has al-ʾaṣl and M2 did not correct it. A and Bakkāʾ have fī. B, L and Kolkata have ʿalā. M1 has ʿalā and M2 did not correct it. This ġayr is inconsistent with the teaching: the verb takes a final kasra before the article and the waṣla, because it is the base form, i.e. a Ḥijāzī majzūm form, and there is no difference with Tamīm in this. This is the rule for all verbs, including non-geminated ones. M2 has struck this ġayr through. It is found in no other version collated here. Kolkata has lam tabqa. M1 had a different example, which has been erased by M2. All the versions collated here have qāḍin. M1 had a different wording. Al-Sīrāfī’s commentary (Šarḥ XIV, 62.9) has a combination of the two readings: ʾiḏā ijtamaʿa sākināni. It is surprising that M2 has struck these words through, because all of A, B, L, Bakkāʾ and al-Sīrāfī’s commentary (XIV, 62.9–10) have them. Kolkata has ʾillā fī al-ʾalif wa-l-ʾalif wa-llām al-ḫafīfa. If it is really the case, as al-Ḫalīl says, that these assimilated verbs have been treated like ʾayna, kayfa and sawfa, i.e. with an invariable ending fatḥa, then M2 is right to strike these words through because these verbs would take a fatḥa in all cases, even before a waṣla and an article. According to Bakkāʾ, Mawṣil adds raḥimahu Allāh taʿālā. Kolkata adds rḥ. Kolkata has ʾiḏā.

148

druel

َ ‫خر‬ ِ ‫ الآ‬86‫[ ي ُت ْب ِعوا‬8] ‫ عربيتّ هُ ولم‬85‫ ممن ترُ تضى‬84‫[ بنو أسٍد وغير ُهم من بني تميم وسمعنا‬7] ‫الأّولون وهم‬ ‫ ومنهم‬92‫[ قالوا اب ْن ُم ٌ واب ْن ٍِم واب ْنمَ ًا‬9] 91(90‫ كما‬89‫ل‬ َ ‫خر َ الأّو‬ ِ ‫ الآ‬88‫ و َأتبعوا‬87‫ئ‬ ٍ ِ ‫ل كما )قالوا ا ِْمرُؤ ٌ وا ِْمر‬ َ ‫الأّو‬ ʘ 95⸂‫ن‬ َ ْ ‫ في جميع الأشياء ⸃ك َأ ي‬94‫يجعله‬ َ ‫[ح ًا‬10]‫ مفتو‬93 ⸂‫من ي َد َعه إذا جاء بالألف واللام ⸃على حاله‬ ‫وزعم يونس أن ّه سمعهم يقولون‬ 96⸉ ً‫ت ولا كلابا‬ َ ‫ف إن ّك من نمُ َي ْرٍ ⸊فلا كعب ًا بلغ‬ َ ْ ‫طر‬ ّ َ ‫[ ال‬11] ‫ض‬ ّ َ ُ‫غ‬ ‫[ مجراها‬12] ‫ ولـكن يجعلها في الفعل تجري‬98⸂‫ البت ّة ⸃من قال هلَ َ ُم ّا وهلَ ُم ّ ِي‬97‫مؤخروهلَ ُ َمّ مقدملا ي َكسر‬ ‫ل حال فيجعله بمنزلة‬ ّ ‫[ ومن العرب من ي َكسر ذا أجمع على ك‬13] ʘ َ ‫في لغة أهل الحجاز بمنزلة ر ُو َي ْد‬

84 85

86 87

88 89 90 91

92 93

94 95

96 97

98

A, B, L, Kolkata and Bakkāʾ have samiʿnāhu. M1 has samiʿnā and M2 has not corrected it. The form samiʿnāhu is probably better. A, Kolkata and Bakkāʾ have turḍā. B and L have turtaḍā. M1 has turtaḍā and M2 has not corrected it. Both forms have the same meaning and construction and it is not clear whether one form is better than the other. Kolkata has yutbiʿūhu, which makes no sense. A, B, L, Kolkata and Bakkāʾ add here wa-mraʾan. M1 does not have it and M2 has not added it (actually, M2 has put the whole expression between brackets, which may explain why he did not correct what was between the brackets). A, B, L, Kolkata and Bakkāʾ have fa-ʾatbaʿū. M1 has wa-ʾatbaʿū and M2 has not corrected it. The lām is unclear. To be checked directly on the manuscript. A, B, L and Bakkāʾ have wa-kamā. Kolkata has only kamā. The text that M2 has put between brackets is found in the matn of A, B, L, Kolkata and Bakkāʾ, with the variants described above. It could be that M2 considered it to be an added gloss. A, B, L and Bakkāʾ have ibnimin wa-bnumun wa-bnaman. Kolkata has ibnimin ibnumun ibnaman. M1 has ibnumun wa-bnimin wa-bnaman and M2 did not correct it. M1 did not have the expression ʿalā ḥālihi, which is found in all the versions collated here. M2 has added it. Without it, the sentence is less clear in M1. Just like in notes 1 and 35 above, this could be a trace of a less “polished” original text. Kolkata has only one dot under the yāʾ, which is probably a typo, not to be read bi-jaʿlihi. The addition of ka-ʾayna by M2, which is found in all the versions collated here, makes the text clearer by providing an example. See notes 1, 35 and 93 above for other cases where M2 makes M1 easier to read. M2 has completed the verse in the margin, but this addition is not found in the other versions collated here. A, Kolkata and Bakkāʾ have wa-lā yaksiru halumma. B and L have wa-lā taksiru halumma. M1 has wa-halumma lā yaksiru and M2 has corrected it into wa-lā yaksiru halumma by adding the two words muʾaḫḫar and muqaddam above the expression. The wording of M1 should probably be read in the passive wa-halumma lā yuksaru, since M1 does not contain the following example added by M2. This could once more be considered a less polished reading in M1. All the versions collated here have the expression man qāla halummā wa-halummī, that M2 added to M1. It modifies the text by adding a condition to the impossibility of an end-

ambrosiana x 56 sup.

149

ِ‫ل ح ُرّك لالتقاء الساكنين‬ ٌ ‫ بالألف واللام لأن ّه فع‬99‫ك وإن لم تجئ‬ َ َ ‫ب اب ْن‬ ِ ِ ‫[ وا ِض ْر‬14] ‫ل‬ َ ُ ‫ب ال َر ّج‬ ِ ِ ‫ا ِض ْر‬ ‫ من‬101‫[ يا فتَ َى‬16] ّ‫ يقول هلَ ُ ِم‬100‫ل ولا يقولها في هلَ ُ َمّ ]و[لا‬ َ ُ ‫ب ال َر ّج‬ ِ ِ ‫ك واض ْر‬ َ َ ‫ب اب ْن‬ ِ ِ ‫[ وكذلك ا ِض ْر‬15] ‫ف الفعل‬ َ ّ ‫[ ت ُصر َف تص ُر‬17] ‫ ولا ي َكسر هلَ ُ َمّ أحدٌ لأّنها لم‬102⸂َ ‫يقول هلَ ُم ّوا فيجعلها بمنزلـ]ـته[←⸃ـة ر ُو َي ْد‬ ‫ على أّنهم يقولون للنساء‬103‫[ وأهل الحجاز وغير ُهم مجتمِعون‬18] ʘ ٌ ّ‫ب وغ َن ِي‬ َ ‫ولم تقَ و قو ّته ومن ي َكسر‬ ٌ ْ‫كع‬ ‫ف‬ ٍ ‫ل حر‬ ّ ‫ هاهنا لأمر ٍ ولا نهٍي وكذلك ك‬104⸂‫ن الدال لم ]××××[←⸃ت َسكن‬ ّ ‫[ وذلك لأ‬19] َ‫ا ُْرد ُْدن‬ ‫[ لازم ٌ له‬1] [‫ظ‬63] ‫ن السكون‬ ّ ‫ ترى أ‬105‫يجزم ألا‬ َ ‫ف‬ ٍ ‫[ قبل نون النساء لا ي َسكن لأمر ٍ ولا لحر‬20] ‫[ )أخبرني‬2] َ‫ أن يرَ ْد ُْدن‬107ّ‫ وهّن يرَ ْد ُْدنَ وعلي‬106َ‫في حال النصب والرفع وذلك قولك ]×[←⸃ر َ⸂د َْدن‬ ‫يجري غير ُ المضاع َف قبل‬ َ [3] ‫ وكذلك‬108(‫ت‬ َ ْ ‫أبو عبيدة أّنهم لا يقولون هلَ ُمّ ز َي ْد ًا إن ّما يقولون هلَ ُمّ َأن‬ ‫ن و ي َْذه َب ْنَ فلماّ كان هذا الحرف‬ َ ْ ‫[ و ي َض ْر ِ ب‬4] ‫ن‬ َ ْ ‫ يح َر ّك في حال وذلك قولك ض َر َب‬109‫نون النساء لا‬ ‫كن فيه‬ ّ ‫[ السكون حاجز ًا عنه ما سواه من الإعراب وتم‬5] ‫ل موضٍع وكان‬ ّ ‫يلزمه السكون في ك‬ ‫ أو لحرف الجزم‬110ٍ ‫يجزم ]لـ[لأمر‬ ُ ‫[ من الفعل كرهوا أن يجعلوه بمنزلة ما‬6] ‫كن في غيره‬ ّ ‫ما لم يتم‬

99 100

101 102 103 104 105 106

107 108 109 110

ing kasra in halumma, whereas M1 teaches that this impossibility is absolute, for some speakers, and that they always treat halumma like ruwayda, i.e. not like an assimilated verb. Could this addition by M2 actually reflect a gloss that was absent in M1? Kolkata has yajiʾ. Unlike all the collated versions, M1 has wa-lā, which seems to be introducing a new argument. The suppression of this wāw makes the text smoother since what follows is not a new argument but an example. Kolkata has two dots below the yāʾ, which are probably a typo, not intended to be read fatayya. M2, just like all the other versions, explicitly adds the example of ruwayda instead of a mere pronoun. This change makes the text easier to follow. B and L are the only ones to have mujmaʿūn. Kolkata is not vocalized. M1 had a different word. All other versions have taskun. The original word may be readable directly on the manuscript. The ʾalif is unreadable. To be checked directly on the manuscript. M1 had a different example. The original word is unreadable. It could be either wadadna or ṣadadna, according to the examples found above in the text. All other versions have radadna. Kolkata has ʿalā with an ending ʾalif maqṣūra and a superscript ʾalif, which is clearly a typo. This gloss was inserted here in the text in M1, although it does not relate to the topic at stake here. M2 has put it between brackets. A and Bakkāʾ are the only ones to have wa-lā. M1 has the definite li-l-ʾamr, which M2 corrected into li-ʾamrin. The difference is insignificant.

150

druel

←[‫[ ]و؟‬8] ‫ ومثل ذلك قولهم‬112‫ يلزمه السكون كلزوم هذا ال ّذي هو غير مضاع َف‬111‫[ فلا‬7] ‫[ السكون فيه‬9] ‫ على هذه التاء كما بنُ ي على النون وصار‬114‫ن الحرف بنُ ي‬ ّ ‫ت لأ‬ ّ ُ ‫ وم َد َْد‬113‫ت‬ ّ ُ ‫⸃ر َ⸂د َْد‬ ‫سا‬ ً ‫ن نا‬ ّ ‫ أ‬116‫ وغيره‬115‫ وزعم الخليل‬ʘ ‫[ فتٍح‬10] ‫بمنزلته فيما فيه نون النساء يدل ّك على ذلك أن ّه في موضع‬ ‫ بمنزلة ر َ َدّ وم ََّد وكذلك‬119‫ جعلوه‬118⸂‫ت‬ ُ ⸃←[××××]ّ‫[ ور َ َد‬11] َ‫ وم ََّرن‬117َ‫من بكر بن وائل يقولون ر َ َدّن‬ َ ‫ ر َ َدّد‬120‫[ فأمّا‬13] ʘ ‫ت لك في لغة أهل الحجاز وغيرهم والبكر ي ّين‬ ُ ‫يجري كما ذكر‬ َ [12] ‫جميع المضاع َف‬ ‫[ ولم يكونوا ليحر ّكوا العين الُأولى‬14] 121‫ن في َلتقيان‬ ِ ‫و يرُ َدِّد ُ فلم ي ُدِغموه لأن ّه لا يجوز أن ي َسكن حرفا‬ ‫[ يرَ فعوا ألسنتهم مّرتينِ فلماّ كان ذلك لا ي ُنجيهم أجروه على‬15] ‫لأّنهم لو فعلوا ذلك لم ي َنجوا من أن‬ [17] ‫ن الشعراء إذا اضط ُر ّوا إلى ما يجتمع أهل الحجاز‬ ّ ‫ واعلم أ‬ʘ ‫ غيره‬122[××]‫[ يجز‬16] ‫الأصل ولم‬ 125⸂‫ب‬ ٍ ‫ب بن ُأمّ صاح‬ ُ َ ‫ قعَ ْن‬124‫ قال الشاعر ⸃وهو‬ʘ ‫ على الأصل‬123‫وغيرهم على إدغامه أجر]×[وه‬ 128‫ضننِ وُ ا‬ َ ‫ وإْن‬127‫جود ُ لَأق ْواٍم‬ ُ ‫ من خ ُل ُقِي َأن ِ ّي َأ‬126‫ت‬ ِ ْ ‫ل قد ج َرّ ب‬ َ ِ‫[ م َه ْل ًا َأعاذ‬18] 129‫[ ⸃وقال⸂ ومثله قول الراجز‬19]

111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129

B, L and Kolkata have fa-lam. Kolkata has al-muḍāʿaf. Again, M1 had a different example, probably wadadtu, which M2 corrected into radadtu, which is the reading in all the other versions collated here. B has yubnā. According to Bakkāʾ, Mawṣil adds raḥimahu Allāh taʿālā. M2 has added a deletion mark above wa-ġayruhu, which was in M1. This wa-ġayruhu is also found in al-Sīrāfī’s commentary (Šarḥ XIV, 64.4). Kolkata vocalized ruddanā. M1 probably had two examples here and M1 has erased the second one and replaced by a large ending tāʾ of the first example. Kolkata has the strange yajʿalūhu. We would have at least expected a marfūʿ form. Kolkata has the obviously erroneous wa-lammā. Kolkata is the only one to have the erroneous final nūn that was in M1 before M2 struck it through. M2 has erased the two ending letters. The original reading of M1 is not clear. M2 has erased a letter. The original reading of M1 is not clear. A, B, L and Bakkāʾ omit wa-huwa. Kolkata has qāla Ibn ʾUmm Ṣāḥib. M1 did not mention the name of the poet. Maybe this is a sign of an earlier, less explicit version of the text. Kolkata has an erroneous yāʾ instead of the bāʾ. Kolkata has al-ʾaqwām. Kolkata has ẓaninū. M2 has put deletion marks above the expression wa-miṯluhu qawl al-rājiz, which is found only in M1, and added wa-qāla, which is the reading all other versions contain.

151

ambrosiana x 56 sup.

‫ل‬ ِ َ ‫ل و َأْظل‬ ٍ َ ‫ الو َجَى م ِْن َأْظل‬130‫ي َْشكُو‬ ʘ ٌ ‫ النحو في الشعر كثير‬131‫[ وهذا‬20] Numerals: Ambrosiana 70 v°.16–71 v°.15 = Derenbourg chapter 412, II, 176.12– 177.19 = Kolkata 721–722 = al-Bakkāʾ V, 257–258 133‫[ لتبي ّن ما العدد‬18] ‫[ المؤن ّث والمذك ّر‬17] ‫ على عّدة‬132‫[ هذا باب الأسماء التّ ي توُ ق َُع‬16] [‫ظ‬70] ِ‫ن ما جاوز الا ِث ْن َي ْن‬ ّ ‫ اعلم أ‬ʘ َ ‫[ وت ِْسَع ع َش ْر َة‬19] َ ‫ الا ِث ْن َي ْنِ والث ِّن ْت َي ْنِ إلى أن تبلغ ت ِْسع َة َ ع َش َر‬134‫إذا جاوز‬ ُ ‫[ التّ ي هي علامة‬21] ُ ‫ن الأسماء التّ ي تبي ّن بها عّدتهَ مؤن ّثة ٌ فيها الهاء‬ ّ ‫[ مماّ واحد ُه مذك ّر ٌ فإ‬20] ِ ‫إلى الع َش َر َة‬ ‫ل‬ ٍ ‫ و َأْر بعَ ة ُ َأْجما‬136‫[ فيها الهاء ُ التّ ي هي علامة‬22] ٌ ‫ ث َلاثةَ ُ ب َن ِينَ بها عّدتهَ مؤ َن ّثة‬135 ⸂‫التأنيث وذلك قولك ⸃له‬ 138⸉‫س َت ّة ُ َأْحم ِرةَ ٍ وكذلك جميع ها⸃ذا⸂ ⸊صح‬ ِ ‫[ و‬1] [‫و‬71] ‫[ إذا كان الواحد ُ مذك ّر ًا‬23] ‫س‬ ٍ ‫ َأف ْرا‬137ُ ‫وخَم ْسة‬ 140[‫تخرج هذه الهاء]ات‬ ُ ‫ كان الواحد ُ مؤن ّث ًا فإن ّك‬139⸂‫[ فإ]ذا[←⸃ن‬2] َ ‫ٺثبت فيه الهاء ُحت ّى تبلغ الع َش َر َة‬ ‫ت‬ ٍ ‫ث ب َنا‬ ُ ‫ ث َلا‬141‫[ له‬4] ‫ت فيها علامة ُ التأنيث وذلك قولك‬ ْ ‫[ الأسماء وتكون مؤن ّثة ً ليس‬3] ‫من هذه‬ ‫ حت ّى‬142‫ت وكذلك جميع ⸃ه⸂ذا‬ ٍ ‫[ وث َمان ِي بغَ لَ ا‬5] ‫ت‬ ٍ ‫سب ُْع ت َمرَ ا‬ َ ‫ت ل ِب ْنٍ و‬ ُّ ‫س‬ ِ ‫قو‬ ٍ ُ ‫س َأي ْن‬ ُ ْ ‫و َأْر ب َُع ن ِْسو َة ٍ وخَم‬ ‫ت‬ َ ‫[ كأن ّك قل‬7] َ ‫ت َأح َد َ ع َش َر‬ َ ‫[ المذك ّر ُ الع َش َر َة َ فزاد عليها واحدًا قل‬6] ‫ فإذا جاوز‬143[‫تبلغ الع َش ْر َ]ة‬

130

131 132 133 134 135 136

137 138 139 140 141 142 143

A has naškū; B has yaškū; L, Kolkata, Bakkāʾ and al-Sīrāfī’s commentary (Šarḥ XIV, 65.6) have taškū (Kolkata even has an ending ʾalif as in plural verbs). M1 has yaškū and M2 has not corrected it. Kolkata has only hāḏā without the wāw. Kolkata has taqaʿu. Kolkata has li-l-ʿadad. Kolkata has jāwazat. M1 did not have this lahu. All the versions collated here have it, and M2 added it to the text. It is not found in al-Sīrāfī’s commentary (Šarḥ XIV, 116.7). M1 has struck through these words that are repeated in M1, probably due to the similarity between the words tabayyana and banīn. It is surprising that M1 did not correct himself, considering that the text stops abruptly (allatī hiya ʿalāma) before continuing with the correct text. If M1 had noticed the error, why didn’t he erase the passage and rewrite over it? Kolkata has ʾaw ḫamsatu. M2 has added ḏā and a collation mark in the margin. A, B, L, Bakkāʾ and Kolkata have wa-ʾin. M1 had fa-ʾiḏā which M2 corrected into fa-ʾin. A, B and Bakkāʾ have the plural. L and Kolkata have a singular. M1 had a plural, which M2 corrected it into a singular. M1 is the only one to have this lahu and M2 did not correct it. A, B, L, Bakkāʾ and Kolkata do not have it. M1 is the only version that did not have hāḏā but only ḏā. M2 has added the hāʾ. M1 had the form al-ʿašara which M2 corrected into al-ʿašr. It is consistent with the exam-

152

druel

‫جعلا اسم ًا‬ ُ 146‫[ ]حرفـ[←⸃اسمـ⸂ان‬8] ‫ وهما‬145⸂‫ف‬ ٌ ‫ت في ع َش َر َ ]×××××××[←⸃أل‬ ْ ‫ وليس‬144‫ل‬ َ َ ‫َأح َد َ جَم‬ ‫[ ال ّذي كان عليه مفرد ًا‬9] ‫ بنائه‬150‫ عن‬149⸂َ ‫ ⸃ َأح َد‬148‫ ولم يغي ّر‬147[‫واحدًا ضم ّوا َأح َد َ إلى ع َش َر َ]ة‬ ‫ كان منفرد ًا‬152 ⸂‫ بنائه حيـ]ـث[←⸃ـن‬151‫خر ُ على غير‬ ِ ‫[م ًا وجاء الآ‬10]‫ت له َأح َدٌ وعِش ْرونَ عا‬ َ ‫حين قل‬ 156‫ فزاد واحدًا‬155[‫ث الع َش ْر َ]ة‬ ُ ّ ‫ جاوز المؤن‬154‫ ع َش َر َة ً وإن‬153‫[ ]تـ[←⸃ يـ⸂ـجاوز‬11] ‫والعد]ة[←⸃د ُ⸂ لم‬

144 145

146

147

148

149 150 151 152

153

154 155

156

ples dealt with here to use the form ʿašr, since the text is dealing with feminine counted objects. But it is more common to use the forms with a tāʾ marbūṭa when referring to numerals in isolation. In the end, both forms can be accounted for and none is better than the other. Kolkata has the incorrect jamalan. M2 has erased the word that was initially found in M1 and replaced it by ʾalif, which is the reading found in all the versions collated here. According to al-Sīrāfī in his commentary (Šarḥ XIV, 119.14–120.2), some people thought that the expression laysa fī ʿašar ʾalif was a mistake and that what was intended was laysa fī ʿašar hāʾ (i.e. tāʾ marbūṭa). Al-Sīrāfī comments by saying that Sībawayhi intends here to correct the erroneous dialectal forms such as ʾaḥadā–ʿašar, with a long ʾalif. M1 had another reading, which is unfortunately lost. A, Bakkāʾ and al-Sīrāfī’s commentary (XIV, 117.6) have ḥarfāni. B, L and Kolkata have ismāni. M1 had ḥarfāni, which M2 corrected into ismāni. It is common to find ḥarf in the meaning of ism, so in the end it is difficult to prefer one reading over the other. As above in note 143, one might assume that M1 intended the isolated form ʿašara before coalescence in the compound numeral, whereas M2 corrected it in the actual form -ʿašar, after coalescence, which is the reading of all the collated versions. A and Bakkāʾ have the plural yuġayyirū. B, L and Kolkata have the singular. M1 also has the singular, to be read in the passive. The plural would imply an active form meaning ‘they’. The addition of ʾaḥad by M2 makes the text easier to follow. All the other versions have it. Kolkata has min. Kolkata omits this ġayr, which is a clear misunderstanding of the text (or simply a typo). In our three chapters, ḥayṯu (6 times) and ḥīna (twice) are apparently interchangeable. In addition to these eight cases, we see here that M2 replaces ḥayṯu in M1 by ḥīna, which is the reading of all the other versions. And in another case, see note 180, M1 and M2 agree on ḥayṯu against most of the versions, that have ḥīna. M2 replaces the expression al-ʿidda lam tujāwiz by al-ʿadad lam yujāwiz, which is the reading of all the other versions. In this chapter, ʿadad is found 5 other times, and ʿidda, also 5 other times, both words being eventually associated with the verb jāwaza. Kolkata has wa-ʾiḏā. Same kind of correction as described in notes 143 and 147. When mentioning the form alone, M1 systematically prefers the isolated form with tāʾ marbūṭa rather than the forms that actually appear in the examples discussed. Bakkāʾ has fa-zādū ʾaḥadan where the initial wāw and ʾalif in wāḥid were confused with the plural mark in the verb.

ambrosiana x 56 sup.

153

‫ و بلغة أهل الحجاز‬158 َ⸂‫ ِإحْد َى ]××××[←⸃ن َبقِ َة‬157⸂‫ت‬ َ ‫[ ِإحْد َى ع َش ِر َة َ بلغة بني تميم ⸃كأن ّما قل‬12] ‫ت‬ َ ‫قل‬ ‫جعلا اسم ًا‬ ُ ‫ن‬ ِ ‫ حرفا‬161‫ وهما‬160⸂َ َ‫ت ِإحْد َى ]××××[←⸃ت َم ْرة‬ َ ‫ قل‬159⸂‫[ ِإحْد َى ع َش ْر َة َكأن ّـ]ك[←⸃ـما‬13] ‫ت‬ َ ‫[ حين قل‬16] 162ً ‫[لها مـ]ـنـ[ـفردة‬15]‫[ واحدًا ضم ّوا ِإحْد َى إلى ع َش ْر َة َ ولم يغُ ي ّروا ِإحْد َى عن حا‬14] ‫ت له ا ِث ْن َا ع َش َر َ قلت له اثنا‬ َ ‫[ َأح َد َ ع َش َر َ قل‬17] ‫ زاد المذك ّر ُ واحدًا على‬163‫سنةَ ً وإْن‬ َ َ‫له ِإحْد َى وعِش ْرون‬ َ ‫ت الواحد‬ َ ّ‫ إذا ثني‬168⸂‫ الا ِث ْن َي ْنِ عن حالـ]ـه؟[←⸃ـهما‬167‫[ لم تغي ِّروا‬18] َ ‫ ع َش َر‬166‫ له ا ِثنْ َْي‬165‫ن‬ ّ ‫ فإ‬164‫عشر‬ ِ‫ الا ِث ْن َي ْن‬170⸂‫ بمنزلة النون والحرف ال ّذي قبل ⸃الـ⸂ نون ⸃في‬169[‫ن ع َش َر َ]ة‬ ّ ‫[ النونَ لأ‬19] ‫ت‬ َ ‫غير أن ّك حذف‬ ‫ لا‬173‫[ فيما ينصرف وما‬21] ‫ كخمَ ْسَة َ ع َش َر َ وقد بينّ اّ ذلك‬172‫ وليس حاله‬171‫[ حرف الإعراب‬20]

157 158

159 160 161 162

163 164 165 166 167

168

169 170 171 172 173

The omission of the expression ka-ʾannamā qulta in M1 makes the text difficult to follow, if not incorrect. M1 had a different example from all the versions collated here. Unfortunately, M2 has erased it and the original word is not readable any more. In his commentary, al-Sīrāfī (Šarḥ XIV, 119.4) has the same example, nabiqa. M1 replaced ka-ʾannaka by ka-ʾannamā, which is the reading of all the other versions. Both forms are found in our chapters. Same case as presented in note 158. In his commentary, al-Sīrāfī (Šarḥ XIV, 119.5) has the same example, tamra. Kolkata has the erroneous form humāni. A and Bakkāʾ have munfaridatan. B, L and Kolkata have mufradatan. M1 had munfaridatan, which M2 corrected into mufradatan. Both mufrad and munfarid are used once each in the lines above in the same meaning of ‘singular’. A, B, L and Kolkata have wa-ʾin and Bakkāʾ has fa-ʾin. M1 has fa-ʾin, which M2 did not correct. M2 has struck through these words, which are repeated in M1, with no apparent explanation. A, B, L, Bakkāʾ and Kolkata have wa-ʾinna. M1 has fa-ʾinna, which M2 did not correct. A has the incorrect iṯnā. A, B, L, Bakkāʾ and Kolkata have the singular lam tuġayyir. M1 has a plural, which M2 did not correct. This plural makes no sense here, since the whole sentence is constructed in the singular (2nd pers. masc.), not the plural. M2 has corrected the singular of M1 into a dual, which is the reading of all the other versions. However, the singular makes more sense here, since the text comments what happens to iṯnāni, not to ʿašar. Same correction as described above in notes 143, 147 and 155. M2 changes the construction of the expression in order to align it on the other versions, without any incidence on the quality of the text. A and Bakkāʾ have ḥarf ʾiʿrāb, without the article. M2 has inserted a deletion mark above ḥāluhu. The text was actually easier to follow in M1. This mā is not found in A, B, L and Bakkāʾ. It is found only in Kolkata. M1 has it and M2 did not correct it.

154

druel

176‫ت له ]ا[ث ِن ْتا‬ َ ‫[ قل‬22] 175⸂َ ‫ث واحدًا على ]أ[←⸃ِإ⸂حْد َ⸃ى⸂ ع َش ْر َ⸃ة‬ ُ ّ ‫ زاد المؤن‬174⸂‫ينصرف وإ]ن[←⸃ذا‬ 178[‫[ و بلغة أهل الحجاز ع َش ْر َة َ ولم تغي ِّر]وا‬23] َ ‫ ع َش ِر َة‬177‫ن له ث ِن ْت َْي ع َش ِر َة َ وا ِ⸃ث ْـ⸂ـنتَ َْي‬ ّ ‫ع َش ِر َة َ وا ِثنْ َت َا ع َش ِر َة َ وإ‬ ‫ت في‬ ْ ‫ت هنا كما ذهب‬ ْ ‫ن النون ذهب‬ ّ ‫ت الواحدة َ إلّا أ‬ َ ّ‫ ثني‬180‫[ حيث‬1] [‫ظ‬71] ‫ عن حالهما‬179ِ‫الث ِّن ْت َي ْن‬ ‫[ غير‬3] 181⸂‫ف ال ّذي بعد ِإحْد َى وث ِن ْت َي ْنِ ⸃على‬ ُ ‫سواء ٌ و بنُ ي الحر‬ َ ‫صة المذك ّر والمؤن ّث‬ ّ ‫[ ق‬2] ‫ن‬ ّ ‫الا ِث ْن َي ْنِ لأ‬ ‫ل فإذا انتقل عن‬ ٍ ‫ظ له بناء ٌ في حا‬ ُ ‫[ اللف‬4] ‫ وقد يكون‬182‫بنائه والعدد ُ لم يجاوز الع َش ْر َ كما فعُ ل ذلك بالمذك ّر‬ َ ‫[ ز َب ِينة‬6] ‫ق َأف َقِ ّ ٌي وفي‬ ِ ُ ‫[ ذلك تغيير ُهم الاسم في الإضافة قالوا في الُأف‬5] ‫ فمن‬ʘ ‫تلك الحال تغي ّر بناؤه‬ ‫ن‬ ّ ‫ العدد ُ واحدًا على ا ِثنْ َْي ع َش َر َ فإ‬184‫[ وإذا زاد‬7] ʘ ‫ كثير ٌ في الإضافة وقد بينّ اّ ه في بابه‬183‫ز َبان ِيّ ٌونحو ذا‬ ‫[لته‬9]‫ العّدة ُ ثلاثة ً والآخر بمنز‬186‫[ بناؤه عن حاله و بنائه حيث لم تجاوز‬8] ‫ لا يتغي ّر‬185‫الحرف الأّول‬ ‫ وكذلك‬189⸂‫[ ع َش َر َ ⸃عبدًا‬10] َ ‫ ث َلاثة‬188 ⸂‫ قولك ⸃له‬187‫ ]فمن[←⸃و⸂ذلك‬ʘ ِ‫حيث كان بعد َأح ٍَد وا ِث ْن َي ْن‬ ‫ف الأّول بمنزلته‬ ُ ‫[ واحدًا فوق ث ِن ْت َْي ع َش ِرة َ فالحر‬11] ُ ‫ما بين هذا العدد إلى ت ِْسعة َ ع َش َر َ وإذا زاد العدد‬ 174

175

176

177 178

179 180 181 182 183 184 185 186 187 188 189

M1 corrected ʾiḏā by ʾin. Above, see note 139, it was the other way round. Altogether, ʾiḏā is found 18 times and ʾin 15 times in our three chapters in equivalent conditional constructions, so that one cannot be said to be preferable to the other. Just like in the four cases above (see notes 143, 147, 155 and 169), M2 corrects the masculine form of M1 into the feminine form which is discussed here. M2 has the same reading of all the other versions collated. Kolkata has the incorrect ṯintā. Apparently, M1 had the same form iṯnatā ʿašira twice. M2 has erased the waṣla in the first occurrence, turning the example into ṯintā ʿašira, which is present in all the other versions. M2 had a missing ṯāʾ in the word iṯnatay. M2 has corrected the plural form of M1 into a singular. The plural makes no sense, since the whole sentence is constructed in the singular. See above, note 167. Did M2 forget to correct this plural form above? Kolkata has iṯnatayni. A, B, L and Bakkāʾ have ḥīna. Kolkata is the only one to have ḥayṯu. M1 has ḥayṯu and M2 did not correct it. See above, note 152. The omission of ʿalā in M1 is clearly a mistake. The sentence makes no sense without it. All the other versions have it. Kolkata has the obvious typo al-ḏikr. A, B, L and Bakkāʾ have fa-naḥwu hāḏā. Kolkata has wa-naḥwu hāḏā. M1 has wa-naḥwu ḏā and M2 did not correct it. No reading can really be said to be better than the others. Kolkata has the erroneous ʾarāda. Kolkata has al-ḥarf allaḏī al-ʾawwal. A has lam yujāwiz. And Kolkata has lam yatajāwaz. M2 corrected the expression fa-min ḏālika by wa-ḏālika, which is found in all the other versions. None can be said to be better. Same correction as above, see note 135. M1 did not have ʿabdan, which is found in all the other versions. Without it, the example is still understandable, although less clear.

ambrosiana x 56 sup.

155

‫[ وذلك قولك‬13] 191ِ‫[ العّدة ُ ثلاثاً والآخر بمنزلته حيث كان بعد ِإحْد َى وث ِن ْت َي ْن‬12] 190‫حيث لم تجاوز‬ َ ‫[ إلى ت ِْسَع ع َش ِرة‬14] ‫ وكذلك ما بين هذه العّدة‬193⸂‫ث ع َش ِر َة َ جار ية ً ⸃وع َش ْر َة َ بلغة أهل الحجاز‬ َ ‫ ث َلا‬192‫له‬ ʘ ‫ هذا الباب‬196‫[ في‬15] ⸂‫ بين التأنيث والتذكير في جميع ما ذكرنا ⸃من‬195⸂‫ ⸃ما‬194‫]فـ[فرقّ وا‬

Bibliographical References A

Primary Sources

Sībawayhi, Kitāb = ʾAbū Bišr ʿAmr ibn ʿUṯmān Sībawayhi (d. ca. 180/796), al-Kitāb. Ed. by Hartwig Derenbourg, Le livre de Sîbawaihi. Paris: Imprimerie nationale, 1881–1889. (Repr., Hildesheim and New York: G. Olms, 1970.)/Ed. by Kabīr al-Dīn ʾAḥmad, Hāḏā Kitāb ismuhu al-Kitāb wa-huwa fī l-naḥw miṯla ʾumm al-Kitāb. Kolkata: Maṭbaʿ Urdū Gāʾīd, 1887./Ed. Būlāq, Kitāb Sībawayhi. Cairo: al-Maṭbaʿa al-ʾAmīriyya, 1898./Ed. by ʿAbd al-Salām Hārūn, Kitāb Sībawayhi, I. Cairo: Dār al-Qalam, 1966; II. Cairo: Dār al-Kitāb al-ʿArabī, 1968. III–V. Cairo: al-Hayʾa al-Miṣriyya al-ʿĀmma li-l-Kitāb, 1973– 1977./Ed. by Émile Badīʿ Yaʿqūb, al-Kitāb. Beirut: Dār al-Kutub al-ʿIlmiyya, 2009./Ed. by Muḥammad Kāẓim al-Bakkāʾ, al-Kitāb. Beirut: Maktabat Zayn, 2015. /Ed. by Muḥammad Fawzī Ḥamza, al-Kitāb. Cairo: Maktabat al-ʾĀdāb, 2015. Suyūṭī, Buġya = Jalāl al-Dīn ʾAbū l-Faḍl ʿAbd al-Raḥmān ibn ʾAbī Bakr al-Suyūṭī (d. 911/ 1505), Buġyat al-wuʿāt fī ṭabaqāt al-luġawiyyīn wa-l-nuḥāt. Ed. by Muḥammad ʾAbū l-Faḍl ʾIbrāhīm. Cairo: ʿĪsā al-Bābī al-Ḥalabī, 1964–1965. Sīrāfī, Šarḥ = ʾAbū Saʿīd al-Ḥasan ibn ʿAbdallāh al-Sīrāfī (d. 368/979), Šarḥ Kitāb Sībawayhi, XIV. Ed. by Hudā Qirāʿa. Cairo: Dār al-Kutub wa-l-Waṯāʾiq al-Qawmiyya, 2010.

190 191 192 193 194 195 196

A has lam yujāwiz. Kolkata has lam yatajāwaz. A has bayna ʾiḥdā wa-ṯalāṯīna, which makes no sense, instead of baʿda ʾiḥdā wa-ṯintayni. M1 is the only one who has this lahu and M2 did not correct it. See the opposite situation above, in notes 135 and 188. M1 did not supply the Ḥijāz form, which is found in all the other versions and added by M2. A and Bakkāʾ have fa-farraqū. B, L and Kolkata have farraqū. M1 had fa-farraqū but M2 corrected it to farraqū. The reading with fāʾ was probably smoother to read. M1 did not have this mā, the addition of which is maybe slightly better grammatically. All the other versions have it. M2 corrected fī by min, which is the reading found in all the other versions. None can be said to be better than the other.

156 B

druel

Secondary Sources

Druel, Jean N. 2012. Numerals in Arabic grammatical theory: An impossible quest for consistency? Ph.D. diss., Nijmegen University. Humbert, Geneviève. 1992. Premières recherches sur le Kitāb de Sībawayhi. I. Les voies de la transmission. II. Les documents. Ph.D. diss., Université de Paris-VIII. Humbert, Geneviève. 1993. “Un témoin fossile du Kitāb de Sībawayhi”. Développements récents en linguistique arabe et sémitique, ed. by Georges Bohas, 121–139. Damascus: Institut français de Damas. Humbert, Geneviève. 1994. “Le Kitāb de Sībawayhi d’ après l’ autographe d’ un grammairien andalou du XIIe siècle”. Le manuscrit arabe et la codicologie, ed. by AhmedChouqi Binebine, 9–20. Rabat: Faculté des lettres et de sciences humaines. Humbert, Geneviève. 1995. Les voies de la transmission du Kitāb de Sībawayhi. Leiden: E.J. Brill. Khalidov, A.B. 2000. “A Kazan manuscript of Sībawayhi”. Manuscripta orientalia 6/2.8– 9. Munajjid, Ṣ. al- 1960. al-Kitāb al-ʿarabī l-maḫṭūṭ ʾilā l-qarn al-ʿāšir al-hijrī. I. al-Namāḏij. Cairo: Jāmiʿat al-Duwal al-ʿArabiyya, Maʿhad al-Maḫṭūṭāt al-ʿArabiyya.

Conditionality: Syntax and Meaning in al-Sīrāfī and Ibn Sīnā Manuela E.B. Giolfo and Wilfrid Hodges

1

Introduction: What Counts as a Conditional Particle?1

In his paper “Two conceptions of irreality in Arabic grammar: Ibn Hišām and Ibn al-Ḥājib on the particle law” Kees Versteegh (1991:79) cites the analysis by Dévényi (1988) for the fact that in the Arabic grammatical tradition “the particle law is not regarded as a conditional particle”. He comments that this is true at least for Sībawayhi, given that “[f]or Sībawayhi there was no syntactic reason to deal with law and in simultaneously, since only one of them governs the jussive”. But, he goes on to say, for later grammarians, such as Ibn Hīšām (d. 761/1360) and Ibn al-Ḥājib (d. 646/1249) the semantic similarities between law and the paradigm conditional particle ʾin made it sensible to treat the two particles under the same head. He traces the reason for this change of view to “the influx of Greek logic” (1991:80), naming al-Fārābī (d. 339/950) as a spokesman for the Greek logical tradition. Versteegh was certainly right that we find grammarians saying that particles that do not govern the apocopate should not be counted as conditional. In Giolfo and Hodges (2017:257) we quoted al-Sakkākī’s (d. 626/1229) remark that [s]ome grammarians have added to the conditional words (kalimātu alšarṭ) kullamā, although the principles of syntax would exclude that, inasmuch as it was established that, rightly, conditional particles operate the apocope of the verb, whilst kullamā does not have anything to do with the apocope of the verb. The particle excluded from the conditionals here is not law but kullamā, a particle which Ibn Sīnā consistently preferred to ʾin in his own conditional logic.

1 Although the ideas of this paper come from a joint research project of both authors, in the present article Manuela E.B. Giolfo is to be held responsible for paragraphs 1, 2, 4 and 6, and Wilfrid Hodges for paragraphs 3 and 5.

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_010

158

giolfo and hodges

However, we think that even in Sībawayhi the position is more complicated than Versteegh (1991) suggested. In the first place one of us (Giolfo 2015) has argued that the fact that law never takes the apocopate is not a purely syntactic fact, as Versteegh (1991) took it to be. From early times the apocopate was used to express that the clause in question is mubham, i.e. lacks certainty.2 When we say ‘ʾin p’ with the apocopate in p, we commonly express lack of certainty about whether or not p. But in a typical counterfactual clause of the form ‘law p’ we express that p is in fact false, so there is no uncertainty and hence no apocopate. Thus there always was a semantic reason for distinguishing between ʾin and law, and quotations from Sībawayhi show that he was very much aware of it. In section 4 of the present paper we attempt a formal analysis of the distinction that al-Ḫalīl was making when he said, in a remark quoted by Sībawayhi, that conditional particles have to be mubham. On our analysis the point that he was making was one of considerable interest to the logicians too, but it represented only one of many aspects of conditional statements. From several points of view it makes perfect sense to treat law and ʾin in parallel, and in fact Sībawayhi himself does so in places where he discusses these aspects. At Sībawayhi (Kitāb I, 114.14, quoted below) he explicitly speaks of, “law in the role of ʾin” (law bi-manzilati ʾin). In sections 2 and 3 below we will take material from Sībawayhi and the commentary of al-Sīrāfī, together with some parallel material from Ibn Sīnā, to build up a picture of aspects of conditional statements that were within the purview of these writers. Much of this material does in fact treat law together with other particles that today we would regard as conditional. It is in any case not completely clear that Sībawayhi has a single consistent distinction between those particles that are conditional and those that are not. Besides speaking of ḥurūf al-šart, we saw above that he speaks of particles that are bi-manzilati ʾin (also at Sībawayhi, Kitāb I, 385.4); elsewhere he talks of ḥurūf al-jazāʾ (Kitāb I, 386.14, quoting al-Ḫalīl), ḥurūf al-mujāzāt (Kitāb I, 57.6), the particles which yujāzu bihā (Kitāb I, 385.19) and maʿnā l-mujāzāt (Kitāb I, 56.12). Some at least of these categories include both law and ʾin, giving scope for them to be treated in parallel.

2 Cf. Sībawayhi (Kitāb I, 385.22) “ʾin is always uncertain, and all the conditional particles are like that” (ʾin ʾabadan mubhamatun wa-ka-ḏālika ḥurūfu l-jazāʾi), and Sībawayhi (Kitāb I, 386.12) “hypothetical particles operate the apocope of the verbs, being the apodosis apocopated by what precedes [i.e. protasis]” (ḥurūfu l-jazāʾi tajzimu l-ʾafʿāla wa-yanjazimu l-jawābu bi-mā qablahu).

conditionality: syntax and meaning in al-sīrāfī and ibn sīnā

159

So Versteegh’s remarks on law provide our introduction to conditionals in Sībawayhi, al-Sīrāfī and Ibn Sīnā. But our main concern is with conditionals in general, not just law and ʾin. The picture we have is that there is a broad class of sentences that might be regarded as conditional, and that observations of the Arabic authors lead to natural classifications of these sentences. Through Ibn Sīnā’s contributions we can see what a significant role some of these classifications play in the logic of conditionals. It is also helpful to note that these classifications fall in line remarkably well with views on conditionals that have become standard in recent linguistics—though we aim to read the Arabic authors on their own terms. We hope that this paper makes clear the richness and sophistication of the insights that these Medieval authors had into conditional statements.

2

Law

Sībawayhi has a section (section 52, Kitāb I, 109–116) which is ostensibly about how implicit verbs can govern the cases of nouns, but as al-Sīrāfī’s commentary (Šarḥ II, 156–169) explains, it is really about what governs the cases of nouns in conditional sentences. The main answers are in terms of the ‘verb of the condition’, which can be either explicit or implicit. In this section both Sībawayhi and al-Sīrāfī consider law and ʾin as parallel expressions. The mixture of syntax and semantics in this section of Sībawayhi needs more unpicking than we have space for here. But we quote one passage from Sībawayhi and one from al-Sīrāfī. First from Sībawayhi: Thus “Bring me a beast of burden, even if (wa-law) a donkey [accusative]”. […] If you say “even if a donkey” with a genitive, it is in the same category as with “if” (ʾin). Likewise someone says “I gave you a dirham, so how about (hallā) a dinar?”, where [“how about”] plays the same role as “if” (ʾin) here, operating the verbs. It is bad to use the nominative in “how about a dinar?” and in “even if a donkey”, because rather than making it depend on an implicit “there is”, it would be better to use a verb addressed to the interlocutor. […] law plays the same role as ʾin, what comes after it are only the verbs; so if what comes after it is a noun, then there is an implicit verb at that place, and the nouns are operated by that verb […] (wa-ʾātinī bi-dābbatin wa-law ḥimāran […] wa-law qulta wa-law ḥimārin fa-jararta kāna bi-manzilatihi fī ʾin wa-miṯluhu qawlu baʿḍihim ʾiḏā qulta jiʾtuka bi-dirhamin fa-hallā dīnārin wa-huwa bi-manzilati ʾin fī hāḏā l-mawḍiʿi tubnā ʿalayhā l-af ʿālu. wa-l-raf ʿu qabīḥun fī fa-hallā dīnārun

160

giolfo and hodges

wa-fī wa-law ḥimārun li-ʾannaka law lam taḥmilha ʿalā ʾiḍmāri yakūnu fafiʿlu al-muḫāṭabi ʾawlā bihi […] wa-law bi-manzilati ʾin lā yakūnu baʿdahā ʾillā l-ʾaf ʿālu fa-ʾin saqaṭa baʿdahā ismun fa-fīhi fiʿlun muḍmarun fī hāḏā l-mawḍiʿi tubnā ʿalayhi l-asmāʾu). Kitāb I, 114.9–15

Presumably the verb addressed to the interlocutor would be ‘give me’, operating ‘dinar’ and ‘donkey’. (A small slice of life here: the dinar/dirham exchange rate fluctuated enormously throughout the Islamic period, but the dinar was the gold coin, which was much more valuable than the dirham, which was the silver coin!) Suppose someone were to say: We saw you claiming that law, when it has a response, is followed only by the verb, because it contains the meaning of condition. Then people say: ‘If only (law) that (ʾanna) Zayd comes to me, I will indeed honor him’, and they do not say: ‘[If only] that (ʾanna) Zayd is standing, then I will indeed honor him’. So they distinguish between the case where the ḫabar is a noun and the case where it is a verb. When the ḫabar is a verb, they give it the role of the verb of the condition, as when you say: ‘If (ʾin) Zayd stood, I would honor him’. Here ‘Zayd’ is the mubtadaʾ and ‘stood’ is its ḫabar and ‘stood’ represents the verb of the condition, so it has the same meaning as if we said: ‘If Zayd stood, we would honor him’. It was reported that one can say: ‘If (law) that (ʾanna) Zayd stood’, because ‘that’ (ʾanna) may have in conjunction to it, after law, an implicit verb in the underlying original sentence, and the verb that is the ḫabar of ‘that’ (ʾanna) is an elucidation thereof [i.e. of this implicit verb], as if we said ‘If only (law) it were a fact that (ʾanna) Zayd stood’ or ‘If only (law) it were known [that (ʾanna) Zayd stood]’ ( fa-ʾin qāla qāʾilun: fa-qad raʾaynākum tazʿumūna ʾanna ‘law’ allatī lahā jawābun lā yalīhā ʾillā l-fiʿlu li-ʾanna fīhā maʿnā l-šarṭi, ṯumma yaqūlūna ‘law ʾanna zaydan ʾatānī la-ʾakramtuhu’ wa-lā yaqūlūna ‘ʾanna zaydan qāʾimun la-ʾakramtuhu’ fa-faṣluhum bayna lḫabari ʾiḏā kāna isman, wa-ʾiḏā kāna fiʿlan fa-jaʿluhum al-ḫabara ʾiḏā kāna fiʿlan bi-manzilati fiʿli l-šarṭi fa-ka-ḏālika taqūlu ‘ʾin zaydun qāma ʾakramnāhu’ wa-yakūnu zaydun mubtadaʾan wa-qāma ḫabarahu, wa-nāba qāma ʿan fiʿli l-šarṭi fa-ka-ʾannā qulnā ‘ʾin qāma zaydun ʾakramnāhu’ fī l-maʿnā, qīla lahu ʾinnamā jāza ‘law ʾanna zaydan qāma’ li-ʾanna ‘ʾanna’ qad waqaʿa ʿalayhā fiʿlun muḍmarun baʿda ‘law’ ʿalā l-ʾaṣli llaḏī qaddamnāhu wa-l-fiʿlu llaḏī huwa ḫabaru ‘ʾanna’ tafsīrun lahu, ka-ʾannā qulnā law ṣaḥḥa ʾanna zaydan qāma ʾaw law ʿurifa). Sīrāfī, Šarh II, 162.1–8

conditionality: syntax and meaning in al-sīrāfī and ibn sīnā

161

We note that in both these passages ʾin and law are treated as having parallel properties. Moreover, the properties are partly syntactic and partly semantic. In fact the two passages, and that of al-Sīrāfī in particular, suggest the beginnings of a semantic theory of conditionals. A key point is the syntactic observation that both ʾin and law need to be followed immediately by a spoken or implied verb. The verb is needed as the ‘verb of the condition’. Both Versteegh (1991:79) and al-Sīrāfī (e.g. Šarḥ II, 162.14) quote Sībawayhi’s explanation (Kitāb II, 334.8) that law is used “for that which was going to happen on account of the occurrence (wuqūʿ) of something else”; al-Sīrāfī paraphrases ‘occurrence of something else’ as wuqūʿ al-šart. So there must be something in the condition to explain what ‘occurs’, and it is a small jump to suppose that the role of the ‘verb of the condition’ is precisely to specify the required type of occurrence. Arguably most nouns do not specify a kind of occurrence. We will come back to these points. But before we leave Sībawayhi’s use of law, we should note that although he does not say very much about the word, he uses it constantly, and in one particular way. Typical examples are: If you said: ‘He was a member (ʾaḥadun) of so-and-so’s family’, it would not be allowed (law qulta kāna ʾaḥadun min ʾāli fulānin lam yajuz). Kitāb I, 20.17

If you say: ‘This Zayd, he is good and beautiful’, this is correct speech (law qulta hāḏā zaydun fa-ḥasanun jamīlun kāna kalāman jayyidan). Kitāb I, 58.21

Don’t you see that if you count fī and law and similar words as nouns, this makes things difficult? (ʾa-lā tarā ʾannaka law jaʿalta fī wa-law wanaḥwahā isman ṯaqqalta). Kitāb II, 331.20

These uses all have it in common that there are no literal ‘occurrences’, either in the antecedent or in the consequent. For example the first quotation says that the sentence ‘He was a member of so-and-so’s family’ is not permissible Arabic. The third says that the theory that fī and law and similar words are nouns creates difficulties. No questions of time past, present or future are involved. Nor is it relevant whether anybody actually has said ‘This Zayd, he is good and beautiful’. And finally Sībawayhi has many similar examples with ʾin, such as: “If (ʾin) you say ‘Hit whichever of them comes to you’, you use the nominative” (ʾin qulta ʾayyuhum jāʾaka fa-ḍrib rafaʿtahu, Kitāb I, 57.5)

162

giolfo and hodges

So it seems that ʾin and law are exchangeable in this usage. (We have not found in Sībawayhi or al-Sīrāfī any examples of this use of ʾin with an apocopated verb.) The usage can be described as follows. The clause ‘law p’ invites the reader to consider some entity or type of entity, usually either a form of words or a context in which somebody speaks that form of words. The consequent makes a statement about the type of entity under consideration. The function of the word law is to be an anchor for the description of the type of entity, and the description is given by a verbal sentence, but the verb usually has the minimum content required to make the entity into a situation: you state a statement, you adopt a theory, etc. We will describe this usage as ‘virtual law’, and similarly ‘virtual ʾin’. We turn to Ibn Sīnā’s use of law. Ibn Sīnā wrote many pages, scattered over many books, about ‘conditional’ (šarṭī) statements and arguments from a logical point of view. We can distinguish between occurrences of conditional words in the sentences of Ibn Sīnā’s formal logic, and occurrences in Ibn Sīnā’s discussion of that logic (his metatheory, if you will). This is like distinguishing between Sībawayhi’s use of law as a part of sentences under discussion, and his use of it in his discussions of sentences. The results are similar for both authors: the vast majority of occurrences of law in Ibn Sīnā’s logic are occurrences of virtual law in his metatheory. Thus for example: “If he says ‘Every person laughs’, he tells the truth” (law qāla kullu ʾinsānin ḍaḥḥākun ṣadaqa, Qiyās IX/1, 422.9). Ibn Sīnā also uses the word law in other ways that are well-known to the linguists. Thus he uses it for counterfactual statements about the past: If this kind of enquiry had been taken into consideration when studying the properties of premises and syllogisms, then people would have said […] But they didn’t do any of this at all ( fa-law kāna hāḏā l-naḥwu min alnaẓari muʿtabiran fī taʿrīfi ʾaḥkāmi l-muqaddamāti wa-l-maqāyīsi, la-qad kāna yuqālu […] lākinnahum lam yafʿalū šayʾan min hāḏā). Qiyās VIII/1, 392.4–11

He also uses it in the sense of ‘even if’: Even if (wa-law) there did exist a vacuum, it does not follow that there would be no humans (lā, wa-law kāna al-ḫalāʾu mawjūdan, yalzamu ʾan lā yakūna l-ʾinsānu mawjūdan). Qiyās V/2, 250.16f.

conditionality: syntax and meaning in al-sīrāfī and ibn sīnā

163

The expression ‘when’ seems not to include this meaning [of ‘whenever’], but rather it requires that the truth of [the consequent] HZ follows even if (wa-law) [it is] from just one posit of [the antecedent] AB (wa-ʾammā lafẓatu ʾiḏā, fa-tušbihu ʾan lā yataḍammana hāḏā l-maʿnā, bal taqḍī bittibāʿin yūjadu min h z, wa-law ʿinda ʾaḥadi ʾawḍāʿi a b). Qiyās V/4, 263.10f.

(Sībawayhi would note that in the second passage, kāna is implicit after walaw.) If we turn to the sentence forms of Ibn Sīnā’s formal logic, we find that he never uses law in these forms when he is discussing their use in arguments. He does very occasionally use it when he is discussing the truth conditions for single conditional sentences. We found just four examples in his major treatment in Qiyās: If (law) five was even then it would be a number (law kānat al-ḫamsatu zawjan la-kāna ʿadadan) Qiyās v/1, 240.13

If (law) someone were to say: ‘If (law) this [number] was even, but not divisible into two equal parts, then it would be odd’, this would be true […] (law qāla qāʾilun: ʾinnahu law kāna hāḏā ʾuṯnuwatan,3 wa-kāna lā yanqasimu bi-mutasāwiyayni, la-kāna takūnu hāḏihi l-ṯanwatu4 fardan, faʾinna hāḏā ḥaqqun […]). Qiyās v/4, 273.16f.

The other two examples are similar; they are found at Qiyās v/5, 283.2f. It makes good sense that law appears in the sentence forms when Ibn Sīnā is discussing their truth conditions, but not when he is discussing their use in formal proofs. There are no special proof rules for false antecedents, and hence the sentence forms in proofs do not need a word like law which is specialized for counterfactual antecedents. But when we are discussing the grounds for believing a conditional sentence, counterfactual conditions do need special consideration, and this is where he uses forms that include law. (Note also the virtual law in the last passage quoted.)

3 ‫[ اثنوة‬sic! transliteration: ʾuṯnuwa?]. 4 ‫[ الثنوة‬sic! transliteration: al-ṯanwa?].

164 3

giolfo and hodges

Quantification over Situations

The previous section allows us to leave the particle law and its peculiarities on one side for the rest of this paper. But it also threw up some ideas about features that law shares with ʾin, and hence probably with other particles that are ‘conditional’ in some broad sense. Some of these ideas prove to be remarkably fruitful in ways that one would hardly have expected at the outset. Take for example the idea that conditional particles are associated with a specification of classes of entities or situations. This same idea became prominent in Western linguistics over the last half century, largely through insights of David Lewis (1975) and Angelika Kratzer (1986); see von Fintel (1994) for a mature account. The following example is due to Kratzer (2012:89), but based on examples of Lewis: Sometimes if a man buys a horse, he pays cash for it. The word ‘if’ heads a phrase ‘if a man buys a horse’, whose role is to name a class of situations (Kratzer says ‘events’), namely those where a man buys a horse. The word ‘sometimes’ quantifies over this class; it says that there are one or more such situations which also have the property that the man pays cash for the horse. The main clause is the consequent ‘he pays cash for it’, and this clause lies within the scope of the quantification over situations. The overall effect is that the sentence expresses ‘There are situations such that S and T’; after the quantifier ‘There are’ we have a conjunction, not an implication. Now consider the sentence forms of the part of Ibn Sīnā’s hypothetical (šarṭī) logic which he calls muttaṣil (the name is not informative—he had historical reasons for it). There are four such forms,5 and he writes them as follows: i. kullamā kāna p fa-q ii. laysa l-batta ʾiḏā p fa-q iii. qad yakūnu ʾiḏā kāna p fa-q iv. laysa kullamā p fa-q.6 5 These forms can be found for example in Ibn Sīnā, Qiyās vi./2. The forms and Ibn Sīnā’s logical use of them are analyzed in detail in Hodges (in preparation). 6 Modern Western readers might be tempted to read Ibn Sīnā’s fa- as the ‘then’ answering to ‘if’ or ‘when’. However, this would be a mistake. See Giolfo (2017:109): “Les mots de Zamaḫšarī “in et law agissent sur deux énoncés en rendant le premier ‘condition’ et le second ‘réponse à la condition’” [Zamaḫšarī (d. 538/1144), Kitāb al-mufaṣṣal fī al-naḥw, J.P. Broch (ed.), Christianiae (1859), quoted in K. Dévényi (1988:19): in wa-law tadḫulāni ʿalā ǧumlatayni fa-taǧʿalāni al-ūlā šarṭan wal-ṯāniyata ǧazāʾan] indiquent clairement que tant in que law ne sont pas du point de vue logique des ‘si’, mais des ‘si … alors’ c’est-à-dire des opérateurs binaires”. In Clas-

conditionality: syntax and meaning in al-sīrāfī and ibn sīnā

165

And there at item iii we have precisely Kratzer’s example, though with ʾiḏā rather than ʾin. Item ii is another of Kratzer’s forms based on Lewis; it comes by replacing ‘Sometimes’ by ‘Never’, and again one can check that the form after the quantifier ‘Never’ is a conjunction and not an implication. Item i (and hence also item iv, which is just item i negated) illustrates the same pattern but with the quantifier ‘Always’ absorbed into the particle kullamā—though in this case we really do have an implication and not a conjunction.7 It is universally agreed that Ibn Sīnā’s four muttaṣil forms above, with their quantifier phrases kullamā, laysa l-batta, qad yakūnu, were his own original contribution.8 What does Ibn Sīnā himself have to say about the relationship between these forms and the more familiar notion of an ‘If … then’ conditional? He attacks an earlier exposition of hypothetical logic as follows: [This book] gives a rotten explanation of conditionals. […] The student should avoid it altogether, because it is full of falsehoods and irrele-

sical Arabic, fa- is used before the consequent of a conditional only if there is some kind of discontinuity between the antecedent and the consequent. The consequent’s being a nominal sentence counts as a discontinuity. See Larcher (2000, 2006). The Arabic hypothetical sentences that Ibn Sīnā uses as illustrations in his logic usually have nominal sentences as their consequents. 7 Readers who want to see more details of the formal analysis can consult von Fintel (1994). In the talk in Genoa we also cited Schubert and Pelletier (1987); their formal analysis of the sentence ‘Usually, when cats drop to the ground, they land on their feet’ duly presents a ∩ and not a ⊆ at the crucial point. Likewise Rescher (1963:54) proposes the formula (∃s)(Ms & Ps) for Ibn Sīnā’s form iii above. Unfortunately the logic required for analysing Ibn Sīnā’s forms and others of the Lewis-Kratzer kind goes beyond what is normally taught in undergraduate logic courses. 8 The most powerful witness to this is Maróth (1989), a book largely devoted to finding Peripatetic origins for Ibn Sīnā’s hypothetical logic wherever possible. Maróth finds hypothetical statements with quantifiers already in the Peripatetic tradition, and particularly in Boethius. However, he comments (1989:115): “Much more important is the fact that in Boethius the quantification means the quantification of the clauses of the [hypothetical] sentence, whereas in Ibn Sīnā the words ‘always’, ‘sometimes’, ‘never’, ‘sometimes not’ that stand at the beginning of the [hypothetical] sentence express a quantification of the way the clauses of the sentence are linked together. The words ‘every’, ‘some’ etc., which appear inside the clauses, do not serve to quantify the hypothetical sentence; their only interest is as features of the clauses, which are categorical sentences” (Viel wichtiger ist der Umstand, dass bei Boethius die Quantifizierung die Quantifikation der Aussagenteile bedeutet, während bei Ibn Sīnā die am Anfang der Aussage stehenden Worte “immer”, “manchmal”, “nie”, “manchmal nicht” die Quantifizierung der Aussagenverknüpfung bedeuten. Die in den Aussagen stehenden Worte “alle”, “manche” usw. spielen für die Quantifizierung der hypothetischen Aussage keine Rolle, sie sind nur aus der Sicht des an sich als kategorische Aussage einstufbaren Aussagenteiles interessant).

166

giolfo and hodges

vances. The reason is that its author does not understand how conditionals are affirmative or negative, and how they are universally or existentially quantified, or unquantified […] ( fa-ʾinnahu fāsidun fī taʿrīfi ḥāli l-qaḍāyā l-šarṭiyyati […] fa-yajibu ʾan lā yaltafita l-mutaʿallimu ʾilā ḏālika l-battata, fa-ʾinnahu muzāġatun wa-muḍallatun wa-ḏālika li-ʾanna muṣannifahu lam yaʿrif al-ʾījāba wa-l-salba fī l-šarṭiyyāti, wa-l-kulliyyata wa-ljuzʾiyyata wa-l-ʾihmāla kayfa yakūnu […]). Qiyās VI/6, 356.12–17

Ibn Sīnā’s forms i and ii illustrate ‘universally quantified’, while iii and iv are examples of existential quantification. In all these cases the quantifier is ostensibly over times, but Ibn Sīnā’s examples make clear that he includes situations in some sense ‘unquantified’ (muhmal to the logicians). The sentences in question are ones that make some reference to a time that is left hanging in the sentence itself; so the reference needs to be fixed by the context in which the sentence is used. Ibn Sīnā claims that conditional sentences with ʾin or ʾiḏa are unquantified: If someone said: ‘If (ʾin) it were so, then so and so’, and ‘When (ʾiḏā) it is so, then so and so’, the sentence is unquantified […] (wa-ʾammā ʾiḏā qīla: ʾin kāna ka-ḏā, fa-ka-ḏā ka-ḏā; wa-ʾiḏā kāna ka-ḏā, fa-ka-ḏā ka-ḏā; fal-qaḍiyyatu muhmalatun […]). Qiyās 263.5f.

The details have not yet been clarified, but we can see some sense in this with English examples. Contrast a promise and a threat: (a) I’ll remind you of the details when I see you. (b) I’ll remind you of the details whenever I see you. Here (b) quantifies over all future occasions on which I see you; this class of occasions is specified by the phrase ‘whenever I see you’. But (a) is more complicated: it carries an implication that I will see you, and maybe that I will see you soon, but it requires only that I will tell you the details on the first occasion when I see you. This may be what Ibn Sīnā has in mind at Qiyās V/4, 263.10f. (quoted in section 2 above) when he talks about ‘just one posit of the antecedent’ in the case of ʾiḏā. At any rate an explanation of ‘when’ is substantially more complicated to spell out than that of ‘whenever’, and it is plausible that Ibn Sīnā thought that the extra needed to be understood from the context rather than from the sense of the words themselves.

conditionality: syntax and meaning in al-sīrāfī and ibn sīnā

167

Assembling what we have so far, it seems that the Lewis-Kratzer theory of conditionals provides a very congenial framework for understanding both Ibn Sīnā’s choice and classification of conditional sentence forms, and those features of conditional particles that we found in both Sībawayhi and al-Sīrāfī in section 2 above. A conditional statement (in a suitably broad sense of ‘conditional’) consists of a statement (the consequent) which is made within the scope of a quantifier over a set of times, situations, events or something similar. We can call this set the ‘domain’ of the quantifier. The domain is specified by a clause (the antecedent) headed by a conditional particle. This framework allows us to classify conditional statements in several ways. We name four below; there are further possibilities. First classification: What is the quantifier? Is it universal, existential, or something subtler like ‘usually’, or does the whole statement leave the quantifier to be inferred from the context of utterance? We have seen Ibn Sīnā go some way along this dimension of classification. Second classification: Is the quantification explicit or implicit, and if explicit, is it separate from the conditional particle or included within the conditional particle? If separate, does it have to come before the conditional particle, or close to it? We have seen some of Ibn Sīnā’s moves towards this classification. The particle kullamā can be taken to include a quantification over all times. The particle ʾiḏā can appear sometimes without a separate explicit quantifier expression, but also (as in Ibn Sīnā’s item iii) it can receive a separate quantifier (qad yakūnu) within the sentence. Third classification: Is the antecedent explicit or implicit? Is it partly explicit and partly implicit? If implicit, is it picked up from a previous statement? Sībawayhi’s section 52, discussed in our section 2 above, bears directly on this classification. When Sībawayhi and al-Sīrāfī give examples where a verb is implicit after the conditional particle, the verb is (in their words) the ‘verb of the condition’, i.e. it is a part of the antecedent. In section 5 below we will analyze some cases where al-Sīrāfī and Ibn Sīnā claim to find antecedents that are understood but not spoken. Fourth classification: How is the domain of the quantifier determined? For example how does it depend on the choice of conditional particle? This is a question addressed directly by al-Sīrāfī in at least two places.

168

giolfo and hodges

In one of them, al-Sīrāfī (Šarḥ III, 263.21) quotes al-Ḫalīl’s statement that “ʾin is the mother of the conditional particles” (wa-ʾammā qawlu l-Ḫalīl: (ʾin) hiya ʾummu ḥurūfi l-jazāʾi), and he continues: [This is] because it attaches to the conditional expression from any aspect. This is not so for the rest of what introduces a conditional sentence, because man introduces conditional sentences about rational beings, mā about irrational ones, ʾayyu about things that [are presented as] divided up, matā for time, ʾayna and ḥayṯumā for place, and ʾannā similarly, and ʾiḏ mā is used only by a few and not all the Arabs know it ( fa-li-ʾannahā tadḫulu ʿalā l-jazāʾi fī jamīʿi wujūhihi, wa-laysat ka-ḏā sāʾiru mā yujāzā bihi, li-ʾanna (man) yujāzā bi-hā fī-mā yuʿqalu, wa-(mā) fīmā lā yuʿqalu, wa-(ʾayyu) fī-mā yubʿaḍu, wa-(matā) li-l-zamāni, wa-(ʾayna) wa-(ḥayṯumā) lil-makāni, wa-(ʾannā) naḥwa min ḏālika, wa-(ʾiḏ mā) yatakallamu bi-hā l-qalīlu minhum, wa-mā kullu l-ʿArabi taʿrifuhā9). Šarḥ III, 263.21–24

Not everyone would count ‘who’ as a conditional particle. Presumably al-Sīrāfī has in mind paraphrases like Whoever you meet, hit him. = If you meet anyone, hit him. Sībawayhi (Kitāb I, 57.5) quotes a similar sentence with ʾayyuhum, and says that ‘hit him’ plays the role of consequent with ʾayyuhum as conditional particle (ḥarf al-mujāzā). We do know that al-Ḫalīl refused to accept ʾiḏā as a particle of consequence (ḥarf al-jazāʾ), and in section 5 below we will turn to the reason that he gave. In the second place where al-Sīrāfī discusses the domain of the quantifier of the conditional, he points out the practical consequences of an ambiguity in an example of Sībawayhi. Sībawayhi had asked al-Ḫalīl to comment on the sentence allaḏī yaʾtīnī fa-lahu dirhamāni ‘Whoever comes to me will get two dirhams’ (Kitāb I, 402.20). Sībawayhi’s question was about why fa- is allowed here, when (for instance) if ʿAbdallah comes to me we cannot simply say ʿabdullāhi yaʾtīnī fa-lahu dirhamāni. Al-Sīrāfī raises a different kind of

9 In this passage jazāʾ is the conditional expression or conditional sentence, not the apodosis as it would usually be in later writers. Al-Sīrāfī writes jawāb for the apodosis. Cf. Dévényi (1988: 14).

conditionality: syntax and meaning in al-sīrāfī and ibn sīnā

169

question. He lowers the number of dirhams per visit but allows more than one visit: kullu rajulin yaʾtīnī fa-lahu dirhamun ‘Every man who comes to me gets a dirham’. Al-Sīrāfī comments that the reason why the man gets a dirham is not specified, and this leaves open what is owed to the same person who visits me twice. If the quantification is over people then he gets just one dirham, but if it is over situations of coming (ʾityān) then he gets two. Don’t you see that when he says: ‘Every man who comes to me gets a dirham’, and two men went to him, each of them gets a dirham; and if he said ‘If Zayd came to me, he would get a dirham’, and he went to him twice, he got only one dirham; whilst his expression ‘Whoever comes to me will get a dirham’ is construed with fa-, to clarify that whoever comes would get a dirham on each situation of coming (ʾityān), and if he had said ‘Whoever comes to me will get a dirham’ [without fa-], it could well be that whoever comes would get a dirham on each situation of coming or not, as when he says: ‘Zayd gets a dirham’, and you do not envisage why, and the verb can be perfect (māḍiyan), as when you say ‘Whoever came (ʾatānī) to me got [construed with fa-, i.e. fa-lahu] a dirham’, specifying that he actually got the dirham […] (ʾa-lā tarā ʾannahu ʾiḏā qāla: kullu rajulin yaʾtīnī fa-lahu dirhamun, fa-ʾatāhu rajulāni, wa-likulli wāḥidin minhumā dirhamun; wa-law qāla ʾin ʾatānī zaydun fa-lahu dirhamun, fa-ʾatāhu marratayni lam yastaḥiqq ʾillā dirhaman wāḥidan; wa-qawluhu: allaḏī yaʾtīnī fa-lahu dirhamun, daḫalat al-fāʾu li-tubayyina ʾanna l-dirhama istaḥaqqahu bi-l-ʾityāni, wa-law qāla: allaḏī yaʾtīnī lahu dirhamun jāza ʾan yakūna l-dirhamu yastaḥiqquhu bi-l-ʾityāni, wa-jāza ʾan yakūna bi-ġayrihi, ka-mā yaqūlu: zaydun lahu dirhamun, wa-lam taḏkur sababa stiḥqāqihi li-l-dirhami, wa-yajūzu ʾan yakūna l-fiʿlu māḍiyan kaqawlika: allaḏī ʾatānī fa-lahu dirhamun, yuṯbitu ʾanna l-dirhama staḥaqqahu […]). Šarḥ III, 309.23–310.7

This ambiguity was known also to the logicians, at least in general terms. AlFārābī (Burhān 44.17f.) observes that if most but not all As are Bs, it could be that all As are Bs but each of them for most of the time, or it could be that most As are Bs all of the time, or that most As are Bs most of the time. It is noticeable that al-Fārābī talks in generalities whereas al-Sīrāfī has a precise and concrete example. We have the impression that this is a general contrast between the intellectual characters of al-Fārābī and al-Sīrāfī; but in this particular case the source of al-Sīrāfī’s precision is known. Sībawayhi had suggested that ‘when-

170

giolfo and hodges

ever you come to me’ can usefully be paraphrased as ‘for every coming (ʾityān) of you to me’ (Kitāb I, 402.18), presumably because this converts the quantifier kullamā specialized for time into a standard quantifier kullu. Al-Sīrāfī’s question points out an ambiguity in the nominalization: is an ʾityān a fact of someone’s coming, or is it an occasion on which someone comes? It would be unsurprising if a similar discussion appeared somewhere in the Arabic legal literature, but this is outside our expertise.

4

Conditional as mubham

Sībawayhi reports al-Ḫalīl as saying that ʾiḏā, unlike ʾin, is not a conditional particle, for the following reason: And I asked him about ʾiḏā, and what prevents it from being used as conditional. He said that the verb with ʾiḏā plays the same role as it does with ʾiḏ when you say: ‘I remember [when] you [were] telling me’. So ʾiḏā plays the same role in relation to the future as ʾiḏ does in relation to the past, and this shows that ʾiḏā [refers to something that] occurs at a known time. Don’t you see that if you say: ‘I will come to you when the unripe dates turn red’, it is good, but if you say: ‘I will come to you if the unripe dates turn red’, it is bad. Thus ʾin is always uncertain (mubhamatun), and all the conditional particles are like that. When ʾiḏā is connected to a verb, the verb with ʾiḏā plays the same role as with ḥīna just as if you say ‘At the time at which you come to me, I shall come to you’ (wa-saʾaltuhu ʿan ʾiḏā mā manaʿahum an yujāzū bi-hā fa-qāla l-fiʿlu fī ʾiḏā bi-manzilatihi fī ʾiḏ ʾiḏā qulta ʾataḏakkaru ʾiḏ taqūlu fa-ʾiḏā fī-mā tastaqbilu bi-manzilati ʾiḏ fīmā maḍā wa-yubayyinu hāḏā ʾanna ʾiḏā tajīʾu waqtan maʿlūman ʾa-lā tarā ʾannaka law qulta ʾātīka ʾiḏā ḥmarra l-busru kāna ḥasanan wa-law qulta ʾātīka ʾin iḥmarra l-busru kāna qabīḥan fa-ʾin ʾabadan mubhamatun wa-kaḏālika ḥurūfu l-jazāʾi wa-ʾiḏā tūṣalu bi-l-fiʿli fa-l-fiʿlu fī ʾiḏā bi-manzilatihi fī ḥīna ka-ʾannaka qulta al-ḥīnu llaḏī taʾtīnī fīhi ʾātīka fīhi). Kitāb I, 385.19–24

Al-Ḫalīl says that ʾin is always ‘uncertain’ (mubhamatun). He presumably means that an antecedent headed by ʾin is uncertain. He may also mean that the entire conditional sentence headed with ʾin is uncertain (i.e. hypothetical). In fact Sībawayhi, after stating that “ʾin is always uncertain, and all the conditional particles are like that” (ʾin ʾabadan mubhamatun wa-ka-ḏālika ḥurūfu l-jazāʾi, Kitāb I, 385.22), adds that “hypothetical particles operate the apocope of the

conditionality: syntax and meaning in al-sīrāfī and ibn sīnā

171

verbs, being the apodosis apocopated by what precedes [i.e. protasis]” (ḥurūfu l-jazāʾi tajzimu l-ʾafʿāla wa-yanjazimu l-jawābu bi-mā qablahu, Kitāb I, 386.12). It is worth noting here that [p]our Blachère et Gaudefroy-Demombynes seuls l’ “apocopé exprime un procès dont la réalisation est incertaine ou conditionnelle” (Blachère & Gaudefroy-Demombynes 1952, p. 255) et ils trouvent justement dans cette affirmation la justification de l’emploi de l’“apocopé dans les phrases contenant une notion d’éventuel ou d’hypothétique” (ibid.). Giolfo 2017:66

Elsewhere Sībawayhi uses mubham to mean demonstrative pronoun, but that is hardly its meaning here. Al-Ḫalīl uses it in contrast with ‘at a known time’, which indicates that we have some lack of knowledge about an antecedent beginning with ʾin. Al-Ḫalīl goes on to give two examples where the domain given by the antecedent is ‘time(s) at which the dates turn red’. It seems that the difference between the case with ʾiḏā and the case with ʾin is that the statement with ʾiḏā implies that the domain is not empty, i.e. there is a time at which the dates turns red, whereas the statement with ʾin has no such implication. We will put al-Ḫalīl’s remarks alongside a comment by Ibn Sīnā. Before Ibn Sīnā, logicians recognized two main kinds of conditional (šarṭī) sentence. One had the form ‘If … then …’ and was known as muttaṣil, and the other had the form ‘Either … or …’ and was known as munfaṣil. Ibn Sīnā observed that these munfaṣil sentences would never be described as šarṭī in ordinary Arabic usage, and he asked whether a motivated definition of šarṭī could be given that would include both the muttaṣil sentences and the munfaṣil. His answer was: But [the Peripatetic logicians] also describe the munfaṣil sentences as conditional. It is as if they mean by ‘conditional’ a sentence that contains a proposition with an attachment which deflects it from being a proposition [in its own right] and makes it [just] a part of a proposition. Don’t you see that the sentence ‘The sun is up’ makes a true or false statement, but when you attach an addition so that you say ‘If the sun is up’, you deflect the proposition so that it ceases to be a proposition in the sense of making a true or false statement? Similarly the sentence ‘It is daytime’ makes a true or false statement, but when you attach to it an addition so that you say ‘then it is daytime’, you deflect the proposition so that it ceases to be a proposition, and your clause ‘Then it is etc.’ […] does not make a true or false statement (lākinnahum yusammūna l-munfaṣilata ʾayḍan

172

giolfo and hodges

šarṭiyyatan wa-ka-ʾannahum yaʿnūna bi-l-šarṭiyyati mā yulḥaqu fīhi biqaḍiyyatin min al-qaḍāyā ziyādatan tuḥarrifuhā ʿan ʾan takūna qaḍiyyatan wa-tajʿaluhā juzʾa qaḍiyyatin ʾa-lā tarā ʾannahu kāna qawluka ‘al-šamsu ṭāliʿatun’ qawlan ṣādiqan ʾaw kāḏiban fa-lammā ʾalḥaqta bihi l-ziyādata faqulta ‘ʾin kānat al-šamsu ṭāliʿatan’ fa-ḥarrafta l-qaḍiyyata fa-ṣārat ġayra qaḍiyyatin ḥīna zāla ʿanhā ʾan takūna ṣādiqatan ʾaw kāḏibatan? wa-kaḏālika kāna qawluka ‘al-nahāru mawjūdun’ qawlan ṣādiqan ʾaw kāḏiban falammā ʾalḥaqta bihi l-ziyādata fa-qulta ‘fa-l-nahāru mawjūdun’ fa-ḥarrafta l-qaḍiyyata fa-ṣārat ġayra qaḍiyyatin, fa-ʾinna qawlaka ‘fa-kāna ka-ḏā’ […] lā ṣādiqun wa-lā kāḏibun). Mašriqiyyūn 61.8–15

In effect Ibn Sīnā here redefines šarṭī to mean ‘containing subclauses which are neither affirmed nor denied when the sentence as a whole is stated’. This new definition has the advantage of including his own new muttaṣil sentence forms. Now we can see that al-Ḫalīl’s comment on the sentence ‘I will come to you if the grapes turn red’ expresses that this sentence is šarṭī in Ibn Sīnā’s new sense, since the clause ‘the dates (will) turn red’ is not stated (and not denied either) when the sentence as a whole is stated. On the other hand the sentence ‘I will come to you when the dates turn red’ is strictly not šarṭī in Ibn Sīnā’s new sense, since a statement of this sentence does imply a statement that the grapes will turn red, and hence also a statement that I will come to you. Since Ibn Sīnā certainly did regard statements beginning with ʾiḏā as šarṭī, it looks as if there is a disagreement between Ibn Sīnā and al-Ḫalīl at this point. But a closer analysis shows that there are some important subtleties involved. They might seem a subtlety too far, but be assured that everything will fall into place. When Ibn Sīnā introduced his four muttaṣil forms as listed in section 3 above, he devised a logic for them (the logic PL2 in Hodges) which was formally an exact copy of Aristotle’s categorical syllogistic. He himself emphasized the formal equivalence. Now the categorical sentence equivalent to Ibn Sīnā’s first form (kullamā kāna p fa-q) was the sentence form Every A is a B. Ibn Sīnā took this form to carry the implication that there is at least one A. He was very explicit about this point (see ʿIbāra II/1, 79.11–81.1 and Hodges 2012). The implication that there is at least one A is known as the ‘existential import’ of the sentence ‘Every A is a B’.

conditionality: syntax and meaning in al-sīrāfī and ibn sīnā

173

If the common formal structure between the two logics carries over to a common semantic structure—and there is every reason to believe it does—then Ibn Sīnā will be reading his muttaṣil sentence form as saying: At every time at which p is true, q is true; and moreover there is a time at which p is true. The clause after the semicolon expresses the existential import. Al-Ḫalīl’s statement with ʾiḏā is a statement of a similar form, so Ibn Sīnā should read it as: For every (or perhaps the first) time at which the dates turn red, I will come to you at that time; and moreover there is a time at which the dates turn red. If we have understood al-Ḫalīl’s remarks correctly so far, they show that al-Ḫalīl ascribed existential import to sentences of the form ‘When p then q’, but not to sentences of the form ‘If p then q’. If so, then his statement that ʾin is mubham has a rather precise meaning, namely that conditionals formed with ʾin have no existential import. The question where existential import applies is quite complicated, even for Ibn Sīnā’s own logics. He certainly assumed it for muttaṣil sentences in his logic PL2, but it was optional in his more advanced logic PL3 (Hodges, in preparation). Chatti (2016) is a recent discussion of existential import in modal sentences.

5

Taqdīr/taḥlīl

Sībawayhi and his successors sometimes make use of a form of linguistic analysis that carries the name taqdīr. More precisely, this form of analysis takes a spoken or written phrase, and replaces it by its taqdīr, which is another phrase that is taken to be a better representation of what the speaker or writer had in mind. This form of analysis has a partial parallel in the procedure that Ibn Sīnā and other logicians call taḥlīl lit. ‘analysis’, which replaces a spoken or written argument by an argument that more accurately represents the logical content of the argument that the speaker or writer intended. Unlike the linguists, Ibn Sīnā has no name for the phrase after the analysis; he is content to have a name for the process of analyzing. On the other hand he explains at some length what needs to be done in analysis, and he devotes several sections of Qiyās to advice and examples. Thus he says:

174

giolfo and hodges

[In the argument to be analyzed] sometimes a part is hidden, or there is a superfluous addition, and sometimes [even] when it is simple, the pieces are jumbled out of their natural order, or a piece is missing or added (muḍmaran fīhi šayʾun, ʾaw mūradan fīhi ziyādatan ġayra muḥtājin ʾilayhā wa-rubbamā kāna basīṭan, wa-muḥarrafan ʾayḍan ʿan tartībihi l-ṭabīʿī ʾaw nāqiṣan, ʾaw zāʾidan). Qiyās IX/6, 460.6–8

By contrast Sībawayhi himself is quite elusive about what counts as a taqdīr, in fact he uses the word only once, though he makes use of the general concept often enough. We paraphrase a summary by Aryeh Levin (1997:144) of what he takes to be the central features of a taqdīr in Sībawayhi himself and later Arabic linguists: The taqdīr contains things ‘concealed’ (muḍmar) in the spoken text. The taqdīr removes ‘superfluous’ (zāʾid) pieces of the spoken text. The taqdīr switches the order of some items. The taqdīr replaces one text by another. The list is notably close to Ibn Sīnā’s list of changes made in analysis, even up to use of some of the same terms. But of course any revision of text is likely to involve additions, subtractions, permutations and so forth; the purposes behind linguistic taqdīr and logical taḥlīl might still be completely different. For example, linguists who invoke taqdīr nearly always do so in order to justify morphological facts about inflections of nouns and conjugations of verbs, or sometimes facts about pronunciation. For a logician like Ibn Sīnā, who claims to be describing phenomena that apply across all ‘possible languages’, facts of these kinds are of no interest at all. Closer inspection shows that there are in fact significant connections between the notions, at least when the linguist in question is al-Sīrāfī. We will concentrate on one example, namely the discussion in Sībawayhi section 253 (Kitāb I, 399–401), where Sībawayhi presents the view of al-Ḫalīl: Al-Ḫalīl stated that all of these primitive forms contain the meaning of ‘if’ (ʾin), and this is the reason why the apodosis ( jawāb) is apocopated. In fact, when he says ‘Come to me, I will come to you’ the meaning of his statement is indeed ‘If (ʾin) a visit from you occurs, I will come to you’. And when he says ‘Where is your house? I will visit you’, it is as if he said ‘If (ʾin) I knew the address of your house I would visit you’, because his phrase ‘Where is your house?’ means ‘Tell me!’ (wa-zaʿama l-Ḫalīl ʾanna hāḏihi l-

conditionality: syntax and meaning in al-sīrāfī and ibn sīnā

175

ʾawāʾila kullahā fīhā maʿnā ʾin fa-li-ḏālika injazama l-jawābu li-ʾannahu ʾiḏā qāla ‘iʾtinī ʾātika’ fa-ʾinna maʿnā kalāmihi ‘ʾin yakun minka ʾityānun ʾātika’ wa-ʾiḏā qāla ‘ʾayna baytuka ʾazurka’ fa-ka-ʾannahu qāla ‘ʾin ʾaʿlam makāna baytika ʾazurka’ li-ʾanna qawlahu ‘ʾayna baytuka’ yurīdu bihi ‘ʾaʿlimnī’). Kitāb I, 399, 12–15

As Sībawayhi presents it, al-Ḫalīl’s argument is not very convincing. It is true that in the context in question you might get the same effect by saying ‘If you come to me’ as you would by saying just ‘Come to me’; but this hardly shows that the imperative contains a hidden ‘if’. One might guess that al-Ḫalīl was content to paraphrase the imperative into a form with ʾin, just because ʾin can take the apocopate. But if we turn to al-Sīrāfī’s account, we find a subtler argument for the same conclusion. Thus: Don’t you see that when he says ‘Come to me, I will come to you’, the one who commands is not obliged to visit the person being commanded unless that person comes to him; and when he says ‘Where is your house? I’ll come and visit you’, that does not oblige him to make the visit unless the person [asked] does tell him his address? The imperative and the interrogative expression do not signify this meaning [of ‘if’], but what does reveal it is the conditional expression, so that it [i.e. the meaning of ‘if’ (ʾin)] has to be recognized in line with these things (ʾa-lā tarā ʾannahu ʾiḏā qāla ‘iʿtinī ʾātika’, lam yulzam al-ʾāmiru ʾan yaʾtiya l-maʾmūra ʾillā baʿda ʾan yaʾtiyahu l-maʾmūru, wa-ʾiḏā qāla ‘ʾayna baytuka ʾazurka’ lam yalzamhu l-ziyāratu ʾillā baʿda ʾan yuʿarrifa baytahu, wa-lafẓu l-ʾamri wa-l-istifhāmi lā yadullu ʿalā hāḏā l-maʿnā, wa-llaḏī yakšifuhu lafẓu l-šarṭi, fa-wajaba taqdīruhu baʿda hāḏihi l-ʾašyāʾi). Šarḥ III, 299.14–17

The argument here is pragmatic. We can recognize that if you say to me ‘Where do you live? I’ll come and see you’, but I decline to tell you where I live, then by not telling you my address I release you from any obligation to keep your promise to visit me. But this implies that your statement ‘I’ll come and see you’ carries only a conditional obligation. So the statement must be subject to an implied condition, and we can see what the condition would say. This argument is not an instance of what Levin (1997:143) cites, that linguists invoke the notion of taqdīr “when they find that the literal construction of a given utterance does not accord with one of their theories”. Al-Sīrāfī’s argument is independent of any theory about the form of the utterance; for example it

176

giolfo and hodges

makes no reference to the fact that the utterance uses the apocopate. Instead his argument is in terms of what is needed for the utterance to fulfill the practical purpose that we can see it has. There may well be many other examples that fit Levin’s account of the motivation behind the notion of taqdīr; al-Sīrāfī’s is simply not one of them. In fact al-Sīrāfī’s argument is remarkably close to the logical argument that underlies a piece of logical analysis by Ibn Sīnā. Just as al-Sīrāfī does, Ibn Sīnā finds that in certain cases a condition is understood but not spoken. At Qiyās VIII/3, 410.10–17, Ibn Sīnā points to arguments in which a conditional antecedent is implicit, having been stated in an earlier sentence. For concreteness, here is an example of the kind of argument that Ibn Sīnā almost certainly had in mind; it is taken from the first book of Euclid’s Elements. Demonstration. If the two [lines] are not parallel then when they are both extended on one of the two sides, they meet. So we extend them on the side BD so they meet in a point K if that is possible, so the angle AHT external to the triangle KTH is greater than the internal angle KTH, … and this is absurd (burhānuhu ʾinnahumā ʾin lam yakūnā mutawāziyayni faʾinnahumā ʾiḏā ʾuḫrijā fi ʾiḥdā l-jihatayni ltaqiyā fa-nuḫrijuhumā fī jihati bd fa-yaltaqiyāni ʿalā nuqṭati k ʾin ʾamkana dālika fa-taṣīru zāwiyatu aḥṭ alḫārijatu min muṯallaṯi htk ʾaʿẓamu min zāwiyati htk al-dāḫilati … wa-hāḏā ḫalfun). Besthorn and Heiberg 1893:114–116, transl. in Hodges 2017

Ibn Sīnā’s claim is that the antecedent ‘If the two lines are not parallel’ should be understood as applying to the second sentence of the demonstration, as well as the first sentence (where it is explicit). There is a clear logical reason for this. Namely, without the assumed antecedent there is no reason to believe that ‘the angle AHT external to the triangle KTH is greater than the internal angle KTH’. Without the implied antecedent there is no obligation on us to believe that statement. But the whole purpose of a geometrical proof is to show that certain things must be true. So in both cases the argument is the same. From our understanding of the use of language, we can see that the speaker by uttering a certain statement puts herself under an obligation to do or accept something. But the obligation is not correctly stated until we add an unspoken condition to the statement; so the unspoken condition must be implicit in making the statement. Incidentally Kratzer (2012:108) also points to examples of conditional statements where the antecedent is implicit, having been stated in an earlier sentence. Unlike in Ibn Sīnā’s examples, her examples introduce a new quantifica-

conditionality: syntax and meaning in al-sīrāfī and ibn sīnā

177

tion, but over the domain introduced in the earlier sentence. We do not know of any similar examples discussed in the Medieval Arabic literature. Al-Sīrāfī also has an interest in conditionals where the consequent is implied and not spoken. As to what al-Ḫalīl said about the suppression of the consequent (ḥaḏfu l-jawāb) in the words of God […] there is a consensus of the grammarians that the explanation of some passages of this kind in the Qurʾān is that the consequent ( jawāb) is suppressed […] and the consequent ( jawāb) is in the taqdīr ( fī-mā yuqdaru). […] They say in the book Meanings [of al-Farrāʾ] that suppression is frequent in the Qurʾān and in the speech of the Arabs, so that one can have the consequent ( jawāb) suppressed (wa-mā ḏakarahu l-Ḫalīl min ḥaḏfi l-jawābi fī qawli llāh […] waqad ijtamaʿa l-naḥwiyyūna wa-jāʾa l-tafsīru fī baʿḍi mā fī l-Qurʾāni min naḥwi ḏālika ʾannahu maḥḏūfu l-jawābi, […] wa-jawābuhā fī-mā yuqdaru […] wa-ḏakarūhā fī Kitābi [l-Maʿānī] ʾanna l-ḥaḏfa kaṯīrun fī l-Qurʾāni wa-kalāmi l-ʿArabi, wa-ʾiḏā kāna ka-ḏālika […] qad inḥaḏafa jawābuhu […]). Šarḥ III, 311.1–23

We know of no examples of this in Ibn Sīnā.

6

Redundant Antecedents

In the previous section we considered antecedents that are not stated as such but have to be understood, given the purpose of the discourse. We turn to the opposite phenomenon: antecedents that are expressed although the consequent does not depend on them (at least semantically). The double-act of Sībawayhi and al-Ḫalīl is extraordinarily effective in finding interesting examples, and in this case they provide us with two examples from the Qurʾān: I asked al-Ḫalīl about mahmā, and he said it is mā where you attach to it mā as a null word, in the same role as it has with matā when you say ‘Whenever (matā mā) you come to me, I will come to you’ and in the same role as it has with ʾin when you say ‘If ever (ʾin mā) you come to me, I will come to you’, and the role it has with ʾayna as the Almighty said ‘Wherever (ʾayna mā) you are, death will find you’ (Q. 4/78), and the role it has with ʾayy when you say ‘Whatever (ʾayyan mā) you call him, he has the most beautiful names’ (Q. 17/110) […] (wa-saʾaltu l-Ḫalīl ʿan mahmā fa-

178

giolfo and hodges

qāla hiya mā ʾadḫalta maʿahā mā laġwan bi-manzilatihā maʿa matā ʾiḏā qulta ‘matā mā taʾtinī ʾātika’ wa-bi-manzilatihā maʿa ʾin ʾiḏā qulta ‘ʾin mā taʾtinī ʾātika’ wa-bi-manzilatihā maʿa ʾayna kamā qāla subḥānahu wa-taʿālā ‘ʾaynamā takūnū yudrikkumu l-mawtu’ wa-bi-manzilatihā maʿa ʾayyin ʾiḏā qulta ‘ʾayyan mā tadʿū fa-lahu l-ʾasmāʾu l-ḥusnā’ […]). Kitāb I, 385.12–15

These two Qurʾānic quotations express that the consequent is independent of the antecedent. Thus: Death will find you, regardless of where you try to hide. Allah’s names are the most beautiful ones, regardless of what you may call him. Even the apocopate in the first quotation does not override this implication. In spite of choosing these interesting examples, al-Ḫalīl as reported by Sībawayhi seems to miss what is interesting about them. His comment is that the mā suffixed to the particle of consequence is ‘null and void’ (laġwan). That seems debatable. Nevertheless Sībawayhi himself never returns to either of these quotations to make any further comments on them; and al-Sīrāfī in his commentary accepts without argument the view that the suffix mā adds nothing to the quantification over places or names (Šarḥ III, 261.9 ff.). Ibn Sīnā by contrast has a number of things to say in his Qiyās about those conditionals that are true because their consequents are true, independent of the truth of the antecedent. Unfortunately his examples are logic-book examples, sentences that no real person would use in real life. Thus: If the human talks then the donkey brays (ʾin kāna l-ʾinsānu nāṭiqan fa-lḥimāru nāhiqun). Qiyās V/4, 265.12

It is not the case, if there are human beings, that there is a vacuum (laysa ʾin kāna l-ʾinsānu mawjūdan fa-l-ḫalāʾu mawjūdun10). Qiyās V/5, 279.7

Ibn Sīnā says that examples like the first of these sentences are true because of ‘correspondence’ or ‘conformity’ (muwāfaqa), presumably correspondence between the consequent and the real world. He describes sentences of the second kind, where the consequent is negated, as being true because of ‘denial

10

For the correct placing of the negation in the second example we are relying on Ibn Sīnā’s general theory of hypothetical logic, cf. Hodges (in preparation).

conditionality: syntax and meaning in al-sīrāfī and ibn sīnā

179

of correspondence’ ( yuslabu l-muwāfaqatu). Both the affirmative and the negative cases are covered by his phrase ‘by way of correspondence’ (ʿalā sabīli l-muwāfaqati). Although Ibn Sīnā’s examples of ‘by way of correspondence’ are painfully artificial, he has a realistic point of language to make about them. We accept that ‘If p then q’ is equivalent to ‘If not q then not p’. (A standard logicians’ example: ‘If it’s daytime then the sun is up’ is equivalent to ‘If the sun is not up then it’s not daytime’.) But this equivalence does not carry over to conditionals by way of correspondence. We can use a conditional ‘If p then q’ to express that q is true regardless of whether p is true; but we cannot use ‘If not q then not p’ to express the same thing. In the construction by way of correspondence, the redundant clause has to be the antecedent, not the consequent. The fact that donkeys bray entitles us to say ‘If humans talk then donkeys bray’, but it does not entitle us to say ‘If donkeys don’t bray then humans don’t talk’—at least not by way of correspondence (Ibn Sīnā, Qiyās V/4, 267.6–11). The Qurʾānic examples add a further point. In constructions by way of correspondence, the antecedent and the consequent do not have to be irrelevant to each other. For example the first Qurʾānic example might be paraphrased as ‘Regardless of where you try to hide, death will find you there’. So we can read a quantification over places as applying to both antecedent and consequent— and to this extent al-Sīrāfī’s analysis is confirmed. Nevertheless what makes the conditional true and worth stating is the fact that death will find you; your being in one place rather than another is irrelevant to this fact. Ibn Sīnā’s examples miss this point, probably because he bases them on what he found already in the logical literature.11 In short, for these conditionals by way of correspondence, al-Ḫalīl and alSīrāfī have the better examples but Ibn Sīnā has the better theory. If only they could have pooled their efforts …

Bibliographical References A

Primary Sources

Euclid, Elementa = Codex Leidensis 339, 1: Euclidis Elementa ex interpretatione AlHadschdschadschii cum commentariis Al-Narizii. Ed. by Rasmus Olsen Besthorn and Johan Ludvig Heiberg. Copenhagen: Hegel and Son, 1893. 11

In our talk in Genoa we presented several of Ibn Sīnā’s examples and asked what the audience thought they had in common. Michael Carter voiced what seemed to be a general impression, that “The two clauses haven’t got anything to do with each other”. It would have been interesting to discuss al-Ḫalīl’s examples along with Ibn Sīnā’s.

180

giolfo and hodges

Ibn Sīnā, ʿIbāra = ʾAbū ʿAlī al-Ḥusayn ibn ʿAbdallāh Ibn Sīnā, al-Šifāʾ: al-ʿIbāra, ed. by Mahmoud El-Khodeiri and Ibrahim Madkour. Cairo, 1970. Ibn Sīnā, Qiyās = ʾAbū ʿAlī al-Ḥusayn ibn ʿAbdallāh Ibn Sīnā, al-Šifāʾ: al-Qiyās. Ed. by Said Zayed and Ibrahim Madkour. Cairo, 1964. Ibn Sīnā, Mašriqiyyūn = ʾAbū ʿAlī al-Ḥusayn ibn ʿAbdallāh Ibn Sīnā, Manṭiq al-mašriqiyyīn. Cairo: al-Maktaba al-Salafiyya, 1910. Fārābī, Burhān = ʾAbū Naṣr Muḥammad ibn Muḥammad al-Fārābī, Kitāb al-burhān. Ed. by Majid Fakhry, Kitāb al-burhān wa-kitāb al-šarāʾiṭ al-yaqīn (Book of demonstration and book of conditions of certainty), 19–96. Beirut: Dar el-Mashreq, 1986. Sībawayhi, Kitāb = ʾAbū Bišr ʿAmr ibn ʿUṯmān Sībawayhi, al-Kitāb. Ed. by Hartwig Derenbourg, Le livre de Sībawaihi. 2 vols. Paris: Imprimerie nationale, 1881–1889. (Repr., Hildesheim: G. Olms, 1970.) Sīrāfī, Šarḥ = ʾAbū Sāʿīd al-Ḥasan ibn ʿAbdallāh al-Sīrāfī, Šarḥ Kitāb Sībawayhi. Ed. by ʾAḥmad Ḥasan Mahdalī. Beirut: Dār al-Kutub al-ʿIlmiyya, 2012. Zamaḫšarī, Mufaṣṣal = ʾAbū l-Qāsim Maḥmūd ibn ʿUmar al-Zamaḫšarī, Kitāb al-mufaṣṣal fī l-naḥw. Ed. by Jens Peter Broch. Christiania: Libraria P.T. Mallingii, 1859.

B

Secondary Sources

Blachère, Régis and Maurice Gaudefroy-Demombynes. 1952. Grammaire de l’ arabe classique (morphologie et syntaxe). 3rd rev. ed. Paris: G.P. Maisonneuve et Larose. Chatti, Saloua. 2016. “Existential import in Avicenna’s modal logic”. Arabic Sciences and Philosophy. 26.45–71. Dévényi, Kinga. 1988. “The treatment of conditional sentences by the Mediaeval Arabic grammarians: Stability and change in the history of Arabic grammar”. Budapest Studies in Arabic 1.11–42. Giolfo, Manuela E.B. 2015. “Real and irreal conditionals in Arabic grammar: From alʾAstarābāḏī to Sībawayhi”. The foundations of Arabic linguistics. II. Kitāb Sībawayhi: Interpretation and transmission, ed. by Amal Marogy and Kees Versteegh, 100–119. Leiden: E.J. Brill. Giolfo, Manuela E.B. 2017. Les systèmes hypothétiques de l’ arabe classique. Étude syntaxique et sémantique: une hypothèse modale. Rome: Aracne Editrice. Giolfo, Manuela E.B. and Wilfrid Hodges. 2016. “The system of the sciences of the Arabic language by Sakkākī: Logic as a complement of rhetoric”. Approaches to the history and dialectology of Arabic in honor of Pierre Larcher, ed. by Manuel Sartori, Manuela E.B. Giolfo, and Philippe Cassuto, 242–266. Leiden: E.J. Brill. Hodges, Wilfrid. 2012. “Affirmative and negative in Ibn Sīnā”. Insolubles and consequences: Essays in honour of Stephen Read, ed. by Catarina Dutilh Novaes and Ole Thomassen Hjortland, 119–134. London: College Publications. Hodges, Wilfrid. 2017. “Ibn Sīnā on reductio ad absurdum”. Review of Symbolic Logic 10:3.583–601.

conditionality: syntax and meaning in al-sīrāfī and ibn sīnā

181

Hodges, Wilfrid, in preparation. “Identifying Ibn Sīnā’s hypothetical logic. I. Sentence forms”. Draft online at wilfridhodges.co.uk/arabic59.pdf. Kratzer, Angelika. 1986. “Conditionals”. Papers from the Parasession on Pragmatics and Grammatical Theory, ed. by Anne M. Farley, Peter T. Farley and Karl-Erik McCullough, 115–135. Chicago: Chicago Linguistics Society. (Rev. version in Kratzer 2012, ch. 4.) Kratzer, Angelika. 2012. Modals and conditionals. New York: Oxford University Press. Larcher, Pierre. 2000. “Subordination vs. coordination ‘sémantiques’: L’exemple des systèmes hypothétiques de l’arabe classique”. Annales Islamologiques 34.193–207. Larcher, Pierre. 2006. “Le ‘segmentateur’ fa-(ʾinna) en arabe classique et moderne”. Kervan. Rivista Internazionale di Studi Afroasiatici 3.1–63. Levin, Aryeh. 1997. “The theory of al-taqdīr and its terminology”. Jerusalem Studies in Arabic and Islam 21.142–166. Lewis, David. 1975. “Adverbs of quantification”. Formal semantics of natural language, ed. by Edward L. Keenan, 3–15. Cambridge: Cambridge University Press. Maróth, Miklos. 1989. Ibn Sina und die Peripatetische ‘Aussagenlogik’. Leiden: E.J. Brill. Rescher, Nicholas. 1963. “Avicenna on the logic of ‘conditional’ propositions”. Notre Dame Journal of Formal Logic 4:1.48–58. Schubert, Lenhart K. and Francis Jeffry Pelletier. 1987. “Problems in the representation of the logical form of generics, plurals, and mass nouns”. New directions in semantics, ed. by Ernest LePore, 385–451. London: Academic Press. Versteegh, Kees. 1991. “Two conceptions of irreality in Arabic grammar: Ibn Hišām and Ibn al-Ḥājib on the particle law”. De la grammaire de l’ arabe aux grammaires des arabes, ed. by Pierre Larcher [= Bulletin d’Études Orientales, 43], 77–92. Damascus: Institut français de Damas. Von Fintel, Kai. 1994. Restrictions on quantifier domains. Ph.D. diss., University of Massachusetts, Amherst.

The Technical Terms taqdīr and taḫfīf in Persian Classical Sources Éva M. Jeremiás

1

Introduction1

In indigenous Persian sources, theoretical or didactic grammatical literary forms as ars were completely absent in the first centuries of the Muslim period. Traces of grammatical thinking, however, can be detected in non-grammatical sources, mainly in literary sciences such as prosody, rhyme, rhetoric or lexicography. I have started many of my earlier studies with this same introduction. In these sources, grammatical issues were treated in a non-technical way in most cases, but Muslim education including logic helped Iranian authors to formalize rules and to describe Persian characteristics by means of Arabic terms and analytical devices. One striking characteristic of Persian is that it readily generates compound forms (tarkīb), sometimes explained by means of taqdīr. The latter term is a well-known term in the Arabic grammatical literature, and is interpreted variously as a “reconstruction of an underlying level” (Versteegh 1994; 2009:IV, 446), an “abstract representation” (Bohas et al. 1990:62 f.), or a “suppletive insertion” (Baalbaki 2008:108)—to quote only some of the possible definitions.2 On the other hand, the term taḫfīf “lightening” (Baalbaki 2008:59) or “phonetic ease” (Carter 1981:484)3 is a very uncommon term in Persian sources.

1 In the transcription of Arabic, I follow the instructions of the editors. Persian, however, is transcribed with a few changes: the vowels are transcribed according to the Classical Persian system, which is very similar to the Classical Arabic; the phonological system of consonants, however, differs from Arabic in the following points: Arabic ṯ = s̱, ḏ= ẕ, ḍ = ż, w = v, but diphthongs are transcribed as aw and ay. The morphological elements such as the ‘objective’ postposition -rā, the binding vowel of the iżāfa-construction -i, the clitic forms of the verb to be and the possessive pronominal clitics are hyphenated. 2 The theory of this underlying structure and the wider extension of the term taqdīr is discussed by Kasher (2009) relying on Levin’s (1997) interpretation as “the speaker’s intention”. 3 Cf. ‘lightening’ in terms of phonological processes in Bohas et al. (1990: 91–93). See Jeremiás (forthcoming).

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_011

the technical terms taqdīr and taḫfīf

183

However, Īravānī, the author of the 19th century grammatical compilation (see Jeremiás 2012), uses this term quite exceptionally as a pragmatic device to classify compounds in Persian. These two terms taqdīr and taḫfīf are both of Arabic origin; yet, their meaning and use differ in Arabic and Persian sources. In the present study, I shall try to analyze them in Medieval Classical and post-Classical Persian sources as they are applied to an Indo-European language. The sources discussed here go far beyond the limits of the 8th–14th centuries set by the organizers of this conference, but the long-lasting, weighty influence of Arabic grammatical thinking, the highly derivative nature of the terminology and consequently, the heterogeneous character of the Persian material, do not allow me to fix timelimits on the sources used in the following analysis.

2

The Notions of taqdīr and taḫfīf in Modern Persian Lexicography

In modern lexicography, these terms do not occur in the sense that concern us. For instance, the largest modern Persian-Persian dictionary, the Farhangnāmayi fārsī by Ṣadrī Afšār, which contains a large number of Classical lexica, does not acknowledge the grammatical meaning of the word taqdīr ‘fate, appreciation’ (III, 843). On the other hand, the traditionally organized Luġatnāma, compiled by ʿAlī Akbar Dihḫudā in the 1930s, deals with its use in grammatical sources, following the enumeration of the various meanings of this word. Its use as a grammatical term is illustrated with a quotation from al-Tahānawī’s Kaššāf (cf. Luġatnāma V, 6873): Sometimes grammarians use taqdīr instead of ḥaẕf and it is said that the difference between taqdīr and ḥaẕf ‘deletion’ is that the trace of the reconstructed element (muqaddar) remains in the lafẓ, in contrast with the eliminated (maḥẕūf ) element; and taqdīr is [also] said to mean to drop something from the lafẓ and its trace remains in the intention (Naḥvīān gāh taqdīr-rā ba jāy-i ḥaẕf ba kār barand va gufta-and farq bayn-i ḥaẕf va taqdīr in ast ki as̱ar-i muqaddar dar lafẓ bāqī ast ba ḫalāf-i maḥẕūf va gufta-and taqdīr ʿibārat az ḥaẕf-i čīzī ast dar lafẓ va ibqāʾ-i ān dar niyyat). See Tahānawī, Kaššāf I, 497

The notion of taḫfīf, like that of taqdīr, occurs in modern lexicography only in its primary, everyday sense of kāhiš ‘reduction, drop-off’ (Afšār 1388/2009:III, 790), but it does not appear as a grammatical term. In Luġatnāma, however, early Medieval sources are quoted (e.g. Tāj al-maṣādir by Bayhaqī or Zawzanī)

184

jeremiás

which had recognized taḫfīf as a common word with the meaning taskīn, sabukī ‘lightness, lightening’. Dihḫudā gives the meaning of the term as “the abridgement of the word in order to make easy the pronunciation by taking tanvīn and tašdīd away” (iḫtiṣār-i kalima barāy-e suhūlat-i talaffuẓ va bar dāštan-i tanvīn va tašdīd az ān; cf. Tahānawī, Kaššāf I, 397; Luġatnāma V, 6527). To summarize, these two terms do not belong to the living language of present-day Persian, neither in common usage, nor in the technical vocabulary of grammatical science, but they were known in earlier Classical sources. In the following sections, I discuss the term taqdīr in two Classical and one post-Classical Persian sources, while the term taḫfīf will be treated only in the final post-Classical source. The first Classical source is a poetic manual of Šamsi Qays from the early 13th century, the best Medieval source where one can find traces of the foundational Arabic grammatical concepts, even though they are not spelled out systematically; the second is a commentary written on Persian Classical poetry such as Ḥāfiẓ and Saʿdī by the Bosnian Muḥammad Sūdī (d. 1000/1591); the third is a grammatical compilation from the mid-19th century, Qavāʿid-i fārsiyya ‘Persian rules’ written by ʿAbd al-Karīm Īravānī in [1846] and published as a lithograph in Tabrīz.

3

The Term taqdīr in Classical Heritage: Šams-i Qays (13th Century)

Šams-i Qays, the author of the first source, mentions this term in connection with phonological and morphological contexts. The first case refers to the obligatory insertion of a prosodic vowel called majhūla, muḫtalasa, or nīmfatḥa (Jeremiás 2016:62). Here the notion taqdīr is used in connection with the phonological interpretation of certain segments when the text is scanned. Šams-i Qays describes the phenomenon clearly: in my first example, the term taqdīr is not mentioned, while in the second example taqdīr is referred to explicitly: Each tā which is preceded by a sākin letter, such as mast, dast, bāḫt and tāḫt, if it occurs in verse, is to be considered a mutaḥarrik letter (har tā ki mā qabl-i ān sākin bāšad čūn mast va dast va bāḫt va tāḫt agar dar miyān-i šiʿr uftad har āyna ba ḥarf-i mutaḥarrik maḥsūb bāšad). Šams-i Qays, Muʿjam 99

Two partly overlapping Arabic phonological rules made it necessary to insert this prosodic vowel. One such basic rule was the abstract phonological interpretation of the long vowel as a sequence of “a vowel followed by a homorganic

the technical terms taqdīr and taḫfīf

185

glide which constitutes the element which prolongs the vowel”,4 where the vowel remains unwritten, and the glide is written by a consonantal letter (ḥarf ). This rule, even though it relied entirely on the Arabic phonological system, helped to distinguish between two differing two-vowel sequences in Persian, the long vowel and diphthong (Jeremiás 2010:157). The other rule prohibited three consecutive ḥarf s, which is very common in Persian, for instance, in words such as andūḫt and rīḫt, with the word-final endings ūḫt or īḫt (in abstract phonological interpretation /uwḫt / or /iyḫt/) (Jeremiás 2010:159f.). This prohibition helped to articulate a peculiar rule of Persian prosody, that is, the obligatory insertion of a prosodic vowel while scanning the verse. Actually, Šams-i Qays offers two solutions to the pronunciation of words with two or three consecutive consonants either in middle or wordfinal positions (-CC, -CC-, or -CCC, -CCC-), either to drop the last consonant(s) or to make the last one mutaḥarrik in scanning. But he cleverly adds that the status of this prosodic vowel is valid only ‘virtually’, that is, ‘in taqdīr’ (dar taqdīr): The sākin letter tā when it is preceded by two other sākin letters and if it occurs in a verse and may be scanned, the letter preceding tā is to be considered virtually mutaḥarrik (tā-i sākin ki piš az [ān] du sākin-i dīgar bāšad agar dar miyān-i bayt uftad va dar lafẓ tavān āvard albatta bā mā qabl-i ḫwīš dar taqdīr ḥarakat bāšad va bad u ḥarf-i mutaḥarrik maḥsūb).5 Šams-i Qays, Muʿjam 100

This example exhibits a strange, non-standard use of the term taqdīr. The Arabic phonological rule adapted to an alien system requires the insertion of this vowel pronounced with a weak articulation and this process is called taqdīr by Šams-i Qays. This vowel, however, does not belong to the material body of the word and it is not pronounced in common usage. Actually, such a word with a shwa-like prosodic vowel (ǝ) does appear in a specific spoken variety, but it is not reconstructed in the underlying level. Consequently, this Arabic phonological rule does not work automatically or predictably on words with three-letter endings in order to “maintain structural coherency” (Versteegh 2009:449): its appearance is restricted to a specific context, when the text is scanned in metrical form. That is, its occurrence is predicted in the dichotomy of scanned vs. non-scanned spoken forms. 4 Jeremiás (2016:62, n. 16), where I quote Bohas et al. (1990:98 f., n. 3) and Versteegh (1997:27). 5 Šams-i Qays gives the example bāḫt, in abstract phonological form /baʾḫt/, where the phoneme /ḫ/ is mutaḥarrik (ḫǝ).

186

jeremiás

The latter dichotomy points to the disparity between spoken and written forms, which appears many times in Šams-i Qays’ manual, expressed by such term-pairs as malfūẓ ‘spoken’ vs. maktūb ‘written’ (Šams-i Qays, Muʿam, 97–102) or taqṭīʿ ‘scanned’ vs. maktūb ‘written’ (Šams-i Qays, Muʿjam 97).6 In the next example, taqdīr occurs in a morphological context in the section on rhyme-science. Discussing rhyme-pairs such as hami na-gardānad/nagardānad/gardānad Šams-i Qays explains (Muʿǧam, 213) that in word-final sequences of a verse, the verbal form without a verbal prefix (gardānad) or the form of the continuous verbal prefix (hamī) and the linear ordering of this prefix and a following negative particle (hamī na), are not standard forms: the correct form of the infinitive gardīdan /gardāndan ‘to spin, turn’ in present continuous 3rd person singular in the negative is ‘virtually’ na-mī-gardānad. This example is not exactly the case that we would like to find when speaking about taqdīr. The verbal prefix (hamī) is a pre-Classical form of the continuous prefix mī- (which was preserved under the requirements of prosody), while the reconstructed or suggested ‘virtual’ form (na-mī) belongs to the formal variety. Therefore, the difference between the two forms is a matter of style or a matter of chronological variation, and does not point to a difference between surface and underlying structures.

4

Muḥammad Sūdī (d. 1000/1591)

The Bosnian Muḥammad Sūdī wrote his commentaries in Osmanli Turkish.7 Here I quote the examples from the Persian translation of his commentaries,8 where grammatical terms are always preserved in the original Arabic form. In Sūdī’s commentaries, there is no detailed theoretical explanation, classification or reference to his sources, but his terminology shows that he may have been well-versed in Arabic grammar.9 He employs the term taqdīr fairly often, but also uses synonymous words, such as yaʿnī or aṣl, and sometimes the latter two terms occur together with taqdīr, for example, taqdīran … yaʿnī (Sūdī, Ḥāfiẓ I, 51), taqdīr … yaʿnī (Sūdī, Ḥāfiẓ I, 75), maʿnā … taqdīr10 (Sūdī, Ḥāfiẓ I, 37), aṣl … taqdīr (Sūdī, Būstān, 4) or,

6 7 8 9 10

See more examples of this dichotomy in Jeremiás (2016:65). See Algar (2003). Full titles of Sūdī, Gulistān, Būstān, Ḥāfiẓ are given in the bibliographical references. It is thought that he may have translated Ibn al-Ḥājib’s Kāfiya into Turkish. Cf. Versteegh (1993:99).

the technical terms taqdīr and taḫfīf

187

yaʿnī … dar aṣl (Sūdī, Ḥāfiẓ I, 50), which appear to signify “the speaker’s intention” (Levin 1997) more clearly. I quote a selection of examples below collected from Sūdī’s commentaries on Ḥāfiẓ’ poems and on Saʿdī’s works Būstān and Gulistān. The term taqdīr and its synonyms are employed in various morpho-syntactic contexts, and they represent specimens on word level (i), phrase level (ii, iii), and on the level of larger utterances (iv, v). (i) On the lowest level, compounds of the Indo-European type are the most typical cases, where the terms yaʿnī, aṣl and taqdīr are used, in this order of frequency. The equivalent structures, that is, their supposed reconstructed deep structures, show that these compounds belong to different types: some are compounds consisting of an adjective and a noun and representing a classical bahuvrīhi type (a) or, the compound is unfolded as an attributive phrase (b) or, they exemplify compounds of a determinative type (c) consisting of a noun and a verbal noun. (a) The compound is interpreted as a bahuvrīhi type: nikū-nām: tarkīb-i vaṣfī ast ba maʿnā dāranda-yi nām-i nīk “ ‘of good reputation’: is a descriptive compound which means ‘having a good name’” (Sūdī, Būstān 114);11 nīk-nām-ī: nīk-nām yaʿnī kasī ki dārā-yi nām-i ḫūb ast; īn az aqsām-i tarkīb-i vaṣfī ast va yāʾ ḥarf-i maṣdar ast “ ‘the status of good reputation’: the person is called nīk-nām who has a good name and it is one of the descriptive compounds and the yāʾ is the letter [sign] of the maṣdar” (Sūdī, Ḥāfiẓ I, 46) tuhī-dast: ba maʿnā ṣafr al-yad “‘empty-handed’: its meaning in Arabic is ‘empty-handed’” (see the commentary: ay ḥarf-i nidā, munādā maḥzūf, taqdīr-aš ‘ay mard-i tuhī dast’ ast “ay is the vocative particle and its vocative noun is deleted; its virtual [underlying] meaning is ‘oh, the man empty-handed’”, Sūdī, Gulistān 64) sabukbārān: yaʿnī kasānī ki bār-išān sabuk ast “ ‘lightly loaded’: that is, the people whose burden is light” (see the commentary: jamʿ-i sabukbār čūn ṣifat ast barāy-i ẕūy-l-ʿuqūl pas bā alif nūn jamʿ šuda ba taqdīr-i kalām murād-i guftan ‘mardān-i sabukbārān’ mī-bāšad “the plural of sabukbār, 11

The two constituents create an adjective, for instance, nikū-nām ‘of good reputation’, whose meaning is further extended to designate ‘someone who has good reputation’.

188

jeremiás

which is an adjective denoting animate, and its plural is made with alif and nūn and its underlying structure [meaning] is ‘lightly loaded people’”, Sūdī, Ḥāfiẓ I, 13) nīk-ḫwāh-ān-rā: nīk-ḫwāh vaṣf-i tarkībī va murād-i nīk ḫwāhande ast ba maʿnā dūst va alif va nūn adāt-i jamʿ va rā adāt-i mafʿūl “ ‘good-willing people [acc.]’: the word nīk-ḫwāh is a compounded description intending to mean a person of good-will and signifies a ‘friend’ and the letters alif and nūn are the plural sign and rā is the sign of the object” (Sūdī, Gulistān 326) (b) The compound is unbounded as an attributive phrase: nīk-mard: dar aṣl mard-i nīk buvad “‘nice-man’ [compound]: its underlying structure is ‘a nice man’” (Sūdī, Būstān 33)12 (c) The compound is interpreted as a determinative type: These compounds consist of a noun and a verbal noun, traditionally called present and past verbal stems. Sūdī interprets the present stem (e.g. gū < guftan ‘to speak’) as the equivalent of the present participle (gūyanda ‘speaking’) and the past stem (ālūd ‘drowned’)13 is the equivalent of the past participle (ālūda). In the first case, however, he went further, saying that the underlying structure of these compounds appears to be a possessive iżāfa-construction, in which the present participle has to be conceived as noun and therefore it may figure as the first member of the iżāfa-construction.14 In the analysis of these compounds, Sūdī skips the transformation ADJ → N, interpreting the participle (and also the infinitive) as a noun.15 This type is one of the most commonly used compounds in Persian poetry, but it occurs only sparsely with an Arabic word as a first constituent. In the examples below, Sūdī specifies only the type of the compound (e.g. tarkīb-i vaṣfī ‘descriptive compound’), and in most cases, he simply gives an equivalent without unfolding the compound, by changing the stem form into a present or past participle, e. g.:

12 13 14 15

Sūdī makes a clear distinction between the two compounds on the basis of their differing underlying levels, see nīk-nām ‘of good reputation’ and nīk-mard ‘a nice man’. This form is used only in compounds. In the English translation, the difference between gū and gūyanda disappears. The status of these participles and infinitives caused Lumsden many headaches when he was dealing with the problems of the parts of speech in indigenous systems (Jeremiás 2012a:135). See Jeremiás (forthcoming).

the technical terms taqdīr and taḫfīf

189

fārsī-gū: yaʿnī kasī ki fārsī mī-gūyad dar aṣl gūyanda-yi pārsī būda “‘Persian-speaking’: that is, someone who speaks Persian, basically, ‘the speaker of Persian’” (Sūdī, Ḥāfiẓ I, 50) dast-gīr: gīr dar aṣl gīranda buvad, dar taqdīr gīranda-yi dast, az qabīli iżāfa-yi ism-i fāʿil ba mafʿūl-aš “‘hand-taking’: gīr is basically gīranda, reconstructed at the underlying level as ‘taking of the hand’, a sort of iżāfa which is composed of ism-i fāʿil (present participle) and its ism-i mafʿūl (object)” (Sūdī, Būstān 3f.) ḫāk-rūb: dar aṣl rūbanda-yi xāk “‘dust-sweeping’: is basically ‘sweeper of dust’” (Sūdī, Ḥāfiẓ I, 72) pūziš-paẕīr: ba maʿnā ʿuẕr … dar aṣl paẕīranda-yi pūziš “ ‘excuseaccepting’: means ʿuẕr [in Arabic] and its underlying meaning is ‘acceptance of excuse’” (Sūdī, Būstān 4) šakar-forūš: yaʿnī šakar-forūšanda “‘sugar-selling’: means ‘selling sugar’ ” (Sūdī, Ḥāfiẓ I, 34) ḫwāb-ālūd:16 yaʿnī ḫwāb-ālūda “‘drowned in sleep’: means ‘drowned in sleep’” (Sūdī, Ḥāfiẓ I, 99) mār-gazīda: taqdīr-aš mardum-i mār-gazīda ast “ ‘snake-bitten’: reconstructed at the underlying level as ‘the person who is bitten’” (Sūdī, Gulistān 235) (ii) Nouns or nominal phrases: the reconstructed underlying level inserts a deleted part of speech or a case marker in order to make the utterance (surface structure) syntactically or semantically clearer (Versteegh 2009:IV, 447). The deleted part of the sentence may be as follows: (a) the deleted element is the vocative noun and in the surface structure it is reconstructed after the vocative particle: ay: ay ḥarf-i nidā va munādā-yaš maḥẕūf ; taqdīran ay jānān ast “ ‘ay’: ay is the vocative particle and its vocative noun is deleted; the virtually reconstructed vocative noun is ‘oh, souls’” (Sūdī, Ḥāfiẓ I, 127) 16

In this locus Sūdī says explicitly that ālūd is an ism (Sūdī, Ḥāfiẓ I, 99).

190

jeremiás

Ḥāfiẓ: munādā va ḥarf-i nidā-yaš ḥaẕf šuda; dar aṣl murād-i guftan ay Ḥāfiẓ “Ḥāfiẓ: is the vocative noun and the vocative particle is deleted; originally, the intention of utterance is ‘oh, Ḥāfiẓ!’” (Sūdī, Ḥāfiẓ I, 16) (b) the deleted element is the object suffix -rā: darvīš namī-pursī: ba taqdīr-i kalām darvīš-rā ast ki ba żarūrat-i vazn adāt-i mafʿūlī ḥaẕf šuda “‘you do not ask the dervish’: with the virtually reconstructed object of the sentence ‘the dervish’ [acc.], where the object suffix is deleted through metrical necessity” (Sūdī, Ḥāfiẓ I, 144) (c) the deleted element is the marker of indeterminacy -ī: rūz: dar aṣl rūzī būda yā-yi vaḥdat ba żarūrat-i vazn ḥaẕf šuda “ ‘day’: the virtually reconstructed phrase is ‘once in a day’, and the marker of indeterminacy is deleted through metrical necessity” (Sūdī, Ḥāfiẓ I, 109) (d) the deleted element is the determinative possessive of the noun expressed as pronominal clitic -am: zǝ bar: mīm-i mutakallim muqaddar ast yaʿnī taqdīran zǝ bar-am būda “‘from the body’: the letter m signifying the first person singular is substituted virtually, and it means at the underlying level ‘from my body’ ” (Sūdī, Ḥāfiẓ I, 65) (e) the deleted element is the first member (noun) of the iżāfa-construction: ba ḥikmat: taqdīran ba ʿilm-i ḥikmat mużāf ḥaẕf šuda “ ‘with wisdom’: virtually reconstructed as ‘with the science of wisdom’ ” (Sūdī, Ḥāfiẓ I, 29). (iii) Verb or verbal phrases: the reconstructed underlying level inserts a deleted part of the verbal phrase, or the equivalents of verbal morphemes of various functions: (a) the deleted element is the nominal part of the verbal phrase: kunand: dar mā qabl-a yak maṣdar maḥẕūf ast ba taqdīr-i kalām yaʿnī ṣarf kunand “’kunand: in front of [the form] kunand a noun is deleted, the virtual meaning of the phrase is ‘to consume, to lose’ ” (Sūdī, Ḥāfiẓ I, 75)

the technical terms taqdīr and taḫfīf

191

(b) the deleted element is the second member of the compounded verbal form called pluperfect: sāḫta: qismatī az fiʿl-i māżī-yi mufrad-i ġāʾib va taqdīr-aš ‘sāḫta būdand’ ast “sāḫta is a part of the past third person singular, and its underlying form is sāḫta būdand ‘had prepared’”. This reconstructed verbal form is made by analogy of the first part of the sentence ba ḥukm-i ānki malāżī manīʿ az qulla-yi kūhī ba dast āvarda būdand va maljaʾ va maʾvā-yi ḫwud sāḫta ‘Because they had already occupied a high mountain fortress and had made asylum for themselves’ (Sūdī, Gulistān 145) (c) the archaic form of the past continuous is expressed by a postponed -ī after the inflected verbal form and it is substituted by the Classical (formal) verbal prefix mī: nadīdam-ī: ḥikāyat-i ḥāl-i māżī yaʿnī namī-dīdam “ ‘I have never seen’: is a past continuous form, that is, namī-dīdam” (Sūdī, Ḥāfiẓ I, 151) (d) the colloquial verbal form is substituted by its formal equivalent: ārad: dar aṣl āvarad … vāv-rā ḥaẕf kunand “[the verbal form] ārad: is basically āvarad … and the letter vāv is deleted” (Sūdī, Ḥāfiẓ I, 24) (iv) Utterance-level: the case of word order in Persian (a) The placement of clitic pronouns šīn-i żamīr: az jihat-i maʿnā marbūṭ ba kalima-i yād mī-bāšad ba taqdīr yādaš ḫuš bād “the [clitic] pronoun -aš: is transferred to the word yād and its underlying form is ‘his memory be happy’” (Sūdī, Ḥāfiẓ I, 20) the end -t (tāʾ-i āḫir) in the word ġayrat-at: tāʾ-i ḫaṭāb dar maʿnā muqayyad ba kalima ‘bi-sūzad’ ast ba taqdīr bi-sūzad-at “the clitic t in the second person: is transferred to the word bi-sūzad, which is interpreted virtually ‘he should burn you’” (Sūdī, Ḥāfiẓ I, 49) (b) the placement of the object suffix: rā: dar kalima-yi ‘ḫāṣ-rā’ rā marbūṭ ba ‘kas’ mī-bāšad ba taqdīr ‘kas-rā namī-bīnam’ “rā: in the word ḫāṣ-rā [the object suffix] -rā is transferred

192

jeremiás

to the word kas ‘somebody, nobody’, with the underlying reconstructed word order kas-rā namī-bīnam ‘I do not see anybody’ ” (Sūdī, Ḥāfiẓ I, 68) (v) Very occasionally, there is a paraphrase of the Persian text in Arabic in order to make the sentence more comprehensible, e.g.: matā mā talqā man tahwā daʿ al-dunyā wa-ʾahmilhā: ba taqdīr-i kalām: bāyad guft tā maʿnā durust dar yābad matā ʾaradta ʾan talqā “ ‘Whenever you meet the person you love, leave the world and forget it’: the sentence should be interpreted with the paraphrase (taqdīr) ‘Whenever you wish to meet’ in order to make the meaning properly” (Sūdī, Ḥāfiẓ I, 16)

5

Īravānī’s Persian Rules (1846): the Terms taqdīr and taḫfīf

Īravānī’s grammar Qavāʿid-i fārsiyya from the mid-19th century seems to be the last representative of the ancient tradition, but the treatment of the true compound of the Indo-European type might be regarded as one of his novelties. In his grammar, the last short chapter is dedicated to this subject,17 where the terms taqdīr and taḫfīf occur abundantly. He opens the discussion with the general statement that the ‘descriptive compound’ (vaṣf-i tarkībī) is a “widely used type” (kas̱īr al-istiʿmāl ast) and therefore “it has a special importance” (ihtimām ba šaʾn-i ān bīštar ast, f. 56r). The summary of his analysis of the compounded entities is as follows:18 Persians easily combine two words (lafẓ), the meaning of which is considered to be derivate (muštaqq) and the derivative entities can be ism-i fāʿil, ism-i mafʿūl, ṣifat-i mušabbaha19 and ism-i mansūb, that is, present or past participles and derived (or secondary) adjectives (used occasionally as nouns),20 and also derived nouns. Then, he gives the following classification, dividing compounds into two main groups in terms of taḫfīf,21 taqdīm, ḥaẕf, taqdīr and aṣl. The abundant employment of the term taḫfīf is a novelty here, although it 17 18 19 20

21

Ch. XIV: dar bayān-i vaṣf-i tarkībī ‘On the description of the compound’ (ff. 56r–57r). See Jeremiás (2012: 112). See more examples in Jeremiás (forthcoming). Cf. Afšār (1388/2009), II:1827; Carter (1981:483): ṣifa mušabbaha bi-l-fāʿil ‘quasi-participial adjective’. If I understand this passage properly, the last two categories might be conceived as subcategories of the ism interpreted in terms of tašbīh ‘formal similarity’ (cf. Bohas et al. 1990:51–53). See above, note 3.

the technical terms taqdīr and taḫfīf

193

occurs sporadically in the previous two sources as well. The rather ambiguous classification of compounds listed by Īravānī appears to be based mainly on operations which create ‘lightened’ forms.22 For instance, the specimens of the first group (i) are produced purely and simply for the sake of ‘lightening’ (taḫfīf ), and they are created by the operation of ‘fronting’ (taqdīm) of the mużāf ilayh, producing a sort of ‘lightened’, more easily pronounced form (nowʿ-i taḫfīfī). The items in the other group (ii), with three varying subgroups, are also produced for the sake of taḫfīf, but in addition they create new meaning, that is, a different part of speech. (i) The compound is unbounded as an attributive phrase: gulāb ki dar aṣl āb-i gul ba taqdīm-i mużāf ilayh nawʿ-i taḫfīfī ḥāṣil gardīd “The word gulāb is basically āb-i gul ‘the water of the rose’, and by means of fronting the mużāf ilayh a sort of lightened form was created” (Īravānī, Qavāʿid f. 56r) (ii) The compounds are unbounded as copulative and determinative phrases: (a) ḫūnāb ki dar aṣl ḫūn va āb būda ast, vāv va alif bi-yuftād va ḫūnāb šud va maʿnā-i ism-i mansūb baḫšīd “The word ḫūnāb ‘blood-water’, which is basically ḫūn va āb ‘blood and water’, the letters vāv and alif are dropped and it becomes ḫūnāb and its meaning is transformed into ism-i mansūb” (Īravānī, Qavāʿid f. 56v)23 (b) suḫan-dān ki dar aṣl dān-i suḫan buvad va čūn mafʿūl-i amr-rā muqaddam kardand maʿnā-i ism-i fāʿil baḫšīd va dar īn lafẓ du iʿtibār namūda-and, yakī ānki mazkūr šud [taqdīm], duyyum ānki taqdīr-i dānanda-yi suḫan kunand va dar īn ṣūrat ḥazf va taqdīm har du hast va ūlaviyyat-i vajh-i avval ba jahat-i taḫfīf va suhūlat va ʿadam-i taqdīr ast va ūlaviyyat-i vajh-i duyyum barā-yi īn ki aqrab ba qiyās va mufīdtar ast “The underlying structure of the word suḫandān ‘eloquence-knowing’ is dān-i suḫan ‘knowledge of eloquence’; and because

22

23

I interpret the passage fāyida-yi dīgar az ḥays̱iyyat-i maʿnā maqṣūd mī-bāšad (f. 56r) as referring to the part of speech, that is, the ‘meaning’ of the word. Cf. one of the various interpretations of maʿānī: “the meaning or function of a word, which is what the grammarians are concerned with” (Versteegh 1997:59). The compounds gulāb and ḫūnāb are also discussed by Šams-i Qays (Muʿjam 214 f.) under the letter b, but he regards their differing behavior in a particular situation, in rhyme.

194

jeremiás

the object of the ‘imperative’ is fronted, the meaning ism-i fāʿil is created.24 The interpretation of suḫan-dān is twofold: one is the aforementioned and the underlying structure (taqdīr) of the other is dānanda-yi suḫan ‘knowledge of eloquence’: in this case both ḥaẕf and taqdīm have operated. The lightening (taḫfīf ) and fluency (suhūlat) due to the lack of taqdīr (ʿadam-i taqdīr) give priority to the former case (dān) over the latter25 (dānanda), while the latter interpretation is to be preferred owing to its [formal] analogical proximity [to present participles], and it is more practical” (Īravānī, Qavāʿid f. 56v) (c) šaqāvat-pīša ki dar aṣl pīša kunanda-yi šaqāvat buvad baʿd az ḥaẕf va taqdīmi šaqāvat-pīša šuda maʿnā-i ṣifat-i mušabbaha baḫšīd “The word šaqāvat-pīša ‘of cruel behavior’, which basically means ‘performing tyranny’ and by means of the operations ‘deletion’ and ‘fronting’ yields the meaning ‘assimilated adjective’ (ṣifat-i mušabbaha)” (Īravānī, Qavāʿid ff. 56v–57r) (d) saʿādat-ḫāna dar aṣl ḫāna-yi saʿādat būda ast va čun taqdīm-i mużāf ilayh-rā bar mużāf muqaddam dārand va kasra-yi iżāfa bi-yuftad ān ṣūrat nīz vaṣf-i tarkībī mī-bāšad va manẓūr-i taḫfīf tanhā buvad hamču gul-āb “The word saʿādatḫāna ‘happiness-mansion’, which basically is ḫāna-i saʿādat ‘mansion of happiness’ and by means of fronting mużāf ilayh and omitting kasra-yi iżāfa it becomes a vaṣf-i tarkībī. The only intention [of these operations] is the ‘lightening’, rather like gul-āb” (Īravānī, Qavāʿid f. 57r). Īravānī cleverly adds, however, that the latter type, which was very common in Persian, for instance żalālat-andīša ‘of deviation thinking’, siyāh-čašm ‘blackeyed’ and tar-dāman ‘of wet-rock, immortal, polluted’, which creates the meaning ism-i mansūb by means of taqdīm, is produced only for the sake of ‘lightening’ (taḫfīf ), as in the case of gul-āb mentioned above. The terms taḫfīf, taḫfīfan and muḫaffaf occur in other chapters, too, but their employment refers to the prosodically shortened lexical items, which are common in Sūdī’s commentaries, for instance, dīgar > digar, afġān > faġān, čūn > ču, agar > ar, etc.

24 25

Cf. above, notes 13 and 14. This passage indicates that the terms aṣl and taqdīr might have different meanings.

the technical terms taqdīr and taḫfīf

6

195

Summary

The eclectic ways of using taqdīr or taḫfīf do not point to a definite interpretation of the original theory, or to any interpretation which might have dealt with these concepts systematically. Nevertheless, from the beginning, the Arabic linguistic tradition is clearly felt in the background, and Īravānī’s case testifies that in the mid-19th century a renewed interest in it is palpable. The Classical notion of taqdīr is used to understand and ‘correct’ poetic style, where grammaticality has failed to be followed properly. The main target of grammatical remarks, however, is the compound, this typically Persian phenomenon. Even though Sūdī’s compounds represent more numerous types than those listed by Īravānī, the difference between them is not a matter of quantity, but that of quality. Tentatively, I would say that with the concept of taqdīr, Sūdī and Īravānī indicate how compounds are to be understood by unfolding their underlying structures, while with the term taḫfīf, Īravānī tries to grasp why they became transformed into one-word-compounds, that is, to make linguistic expression more concise. This latter term has been well-known since Sībawayhi’s time as referring to “phonetically tolerable or more syntactically economical” forms (as Baalbaki argues), an uncommon term in Persian sources, but its adaptation to the process of creating compounds seems to be unique. As I argued in an earlier study (Jeremiás, forthcoming), Īravānī’s focus is directed more towards the operation itself which generates compounds. In the present study, greater emphasis is given to the ‘lightening’ aspect of his approach, by which the author was able to understand the concise and compound-like character of Persian poetic language. Yet, since this approach has not been adopted in either earlier or later elaborations of Persian, it must be regarded as Īravānī’s own unique contribution, one which reveals his proficiency in Classical doctrine.

Bibliographical References A

Primary Sources

Afšār, Farhangnāma = Ṣadrī Afšār (ed.), Farhangnāma-yi Fārsī. 3 vols. Tehran: Farhang-i Muʿāṣir, 1388/2009. Dihḫudā, Luġatnāma = ʿAlī Akbar Dihḫudā, Luġatnāma. Ed. by M. Muʿīn and J. Šahīdī. New ed. 16 vols. Tehran: Tehran University Publication, 1377/1998. Īravānī, Qavāʿid = ʿAbd al-Karīm Īravānī, Qavāʿid-i fārsiyya. Lithograph [Tabriz], [1846]. Sūdī, Būstān = Muḥammad Sūdī, Šarḥ-i Sūdī bar Būstān-i Saʿdī. Ed. and transl. by Akbar Bihrūz. 2 vols. Tabriz, 1352/1973. Sūdī, Gulistān = Muḥammad Sūdī, Šarḥ-i Sūdī bar Gulistān-i Saʿdī. Transl. by Ḥaydar

196

jeremiás

Khuš Ṭīnat, Zayn al-ʿĀbidīn Čāvušī, ʿAlī Akbar Kāẓimī. Tabriz and Tehran: Bihtarīn, 1349/1970. Sūdī, Ḥāfiẓ = Muḥammad Sūdī, Šarḥ-i Sūdī bar Ḥāfiẓ. Transl. by ʿIsmat Sittārzāde. 4 vols. 5th ed. Tehran, 1378/1999. Šams-i Qays, Muʿjam = Šams-i Qays, al-Muʿjam fī maʿāyīr ʾašʿār al-ʿAjam. Ed. by Muḥammad Qazvīnī and Mudarris Rażavī. [Tehran]: Intišārāt-i Dānišgāh-i Tihrān, [1338/ 1959]. Tahānawī, Kaššāf = Muḥammad ʾAʿlā Tahānawī, Kaššāf iṣṭilāḥāt al-funūn wa-l-ʿulūm (1745). Ed. by Rafīq al-ʿAjam. 2 vols. 1st ed. Beirut: Maktabat Lubnān Nāširūn. 1996.

B

Secondary Sources

Algar, Hamid. 2003. “Bosnia and Herzegovina”. Encyclopaedia Iranica, http://www.iranicaonline.org/articles. Baalbaki, Ramzi. 2008. The legacy of the Kitāb: Sībawayhi’s analytical methods within the context of the Arabic grammatical theory. Leiden: E.J. Brill. Bohas, Georges, Jean-Patrick Guillaume, and Djamel Eddine Kouloughli. 1990. The Arabic linguistic tradition. London and New York: Routledge. Carter, Michael G. 1981. Arab linguistics: An introductory classical text with translation and notes. Amsterdam: J. Benjamins. Jeremiás, Éva M. 2010. “The grammatical tradition in Persian: Shams-i Fakhrī’s rhyme science in the fourteenth Century”. IRAN, The British Institute of Persian Studies 48.153–162. Jeremiás, Éva M. 2012. “ʿAbd al-Karīm Īrawānī’s ‘Persian Rules’”. At the gate of Modernism, ed. by Éva M. Jeremiás, 85–116. Piliscsaba: The Avicenna Institute of Middle Eastern Studies. Jeremiás, Éva M. 2012a. “Matthew Lumsden’s Persian Grammar (Calcutta, 1810), Part I”. IRAN, The British Institute of Persian Studies 50.129–140. Jeremiás, Éva M. 2016. “The history of grammatical ideas in Persian: kitābatan-lafẓan in Classical Persian sources”. Further topics in Iranian linguistics, ed. by Jila Gomeshi, Carina Jahani and Agnès Lenepveu-Hotz. Studia Iranica Cahier (Paris) 58.55–70. Jeremiás, Éva M., forthcoming. “The technical term tarkīb ‘compound’ in the indigenous Persian ‘scientific’ literature”. Kasher, Almog. 2009. “Two types of taqdīr? A study in Ibn Hišām’s concept of ‘speaker’s intention’”. Arabica 56.360–380. Levin, Aryeh. 1997. “The theory of al-taqdīr and its terminology”. Jerusalem Studies in Arabic and Islam 21.142–166. Versteegh, Kees. 1993. Arabic grammar and Qurʾānic exegesis in early Islam. Leiden: E.J. Brill. Versteegh, Kees. 1994. “The notion of ‘underlying levels’ in the Arabic grammatical tradition”. Historiographia Linguistica 21.271–296.

the technical terms taqdīr and taḫfīf

197

Versteegh, Kees. 1997. The Arabic linguistic tradition. London and New York: Routledge. Versteegh, Kees. 2009. “Taqdīr”. Encyclopedia of Arabic language and linguistics, ed. by Mushira Eid, Alaa Elgibali, Kees Versteegh, Manfred Woidich, and Andrzej Zaborski, IV, 446–449. Leiden: E.J. Brill.

How to Parse Effective Objects according to Arab Grammarians? A Dissenting Opinion on al-mafʿūl al-muṭlaq Almog Kasher

1

The Meaning of the Term al-mafʿūl al-muṭlaq

In 1991 two articles were published in which the technical term al-mafʿūl almuṭlaq was examined. Levin’s article (1991), entitled “What is meant by almafʿūl al-muṭlaq?”, is dedicated in its entirety to this issue, whereas Larcher (1991:153 [= 2014:292f.]) discusses it succinctly in a single paragraph. Their findings are virtually the same:1 The element mafʿūl in the phrase al-mafʿūl almuṭlaq is a personal passive participle. It means ‘that which is done’, designating the action done by the agent. For instance, in the sentence qumtu qiyāman lit. ‘I rose a rising’, qiyām denotes the action done by the speaker. The word muṭlaq here means ‘unqualified’, and should be understood in contradistinction to the other terms in the category of mafʿūlāt/mafāʿīl, e.g. almafʿūl fīhi, lit. ‘that in which the action is done’ (technically: the locative/temporal accusative), terms in which the word mafʿūl is an impersonal passive participle, qualified (muqayyad) by a prepositional phrase (or a ẓarf phrase). By contrast, in the case of al-mafʿūl al-muṭlaq, the passive participle mafʿūl is not qualified by any phrase, as it designates, by itself, the intended meaning, viz. that which is done. Some grammarians discuss the difference between al-mafʿūl al-muṭlaq and al-mafʿūl bihi, direct object, lit. ‘that to whom, or to which, the action is done’. They state that whereas al-mafʿūl al-muṭlaq designates what is done by the agent, viz. the action, al-mafʿūl bihi designates to whom, or to what, the action is done. For this reason, some of them regard al-mafʿūl al-muṭlaq as the ‘real’ mafʿūl (al-mafʿūl al-ḥaqīqī, al-mafʿūl fī l-ḥaqīqa). For instance, in the sentence ḍaraba zaydun ʿamran ḍarban lit. ‘Zayd hit ʿAmr a hitting’, ḍarb designates the action done and ʿamr refers to the person to whom it is done. Zayd (the referent of the fāʿil) produced the action; he did not produce ʿAmr. Furthermore, gram-

1 The reader is referred to Levin (1991) for references to primary sources.

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_012

how to parse effective objects according to arab grammarians? 199

marians sometimes state that qāma zaydun, for example, conveys the same meaning as faʿala zaydun qiyāman lit. ‘Zayd did a standing’.2 Levin’s (1991:920f.) definition of the term al-mafʿūl al-muṭlaq runs as follows: the accusative which is a maṣdar expressing the act performed by the fāʿil, and which is denoted in grammatical terminology by the word al-mafʿūl only, without adding to it a restrictive phrase, such as a preposition + genitive. This was indeed the most common interpretation of the term al-mafʿūl almuṭlaq by Medieval grammarians, and it will also be the axis around which our discussion will revolve. In the Excursus at the end of this article I will discuss alternative explanations for this term by Arab grammarians.

2

Effective Objects in Western Linguistics

The following discussion can benefit from a distinction made in Western linguistics between ‘effective object’ (also: ‘effected object’, ‘object of result’) and ‘ordinary’ objects,3 illustrated by Lyons (1968:439) with the sentences in (1) and (2). (1) ‘He is reading a book’ (an ‘ordinary’ object) (2) ‘He is writing a book’ (an ‘effective object’) He explains that “in (1) the book referred to exists prior to, and independently of, its being read; but the book referred to in (2) is not yet in existence—it is brought into existence by the completion of the activity described by the sentence”.4

2 See, in this regard, al-Suhaylī’s (d. 581/1185) unique view, discussed in Baalbaki (1999:31 f.). 3 ‘Effective object’ is often contrasted with ‘affective object’ (also: ‘affected object’); see in what follows, and also Baalbaki (1990:34, 166), who translates ‘affected object’ (in e.g. ‘He cut the apple’) as mafʿūl bihi mutaʾaṯṯir. I shall refrain from using this term here, as it implies the preclusion of objects of verbs such as ḥasiba ‘he thought’ (on which see Kasher 2012a, and the references therein), which will be regarded here as subsumed under ‘ordinary’ objects. 4 Baalbaki (1990:166) translates ‘effected object’ (in e.g. ‘They erected a monument’) as mafʿūl bihi muḥdaṯ, and ‘object of result’ as mafʿūl al-natīja (ibid., 343).

200

kasher

In the Arabic grammatical tradition such a distinction was almost never made, so the question of how one is to parse effective objects was seldom raised; these are generally parsed, as one might expect, as al-mafʿūl bihi. For instance, commenting on the Qurʾānic verse huwa llaḏī ḫalaqa lakum mā fī lʾarḍi jamīʿan … (Q. 2/29) ‘It was He who created all that is on the earth for you …’,5 al-Zajjāj (d. 311/923) says: “The position of mā is direct object” (mawḍiʿu mā mafʿūlun bihi),6 that is, mā, being a relative pronoun (ism mawṣūl), hence filling nominal positions, fills in this verse the position of the direct object of ḫalaqa.

3

Effective Objects Parsed as al-mafʿūl al-muṭlaq

Yet, there are exceptions; some grammarians parsed effective objects as almafʿūl al-muṭlaq. While this analysis is interesting in itself and deserves our attention on its own merits, the passages in which it is discussed have further importance, since they raise issues pertaining to categorization in the Arabic grammatical tradition. Here, we will discuss the relevant passages in two well-known treatises, al-Jurjānī’s (d. 471/1078) ʾAsrār al-balāġa and Ibn Hišām’s (d. 761/1360) Muġnī l-labīb ʿan kutub al-ʾaʿārīb.7 These passages are exceptional vis-à-vis the view of the majority of grammarians towards al-mafʿūl al-muṭlaq, which both grammarians (to the best of my knowledge) advance in their other treatises. For instance, in his Muqtaṣid, al-Jurjānī identifies al-mafʿūl al-muṭlaq with the maṣdar, in line with the abovementioned explanation: muṭlaq means that it (i.e., the term mafʿūl) is not qualified (lā yuqayyadu) by any preposition, e.g. bihi in al-mafʿūl bihi; here one uses the term mafʿūl unqualified (ʿalā l-ʾiṭlāq)—and this, he says, is the maṣdar.8 Ibn Hišām’s adherence to the common view can be nicely (albeit ex silentio) illustrated with his criticism of ʾAbū Ḥayyān’s (d. 745/1344) use of maṣdar, instead of al-mafʿūl al-muṭlaq, as the term 5 Abdel Haleem (2005:6). 6 Zajjāj, Maʿānī I, 107. 7 Several other scholars dealt with this issue, but their discussions were theologically oriented; these will be dealt with elsewhere. Our paragraphs from al-Jurjānī’s ʾAsrār al-balāġa and Ibn Hišām’s Muġnī l-labīb were briefly discussed by Larkin (1995:95) and Gully (1995:249f.), respectively; yet, they misinterpreted the use of the term al-mafʿūl al-muṭlaq by these scholars. It should be noted that al-Jurjānī’s and Ibn Hišām’s discussions differ with respect to their orientations. Whereas Ibn Hišām’s goal is to correct what he regards as an erroneous parsing, in conformity with the large-scale aim of his Muġnī, al-Jurjānī’s purpose here is to differentiate between two types of mafʿūl, as part of his discussion of majāz (see Abu Deeb 1979:231–233; Larkin 1995:93ff.); the terminological question is certainly a marginal issue for him. 8 Jurjānī, Muqtaṣid I, 580ff.

how to parse effective objects according to arab grammarians? 201

for this bāb (see below). What is of interest is that in none of Ibn Hišām’s three reasons for the inadequacy of the former is any mention made of effective objects.9 Al-Jurjānī distinguishes between two types of mutaʿaddin verbs: those taking a mafʿūl bihi, e.g. ḍarabtu zaydan ‘I hit Zayd’ and those taking a mafʿūl ʿalā l-ʾiṭlāq, e.g. faʿala zaydun-i l-qiyāma ‘Zayd did the rising’ and ḫalaqa llāhu lʾanāsiyya ‘God created humankind’. In the latter cases, he asserts, the accusative is mafʿūlun muṭlaqun lā taqyīda fīhi, as it is absurd to claim that ḫalaqa l-ʿālama ‘He created the world’ means faʿala l-ḫalqa bihi ‘He did the creation to it’, and faʿala l-qiyāma ‘he did the rising’ means faʿala šayʾan bi-l-qiyāmi ‘he did something to the rising’, just as ḍarabtu zaydan means faʿaltu l-ḍarba bi-zaydin (see above).10 In a similar vein, Ibn Hišām, in a passage subsumed under the category of common wrong parsings,11 criticizes the parsing of the Qurʾānic phrase Allāhu llaḏī ḫalaqa l-samāwāti (e.g. Q. 7/54)12—‘… God, who created the heavens …’13 as if al-samāwāt were mafʿūl bihi. He maintains that it should be parsed as al-mafʿūl al-muṭlaq. The reason is that al-mafʿūl al-muṭlaq is the constituent to which the term mafʿūl is applied without qualification (bi-lā qayd), as in ḍarabtu ḍarban. In contrast, al-mafʿūl bihi is a constituent to which the term mafʿūl is applied only when qualified by bihi, as in ḍarabtu zaydan. Now, alsamāwāt in the verse in question is indeed mafʿūl, in the sense of ‘that which is done’, just like ḍarb in ḍarabtu ḍarban; it is therefore not a mafʿūl bihi, in the sense of ‘that to whom, or to which, the action is done’.14 Furthermore, Ibn Hišām asserts that whereas the referent of a constituent parsed as al-mafʿūl bihi had already existed prior to the action denoted by its verb, the verb of al-mafʿūl al-muṭlaq denotes the action of bringing into being the referent in question.15 This argument also appears in a statement ascribed by al-Suyūṭī (d. 911/1505) to al-Jurjānī. Al-Suyūṭī says that al-Jurjānī disagreed with the parsing of ḫalaqa llāhu l-samāwāti wa-l-ʾarḍa (Q. 29/44) ‘God has created the heavens and earth’16 as al-mafʿūl bihi, for this term applies to what already exists, and the fāʿil brings about something else in it. For instance, in ḍarabtu zaydan ‘I hit Zayd’, Zayd already exists, and the fāʿil brings about the 9 10 11 12 13 14 15 16

Ibn Hišām, Šarḥ II, 158f. Jurjānī, ʾAsrār 340f. Ibn Hišām, Muġnī VI, 541ff. Or, according to a different edition, ḫalaqa llāhu l-samāwāti (Q. 29/44) ‘God has created the heavens’ (Abdel Haleem 2005:254). See Ibn Hišām, Muġnī (ed. Damascus) 736. Abdel Haleem (2005:98). Ibn Hišām, Muġnī VI, 578f. Ibn Hišām, Muġnī VI, 579. Abdel Haleem (2005:254).

202

kasher

hitting in him. Al-mafʿūl al-muṭlaq, on the other hand, does not exist yet, but is brought about by the fāʿil, and this applies also to al-samāwāt in the abovementioned verse.17 Ibn Hišām also explains why “most grammarians” parsed such effective objects as al-mafʿūl bihi: people can only produce actions (ʾafʿāl), not substances (ḏawāt), while God produces both. And since the grammarians illustrated al-mafʿūl al-muṭlaq with the former, they believed that it designated only events.18 In our paragraph, Ibn Hišām raises a possible difficulty for his parsing. One can apply the passive participle of the verb ḫalaqa, namely maḫlūq, to alsamāwāt, which is a property of al-mafʿūl bihi. That is, just as zayd is maḍrūb (the passive participle of ḍaraba) in ḍarabtu zaydan, al-samāwāt are maḫlūqa in ḫalaqa llāhu l-samāwāti.19 Ibn Hišām does not refute this argument.20 Al-Jurjānī’s and Ibn Hišām’s characterization of the contrast al-mafʿūl almuṭlaq vs. al-mafʿūl bihi corresponds to Goldenberg’s (2003:167) discussion of “… inner objects, whose sense is contained in the verb, that is Objects Effected (not affected) …”. He explains (ibid., 169): These are in fact periphrastic renderings of the verb in question […] which should in principle separate the general statement of ‘doing, making, performing’ from the lexeme specifying what it is that is said to be done, or made, or performed, as, e.g., ‘fight (with) → make war (with)’ […] In Semitic languages, however, the use of such general vicarious verbs is mostly limited, and reiteration of the specific related verb is substituted. This is also in line with Jespersen’s (2007:159f.) classification of inner objects, e.g. ‘dream a strange dream’, as objects of result (= effective objects), which he exemplifies with sentences such as ‘He built a house’. The conundrum this type of effective objects poses to the grammarians is represented in Table 1.

17 18

19 20

Suyūṭī, ʾAšbāh VII, 140. Ibn Hišām, Muġnī VI, 579f. The parsing in question undoubtedly also has theological aspects. In this discussion Ibn Hišām expresses the orthodox view, that it is God alone who is really the creator of both substances and actions. This, however, has no effect on the issue at hand, with which he deals as a purely linguistic matter. He even states that his discussion applies also to sentences such as ʾanšaʾtu kitāban ‘I composed a book’, where the producer is a human being rather than God. Cf. Kasher (2012a). Ibn Hišām, Muġnī VI, 579.

how to parse effective objects according to arab grammarians? 203 table 1

effective maṣdar

The parsing of effective objects

ḍaraba zaydun ʿamran

ḍarabtuhu ḍarban

ḫalaqa llāhu l-samāwāti

– –

+ +

+ –

Constituents parsed as al-mafʿūl al-muṭlaq are normally [+maṣdar] and [+effective], whereas those parsed as al-mafʿūl bihi are prototypically [-maṣdar] and [-effective]. Yet, effective objects of the type ḫalaqa llāhu l-samāwāti are characterized as [-maṣdar] and [+effective]. It is now up to the grammarian to decide which of the two features should be regarded as the decisive feature with respect to their classification. For most grammarians, as Ibn Hišām states, it is [+maṣdar], while for Ibn Hišām himself, as well as for al-Jurjānī, it is [+effective]. According to the counter-argument raised by Ibn Hišām, a sufficient condition for al-mafʿūl bihi is the possibility to apply the passive participle to the constituent in question, and this condition is satisfied with al-samāwāt. Although Ibn Hišām does not solve this difficulty, it is safe to assume that he simply does not take this feature to be a sufficient condition for al-mafʿūl bihi.

4

The Literal Meaning of Technical Terms

An interesting issue which al-Jurjānī’s and Ibn Hišām’s discussions raise is the extent to which grammarians take into consideration the literal meaning of a technical term in matters of categorization. This is a very complicated question, which merits a separate study. For now I will restrict myself to some observations. The first is an obvious one: the basic reason for using a certain word or phrase as a technical, or semi-technical, term is its literal meaning. This point is emphasized by Carter (1994:400), who notes that “their [sc. technical terms’] most significant feature for students of Arabic grammar or indeed of any foreign science” is “that their creation often involves a metaphorical extension”. He adds (1994:400f.) that the literal meaning of technical terms should not be ignored; the term fiʿl, for instance, can be used to denote both ‘act’ and ‘verb’. This last point is developed by Peled (1999), who demonstrates that grammarians very frequently employ what he calls ‘metagrammatical intuitive terms’, that is, words whose semantic scope covers both their meaning as technical term and the everyday concept underlying them.

204

kasher

In consequence, one may expect some correlation between a category and the literal meaning of its term. A case in point is Sībawayhi’s (d. 177/793?) assertion that al-ṯawb ‘the garment’ in kasawtu zaydan-i l-ṯawba ‘I clothed Zayd in a garment’ is not a circumstantial qualifier (ḥāl). Rather, it is a mafʿūl. Sībawayhi’s initial argument, before delving into syntactic and semantic proofs, is simply that inasmuch as al-ṯawb does not refer to ḥāl (in the everyday meaning of the word), it does not qualify as a ḥāl.21 However, the issue is far more intricate than that, since the grammarians were fully cognizant of the distinction between the literal and the technical meaning of terms, as shown by the contrast they frequently made between luġatan and iṣṭilāḥan.22 But far more relevant for us is the grammarians’ recognition that the literal meaning does not dictate the extension of the technical term. An obvious illustration is the term fāʿil (the subject in verbal sentences). Some grammarians state explicitly that the constituent parsed as fāʿil is not necessarily a ‘doer’ (i.e. Agent), e.g. in māta zaydun ‘Zayd died’. For instance, Ibn al-ʾAnbārī (d. 577/1181) states that zayd in this sentence is a fāʿil, although Zayd did not bring about the death.23 Ibn Ḫālawayhi (d. 370/980–981) (ʾIʿrāb 70)24 goes so far as to assign the term fāʿil to the subject of the passive verb as well.25 An explicit statement to the effect that the literal meaning of a term does not dictate its extension is made by al-Jurjānī, when he discusses the practice of using the term istiʿāra too broadly, and not restricting it to metaphors.26 He compares this with a hypothetical application of the grammatical term tamyīz lit. ‘specification’ also to ḥāl: just as it is incorrect to apply the technical term istiʿāra to non-metaphors, on the ground of its literal meaning, i.e. ‘borrowing’, it would be incorrect to apply the term tamyīz to ḥāl as well, on the ground that by saying rākiban ‘while riding’, you (viz. the speaker) mayyazta and bayyanta lit. ‘clarified’,27 just as it is, for instance, in ʿišrūna dirhaman ‘twenty dirhams’,

21 22 23 24 25 26

27

Sībawayhi, Kitāb I, 15: li-ʾanna l-ṯawba laysa bi-ḥālin waqaʿa fīhā l-fiʿlu. See e.g. Carter (1981:8f.): maʿnā fī l-luġa / maʿnāhu luġatan vs. maʿnā fī l-iṣṭilāḥ / maʿnāhu iṣṭilāḥan. See also Kasher (2012b:157). Ibn al-ʾAnbārī, ʾAsrār 88f. See also Peled (1994:141, 1999:68); Guillaume (1998:56 f.). Quoted in Hamzé (2007:82). However, in contrast to Hamzé’s categorical assertion, the subject of the passive verb is usually not parsed as fāʿil. Ibn Ḫālawayhi, ʾIʿrāb 70. See also Peled (1994:141, 1999:54 f.). Al-Jurjānī criticizes here Ibn Durayd ( Jamhara III, 1255 ff.) for subsuming cases such as ġayṯ ‘rain’ → ‘what vegetates due to rain’ under bāb al-istiʿārāt. See also Heinrichs (1991– 1992). The term tabyīn lit. ‘clarification’, is synonymous with tamyīz, in the technical sense (see

how to parse effective objects according to arab grammarians? 205

which is, according to the grammarians, a case of tamyīz.28 It is important to note that the parallel between tamyīz and ḥāl with respect to their functioning as tabyīn is not an ad hoc invention of al-Jurjānī for the sake of argument. Al-Fārisī (d. 377/987) uses the verb bayyana, with respect to the ḥāl, thus comparing it with the tamyīz. According to al-Fārisī, this analogy explains the indefiniteness of the ḥāl, based on the indefiniteness of the tamyīz.29 Al-Jurjānī, on the other hand, notes both indefiniteness and bayān as two separate points of similarity between the ḥāl and the tamyīz.30 From al-Jurjānī’s and Ibn Hišām’s discussions it may be inferred that the literal meaning of the term al-mafʿūl al-muṭlaq carried more weight for them than for most grammarians. This also applies to the term al-mafʿūl bihi, which they contrast with al-mafʿūl al-muṭlaq, while considering the literal meaning of both: whereas al-mafʿūl al-muṭlaq is the constituent referring to that which (or even: who) is produced, al-mafʿūl bihi is the constituent referring to that to which, or to whom, the action is done. As we have seen, al-Jurjānī and Ibn Hišām state that it is wrong to apply al-mafʿūl bihi to al-ʿālam and al-samāwāt (respectively) after the verb ḫalaqa, in contrast to its application to zayd (in ḍarabtu zaydan), since the former is not a mafʿūl bihi in the literal sense. Here also it seems that the literal meaning of the term is more decisive for them with regard to the extension of the category in question than it is for most grammarians, who apply the term al-mafūl bihi also to effective objects, notwithstanding the fact that the literal meaning of this term does not apply to them very well. Interestingly, this corresponds to Lyons’ (1977:492) comment on objects of result (= effective objects): “it does not make sense to say What God did to Adam was to create him.” It should be kept in mind, however, that Arab grammarians never restricted the extension of al-mafʿūl bihi to affective objects in the first place, for they applied it also to accusative nominals which are neither affective nor effective objects, e.g. the objects of ḥasiba ‘he thought’.31 The literal meaning of the term al-mafʿūl bihi was thus never regarded as dictating the extension of the category; it only characterizes its prototypical cases. On the other hand, with respect to the characterization by the majority of grammarians of al-mafʿūl al-muṭlaq as maṣdar, this may also be connected to

28 29 30 31

e.g. Carter (1981:380)). This term would seem to constitute a better illustration of alJurjānī’s argument. Jurjānī, ʾAsrār 369f. Jurjānī, Muqtaṣid I, 675. See also Ibn al-ʾAnbārī, ʾAsrār 199. Jurjānī, Muqtaṣid I, 675f. Ibn al-Warrāq (ʿIlal 371) explains the indefiniteness of the ḥāl on the ground of this very analogy. See Kasher (2012a), and the references therein.

206

kasher

the terminological level, as well as to the term’s history: as is well known, grammarians make extensive use of maṣdar as a syntactic term, on a par with other mafʿūlāt terms. Such is the case, for instance, in the list of nouns taking the accusative (manṣūbāt al-ʾasmāʾ) given in Ibn ʾĀjurrūm’s (d. 723/1323) celebrated al-ʾĀjurrūmiyya. Here the term maṣdar is simply interposed between the terms al-mafʿūl bihi and ẓarf al-zamān (the temporal qualifier),32 a practice criticized by al-Širbīnī (d. 977/1570), a commentator on this short grammar, who states that its author should have given the relevant chapter the title of Bāb al-mafʿūl al-muṭlaq instead of Bāb al-maṣdar, since maṣdars can assume a number of different functions.33 It makes sense that the grammarians were disinclined to subsume effective objects of the al-samāwāt type under maṣdar.34 Furthermore, the first extant grammar in which the term al-mafʿūl al-muṭlaq appears is, as far as we know,35 Ibn al-Sarrāj’s (d. 316/928) al-ʾUṣūl fī l-naḥw; the term is thus probably an innovation, coined long after the category had already been well established.

5

Two Remarks on Ibn Hišām’s Argument

We return now to Ibn Hišām’s discussion. Because it is not a maṣdar, alsamāwāt is excluded from being parsed as al-mafʿūl al-muṭlaq by the majority of the grammarians. Ibn Hišām, however, also mentions their positive argument for parsing it as al-mafʿūl bihi, namely the possibility to apply to it the passive participle of its verb, maḫlūq. This property cuts across the effective/noneffective opposition. Needless to say, the categories al-mafʿūl al-muṭlaq and al-mafʿūl bihi are mutually exclusive, and hence a sufficient condition to be parsed as mafʿūl bihi constitutes also a sufficient condition to be excluded from al-mafʿūl al-muṭlaq.

32 33 34

35

Carter (1981:324ff.). Carter (1981:344). See also Ibn Hišām, Šarḥ II, 158f. (see above). See Peled (1999:72, 85, n. 18). As is well known, the syntactic position of al-mafʿūl al-muṭlaq is not always filled by a maṣdar. However, the phrase assuming this function is always (apart from the effective objects dealt with here, of course) related to the maṣdar, conveying, or qualifying, the action in question. To take an extreme case, ḍarabtuhu sawṭan ‘I struck him with a whip’, Ibn Yaʿīš (Šarḥ I, 112f.) maintains that here sawṭan takes the accusative ʿalā l-maṣdar, although it is not a maṣdar, since the underlying form of this sentence is ḍarabtuhu ḍarbatan bi-lsawṭi. See also al-Zamaḫšarī’s general statement regarding cases where the constituent in question is not the verb’s morphological maṣdar (Ibn Yaʿīš, Šarḥ I, 111). As noted in Levin (1991:917).

how to parse effective objects according to arab grammarians? 207

A note about Ibn Hišām’s line of argumentation is in order here. His counterargument is placed after the first argument, to wit, that al-samāwāt should be parsed as al-mafʿūl al-muṭlaq since one can apply mafʿūl to it. As a counterargument it is said that one can also apply the passive participle maḫlūq to it. The argument and the counter-argument seem to be conceived as placed on the same level (i.e. application of mafʿūl vs. application of a passive participle), thus an inconclusive result is reached, a tie, as it were, which leads Ibn Hišām to add another ‘clarification’, regarding the existence of the referent of the mafʿūl bihi prior to the action vs. the fact that the referent of the mafʿūl muṭlaq exists only through the action designated by the verb. Yet, Ibn Hišām’s two arguments constitute merely two sides of the same coin.

6

Conclusion

The issue at stake can be formulated as follows: how should a constituent be parsed so that it is (i) effective, (ii) non-maṣdar and (iii) one to which the passive participle of its verb can be applied. Being ‘effective’ is a property of the ‘regular’ al-mafʿūl al-muṭlaq; it tallies with this term’s literal meaning, and it is problematic with respect to al-mafʿūl bihi as far as this term’s literal meaning is taken into consideration. Being a maṣdar is considered by most grammarians to be a necessary condition for a constituent being parsed as al-mafʿūl al-muṭlaq, whereas for Ibn Hišām it is merely characteristic of the illustrations given by grammarians to this category. The possibility to apply the passive participle of the verb to the constituents in question may be regarded by some grammarians as a sufficient condition for parsing them as al-mafʿūl bihi. It is inferred that for al-Jurjānī and Ibn Hišām this is not the case. Whereas Levin’s and Larcher’s accounts of the term al-mafʿūl al-muṭlaq apply to the great majority of grammarians, at least for al-Jurjānī and Ibn Hišām the opening of Levin’s definition (see above) should be reformulated in the following way: “the accusative expressing that which (or who) is produced …”, be it a maṣdar or not.

Excursus: Alternative Explanations for the Term al-mafʿūl al-muṭlaq It turns out that the interpretation discussed by Levin and Larcher was not the only one furnished in the Arabic grammatical tradition. Al-ʿUkbarī (d. 616/1219) offers two explanations of the term, the second of these being identical with

208

kasher

the abovementioned one. According to the first explanation, the maṣdar is termed al-mafʿūl al-muṭlaq because it is al-mafʿūl ʿalā l-taḥqīq (for the expressions al-mafʿūl al-ḥaqīqī and al-mafʿūl fī l-ḥaqīqa see above), in the sense that the sentence ḍarabtu zaydan does not mean that the fāʿil produced Zayd, but rather that he produced an action, and that he produced it upon Zayd.36 This assertion is indeed commensurate with the more generally accepted explanation of the term al-mafʿūl al-muṭlaq, as shown above. Yet, from the fact that al-ʿUkbarī presents it as a separate option, it is inferred that he proposes here to relate al-mafʿūl al-muṭlaq with al-mafʿūl ʿalā l-taḥqīq. In light of this alternative explanation, one wonders how a grammarian such as Ibn al-Warrāq (d. 381/991) apprehended the item muṭlaq when he explained that the term al-mafʿūl al-muṭlaq was used “because the doer produced it” (li-ʾanna l-ʿāmila37 ʾaḥdaṯahu).38 Yet another alternative is found in Ibn ʿUṣfūr’s (d. 670/1271) discussion of the term al-mafʿūl al-muṭlaq. This grammarian presents two explanations, as a disagreement among grammarians, the first being the common one given above. According to the second explanation, it is so named since the verb is connected with it (or: reaches it)39 by itself, while it connects with the other mafʿūlāt by means of a particle, overt or underlying.40 For instance, the verb is connected with al-mafʿūl lahu by means of an underlying li-. As for al-mafʿūl bihi, Ibn ʿUṣfūr himself raises the question, for the verb is neither connected with it with a particle, nor with an underlying particle. However, says Ibn ʿUṣfūr, there are indeed cases where al-mafʿūl bihi is the object of a preposition, e.g. marartu bi‘I passed by’.41

36 37

38 39 40

41

ʿUkbarī, Lubāb I, 261f. Here, al-ʿāmil does not have the meaning of ‘operator’. It is either used in its literal sense, as ‘doer’, or rather should be corrected to al-fāʿil, a term used in the immediately following explanation of al-mafʿūl bihi (li-ʾanna l-fāʿila lam yafʿal zaydan …). Ibn al-Warrāq, ʿIlal 320. The expression used is yaṣilu ʾilayhi. On this notion, see Kasher (2013). The text here is probably corrupt, as it reads bi-taqdīri fī. However, on the next page (Ibn ʿUṣfūr, Šarḥ II, 450), the same notion is reiterated: … bi-ḥarfi jarrin ʾaw bi-taqdīrihi (this to the exclusion of al-mafʿūl bihi, on which see in what follows). The word jarr in this phrase is dubious, since in the case of al-mafʿūl maʿahu it is wa- which connects the verb with the mafʿūl in question (Ibn ʿUṣfūr, Šarḥ II, 451). Ibn ʿUṣfūr, Šarḥ II, 449. On Ibn ʿUṣfūr’s classification of objects of prepositions as al-mafʿūl bihi, see e.g. Šarḥ I, 300 (but see Šarḥ I, 161).

how to parse effective objects according to arab grammarians? 209

Bibliographical References A

Primary Sources

Ibn al-ʾAnbārī, ʾAsrār = Kamāl al-Dīn ʾAbū l-Barakāt ʿAbd al-Raḥmān ibn Muḥammad ibn ʾAbī Saʿīd al-ʾAnbārī, Kitāb ʾasrār al-ʿarabiyya. Ed. by Muḥammad Bahjat alBayṭār. Damascus: al-Majmaʿ al-ʿIlmī al-ʿArabī, 1957. Ibn al-ʾAnbārī, ʾInṣāf = Kamāl al-Dīn ʾAbū l-Barakāt ʿAbd al-Raḥmān ibn Muḥammad ibn ʾAbī Saʿīd al-ʾAnbārī, al-ʾInṣāf fī masāʾil al-ḫilāf bayna l-naḥwiyyīna l-baṣriyyīna wa-l-kufiyyīna. Ed. by Muḥammad Muḥyī l-Dīn ʿAbd al-Ḥamīd. 2 vols. 4th ed. Egypt: al-Maktaba al-Tijāriyya al-Kubrā, 1961. Ibn al-Warrāq, ʿIlal = ʾAbū l-Ḥasan Muḥammad ibn ʿAbdallāh al-Warrāq, ʿIlal al-naḥw. Ed. by Muḥammad Jāsim Muḥammad al-Darwīš. Riyadh: Maktabat al-Rušd, 1999. Ibn Durayd, Jamhara = ʾAbū Bakr Muḥammad ibn al-Ḥasan Ibn Durayd, Kitāb jamharat al-luġa. Ed. by Ramzī Munīr Baʿalbakī. 3 vols. Beirut: Dār al-ʿIlm li-l-Malāyīn, 1987. Ibn Ḫālawayhi, ʾIʿrāb = ʾAbū ʿAbdallāh al-Ḥusayn ibn ʾAḥmad al-maʿrūf bi-Ibn Ḫālawayhi, Kitāb ʾiʿrāb ṯalāṯīna sūra min al-Qurʾān al-Karīm. Cairo: Maktabat al-Zahrāʾ, n.d. Ibn Hišām, Muġnī = Jamāl al-Dīn ʿAbdallāh ibn Yūsuf Ibn Hišām al-ʾAnṣārī, Muġnī llabīb ʿan kutub al-ʾaʿārīb. Ed. by ʿAbd al-Laṭīf Muḥammad al-Ḫaṭīb. 7 vols. Kuwait: al-Majlis al-Waṭanī li-l-Ṯaqāfa wa-l-Funūn wa-l-ʾĀdāb, al-Turāṯ al-ʿArabī, 2000–2002. Ibn Hišām, Muġnī (ed. Damascus) = Jamāl al-Dīn ʿAbdallāh ibn Yūsuf Ibn Hišām alʾAnṣārī, Muġnī l-labīb ʿan kutub al-ʾaʿārīb. Ed. by Māzin al-Mubārak and Muḥammad ʿAlī Ḥamd Allāh. 2 vols. 2nd ed. [Damascus]: Dār al-Fikr, 1969. Ibn Hišām, Šarḥ = Jamāl al-Dīn ʿAbdallāh ibn Yūsuf Ibn Hišām al-ʾAnṣārī, Šarḥ al-Lamḥa al-badriyya fī ʿilm al-luġa al-ʿarabiyya. Ed. by Hādī Nahar. 2 vols. ʿAmmān: al-Yāzūrī, [2007]. Ibn ʿUṣfūr, Šarḥ = ʾAbū l-Ḥasan ʿAlī ibn Muʾmin Ibn ʿUṣfūr al-ʾIšbīlī, Šarḥ Jumal al-Zajjājī. Ed. by Ṣāḥib ʾAbū Janāḥ. n.p. n.d. Ibn Yaʿīš, Šarḥ = Muwaffaq al-Dīn Yaʿīš ibn ʿAlī Ibn Yaʿīš, Šarḥ al-Mufaṣṣal. 10 vols. Egypt: ʾIdārat al-Ṭibāʿa al-Munīriyya, n.d. Jurjānī, ʾAsrār = ʾAbū Bakr ʿAbd al-Qāhir ibn ʿAbd al-Raḥmān al-Jurjānī, Kitāb ʾasrār albalāġa. Ed. by Hellmut Ritter. Istanbul: Government Press, 1954. Jurjānī, Muqtaṣid = ʾAbū Bakr ʿAbd al-Qāhir ibn ʿAbd al-Raḥmān al-Jurjānī, Kitāb almuqtaṣid fī šarḥ al-ʾĪḍāḥ. Ed. by Kāẓim Baḥr al-Marjān. 2 vols. [Baghdad]: Dār alRašīd, 1982. Sībawayhi, Kitāb = Le livre de Sībawaihi. Ed. by Hartwig Derenbourg. 2 vols. Paris: Imprimerie nationale, 1881–1889. (Repr. Hildesheim: G. Olms, 1970.) Suyūṭī, ʾAšbāh = Jalāl al-Dīn ʾAbū l-Faḍl ʿAbd al-Raḥmān ibn ʾAbī Bakr al-Suyūṭī, alʾAšbāh wa-l-naẓāʾir fī l-naḥw. Ed. by ʿAbd al-ʿĀl Sālim Mukarram. 9 vols. Beirut: Muʾassasat al-Risāla, 1985.

210

kasher

ʿUkbarī, Lubāb = ʾAbū l-Baqāʾ ʿAbdallāh ibn al-Ḥusayn al-ʿUkbarī, al-Lubāb fī ʿilal al-bināʾ wa-l-ʾiʿrāb. Ed. by Ġāzī Muḫtār Ṭulaymāt and ʿAbd al-ʾIlāh Nabhān. 2 vols. Beirut: Dār al-Fikr al-Muʿāṣir, 1995. Zajjāj, Maʿānī = ʾAbū ʾIsḥāq ʾIbrāhīm ibn al-Sarī al-Zajjāj, Maʿānī l-Qurʾān wa-ʾiʿrābuhu. Ed. by ʿAbd al-Jalīl ʿAbduh Šalabī. 5 vols. Beirut: ʿĀlam al-Kutub, 1988.

B

Secondary Sources

Abdel Haleem, Muhammad A.S., trans. 2005. The Qurʾān. Oxford: Oxford University Press. Abu Deeb, Kamal. 1979. Al-Jurjānī’s theory of poetic imagery. Warminster: Aris and Phillips. Baalbaki, Ramzi. 1990. Dictionary of linguistic terms: English—Arabic, with sixteen Arabic glossaries. Beirut: Dār al-ʿIlm li-l-Malāyīn. Baalbaki, Ramzi. 1999. “Expanding the maʿnawī ʿawāmil: Suhaylī’s innovative approach to the theory of regimen”. Al-Abhath 47.23–58. Carter, Michael G. 1981. Arab linguistics: An introductory classical text with translation and notes. Amsterdam: J. Benjamins. Carter, Michael G. 1994. “Writing the history of Arabic grammar”. Historiographia Linguistica 21.385–414. Goldenberg, Gideon. 2013. Semitic languages: Features, structures, relations, processes. Oxford: Oxford University Press. Guillaume, Jean-Patrick. 1998. “Les discussions des grammairiens arabes à propos du sens des marques d’iʿrab”. Histoire Épistémologie Langage 20.43–62. Gully, Adrian. 1995. Grammar and semantics in Medieval Arabic: A study of Ibn Hisham’s ‘Mughni l-Labib’. Richmond: Curzon. Hamzé, Hassan S.B. 2007. “Fāʿil”. Encyclopedia of Arabic Language and Linguistics, ed. by Mushira Eid et al., II, 82–84. Leiden: E.J. Brill. Heinrichs, Wolfhart. 1991–1992. “Contacts between scriptural hermeneutics and literary theory in Islam: The case of majāz”. Zeitschrift für Geschichte der arabischislamischen Wissenschaften 7.253–284. Jespersen, Otto. 2007. The philosophy of grammar. London: Routledge. Kasher, Almog. 2012a. “The term mafʿūl in Sībawayhi’s Kitāb”. The foundations of Arabic linguistics: Sībawayhi and the early Arabic grammatical theory, ed. by Amal E. Marogy, 3–26. Leiden: E.J. Brill. Kasher, Almog. 2012b. “A note on the literal meaning(s) of the term (ḍamīr al-) faṣl”. Journal for Semitics 21.157–166. Kasher, Almog. 2013. “The term al-fiʿl al-mutaʿaddī bi-ḥarf jarr (lit. ‘the verb which ‘passes over’ through a preposition’) in Medieval Arabic grammatical tradition”. Journal of Arabic and Islamic Studies 13.115–145. Larcher, Pierre. 1991. “Les mafʿûl mut‘laq ‘à incidence énonciative’ de l’ arabe classique”.

how to parse effective objects according to arab grammarians?

211

L’adverbe dans tous ses états: Travaux linguistiques du CERLICO 4, ed. by Claude Guimier and Pierre Larcher, 151–178. Rennes: PUR 2. [= Pierre Larcher, Linguistique arabe et pragmatique, 291–316. Beyrouth: Presses de l’ Ifpo, 2014.] Larkin, Margaret. 1995. The theology of meaning: ʿAbd al-Qāhir al-Jurjānī’s theory of discourse. New Haven: American Oriental Society. Levin, Aryeh. 1991. “What is meant by al-mafʿūl al-muṭlaq?”. Semitic studies in honor of Wolf Leslau on the occasion of his eighty-fifth birthday, November 14th, 1991, ed. by Alan S. Kaye, II, 917–926. Wiesbaden: O. Harrassowitz. Lyons, John. 1968. Introduction to theoretical linguistics. Cambridge: Cambridge University Press. Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press. Owens, Jonathan. 1990. Early Arabic grammatical theory: Heterogeneity and standardization. Amsterdam: J. Benjamins. Peled, Yishai. 1994. “Aspects of case assignment in Medieval Arabic grammatical theory”. Wiener Zeitschrift für die Kunde des Morgenlandes 84.133–158. Peled, Yishai. 1999. “Aspects of the use of grammatical terminology in Medieval Arabic grammatical tradition”. Arabic grammar and linguistics, ed. by Yasir Suleiman, 50–85. Richmond: Curzon.

The Phenomenon of ittisāʿ al-kalām in Old Arabic Aryeh Levin

1

Introduction

In the terminology of Sībawayhi, al-Mubarrad, Ibn al-Sarrāj and a number of other grammarians, the terms ittisāʿ al-kalām and saʿat al-kalām refer to some syntactic phenomena that do not occur in the ordinary way of speech. The Old Arabic grammarians who deal with these phenomena in detail, say that the phenomenon of ittisāʿ al-kalām was widespread in Bedouin speech.1 Some of the grammarians’ examples of this phenomenon occur in the text of the Qurʾān2 and in Old Arabic poetry.3 The phenomenon of ittisāʿ al-kalām has been studied in the past by Ramzi Baalbaki and Kees Versteegh.4 I myself have briefly mentioned it in an earlier article.5 After finishing the present article for the proceedings of the Fourth Conference on the Foundations of Arabic Linguistics, I was informed by the editors that Hanadi Dayyeh of The American University of Beirut had published in the proceedings of the Second Conference on the Foundations of Arabic Linguis-

1 See, e.g. Sībawayhi, Kitāb I, 89.17f. In referring to two examples of ittisāʿ al-kalām occurring in the above text, Sībawayhi says: “Such phenomena are innumerable” (wa-hāḏā ʾakṯaru min ʾan yuḥṣā). Ibn al Sarrāj (ʾUṣūl II, 255.15) says: “These occurrences of deviation from the ʾaṣl [are innumerable] and cannot be thoroughly acquainted” (wa-hāḏā l-ittisāʿ ʾakṯar min ʾan yuḥāṭa bihi). 2 See e.g. Sībawayhi, Kitāb I, 88.16–89.4. 3 See e.g. Sībawayhi, Kitāb I, 69.6–10; 75.20; 89.10–17. 4 See Baalbaki (1988); Versteegh (1990). 5 See Levin (1997:154–157, §4.3). The present article does not deal with the semantic phenomenon of the extension of the primary sense of certain words, by using them in a figurative sense. Irrespective of the fact that the grammarians refer to this phenomenon by using the verb ittasaʿū as a technical phrase, meaning “they [i.e.the Beduin speakers] extended the primary sense of a certain word by using it in a figurative sense”. This semantic phenomenon completely differs from the syntactic phenomenon called ittisāʿ al-kalām, which is the topic of this article. The statement in my discussion of ittisāʿ al-kalām (Levin 1997:156, n. 81) that this term refers to both syntactic and semantic phenomena is incorrect. It is also incorrect that the phenomenon of ittisāʿ al-kalām belongs to the theory of al-taqdīr (Levin 1997:155.26– 157.28). For the correct views on these points see below 3.3. It should be emphasized that the view that in the above text the main notion of the theory of al-taqdīr is compared by al-Ḫalīl to examples of the phenomenon of ittisāʿ al-kalām is correct (see below 3.3).

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_013

the phenomenon of ittisāʿ al-kalām in old arabic

213

tics an article on ittisāʿ al-kalām (Dayyeh 2015). I was not aware of this article since I did not participate in the second conference and I had not seen its proceedings.6

2

The Sense of the Term ittisāʿ al-kalām

Although the grammarians do not explain and do not define this term, it is possible to infer its sense and its definition from the grammatical texts. In the works of al-Mubarrad and Ibn al-Sarrāj the term ittisāʿ al-kalām is opposed to al-ʾaṣl,7 which denotes “the ordinary way of speech”. This sense of ʾaṣl is attested by Baalbaki’s definition of one of the meanings of this term as “the form, pattern, case-ending etc. which agrees with the qiyās, i.e., with the norm and with the usage which is most frequently attested in accepted dialects”.8 In Baalbaki’s view, the literal sense of ittisāʿ al-kalām is “extension, latitude of speech”,9 i.e., the extension of the ordinary way of speech. The significance of this extension is that the speaker applies it by using syntactic constructions that deviate from the ʾaṣl. Hence, it is possible to define ittisāʿ al-kalām as “the extension of syntactic constructions occurring in ordinary speech, by using some structures deviating from the norm”. Ibn al-Sarrāj, who conceives of the phenomenon of ittisāʿ al-kalām mainly as a phenomenon of ‘omission’ (ḥaḏf ),10 says that the occurrence of sentences characterized by phenomena of ittisāʿ al-kalām is brought about by making these utterances shorter than their underlying and primary utterances, as they occur in the ordinary way of speech. An example is the utterance ṣīda ʿalayhi yawmāni lit. ‘Two days were hunted in it’ (Kitāb I, 88.11), which originates in the utterance ṣīda ʿalayhi l-waḥšu fī yawmayni ‘Wild animals were hunted in it for two days’ (Kitāb I, 88.11). The extension of the ordinary way of speech is achieved by dropping the nominative al-waḥšu, irrespective of the fact that it is the indispensable subject of the verb ṣīda, and then putting the adverb of time yawmāni in its place. Thus, the utterance ṣīda ʿalayhi yawmāni becomes shorter 6

7 8 9 10

Hanadi Dayyeh’s (2015) contribution to the proceedings of FAL 2 is a good article. Irrespective of this, I cannot accept her approach and some of her notions concerning ittisāʿ al-kalām. Her article deserves a detailed answer, which is beyond the limits of my current paper. Hence, I hope to write in the future an article referring to her concept of ittisāʿ alkalām. See e.g. Mubarrad, Muqtaḍab IV, 330.7–9. See Baalbaki (1988:163.26–28). See Baalbaki (1988:129.28f.). See Ibn al-Sarrāj, ʾUṣūl I, 255.1f.

214

levin

than its underlying and primary form, and its syntactic construction deviates from the one occurring in the ordinary way of speech. The construction of ittisāʿ al-kalām is also achieved by the omission of one of the ḥurūf al-jarr from the sentence (see below). Apart from the mechanism of ḥaḏf, there are some other phenomena that occur in utterances characterized by ittisāʿ al-kalām, most of them relevant to the division of words into parts of speech, especially into secondary parts of speech: words occurring in the ʾaṣl as adverbs of time and place occur in the lafẓ of ittisāʿ al-kalām as nouns; maṣdar forms in the ʾaṣl occur sometimes as adverbs of time or as a direct object, and adverbs of time frequently occur as a direct object (mafʿūl ʿalā l-saʿa), or as a nominative assigned as a subject to a passive verb, and even as a nominative assigned as a subject to an active participle (see below).

3

Some Features of Utterances Characterized by the Phenomenon of ittisāʿ al-kalām

3.1 Lafẓ and maʿnā In Sībawayhi’s view, every utterance characterized by the phenomenon of ittisāʿ al-kalām has a lafẓ and a maʿnā. The lafẓ, which is the shorter utterance pronounced by the speaker, is the literal structure of the utterance, while the maʿnā is the sense of the utterance, as it is intended by the speaker.11 The lafẓ expresses a literal sense, which is not regarded by the speaker as the sense of the utterance. For example, the literal sense of the utterance banū fulānin yaṭaʾuhumu l-ṭarīqu is ‘The sons of such-and-such a tribe, the road treads them’ (Sībawayhi, Kitāb I, 89.4f.). Since the literal sense of this utterance is illogical and even designates an absurdity, the speaker feels and knows that it does not express the sense intended by him. Hence, when pronouncing this utterance, the speaker intends the sense of his literal utterance to be that of another utterance, namely [banū fulānin] yaṭaʾuhum ʾahlu l-ṭarīqi ‘[The sons of such-and-such a tribe], the people [traveling] on the road tread

11

This conclusion is inferred from the title of chapter 42 of the Kitāb (I, 88.9), and from the fact that Sībawayhi’s explanations of the sense of the literal form of sentences characterized by the phenomenon of ittisāʿ al-kalām is usually preceded by the words al-maʿnā or wa-ʾinnamā or wa-ʾinnamā l-maʿnā (see e.g. Kitāb I, 88.11–89). In this context wa-ʾinnamā is a particle indicating that the following text is an explanation of the sense of a word or of a sentence or of a notion. I am indebted to my late teacher H.J. Polotsky for this sense of ʾinnamā.

the phenomenon of ittisāʿ al-kalām in old arabic

215

them’ (Sībawayhi, Kitāb I, 89.4f.), i.e., the travelers become the guests of suchand-such a tribe, encamping on their road. The above expression expresses praise, referring to the generosity of the tribe encamping on the travelers’ road. The maʿnā, i.e., the sense intended by the speaker when pronouncing the lafẓ, is actually the sense of the longer utterance, as it is pronounced in ordinary speech. It should be noted that Sībawayhi says that the interlocutor understands sentences characterized by the phenomenon of ittisāʿ al-kalām, “because the interlocutor knows the sense [of the speaker’s shortened utterance]” (li-ʿilm almuḫāṭab bi-l-maʿnā), irrespective of the fact that part of it was dropped.12 This view of Sībawayhi is important evidence that utterances characterized by phenomena of ittisāʿ al-kalām frequently occurred in Bedouin speech, irrespective of the fact that their syntactic structure contains phenomena that deviate from the ordinary way of speech. It is also evidence that utterances belonging to this category were easily understood by Bedouin speakers. 3.2 Istiʿmāl al-fiʿl fī l-lafẓ lā fī l-maʿnā Another feature of the lafẓ is that the verb contained in it governs a nominative or an accusative only from the grammatical point of view, but not as regards the sense of the utterance. In the title of chapter 42 of the Kitāb, which is Sībawayhi’s first chapter dealing with ittisāʿ al-kalām, he says in this respect: This is the chapter dealing with the application of the ʿamal of the verb as regards the lafẓ [of the utterance], but not as regards the sense intended by the speaker, because they [i.e., the Bedouin speakers] extend the ordinary way of speech, and because of their intention to make the utterance shorter (hāḏā bāb istiʿmāl al-fiʿl fī l-lafẓ lā fī l-maʿnā li-ttisāʿihim fī l-kalām wa-li-l-ʾījāz wa-l-iḫtiṣār). Kitāb I, 88.9

In the utterance ṣīda ʿalayhi yawmāni (Kitāb I, 88.11), for instance, the ʿamal of the verb ṣīda produces the nominative in the noun yawmāni. This ʿamal affects yawmāni because any verb occurring in a sentence must produce the nominative in one of the nouns occurring in it (Kitāb I, 88.11). The sense of ṣīda yawmāni is illogical, but ṣīda affects yawmāni because it is the only noun occurring in the sentence. However, the combination ṣīda yawmāni occurs only in the lafz of the

12

See Sībawayhi, Kitāb I, 89.f.

216

levin

sentence, not in the maʿnā intended by the speaker. In the maʿnā, which is ṣīda ʿalayhi l-waḥšu fī yawmayni, the verb ṣīda is assigned as a predicate to the noun al-waḥšu, and fī yawmayni is a ẓarf. 3.3 Ittisāʿ al-kalām and taqdīr It may be inferred from the texts of the early grammarians that the notion of taqdīr and the notion of ittisāʿ al-kalām differ from each other in some essential respects (see below).13 The theory of taqdīr offers solutions to the grammatical difficulties in the structure of utterances occurring in Bedouin speech, in the Qurʾān and in Old Arabic poetry, while the theory of ittisāʿ al-kalām deals with the logical and semantic problems created by the extension of the ordinary way of speech, especially by the omission (ḥaḏf ) of certain parts of the sentence. The notion of taqdīr was developed by the grammarians in order to solve a theoretical difficulty, and they apply it when they find that the literal construction of the utterance includes a grammatical phenomenon that does not accord with one of their theories. In the grammarians’ view, when pronouncing such an utterance, the speaker simultaneously intends it to be as if he were expressing another utterance,14 differing from his literal utterance in construction, but not in intended meaning.15 Thus, when the speaker pronounces such a literal utterance, a corresponding imaginary utterance exists in his mind. If we designate the literal utterance as ‘X’ and its corresponding imaginary utterance as ‘Y’, we can say that the main notion of the theory of al-taqdīr is that the speaker intends, or imagines, that when he says ‘X’, it is as if he were saying ‘Y’. For example, the grammarians hold that when saying zaydun fī l-dāri (= ‘X’) ‘Zayd is in the dwellings of the tribe’, the speaker intends that it is as if he were saying zaydun istaqarra fī l-dāri (= ‘Y’) lit. ‘Zayd settled in the dwellings of the tribe’ (Jurjānī, Muqtaṣid I, 275.3f.). The imaginary utterance zaydun istaqarra fī l-dāri is given the name al-taqdīr. In this use, taqdīr is to be defined as “the imaginary utterance which the speaker intends as if he were saying it, when expressing his literal utterance”.16 For the later grammarians’ view that the taqdīr of zaydun fī l-dāri is zaydun istaqarra fī l-dāri see Levin (1997:143.14–29). Irrespective of the fact that the lafẓ of utterances belonging to the category of ittisāʿ al-kalām deviates from the ordinary way of speech (= al-ʾaṣl), Sīb-

13 14 15 16

Versteegh (1990:285, 293) was the first to point out that “there is a fundamental difference between this term [i.e., ittisāʿ al-kalām] and taqdīr”. See Sībawayhi, Kitāb II, 137.8–15; cf. Jurjānī, Muqtaṣid I, 333.7–10. See Levin (1997: 154–157, §4.3). This point is inferred from Sībawayhi, Kitāb I, 372.17–373.17. For this definition see Levin (1997: 21, 151–157, §4).

the phenomenon of ittisāʿ al-kalām in old arabic

217

awayhi and the early grammarians do not say that the lafẓ has a corresponding taqdīr construction, which illustrates the sentence as it occurs in the speaker’s mind. It is evident that the early grammarians believe that the construction of utterances characterized by the phenomenon of ittisāʿ al-kalām is the construction intended by the speaker, hence it does not have, nor does it need to have, any corresponding taqdīr construction. This is evidenced by the following points: (i) As Baalbaki says, one of the characteristics of the utterances characterized by phenomena of ittisāʿ al-kalām is grammatical correctness. Hence, the lafẓ of these utterances does not need any corresponding taqdīr construction. (ii) The title of chapter 42 of the Kitāb17 is evidence that the syntactic construction of the lafẓ of such utterances accords with the intention of the speaker: when producing utterances of this type, the speaker deliberately deviates from the construction of the ʾaṣl in order to pronounce a shorter sentence than the one occurring in the ʾaṣl. Hence, it is evident that the syntactic construction of the lafẓ of such utterances is the construction intended by the speaker, so there is no need, nor is it possible to hold that these utterances have a corresponding taqdīr construction. Although the utterances belonging to the category of ittisāʿ al-kalām do not need any grammatical explanation, their sense does need some explanation. This explanation is what the grammarians call al-maʿnā “[the explanation of] the sense [intended by the speaker]”. This maʿnā is based on the ʾaṣl of the utterances characterized by ittisāʿ al-kalām, and it is even identical to it. For example, in referring to the utterance ṣīda ʿalayhi yawmāni, Sībawayhi says: “The maʿnā is ṣīda ʿalayhi l-waḥšu fī yawmayni ‘Wild animals were hunted in it [i.e., in a certain place] for two days’” (wa-l-maʿnā ṣīda ʿalayhi l-waḥšu fī yawmayni, Kitāb I, 88.11). In contrast to the taqdīr, the utterance illustrating the maʿnā is not an utterance intended by the speaker as if he were saying it. It is only the explanation of the lafẓ, which is pronounced by the speaker, and it has no grammatical significance. Ibn al-ʾAnbārī (d. 577/1181) calls this type of maʿnā “a taqdīr that occurs in order to explain the sense of the utterance, and not in order to illustrate the ʿāmil [affecting the ʾiʿrāb occurring in the utterance]” (taqdīr li-maʿnā l-kalām lā li-ʿāmilihi).18 In one of his texts, Ibn al-Sarrāj (d. 316/928) calls this type of maʿnā

17 18

Chapter 42, Sībawayhi, Kitāb I, 88.9–90.3. This title has been quoted, translated, and discussed in 3.2 above. See Ibn al-ʾAnbārī, ʾInṣāf 122.13–15.

218

levin

“the explanation” (al-taʾwīl).19 His teacher, al-Mubarrad (d. 285/898), uses the expression fa-ʾinna taʾwīlahu … in the same meaning.20 It should be emphasized that the grammarians refer to the utterance denoting the maʿnā by items belonging to the terminology of the theory of taqdīr, such as yurīdu ‘he intends’.21 The later grammarians Ibn al-ʾAnbārī and Ibn Yaʿīš (d. 643/1245) even sometimes call the maʿnā by the name of al-taqdīr.22 It seems that Ibn al-ʾAnbārī uses the term al-taqdīr in this sense as an abbreviation of the longer expression quoted above, taqdīr li-maʿnā l-kalām lā li-ʿāmilihi. 3.4 Mafʿūl ʿalā l-saʿa In sentences characterized by the phenomenon of ittisāʿ al-kalām, adverbs of time and place, such as al-yawma, al-layla, ʾamsi, ġadan, makān and farsaḫ,23 frequently occur in syntactic positions that are typical of nouns, rather than ẓurūf. The occurrence of these adverbs and of some maṣdar forms in these positions, is a deviation from ordinary speech. When the ẓurūf occur in the sentence in the syntactic position of a noun, they are not regarded as ẓurūf, but as nouns. The grammarians conceive of these adverbs and maṣdar forms as nouns whose syntactic function is that of a direct object, due to the extension of the ordinary way of speech. This extension, they say, takes place because of “the wideness of the language” (saʿat al-kalām). Hence, a mafʿūl that originates in an adverb of time or place or a maṣdar is called “a mafʿūl by deviation from the ordinary way of speech, due to the wideness of the language” (mafʿūl ʿalā lsaʿa, or mafʿūl ʿalā saʿat al-kalām). A mafʿūl ʿalā l-saʿa may occur in one of the following syntactic constructions: (i) as the second part of the construction of ʾiḍāfa. In the verse yā sāriqa llaylati ʾahla l-dāri lit. ‘O you who stole the night from the people of the dwellings of the tribe!’24 (Sībawayhi, Kitāb I, 75.3), the first part of the ʾiḍāfa construction sāriqa l-laylati is the form sāriqa, which is the active participle of the transitive verb saraqa, and the second part is the ẓarf denoting time, al-laylati. In the ordinary way of speech, the second part of an ʾiḍāfa whose first part is a participle, must be a noun, whose function is that of a direct object, as in muʿṭī zaydin, e.g. hāḏā muʿṭī zaydin

19 20 21 22 23 24

See Ibn al-Sarrāj, ʾUṣūl II, 255.11. See al-Mubarrad, Muqtaḍab IV, 331.9–11. See Levin (1997:156.3–19). See Levin (1997:156.20–157.7). These adverbs of time and place belong to the category called al-ẓurūf al-mutamakkina. The maʿnā of this verse is ‘O you who stole in the night from the people of the dwellings of the tribe’.

the phenomenon of ittisāʿ al-kalām in old arabic

219

dirhaman ‘This person gave Zayd a dirham’ (Sībawayhi, Kitāb I, 75.5f.).25 Since in sāriqa l-laylati, al-laylati is a ẓarf, which in ordinary speech cannot occur in this position, the grammarians hold that the Bedouin speakers conceive of it as a noun with the function 0f a direct object (mafʿūl), or more precisely, according to Ibn al-Sarrāj, it is a mafʿūl ʿalā l-saʿa.26 Similar examples are hāḏā muḫriju l-yawmi l-dirhama ‘This person spent today a dirham’, and [hāḏā] ṣāʾidu l-yawmi l-waḥša ‘This person hunted today wild animals’, where al-yawmi is a mafʿūl ʿalā l-saʿa.27 (ii) The ẓarf can be a mafʿūl ʿalā l-saʿa of a transitive verb, as in saraqtu llaylata ʾahla l-dāri lit. ‘I stole the night from the people of the dwellings of the tribe’ (Kitāb I, 75.4). The maʿnā is saraqtu fī l-laylati ʾahla l-dāri. Sībawayhi believes that since al-laylati in yā sāriqa l-laylati ʾahla l-dāri is a mafʿūl, the form al-laylata in saraqtu l-laylata ʾahla l-dāri is also a mafʿūl ʿalā l-saʿa.28 He says that the speaker implies the effect of the verb saraqtu on al-laylata, just as he does so in other constructions characterized by the phenomenon of saʿat al-kalām.29 In the same discussion, Sībawayhi also mentions the Qurʾānic verse bal makru l-layli wa-l-nahāri lit. ‘Nay, the cheating of the night and the day’ (Q. 34/33), where al-layli wa-l-nahāri occurs as nouns characterized by the phenomenon of saʿat al-kalām.30 Sībawayhi notes that the night and the day do not cheat, but the act of cheating takes place during the night and the day. (iii) In Sībawayhi’s view, when a passive verb is assigned as a predicate to a nominative, this nominative is conceived of as a direct object, called almafʿūl. For example, in the utterance ḍuriba zaydun ‘Zayd was hit’, zaydun is a mafʿūl, just as the accusative zaydan in ḍarabtu zaydan ‘I hit Zayd’ is called al-mafʿūl.31 When a verb in the passive voice is assigned as a predicate to one of the ẓurūf mutamakkina, or to a maṣdar, Ibn al-Sarrāj calls these ẓurūf and maṣdar forms “direct objects due to the extension of the ordinary way of speech” (mafʿūlāt ʿalā l-saʿa).32 In referring to the exam25 26

27 28 29 30 31 32

See Sībawayhi, Kitāb I, 75.1–9. It is understood that Sībawayhi conceives of al-laylati as a mafʿūl ʿalā l-saʿa, although he does not use this term (see Kitāb I, 75.1–9). Ibn al-Sarrāj explicitly says that the poet who used in his verse the words sāriqa l-laylati conceived of al-laylati as a mafʿūl bihā ʿalā lsaʿa (ʾUṣūl I, 195.22–196.3). The term mafʿūl ʿalā l-saʿa frequently occurs in Ibn al-Sarrāj, as shown in this article. Sībawayhi, Kitāb I, 75.1–9. Sībawayhi, Kitāb I, 75.1–8. Sībawayhi, Kitāb I, 75.f. Sībawayhi, Kitāb I, 75.12f. See Levin (1979:199.18–200.2). Ibn al-Sarrāj, ʾUṣūl I, 79.9–13. Cf. Ibn al-Ḫaššāb, Murtajil 123.4–10.

220

levin

ples containing some of the adverbs of time and place, he calls them mafʿūl ʿalā l-saʿa.33 Ibn al-Sarrāj believes that when these adverbs and maṣdar forms occur in the position of a noun, they are not ẓurūf anymore, but become nouns whose syntactic function is that of a mafʿūl ʿalā l-saʿa. Sībawayhi does not use the term mafʿūl ʿalā l-saʿa, but he holds the view later expressed by Ibn al-Sarrāj, without explicitly using this term. This point can also be inferred from the text of Kitāb I, 75.1–9, where he compares the phenomenon of ittisāʿ al-kalām in saraqtu l-laylata ʾahla ldāri on the one hand, with ṣīda ʿalayhi yawmāni and wulida lahu sittūna ʿāman on the other. (iv) Maṣdar forms can occur as a mafʿūl ʿalā l-saʿa, as in the example sīra bizaydin sayrun šadīdun ‘Zayd was made to walk an intensive walking’ (Ibn al-Sarrāj, ʾUṣūl I, 79.11) (For this construction see Ibn al-Sarrāj, ʾUṣūl I, 79.9– 12). (v) The bound pronoun suffix as a mafʿūl ʿalā l-saʿa. The following passages illustrate bound pronoun suffixes occurring as a mafʿūl ʿalā l-saʿa. These bound pronoun suffixes originate in combinations of fī + a pronoun suffix: a. “As for [an example of a bound pronoun suffix occurring as a] [mafʿūl] ʿalā l-saʿa [there] is the utterance yawmu l-jumʿati ḍarabtuhu zaydan lit. ‘Friday, I hit it Zayd’. You intend to say: ḍarabtu fīhi ‘I hit in it’, and you join the verb to it [i.e., to zayd]” (wa-ʾammā ʿalā l-saʿa fa-qawluka yawmu l-jumʿati ḍarabtuhu zaydan turīdu ḍarabtu fīhi zaydan fa-ʾawṣalta l-fiʿla ʾilayhi; Mubarrad, Muqtaḍab IV, 332.5 f.). b. “Know that one may treat these [words belonging to the category of] al-ẓurūf al-mutamakkina as nouns and say yawmu l-jumʿati qumtuhu instead of qumtu fīhi ‘I prayed in it’,34 and al-farsaḫu sirtuhu ‘The [distance of the] farsaḫ, I moved along it’, and makānukum jalastuhu ‘Your place, I sat in it’. This is extension of speech.” (wa-ʿlam ʾanna hāḏihi l-ẓurūfa l-mutamakkina yajūzu ʾan tajʿalahā ʾasmāʾan fataqūlu yawmu l-jumʿati qumtuhu fī mawḍiʿ qumtu fīhi, wa-l-farsaḫu sirtuhu wa-makānukum jalastuhu wa-ʾinnamā hāḏā ttisāʿun; Mubarrad, Muqtaḍab IV, 330.7–9)

33 34

Ibn al-Sarrāj, ʾUṣūl II, 294.9–14. I am indebted to my friend and colleague Professor Yohanan Friedmann, who told me that qāma occurs in certain texts in the sense of ‘he prayed’. See also above, n. 40. Cf. Lane (1863–1893: VIII, 2995 B, 9–12): qāma ramaḍāna ‘He passed the nights of Ramaḍān in prayer’.

the phenomenon of ittisāʿ al-kalām in old arabic

c. d. e.

f.

221

yawmu l-jumʿati sirtuhu ‘Friday, I walked in it’ (Sībawayhi, Kitāb I, 33.14). The maʿnā is yawmu l-jumʿati sirtu fīhi. yawmu l-jumʿati ṣumtuhu ‘Friday, I fasted in it’ (Sībawayhi, Kitāb I, 33.14). The maʿnā is yawmu l-jumʿati ṣumtu fīhi. ʾammā l-laylatu fa-ʾanta sāriquhā zaydan lit. ‘As for the night, you stole it from Zayd’ (Ibn al-Sarrāj, ʾUṣūl I, 196.1f.). The maʿnā is ʾammā l-laylatu fa-ʾanta sāriqun fīhā zaydan ‘As for the night, you stole in it from Zayd’. In the verse wa-yawmin šahidnāhu sulayman wa-ʿāmiran … “Many a day of war we saw [the people of the tribes] Sulaym and ʿĀmir …”.35 The form šahidnāhu occurs instead of šahidnā fīhi. The pronoun -hu in šahidnāhu is a mafʿūl ʿalā l-saʿa.

3.4.1 Mafʿūl ʿalā l-saʿa and ẓarf In Sībawayhi’s view, adverbs of time and place, like al-yawm, al-layla, ʾamsi, ġadan, makān and farsaḫ, which belong to the category of the ẓurūf mutamakkina, take the accusative because of the effect of the verb preceding it, as in ḏahabtu ʾamsi36 ‘I went away yesterday’ (Sībawayhi, Kitāb I, 11.15) and saʾaḏhabu ġadan ‘I’ll go away tomorrow’ (Sībawayhi, Kitāb I 11.15 f.).37 In referring to the above examples, Sībawayhi says: “If you wish, you don’t have to conceive of these two nouns as a ẓarf ” ( fa-ʾin šiʾta lam tajʿalhumā ẓarfan),38 i.e., “If you wish you can conceive of ʾamsi and ġadan not as a ẓarf, [but as a mafʿūl ʿalā l-saʿa]”, i.e., you can assume that ʾamsi and ġadan take the accusative because of the ʿamal of the verb, not as a ẓarf, but as a mafʿūl ʿalā l-saʿa. The following texts from the Kitāb are evidence that this explanation of the words fa-ʾin šiʾta etc. is correct. In Kitāb I, 14.13f., Sībawayhi compares two of his examples characterized by the phenomenon of ittisāʿ al-kalām: wa-saraqtu ʿabdallāhi l-ṯawba l-laylata ‘And I stole from ʿAbdallāh the garment in the night’. When saying this, you don’t conceive of [the word al-laylata] as a ẓarf, just as when saying yā sāriqa l-laylati zaydan l-ṯawba ‘O, you who stole in the night the garment from Zayd’, you don’t conceive of it [i.e., of al-laylati]39 as a ẓarf [but as a mafʿūl ʿalā l-saʿa] (wa-saraqtu

35 36 37 38 39

See Sībawayhi, Kitāb I, 75.20. ʾamsi is virtually in the accusative. See Sībawayhi, Kitāb I, 11.12–20. See Sībawayhi, Kitāb I, 11.15f. It is impossible to conceive of al-laylati as a ẓarf, because a ẓarf cannot occur in the position of the second part of ʾiḍāfa.

222

levin

ʿabdallāhi l-ṯawba l-laylata lā tajʿaluhu ẓarfan wa-lākin kamā taqūlu yā sāriqa l-laylati zaydan al-ṯawba lam tajʿalhā ẓarfan). It is evident that since Sībawayhi does not conceive of al-laylati and al-laylata as a ẓarf, he conceives of them as what is called in later terminology a mafʿūl ʿalā l-saʿa, which takes the accusative because of the ʿamal of the verb. The expression lā tajʿaluhu ẓarfan, in the above sense, occurs also in Sībawayhi’s discussions of the ʿamal of passive verbs and participles.40 In the above examples, ʾamsi and ġadan occur as a ẓarf zamān, i.e., as an adverb of time, which in the ordinary way of speech takes the accusative because of the effect of the verbs ḏahabtu and sa-ʾaḏhabu.41 By contrast, the text of Sībawayhi (Kitāb I, 253.10–14) is evidence that al-Ḫalīl believed that ʾamsi in laqītahu ʾamsi is a contracted form of bi-l-ʾamsi, as in his example laqītahu bi-l-ʾamsi ‘You met him yesterday’. The Bedouin, he says, dropped the particle bi- and the article -l “in order to make it easier for the tongue” (taḫfīfan ʿalā l-lisān). The dropping of one of the ḥurūf al-jarr in order to make it easier for the speaker to pronounce an utterance is one of the characteristics of the phenomenon of saʿat al-kalām.42 Hence, it is understood that in al-Ḫalīl’s view, the form ʾamsi is regarded as an abbreviation of the combination bi-l-ʾamsi.43 According to this view, in the above example ʾamsi is not a ẓarf which virtually takes the accusative because of the ʿamal of the verb, but it is a mafʿūl ʿalā l-saʿa. This conclusion is an argument supporting the above explanation of the expression fa-ʾin šiʾta lam tajʿalhumā ẓarfan.44 3.4.2 The ẓarf as a mubtadaʾ In the following examples the mubtadaʾ is a ẓarf. In nahāruhu ṣāʾimun walayluhu qāʾimun lit. ‘His day fasts and his night prays’ (Sībawayhi, Kitāb I, 69.5f.),45 the maʿnā is ʾinnaka ṣāʾimun fī l-nahāri wa-qāʾimun fī l-layli ‘You fast

40 41 42 43 44

45

Sībawayhi, Kitāb I, 14.23–15.1; 15.12–15. Sībawayhi, Kitāb I, 11.12–16. See Ibn al-Sarrāj, ʾUṣūl I, 171.4–7, 18–22. It should be emphasized that Sībawayhi does not accept al-Ḫalīl’s explanation of the form ʾamsi (Kitāb I, 254, 1f.). In my second article on Sībawayhi’s terminology (Levin 1979), I attempted to explain the sense of Sībawayhi’s expression fa-ʾin šiʾta lam tajʿalhumā ẓarfan, but my explanation there is incorrect (see Levin 1979:195f., n. 11). The correct sense of this expression is given here. For this sense of qāma l-laylu see above, n. 32.

the phenomenon of ittisāʿ al-kalām in old arabic

223

at day time and pray in the night’.46 In a verse of Jarīr the following example occurs: … wa-nimti wa-mā laylu l-maṭiyyi bi-nāʾimin lit. ‘And you were asleep, while the night of the riding beast was not asleep’ (Sībawayhi, Kitāb I, 69.7). The maʿnā is: ‘And you were asleep, while the riding beast was not asleep at night’.

Bibliographical References A

Primary Sources

Ibn al-ʾAnbārī, ʾInṣāf = Kamāl al-Dīn ʾAbū l-Barakāt ʿAbd al-Raḥmān ibn Muḥammad Ibn al-ʾAnbārī, Kitāb al-ʾinṣāf fī masāʾil al-ḫilāf bayna l-naḥwiyyīna l-Baṣriyyīna wa-lKūfiyyīn. Ed. by Gotthold Weil. Leiden: E.J. Brill, 1913. Ibn Ḫaššāb, Murtajil = ʾAbū Muḥammad ʿAbdallāh ibn ʾAḥmad Ibn al-Ḫaššāb, al-Murtajil. Ed. by ʿAlī Ḥaydar. Damascus, 1972. Ibn al-Sarrāj, ʾUṣūl = ʾAbū Bakr Muḥammad ibn Sahl Ibn al-Sarrāj al-Naḥwī al-Baġdādī, Kitāb al-ʾuṣūl fī l-naḥw. Ed. by ʿAbd al-Ḥusayn al-Fatlī. 3 vols. Beirut: Muʾassasat alRisāla, 1987. Ibn Yaʿīš, Šarḥ = Muwaffaq al-Dīn ʾAbū l-Baqāʾ Yaʿīš Ibn Yaʿīš, Šarḥ al-Mufaṣṣal. Ed. by Gustav Jahn, Ibn Yaʿîś’ Commentar zu Zamachśarî’s Mufaṣṣal. 2 vols. Leipzig, 1882– 1888. Jurjānī, Muqtaṣid = ʾAbū Bakr ʿAbd al-Qāhir ibn ʿAbd al-Raḥmān al-Jurjānī, Kitāb almuqtaṣid fī šarḥ al-ʾĪḍāḥ. Ed. by Kāẓim Baḥr al-Murjān. 2 vols. Baghdad: Wizārat al-Ṯaqāfa wa-l-ʾIʿlām, 1982. Mubarrad, Muqtaḍab = ʾAbū l-ʿAbbās Muḥammad ibn Yazīd al-Mubarrad, Kitāb almuqtaḍab. Ed. by Muḥammad ʿAbd al-Ḫāliq ʿUḍayma. 4 vols. Cairo: Dār al-Taḥrīr, 1385–1388/1965–1968. Sībawayhi, Kitāb = ʾAbū Bišr ʿAmr ibn ʿUṯmān Sībawayhi, al-Kitāb. Ed. by Hartwig Derenbourg, Le livre de Sîbawaihi: Traité de grammaire arabe. 2 vols. Paris: Imprimerie nationale, 1881–1889.

B

Secondary Sources

Baalbaki, Ramzi. 1988. “A contribution to the study of technical terms in early Arabic grammar: The term aṣl in Sībawayhi’s Kitāb”. In Memoriam Thomas Muir Johnstone 1924–1983. Professor of Arabic in the University of London 1970–82, ed. by A.K. Irvine, R.B. Serjeant and G. Rex Smith, 163–177. Essex: Longman Group.

46

Compare Ibn al-Sarrāj, ʾUṣūl II, 255.12f.

224

levin

Baalbaki, Ramzi. 2008. The legacy of the Kitāb: Sībawayhi’s analytical methods within the context of the Arabic grammatical theory. Leiden: E.J. Brill. Dayyeh, Hanadi. 2015. “Ittisāʿ in Sībawayhi’s Kitāb: A semantic ʿilla for disorders in meanng and form”. The foundations of Arabic linguistics. II. Kitāb Sībawayhi: Interpretation and transmission, ed. by Amal Elesha Marogy and Kees Versteegh, 66–80. Leiden: E.J. Brill. Lane, Edward William. 1863–1893. An Arabic-English lexicon. London: Williams and Norgate. 8 vols. (Repr., Beirut: Librairie du Liban.) Levin, Aryeh. 1979. “The meaning of taʿaddā al-fiʿl ilā in Sībawayh’s al-Kitāb”. Studia Orientalia Memoriae D.H. Baneth Dedicata, 193–210. Jerusalem: The Magnes Press. Levin, Aryeh. 1997. “The theory of al-taqdīr and its terminology”. Jerusalem Studies in Arabic and Islam 21.142–166. Versteegh, Kees. 1990. “Freedom of the speaker: The term ittisāʿ and related notions in Arabic grammar”. Studies in the history of Arabic grammar, II, ed. by Michael G. Carter and Kees Versteegh, 281–293. Amsterdam and Philadelphia: J. Benjamins.

Which Verbal Nouns Can Function as Adverbial Accusatives of State or Condition (ḥāl) according to Sībawayhi and Later Grammarians? Arik Sadan

Ḥāl is an adverbial accusative of state or condition that describes the state of one of the nouns in the sentence.1 The term ḥāl lit. ‘state’ derives from Sībawayhi’s usage in describing this phenomenon, for example “a state in which the matter, or action, occurred” (ḥālun waqaʿa fīhi l-ʾamru);2 “states in which the matters, or actions, occur” (ʾaḥwālun taqaʿu fīhā l-ʾumūru).3 Of the parts of speech which can function as adverbial accusatives of state or condition, the first to come to mind is the participle, as in jāʾa zaydun rākiban ‘Zayd came riding’.4 More rarely, adjectives and nouns can also function as adverbial accusatives of state or condition, as in the examples yuqawwamu bi-l-ṯiqāfi l-ʿūdu ladnan ‘a piece of wood can be straightened with the ṯiqāf [a kind of tool], while it is soft’ and ṭalaʿa l-qamaru badran ‘the moon rose full’, respectively. Lastly, some verbal nouns can appear as adverbial accusatives of state or condition, in which case they are usually perceived and explained as equivalent in meaning to participles, as in ṭalaʿa baġtatan ‘he came into view suddenly’, where the accusative verbal noun baġtatan is usually perceived and explained as equivalent in meaning to the participle bāġitan. In chapter 82 of his Kitāb Sībawayhi begins a discussion of verbal nouns that take the accusative case due to their functioning as ḥāl. The chapter opens with the title and relevant examples: This is the chapter of the verbal nouns that take the accusative case, because they are ḥāl—a state, in which the matter occurred—and it received the accusative case, because the matter happened in them, as in the examples qataltuhu ṣabran ‘I killed him in captivity’, laqītuhu fujāʾatan

1 2 3 4

On this phenomenon see Bernards (2007) and Carter (2002). See Sībawayhi, Kitāb I, 155.18. See Sībawayhi, Kitāb I, 169.6. All examples and translations in this paragraph are taken from Wright (1997:II, 112–115, § 44 c).

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_014

226

sadan

wa-mufājaʾatan ‘I met him suddenly/unexpectedly’, [laqītuhu] kifāḥan wa-mukāfaḥatan ‘[I met him] battling/fighting’, laqītuhu ʿiyānan ‘I met him eye to eye’, kallamtuhu mušāfahatan ‘I spoke to him mouth to mouth’, ʾataytuhu rakḍan ‘I came to him rushing’, [ʾataytuhu] ʿadwan ‘[I came to him] running’, [ʾataytuhu] mašyan ‘[I came to him] walking’ and ʾaḫaḏtu ḏālika ʿanhu samʿan wa-samāʿan ‘I received that from him by hearsay, or hearing it from him’ (hāḏā bābu mā yantaṣibu mina l-maṣādiri li-ʾannahu ḥālun waqaʿa fīhi l-ʾamru fa-ntaṣaba li-ʾannahu mawqūʿun5 fīhi l-ʾamru wa-ḏālika qawluka qataltuhu ṣabran wa-laqītuhu fujāʾatan wa-mufājaʾatan wa-kifāḥan wa-mukāfaḥatan wa-laqītuhu ʿiyānan wa-kallamtuhu mušāfahatan wa-ʾataytuhu rakḍan wa-ʿadwan wa-mašyan wa-ʾaḫaḏtu ḏālika ʿanhu samʿan wa-samāʿan).6 Following these twelve examples of verbal nouns functioning as ḥāl, Sībawayhi says that not all verbal nouns can function as ḥāl, even if they seem similar to one of the above-mentioned examples, because for it to function as ḥāl a verbal noun must take the place of an active participle, and this is not always possible, as in the two impermissible examples *ʾatānā surʿatan and *ʾatānā rujlatan: And not every verbal noun, even if it were by way of an analogy like the previous examples [lit. ‘what has been before’] of this chapter, can be put in this position [i.e., of a ḥāl], because the verbal noun here takes the place of an active participle, when it is a ḥāl. Don’t you see that [the two impermissible examples] *ʾatānā surʿatan ‘he came to us quickly’ and *ʾatānā rujlatan ‘he came to us on foot’ are not appropriate (wa-laysa kullu maṣdarin wa-ʾin kāna fī l-qiyāsi miṯla mā maḍā min hāḏā l-bābi yūḍaʿu hāḏā l-mawḍiʿa li-ʾanna l-maṣdara hāhunā fī mawḍiʿi fāʿilin ʾiḏā kāna ḥālan ʾa-lā tarā ʾannahu lā yaḥsunu ʾatānā surʿatan wa-lā ʾatānā rujlatan).7 About a century after Sībawayhi (d. 177/793?), al-Mubarrad (d. 285/898) gave a different analysis of the issue. Like Sībawayhi, he does not permit all verbal nouns to function freely as ḥāl, but he seems to permit more cases and also provides a tool for verifying whether or not an utterance in which a verbal noun functions as ḥāl is permissible. After quoting some permissible examples, identical or similar to those quoted by Sībawayhi, al-Mubarrad explains that in each case the verbal noun can function as ḥāl because it indicates the 5 According to another version: mawqaʿun. 6 See Sībawayhi, Kitāb I, 155.18–156, 1. 7 See Sībawayhi, Kitāb I, 156.1–4.

which verbal nouns can function as ḥāl?

227

type or manner of action denoted by the main verb. In jiʾtuhu mašyan ‘I came to him walking’, for instance, the action of mašy ‘walking’ is a certain type or manner of majīʾ ‘coming’, which the main verb jiʾtu conveys.8 In contrast, the sentence *jiʾtuhu ʾiʿṭāʾan ‘I came to him giving [something]’, is impermissible, because the action of ʾiʿṭāʾ ‘giving’ is not a certain type or manner of majīʾ ‘coming’, which the main verb jiʾtu conveys.9 Al-Mubarrad provides two theoretical explanations for this distinction. His first explanation is identical to that of Sībawayhi, namely that the accusative verbal noun is in the position of an active participle; but then he adds a second explanation, that the verbal noun is in fact the mafʿūl muṭlaq10 of an omitted verb: jiʾtuhu mašyan, the first example mentioned above, is explained first as jiʾtuhu māšiyan and then as jiʾtuhu ʾamšī mašyan. From the preceding discussion of Sībawayhi’s and al-Mubarrad’s views in the matter, we can see that there are two main points of dispute: (i) Are the examples of accusative verbal nouns that function as ḥāl samāʿiyya or qiyāsiyya? In other words, must we restrict ourselves to examples heard by al-ʿArab and refrain from constructing new examples by way of analogy (qiyās), as Sībawayhi maintains, or, on the contrary, should existing examples serve as templates for the creation of new constructions by qiyās, as al-Mubarrad claims? (ii) What is the grammatical explanation for the verbal nouns functioning as ḥāl? What is the taqdīr structure? Is it a participle, as Sībawayhi says, or a mafʿūl muṭlaq of an omitted verb, as al-Mubarrad believes? For example, if we ask about the taqdīr structure of ʾataytuhu mašyan, is it ʾataytuhu māšiyan or ʾataytuhu ʾamšī mašyan? For the purpose of this paper I examined the opinions of grammarians from the time of Sībawayhi until the end of the 7th/14th century. Of the thirty-six grammarians examined (in more than fifty works, since many grammarians composed more than one work), sixteen explicitly discuss the possibility of a verbal noun functioning as ḥāl,11 whereas twenty do not.12 Note that in each list 8 9 10 11

12

See Mubarrad, Muqtaḍab III, 234.5. See Mubarrad, Muqtaḍab III, 234.7. On this phenomenon see Levin (1991), and above, Kasher’s contribution to the present volume. These are Sībawayhi (d. ca. 180/796), al-Mubarrad (d. 286/900 or 285/899), ʾAbū Saʿīd al-Sīrāfī (d. 368/979), al-Fārisī (d. 377/987), al-Ṣaymarī (d. in the 4th/10th century), alŠantamarī (d. 476/1083), al-Zamaḫšarī (d. 538/1144), Ibn al-Dahhān (d. 569/1174), al-Yamanī (d. 599/1202), Ibn Yaʿīš (d. 643/1245), al-Šalawbīnī (d. 645/1247), Ibn al-Ḥājib (d. 646/1249), al-Zamalkānī (d. 651/1253), Ibn Mālik (d. 672/1274), Ibn al-Nāẓim (d. 686/1287), al-ʾAstarābāḏī (d. 684/1285 or 686/1288). These are Ṯaʿlab (d. 291/904), Ibn al-Sarrāj (d. 316/929), Ibn Šuqayr (d. 317/929), al-Zajjājī

228

sadan

there are more and less prominent grammarians, for example in the former we find Sībawayhi, al-Mubarrad, al-Zamaḫšarī and al-ʾAstarābāḏī while the latter contains the prominent grammarians Ibn al-Sarrāj, al-Zajjājī, Ibn Jinnī and alJurjānī. In any case, it seems rather striking that so many grammarians do not even mention the possibility of a verbal noun functioning as ḥāl, although the construction is not at all rare. A perusal of the grammarians’ views shows that although many adhere to Sībawayhi’s view, some adopt al-Mubarrad’s, with some variation. ʾAbū Saʿīd al-Sīrāfī (d. 368/978) in his commentary on Sībawayhi’s Kitāb says that in all of Sībawayhi’s examples of a verbal noun functioning as ḥāl, it substitutes for an active or passive participle. He then explicitly says that these examples, of actual usage by Bedouins, must not be extended by analogy (qiyās). The reason why analogical extension is rejected, says al-Sīrāfī, is that in these examples there was a replacement, that is, the verbal noun in the accusative case replaced the part of speech that normally functions as ḥāl, namely the active or passive participle: This [i.e. the verbal noun functioning as ḥāl] is not a regular analogy, but rather should be used in the instances that the Bedouins used, because it is a thing that was used in a construction that belonged to something else (wa-laysa ḏālika bi-qiyāsin muṭṭaridin wa-ʾinnamā yustaʿmalu fīmā staʿmalathu l-ʿArabu li-ʾannahu šayʾun wuḍiʿa fī mawḍiʿi ġayrihi).13 ʾAbū Saʿīd al-Sīrāfī then describes the opinion of ʾAbū l-ʿAbbās, that is alMubarrad, which I have introduced above, based on his Muqtaḍab.14 He rejects al-Mubarrad’s view that jiʾtuhu mašyan has a taqdīr of ʾamšī mašyan, that is, that the ḥāl originates as the mafʿūl muṭlaq of a missing verb ʾamšī. Rather, it should be understood as a mafʿūl muṭlaq of the sentence’s main verb, since the accusative verbal noun reflects a manner of the action denoted by the main verb. Thus with reference to the example ʾatānā zaydun mašyan, al-Sīrāfī says that the main verb ʾatā can be perceived as having the meaning maḍā ʾilayya ‘he

13 14

(d. between 337–340/948–950), al-Rummānī (d. 384/994), Ibn Jinnī (d. 392/1002), alHarawī (d. in the 5th/11th century), al-Ḍarīr (d. in the 5th/11th century), Ibn Bābašāḏ (d. 469/1077), al-Jurjānī (d. 471/1078), al-Mujāšiʿī (d. 479/1086), al-Ḥarīrī (d. 516/1122), alBaṭalyawsī (d. 521/1127), al-Šarīf al-Kūfī (d. 539/1145), Ibn al-Ḫaššāb (d. 567/1172), Ibn alʾAnbārī (d. 577/1181), Ibn Barrī (d. 582/1187), al-Jazūlī (d. 607/1210), ʾAbū l-Baqāʾ al-ʿUkbarī (d. ca. 616/1219), Ibn Muʿṭī (d. 628/1231). See Sīrāfī, Šarḥ II, 257, last two lines. Cf. Šantamarī, Nukat I, 531, last two lines. See Sīrāfī, Šarḥ II, 258.2–7. Cf. Šantamarī, Nukat I, 532.

which verbal nouns can function as ḥāl?

229

turned to me’, so that mašyan expresses a manner of carrying out an action.15 ʾAbū Saʿīd al-Sīrāfī thus partly adopts Sībawayhi’s view by rejecting qiyās, but also partly accepts al-Mubarrad’s view according to which the accusative verbal noun is mafʿūl muṭlaq. Ibn Yaʿīš supports Sībawayhi’s opinion and rejects the views of both alMubarrad and ʾAbū Saʿīd al-Sīrāfī.16 The latter cannot be right, says Ibn Yaʿīš, because the verbal noun must be indefinite. Because of the ungrammaticality of *ʾatānā Zayduni l-mašya, which according to al-Mubarrad could be derived from the grammatical ʾatānā Zaydun yamšī l-mašya, Ibn Yaʿīš concludes that the correct interpretation is Sībawayhi’s.17 ʾAbū ʿAlī al-Fārisī does not refer to this topic in any of his books except alMasāʾil al-manṯūra. There he devotes chapter 8 to verbal nouns functioning as ḥāl and describes the two opinions on it and the issue of qiyās: according to the first opinion, only existing examples of what Bedouins were heard to say are valid and one should not extend the construction by qiyās; the second opinion allows creating new examples by qiyās.18 ʾAbū ʿAlī al-Fārisī does not attribute these two opinions to any specific grammarians, nor does he express his support for either. As we have seen, the first opinion is Sībawayhi’s and al-Mubarrad adopts the second. In the same manner al-Ṣaymarī mentions these two opinions without expressing his support of either, but unlike al-Fārisī, he explicitly attributes them to Sībawayhi and al-Mubarrad, respectively.19 Some grammarians merely mention briefly that a verbal noun can function as ḥāl.20 Others add that similar examples “cannot be created by analogy” (lā yuqāsu ʿalayhi), in line with Sībawayhi’s view.21 Still others add, again in agreement with Sībawayhi, that this verbal noun replaces a participle, without

15

16 17 18 19

20 21

See Sīrāfī, Šarḥ II, 258.8–19. This also seems to be al-Suhaylī’s view; see Suhaylī, Natāʾij 183.1f.; 303.4–2 from the end. Al-Zamalkānī, on the contrary, rejects al-Sīrāfī’s view, saying: “The verbal noun which clarifies the type of the action [i.e. al-mafʿūl al-muṭlaq li-l-nawʿ] is not dependent on the heard [examples] nor does it contradict the analogy” ( fa-ʾinna l-maṣdara l-mubayyina li-l-nawʿi laysa mawqūfan ʿalā l-samāʿi wa-lā muḫālifan li-l-qiyāsi). (Zamalkānī, Mufḍil, 299.3; for the whole discussion see ibid., 297.3–301.4). See Ibn Yaʿīš, Šarḥ II, 12f. See Ibn Yaʿīš, Šarḥ II, 12.6–4 from the end. See Fārisī, Manṯūra, 16 (al-masʾala 8). See Ṣaymarī, Tabṣira I, 299.9–300.5. See also Zamaḫšarī, Mufaṣṣal 28.10 f.; Ibn al-Ḥājib, ʾĪḍāḥ, 334.12–335.9, where the author says that the majority of grammarians view the ḥāl examples with verbal nouns as “heard [examples] not made by analogy” (samāʿiyya lā qiyāsiyya). See, for example, Yamanī, Kašf I, 480.7f.; 483.7f. See, for example, Šalawbīnī, Tawṭiʾa, 212.6–4 from the end.

230

sadan

elaborating further.22 Ibn Mālik is among the few grammarians who mention both Sībawayhi’s and al-Mubarrad’s views, and agrees with the latter to some extent: following his presentation of al-Mubarrad’s view he says: “This [i.e. alMubarrad’s view in favor of qiyās] is not inconceivable” (wa-laysa ḏālika bibaʿīdin).23 However, in another book of his, Šarḥ al-tashīl, he harshly criticizes al-Mubarrad.24 A neat summary of the issue, in which Sībawayhi’s approach is defended and al-Mubarrad’s is rejected, appears in al-ʾAstarābāḏī’s Šarḥ alKāfiya.25 To sum up, an examination of grammarians’ writings on the possibility of a verbal noun functioning as ḥāl, reveals that many do not refer to this issue at all. Many of those who do mention and discuss the topic agree with Sībawayhi’s opinion that one should regard the existing examples as samāʿiyya and not qiyāsiyya and that the accusative verbal noun replaces a participle. However, other views on the topic also exist, in particular that of al-Mubarrad, whose opinion is either mentioned neutrally by later grammarians (for example, by ʾAbū ʿAlī al-Fārisī), modified (for example, by ʾAbū Saʿīd al-Sīrāfī) or criticized (for example, by Ibn Yaʿīš).

Bibliographical References A

Primary Sources

ʾAstarābāḏī, Šarḥ = Raḍī l-Dīn Muḥammad ibn al-Ḥasan al-ʾAstarābāḏī, Šarḥ al-Raḍī li-Kāfiyat Ibn al-Ḥājib. Ed. by Ḥusayn ibn Muḥammad ibn ʾIbrāhīm al-Ḥifẓī. Riyad: Jāmiʿat al-ʾImām Muḥammad ibn Saʿūd al-ʾIslāmiyya, 1414/1993. Fārisī, Manṯūra = ʾAbū ʿAlī al-Ḥasan ibn ʾAḥmad ibn ʿAbd al-Ġaffār al-Fārisī, al-Masāʾil al-manṯūra. Ed. by Muṣṭafā al-Ḥudarī. Damascus: Majmaʿ al-Luġa al-ʿArabiyya, 1406/ 1986. Ibn al-Dahhān, Fuṣūl = ʾAbū Muḥammad ibn al-Mubārak Ibn al-Dahhān, al-Fuṣūl fī lʿarabiyya. Ed. by Fāʾiz Fāris. Beirut: Muʾassasat al-Risāla, 1409/1988. Ibn al-Dahhān, Šarḥ = ʾAbū Muḥammad ibn al-Mubārak Ibn al-Dahhān, Šarḥ al-durūs fī l-naḥw. Ed. by ʾIbrāhīm Muḥammad ʾAḥmad al-ʾIdkāwī. Cairo: Maṭbaʿat al-ʾAmāna, 1411/1991.

22 23 24 25

See, for example, Ibn al-Dahhān, Fuṣūl, 25.3. In his much more detailed book there is no mention of the issue: Ibn al-Dahhān, Šarḥ, 249–261. See Ibn Mālik, Šarḥ al-Kāfiya I, 233.11. For the whole discussion, see ibid. I, 233.6–11. See Ibn Mālik, Šarḥ al-tashīl II, 328.14–18. For the whole discussion, see ibid. II, 327–331. See also Ibn al-Nāẓim, Šarḥ, 231.14–232.3. See ʾAstarābāḏī, Šarḥ, 671f.

which verbal nouns can function as ḥāl?

231

Ibn al-Ḥājib, ʾĪḍāḥ = Jamāl al-Dīn ʾAbū ʿAmr ʿUṯmān ibn ʿUmar ibn ʾAbī Bakr al-Mālikī Ibn al-Ḥājib, al-ʾĪḍāḥ fī šarḥ al-Mufaṣṣal. Ed. by Mūsā Bannāy al-ʿAlīlī. Baghdad: Maṭbaʿat al-ʿĀnī, 1402/1982. Ibn Mālik, Šarḥ al-Kāfiya = ʾAbū ʿAbdallāh Jamāl al-Dīn Muḥammad ibn ʿAbdallāh Ibn Mālik al-Ṭāʾī al-Jayyānī, Šarḥ al-Kāfiya al-šāfiya. Ed. by ʿAlī Muḥammad Muʿawwaḍ and ʿĀdil ʾAḥmad ʿAbd al-Mawjūd. 2 vols. Beirut: Dār al-Kutub al-ʿIlmiyya, 1420/2000. Ibn Mālik, Šarḥ al-Tashīl = ʾAbū ʿAbdallāh Jamāl al-Dīn Muḥammad ibn ʿAbdallāh Ibn Mālik al-Ṭāʾī al-Jayyānī, Šarḥ al-tashīl. Ed. by ʿAbd al-Raḥmān al-Sayyid and Muḥammad Badawī al-Maḫtūn. 4 vols. Cairo: Hajr, 1410/1990. Ibn al-Nāẓim, Šarḥ = ʾAbū ʿAbdallāh Badr al-Dīn Muḥammad Ibn al-Nāẓim, Šarḥ Ibn al-Nāẓim ʿalā ʾAlfiyyat Ibn Mālik. Ed. By Muḥammad Bāsil ʿUyūn al-Sūd. Beirut: Dār al-Kutub al-ʿIlmiyya, 1420/2000. Ibn Yaʿīš, Šarḥ = Muwaffaq al-Dīn ʾAbū l-Baqāʾ Yaʿīš ibn ʿAlī Ibn Yaʿīš al-Ḥalabī, Šarḥ al-Mufaṣṣal li-l-Zamaḫšarī. Ed. by ʾImīl Badīʿ Yaʿqūb. 6 vols. Beirut: Dār al-Kutub alʿIlmiyya, 1422/2001. Mubarrad, Muqtaḍab = ʾAbū l-ʿAbbās Muḥammad ibn Yazīd ibn ʿAbd al-ʾAkbar alṮumālī al-ʾAzdī al-Mubarrad, al-Muqtaḍab. Ed. by Muḥammad ʿAbd al-Ḫāliq ʿUḍayma. 4 vols. Cairo: Muʾassasat Dār al-Taḥrīr, 1385–1388/1965–1968. Šalawbīnī, Tawṭiʾa = ʾAbū ʿAlī ʿUmar ibn Muḥammad ibn ʿUmar ibn ʿAbdallāh al-Šalawbīnī al-ʾIšbīlī, al-Tawṭiʾa. Ed. by Yūsuf ʾAḥmad al-Muṭawwiʿ. Cairo: Dār al-Turāṯ alʿArabī, 1401/1981. Šantamarī, Nukat = ʾAbū l-Ḥajjāj Yūsuf ibn Sulaymān al-ʾAndalusī al-Naḥwī al-Šantamarī, al-Nukat fī tafsīr Kitāb Sībawayhi. Ed. by Rašīd Balḥabīb. 3 vols. Al-Maḥmadiyya: Maṭbaʿat Faḍāla, 1420/1999. Ṣaymarī, Tabṣira = ʾAbū Muḥammad ʿAbdallāh ibn ʿAlī ibn ʾIsḥāq al-Ṣaymarī, al-Tabṣira wa-l-taḏkira. Ed. by Fatḥī ʾAḥmad Muṣṭafā ʿAlī al-Dīn. 2 vols. Damascus: Dār al-Fikr, 1402/1982. Sībawayhi, Kitāb = ʾAbū Bišr ʿAmr ibn ʿUṯmān Sībawayhi, al-Kitāb. Ed. by Hartwig Derenbourg, Le livre de Sībawaihi, traité de grammaire arabe. 2 vols. Paris: Imprimerie nationale, 1881–1889. Sīrāfī, Šarḥ = ʾAbū Saʿīd al-Ḥasan ibn ʿAbdallāh ibn al-Marzubān al-Sīrāfī, Šarḥ Kitāb Sībawayhi. Ed. by ʾAḥmad Ḥasan Mahdalī and ʿAlī Sayyid ʿAlī. 5 vols. Beirut: al-Kutub al-ʿIlmiyya, 2008. Suhaylī, Natāʾij =ʾAbū l-Qāsim ʿAbd al-Raḥmān ibn ʿAbdallāh al-Suhaylī, Natāʾij al-fikr fī l-naḥw. Ed. by ʿĀdil ʾAḥmad ʿAbd al-Mawjūd and ʿAlī Muḥammad Muʿawwaḍ. Beirut: Dār al-Kutub al-ʿIlmiyya, 1412/1992. Yamanī, Kašf = ʿAlī ibn Sulaymān ibn ʾAsʿad ibn ʾIbrāhīm ibn ʿAlī ibn Tamīm al-Yamanī al-Ḥāriṯī, Kašf al-muškil fī l-naḥw. Ed. by Hādī ʿAṭiyya Maṭar. 2 vols. Baghdad: Maṭbaʿat al-ʾIršād, 1404/1984. Zamaḫšarī, Mufaṣṣal = ʾAbū l-Qāsim Maḥmūd ibn ʿUmar al-Zamaḫšarī, al-Mufaṣṣal

232

sadan

fī ṣunʿat al-ʾiʿrāb. Ed. by Jens Peter Broch. 2nd edition. Christiania: Libraria P.T. Mallingii, 1879. Zamalkānī, Mufḍil = Kamāl al-Dīn ʾAbū l-Makārim al-Zamalkānī, al-Mufḍil ʿalā l-mufaḍḍal fī dirāyat al-Mufaṣṣal. Ed. by ʿAbd al-ʿAzīz ibn Muḥammad al-Ḥikmī. Mecca: Jāmiʿat ʾUmm al-Qurā, 1426/2005.

B

Secondary Sources

Bernards, Monique. 2007. “Ḥāl”. Encyclopedia of Arabic language and linguistics, ed. by Mushira Eid, Alaa Elgibali, Kees Versteegh, Manfred Woidich, and Andrzej Zaborski, II, 224–228. Leiden: E.J. Brill. Carter, Michael G. 2002. “Patterns of reasoning: Sībawayhi’s analysis of the ḥāl”. Proceedings of the 20th Congress of the Union Européenne des Arabisants et Islamisants, Budapest, 10–17 September 2000, I, ed. by Kinga Dévényi, 3–15. Budapest. Levin, Aryeh. 1991. “What is meant by al-mafʿūl al-muṭlaq?”. Semitic studies in honour of Wolf Leslau on the occasion of his eighty-fifth birthday, ed. by Alan S. Kaye, II, 917– 926. Wiesbaden: O. Harrassowitz. (Repr. Aryeh Levin, Arabic linguistic thought and dialectology, art. XI. Jerusalem: The Hebrew University, 1998.) Wright, William. 1997. A grammar of the Arabic language. 2 vols. Cambridge: Cambridge University Press.

What is Definiteness in Arabic? Focusing on Proper Nouns for Genera and ʾasmāʾ mubhama ‘Ambiguous Nouns’ Haruko Sakaedani

1

Introduction1

This paper analyzes the changes in the Arabic grammarians’ descriptions of definite noun phrases. First, we will begin with the changes in the list of items included in the class of definite nouns. Definite noun phrases in Arabic traditional grammar include pronouns, proper nouns, demonstratives, relatives, nouns with al-, and nouns followed by another definite. But, in this classification, the discussion of the following two points is particularly relevant in this connection: referentiality, mainly with respect to proper nouns for genera; and the treatment of the ʾasmāʾ mubhama ‘ambiguous nouns’. Secondly, we will examine changes in the description of the hierarchy of the items included in the class of definite noun phrases. Pronouns are nowadays regarded as the most definite, and nouns with al- are regarded as the least definite. Although this hierarchy is defined by their formal aspect, i.e., their modifying relations, it is also supported by the cognitive criteria viewpoint. 1.1 Referentiality To begin with, we inquire into referentiality, which presents a problem when we are dealing with proper nouns for genera. According to Tanaka (1981:16), noun phrases are classified as shown in Figure 1. First, noun phrases are divided into two kinds of phrases, non-referential and referential, depending on the presence or absence of a referential function. Referential noun phrases are then further divided based on whether they have a specific or unspecific reference. Finally, specific references are divided into

1 I would like to thank the editors, and the audiences for their comments when I presented a paper on definiteness in Arabic at the Foundations of Arabic linguistics IV meeting in Genoa, and a paper on ambiguous nouns at the 58th meeting of the Society for Near Eastern Studies in Japan (Nippon Oriento Gakkai) in Tokyo. However, any and all possible mistakes are mine.

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_015

234

sakaedani non-referential

Noun phrase

unspecific reference referential

generic reference specific reference homophonic reference

figure 1

Classification of noun phrases after Tanaka 1981:16

generic and homophonic references. Although this study will elucidate that a generic reference is one of the subdivisions of a specific reference, most Arabic grammarians have a different perspective on this point. To understand the difference between an unspecific reference and a generic reference, consider examples (1a) and (1b). (1) a. The elephant has a long trunk. ‘The elephant’ here refers to the species elephant holistically, not to a specific elephant, but it is different from unspecific reference as it has a real referent. b. If I were a zookeeper I would want to keep an elephant at the head. ‘An elephant’ here refers neither to a specific elephant nor to the elephant species, but rather to any elephant. Of course, referentiality and definiteness are different concepts, as is clear from (2). (2) She wants to marry an American. In this utterance ‘an American’ is an indefinite noun phrase, but the sentence can be interpreted in two different ways: a) she wants to marry a specific American man (homophonic reference); and b) she wants to select her husband from among Americans (unspecific reference). 1.2 Deictics Deictics are linguistic forms whose exact meaning cannot be understood without referring to the context in which they are uttered. Pronouns and demonstratives are representative examples. Apart from these, movement verbs such as ‘to go’ and ‘to come’, and adverbs of time are also included into deictics because of their connections with the context.

what is definiteness in arabic?

235

Traditionally, deictic and anaphoric usage of words have been distinguished, but Tanaka (1981:22f.) insists that there is no clear argument to exclude anaphoric usage from deictic usage of a word. As mentioned above, deictics are understood by referring to the context in which they are uttered. Tanaka states that the ‘context’ here must include both the circumstantial and the linguistic context, and there is no clear argument to exclude deixis from the elements of the linguistic context. Another issue is that of relative pronouns. Since relative pronouns mediate between their antecedent and the relative clause, there are some researchers like Naiki (1983:201–217), who call them relative ‘conjunctions’. A relative pronoun without either its antecedent or its relative clause is ambiguous, and it is hard to understand its referent or meaning. Thus, many grammarians did not regard relative pronouns as part of definite noun phrases, and Ibn ʿAqīl had his own idea regarding relatives, as we will see later.

2

The Grammarians’ Description of Definite Noun Phrases

2.1 Sībawayhi Sībawayhi (d. ca. 180/796) names five kinds of noun phrases as definite noun phrases (Kitāb II, 5f.): i. individual proper nouns ii. nouns followed by definite noun phrases iii. al- (meaning nouns prefixed with al-) iv. ʾasmāʾ mubhama ‘ambiguous nouns’ (i.e., demonstratives and similar elements)2 v. pronouns Interestingly, Sībawayhi does not deal with proper nouns identifying a genus. This may indicate that he does not differentiate between unspecific and generic reference. When he lists definite items, he often uses the expression dūna sāʾir ʾummati-hi ‘with the exclusion of the rest of its community’, for instance (Kitāb II, 5.6): “It [an individual proper noun] is definite as it is someone’s name, and the one with the exclusion of the rest of its community, is known by it” (ʾinnamā ṣāra maʿrifatan li-ʾannahu smun waqaʿa ʿalayhi yuʿrafu bihi bi-ʿaynihi dūna sāʾir ʾummatihi). However, Sībawayhi considers pronouns to be exceptions, since he regards only homophonic references as definite.

2 Sadan (2018) argues about this point in detail.

236 table 1

sakaedani ‘Ambiguous nouns’ (only demonstratives) in Sībawayhi

masculine feminine masculine dual feminine dual plural singular singular nominative nominative close objects hāḏā

hāḏihi

hāḏāni

hātāni

hāʾulāʾi

far objects

tilka

ḏānika

tānika

ʾulāʾika

table 2

ḏālika

‘Ambiguous nouns’ (demonstratives and pronouns) in Sībawayhi

masculine masculine dual feminine feminine dual plural singular nominative singular nominative close objects hāḏā

hāḏāni

far objects

ḏānika

hāḏihi

hātāni

hāʾulāʾi

tānika

ʾulāʾika

tilka ḏālika

tīka

masculine feminine dual singular singular huwa

hiya

masculine feminine plural plural

humā hum

hunna

In the second place, relative pronouns are not included in his list. A relative pronoun only mediates between its antecedent and the relative clause and may therefore be difficult to recognize as a definite noun phrase, especially without its relative clause, as noted above. As for ʾasmāʾ mubhama ‘ambiguous nouns’, Sībawayhi only refers to demonstratives in this section (Kitāb II, 5.16f.), as shown in Table 1. He says that these and similar terms are ʾasmāʾ mubhama ‘ambiguous nouns’. Although he recognizes an independent category of the pronoun (ʾiḍmār) (Kitāb II, 5.4), he includes third person pronouns in the category of ambiguous nouns (Kitāb II, 77.12–78.2), as shown in Table 2, and adds the first person singular ʾanā and the second person singular ʾanta (Kitāb II, 79.8).

237

what is definiteness in arabic? table 3

Diminutives of ‘ambiguous nouns’3

ambiguous nouns diminutives

ambiguous nouns diminutives

hāḏā

hāḏayyā

ḏāka

ḏayyāka

ʾulā(ka)

ʾulayyā(ka)

tā

tayyā

ʾulāʾi(ka)

ʾulayyāʾi(ka)

allaḏī

allaḏayyā

allatī

allatayyā

(plural)

allaḏayyūna (plural) allātī

allatayyātu

(dual)

allaḏayyāni

allatayyāni

(dual)

We have seen above that Sībawayhi does not include relative pronouns into the category of definite noun phrases, but he includes them with the demonstratives into the ambiguous nouns in the chapter ‘Diminutives of ambiguous nouns’ (Taḥqīr al-ʾasmāʾ al-mubhama), where he discusses their diminutive forms (Kitāb III, 487–489). These forms are given in Table 3. The fact that Sībawayhi included relatives into the category of the ʾasmāʾ mubhama shows that he found similarities between demonstratives and relatives,4 although he does not express them clearly. Thus, it seems he had an ambiguous attitude by not including relatives into the category of definite nouns, but by including them into that of the ambiguous nouns. 2.2 Al-Mubarrad Al-Mubarrad (d. 285/898) was born in Basra; he introduced the concept of a Basran school by promoting Sībawayhi’s Kitāb in Baghdad. In the chapter ‘Definite and indefinite’ (al-maʿrifa wa-l-nakira) of his Muqtaḍab (IV, 276–285), he names four kinds of noun phrases as definite noun phrases:5 3 Sībawayhi does not mention allaḏīna, allaḏāni, and allatāni in his Kitāb. 4 Ḥassānīn (1990:62) describes the fact that most grammarians regard al-, the head of allaḏī and allatī, as a definite article. ʿAbduh (1973: 65) thinks that the second lām, that is, -la- of al-la-ḏī and al-la-tī, has some relationship with the suffixes of demonstratives. From this viewpoint, the meanings of both demonstratives and relatives depend on a context that includes both circumstantial and linguistic contexts, as argued by Tanaka (1981:22 f.). 5 He includes vocatives, too, as definites, but does not refer to them in this chapter.

238

sakaedani

i. individual proper nouns ii. nouns prefixed with aliii. al-ʾasmāʾ al-mubhama ‘ambiguous nouns’ iv. pronouns He also removes proper nouns identifying a genus from the category of definite noun phrases. Ambiguous nouns, which are listed as the third category of definite nouns (Muqtaḍab IV, 277.5), are demonstratives (Muqtaḍab IV, pp. 227–279). However, in the chapter ‘Diminutives of ambiguous nouns’ (Taḥqīr al-ʾasmāʾ almubhama, Muqtaḍab II, 286–290), he describes relatives as well. Al-Mubarrad claims that ambiguous nouns are different from others from the point of view of their meaning and, in many cases, from a formal point of view as well. As for the semantic point of view, ambiguous nouns mean what they indicate, while from the formal point of view, they are made of two consonants like ḏā and tā, one of which is a weak consonant (Muqtaḍab II, 286.5 f.). The formation of their diminutives is also different, as shown in Table 4 (Muqtaḍab II, pp. 286–289). He also refers to the form allaḏayyayni, which al-ʾAḫfaš suggested. Likewise, he, too, includes relatives into ambiguous nouns and argues about their diminutives. At the same time, al-Mubarrad asserts that man, mā, ʾayy, kam, kayfa, ʾayna, matā, kull, and kilā have no diminutive (Muqtaḍab II, 289 f.). Except for kull and kilā, they are interrogatives, but they are mentioned in the chapter on diminutives as ‘ambiguous nouns’. 2.3 Ibn al-Sarrāj Ibn al-Sarrāj (d. 316/929), who learned directly from al-Mubarrad, also puts forth five kinds of noun phrases as definite noun phrases (ʾUṣūl I, 149): i. al-maknī, metonymically expressed nouns, meaning pronouns in the terminology of the Kufan school ii. al-mubham, ambiguous nouns, meaning demonstratives6 iii. (individual) proper nouns iv. nouns with the prefix alv. nouns followed by definite noun phrases On the other hand, he also points out that there are some nouns that do not specify anything and that cannot be definite, even if they are followed

6 He does not mention other categories such as relatives, that are discussed in the following section.

239

what is definiteness in arabic? table 4

Diminutives of relatives

ambiguous nouns diminutives

ambiguous nouns diminutives

ḏā

ḏayyā

ḏāka

ḏayyāka

ḏihi (= tā)

tayyā

ḏī (= tī)

tayyā

ḏālika

ḏayyālika

ʾulā

ʾulayyā

ʾulāʾika

ʾulayyālika

hāʾulāʾi (long)

hāʾulayyālika hāʾulāʾi (short)

hāʾulayyā

allaḏī

allaḏayyā

allatī

allatayyā

allātī

allatayyātu,a allawayyāb

allaḏūna

allaḏīyūna

allaḏāni

allaḏayyāni

allaḏīna

allaḏīyīna

a Sībawayhi’s view. b al- ʾAḫfaš’s view.

by definite noun phrases. They include: miṯl ‘something like …’, šibh ‘something similar to …’, and ġayr ‘something other than …’ (ʾUṣūl I, 153.4 f.), as in (3). (3) marartu bi-rajulin miṯli-ka wa-bi-rajulin šibhi-ka wa-bi-rajulin ġayri-ka ‘I passed by a man like you, a man similar to you, and a man who is not you’ He adds (ʾUṣūl I, 153) that if miṯl, šibh, and ġayr were not indefinite, they could not modify indefinite noun phrases, however, they actually modify the indefinite noun phrase rajulin ‘a man’, and thus they are indefinite. From a semantic point of view, miṯlu-ka is one who may be like you with regard to your body, height, skin colour, or knowledge, but he is not defined by these elements because the elements that define a thing as itself and none other, are numer-

240

sakaedani

ous. Šibhu-ka functions the same way. As for ġayru-ka, it is indefinite because all those entities that are not you, are ġayru-ka. As for ambiguous nouns, Ibn al-Sarrāj mentions hāḏā, tilka, and ʾulāʾika as examples (ʾUṣūl I, 149.7); that is, he regards ambiguous nouns as demonstratives. He does not refer to relatives in his five categories of definite noun phrases, but says that allaḏī (masculine singular) and allatī (feminine singular) are definite and points out that they are only complete when combined with a relative clause (ṣila) (ʾUṣūl I, 158.8). According to Tanaka (1981:22f.), both demonstratives and relatives are related to the context, in addition to the notions of deictic and anaphoric usage. 2.4 Al-Sīrāfī Al-Sīrāfī (d. 368/979) learned grammar from Ibn al-Sarrāj in Baghdad. First, he quotes Sībawayhi (Sībawayhi, Kitāb II, pp. 77–78) and mentions as ambiguous nouns hāḏā, hāḏihi, hātāni, hāʾulāʾi, ḏālika, ḏānika, tilka, tānika, tīka, ʾulāʾika, huwa, hiya, humā, hunna, and terms that resemble these, while omitting hum (Šarḥ II, 405.6–8). Then, he insists in his commentary on this section (Šarḥ II, 405 f.) that pronouns are not ambiguous nouns; however, they are often confused with them because of the similarities between the two: according to Sībawayhi, when an ambiguous noun becomes a theme, i.e., the subject of a nominal sentence, its rheme becomes an unambiguous noun. Al-Sīrāfī echoes al-Mubarrad’s belief that there are two kinds of mubhama, namely one that appears as a pronoun and another that appears not as a pronoun. According to Al-Sīrāfī, all of them become mubhama because pronouns and demonstratives indicate everything and do not distinguish among inanimates, animates, and others. The exclusion of pronouns from ambiguous nouns in al-Sīrāfī’s commentary on the Kitāb is the first turning point in the theories about the ambiguous nouns. 2.5 Al-Zamaḫšarī and Ibn Yaʿīš In his Mufaṣṣal, al-Zamaḫšarī (d. 538/1144) lists the following five kinds of noun phrases as definite noun phrases (ap. Ibn Yaʿīš, Šarḥ V, 85.15–17): i. individual proper nouns ii. pronouns iii. ambiguous nouns (i.e. demonstratives and relatives) iv. nouns prefixed with alv. nouns followed by definite noun phrases Al-Zamaḫšarī mentions the following proper nouns which identify a genus (ap. Ibn Yaʿīš, Šarḥ I, 34.26f.):

what is definiteness in arabic?

Lion species: Fox species: Hyena species: Scorpion species:

ism

kunya

ʾUsāma Ṯuʿāla Ḥaḍājir Šabwa

ʾAbū l-Ḥāriṯ ʾAbū l-Ḥuṣayn ʾUmm ʿĀmir ʾUmm ʿIryaṭ

241

Some species are such that they have no kunya, but they do have an ism, while others have no ism while having a kunya. In his commentary on the Mufaṣṣal, Ibn Yaʿīš (d. 643/1245) explains that a proper name that indicates a genus distinguishes every species that belongs to that genus; for example, ʾUsāma (lion species) and Ṯuʿāla (fox species) are found in every reference of lion or fox, respectively. Ibn Yaʿīš also explains the need for proper nouns to denote animal species as follows. With regard to human beings, the need to distinguish one person from another led to the introduction of proper names (ism). For domestic animals, in order to distinguish one horse from another, one camel from another, and one dog from another, the Arabs used a laqab like ʾAʿwaj ‘Bow-backed’, Lāḥiq ‘Attached’, and so on. However, since wild animals do not have to be distinguished from each other, even if an animal is referred to with the word laqab, this refers to the whole species. (Šarḥ I, 35.2f.) According to Ibn Yaʿīš, proper names for genera are definite from the viewpoint of their forms because indefinites occur after them as a circumstance (ḥāl); for example, hāḏā ʾUsāma muqbilan ‘This is the lion (species) coming’. He insists that if proper nouns for genera were indefinite, they could not be followed by the circumstance. However, he also describes that from the viewpoint of their meaning they are indefinite because a proper noun for a genus is shared by every species of the genus without any distinction (Šarḥ I, 35.22 f.). Another noteworthy point is that relatives are included by Ibn Yaʿīš in the category of ambiguous nouns. He describes demonstratives (ʾasmāʾ al- ʾišāra) and relatives (al-mawṣūlāt) by saying that demonstratives are words like ḏā, ḏihi, ḏāni, tāni, and ʾulāʾi (Šarḥ V, 86.8f.). Ibn Yaʿīš adds (Šarḥ V, 86.9 f.): Indication (ʾišāra) implies suggesting what exists, so when you call attention to something nearby, you indicate it by using -hā, as in hāḏā or hātā, and when you call attention to something far away, you attach to it the addressee’s kāf [i.e., the personal suffix -ka of the second person masculine singular], as in ḏāka, in order to indicate the difference between the

242

sakaedani

table 5

al-mubhama

‘before it’ (i.e., anaphora) pronouns demonstratives (i.e., endophora) ‘after it’ (i.e., cataphora) (relatives)

two (maʿnā l-ʾišāra al-ʾīmāʾ ʾilā ḥāḍir fa-ʾin kāna qarīban nabbahta ʿalayhi bi-hā naḥw hāḏā wa-hātā wa-ʾin kāna baʿīdan ʾalḥaqtahu kāf al-ḫiṭāb fī ʾāḫirihi naḥwa ḏāka li-l-farq baynahumā). The difference between pronouns (muḍmar) and ambiguous nouns (mubham) is that a third person pronoun is clarified by what appears before it, but that a demonstrative (ism al-ʾišāra) is explained by what appears after it, i.e. a ‘noun showing a genus’ (ism al-jins), such as hāḏā l-rajulu ‘this man’, hāḏā l-ṯawbu ‘this cloth’, and so on (Šarḥ V, 86.11f.), as shown in Table 5. As for ʾibhām ‘ambiguity’, since demonstratives are used for animates (ḥayawān), inanimates ( jamād), and all others (ġayruhumā), you cannot specify ‘what is determined’ (musamman) without ‘what is determined’ (musamman). In other words, their referents are not specified only by the demonstratives, but rather by demonstratives and by what comes after them. (Šarḥ V, 86.14 f.) Relatives are words like allaḏī, allatī, man, and mā, all of which are defined by their relative clauses, that is, what appears after them. Demonstratives are clarified by nouns denoting a genus, while relatives are clarified by the sentence after them. All relatives are ambiguous (mubhama), as they do not distinguish ‘what is determined’ (musamman) without ‘what is determined’ (musamman) like demonstratives do.7 Here, the category of ambiguous nouns becomes something like prepacked deictics containing demonstratives, relatives and, occasionally, pronouns (Šarḥ V, 86.16f.) The common factor seems to be that they are all contextually dependent (see Tanaka 1981).

7 However, ambiguous nouns are treated just as demonstratives in the argument of the definiteness hierarchy, which we will see in the following section.

what is definiteness in arabic?

243

2.6 Ibn Mālik and Ibn ʿAqīl Ibn Mālik (d. 672/1274) studied al-Zamaḫšarī’s Mufaṣṣal under Ibn Yaʿīš. Yet, in his ʾAlfiyya, he distinguishes clearly between demonstratives (ism al-ʾišāra) and relatives (mawṣūl), and he does not use the category or name of ambiguous nouns. This is an important difference between the two grammarians. Demonstratives are referred to by Ibn Mālik from line 82 to 87 (Ibn ʿAqīl, Šarḥ 71–74). He identifies singulars (masculine and feminine), duals (nominatives and genitives/accusatives of each of masculine and feminine), and plurals, and then concisely describes how to form demonstratives which show objects in the distance. Ibn Mālik (ʾAlfiyya lines 86f.) also includes within the class of pronouns the following deictics for objects close by: hunā, hāhunā, and for distant objects: ṯamma, hunnā, hunālika, hinnā. Relatives are discussed from line 88 to 105 (Ibn ʿAqīl, Šarḥ 75–93). Ibn Mālik specifies, in lines 88–90 (Ibn ʿAqīl, Šarḥ 75), the relatives allaḏī and allatī, and then says that duals are formed by removing the yāʾ at the endings and adding a dual ending. To do this, you can geminate the nūn in the endings (allaḏānni, allatānni, allaḏaynni, allataynni) as compensation for removing yāʾ. In lines 91f., he suggests al-ʾulā as a plural. His commentator Ibn ʿAqīl (d. 769/ 1367) explains that this is a common form between masculine and feminine. Ibn Mālik further mentions allaḏīna as an invariable plural relative; however, with regard to the latter, some tribes use allaḏūna as a nominative. He suggests, as feminine forms, allāʾi and allāti (these do not have yāʾ in their endings). In lines 93f., Ibn Mālik refers to man, mā, al-, and to ḏū of the tribe Ṭayyiʾ, ḏātu (feminine singular) and ḏawātu (feminine plural, i.e., corresponding to allātī). He suggests ḏā, which follows the interrogatives mā and man as a relative in line 95 (Ibn ʿAqīl, Šarḥ 82). In line 96 (Ibn ʿAqīl, Šarḥ 83), Ibn Mālik insists that a relative clause (ṣila) and an ‘appropriate pronoun’ (ḍamīr lāʾiq) referring to the relative, are necessary for all relatives. Then, in line 97 (Ibn ʿAqīl, Šarḥ 84), he defines a relative clause as a sentence or something resembling a sentence. Ibn ʿAqīl explains that something resembling a sentence refers to adverbs (ẓarf ) that express time or place and a prepositional phrase, as in jāʾa llaḏī ʿindaka wa-llaḏī fī l-dār ‘The one who is with you and the one who is in the residence has come’. In line 98 (Ibn ʿAqīl, Šarḥ 85), he mentions the relative clause of al- and explains that it is ṣifa ṣarīḥa. Ibn ʿAqīl further explains that ṣifa ṣarīḥa means an active participle like al-ḍārib ‘the one who hits’, and a passive participle like al-maḍrūb ‘the one who is hit’, and it is a rare occurrence for an imperfect form of a verb to appear in a relative clause. Ibn Mālik, in line 99 (Ibn ʿAqīl, Šarḥ 86) regards ʾayy as similar to mā (Ibn ʿAqīl explains that this means that it does not change according to gender and

244

sakaedani

number), but it is declined when ʾayy is not followed by a (genitive) pronoun suffix, and when the nominative pronoun (ṣadr) in the relative clause is not omitted. However, at the beginning of line 100, he mentions that ʾayy is always declined in the language of some Bedouin tribes. In his commentary on line 53 of the ʾAlfiyya, Ibn ʿAqīl suggests six kinds of noun phrases as definite noun phrases (Šarḥ 50): i. pronouns ii. demonstratives iii. proper nouns iv. nouns prefixed with alv. relatives vi. nouns followed by definite noun phrases In this list, proper nouns denoting a genus are included among proper nouns for the first time, but Ibn ʿAqīl says that they are like individual proper nouns, i.e., definite, from a formal point of view, while they are indefinite from a semantic perspective (Šarḥ 69, on lines 79–81 of the ʾAlfiyya). Moreover, relatives have been assigned to an independent category and the category of prepacked deictics, ‘the ambiguous nouns’, has disappeared. However, Ibn ʿAqīl adds five conjunctions as ‘particle relatives’, namely ʾan, ʾanna, kay, mā, and law to the ordinary relatives, which he calls ‘nominal relatives’ (Šarḥ 75–77, on lines 88–90 of the ʾAlfiyya). Both particle relatives and nominal relatives need subordinate clauses. As for the nominal relatives (mawṣūl ismī), he also deals with ʾayy, al-, etc. as relatives adding to allaḏī, allatī, and so on. Besides, he explicitly mentions that a relative must be followed by a relative clause and that it is imposed as a condition that a relative clause must include a resumptive pronoun. This is important since relatives are regarded as definite nouns because of their relative clauses (Šarḥ 83, on line 96 of the ʾAlfiyya). The distinction between ‘nominal relative’ (mawṣūl ismī) and ‘particle relative’ (mawṣūl ḥarfī) is not found in Ibn Mālik; the latter in particular is not included among the relatives in the body of Ibn Mālik’s text. According to Ibn ʿAqīl, Ibn Mālik uses the expression ‘nominal relative’ (mawṣūl ismī) in order to clearly distinguish between ‘nominal relative’ and ‘particle relative’ (mawṣūl ḥarfī). In his commentary on line 96 of the ʾAlfiyya he also says that all relatives, regardless of whether they are particle or nominal, need a relative clause (ṣila) following them in order to clarify their meaning8 (Šarḥ 83). In other words, he describes clauses which follow particle relatives as similar to clauses which follow nominal relatives. 8 Giolfo (2012: 144, n. 38) says “Arabic ṣila designates a sentence after a mawṣūl either ismī (rel-

what is definiteness in arabic?

245

2.7 Ibn Hišām Ibn Hišām (d. 761/1359) lists six kinds of noun phrases as definite noun phrases (Šuḏūr 134–157): i. pronouns ii. proper nouns iii. demonstratives iv. relatives v. nouns prefixed with alvi. nouns followed by definite noun phrases Ibn Hišām explicitly includes proper nouns denoting a genus within the class of proper nouns when he says that an expression such as ʾusāmatu ʾašjaʿu min ṯuʿālata means al-ʾasadu ʾašjaʿu mina l-ṯaʿlabi ‘The lion is braver than the fox’. In such expressions, the article al- defines a genus (Šuḏūr 138 f.). As for relatives, Ibn Hišām does not include any particle relatives, but regards only nominal relatives as relatives. He describes them in detail, borrowing many views from Ibn ʿAqīl. He does not use the term ‘ambiguous nouns’, either. According to Ibn Hišām, relatives need two things: a relative clause and a pronoun within the relative clause referring to the relative (Šuḏūr 141). There are four types of relative clauses (Šuḏūr 141): i. a predication sentence, whose authenticity can be determined, as in (4a), whereas the sentence in (4b) is unacceptable because the authenticity of interrogatives or relatives cannot be checked (4) a. jāʾa-nī llaḏī qāma/allaḏī ʾabūhu qāma ‘The man who stood up/ whose father stood up came to me’ b. *jāʾa llaḏī hal qāma/allaḏī lā taḍribhu ii. complete (tāmm) ẓarf (accusative nouns that indicate time or place), and iii. complete (tāmm) prepositional phrase, as in (5a), whereas the sentences in (5b) and (5c) are unacceptable (5) a. wa-lahu man fī l-samāʾi wa-l-ʾarḍi wa-man ʿindahu lā yastakbirūna ʿan ʿibādatihi (Q. 21/19) ‘And those who are in the heavens and the earth belong to Him, and those who are with Him are never too proud to worship Him.’ b. *jāʾa llaḏī al-yawma9 c. *jāʾa llaḏī bika ative pronouns) or ḥarfī (particles of conjugation). The expression ‘is not ṣila’ could be explained as ‘is not in relation with what precedes’, where the concept of ‘being in relation with what precedes’ is wider than the concept of ‘relative clause’ […]”. 9 Both of al-yawma and bika are not complete (tāmm). Ibn Hišām says that if ẓarf and jārr

246

sakaedani

iv.

explicit adjective (waṣf ṣarīḥ), a participle which loses its character as a noun becoming a relative clause headed by al-,10 e.g. al-ḍāribu, almaḍrūbu. In (6) the pronoun -hu is an example of a pronoun in the relative clause referring to the relative (ḍamīr ʿāʾid) (Šuḏūr 141–143). (6)

jāʾa llaḏī qāma ʾabūhu

Pronouns in the relative clause referring to the relative must correspond to it in gender, number, and so on. However, an overt (ẓāhir) noun can change the place of relatives, as in (7) and (8). (7) suʿādu llatī ʾaḍnāka ḥubbu suʿādā (8) al-ḥamdu li-llāhi llaḏī ḫalaqa l-samāwāti wa-l- ʾarḍa wa-jaʿala l-ẓulumāti wa-l-nūra ṯumma llaḏīna kafarū bi-rabbi-him yaʿdilūna (Q 6/1) In (7) ḥubbu suʿādā is to be considered ḥubbuhā (Šuḏūr 143.6 f.). At this instance, Ibn Hišām presents al-Zamaḫšarī’s interpretation, according to which kafarū bi-rabbi-him yaʿdilūna means kafarū bi-hi yaʿdilūna (Šuḏūr 143.1–6); therefore, this verse translated as ‘Praise to God Who created the heavens and the earth and made darkness and light, then (made) those who disbelieve their Lord setting up equals to Him’.

10

wa-majrūr are relative clauses they must be related to an omitted verb, which is assumed to be istaqarra (to settle). (Qaṭr 194) Here *istaqarra l-yawma ‘?He settled today’ and *istaqarra bika ‘?He settled by you’ do not make sense, and therefore they are not complete (tāmm). Ibn Hišām refutes two opinions about this al- (Šuḏūr 148). One view regards this as a particle relative. Ibn Hišām denies this opinion because the al- clause cannot be paraphrased with a verbal noun (maṣdar); in other words, it is different from the ʾan clause. Additionally, al- has a pronoun referring to the relative. The second view regards this as a definite article. He denies this opinion, too, because al- as a relative involves a verb in its relative clause, as in Q. 100/3f.: fa-l-muġīrāti ṣubḥan fa-ʾaṯarna bihi naqʿan ‘And those that raid at dawn, and raise dust in it [the morning, or the place]’. Here al-muġīrāti is a relative al- and an active participle muġīrāti is attached to the relative al-. Moreover, muġīrāti and ʾaṯarna are connected in a series, namely, ʾaṯarna is also attached to the relative al-.

what is definiteness in arabic?

3

247

Definiteness Hierarchy

Within the category of definite nouns the grammarians posited a hierarchy of definiteness, dependent on the kind of definite noun that is modified and by which kind of definite nouns it is modified. Sībawayhi describes the following modifiers of the noun. As for noun phrases modifying definite noun phrases, the following shows how Sībawayhi analyzes the modification of definite noun phrases by other noun phrases: i. individual proper nouns are modified by a) noun phrases followed by ‘such ones’ (i.e., pronouns and demonstratives); b) (nouns prefixed with) al-; and c) ambiguous noun phrases (Kitāb II, 6). ii. Nouns followed by definite noun phrases are modified by a) nouns followed by something in the same way; b) (nouns prefixed with) al-; and c) ambiguous noun phrases (Kitāb II, 7). iii. Nouns prefixed by al- are modified by a) (nouns prefixed with) al-; and b) nouns added by (that is, nouns prefixed with) al- (Kitāb II, 7). iv. Ambiguous nouns are modified by a) nouns with the addition of (nouns prefixed with) al-; and b) adjectives with the addition of (nouns prefixed with) al- (Kitāb II, 7f.). Al-Zamaḫšarī states that more specific noun phrases are also increasingly definite, and notes the following hierarchy (ap. Ibn Yaʿīš, Šarḥ V, 87): i. pronouns a) 1st person, which is the most definite b) 2nd person c) 3rd person ii. individual proper nouns iii. ambiguous nounn iv. nouns prefixed with al-, which is the least definite As for nouns followed by definite noun phrases, they are as definite as the latter. In his commentary on the Mufaṣṣal, Ibn Yaʿīš says that the ‘more specific’ (ʾaḫaṣṣ) a noun is, the more definite it is. He gives two reasons as to why pronouns are more definite than individual proper nouns. In the first place, a pronoun has only one referent, although a proper noun can have plural referents. In the second place, a pronoun cannot modify another noun as an adjective and cannot be modified by an adjective, but a proper noun can be modified by an adjective. However, according to Ibn Yaʿīš, Kufan grammarians, and ʾAbū Saʿīd al-Sīrāfī, who also followed the same grammar rules, insist that ambiguous nouns are the most definite, followed by pronouns as the next most definite, and finally,

248

sakaedani

proper nouns as the third most definite. Their argument is that a proper noun has indeed only one referent, but a referent of a pronoun changes according to the context and cannot be defined. Besides, when an antecedent is indefinite, the pronoun which refers to it is also indefinite. Pronouns prefixed with rubba ‘many a …’ are indefinite. Counterarguments to this line of reasoning are given by Ibn Yaʿīš, who states that it has already been said that one proper noun can have plural referents. Even if the antecedent is indefinite, it cannot be said that the pronoun is also indefinite, because we can surely know what the pronoun means. As for the pronouns prefixed with rubba, these are exceptional. In addition, some other grammarians, among them Ibn al-Sarrāj, also insist that ambiguous nouns are the most definite, pronouns are the second most definite, and pronouns third, because referents of ambiguous nouns can be recognized both visually and instinctively (that is, through sight and conception). However, others cannot be recognized except through one method, i.e., conception, and thus are weak from the viewpoint of definiteness. Ibn Yaʿīš refutes this argument by stating that proper nouns are more definite than ambiguous nouns because the former can be modified by adjectives, but there are no nouns modified by a proper noun. Furthermore, he points out that demonstratives can be modified by adjectives, while nouns can be modified by demonstratives. Adjectives cannot be ‘more specific’ (ʾaḫaṣṣ) than their modified nouns, therefore demonstratives are weak in terms of definiteness. In addition, Ibn Yaʿīš explains why demonstratives are more definite than nouns prefixed by al-. He says that al- is the most ambiguous and the closest to indefinite nouns, as in (9). (9) ʾinnī ʾamurru bi-l-rajuli ġayri-ka fa-yanfaʿu-nī wa-bi-l-rajuli miṯli-ka fa-yuʿṭīnī ‘I pass by the man who is not you, then he is of help to me, and by the man like you, then he gives me’ (A)l-rajuli is modified by ġayr and miṯl. As mentioned above, miṯl, šibh, and ġayr do not specify anything and cannot be definite, even if they are followed by definite noun phrases. Example (10) shows that nouns prefixed by al- are very close to indefinite nouns. (10) (i)hdinā l-ṣirāṭa l-mustaqīma ṣirāṭa llaḏīna ʾanʿamta ʿalayhim ġayri l-maġḍūbi ʿalayhim

what is definiteness in arabic?

249

‘Guide us to the straight path—The path of those upon whom You have bestowed favour, not of those who have evoked [Your] anger’ (Q. 1/6f.) In this verse, ġayr modifies (a)llaḏīna. This is caused by the use of al-, which does not refer to something specific. Thus, nouns prefixed by al- cannot be modified by ambiguous nouns, he says. Ibn Hišām presents the definiteness hierarchy as follows: pronouns; proper nouns; demonstratives; relatives; nouns prefixed with al-. He notes that a noun followed by another definite noun to form an ʾiḍāfa takes the latter’s level of definiteness, but that only nouns followed by pronouns take on the same level as proper nouns, as in (11). (11) marartu bi-zaydin ṣāḥibika ‘I passed by Zayd, your friend’ (Šuḏūr 157.2) If a noun followed by a pronoun takes on the same level as pronouns, ṣāḥibika cannot modify zaydin in this case. However, Ibn Yaʿīš regards a noun followed by a pronoun as on the same level as pronouns, and considers that such an example is an apposition (badal). Ibn Hišām himself also mixes this up sometimes, and adduces the sentence jāʾa-nī zaydun ʾaḫūka ‘Zayd, your brother, came to me’ as an example of an apposition (badal). Some grammarians assume that a noun followed by another definite noun always takes a lower level than the definite noun by one level. However, Ibn Hišām adduces example (12). (12) ka-ḫuḏrūfi l-waldi l-muṯaqqabi ‘Like the pierced baby’s spinning top’ (Šuḏūr 156.14) He says that (a)l-muṯaqqabi, which is a noun prefixed by al-, modifies ḫuḏrūfi l-walīdi, so ḫuḏrūfi l-walīdi cannot be less definite than nouns prefixed by al-, that is, it takes the same level as (a)l-muṯaqqabi. The definiteness hierarchy as described thus far is determined from a formal point of view, but it is well supported from the perspective of cognition as well. In the Givenness Hierarchy proposed by Gundel et al. (1993:275), the status of each category is characterized in Figure 2. The following examples illustrate each category: type identifiable: I couldn’t sleep last night. A dog (next door) kept me awake.

250

sakaedani

in > Activated focus

{it} figure 2

that ⎫ ⎧ { } this ⎬ ⎨ { } ⎩ this N ⎭

> familiar >

{that N}

uniquely type > referential > identifiable identifiable

{the N}

{indefinite this N}

{a N}

Givenness Hierarchy after Gundel et al. 1993:275

referential: I couldn’t sleep last night. This dog (next door) kept me awake. uniquely identifiable: I couldn’t sleep last night. The dog (next door) kept me awake. familiar: I couldn’t sleep last night. That dog (next door) kept me awake. activated: I couldn’t sleep last night. That kept me awake. in focus: a. My neighbour’s bull mastiff bit a girl on a bike. It’s } the same dog that bit Mary Ben last summer. b. { That’s According to Gundel et al. (1993), ‘type identifiable’ means that the addressee is able to access a representation of the type of object described by the expression. When using a ‘referential’, the speaker intends to refer to a particular object or objects. As for ‘uniquely identifiable’, the addressee can identify the speaker’s intended referent on the bases of the nominal alone. In a ‘familiar’ utterance, addressees are able to uniquely identify the intended referent because they already have a representation of it in memory. In an ‘activated’ utterance, the referent is represented in current short-term memory. Activated representations may have been retrieved from long-term memory, or they may arise from the immediate linguistic or extralinguistic context. As for ‘in focus’, the referent is not only in short-term memory, but also at the current center of attention. We apply this Givenness Hierarchy to the Arabic Definiteness Hierarchy. i. raʾaytu rajulan ‘I saw a man’. At this stage, one can identify only the type of what one saw: it was neither a woman, nor a little boy, but a man. ii. raʾaytu l-rajula ‘I saw the man’. This is used when one can specifically identify the person one saw. iii. raʾaytu l-rajula llaḏī kāna ʿindaka ‘I saw the man who was with you’. Here, the addressee can also identify the person one saw because of the information provided by the relative clause. iv. raʾaytu haḏā l-rajula ‘I saw this man’. At this instance, the addressee can uniquely identify the man; however, if the addressee is not able to iden-

what is definiteness in arabic?

251

tify the man using the information provided, one can say ‘I saw this man’, if the man is actually present. v. raʾaytu muḥammadan ‘I saw Muḥammad’. If the man one saw is activated by the addressee, he can be referred to by his name. However, if the man has not been activated, the addressee cannot identify who Muḥammad is, or which Muḥammad one saw. vi. raʾaytuhu ‘I saw him’. At this stage the man one saw is now in focus. Thus the Arabic Definiteness Hierarchy seems to be compatible with the Givenness Hierarchy.

4

Conclusions

This study has shown that the definition of the definite noun phrase has changed with the times, particularly the meaning of the terms al-ʾasmāʾ almubhama (the ambiguous noun phrases) and al-ʿalam li-l-jins (the proper nouns for genera). The former include deictics such as demonstratives and relatives, and sometimes pronouns; the meaning of these deictics are determined according to the wider context. The latter covers the distinction between unspecific and generic reference. Some grammarians regarded these as definite from the viewpoint of forms and as indefinite from the viewpoint of meanings; however, others regarded them as definite noun phrases, just like proper nouns for individuals. The definiteness hierarchy from pronouns to nouns prefixed with al- was defined by the grammarians mainly from a formal point of view, but it roughly fits the cognitive order in the Givenness Hierarchy.

Bibliographical References A

Primary Sources

Ibn ʿAqīl, ʾAlfiyya = Bahāʾ al-Dīn ʿAbdallāh Ibn ʿAqīl, Šarḥ Ibn ʿAqīl ʿala ʾAlfiyyat Ibn Mālik. Ed. by ʾAḥmad Salīm al-Ḥumṣī and Muḥammad ʾAḥmad Qāsim. Tripoli (Lebanon): Dār Jarrūs, 1990. Ibn Hišām, Qaṭr = Jamāl al-Dīn ʾAbū Muḥammad ʿAbdallāh ibn Yūsuf Ibn Hišām alʾAnṣārī, Šarḥ Qaṭr al-nadā wa-ball al-ṣadā. 1st ed. Saida and Beirut: al-Maktaba alʿAṣriyya, 1994. Ibn Hišām, Šuḏūr = Jamāl al-Dīn ʾAbū Muḥammad ʿAbdallāh ibn Yūsuf Ibn Hišām alʾAnṣārī, Šarḥ Šuḏūr al-ḏahab fī maʿrifat kalām al-ʿArab. Saida and Beirut: al-Maktaba al-ʿAṣriyya, n.d.

252

sakaedani

Ibn al-Sarrāj, al-ʾUṣūl = ʾAbū Muḥammad ibn al-Sarī Ibn al-Sarrāj, al-ʾUṣūl fī l-naḥw. Ed. by ʿAbd al-Ḥusayn al-Fatlī. 3 vols. 3rd ed. Beirut: Muʾassasat al-Risāla, 1988. Ibn Yaʿīš, Šarḥ = Muwaffaq al-Dīn Yaʿīš Ibn Yaʿīš, Šarḥ al-Mufaṣṣal. 10 vols. Beirut: ʿĀlam al-Kutub, n.d. Mubarrad, Muqtaḍab = ʾAbū l-ʿAbbās Muḥammad ibn Yazīd al-Mubarrad, Kitāb almuqtaḍab. Ed. by Muḥammad ʿAbd al-Ḫāliq ʿUḍayma. 4 vols. 6th ed. Cairo: Wizārat al-ʾAwqāf, Lajnat ʾIḥyāʾ al-Turāṯ al-ʾIslāmī, 1966–1979. Sībawayhi, Kitāb = ʾAbū Bišr ʿAmr ibn ʿUṯmān ibn Qanbar Sībawayhi, al-Kitāb. 5 vols. 3rd edition. Ed. by Muḥammad ʿAbd al-Salām Hārūn. Cairo: Maktabat al-Ḫānjī, 1988. Sīrāfī, Šarḥ = ʾAbū Saʿīd al-Ḥasan ibn ʿAbdallāh al-Sīrāfī, Šarḥ Kitāb Sībawayhi. 5 vols. Ed. by ʾAḥmad Ḥasan Mahdalī and ʿAlī Sayyid ʿAlī. Beirut: Dār al-Kutub al-ʿIlmiyya, 2008

B

Secondary Sources

ʿAbduh, Dāwūd. 1973. ʾAbḥāṯ fī l-luġa al-ʿarabiyya. Beirut: Maktabat Lubnān. Giolfo, Manuela E.B. 2012. “Yaqum vs qāma in the conditional context: A relativistic interpretation of the frontier between the prefixed and the suffixed conjugations of the Arabic language”. The foundations of Arabic linguistics: Sībawayhi and early Arabic grammatical theory, ed. by Amal Elesha Marogy, 135–160. Leiden: E.J. Brill. Gundel, Jeanette K., Nancy Hedberg, and Ron Zacharski. 1993. “Cognitive status and the form of referring expressions in discourse”. Language 69.274–307. Ḥassānīn, Fatḥī ʿAlī. 1990. ʾAdāt al-taʿrīf fī l-naḥw al-ʿarabī: Dalālāt wa-stiʿmālāt. Cairo: Maṭbaʿat al- ʾAmāna. Naiki, Ryoichi. 1982. Kiso Arabiyago [Basic Arabic]. Tokyo: Daigakushorin. Sadan, Arik. 2018. “Demonstratives in Sībawayhi’s Kitāb”. The foundations of Arabic linguistics. III. The development of a tradition: Continuity and change, ed. by Georgine Ayoub and Kees Versteegh, 178–189. Leiden: E.J. Brill. Tanaka, Nozomi. 1981. “‘Ko-, so-, a-’ wo meguru shomondai [Some issues of ‘ko-, so, and a-’]”. Nihongo no shijishi, ed. by National Institute for Japanese Language, 1–50. Tokyo: National Institute for Japanese Language.

Definition and Determination in Medieval Arabic Grammatical Thought Manuel Sartori

1

Introduction

The grammatical tradition of Arabic may seem to be based entirely on the oldest grammar book to have reached us, the Kitāb of Sībawayhi (d. 180/796?), its later development being nothing more than a reinterpretation and/or a reorganization of this first material. Yet, fundamental and significant additions have been brought to bear on this first and crucial work. Suffice it to mention here the category of ʾinšāʾ, which emerged in the post-Classical period of Arabic grammar (after the first half of the 5th/11th century), probably as the result of influence from the legal sciences.1 The term ʾinšāʾ is not the only one to have remained unrecognized for a long time. This is also the case of taḫṣīṣ, usually rendered by ‘particularization’ which, without being totally unknown, has remained largely unrecognized. It has been shown to appear in Arabic grammar relatively late, around the end of the 4th/10th century. Its first occurrences are in the form of nouns derived from the consonantal root ḫ-ṣ-ṣ, first in the form of iḫtiṣāṣ with al-Sīrāfī (d. 368/979), then of muḫaṣṣiṣ with ʾAbū ʿAlī al-Fārisī (d. 377/987). The term taḫṣīṣ itself appears with Ibn al-Warrāq (d. 381/991), similar in sense to its later use, but it is primarily with Ibn Jinnī (d. 392/1002), then with ʿAbd al-Qāhir al-Jurjānī (d. 471/1078), and especially with al-Zamaḫšarī (d. 538/1144) that taḫṣīṣ acquires its technical and grammatical meaning of ‘particularization’.2 After studying the term taḫṣīṣ, I turn to the term taḫlīṣ that occurs in combination with it, in order to specify the time of its appearance, and then we analyze the terminological distributions in connection with the opposition definition/indefinition (taʿrīf/tankīr) in Arabic. First, we shall take a look at the concepts of definition/indefinition and determination/indetermination.

1 On the origin of the ʾinšāʾ category, see Larcher’s articles, republished in Larcher (2014). 2 For these historical data and the technical sense of taḫṣīṣ, see Sartori (2018).

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_016

254 2

sartori

Definition/Indefinition and Determination/Indetermination

Two pairs of notions should be effectively and logically distinguished on the following basis: all that is definite is determinate, but all that is determinate is not necessarily definite.3 In this context, the terms of the first pair, definition/indefinition, refer to definite and indefinite expressions. An example of an indefinite and indeterminate expression is man, i.e. a single noun devoid of any marker of definition or determination. From there, a move towards definition begins: a man is an indefinite and determinate expression (here by a quantification, the article a, which is an indefinite determiner), while a tall man or a man of science remain indefinite expressions, which are, however, more determinate than the first one (for they have a quantification, a, and a qualification, tall or of science). These expressions are not yet as definite as the man, which by itself is a definite expression (where the article the is a definite determiner). This expression is then both definite and determinate. However, it is less determinate than for example the tall man, which adds a determination (tall or the man of science) for the same reasons. In Arabic, the pair definition/indefinition is identified easily with that of taʿrīf/tankīr, whose terms are connected to maʿrifa and nakira, respectively. In Arabic terminology maʿrifa is a ‘definite expression’, while nakira is an ‘indefinite expression’. The definite term in Arabic is so either by nature (e.g., a proper name like Zayd), or by the article (al-rajul) or by annexation (rajul al-madīna). As for the indefinite term, whereas in English a term may be indefinite and indeterminate (man), in Arabic a term is minimally determinate since rajul equals ‘a man’ and not ‘man’. In technical terms, indefinition is thus tankīr and definition taʿrīf. Could it be the case that taḫṣīṣ, whose meaning is ‘particularization’, is a form of determination? This is precisely the question that the present article wishes to answer.

3

Taḫṣīṣ and Its Complementary Term

3.1 The Technical Sense of taḫṣīṣ in Arabic Grammar As ‘particularization’, taḫṣīṣ is not entirely unknown in the secondary literature. However, when it appears, it is always incidentally, no special section being dedicated to it. This is the case of Wright (since it is an addition made by Wright

3 Which opposes the scheme proposed by Kouloughli (2001:40), who claims that a definite term can at the same time be indeterminate.

definition and determination in medieval arabic thought

255

in a footnote, it does not derive from Caspari, see Sartori 2018: 205, n. 5), Reckendorf, Gätje, Fleisch, Troupeau, Carter, Badīʿ Yaʿqūb and ʿĀṣī, and Brustad.4 One even finds the concept referred to without the term in the articles ʾiḍāfa and specificity of the Encyclopedia of Arabic language and linguistics.5 In the vast majority of cases, the term and its technical scope are ignored.6 On the basis of the scattered data I have collected the following may be said about the technical meaning of the word: taḫṣīṣ appears in connection with semantic annexation (ʾiḍāfa maʿnawiyya), with qualification (naʿt), and even in connection with the explanatory apposition (ʿaṭf bayān), albeit merely as an extension of qualification in Ibn Mālik (d. 672/1274) (Šarḥ I, 533; see Sartori 2018). Regarding taḫṣīṣ, Carter speaks of a “weaker type of definition”. For Wright it is a “partial determination”, and for Reckendorf a nähere Bestimmung (almost determination/definition).7 Apparently, taḫṣīṣ is assigned two significations (determination and definition), which it would seem useful to distinguish.8 For reasons to be explained below, confusing the two leads to inadequacy. It seems that Arab grammarians were aware of a distinction to be made between definition and determination. Suffice it for now to say that as a technical term, taḫṣīṣ has the meaning of particularizing an indefinite term by another one, 4 See Wright (1996:II, 198D, 199A, 260–261D); Reckendorf (1921:57, 193, 200, 218); Gätje (1970:221, 235); Carter (1981:377, 461); Fleisch (1986:1008b); Badīʿ Yaʿqūb and ʿĀṣī (1987:I, 154, 367, II, 868, 1254); Troupeau (1993:1034a); Carter (2000:241b); Brustad (2000:21). 5 See Ryding and Versteegh (2007:295b), Hoyt (2009:316b). 6 See Silvestre de Sacy (1831); Forbes (1863); Palmer (1874); Socin (1885); Vernier (1891); Howell (1911); Fleisch (1961, 1979), and finally Blachère and Gaudefroy-Demombynes (1975). Regarding recent grammars of Arabic, it is still completely absent. See Cantarino (1974); Kouloughli (1994); Neyreneuf and Al-Hakkak (1996); Badawi et al. (2004); Buckley (2004); Holes (2004); Alosh (2005); Ryding (2005); Hassanein (2006); McCarus (2007); Imbert (2008); Schulz et al. (2008); El-Ayoubi et al. (2010). 7 See Carter (2000:241b); Wright (1996:II, 261D); Reckendorf (1921:200). German dictionaries indicate that Bestimmung means both ‘definition’ and ‘determination’, which demonstrates its vagueness from a terminological point of view. 8 The confusion between definition/indefinition and determination/indetermination is fairly common. Some authors speak of Determination und Indetermination for taʿrīf and tankīr and of Qualifikation for taḫṣīṣ (see Gätje 1970: 226). This is also the case with Wensinck (1931), whose study is entitled “The article of determination in Arabic”, whereas Heselwood and Watson (2015) speak of “The Arabic definite article”. As noted by Jan Retsö, “they [the FrancoGerman school] use the term ‘indetermination’ variously for indefiniteness, non-definiteness (or both), ‘indefinite article’, or the ending -n”, where he distinguishes between “non-definite” = indefinite and indeterminate (e.g. house) and “indefinite” = indefinite and determinate (e.g. a house) (Retsö 1986:342f.). One can regret with Pierre Larcher that the terms of the couple taʿrīf/tankīr are renamed “détermination” and “indétermination” in Arabist grammars (see Larcher 1991:146, n. 18). See also Kouloughli (2001, especially 39 f.).

256

sartori

itself indefinite, whether it is the second term of an annexation (zāranī rajulu falsafatin ‘a man of philosophy paid me a visit’), or a qualifier in the broadest sense of the term, that is to say, an attributive adjective (rajulun ṭawīlun ‘a tall man’), including qualifying clauses (rajulun yaktubu risālatan ‘a man who writes a letter’) and prepositional phrases (rajulun min banī tamīmin ‘a man from the Banū Tamīm’), or an explanatory apposition (ištaraytu ḥalyan siwāran ‘I bought jewelry, a bracelet’).9 In this sense, taḫṣīṣ, without being properly speaking intermediate to tankīr/taʿrīf, is connected with these terms, according to a suggestion made by the rhetorician al-Qazwīnī (d. 739/1338), for whom “the completion of the particularization is [made] by definition” (al-taḫṣīṣ kamāluhu bi-l-taʿrīf, ʾĪḍāḥ 41). From this reading, it is possible to understand those of Wright, Reckendorf and Carter. In this sense, taḫṣīṣ, as ‘particularization’, would indeed be a form of ‘determination’. 3.2 The Complementary Terms to taḫṣīṣ While the technical term taḫṣīṣ was not entirely unknown, the situation is different for its complementary terms, which are nowhere treated in the secondary sources cited above, nor in the primary sources. 3.2.1 Taḫlīṣ The first appearance of a term used as complementary to taḫṣīṣ in its technical sense is apparently in Ibn Jinnī’s Ḫaṣāʾiṣ (I, 392), when he speaks about the different case endings in the expression bi-smi llāhi l-raḥmānu/a l-raḥīmu/a: and this is because when [the noun] Allāh is qualified, the goal is not to define it by what follows in terms of qualifiers, since concerning this name there is no doubt that it would need to be qualified in order to specify it, for it is the name of one with whom no-one is associated […]. Thus, since it is not exposed to doubt, its qualification does not intervene in order to specify, but to praise Allāh […], and so making it follow its declension formally takes the same course as that what follows for purposes of specification or particularization (wa-ḏālika ʾanna Allāh taʿālā ʾiḏā wuṣifa fa-laysa l-ġaraḍ fī ḏālika taʿrīfahu bi-mā yatbaʿuhu min ṣifatihi li-ʾanna hāḏā l-ism lā yaʿtariḍu šakk fīhi fa-yaḥtāja ʾilā waṣfihi li-taḫlīṣihi li-ʾannahu l-ism al ladī lā yušāraku fīhi ʿalā wajh wa-baqiyyat ʾasmāʾihi— ʿazza wa-jalla—ka-l-ʾawṣāf al-tābiʿa li-hāḏā l-ism wa-ʾiḏā lam yaʿtariḍ šakk fīhi lam tajiʾ ṣifatuhu li-taḫlīṣihi bal li-l-ṯanāʾ ʿalā llāh taʿālā […] wa-ḏālika ʾanna ʾitbāʿahu ʾiʿrābahu jārin fī l-lafẓ majrā mā yatbaʿ li-l-taḫlīṣ wa-l-taḫṣīṣ) 9 See about these points Sartori (2018).

definition and determination in medieval arabic thought

257

The term taḫlīṣ ‘specification,’10 here is used by the same author along with taḫṣīṣ in a passage related to qualification (Ḫaṣāʾiṣ II, 146, see also II, 447): “and this is because qualifications in speech are of two types, either for specification and particularization, or for praise and eulogy” (wa-ḏālika ʾanna l-ṣifa fī l-kalām ʿalā ḍarbayn ʾimmā li-l-taḫlīṣ wa-l-taḫṣīṣ wa-ʾimmā li-l-madḥ wa-l-ṯanāʾ). Nevertheless, it is not yet possible to comprehend either term in a very precise way since Ibn Jinnī does not say more than this. Thus, the only certain thing is that these two terms operate together at the level of qualification. The term taḫlīṣ is absent from Sībawayhi’s Kitāb,11 appearing for the first time in al-Mubarrad’s (d. 285/898) Muqtaḍab, and later in al-Zajjājī’s (d. 337/949) Kitāb al-lāmāt in a non-technical sense, without any connection either with taḫṣīṣ, 0r with annexation or qualification.12 At the same time as Ibn Jinnī,13 we find taḫṣīṣ in Ibn Fāris (d. 395/1004), once again in connection with adjectives (Ṣāḥibī 52): The adjective follows two courses. One of them is to distinguish a noun from a noun, as when we say zaydun al-ʿaṭṭāru ‘Zayd the perfumer’ and zaydun al-tamīmiyyu ‘Zayd the Tamīmite’, distinguishing it by means of its adjective from others sharing the same name. The other [course] has the meaning of praise and of blame, like al-ʿāqil ‘the judicious’ and al-jāhil ‘the ignorant’ (wa-l-naʿt yajrī majrayayn ʾaḥaduhumā taḫlīṣ ism min ism ka-qawlinā zaydun al-ʿaṭṭāru wa-zaydun al-tamīmiyyu ḫallaṣnāhu bi-naʿtihi min allaḏī šārakahu fī ismihi wa-l-ʾāḫār ʿalā maʿnā l-madḥ waḏamm naḥwa al-ʿāqil wa-l-jāhil) In the examples produced by Ibn Fāris, the aim is indeed to complete a definite term (here the proper name Zayd) by a term that itself is definite (al-ʿaṭṭār and al-tamīmī).

10

11 12 13

I have chosen this translation for the term in order to retain the etymology of Latin species, which denotes an element within a class at a lower level/from a lower level, on the understanding that a species, i.e. an element within a class, is less general than the class itself and thus is more definite, which is what is at stake with concepts of taḫlīṣ (and taḫṣīṣ, and so on). Troupeau (1976:85) records only one occurrence each of ḫallaṣa and ḫallaṣa min, the former in the sense of ‘to clarify’, the latter in that of ‘to get free from’. See Mubarrad, Muqtaḍab II, 567 and Zajjājī, Lāmāt 114. The appearance, at that time, of these words is a striking manifestation of the introduction of logic in the Arab world. This is true of taḫlīṣ understood as ‘specification’ (i.e., from genus to species).

258

sartori

In this connection it is important to note that in other authors, particularly in the family of treatises related to al-Zajjājī’s Jumal, taḫlīṣ, as a single term, is replaced by a noun phrase from the root š-r-k, whose trace is found in the quotation from Ibn Fāris cited above. This is the case of Ibn Ḫarūf (d. 609/1212), who uses the expression rafʿ al-ištirāk ‘to remove the equivocity’ in connection with naʿt (Šarḥ I, 300):14 “The adjective serves to particularize the indefinite expression and to remove the supposed equivocity concerning the definite qualified expression” (wa-fāʾidat al-naʿt taḫṣīṣ al-nakira wa-rafʿ al-ištirāk al-mutawahham fī l-manʿūt al-maʿrifa). In a similar context, Ibn ʿUṣfūr (d. 669/1271) uses the expression ʾizālat ištirāk (Šarḥ I, 142): “According to the grammarians, the adjective designates a noun or what equals a noun that follows what precedes it in order to particularize an indefinite expression or to dispel the possible equivocity of a definite expression” (al-naʿt ʿinda al-naḥwiyyīn ʿibāra ʿan ism ʾaw mā huwa fī taqdīr ism yatbaʿu mā qablahu li-taḫṣīṣ nakira ʾaw li-ʾizālat ištirāk ʿāriḍ fī maʿrifa).15 The transition is found in the writings of ʿAlī ibn Muḥammad al-Jurjānī al-Sayyid al-Šarīf (d. 816/1413), in which the same identification of the phenomenon is present, but with a different term. He writes (Taʿrīfāt 73): “Clarification denotes the fact of removing the ellipsis that happens in definite expressions” (al-tawḍīḥ ʿibāra ʿan rafʿ al-ʾiḍmār al-ḥāṣil fī l-maʿārif ). 3.2.2 Tawḍīḥ After Ibn Fāris, taḫlīṣ apparently disappears in favor of tawḍīḥ ‘clarification’.16 Its first occurrence is in ʿAbd al-Qāhir al-Jurjānī (d. 471/1078). Dealing with qualifiying praise, in particular praise reserved for Allāh, he writes (Dalāʾil 44), following Ibn Jinnī: “To qualification belongs that qualification which contains neither particularization nor clarification” (wa-ʾanna min al-ṣifa ṣifa lā yakūnu fīhā taḫṣīṣ wa-lā tawḍīḥ). We indeed identify here a pair formed of taḫṣīṣ on the one hand and of tawḍīḥ on the other instead of taḫlīṣ. The author is even more precise about the adjective in his Šarḥ al-Jumal (Šarḥ 276, see also Muqtaṣid II, 175):

14 15

16

‘Polysemy’ in linguistics (see Larcher 2011:307, n. 4), is what in logic is called ‘equivocity’. In Ibn ʿUṣfūr, taḫlīṣ does not appear with the technical sense identified elsewhere, as evidenced by the following passage where it has the general meaning of ‘specification’, but not the technical one as connected to taḫṣīṣ and taʿrīf/tankīr: “and it is the specification of the future” (wa-huwa al-taḫlīṣ li-l-istiqbāl, Šarḥ II, 74). This term is cited twice by Gätje (1970:235, 239), who translates it similarly as “Verdeutlichung oder Explikation”, i.e. ‘clarification’ or ‘explication’.

definition and determination in medieval arabic thought

259

Know then that, with respect to indefinite expressions, the adjective conveys particularization and, with respect to definite expressions, clarification. The explanation for this is that when you say marartu bi-rajulin ṭawīlin ‘I passed by a tall man’, you reduce the generality of the noun, applying it to only some of [its] species rather than to its entirety as you do not include in it any man who is not tall. This is what is meant by particularization, and it only occurs with the indefinite expression […]. Clarification, on the other hand, occurs in definite expressions. When you say, for instance, jāʾanī zaydun al-ṭawīlu ‘the tall Zayd came to me’, you only need the qualification when there are two men, each one of them called Zayd, and you want to make clear to the interlocutor that you are referring to the taller one of the two. This is elimination of ambiguity and clarification, rather than particularization, since particularization, as we have mentioned, means to single out one part from a genus. A proper name is a noun referring to a thing in itself; it does not signify the genus, which would make it possible to imagine its particularization (ṯumma iʿlam ʾanna l-ṣifa tufīdu fī l-nakira al-taḫṣīṣ wa-fī l-maʿrifa al-tawḍīḥ tafsīr hāḏā ʾannaka ʾiḏā qulta marartu bi-rajulin ṭawīlin kunta qad naqaṣta min ʿumūm al-ism fa-jaʿaltahu yaqaʿu ʿalā baʿḍ al-jins dūna kullihi min ḥayṯu lā tudḫilu man lā yakūnu ṭawīlan min al-rijāl fīhi fa-hāḏā huwa l-murād bil-taḫṣīṣ wa-lā yakūnu ʾillā fī l-nakira […] wa-l-tawḍīḥ fī l-maʿrifa fa-huwa ʾannaka ʾiḏā qulta jāʾanī zaydun al-ṭawīlu fa-ʾinnaka ʾinnamā taḥtāju ʾilā l-ṣifa ʾiḏā kāna hunāka rajulāni kull wāḥid minhumā yusammā zaydan faʾanta turīdu ʾan tubayyina li-l-muḫāṭab ʾannaka ʿanayta minhumā allaḏī huwa ṭawīl fa-kāna ḏālika ʾizāla li-l-labs wa-tawḍīḥan wa-lā yakūnu taḫṣīṣan li-ʾanna l-taḫṣīṣ kamā ḏakarnā huwa ʾan naḫuṣṣa min al-jins baʿḍahu wa-l-ʿalam yakūnu sman li-šayʾ bi-ʿaynihi wa-lā yadullu ʿalā jins ḥattā yataṣawwara fīhi l-taḫṣīṣ). In doing so, al-Jurjānī is the first to be clear about the distinction to be made between taḫṣīṣ and tawḍīḥ. We find the same two notions being used by alZamaḫšarī (Mufaṣṣal 148), who writes about the adjective: “It is said that it [the qualification] is used for particularization within the indefinite expressions and for clarification within the definite ones” (wa-yuqālu ʾinnahā li-l-taḫṣīṣ fī l-nakirāt wa-li-l-tawḍīḥ fī l-maʿārif ).17 The same distribution is found in Ibn

17

Incidentally, one may note that al-Zamaḫšarī (Mufaṣṣal 158) uses the same lexical root in Form II, in the shape of a conjugated verb, when he talks about the explanatory apposition: wa-wurūd al-ṯānī min ʾajl ʾan yuwaḍḍiḥa ʾamrahu.

260

sartori

Yaʿīš’ (d. 643/1245) commentary on the Mufaṣṣal (Šarḥ II, 233), again about the adjective: “The fact is that it [the qualification] is used for particularization at the level of indefinite expressions and for clarification at the level of definite expressions, as we have mentioned” (ʾinnahā li-l-taḫṣīṣ fī l-nakirāt wali-l-tawḍīḥ fī l-maʿārif ʿalā mā ḏakarnāhu). Likewise, Ibn al-Ḥājib (d. 646/1249) in his ʾImlāʾ ʿalā l-Kāfiya (the autocommentary he made of his Kāfiya, which in its turn is an epitome extracted from al-Zamaḫšarī’s Mufaṣṣal), uses the same pair of terms with respect to the adjective (ʾImlāʾ 48a/3; Kāfiya 129): “His words ‘it conveys particularization or clarification’ [mean that] particularization concerns indefinite expressions and that clarification concerns definite expressions” (qawluhu wa-fāʾidatuhu taḫṣīṣ ʾaw tawḍīḥ fa-l-taḫṣīṣ fī l-nakirāt wal-tawḍīḥ fī l-maʿārif ). Finally, to conclude with the family of treatises related to the Mufaṣṣal, Raḍī l-Dīn al-ʾAstarābāḏī (d. 686/1287 or more likely 688/1289) states (Šarḥ III, 314): The meaning of ‘particularization’ in their [i.e., the Arab grammarians’] terminology is to restrict the equivocity that occurs at the level of indefinite expressions. Thus, when you say jāʾanī rajulun ṣāliḥun ‘a pious man came to me’, according to the imposition of language the word rajul is applicable to all individuals of this species, and by saying ṣāliḥ ‘pious’, you reduce the possible equivocity. According to them [the grammarians], the meaning of ‘clarification’ is to remove the possible equivocity occuring in definite expressions, regardless of whether or not they are proper names, as in zaydun al-ʿālimu ‘Zayd the scholar’ or al-rajulu l-fāḍilu ‘the virtuous man’ (maʿnā taḫṣīṣ fī ṣṭilāḥihim taqlīl al-ištirāk al-ḥāṣil fī l-nakirāt wa-ḏālika ʾanna rajul fī qawlika jāʾanī rajulun ṣāliḥun kāna bi-waḍʿ al-wāḍiʿ muḥtamalan li-kull fard min ʾafrād hāḏā l-nawʿ fa-lammā qulta ṣāliḥ qallalta l-ištirāk wa-l-iḥtimāl wa-maʿnā l-tawḍīḥ ʿindahum rafʿ al-ištirāk alḥāṣil fī l-maʿārif ʾaʿlāman kānat ʾaw lā naḥwa zaydun al-ʿālimu wa-l-rajulu l-fāḍilu). In another family of Arabic grammatical treatises, that of the ʾAlfiyya, the term tawḍīḥ is used in the same way by Ibn Hišām al-ʾAnṣārī (d. 761/1360), who writes about the adjective (Sabīl 416): It conveys particularization, description, praise, blame, pity, or corroboration. The adjective conveys either particularization of an indefinite expression, as in marartu bi-rajulin kātibin ‘I passed by a writing man’, or clarification of a definite one, as in marartu bi-zaydin al-ḫayyāṭi ‘I passed by Zayd the tailor’ (wa-fāʾidatuhu taḫṣīṣ ʾaw tawḍīḥ ʾaw madḥ ʾaw ḏamm

definition and determination in medieval arabic thought

261

ʾaw taraḥḥum ʾaw tawkīd. fāʾidat al-naʿt ʾimmā taḫṣīṣ nakira ka-qawlika marartu bi-rajulin kātibin ʾaw tawḍīḥ maʿrifa ka-qawlika marartu bi-zaydin al-ḫayyāṭi ʾaw madḥ …)18 Finally, two features of the complementary term taḫṣīṣ may be noted here. Firstly, tawḍīḥ can be replaced by ʾīḍāḥ, a term derived from the same consonantal root, but derived from Form IV, which is found especially in Ibn Hišām al-ʾAnṣārī (Sabīl 435). He states about the explanatory apposition that “of every noun we can say that it is an explanatory apposition conveying elucidation or particularization” (kull ism ṣaḥḥ al-ḥukm ʿalayhi bi-ʾannahu ʿaṭf bayān mufīd lil-ʾīḍāḥ ʾaw li-l-taḫṣīṣ). Likewise, Ibn ʿAqīl (d. 769/1367) writes (Šarḥ II, 57f.): “The explanatory apposition is the frozen apposition that looks like a qualification in elucidating the element to which it is apposed […], since it is a clarifier” (wa-ʿaṭf al-bayān huwa l-tābiʿ al-jāmid al-mušbih li-l-ṣifa fī ʾīḍāḥ matbūʿihi […] li-ʾannahu muwaḍḍiḥ).19 Secondly, we should note two significant exceptions. The first is represented by Ibn al-ʾAnbārī (d. 577/1181), who writes in bāb al-waṣf (ʾAsrār 155): If someone asks ‘what is the purpose of qualification?’, he is told that it is particularization and distinction. Thus, if it is a definite expression, the purpose of qualification is particularization, because of the inherent equivocity. Don’t you see that there are many people called ‘Zayd’, or something similar, so that when we say jāʾanī zaydun ‘Zayd came to me’, it is not known which one of them we mean? Thus, when we say zaydun al-ʿāqilu ‘Zayd the intelligent’ or al-ʿālimu ‘the learned’ or al-ʾadību ‘the educated’, or something similar, we single him out from among the others. Now, if the noun is an indefinite expression, the purpose of qualification is distinction. Don’t you see that when you say jāʾanī rajulun ‘a man came to me’, it is not known which man is meant, and that when you

18

19

Here is the translation in French by Goguyer (1887:323f.): “Il sert à particulariser, décrire, louer, blâmer, apitoyer, corroborer. Le qualificatif sert à particulariser un nom indéterminé, ex. marartu bi-rajulin kātibin, décrire l’objet d’ un nom déterminé, ex. marartu bizaydin al-ḫayyāṭi”. I do not choose to translate tawḍīḥ by ‘to describe’, as Goguyer does, since ‘to clarify’ is more appropriate, nor to translate nakira and maʿrifa by ‘indeterminate’ and ‘determinate’ (see above, p. 254 and n. 8). Besides, this is what we read in a contemporary dictionary of grammatical terms about the explanatory apposition (Badīʿ Yaʿqūb and ʿĀṣī 1987:II, 868): “the explanatory apposition serves to clarify the term to which it is attached if it is a definite expression” ( yufīdu ʿatf al-bayān ʾīḍāḥ matbūʿihi ʾin kāna l-matbūʿ maʿrifa).

262

sartori

say rajulun ʿāqilun ‘an intelligent man’, you distinguish him from those who do not possess this qualification, and that it is not a matter of particularizing him, because by distinguishing we mean a specific entity, which was not intended here? (ʾin qāla qāʾil mā al-ġaraḍ fī l-waṣf qīla al-taḫṣīṣ wa-l-tafḍīl fa-ʾin kāna maʿrifa kāna l-ġaraḍ min al-waṣf al-taḫṣīṣ li-ʾanna lištirāk yaqaʿu fīhā ʾa-lā tarā ʾanna l-musammīn bi-zayd wa-naḥwihi kaṯīr fa-ʾiḏā qāla jāʾanī zaydun lā yuʿlamu ʾayyuhum yurīdu fa-ʾiḏā qāla zaydun al-ʿāqilu ʾaw al-ʿālimu ʾaw al-ʾadību wa-mā ʾašbaha ḏālika fa-qad ḫaṣṣahu min ġayrihi wa-ʾin kāna l-ism nakira kāna l-ġaraḍ min al-waṣf al-tafḍīl ʾalā tarā ʾannaka ʾiḏā qulta jāʾanī rajulun lam yuʿlam ʾayy rajul huwa fa-ʾiḏā qulta rajulun ʿāqilun fa-qad faḍḍaltahu ʿalā man laysa lahu hāḏā l-waṣf wa-lam taḫuṣṣahu li-ʾannā naʿnī bi-l-tafḍīl šayʾan bi-ʿaynihi wa-lam nuridhu hāhunā) This is indeed a remarkable exception to the extent that it implies a reversal compared to all other grammarians, since taḫṣīṣ designates here what the others call taḫlīṣ, or later tawḍīḥ (and even ʾīḍāḥ) and since tafḍīl, a term never met in other grammarians in the technical sense that concerns us, refers precisely to what others call taḫṣīṣ. The second is to be found in Ibn Mālik. Indeed, whereas tawḍīḥ is found in Ibn Hišām al-ʾAnṣārī, as we have seen, when he is commenting Ibn Mālik, the latter explicitly proposes another pair of terms, in which taḫṣīṣ is opposed to tawkīd. Thus, he writes (Šarḥ II, 489): The qualification is generally used to particularize what it follows as in uhjuranna zaydan al-baḏī ‘get away from Zayd the obscene!’ and it can convey […] the confirmation of what precedes (wa-l-naʿt ġāliban li-taḫṣīṣ allaḏī yatlūhu ka-hjuranna zaydan l-baḏī wa-qad yufīdu […] tawkīd mā taqaddama). Here, the particularity is not only the appearance of a new term. The terms appear in fact to be reversed, compared to taḫṣīṣ-tawḍīḥ as it is found elsewhere, in particular in his commentator Ibn Hišām al-ʾAnṣārī. This is confirmed by what Ibn Mālik writes in the commentary on his Kāfiya al-Šāfiya, since in connection with the adjective, taḫṣīṣ is used within the framework of definition, whereas tawkīd is used within the framework of indefinition (Šarḥ I, 520): “Particularization is like al-šiʿrā al-ʿabūr ‘Canis Minor’20 […] and simple confir-

20

Name of the constellation, see Kazimirski (1860:II, 154).

definition and determination in medieval arabic thought

263

mation is like lā tattaḫiḏū ʾilāhayni ṯnayni ‘Take not for worship two gods, two!’” (al-taḫṣīṣ ka-l-šiʿrā al-ʿabūr […] wa-mujarrad al-tawkīd naḥwa lā tattaḫiḏū ʾilāhayni ṯnayni).21 On the basis of all of these sources, except for the special cases of Ibn alʾAnbārī and Ibn Mālik, which however concern only the terminological level, the following technical definition may be given of taḫlīṣ, and later of tawḍīḥ: taḫlīṣ means to specify a definite term by another one, itself definite, within the framework of a qualification in the broad sense, that is to say an attributive adjective (al-rajul al-ṭawīl), including a relative sentence (al-rajul allaḏī yaktubu risāla), or an explanatory apposition (ʾaqsama bi-l-Lāhi ʾabū ḥafsin ʿumarin). We note the asymmetry between this definition and that of taḫṣīṣ (see above), since annexation is not mentioned in the definition of taḫlīṣ. 3.3 Taʿrīf Among the authors using taḫṣīṣ, the complementary term to it within the special framework of annexation, is not taḫlīṣ nor tawḍīḥ, as we have seen within the framework of (broad) qualification, rather, it is taʿrīf. Thus, Ibn Jinnī writes (Ḫaṣāʾiṣ II, 267): “It has been said that the purpose of annexation is only to define or to particularize” (qīla li-ʾanna l-ġaraḍ fī l-ʾiḍāfa ʾinnamā huwa l-taʿrīf wa-l-taḫṣīṣ). Here, the pair of terms consists of taʿrīf and taḫṣīṣ and, therefore, in annexation taʿrīf seems to be to taḫṣīṣ what taḫlīṣ is to taḫṣīṣ in qualification. Accordingly, taʿrīf is in a situation of structural homology with taḫlīṣ. Ibn Jinnī says elsewhere (Sirr II, 37) that “annexation imparts definition and particularization” (al-ʾiḍāfa tuksibu l-taʿrīf wa-l-taḫṣīṣ). Ibn Mālik uses the same pair of terms, but is clearer about the identity of the terms involved from the point of view of definiteness (Sarḥ I, 408): All of this belongs to those things whose annexation is semantic, real, and pure, since it has the effect of defining the first term in an annexation, if the second term is a definite expression, and of particularizing the first term, if the second is an indefinite one ( fa-hāḏā kulluhu mimmā ʾiḍāfatuhu maʿnawiyya wa-ḥaqīqiyya wa-maḥḍa li-ʾannahā muʾaṯṯira fī lmuḍāf taʿrīfan ʾin kāna l-ṯānī maʿrifa wa-taḫṣīṣan ʾin kāna l-ṯānī nakira). He is followed in this by Ibn Hišām al-ʾAnṣārī (Sabīl 377 f.):

21

Q. 16/51. Here, tawkīd applies to a definite expression (ʾilāhayn), qualified by an element itself indefinite (iṯnayn), where the other grammarians use taḫlīṣ-tawḍīḥ.

264

sartori

It is called semantic annexation because it defines or particularizes […]. It is definition if the second term of annexation is a definite expression, as in ġulāmu zaydin ‘Zayd’s servant’, and it is particularization if it is an indefinite one, as in ġulāmu mraʾatin ‘a woman’s servant’ […]. It [i.e. the ʾiḍāfa lafẓiyya ‘formal annexation’] conveys neither definition nor particularization (tusammā maʿnawiyya li-ʾannahā li-l-taʿrīf ʾaw al-taḫṣīṣ […] wa-huwa al-taʿrīf ʾin kāna l-muḍāf ʾilayhi maʿrifa naḥwa ġulāmu zaydin wa-l-taḫṣīṣ ʾin kāna l-muḍāf ʾilayhi nakira ka-ġulāmu mraʾatin […] wa-lā tufīdu taʿrīfan wa-lā taḫṣīṣan) Likewise, Ibn ʿAqīl (Šarḥ I, 368f.) states: Pure [annexation] is what is not like this. It conveys the first term of the annexation with particularization if the second term of the annexation is an indefinite expression, as in hāḏā ġulāmu mraʾatin ‘this is a woman’s servant’, and [it conveys] definition if the second term of annexation is a definite expression, as in hāḏā ġulāmu zaydin ‘this is Zayd’s servant’. Thus, it [the first class, i.e. pure annexation] conveys particularization or definition (wa-l-maḥḍa [al-ʾiḍāfa] mā laysat ka-ḏālika wa-tufīdu l-ism al-ʾawwal taḫṣīṣan ʾin kāna l-muḍāf ʾilayhi nakira naḥwa hāḏā ġulāmu mraʾatin wataʿrīfan ʾin kāna l-muḍāf ʾilayhi maʿrifa naḥwa hāḏā ġulāmu zaydin […] fa-yufīdu taḫṣīṣan ʾaw taʿrīfan) The same view on annexation in found in al-Zamaḫšarī (Mufaṣṣal 119). Annexation of a noun to a noun is of two types, semantic and formal. Semantic annexation conveys definition, as in dāru ʿamrin ‘ʿAmr’s house’, and [it conveys] particularization, as in ġulāmu rajulin ‘a man’s servant’ (ʾiḍāfat al-ism li-sm ʿalā ḍarbayn maʿnawiyya wa-lafẓiyya fa-l-maʿnawiyya mā ʾafāda taʿrīfan ka-qawlika dāru ʿamrin ʾaw taḫṣīṣan ka-qawlika ġulāmu rajulin) Ibn Yaʿīš (Šarḥ II, 126) says the same about annexation, and so do Ibn al-Ḥājib (Kāfiya 122) and Raḍī l-Dīn al-ʾAstarābāḏī (Šarḥ I, 202; II, 238f.). Finally, we find the same view in later authors like al-Jārburdī (d. 746/1346), who states (Muġnī 35): Semantic annexation conveys definition of the first term of the annexation when it is annexed to a definite expression, like ġulāmu zaydin ‘Zayd’s servant’, and particularization of it when it is annexed to an indefinite

definition and determination in medieval arabic thought

265

expression, like ġulāmu rajulin ‘a man’s servant’ (wa-l-ʾiḍāfa al-maʿnawiyya tufīdu taʿrīf al-muḍāf ʾiḏā ʾuḍīfa ʾilā l-maʿrifa naḥwa ġulāmu zaydin wa-taḫṣīṣahu ʾiḏā ʾuḍīfa ʾilā l-nakira naḥwa ġulāmu rajulin). Likewise, al-Sayyid al-Šarīf (Taʿrīfāt 32) defines annexation as follows: “annexation is joining two nouns in such a way that it conveys definition or particularization” (al-ʾiḍāfa hiya imtizāj ismayn ʿalā wajh yufīdu taʿrīfan ʾaw taḫṣīṣan). Thus, it appears that within the framework of annexation, the terminological pair is indeed taʿrīf/taḫṣīṣ, of which the former corresponds to the annexation of a definite term to an indefinite one (ġulāmu l-rajuli), and the latter to the annexation of an indefinite term to an indefinite one (ġulāmu rajulin). Accordingly, in annexation taʿrīf is to taḫṣīṣ what taḫlīṣ is to taḫṣīṣ in qualification. Moreover, the term taḫṣīṣ, used for both annexation and (broad) qualification in the indefinite framework, the terms taḫlīṣ-tawḍīḥ-ʾīḍāḥ used exclusively for (broad) qualification in the definite framework, and the term taʿrīf used exclusively for annexation, appear to be as many forms of ‘determination’, the last mentioned case conveying definition and determination at the same time. At this point, no single term seems therefore to express the concept of determination exclusively; rather, this concept is distributed between taḫṣīṣ, on the one hand, and taḫlīṣ-tawḍīḥ (and, more marginally, ʾīḍāḥ), on the other.

4

Takmīl, or Completion as ‘Determination’

There remains a final term to be studied in relation to the categories of definition and indefinition (taʿrīf/tankīr). Ibn Hišām al-ʾAnṣārī is apparently the first to subsume explicitly under the term of takmīl ‘completion’22 the processes of tawḍīḥ and taḫṣīṣ, since he writes (ʾAwḍaḥ III, 223): The coordinated appositive and the permutative are excluded, by the restriction of completion. […] What is meant by ‘what completes’ is what clarifies a definite expression, such as jāʾa zaydun al-tājiru ‘Zayd the merchant came’ or al-tājiru ʾabūhu ‘whose father is the merchant’, and what particularizes an indefinite expression, such as jāʾanī rajulun tājirun ‘a merchant man came to me’ or tājirun ʾabūhu ‘whose father

22

The term mukammil is also found in Ibn al-Dahhān (d. 569/1174), Ġurra II, 854, in connection with ʿaṭf al-bayān.

266

sartori

is a merchant’ ( fa-ḫaraja bi-qayd al-takmīl al-nasaq wa-l-badal […] wal-murād bi-l-mukammil al-muwaḍḍiḥ li-l-maʿrifa ka-jāʾa zaydun al-tājiru ʾaw al-tājiru ʾabūhu wa-l-muḫaṣṣiṣ li-l-nakira ka-jāʾanī rajulun tājirun ʾaw tājirun ʾabūhu) It thus appears that takmīl represents indeed the generic term and hyperonym of both processes of tawḍīḥ and taḫṣīṣ. It seems that the first appearance of takmīl (in the technical sense as well as absolutely) is found in Ibn Mālik. It appears in connection with the adjective (naʿt), on the one hand, and with the explanatory apposition (ʿaṭf al-bayān), on the other: “except that the adjective leads to this completion because it indicates a meaning in the qualified element […]; the qualification is then what completes the term it follows, and the completed item is what is followed [by the adjective]” (ʾillā ʾanna l-naʿt yuwaṣṣilu ʾilā ḏālika l-takmīl bi-dalālatihi ʿalā maʿnan fī l-manʿūt […] fa-l-naʿt al-mukammil matbūʿahu […] wa-l-mukammal matbūʿuhu, Šarḥ I, 516) and “the explanatory apposition is an appositive term which follows the course of the qualification in terms of completion of the element it follows” (ʿaṭf al-bayān tābiʿ yajrī majrā l-naʿt fī takmīl matbūʿihi, Šarḥ I, 532). The term takmīl seems to be used only by these two authors, but it is an interesting term because it encompasses taḫṣīṣ and tawḍīḥ. This applies, however, only to the framework of (broad) qualification, not to that of annexation. This prompts us to distinguish, under taḫṣīṣ, that which is opposed and complementary to taḫlīṣ-tawḍīḥ-ʾīḍāḥ (= taḫṣīṣ1), from that which is opposed and complementary to taʿrīf (= taḫṣīṣ2).

5

Conclusion

As I have noted elsewhere, though less precisely (Sartori 2018), taḫṣīṣ is an intersection to tankīr and taʿrīf. As a matter of fact, if the process of taḫṣīṣ applies indeed to an indefinite noun, it does not fall under indefiniteness.23 However, it does not belong to the domain of definition (taʿrīf ) either, since for the latter it constitutes the complementary term. The question arises whether this makes it an equivalent of ‘determination’ (whether almost or partial determination, as 23

As may be seen, among other authors, in Ibn Jinnī (Ḫaṣāʾiṣ II, 447): “and also, the fact is that nunation indicates indefinition and that annexation is instituted for particularization, so how can you combine them despite what we have remarked about them?” (wa-ʾayḍan fa-ʾinna l-tanwīn dalīl al-tankīr wa-l-ʾiḍāfa mawḍūʿa li-l-taḫṣīṣ fa-kayfa laka bi-jtimāʿihimā maʿa mā ḏakarnā min ḥālihimā).

definition and determination in medieval arabic thought

267

proposed by Reckendorf or Wright) or a weak definition (according to Carter). It seems that this question should be answered in the negative. Rather, it ought to be reaffirmed that the couple taʿrīf/tankīr is indeed that of definition/indefiniteness. Under this pair, while no term seems to exist in the Arabic grammatical metalanguage to signify ‘indetermination’, the second member of the pair, ‘determination’, seems to correspond to many Arabic terms. It is taḫṣīṣ ‘particularization’ within the framework of indefiniteness, and taḫlīṣ-tawḍīḥ (and even, but more marginally, ʾīḍāḥ) ‘specification’, ‘clarification’ within that of definition, but also taʿrīf ‘definition’ for the particular case of annexation within the definite framework. Only one author, Ibn Hišām al-ʾAnṣārī, assigns a special status to taḫlīṣ-tawḍīḥ-ʾīḍāḥ and to taḫṣīṣ1 (outside the framework of annexation), bringing together these terms under the label of takmīl ‘completion’. As for taʿrīf understood as definition, within the framework of annexation it is complementary to taḫṣīṣ2. It then applies to an indefinite term within the annexation construct, which is transferred by it from indefiniteness to definition.24 Therefore, Arab grammarians felt that something else than the mere opposition tankīr/taʿrīf was at work. Obviously without using terms equivalent to our pair ‘indetermination/determination’, they came close to ‘determination’ through taḫṣīṣ ‘particularization’, taḫlīṣ ‘specification’, tawḍīḥ-ʾīḍāḥ ‘clarification-elucidation’ and takmīl ‘completion’. In the absence of any other, takmīl seems to be best able to evoke a generic form of ‘determination’, understood as a predicative determination.25 This is represented in Figure 1. This is opposed to Kouloughli’s (2001:40) reading, for whom specification (particularization-taḫṣīṣ) is not determination, and for whom kalb in expressions like kalbu zaydin ‘Zayd’s dog’ is definite and indeterminate, while in expressions like kalbu ṣaydin ‘a hunting dog’ it is indefinite and indeterminate. We should rather consider them both determinate, accepting that taḫṣīṣ within the domain of annexation is indeed a determination. The schema proposed by Kouloughli would therefore be replaced by the one in Figure 2.

24

25

Note that an indefinite expression can be determinate (a tall man, where man is determinate by a (quantification) and tall (qualification)) or indeterminate (man), while a definite expression is necessarily determinate (the man). See Morais Barbosa (1998).

268

sartori tankīr

taʿrīf ↗ al-rajul

rajul annexation

→ taʿrīf

rajul → taḫṣīṣ2 → rajul madīna

↘ rajul almadina

qualification rajul → taḫṣīṣ1 → rajul ṭawīl al-rajul → taḫlīṣ→ al-rajul altawḍīḥ ʾīḍāḥ ṭawīl explanatory rajul → → rajul tājir zayd → → zayd al-tājir apposition takmīl figure 1

tankīr ‘indefiniteness’; taʿrīf ‘definition’; taḫṣīṣ ‘particularization’; taḫlīṣ ‘specification’; tawḍīḥ ‘clarification’; ʾīḍāḥ ‘elucidation’; takmīl ‘completion’

Determinate ↙ Indefinite kalb(un) kalb(un) jamīl(un) kalb(u) ġulām(in) figure 2

↘ Definite al-kalb(u) al-kalb(u) al-jamīl(u) kalb(u) al-ġulām(i) / kalb(u) zayd(in)

Schema of ‘determination’

Bibliographical References A

Primary Sources

ʾAstarābāḏī, Šarḥ = Raḍi l-Dīn Muḥammad ibn al-Ḥasan al-ʾAstarābāḏī, Šarḥ Kāfiyat Ibn al-Ḥājib. Ed. by ʾImīl Badīʿ Yaʿqūb. 5 vols. Beirut: Dār al-Kutub al-ʿIlmiyya, 1998. Ibn ʿAqīl, Šarḥ = Bahāʾ al-Dīn ʿAbdallāh ibn ʿAbd al-Raḥmān ibn ʿAbdallāh ibn Muḥammad Bahāʾ al-Dīn al-Qurašī al-Hāšimī al-ʿAqīlī al-Hamdānī al-Miṣrī Ibn ʿAqīl, Šarḥ Ibn ʿAqīl ʿalā ʾAlfiyyat Ibn Mālik. Ed. ʾĪmīl Badīʿ Yaʿqūb. 7th ed. 2 vols. Beirut: Dār alKutub al-ʿIlmiyya, 2010. Ibn al-Dahhān, Ġurra = ʾAbū Muḥammad Saʿīd ibn al-Mubārak ibn ʿAlī al-ʾAnṣārī almaʿrūf bi-Ibn al-Dahhān al-Baġdādī, al-Ġurra fī šarḥ al-Lumaʿ min ʾawwal bāb ʾinna wa-ʾaḫawātihā ʾilā ʾāḫir bāb al-ʿaṭf. Ed. by Farīd ʿAbd al-ʿAzīz Al-Zālim al-Sulaym. 2 vols. Riyadh: Dār al-Tadmūriyya, 2011. Ibn al-Ḥājib, ʾImlāʾ = Jamāl al-Dīn ʾAbū ʿAmr ʿUṯmān ibn ʿUmar ibn ʾAbī Bakr Ibn alḤājib al-Miṣrī al-Dimašqī al-Mālikī, al-ʾImlāʾ ʿalā l-Kāfiya fī l-naḥw. Ed. by Manuel Sartori. [unpublished], 2012.

definition and determination in medieval arabic thought

269

Ibn al-Ḥājib, Kāfiya = Jamāl al-Dīn ʾAbū ʿAmr ʿUṯmān ibn ʿUmar ibn ʾAbī Bakr Ibn alḤājib al-Miṣrī al-Dimašqī al-Mālikī, al-Kāfiya fī l-naḥw. Ed. by Ṭāriq Najm ʿAbdallāh. Jeddah: Maktabat Dār al-Wafāʾ, 1986. Ibn al-ʾAnbārī, ʾAsrār = Kamāl al-Dīn ʾAbū l-Barakāt ʿAbd al-Raḥmān ibn Muḥammad ibn ʿUbayd Allāh al-ʾAnṣārī al-ʾAnbārī, ʾAsrār al-ʿarabiyya. Ed. by Muḥammad Ḥusayn Šams al-Dīn. Beirut: Dār al-Kutub al-ʿIlmiyya, 1997. Ibn Fāris, Ṣāḥibī = ʾAbū l-Ḥusayn ʾAḥmad Ibn Fāris ibn Zakariyāʾ al-Qazwīnī al-Rāzī, alṢāḥibī fī fiqh al-luġa al-ʿarabiyya wa-masāʾilihā wa-sunan al-ʿArab fī kalāmihim. Ed. by ʾAḥmad Ḥasan Basj. Beirut: Dār al-Kutub al-ʿIlmiyya, 1997. Ibn Jinnī, Ḫaṣāʾiṣ = ʾAbū l-Fatḥ ʿUṯmān Ibn Jinnī al-Mawṣilī, al-Ḫaṣāʾiṣ. Ed. by ʿAbd alḤamīd Hindāwī. 3rd ed. 3 vols. Beirut: Dār al-Kutub al-ʿIlmiyya, 2008. Ibn Jinnī, Sirr = ʾAbū l-Fatḥ ʿUṯmān Ibn Jinnī al-Mawṣilī, Sirr ṣināʿat al-ʾiʿrāb. Ed. by Muḥammad Ḥasan Muḥammad Ḥasan ʾIsmāʿīl and ʾAḥmad Rušdī Šaḥāta ʿĀmir. 2nd ed. 2 vols. Beirut: Dār al-Kutub al-ʿIlmiyya, 2007. Ibn Ḫarūf, Šarḥ = ʾAbū l-Ḥasan ʿAlī ibn Muḥammad ibn ʿAlī Ibn Ḫarūf al-ʾIšbīlī, Šarḥ Jumal al-Zajjājī. Ed. by Salwā Muḥammad ʿUmar ʿArab. Mecca: Jāmiʿat ʾUmm alQurā, 1998. Ibn Hišām, Sabīl = Jamāl al-Dīn ʾAbū Muḥammad ʿAbdallāh ibn Yūsuf al-ʾAnṣārī Ibn Hišām, Sabīl al-hudā ʿalā šarḥ Qaṭr al-nadā wa-ball al-ṣadā wa-maʿa-hu Risāla fī madḥ al-naḥw. Ed. by Muḥammad Muḥyī l-Dīn ʿAbd al-Ḥamīd and ʿAbd al-Jalīl alʿAṭā al-Bakrī. Damascus: Maktabat Dār al-Fajr, 2001. Ibn Hišām, ʾAwḍaḥ = Jamāl al-Dīn ʾAbū Muḥammad ʿAbdallāh ibn Yūsuf al-ʾAnṣārī Ibn Hišām, ʾAwḍaḥ al-masālik ilā ʾAlfiyyat Ibn Mālik. Ed. by H. al-Fāḫūrī. 4 vols. Beirut: Dār al-Jīl, 1989. Ibn Mālik, Šarḥ = Jamāl al-Dīn ʾAbū ʿAbdallāh Muḥammad ibn ʿAbdallāh ibn ʿAbd Allāh al-Ṭāʾī al-Jayyānī al-ʾAndalusī Ibn Mālik, Šarḥ al-Kāfiya al-Šāfiya. Ed. by ʿAlī Muḥammad Muʿawwaḍ and ʿĀdil ʾAḥmad ʿAbd al-Mawjūd. 2nd ed. 2 vols. Beirut: Dār alKutub al-ʿIlmiyya, 2010. Ibn Yaʿīš, Šarḥ = Muwaffaq al-Dīn ʾAbū l-Baqāʾ Yaʿīš ibn ʿAlī Ibn Yaʿīš al-ʾAsadī al-Ḥalabī, Šarḥ al-Mufaṣṣal li-l-Zamaḫšarī. 2nd ed. 6 vols. Ed. by ʾImīl Badīʿ Yaʿqūb. Beirut: Dār al-Kutub al-ʿIlmiyya, 2011. Ibn ʿUṣfūr, Šarḥ = ʾAbū l-Ḥasan ʿAlī ibn al-Muʾmin ibn Muḥammad al-Ḥaḍramī al-ʾIšbīlī Ibn ʿUṣfūr, Šarḥ Jumal al-Zajjājī. Ed. by Fawwāz al-Šaʿʿār and ʾĪmīl Badīʿ Yaʿqūb. 3 vols. Beirut: Dār al-Kutub al-ʿIlmiyya, 1998. Jārburdī, Muġnī = Faḫr al-Dīn ʾAbū l-Makārim ʾAḥmad ibn al-Ḥasan ibn Yūsuf alJārburdī al-Šāfiʿī, al-Muġnī fī ʿilm al-naḥw. Ed. by Qāsim al-Mūšī ʾAbū Muḥammad ʾAnas. Beirut: Dār Ṣādir and Istanbul: Maktabat al-ʾIršād, 2007. Jurjānī, Dalāʾil = ʾAbū Bakr ʿAbd al-Qāhir ibn ʿAbd al-Raḥmān ibn Muḥammad alJurjānī, Dalāʾil al-ʾiʿjāz. Ed. by Muḥammad al-Tunjī. 3rd ed. Beirut: Dār al-Kitāb alʿArabī, 1999.

270

sartori

Jurjānī, Muqtaṣid = ʾAbū Bakr ʿAbd al-Qāhir ibn ʿAbd al-Raḥmān ibn Muḥammad alJurjānī, al-Muqtaṣid fī šarḥ risālat al-ʾĪḍāḥ. Ed. by al-Širbīnī Šarīda. 2 vols. Cairo: Dār al-Ḥadīṯ, 2009. Jurjānī, Šarḥ = ʾAbū Bakr ʿAbd al-Qāhir ibn ʿAbd al-Raḥmān ibn Muḥammad al-Jurjānī, Šarḥ al-jumal fī l-naḥw. Ed. by Ḫalīl ʿAbd al-Qādir ʿĪsā. 10th ed. Beirut: Dār Ibn Ḥazm and Amman: al-Dār al-ʿUṯmāniyya, 2011. Mubarrad, Muqtaḍab = ʾAbū l-ʿAbbās Muḥammad ibn Yazīd al-Mubarrad, al-Muqtaḍab. Ed. by Ḥasan Ḥamad and ʾImīl Badīʿ Yaʿqūb. 5 parts in 3 vols. Beirut: Dār al-Kutub al-ʿIlmiyya, 1999. Qazwīnī, ʾĪḍāḥ = Jalāl al-Dīn Muḥammad ibn ʿAbd al-Raḥmān al-Šāfiʿī al-Dimašqī almaʿrūf bi-l-Ḫaṭīb al-Qazwīnī, al-ʾĪḍāḥ fī ʿulūm al-balāġa. al-maʿānī wa-l-bayān wa-lbadīʿ. Ed. by ʾIbrāhīm Šams al-Dīn. Beirut: Dār al-Kutub al-ʿIlmiyya, 2003. al-Sayyid al-Šarīf, Taʿrīfāt = ʿAlī ibn Muḥammad ibn ʿAlī al-Sayyid al-Šarīf al-Ḥusaynī alJurjānī al-Ḥanafī, al-Taʿrīfāt. Ed. Muḥammad Bāsil ʿUyūn al-Sūd. 2nd ed. Beirut: Dār al-Kutub al-ʿIlmiyya, 2003. Zajjājī, Lāmāt = ʾAbū l-Qāsim ʿAbd al-Raḥmān ibn ʾIsḥāq al-Nahāwandī al-Zajjājī, Kitāb al-lāmāt. Ed. by Māzin al-Mubārak. 2nd ed. Beirut: Dar Ṣādir, 1992. Zamaḫšarī, Mufaṣṣal = Jār Allāh ʾAbū l-Qāsim Maḥmūd ibn ʿUmar ibn Muḥammad ibn ʾAḥmad al-Ḫwārizmī al-Zamaḫšarī, al-Mufaṣṣal fī ṣanʿat al-ʾiʿrāb. Ed. ʾĪmīl Badīʿ Yaʿqūb. Beirut: Dār al-Kutub al-ʿIlmiyya, 1999.

B

Secondary Sources

Alosh, Mahdi. 2005. Using Arabic: A guide to contemporary usage. Cambridge: Cambridge University Press. Badawi, El-Said, Michael G. Carter, and Adrian Gully. 2004. Modern written Arabic: A comprehensive grammar. London: Routledge. Badīʿ Yaʿqūb, ʾĪmīl and Mišāl ʿĀṣī. 1987. al-Muʿjam al-mufaṣṣal fī l-luġa wa-l-ʾadab. 2 vols. Beirut: Dār al-ʿIlm li-l-Malāyīn. Blachère, Régis and Maurice Gaudefroy-Demombynes, Maurice. 1975. Grammaire de l’arabe classique (morphologie et syntaxe). 3rd. revised ed. Paris: Maisonneuve et Larose. Brustad, Kristen E. 2000. The syntax of Spoken Arabic: A comparative study of Moroccan, Egyptian, Syrian and Kuwaiti dialects. Washington, D.C.: Georgetown University Press. Buckley, Ronald Paul. 2004. Modern Literary Arabic: A reference grammar. Beirut: Librairie du Liban. Cantarino, Vicente. 1974–1975. Syntax of modern Arabic prose. 2 vols. Bloomington and London: Indiana University Press. Carter, Michael G. 1981. Arab linguistics: An introductory Classical text with translation and notes. Amsterdam: J. Benjamins.

definition and determination in medieval arabic thought

271

Carter, Michael G. “Taʿrīf ”. Encylopaedia of Islam, 2nd ed., X, 241a–b. Leiden: E.J. Brill. El-Ayoubi, Hashem, Dieter Blohm and Wolfdietrich Fischer. 2010. Syntax der arabischen Schriftsprache der Gegenwart. II. Die Verbalgruppe. Wiesbaden: O. Harrassowitz. Fleisch, Henri. 1961. Traité de philologie arabe. I. Préliminaires, phonétique, morphologie nominale. Beirut: Imprimerie catholique. Fleisch, Henri. 1979. Traité de philologie arabe. II. Pronoms, morphologie verbale, particules. Beirut: Dar al-Machreq. Fleisch, Henri. 1986. “Iḍāfa”. Encyclopaedia of Islam, 2nd ed. III, 1008a–1009a. Leiden: E.J. Brill. Forbes, Duncan. 1863. Grammar of the Arabic language intended more especially for the use of young men preparing for the East India civil service; and also for the use of selfinstructing students in general. London: Wm. H. Allen & Co. Gätje, Helmut. 1970. “Zum Begriff der Determination und Indetermination im Arabischen”. Arabica 17.225–251. [Available at: http://www.jstor.org/stable/4055880]. Hassanein, Azza. 2006. Modern Standard Arabic grammar: A concise guide. Cairo and New York: The American University Press in Cairo. Heselwood, Barry and Janet Watson. 2015. “The Arabic definite article: A synchronic and historical perspective”. Arabic and Semitic linguistics contextualized: A festschrift for Jan Retsö, ed. by Lutz Edzard, 157–176. Wiesbaden: O. Harrassowitz. Holes, Clive. 2004. Modern Arabic: Structures, functions and varieties. 2nd revised ed. Washington, D.C.: Georgetown University Press. (1st ed. 1995.) Howell, Mortimer Sloper. 1911 [1880–1911]. A grammar of the Classical Arabic language, translated and compiled from the works of the most approved native or naturalized authorities. 4 vols. Allahabad. Hoyt, Frederick M. 2009. “Specificity”. Encyclopedia of Arabic language and linguistics, ed. by Mushira Eid et al., IV, 315–320. Leiden: E.J. Brill. Imbert, Frédéric. 2008. L’arabe dans tous ses états! La grammaire arabe en tableaux. Paris: Ellipses. Kazimirski, Adrien de Biberstein. 1860. Dictionnaire arabe-français. 2 vols. Beirut: Maisonneuve. Kouloughli, Djamel Eddine. 1994. Grammaire de l’arabe d’aujourd’hui. Paris: Pocket, “Langues pour tous”. Kouloughli, Djamel Eddine. 2001. “Sur le statut linguistique du tanwīn: Contribution à l’étude du système déterminatif de l’arabe”. Arabica 48.20–50. [Available at: http:// www.jstor.org/stable/4057589]. Larcher, Pierre. 1991. “D’une grammaire l’autre: Catégorie d’ adverbe et catégorie de mafʿūl muṭlaq”. De la grammaire de l’arabe aux grammaires des arabes, ed. by Pierre Larcher, Bulletin d’Études Orientales 43.139–159. [Available at: http://www.jstor.org/ stable/41608973].

272

sartori

Larcher, Pierre. 2011. “Un texte arabe sur le métalangage”. A festschrift for Nadia Anghelescu, ed. by Andrei A. Avram et al., 306–317. Bucharest: Editura Universităţii din Bucureşti. Larcher, Pierre. 2014. Linguistique arabe et pragmatique. Beirut: Presses de l’ Ifpo. McCarus, Ernest N. 2007. English grammar for students of Arabic: The study guide for those learning Arabic. Ann Arbor: The Olivia and Hill Press. Morais Barbosa, Jorge. 1998. “Détermination épithétique et détermination prédicative”. La Linguistique 34:2.15–20. Neyreneuf, Michel and Ghalib Al-Hakkak. 1996. Grammaire active de l’ arabe. Paris: Le Livre de Poche. Palmer, Edward Henry. 1874. A grammar of the Arabic language. London: Wm. H. Allen & Co. Reckendorf, Hermann. 1921. Arabische Syntax. Heidelberg: Carl Winter’s Universitätsbuchhandlung. Retsö, Jan. 1986. “State, determination and definiteness in Arabic: A reconsideration”. Orientalia Suecana 33–34.341–346. Ryding, Karin C. 2005. A reference grammar of Modern Arabic. Cambridge: Cambridge University Press. Ryding, Karin C. and Kees Versteegh. 2007. “ʾIḍāfa”. Encyclopedia of Arabic language and linguistics, ed. by Mushira Eid et al., II, 294–298. Leiden: E.J. Brill. Sartori, Manuel. 2018. “Origin and conceptual evolution of the term taḫṣīṣ in Arabic grammar”. Foundations of Arabic linguistics. III. The development of a tradition: Continuity and change, ed. by Georgine Ayoub and Kees Versteegh, 203–228. Leiden: E.J. Brill. Schulz, Eckehard et al. 1996. Lehrbuch des modernen Arabisch. Berlin and Munich: Langescheidt KG. (English transl., Standard Arabic. An elementary-intermediate course, New York: Cambridge University Press, 2008 [2000].) Silvestre de Sacy, Antoine-Isaac. 1831. Grammaire arabe à l’ usage des élèves de l’ école spéciale des langues orientales vivantes, avec figures. 2nd revised ed. 2 vols. Paris: Imprimerie royale. (3rd ed., revised by L. Machuel. Tunis: Institut de Carthage, 1904.) Socin, Albert. 1885. Arabische Grammatik: Paradigmen, Litteratur, Chrestomathie und Glossar. Karlsruhe and Leipzig: Reuther. Troupeau, Gérard. 1976. Lexique-index du Kitāb de Sībawayhi. Paris: Klincksieck. Troupeau, Gérard. 1993. “Naʿt”Encyclopaedia of Islam, VII, 1034a. Leiden: E.J. Brill. [Available at: http://referenceworks.brillonline.com.lama.univ‑amu.fr/entries/ encyclopedie‑de‑l‑islam/nat‑SIM_5848]. Vernier, Donat. 1891. Grammaire arabe composée d’après les sources primitives. 2 vols. Beirut: Imprimerie catholique. Wensinck, Arent Jan. 1931. “The article of determination in Arabic”. Mededeelingen der Koninklijke Akademie van Wetenschappen, Afdeeling Letterkunde 71, serie A, no. 3.

definition and determination in medieval arabic thought

273

Wright, William. 1996. A grammar of the Arabic language. 2 vols. Repr., Beirut: Librairie du Liban. (1st ed., 1859–1862; 3rd ed., revised by W. Robertson Smith and M.J. de Goeje. Cambridge: Cambridge University Press, 1896–1898.)

The Concept of tawṭiʾa in the Medieval Arabic Grammatical Tradition Beata Sheyhatovitch

1

Introduction

The term tawṭiʾa lit. ‘preparation’ in Medieval Arabic grammatical literature refers to the function of an element which ‘prepares the way’ for another element (or phenomenon), either in a given structure or in the structure that can be derived from the given one. Its use is not common, but the term seems essential for understanding the discussions in which it appears. Moreover, its systematic study can contribute to our understanding of the grammarians’ perception of language. Like many other terms, tawṭiʾa is neither defined nor explained by the grammarians. Furthermore, its use is not confined to particular chapters in their books, thus its study requires finding all occurrences of the term followed by their analysis and categorization. This article is divided into five main sections that correspond to the five main contexts in which the term is used by the grammarians: tawṭiʾa in discussions on ‘the six nouns’, ḥāl muwaṭṭiʾa, ‘preparing’ particles in discussions on oath sentences, tawṭiʾa in discussions on adjectival phrases and, finally, tawṭiʾa in discussions on doubly-transitive verbs.

2

Tawṭiʾa in Discussions on ‘the Six Nouns’

The prominent use of the term tawṭiʾa in relatively early grammatical literature is in discussions on ‘the six nouns’, namely ʾaḫ ‘brother’, ʾab ‘father’, ḥam ‘fatherin-law’, ḏū ‘a possessor [of something]’, fū ‘[someone’s] mouth’; han ‘thing’ (the last one is not always mentioned as part of this group because it does not always behave like the other ones do).1 The grammarians have to explain why these nouns, when appearing in an annexation or taking bound pronouns, receive

1 See Wright (1967:I, 249) for a discussion on these nouns.

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004389694_017

the concept of tawṭiʾa

275

the case markers in the form of letters (ḥurūf ),2 unlike all other nouns, which receive the case markers in the form of vowels.3 The clearest discussion that makes use of the term tawṭiʾa is found in Ibn alWarrāq (d. 381/991) He explains (ʿIlal 150.2–8) that ʾaḫ etc. receive case markers in the form of letters “as a preparation for dual and plural forms” (tawṭiʾatan limā yaʾtī min al-taṯniyati wa-l-jamʿi). He does not explain the relation between letters serving as case markers and dual/plural forms, but it is evident that dual/plural forms take this type of case markers, e.g., muslimāni/muslimayni ‘two Muslims’: ʾalif is the rafʿ marker, and yāʾ is naṣb/ jarr marker. It is apparent that, by talking about “a preparation for dual/plural forms”, Ibn al-Warrāq means that these forms can be created from ʾaḫ and its likes. Naturally, one may claim that this argument fails to explain the special behavior of ʾaḫ and its likes, as dual/plural forms can be created not only from these particular ones but from any noun. To explain this point, Ibn al-Warrāq says that the nouns in question “are more deserving than others to receive [the case markers in the form of letters] as preparation” (ʾawlā bi-l-tawṭiʾati min ġayrihā), because these are nouns that “must semantically function as annexed ones” (lā tanfakku min ʾiḍāfati l-maʿnā). In other words, one cannot be just father/brother/father-in-law, one must be the father/brother of someone. The governed noun is present at some level, even if it is not mentioned (whereas ḏū and fū cannot appear outside the annexation structure). Being an annexed noun is a deviation from the basic structure, which is an independent word, just as dual/plural forms are a deviation from the basic structure, which is the singular form. This is a characteristic that applies to the six nouns and the dual/plural forms, and that supposedly causes the speaker to consider the six nouns in an annexation as a preparation for these forms and to treat them accordingly.4

2 The term ḥarf may denote a letter of the alphabet, but as a phonetic term it denotes “a sound which is represented in Arabic orthography by a letter” (Levin 1986:425, n. 13). In the present article the term ‘letter’ is used as a short form of the latter formulation. 3 Al-Zajjājī states in ʾĪḍāḥ 72.1–19 that the case markers are supposed to be vowels in principle, but some words may receive their case markers in the form of letters, which requires special explanations. 4 See also Ibn Jinnī, Ḫaṣāʾiṣ I, 310.7–10 and Ibn al-ʾAnbārī, ʾInṣāf I, 27.9–28.4.

276 3

sheyhatovitch

Ḥāl muwaṭṭiʾa

This term was already used by al-ʿUkbarī (d. 616/1219) in Lubāb (I, 295.2–3). Ibn al-Ḥājib (d. 646/1249; the author of Kitāb al-Kāfiya, commented on by alʾAstarābāḏī in his Šarḥ al-Kāfiya) and al-ʾAstarābāḏī (d. probably in 688/1289) disagree with most grammarians who hold that, in principle, the ḥāl is supposed to be morphologically derived (muštaqq, i.e., active/passive participle or an adjective resembling them), and try to paraphrase cases in which underived nouns function as a ḥāl so that they would fit into this model.5 Ibn al-Ḥājib states (ʾAstarābāḏī, Šarḥ II, 32.4) that “any expression signifying state [of the ṣāḥib al-ḥāl during the action mentioned in the sentence]” (kullu mā dalla ʿalā hayʾatin) can function as a ḥāl. Al-ʾAstarābāḏī (Šarḥ II, 32.13–18) agrees with Ibn al-Ḥājib (although he mentions that, undoubtedly, in most cases both ḥāl and waṣf ‘qualifier’ are morphologically derived). He adds that there are some categories of ḥāl in which an underived noun is consistently used. One of these categories is ḥāl muwaṭṭiʾa, which consists of an underived noun followed by an adjective. This adjective is “the true ḥāl” (al-ḥāl fī l-ḥaqīqati), whereas the main noun in the phrase “prepares the way for the real ḥāl” (waṭṭaʾa l-ṭarīqa li-mā huwa l-ḥālu fī l-ḥaqīqati). This is what happens in sentences such as ʾinnā ʾanzalnāhu qurʾānan ʿarabiyyan (Q. 12/2) ‘We have sent it down as an Arabic Qurʾān’ and jāʾanī zaydun rajulan bahiyyan ‘Zayd came to me as a beautiful man’. Al-ʾAstarābāḏī does not add any explanation about the semantic function of the underived nouns in the abovementioned sentences (except for saying that they “prepare the way for the real ḥāl”). These underived nouns are co-referent with ṣāḥib al-ḥāl, and (at least in these examples) do not appear to add any significant information. Thus, the function of the underived noun seems to be to emphasize the meaning by strengthening the link between the ṣāḥib al-ḥāl and the adjective describing its state. Ibn Hišām (d. 761/1360) suggests (Muġnī, 604.4–606.6) several possible categorizations of ḥāl: (i) distinguishing between ḥāl denoting transient and permanent states; (ii) distinguishing “between [ḥāl] that is intended for its own sake and [ḥāl that is intended] for the preparation” (bi-ḥasbi qaṣdihā li-ḏātihā

5 For instance, al-Jurjānī (Muqtaṣid I, 676.5–16) explains that the main difference between the ḥāl and the tamyīz is that, in principle, the ḥāl is supposed to be an adjective, and the tamyīz is supposed to be a noun proper. This point appears also in Ibn Hišām’s Muġnī l-labīb (the relevant fragment is presented in Bernards 2007:225). See Ibn Yaʿīš, Šarḥ II, 60.7–63.29 for examples in which Ibn Yaʿīš paraphrases problematic examples of ḥāl with verbs or derived nouns. As for the term muštaqq, see Larcher (2006).

the concept of tawṭiʾa

277

wa-li-l-tawṭiʾati); (iii) distinguishing between ḥāl that is “simultaneous (with the action signified by the governing verb)” (muqārina), ḥāl that is called “the planned one” (muqaddara), referring to a time subsequent to the one signified by the governing verb, and ḥāl that is called “the quoted one” (maḥkiyya), referring to the past; (iv) distinguishing between ḥāl that clarifies and the one that emphasizes. This discussion proves that Ibn Hišām considers ḥāl muwaṭṭiʾa a type of ḥāl on its own (although it is less common than the ḥāl that is intended for its own sake). Like al-ʾAstarābāḏī, Ibn Hišām (Muġnī, 605.10–11) states that ḥāl muwaṭṭiʾa appears as an underived noun followed by an adjective. His examples are: fa-tamaṯṯala lahā bašaran sawiyyan (Q. 19/17) ‘presented himself to her as a man without fault’ and jāʾanī zaydun rajulan muḥsinan ‘Zayd came to me as a charitable man’.

4

‘Preparing’ Particles in Discussions on Oath Sentences

4.1 The ‘Preparing’ laThe term [al-lām] al-muwaṭṭiʾa li-l-qasam6 was already used by al-Zamaḫšarī (d. 538/1144) and explained by Ibn Yaʿīš (d. 643/1246).7 However, al-ʾAstarābāḏī’s discussion on the topic (found in his treatment of sentences containing a combination of conditional and oath) is particularly intricate. He explains that if an oath precedes two clauses, of which the second (the apodosis) refers to the future and represents something that one swears to do, and the first is preceded by a conditional particle, the conditional particle is often prefixed by la-. This la- is called muwaṭṭiʾa (al-ʾAstarābāḏī glosses this term with the word mumahhida, to mean that it prepares the way for the apodosis). La- indicates that the apodosis belongs specifically to the oath, not to the conditional. An example is: wa-llāhi la-ʾin ʾataytanī la-ʾātiyannaka ‘I swear by God, if you come to me, I come to you!’. The la- of la-ʾin is al-lām al-muwaṭṭiʾa. One could express the same idea without this particle and say wa-llāhi ʾin taʾtinī laʾātiyannaka.8

6 For a discussion of this phenomenon according to Ibn Hišām see Testen (1998:28–32). Testen translates al-lām al-muwaṭṭiʾa as ‘the l which paves the way for the oath’, and al-lām al-muʾḏina (a synonymous term) as ‘the foreshadowing l’. 7 See Ibn Yaʿīš, Šarḥ IX, 22.1–6. The relevant passage is translated and discussed in Testen (1998:29–30). 8 ʾAstarābāḏī, Šarḥ IV, 314.12–315.2. For a discussion of a similar passage from Ibn Hišām see Testen (1998:29).

278

sheyhatovitch

However, “if [the oath] is omitted, but [the speaker] still has it in mind”9 (ʾin ḥuḏifa wa-quddira), al-lām al-muwaṭṭiʾa is usually used “to inform [the addressee] of the oath, which [the speaker] has in mind from the outset” (tanbīhan ʿalā l-qasami l-muqaddari min ʾawwali l-ʾamri).10 Al-ʾAstarābāḏī (Šarḥ IV, 315.5) notes that al-lām al-muwaṭṭiʾa may be omitted in this case as well, as in the Qurʾānic verse wa-ʾin ʾaṭaʿtumūhum ʾinnakum la-mušrikūna (Q. 6/121) ‘if you obey them, you are idolaters’. The relation between this verse and oath sentences is not obvious and calls for additional discussion. Although a detailed analysis is hard to find in the early Medieval grammatical and exegetical literature,11 based on the basic rules of Arabic syntax, it is apparent that ʾinnakum la-mušrikūna cannot be considered as an apodosis of the conditional, because in conditional sentences a nominal clause cannot function as an apodosis, unless it is prefixed by fa(which does not appear here).12 The clearest presentation of this issue appears to be that of ʾAbū Ḥayyān (d. 745/1344), who objects to the claim that this is a simple omission of fa-, by claiming (Baḥr IV, 215.27–28) that such an omission belongs to the realm of “[poetic] licenses” (ḍarāʾir), and is not to be found in the Qurʾān (as the Qurʾānic text is not considered poetry;13 moreover, it is unthinkable that God would be forced to break the rules of the language that He Himself created). ʾAbū Ḥayyān (Baḥr IV, 215.28) claims that the apodosis belongs to the omitted oath expression wa-llāhi ‘I swear by God’, while the apodosis of the conditional is omitted. He notes (Baḥr IV, 216.2–3) that, usually, in these cases a la- “that announces the omitted oath” (al-muʾaḏḏina bi-l-qasami l-maḥḏūfi) is prefixed to the conditional ʾin—e.g., la-ʾin ʾuḫrijū lā yaḫrujūna maʿahum (Q. 59/12) ‘If those are expelled, they will not go forth with them’, and adds that an apodosis of the conditional can be omitted because the apodosis of the oath allows its reconstruction.

9

10 11

12 13

In this context Levin’s interpretation (1997:142f.) of taqdīr as speaker’s intention seems appropriate, although in the later grammatical literature the term usually means a reconstruction by the grammarians (see Versteegh 2009b). ʾAstarābāḏī, Šarḥ IV, 315.3f. For instance, Zajjāj (Maʿānī II, 287.17–21), Zamaḫšarī (Kaššāf II, 58.2–59.5), Rāzī (Tafsīr XIII, 168.9–170.16) and Qurṭubī ( Jāmiʿ VII, 77. 19–78.2) discuss the meaning of the verse without analyzing the function of la- or mentioning an oath. See, e.g., Peled (1992:2–5) for a discussion of the basic patterns of the conditional sentences in the view of the Arab grammarians. See Jones (2004) for a discussion of the Qurʾānic verses condemning poetry and stating that poetry was not a suitable register for the revelation.

the concept of tawṭiʾa

279

To summarize, in wa-ʾin ʾaṭaʿtumūhum ʾinnakum la-mušrikūna not only the apodosis of the conditional is omitted, but also omitted are the oath expression and al-lām al-muwaṭṭiʾa (which is supposed to precede the conditional word in order to signal that the apodosis belongs to the oath and not to the conditional, and whose presence is even more crucial if the oath expression is omitted). Thus, the only considerations that allow the grammarians and the commentators to interpret this verse as an oath sentence are the presence of a structure which is irregular for conditional sentences and the principle according to which the apodosis of an oath may replace the apodosis of the conditional and thus allow its omission.14 4.2 The ‘Preparing’ ʾan 4.2.1 Al-Ḫalīl’s and Sībawayhi’s Analysis Sībawayhi does not use the term tawṭiʾa. However it seems that the later grammarians’ views on the ‘preparing’ ʾan (and probably on the ‘preparing’ particles in oath sentences in general) were inspired by the following discussion from al-Kitāb. The starting point of the discussion (Kitāb I, 404.24) is the following Qurʾānic verse: (i)

wa-ʾiḏ ʾaḫaḏa llāhu mīṯāqa l-nabiyyīna la-mā ʾātaytukum min kitābin waḥikmatin ṯumma jāʾakum rasūlun muṣaddiqun li-mā maʿakum la-tuʾminunna bihi wa-la-tanṣurunnahu (Q. 3/81) ‘And when God took the covenant of the prophets, [saying], “Whatever I give you of the Scripture and wisdom and then there comes to you a messenger confirming what is with you, you [must] believe in him and support him” ’.

Sībawayhi presents al-Ḫalīl’s analysis of (i): The mā [in la-mā in (i)] has the same status as allaḏī [i.e., it functions as a relative pronoun], and is preceded by la-, just like the ʾin in (ii) wallāhi la-ʾin faʿalta la-ʾafʿalanna ‘By God, if you do, I will do!’ is preceded [by la-]. The la- that precedes mā [in (i)] is analogous to the one that precedes ʾin [in (ii)], and the la- that precedes the verb [in la-tuʾminunna/ la-tanṣurunnahu in (i)] is analogous to the one that precedes the verb [in la-ʾafʿalanna in (ii)] (mā hāhunā bi-manzilati llaḏī wa-daḫalathā l-lāmu

14

See Testen (1998:21) for a discussion of this phenomenon.

280

sheyhatovitch

kamā daḫalat ʿalā ʾin ḥīna qulta wa-llāhi la-ʾin faʿalta la-ʾafʿalanna wa-llāmu llatī fī mā ka-hāḏihi llatī fī ʾin wa-l-lāmu llatī fī l-fiʿli ka-hāḏihi llatī fī l-fiʿli hunā).15 It can be concluded from the analogy between (i) and (ii) that al-Ḫalīl interprets (i) as a combination of a conditional and an oath, although neither of the two is explicit in the verse. Al-Zajjāj (Maʿānī I, 436.14 f.) says that the mā in la-mā ʾātaytukum in (i) functions as a conditional, and the meaning is that “everything related to messengers that happens, happens this way [i.e., each time God sends a messenger, the people shall believe in him and help him]” (kullu mā waqaʿa min ʾamri l-rusuli fa-hāḏihi ṭarīqatuhu).16 As for the implicit oath, al-Zajjāj (Maʿānī I, 437.8) explains that ʾaḫaḏa mīṯāqahum means istaḥlafahum ‘made them swear’.17 Al-Qurṭubī ( Jāmiʿ IV, 125.10) states explicitly that ʾaḫaḏtu mīṯāqaka ‘I took your covenant’ can be followed by a clause built as an apodosis of an oath.18 Following this excerpt, in Sībawayhi/al-Ḫalīl’s discussion (Kitāb I, 404.22–24) an analogy is drawn between the first la- in (i) and (ii) (i.e., in la-mā and la-ʾin) on the one hand, and, on the other hand, ʾan in (iii) wa-llāhi ʾan law faʿalta lafaʿaltu ‘By God, had you done [it], I would have done [it]’ and in the following verse by Musayyib ibn ʿAlas: (iv) wa-ʾuqsimu ʾan law iltaqaynā wa-ʾantumu / la-kāna lakum yawmun min alšarri muẓlimu ‘I swear that if we were to meet you,/ it would have been a dark day of evil for you’.19

15 16

17

18 19

Sībawayhi, Kitāb I, 404.20–23. Al-Zajjāj (Maʿānī I, 437.6–7) mentions the option of a relative pronoun as a second, less preferable possibility to interpret the mā, whereas al-Ḫalīl apparently does not see a contradiction between saying that mā in the verse is equivalent to allaḏī and interpreting the verse as an implicit conditional sentence. Sībawayhi (Kitāb I, 389.6–14) speaks of “nouns that open the protasis [of a conditional sentence] and have the same status as allaḏī” (alʾasmāʾ llatī yujāzā bihā wa-takūnu bi-manzilati llaḏī), and mentions mā in this context. Sībawayhi (Kitāb I, 404.8–14) explains that when a speaker recalls an oath made by someone else, or speaks of someone else who made a third party swear, a sentence is still produced as an oath sentence. One of his examples in this context is the Qurʾānic verse wa-ʾiḏ ʾaḫaḏnā mīṯāqa Banī ʾIsrāʾīla lā taʿbudūna ʾillā llāha (Q. 2/83) ‘And [remember] when We made a covenant with the Children of Israel, (saying): Worship none save God [only]’, in which the use of ʾaḫaḏnā mīṯāqa … is equivalent to the use of ʾaḫaḏa llāhu mīṯāqa lnabiyyīna in (i). See Qurṭubī ( Jāmiʿ IV, 124.12–126.12) for a detailed discussion on the verse, including grammatical analyses that evoke various interpretations of the text. See Baġdādī, Ḫizāna X, 80.3–84.14 for a discussion of this verse and its context.

the concept of tawṭiʾa

281

Sībawayhi (Kitāb I, 405.1–3) presents al-Ḫalīl’s analysis as follows: the ʾan that precedes the law in (iv) has the same status as la- that precedes the mā in (i). He does not refer explicitly to the functions of ʾan and la- in the examples, but obviously connects them to the sentences that combine the oath and the conditional. 4.2.2 Al-ʾAstarābāḏī’s and al-Baġdādī’s Approach Al-ʾAstarābāḏī (Šarḥ IV, 456.10–14) says that Sībawayhi views the ʾan in sentences such as (v) wa-llāhi ʾan law jiʾtanī la-jiʾtuka ‘I swear by God, had you come to me, I would have come to you!’ as muwaṭṭiʾa, whereas other grammarians consider it zāʾida ‘redundant’. In this sentence la-jiʾtuka is the apodosis of the oath, not of the conditional (this claim is supported by the fact that the lahere cannot be omitted, unlike the la- that precedes the usual apodosis of law). Al-ʾAstarābāḏī (Šarḥ IV, 313.11–13) mentions al-ʾan al-muwaṭṭiʾa also when presenting Sībawayhi’s opinion on (iv). The fact that Sībawayhi draws an analogy between the ʾan in (iii) and (iv) and the more widespread la- in (i) probably leads the later grammarians to assume that he views that ʾan as muwaṭṭiʾa (as they hold that la- to be muwaṭṭiʾa and conceivably ascribe this view to Sībawayhi). However, Sībawayhi (Kitāb I, 424.7 f.) says that in (vi) ʾa-mā wa-llāhi ʾan law faʿalta la-ʾakramtuka ‘By God, had you done this, would not I honor you?’ ʾan is laġw (in this context the term laġw seems to be close to zāʾida20). Given that claim, it may be concluded that, in Sībawayhi’s view, there is no contradiction between the analogy ʾan—la- and analyzing ʾan as laġw.21 Thus, the contrast that al-ʾAstarābāḏī’s creates between Sībawayhi’s and other grammarians’ opinion on ʾan that appears in the combination of oath and conditional does not seem justified. Ibn ʿUṣfūr states that in sentences that combine oath and law/lawlā, law is sometimes preceded by ʾan, whose function is to act as “preparation for interpreting the following verb as an apodosis of the oath [and not the con-

20

21

Peled (2009:150) interprets laġw as “a constituent that can function neither as ʿāmil nor as maʿmūl”. Cachia (1973:89) suggests “redundant” as one of the senses of the term laġw. Ibn Yaʿīš (Šarḥ VIII, 128.9f.) presents ziyāda and ʾilġāʾ (lit. ‘cancellation’; derived from the same root as laġw) as Baṣran terms for redundancy, whose Kūfan counterparts are ṣila and ḥašw. See Versteegh (2009a) for a discussion of the term ṣila in the sense of a redundant element (and other senses of the term). Sībawayhi views certain particles simultaneously as redundant (zāʾid) and as indicating a reinforcing sense (tawkīd). The relevant passages are discussed in Baalbaki (2008:96 f.). Thus, it is not surprising that Sībawayhi can view ʾan simultaneously as redundant and as playing a certain role in combinations of conditional and oath structures.

282

sheyhatovitch

ditional]” (tawṭiʾatan li-jaʿli l-fiʿli l-wāqiʿi baʿdahā li-l-qasami).22 Like Sībawayhi, Ibn ʿUṣfūr draws an analogy between ʾan+law and la+ʾin. In Šarḥ al-Jumal (I, 528.18–529.2) Ibn ʿUṣfūr suggests that the ʾan replaces the more regular la- in the function of “connecting the oath expression with the content of the oath” ( yarbuṭu l-muqsama bi-l-muqsami ʿalayhi).23 La- is used in this structure for phonetic reasons, namely to avoid creating a sequence of two lām’s. Thus, the speakers say wa-llāhi ʾan law qāma zaydun la-qāma ʿamrun ‘By God, had Zayd risen, ʿAmr would have risen’ (instead of *wa-llāhi la-law qāma zaydun la-qāma ʿamrun). Al-Baġdādī (Ḫizāna X, 81.16–82.1) criticizes Ibn Hišām for claiming that Sībawayhi viewed the aforementioned ʾan as redundant.24 This means that alBaġdādī (in agreement with al-ʾAstarābāḏī) interprets preparing the way and redundancy as two contradictory functions. However, Ibn Hišām has a different view, which will be discussed below in subsection 4.3. 4.3 The ‘Preparing’ lā Ibn Hišām (Muġnī 328.4–329.11) mentions several opinions on the function of the particle lā in the verse lā ʾuqsimu bi-yawmi l-qiyāmati (Q. 75/1) ‘No, I swear by the Day of Resurrection’. Some scholars hold that it is a negative particle (and suggest various assumptions concerning the element negated by it), whereas others claim that it is a redundant element (zāʾida). Those who consider it redundant still disagree on its fāʾida ‘contribution to the text’.25 There are two opinions on this point: i. “It was added as a preparation to the negative apodosis” (zīdat tawṭiʾatan wa-tamhīdan li-nafyi l-jawābi). According to this approach, the intention is: lā ʾuqsimu bi-yawmi l-qiyāmati lā yutrakūna sudan ‘No, I swear by the Day of Resurrection, they will not be left aimless’.26 Other examples of lā with the same function are: fa-lā wa-rabbika lā yuʾminūna ḥattā

22 23 24 25 26

See Baġdādī (Ḫizāna X, 81.10–15) where he cites Ibn ʿUṣfūr’s commentary on Kitāb al-ʾĪḍāḥ written by ʾAbū ʿAlī al-Fārisī. See Baġdādī, Ḫizāna X, 82.8–15 for the grammarians’ criticisms of this formulation. The relevant fragment appears in Ibn Hišām, Muġnī, 50.6–51.1. This sense of the term fāʾida seems to be related to ‘fāʾida as an addition to the message’, one of the main senses of the term discussed in Sheyhatovitch (2012). This interpretation is based on Q. 75/36: ʾa-yaḥsabu l-ʾinsānu ʾan yutraka sudan ‘Does man think that he is to be left aimless?’. Al-ʾAstarābāḏī (Šarḥ IV, 436.14–437.2) gives this example and the following poetic line by Imruʾ al-Qays in explaining that a redundant lā is frequently added in oath sentences “in order to announce that the apodosis is negative” (li-l-ʾīḏāni bi-ʾanna jawāba l-qasami manfiyyun). He notes that it appears before ʾuqsimu less frequently than before other oath expressions.

the concept of tawṭiʾa

283

yuḥakkimūka fīmā šajara baynahum (Q. 4/65) ‘But nay, by thy Lord, they will not believe until they make you judge of what is in dispute between them’, and the line by Imruʾ al-Qays: lā wa-ʾabīki bnata l-ʿĀmiriyyi / lā yaddaʿī l-qawmu ʾannī ʾafirru ‘No, I swear by your father, Oh daughter of al-ʿĀmirī,/ the people will not claim that I am running away’.27 ii. It was added only for the purpose of emphasis and to strengthen the speech. Ibn Hišām suggests ‘preparation’ as one of the possible functions of redundant elements. It can be concluded that, according to Ibn Hišām, an element does not have to govern grammatically in order not to be considered redundant (e.g., the negative lā does not have any grammatical influence, like the redundant lā does), but its contribution to the message of an utterance has to be significant. The contribution of a negative particle is considered sufficiently significant (which is natural, since that reverses the message of an utterance into its opposite), whereas the contribution of an emphasizing or ‘preparing’ element is not considered to be as significant. 4.4 Redundancy versus Preparation In the previous subsections it has been demonstrated that al-ʾAstarābāḏī and al-Baġdādī see a contradiction between redundancy and the function of ‘preparation’, whereas Ibn Hišām (and, supposedly, also Sībawayhi) do not. This disagreement calls for a closer look into the grammarians’ notion of redundancy. Ibn al-Sarrāj (ʾUṣūl II, 257.1–261.13) dedicates a chapter to the phenomenon of “redundancy and cancellation” (al-ziyāda wa-l-ʾilġāʾ), in which he distinguishes between redundant nouns, verbs, particles and clauses. He explains (ʾUṣūl II, 257.1–3) that a redundant constituent (a) does not have a syntactical position that necessitates a case marker; (b) can be removed from an utterance without corrupting it; and (c) functions for “emphasis or clarification” (taʾkīdan ʾaw tabyīnan).28 Ibn al-Sarrāj’s view of redundant constituents can easily include the ‘preparing’ particles presented above. Types of clarification can definitely include indicating to the addressee that the second clause is the apodosis of the oath (which

27 28

See Baġdādī, Ḫizāna XI, 221.5–224.6 for a discussion of this verse. See Versteegh (2007) for a discussion on the term ʾilġāʾ in Medieval Arabic grammatical theory (including the abovementioned passage from Ibn al-Sarrāj). Versteegh holds that the prototypical use of ʾilġāʾ occurs in cases of disrupted government of cognitive verbs.

284

sheyhatovitch

is the function of ‘preparing’ la- and ʾan) and preparing the addressee psychologically for the negative apodosis (which is the function of the ‘preparing’ lā). Ibn Yaʿīš (Šarḥ VIII, 128.9) defines a redundant constituent as something that can be both inserted or omitted without creating a new meaning. He notes (Šarḥ VIII, 129.1–3) that a redundant constituent can still function as emphasis. Like Sībawayhi, he (Šarḥ VIII, 131.2–3) views the ʾan in ʾa-mā wa-llāhi ʾan law faʿalta la-ʾakramtuka as zāʾida. As for the lā of the type discussed in subsection 4.3 above, Ibn Yaʿīš (Šarḥ VIII, 136.3–18) classifies it as redundant and its function as emphasis (since he considers emphasis to be the only function that a redundant constituent can have). Al-ʾAstarābāḏī (Šarḥ IV, 432.12–17) summarizes the grammarians’ views on redundant constituents by saying that the contribution ( fāʾida) of a redundant particle can be either semantic (maʿnawiyya) or formal (lafẓiyya). He explains that the semantic contribution is in “emphasizing the meaning”, and that a constituent with such a contribution can be designated as redundant, since it does not change the original meaning of the utterance and adds to the existing meaning nothing but emphasis and strength. The formal contribution (Šarḥ IV, 433.5–7) of the particle is to adorn the expression and make it more eloquent, or to fix the poem’s rhythm, or something similar. Al-ʾAstarābāḏī (Šarḥ IV, 433.8–9) notes that if a redundant particle had neither a semantic nor a formal contribution, it would not be acceptable, since a speech of eloquent people (let alone God and His prophets) should not include useless elements. Al-ʾAstarābāḏī criticizes the accepted notion of redundancy when he says (Šarḥ IV, 433.1–2) that according to the abovementioned definitions, ʾinna and lām al-ibtidāʾ should be considered redundant (since their semantic contribution is restricted to emphasis), but no other grammarian supports this. Moreover, he expresses (Šarḥ IV, 436.7–12) surprise at the fact that the grammarians consider the lā redundant in sentences such as mā jāʾanī zaydun wa-lā ʿamrun ‘Neither Zayd nor ʿAmr came to me’, although it serves to make the sentence unequivocal, whereas mā l-kāffa, whose only contribution is formal, is not considered redundant. Despite these points of criticism, al-ʾAstarābāḏī does not offer an alternative terminology or definition. Given al-ʾAstarābāḏī’s critical approach to the terminology of redundancy, it is understandable that he mentions the grammarians’ opinions on the ʾan discussed in subsection 4.2 without bothering to give his own opinion. It may be concluded that the varying opinions on the function of that ʾan reflect different terminological and wording preferences rather than theoretical controversy.

the concept of tawṭiʾa

5

285

Tawṭiʾa in Discussions on Adjectival Phrases

Al-ʾAstarābāḏī (Šarḥ III, 435.10–19) starts his discussion of the grammatical government of participles and adjectives by counting the theoretically possible combinations of different types of adjectives and nouns that they govern: the governing adjective can be either prefixed by the definite article or free of it; independently of that, the governed noun can be annexed to another noun, or be prefixed by the definite article, or be free of both. Additionally, the governed noun can take any of the three grammatical cases. Thus, the number of theoretically possible combinations amounts to 18 (2*3*3).29 Al-ʾAstarābāḏī (Šarḥ III, 435.20–436.15) immediately rejects two possibilities, al-adjective annexed to a noun that is in turn annexed to a personal pronoun (e.g., *al-ḥasanu wajhihi), and al-adjective annexed to a noun that is free of al- and a personal pronoun (e.g., *al-ḥasanu wajhin). Al-ʾAstarābāḏī’s basic opinion (Šarḥ III, 438.20–439.3) is that all the aforementioned possibilities originate in al-ḥasanu wajhuhu and ḥasanun wajhuhu. Wajh is supposed to take rafʿ because it is equivalent to a subject in a verbal sentence, and is also supposed to include a pronoun (referring to something described by the adjectival phrase) because ḥasan itself (unlike a verb) cannot include a personal pronoun. The structures with noun in rafʿ are considered to be the origin, because the primary idea of all the adjectival phrases mentioned here is to ascribe the attribute denoted by the adjective to the referent of the noun. A predicative relation then exists between the two constituents of the phrase, as seen in the English translation ‘[the one] whose face is beautiful’,30 and the predicative constituents are in principle supposed to take rafʿ. Among the constructions created from the ‘original’ two, there are four that are frequently used and widely accepted: al-ḥasanu wajhan, ḥasanun wajhan, al-ḥasanu l-wajhi, and ḥasanu l-wajhi.31 The rest are problematic for one reason or another, and thus rarely used.32 These include al-ḥasanu wajhahu/ḥasanun

29 30

31 32

See al-Ġaḍḍāb (2008:100–109) for a detailed description of all these possibilities. Such structures usually function as naʿt sababī, an adjective that refers to the main noun “in virtue of a following word which is connected with it”. See Wright (1967:II, 283–284). Also see Diem (1998) for a detailed discussion on the uses of naʿt sababī. Diem (1998:12) renders this term as “adjektivischer Satz”, since, on the one hand, it is equivalent to a sentence, and on the other hand it has the same syntactic distribution as adjectives and participles. See ʾAstarābāḏī, Šarḥ III, 439.3–19 for discussion. See ʾAstarābāḏī, Šarḥ III, 439.20–442.8 for discussion.

286

sheyhatovitch

wajhahu, which were not approved by the mainstream grammarians although they were accepted by Ibn al-Ḥājib. Al-ʾAstarābāḏī uses the term tawṭiʾa in his discussion of the latter two. His discussion is better understood once it is kept in mind that, according to the grammarians, the speakers tend to transform ‘heavy’ constructions into ‘lighter’ ones. Their writings contain multiple examples of this tendency, which are found at literally every level of linguistic analysis.33 The same principle is at work here: the speakers seek to transform the heavy constructions al-ḥasanu wajhuhu/ḥasanun wajhuhu into lighter ones. This can be done in several ways (e.g., by omitting the definite article or the bound pronoun). Creating an annexation construction can be considered optimal for achieving this lightening, since an annexation is viewed as a partial merger of two words into one.34 There is a problem, however, with transforming a construction with rafʿ directly into an annexation, as the following excerpt demonstrates: This is because annexing an adjective to a constituent that [originally] received its rafʿ from [that adjective] is overtly unacceptable, since an adjective assigning rafʿ to an overt noun is co-referential with the constituent to which it assigns the rafʿ. For instance, in the sentence zaydun ḍāribun ġulāmuhu ʿamran ‘Zayd, his servant is the one hitting ʿAmr’, ḍārib ‘the one hitting’ is co-referential with ġulāmuhu ‘his servant’. Thus, [annexing the adjective to the noun to which it originally assigned rafʿ] amounts to annexing a word to something co-referential with it. This is unacceptable in a real [annexation], but it is a basic structure in an unreal one. [The Arabic speakers] make [the constituent to which the adjective

33

34

Baalbaki (2008:59–61) considers “lightening” to be one of the most frequent grammatical explanations in Sībawayhi’s al-Kitāb. Ibn Jinnī (Ḫaṣāʾiṣ I, 49.3) says that the grammatical explanations are related to lightness/heaviness. Among the phenomena that he explains by the speakers’ tendency towards maximal lightness are the fact that the agent takes the rafʿ and the direct object takes the naṣb (Ḫaṣāʾiṣ I, 50.1–3); the morpho-phonological shift *miwzān