The Foundations of Arabic Linguistics IV The Evolution of Theory PDF

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 1

Manuela E.B. Giolfo and Kees Versteegh
Contribution to a Modern Reading of Sībawayhi 10

Hassina Aliane
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 61

Ramzi Baalbaki
Sallaṭa/tasallaṭa, a Possible Parallel for ‘Govern’? 89

Michael G. Carter
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 157

Manuela E.B. Giolfo and Wilfrid Hodges
The Technical Terms taqdīr and tahfīf in Persian Classical Sources 182
Éva M. Jeremiás
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 212

Aryeh Levin
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 274

Beata Sheyhatovitch
Malay Grammar between Arab and Western Model 295

Kees Versteegh
Index 319
Notes on Contributors
Hassina Aliane
holds a Ph.D. in Computational Linguistics. She is director of the Digital Hu-
manities 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) lin-
guistics and (computational) linguistics more generally.
Georgine Ayoub
is professor of Arabic linguistics at the Institut national des langues et civil-
isations 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 Ara-
bic 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 Univer-
sity 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 gram-
matical 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 gram-
matical 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 his-
torical 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 coordi-
nated 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 Octo-
ber 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 lin-
guistics from Aix-Marseille Université. She was lecturer in Arabic at Exeter Uni-
versity (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 clas-
sique (Rome, 2017).
notes on contributors ix
Wilfrid Hodges
FBA is Emeritus Professor of Mathematics at Queen Mary, University of Lon-
don, specializing in mathematical logic and logical semantics. Since his retire-
ment 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 Zsig-
mond 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), Classi-
cal 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 Ara-
bic 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 Uni-
versity 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 grammat-
ical treatise Taḏkirat jawāmiʿ al-ʾadawāt by Muḥammad b. Aḥmad b. Maḥmūd
(Wiesbaden, 2012); The subjunctive mood in Arabic grammatical thought (Lei-
den, 2012); and, together with Almog Kasher, A critical edition of the grammat-
ical 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 lan-
guage.
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 linguis-
tics, 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 Ara-
bic grammatical and lexicographical tradition, in particular during the foun-
dational 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 sec-
ond 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 vol-
ume, 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 devel-
oping 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 manu-
script, which has long been hailed (see Humbert 1995) as one of the most inter-
esting 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 pro-
vides 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 under-
standing 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 con-
cepts 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 lin-
guistics, as it was first developed by Zellig Harris in 1946, and then concludes
that Mathematical Category Theory provides an even better model for the anal-
ysis of the system of hierarchically organized levels in Arabic grammatical the-
ory. Although she does not focus specifically on Sībawayhi’s Kitāb, much of her
analysis aims at a better understanding of the underlying theoretical presup-
positions 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’ trea-
tises on grammar too difficult to understand. This is why at a relatively early
stage textbooks for beginners began to be composed. After presenting a gen-
1 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 inflec-
tional 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 al-
Sarrāj (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 Confer-
ence 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 gram-
marian 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 disap-
peared 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 gram-
marians 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 construc-
tions 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 par-
ticiple (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 fol-
lows 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-
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 Gun-
del 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 ‘particulariza-
tion’. In the present contribution he focuses on the complementary term of
taʿrīf. Underlying his contribution is the idea that even though the Arabic gram-
marians did not have specific terms to distiguish between definiteness and
determination, they were aware of the difference between the two and devel-
oped 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 Con-
ference on the Foundations of Arabic Linguistic (Ayoub 2018) she had stud-
ied 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 per-
sons (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 explic-
itly 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
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 ‘prepa-
ration’ (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 sys-
tem, 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 grammat-
ical 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 Ara-
bic 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 consultation-
oriented 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 contribu-
tion 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
play a role), pointing out that al-Sīrāfī was more directly interested in ques-
tions of logic than other grammarians, while Ibn Sīnā had a deeper insight in
the relation between language and logic than other logicians. Their contribu-
tion 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 rec-
ognized even if it is only implicit, serves as a key to underlying assumptions
about conditionals apparently shared by al-Ḫalīl and Ibn Sīnā. These underly-
ing 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 Lewis-
Kratzer 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 Aus-
sant; 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 publica-
tions (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
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 mark-
ing. 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 sub-
title 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 pro-
fessionals, 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 gram-
marians 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 iden-
tity 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 Arabic-
speaking 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 com-
mon to the Arabic tradition and modern/Western linguistics, or even beyond
linguistics, if we follow Aliane’s paper. Comparisons between different linguis-
tic frameworks have long been regarded as hazardous because of the risk of
presentism. But some comparisons, as Owens (1988) has shown, can be illu-
minating. 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 lexicogra-
phy, 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 foun-
dations of Arabic linguistics that was initiated eight years ago by Amal Marogy
will bring more results in, hopefully, unexpected ways. The first new develop-
ment 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 activi-
ties. 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

Bibliographical References
Ayoub, Georgine. 2018. “Case and reference: The theory of mā yanṣarif wa-mā lā yan-
ṣarif 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 gén-
érale, 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’ ”. Specu-
lum 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 Ver-
steegh, 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 linguis-
tics. 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 state-
ment usually appears in Arabic Natural Language Processing literature. How-
ever, 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, compu-
tational 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, neverthe-
less 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 mathemati-
cians 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 how-
ever 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. To-
day, 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

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 struc-
tures.
This way of doing seems to correspond to the way Sībawayhi and his fol-
lowers deal with language structures: every language item has a position in an
abstract structure in Sībawayhi’s approach and the ultimate goal of the proce-
dures 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 appa-
ratus of the grammatical tradition by means of the tools of Mathematical Cat-
egory 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 grammat-
ical 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, mod-
ern studies of the Arabic grammatical tradition have as one of their primary
concerns the reconstruction of the methodological, theoretical and epistemo-
logical foundations that are thought to have informed the Arab grammari-
ans 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 analy-
sis unknown to the West until the 20th century”. In the same paper, Carter
remarks that “Sībawayhi’s method is remarkably similar to Immediate Con-
stituent Analysis, with which it shares both common techniques and inade-
quacies”.
The Arabic grammatical tradition constitutes for Arabic linguistics a—per-
haps sometimes heavy and unbearable but very precious—unavoidable her-
itage. 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 theoreti-
cal 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 Sīb-
awayhi’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 tradi-
tion.
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 con-
sidered 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 lan-
guage 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 Ara-
bic grammarians, on the contrary, is universal because it is the only linguistic
approach that is founded on universal methodological principles instead of the-
oretical 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 pri-
mary 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 tra-
dition 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-
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 Leo-

nard Bloomfield (1933), and then formalized and popularized by Zellig Har-
ris (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 environ-
ments 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 environ-
ments. 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 distri-
butional 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 rela-
tive 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 lin-
guiste opère des segmentations et des classements; mais les arguments
qu’il avance en faveur de ces opérations sont de nature distribution-
nelle. 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 gram-
marians 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 fol-
lowers 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 impor-
tance, 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:
It is to Sībawayhi’s credit that if one sets out to examine one of the con-
cepts 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, some-
thing 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 cri-
terion, 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 eco-
nomic 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. Sīb-
awayhi 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 realiza-
tion 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 actu-
alizable, while kitāb or kitābu zaydin or jāʾa zaydun are. An actualizable lan-
guage 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 cri-
terion to delimit linguistic structures and has the advantage of avoiding the
problem of what accounts for an environment in modern distributional analy-
sis.
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 nat-
ural 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 fol-
lows: “His belief in the existence of a norm, the force makes him regard certain
forms, constructions, etc. being basic or central, while others become periph-
eral.” 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 utter-
ances 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-
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 find-
ing 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 posi-
tion 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 abstrac-
tion 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. Accord-
ing 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 oper-
ations 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 real-
izations 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 (muqad-
dar), 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 con-

cerns 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 real-
izations 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 compo-
sitional. 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 transforma-
tions 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 Sīb-
awayhi’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 transforma-
tions. 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 pre-
served 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 los-
ing their properties) into the new utterance’s miṯāl, for instance when enter-
ing 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
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 con-
cept.
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 Sīb-
awayh’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 coales-
cence can be used as an extremely helpful tool of grammatical analysis.
Rediscussing all concepts making up the sophisticated apparatus of Sībaway-

hi’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 struc-
ture, 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 grammari-
ans’ analysis:
Sībawayh seems to be aware of the notion of distribution and environ-

ment 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 Ara-
bic 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 tradi-
tion, 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 admis-
sible 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 lan-
guage 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, when-
ever 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 obser-
vation 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, al-
kitāb, bi-kitāb …
– correspondence of different level structures: for instance the abstraction of
the miṯāl describing the construction of a syntactic unit from lexical items:
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 invari-
ants or their environments in themselves, we consider the transformations on
these invariants. At the structural level, in modification in form or environ-
ments 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 inves-
tigated 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 tra-
dition, 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 con-
cepts 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 equiva-
lence 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 Ara-
bic 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 dis-
covery 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 con-
ceptual 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 consid-
ered 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 generaliz-
ing 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 transfor-
mation, 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 nat-
ural transformations. Properties of objects are uncovered only through these
transformations. A category is a collection of objects and arrows which satis-
fies 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. Cate-
gory 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 con-
stituted of entities called objects. Constructions over objects and categories
allow us to determine these objects through universal participation in a cat-
egory, 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 pro-
vides tools to study concepts relevant to complexity, such as:
– 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 func-
tors) 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 men-
tioned 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 pro-
cess of category construction which is surprisingly close to the process of con-
struction of bāb in qiyās. Moreover, he even uses the word ‘measure’. The fol-
lowing 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 for-
mal 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 spe-
cialization, 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 dif-
ferent 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/cat-
egory 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 cor-
respondence of similar forms, correspondence of structures, and correspon-
dence 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 math-
ematical 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 math-

ématiques, 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 with-
out addressing the implications of those forms of structural analysis which are
clearly different from the usual class/membership-based analysis. Indeed, Cat-
egory Theory constitutes an important epistemological turn regarding struc-

turalism. By stating that language is a system of structures, structuralism cer-
tainly 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 subor-
dinate 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 par-
ticipates 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 ‘partici-
pation by universality’, and it is determined by the properties of the admissible
interactions that relate an object to a category. The conceptual tool that cor-
reponds 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 sen-
tence 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 for-
mal 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 founda-
tions 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 struc-
ture 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 distribution-
alism to stand as a theory of language is to consider operations resulting in form
changes, rather than considering the forms themselves (distributions and envi-
ronments), 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 linguis-
tic 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 com-

position 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 dis-
cover 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 Ara-
bic speakers manipulate the code of their language. This is very important since
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 proper-
ties of the structure.
Secondly, and this principle is a consequence of the first one, objects partic-
ipate 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 participat-
ing 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 serv-
ing their (communicative) intentions. In fact, as we have seen above, the idea
of considering language objects as participating in structures rather than con-
sidering 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 Mathemat-
ical Category Theory may be the adequate framework to express the formal
premises of Arabic grammatical tradition.
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”. Ap-
proaches 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 gram-
matical analysis”. Zeitschrift für arabische Linguistik 2.7–22.
Baalbaki, Ramzi. 1999. “Coalescence as a grammatical tool in Sībawayhi’s Kitāb”. Ara-
bic 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 Ara-
bic 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 the-
ory”. Proceedings of the National Academy of Sciences 28:12.537–543.
Eilenberg, Samuel and Saunders Mac Lane. 1945. “General theory of natural equiva-
lences”. 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’ oeu-
vre 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.
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 na-
ture, 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 Con-
grès 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 tra-
dition”. 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 gram-

matical 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 per-
sons 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 “psy-
chic 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 par-
ticipants, while the third one is ‘absent’ in enunciation. Benveniste, too, dis-
tinguishes 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 indi-
vidual discourse where it is pronounced, and by this, it designates the speaker
[…]”.2
1 Benveniste (1966: I, 252).

2 Benveniste (1966: I, 252).

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 ʿalay-
ka, fa-huwa ʾawwalu kulli kalāmin la-ka, bi-hi taʿṭifu l-mukallama ʿalay-
ka).
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 state-
ment 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 allo-
cutaire).
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 nec-
essary 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 con-
sequently, 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 artic-
ulating 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 men-
tioned 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, pred-
ication. The term al-ġāʾib is always associated with a performative notion des-
ignating 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 con-
text, ‘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, respec-
tively, 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).
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’ (ʾinna-
mā 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 sub-
ject 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 utter-
ance 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 asym-
metry 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 appro-
priation 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 (ʿalā-
mat 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 prag-
matic 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 al-
marfūʿīn)
– ch. 206–209: the accusative pronouns (hāḏā bābu ʿalāmati l-muḍmarīn al-
manṣūbīn)
– 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 hier-
archy is established between pronouns with respect to enunciation.
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 dif-
ferent 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 appo-

sition to hum, because the verb must necessarily have a subject. The sub-
ject8 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 wa-
ḍamīru 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 qaw-
maka, 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 ḍamīr-
ihi; 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 l-
fiʿlu 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, raʾaytu-
humā, 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 This expression may be translated as follows: “Their beasts, on which they rode, were
swift”.
10 “I spur the animal on”.
11 See below for a comment on these two translations.
between the strong nominative pronouns (e.g. ʾantum ḏāhibūn) and the nomi-
native 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 nomina-
tive 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 ʿalā-
matu 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 proba-
bly 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 techni-
cal 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 pro-
nouns, 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 con-
sequence 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 pur-
poses. 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 exam-
ple “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 al-
ʿArab. 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 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 expla-
nation of the surface structure”.
13 Cf. Ayoub (1990:3f.).
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 l-
niyya). 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 usu-
ally 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 quo-
tation in 3.2.1, which illustrates perfectly the opposition between muẓhar/muḍ-
mar, 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 ( fa-
ʿraf 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 tam-
run ‘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 yadul-
luka 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 fa-
ḍribhu, ʾiḏā kāna mabniyyan ʿalā mubtadaʾin muẓharin ʾaw muḍmarin fa-
ʾammā 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
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 pas-
ture 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 utter-
ance, 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 state-
ment.
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 utter-
ance that have an empirical effect on the terms of the utterance. These non-
observable elements are also supposed to be mental representations of the
speaker. In other words, the mental representation of the utterance may be dif-
ferent 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 unex-
plained 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 struc-
tural 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 rep-
resentations.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 al-
ʿayn 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 al-
ʾAʿšā 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/wa-
lā 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 al-
baqiyyat-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 ele-
ment is known by the addressee, it can be hidden (or ‘killed’: fa-lammā ʿurifa
maʿnāhu l-murādu ʾumīta l-fiʿlu; ʿAyn III, 215).
16 See for instance Chomsky (1981).

17 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).
18 The published text gives: ʾalqaw. But the meaning of the word makes this reading diffi-
cult 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: al-
baqiyya min ʾabqayta ʿalayhi wa-stabqaytahu ʾiḏā raʿaytahu wa-raḥimtahu (Dīwān I, 311).
The exact same reason for ellipsis or elision (ḥaḏf, ḫazl) in found in Sīb-
awayhi’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 zay-
dan 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” ( fa-
kullu 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 qal-
bika 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 exam-
ple 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 con-
cealed 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 in-
cludes, 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 al-
hilālu 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-mu-
ẓhari fa-qawluka hāḏā zaydun fa-ḍribhu wa-ʾin šiʾta lam tuẓhir hāḏā wa-
yaʿmalu 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 interpre-
tations 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 pro-
noun. 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 necessar-
ily 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.
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 fol-
lowing 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 expres-
sion 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] ʿalā-
matu l-mutamakkin (Kitāb II, 157); ʿalāmat al-tawkīd (Kitāb II, 157); al-ʿalāma al-
muḫtaṣṣa (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 The example analyzed is ʿalaykum ʾanfusikum, and the muḍmar in question is -kum in
ʿalaykum.
22 We will see that this interpretation is the one retained by some of the later grammarians.
23 Cf. also Lane 1863 s.v. ʿalāma. http://www.tyndalearchive.com/tabs/lane/.
24 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 him-

self, 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 l-
muḫāṭabu 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 ḍamī-
run 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.

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 ʿalā-
matan li-l-jamʿi, kamā faʿalta ḏālika bi-ḍarabat ḥīna kānat ʿalāmatan li-l-
taʾnīṯi).
Kitāb II, 7
This assigns a morphological meaning to the notion of ʿalāma used for pro-
nouns. 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 fre-
quently 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 trans-
late 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 (mas-
culine, 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 sub-
ject 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 realiza-
tion’. 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 under-
stand 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 consid-
ers 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 sig-
nificance 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.
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 al-
muḍmar 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 ʿalā-
matu 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 pro-
nouns 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 pho-
netic 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 ref-
erence. What is muḍmar, whether it is an explicit or implicit element, keeps its
reference implicit or concealed: its form does not indicate its specified refer-
ence 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 fol-
lowing 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 deter-
mining 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 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 con-
sistent with all definitions of definite elements in the same chapter. In each definition,
Sībawayhi introduces the formula “you intend to designate something [specific], exclu-
sively” (turīdu šayʾʾan bi-ʿayni-hi). Moreover, yaʿlamu-hu would be redundant with the first
part of the quotation (man tuḥaddiṯu qad ʿarafa man taʿnī).
28 For a translation and discussion of this passage, see also Marogy (2010:102, 107 f.).
29 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 wa-
ʾašbāha hāḏā lam yakun kalāman.
speaker pronominalizes a noun when he knows that his addressee has iden-
tified 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 pronominal-
ization 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 l-
muḍmara 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 sen-
tence 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 consti-
tute 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-li-
ḏālika 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 naḥ-
wiyyūn, 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 per-
son 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 qabī-
ḥun 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 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 con-
text 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 pro-
noun, 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.
32 See also for a discussion Peled (2009).
before the nearest” (wa-ʾinnamā qabuḥa ʿinda l-ʿArabi karāhiyata ʾan yabdaʾa l-
mutakallimu 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 situa-
tion: 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 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 exclu-
sively 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 Mar-
ogy 2010:88 for another interpretation).
34 Sībawayhi believes there is no difference between the three persons regarding the obliga-
tion 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 expres-
sions 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, al-
muḥaddaṯ ʿanhu.
7 The Development of the Terminology
The ambiguity of the reference of the term muḍmar, which refers to two differ-
ent 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 spe-
cific 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 mean-
ing 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. More-
over, 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 occur-
rences 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).

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 prece-
dence, 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 al-
Māzinī’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 l-
mutakallimi wa-ḍamīruhu) ( fa-hāḏihi l-ḥurūfu ʿinda Sībawayhi fī wuqūʿihā
ʾasmāʾa marratan wa-ḥurūfan marratan bi-manzilati l-tāʾi fī qawli-ka qul-
tu wa-qāl-at fa-l-tāʾu fī qul-tu ismu l-mutakallimi, wa-l-tāʾu fī qālat ʿalā-
matun tuʾḏinu bi-ʾanna l-fiʿla li-l-muʾannaṯi wa-qad qāla ʾAbū ʾUṯmāna wa-
ġayruhu min al-naḥwiyyīna ʾinna l-ʾalifa fī qāmā wa-l-wāwa fī qāmū ḥar-
fāni 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 syn-
tactic 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 al-
qawlu 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 l-
bābi 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 (al-
mutakallim and al-muḫāṭab), on the one hand, and the third person
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 pro-
nouns according to the speech situation are firmly established. This hier-
archy determines some morphological facts, namely the order of suffixa-
tion of pronouns. It seems that there was a debate about these facts before
Sībawayhi, and after him.
iii. 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.
iv. 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 mark-
ers 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.
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 al-
muqtaḍab. 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 al-
ʿĀmma 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-naḥwiy-
yīn 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 Ara-
bist, 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 utter-
ance 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. 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 gén-
érale, I, 251–257. Paris: Gallimard, 1966.)
Benveniste, Émile. 1958. “De la subjectivité dans le langage”. (Repr., Problèmes de lin-
guistique 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: Interpre-
tation 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”. Stud-
ies 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 standard-
ization. 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 lin-
guistics. 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 tradi-
tion. 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 under-
stood 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 grammar-
ians 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, al-
Jāḥiẓ (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 Mus-
lim’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).

62 baalbaki
beneficial only if it did not extend beyond certain prescribed limits.5 Accord-
ing 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 sur-
prise. With the growing need for teaching grammar, it must have become obvi-
ous 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 begin-
ners. 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 pur-
poses.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 peda-
gogical 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 prece-
dence 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. Al-
ʾAḫfaš’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 incompre-
hensible 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 peda-
gogization 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 deter-
mine. 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 descrip-
tions) 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 al-
Nadīm also refers to al-Jarmī’s book as Muḫtaṣar al-naḥw11 since this strength-
ens the possibility that books with a similar title were indeed meant for begin-
ners. 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 al-
ʿAbbās 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 Ġas-
sān 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 al-
Muhallabī (d.?).17
10 Ibn al-Nadīm, Fihrist I, 162, 260.

11 Ibn al-Nadīm, Fihrist I, 267.
12 Ibn al-Nadīm, Fihrist I, 211.
13 Ibn al-Nadīm, Fihrist I, 140, 141, 202, 240, 141, 177, 255, 252 respectively.
14 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.
15 Ibn al-Nadīm, Fihrist I, 196, 248, 265.
16 Ibn al-Nadīm, Fihrist I, 263, 265.
17 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, al-
Mubarrad’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 Ḫāl-
awayhi’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 al-
mutaʿallim 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 simpli-
fied 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 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.
19 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.
20 Cf. Carter (1979:267–273).
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 ped-
agogical 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 peda-
gogical manuals in our possession, and thus, in comparison, its appropriateness
for beginners who have little or no prior knowledge of grammar is question-
able. 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 sim-
ple 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 Ibn al-Ṭarāwa, ʾIfṣāḥ 16.

22 Ibn al-ʾAnbārī, Nuzha 233; Yāqūt, Muʿjam II, 813; Suyūṭī, Buġya I, 496.
23 Cf. Šalabī (1958:516); Carter (1990:131); Baḥr al-Marjān’s introduction to Fārisī, ʾĪḍāḥ 22 f.
24 Zubaydī, Ṭabaqāt 120; Šalabī (1958:517); Baḥr al-Marjān’s introduction to Fārisī, ʾĪḍāḥ 22.
25 As in references to the Baṣriyyūn; cf. Fārisī, ʾĪḍāḥ 117, 185, 252.
26 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).
27 See, in particular, Ibn al-Ṭarāwa, ʾIfṣāḥ 37; also 31, 34.
28 Ibn al-Ṭarāwa, ʾIfṣāḥ 35, 75, 81.
66 baalbaki
2 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 com-
mentaries 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 stu-
dents, 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 al-
ʾAzharī’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 fol-
lowing the discussion of its phonetic and morphological aspects, which obvi-
ously 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 al-
ʾiʿrāb, 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 al-
29 Sezgin (1984:88–94, 174–176). Note that some commentaries are too lengthy and compli-
cated 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.
30 Cf. Zajjājī, ʾIḍāḥ 91: wa-yusammā l-naḥw ʾiʿrāb.
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 al-
naḥwiyya 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. Chap-
ter 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) lex-
ical 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 intro-
ductory 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 Zamaḫšarī, Mufaṣṣal 5. Note that the book was published under the title al-Mufaṣṣal fī ʿilm
al-ʿArabiyya.
32 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).
33 For a detailed study of both types, see Baalbaki (2014:213–225).
34 Ibn Hišām, Muġnī I, 9–10.
35 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 cor-
responds 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 sen-
tence ( 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 var-
ious perspectives. Given that the second chapter, which discusses the jumla,
is essentially a study of ʾiʿrāb as applied to sentences, and that other chap-
ters 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.

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 appreci-
ated 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 tradi-
tion.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 real-
ize 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 chap-
ters 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 accom-
plished 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 commit-
ted 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 tech-
niques 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 Ibn Hišām, Muġnī II, 382–428.

38 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.
39 See, for example, the long list of commentaries on the ʾIʿrāb prepared by Nīl (1985:26–43).
40 Ibn Hišām, Muġnī II, 664, 651, 653 respectively.
70 baalbaki
concept of the ṣināʿa in ʾiʿrāb; the level of proficiency among students and mas-
ters 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 pertain-
ing 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 com-
plement 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 expres-
sions are the essential building blocks in the establishment of a solid ṣināʿa that
is widely taught to beginners. Furthermore, the perfection of the ṣināʿa necessi-
tates the establishment of certain conditions in formulating ʾiʿrāb expressions.
Accordingly, Ibn Hišām highlights the importance of two such conditions that
41 See the term in Muġnī II, 527, 539, 649; ʾIʿrāb 107.
42 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 tradi-
tional study of grammar.
43 Ibn Hišām, Muġnī I, 284; II, 664; ʾIʿrāb 105f.
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 con-
cerned, 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 wa-
qalbihi 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 Sim-
ilarly, 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 expres-
sions 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 Note that both terms are used in the title of the fourth chapter of Ibn Hišām’s ʾIʿrāb 103.
45 Ibn Hišām, Muġnī II, 651, 664; cf. ʾIʿrāb 105.
46 Ibn Hišām, Muġnī II, 651.
47 Ibn Hišām, ʾIʿrāb 105f.; cf. Muġnī I, 227, 56, 117 respectively.
48 Ibn Hišām, ʾIʿrāb 107; Muġnī II, 672. The expression uniformly used in this case is wa-
fāʿiluhu ḍamīr mustatir taqdīruhu …
49 Ibn Hišām, ʾIʿrāb 107; Muġnī II, 667, 672.
50 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 (imper-
ative); (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 Pre-
cision 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 demon-
strate the intricate relationship between the ṣināʿa and meaning (maʿnā)—an
issue that is frequently faced in deciding the most appropriate ʾiʿrāb, particu-
larly in those cases where the construction is tricky or requires analysis that
goes beyond what may seem to be apparent or commonplace. The first objec-
tion 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 Ibn Hišām, Muġnī II, 667f.

52 Chapter Five (II, 527–650).
53 Chapter Five II, 527.
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 wa-
ʾannahu 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 li-
mā 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 sen-
tence.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 con-

56 Chapter Five II, 546, 551.
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 syntac-
tical analysis in the study of ʾiʿrāb and ability to analyze structure from sev-
eral 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 expres-
sion 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 mub-
tadiʾūn 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

59 Ibn Ḥajar, Durar II, 309; Suyūṭī, Buġya II, 69.
apply to the inability to differentiate between the -in that indicates the geni-
tive 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 pronomi-
nal 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 nomi-
nal 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, obvi-
ously 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 gram-
mar 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 indica-
tive and subjunctive respectively, whereas the prefix in each of the two verbs
indicates person.63 It is noteworthy that although Ibn Hišām reports this anec-
dote 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 wide-
spread weakness among students of grammar, it is clear that he was not satis-
fied 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 Ibn Hišām, Muġnī II, 670f.

64 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.).
65 Ibn Ḥajar, Durar II, 308; Suyūṭī, Buġya II, 68.
66 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 (simpli-
fication) 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). Simi-
larly, 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 qui-
escence, 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 crooked-
ness’ (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 Ara-
bic 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. al-
Zubaydī (d. 379/989)71—with the general ignorance of their contemporaries,

70 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 ʾanza-
lahu qayyiman, or to the Holy Book, hence ʾanzala l-kitāba qayyiman.
71 Ibn Durayd, Jamhara I, 39–40; Zubaydī, Laḥn 8.
their lack of interest in seeking knowledge, and the need of authors to sim-
plify 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 com-
ments 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 draw-
ing 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 cer-
tain 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 notwithstand-
ing, 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 erro-
neous.75 He notes that mistakes of this type are abundant (kaṯīra) and cites
twenty cases by way of illustration. Some of these illustrations, however, per-
tain to grammatical issues that are related to ʾiʿrāb but are not, strictly speaking,
72 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.
73 Ibn al-ʾAṯīr, Nihāya I, 11.
74 Ibn Manẓūr, Lisān I, 8.
75 The title of this chapter is Fī l-taḥḏīr min ʾumūr uštuhirat bayna l-muʿribīn wa-l-ṣawāb ḫilā-
fuhā; 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 expres-
sion 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 experi-
ence during my training in ʾiʿrāb in secondary school as well as my undergrad-
uate study as a student majoring in Arabic language, let alone my teaching
experience of ʾiʿrāb for about four decades, but more objectively on a rep-
resentative 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

77 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.
79 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.
80 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).
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 sev-
eral 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 ġāl-
iban (an adverbial denoting anticipated time and mostly indicates the
conditional). Ibn Hišām finds fault with this expression from several per-
spectives. 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 fol-
lows 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 Ibn Hišām, Muġnī I, 257–260; cf. ʾIʿrāb 83–85.

82 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).
83 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).
84 Note that Ibn Hišām’s use of the word rābiṭa (Muġnī II, 652) and his view that the sen-
tence ( 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.
85 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 prohibi-
tion. 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 jus-
sive 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 sub-
stitute 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 inter-
changeable.
(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 pur-
pose 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 al-
ism 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 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).
87 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.
88 Cf. Qabāwa (1978:530).
89 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).
90 Other than Muġnī II, 663, see I, 138.
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 sec-
tions 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 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).
92 Ibn Hišām, Muġnī II, 664; cf. ʾIʿrāb 106.
93 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 demon-
strates 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 syn-
tactical issues that make up the traditional grammars—on the basis of the ʾiʿrāb
of specific particles, constructions, etc., ample illustrations, including exam-
ples 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 por-
tion of the material that makes up the works which follow the prevalent tradi-
tional 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 founda-
tions 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 gov-
erns 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 ele-
ments 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 verba-
tim 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 Ibn Hišām, Muġnī I, 37; II, 668; ʾIʿrāb 106.

95 Ibn Hišām, ʾIʿrāb 106.
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 teach-
ing 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 al-
zamān, 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 li-
l-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 inad-
vertently 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 ʾAsmar (1969:9); Rājiḥī (1979:247); ʿĀṣī and Yaʿqūb (1987: I, 69); Yaʿqūb (1988:78); ʾAltōnjī and
ʾAsmar (1993: I, 25).
98 Qabāwa (1978:411, 420); ʿĀṣī and Yaʿqūb (1987: II, 937f.).
99 Ibn Hišām, ʾIʿrāb 69; but see Muġnī I, 80, where ism li-l-zaman al-māḍī is used.
100 Ibn Hišām, Muġnī I, 81.
84 baalbaki
A Primary Sources
ʾAbū l-Ṭayyib, Marātib = ʾAbū l-Ṭayyib ʿAbd al-Raḥmān ibn ʿAlī al-Luġawī, Marātib al-
naḥwiyyīn. 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-naḥ-
wiyya 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 al-
Marjān. 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 al-
Qawmī, 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: Mak-
tabat 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-
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 al-
luġa. 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, al-
Durar 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 al-
Zajjājī. 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, al-
ʾIʿrāb 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 al-
Sayyid 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: Maṭ-
baʿat 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 al-
ʾaʿẓam. 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, Muḫ-
taṣar 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-Baṣ-
riyyīn. 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 al-
Kutub 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.
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.
Zubaydī, Laḥn = ʾAbū Bakr Muḥammad ibn al-Ḥasan al-Zubaydī al-ʾAndalusī, Laḥn al-
ʿawāmm. Ed. by Ramaḍān ʿAbd al-Tawwāb. Cairo: al-Maṭbaʿa al-Kamāliyya, 1964.
Zubaydī, Ṭabaqāt = ʾAbū Bakr Muḥammad ibn al-Ḥasan al-Zubaydī al-ʾAndalusī, Ṭaba-
qāt 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 (al-
ʾalsuniyyāt). 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. 2009. “The place of al-Jāḥiẓ in the Arabic philological tradition”. Al-
Jāḥiẓ: 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 litera-
ture: 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 Ori-
ental 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-ṣarfiy-
ya. 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: Jā-
miʿat 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 demon-
strated 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 sal-
laṭa/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 cate-
gory 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.).

90 carter
1 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 rep-
resents 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 with-
out saying that the Arabic spellings here,4 not to mention the linguistic argu-
mentation, are highly suspect, which is one reason why the attribution to Ibn
Qutayba can be safely rejected. Nevertheless it is an interesting example of sal-
laṭa 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 chap-

ter 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 equa-
tional 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 obscu-
rity 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 ḥay-
ṯumā 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 construc-
tion ʾaynamā takūnanna ʾātika ‘wherever you may be I shall come to you’, illus-
trating 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 grammat-
ical 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 tech-
nical 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
Leaving aside the dubious attribution to al-Zajjāj,6 there is at least an unmis-

takable association of sallaṭa with grammatical operation, ʿamal. The issue
here is the apparent paradox that the terms denoting the antecedents of adjec-
tival and relative constructions are passive in form, mawṣūf and mawṣūl, imply-
ing 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 oper-
ation 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 pre-
vents 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 syntac-
tical 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.
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 sen-
tence 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 zay-
dan 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 con-
flict of operators in the sentence type ḍarabanī wa-ḍarabtu zaydun or zaydan.
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 muṭ-
laq, 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 trans-
late 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 ex-
pressed 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 veri-
fied: 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 obvi-
ous 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 pref-
ace 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.
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, sug-
gesting 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 evi-
dence 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), “Sim-
ples 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 sal-
laṭa.
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 exam-
ples 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 evi-
dence 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 imper-
sonal 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 gram-
matical application.
I had hoped to show that sallaṭa was brought into the grammatical vocabu-
lary 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
prefers ʾaʿmala in the context of tanāzuʿ, the conflict of operators seen in the
sentence type ḍarabtu wa-ḍarabanī ʾaḫawāka. His analysis is constructed with-
out 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 language-
user, 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 Kul-
liyyāt 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 lin-
guistic 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.
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 grammari-
ans 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 Soci-
ety, 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’Ibnu-
Malik. 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. Bagh-
dad, 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 al-
ʾĀjurrūmiyya. 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 al-
Qurʾān. 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”. Ara-
bica 26.267–273.
Carter, Michael G. 1989. “The Arabic and Medieval Latin terms for ‘governing’ ”. Specu-
lum 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 arabis-
chen zum ersten Mal übersetzt. Berlin: F. Dümmler.
Gille, Christiane. 1995. Das Kapitel al-Mauṣūl (“Das Relativum”) aus dem Manhaǧ al-
sālik 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 syn-
taxe par Ibnu Hijām. Leiden: E.J. Brill.
Guillaume, Jean-Patrick. 1983. “Fragments d’une grammaire oubliée: Relations prédica-
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 ‘hysteron-

proteron’ 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 utter-
ances to be the result of specific mental strategies that the speaker follows,
choosing between various possible operations to convey the intended mean-
ing (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 taq-
dīm wa-taʾḫīr and the meaning of the utterance is evident in al-Jurjānī’s treat-
ment (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-

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 complemen-
tarity 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 cen-
tury 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 suc-
cessors’ approach to the same linguistic phenomenon shows that the speaker
is present there as a learner who is dictated what is permissible and not per-
missible.
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 utter-
ance] 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 wa-
ʾaḫḫarta 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] (ka-
ʾannahum 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 exam-
ined. 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 classifica-
tion 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

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 recipi-
ent has met Zayd or Bišr, so the name is placed before the verb to communi-
cate 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 bet-
ter 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 wa-
ʾinnamā 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.
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 interroga-
tive 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” (wa-
ʾinnamā 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 success-
ful 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 pro-
hibitive 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 command-
ing 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 command-
ing and prohibiting act is addressed. The speakers’ interest in successful com-
munication 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 convey-
ing 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

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 ʾaʿlam-
tahu miṯla mā ʿalimta).11
On the other hand, when the speaker chooses to advance ḥalīman, the inten-
tion is to communicate first the attribute of the person talked about. The lis-
tener 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 per-
son 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 con-
form 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. How-
ever, 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 Sībawayhi, Kitāb I, 22. See a similar example of hysteron-proteron with ʾinna in Sībawayhi,
Kitāb I, 285.
13 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 nom-
inative 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-mub-
tadaʾ).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
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).
14 On the terms musnad, musnad ʾilayhi and ʾisnād see Levin (1981:145–165).
15 Sībawahi, Kitāb I, 278.
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 pred-
icate. 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 utter-
ance. 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), singu-
lar and plural (wāḥid wa-jamʿ), and definite and indefinite (maʿrifa wa-nakira).
According to Sībawayhi, “verbs are heavier than nouns because nouns are [hier-
archically] 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 mas-
culine 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 deal-
ing with the notions of number and (in)definiteness: “The topic is first in the

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 l-
maʿrifa).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 relation-
ship 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 al-
musnad 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 al-
ism 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 inter-
rogative words, such as ʾayna zaydun ‘where is Zayd?’, kayfa ʿabdullāhi ‘how
is ʿAbdallāh’, since, as Sībawayhi explains “the meaning of ʾayna is ‘in what
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 ṣāhi-
bihi fa-lammā jumiʿā staġnā ʿalayhimā l-sukūtu ḥattā ṣārā fī l-istiġnāʾ ka-
qawlika 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 utter-
ance 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 communi-
cate a certain mode of speech to the listener. Thus, the notion of al ḥaqq bi-l-
ṣadāra 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 analyz-
ing 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 cen-
turies.

116 dayyeh
3 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 linguis-
tic 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 al-
ʾUṣūl 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 foun-
dations 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-Muqta-
ḍab, 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 appro-
priate when the utterance clarifies the meaning” (wa-ʾinnamā yaṣluḥu l-taqdīm
29 Cf Bohas et al. (1990:4f.) and Baalbaki (2008:236f.).

30 Cf. Owens (1990:242f.) and Bohas et al. (1990:10f.).
31 Mubarrad, Muqtaḍab III, 95.
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-l-
mafʿūl).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 antepos-
ing 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 cer-
tainty:
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 Mubarrad, Muqtaḍab III, 95f.

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-l-
yaqīn).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] your-
self’; 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 vari-
ations 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 struc-
tures 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 speak-
ers’ 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

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 al-
Muqtaḍab’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 devel-
opment 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 al-
ʾUṣūl, 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 hysteron-
proteron. He starts this chapter by identifying thirteen cases where anteposing
a certain word in the utterance is not allowed. Then he delves into explain-
ing 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 post-
posing 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 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’.
38 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 chang-
ing 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 al-
Sarrāj 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 disci-
pline that the advanced [speakers] deduced by examining the speech of Arabs
in order to grasp the purposes intended by the originators of the language” (wa-
huwa 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 Ara-
bic 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

41 Ibn al-Sarrāj, ʾUṣūl I, 35.
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 wa-
ḏikr 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.
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.

122 dayyeh
B Secondary Sources
Ahmar, May. 2001. al-Taqdīm wa-l-taʾḫīr bayna al-naḥw wa-l-balāġa. M.A. thesis, Amer-
ican University of Beirut.
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.)
context of the Arabic grammatical tradition. Leiden: E.J. Brill.
Bohas, Georges, Jean-Patrick Guillaume, and Djamel Eddine, Kouloughli. 1990. The Ara-
bic 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.
ization. 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 mathemat-
ical 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 compre-
hensive 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 design-
ing 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 al-
Ḫalīl’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 dictio-
nary 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 distin-
guish 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’, ‘under-
lying 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.

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 corol-
lary 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 tradi-
tional 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 letter-
segments 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 math-
ematician, 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 125
2 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. Baal-
baki (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 tradi-
tional 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 lin-
guists 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 con-
sidering 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 princi-
ple). 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 dictionar-

ies 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 vari-
ation 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 aug-
mented letters (ḥurūf al-ziyāda, Jamhara I, 10), and on a number of morpho-
logical 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 al-
Bāriʿ 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 descrip-
tion 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 accord-
ing 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 per-
6 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 cen-
turies.
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 begin-
ning 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 letter-
segment 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 aim-
ing 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 letter-
segment /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 letter-
segment 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 mor-
phological criteria (nouns and adjectives on one side, verbs on another), which
renders the operation of finding a given root or word very difficult. The differ-
ence 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 dictio-
nary 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 phonolog-
ical 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 exhaus-
tively 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 elab-
oration 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 stand-
point 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 arrange-
ments. 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.
6 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 dictionar-
ies 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 subse-
quently 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 mem-
ories 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 mem-
ory 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, liter-
ate 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 con-
temporary memory skills. Manuscript dictionaries were read aloud in ‘reading
sessions’ (majālis), and their contents were remembered by learned partici-
pants, 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 intro-
duction 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 al-
Jawharī’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 fol-
lowed 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 cov-
erage of lexical knowledge; and (ii) oral and manuscript transmission, associ-
ated with a high level of memory skills.
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 al-
ʿĀmma 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
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 al-
ʿArabiyya.
Ṣāḥib ibn ʿAbbād, Muḥīṭ = ʾAbū l-Qāsim ʾIsmāʿīl al-Ṣāḥib ibn ʿAbbād (d. 385/995), al-
Muḥīṭ 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 laby-
rinthe 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 fige-
ment 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 theo-
retical 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 his-
tory 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 attribu-
tion 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 Sīb-
awayhi’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 Hum-
bert (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 Deren-
bourg’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 Repub-
lic 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 “authorita-
tively 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 inter-
nal glosses” that are found in all other manuscripts, and its text seems to con-

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 inter-
mediaries 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 conclu-
sion 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 read-
ings 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 manu-
script, 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 Deren-
bourg’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 Muḥam-
mad 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 iden-
tifications of the poetical verses. As for Ḥamza, he simply reproduced Būlāq’s
edition and added in the margins a selection taken from two different com-
mentaries.
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 Deren-
bourg 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 18th-
century copy made on a descendant of an autograph of al-Zamaḫšarī
(d. 538/1144). This descendant (= Humbert 2Ç), dated 647/1249 was prob-
ably 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 recen-
sion 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 prin-
ceps 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 Deren-
bourg, 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 edi-
tion 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 Hum-
bert = 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 Hum-
bert (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 stud-
ied 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 men-
tioned.
– I will follow Humbert and call M1 the first hand in Ambrosiana (5th/11th cen-
tury?) and M2 the corrector (715/1315).
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 con-
sideration 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 fur-
thest 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 (repre-
sented 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.
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 ver-
sions, 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 al-
ḫafīfa) 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 gen-
eral. 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 col-
lated 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 gram-
marians 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 grammati-
cal 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 mis-
takes 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
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 excep-
tional manuscript, that has not received the attention it deserves is the auto-
graph by the Andalusian grammarian ʾAbū l-Ḥasan ʿAlī Ibn Ḫarūf al-Ḥaḍramī
(d. ca. 609/1212), kept in Paris, Bibliothèque nationale, arabe 6499, which con-
tains a sophisticated critical recension. Even though our understanding of Sīb-
awayh’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 discrep-
ancy 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 con-
sistently 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 muqad-
dam 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 ُ ‫[ والتضعي‬3] ‫[ هذا باب مضاع َف الفعل واختلاف العرب فيه‬2] [‫ظ‬61]
ِ ‫ف أن يكون آ‬
ّ ُ ‫ت وا ِن ْق َد َْد‬
3‫ت‬ ُ ‫ وا ِجْترَ َْر‬2‫ت‬ ّ ُ ‫ ر َد َْد‬1 ⸂‫ن من موضٍع واحد وذلك ⸃نحو‬
ّ ُ ‫[ و]صـ؟[←⸃و⸂د َْد‬4] ‫ت‬ ِ ‫حرفا‬
ُ ‫خر‬
ِ ‫ف الآ‬ ُ ْ ‫ت وا ِْطم َ ْأن َن‬
ُ ‫ت فإذا تحر ّك الحر‬ ُ ‫[ وا ِْحمار َْر‬5] ‫ت‬
ُ ‫ وا ِْحمرَ َْر‬5‫ وترَ اد َْدنا‬4‫ت‬ ّ ُ ‫وا ِْستعَ ْد َْد‬
ُ ‫ت وصار َْر‬
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.
‫ من موضٍع واحٍد‬9‫ لأن ّه لم ّا كان‬8⸂‫ ⸃ َأولى به‬7‫[ الإدغام وذلك فيما زعم الخليل‬6] ‫ على‬6‫ب م ُج ْم ِعون‬
ُ ‫فالعر‬
ِ ‫[ الآ‬8] ‫ أن يرَ فعوا ألسنتهم من موضٍع ثم يعُ يدوها إلى ذلك الموضع للحرف‬10‫[ عليهم‬7] ‫ل‬
ّ‫خر فلما‬ َ ُ َ‫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
ِ ‫ ي َسكـنا‬22‫[ لأّنهما لا‬3] ‫خر‬
‫ن جميع ًا وهو قول غيرهم‬ ِ ‫ذكرنا من المتحر ّكين في ُسكنون الأّول و يحرِ ّكون الآ‬
َ ْ ‫[ فإذا كان الحرف ال ّذي قبل الحرف الأّول من الحرفينِ ساكناً ألقي‬4] ʘ ٌ ‫من العرب وهم كثير‬
‫ت‬
6 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.
7 According to Bakkāʾ, Mawṣil adds raḥimahu Allāh taʿālā.
8 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.
9 A, B, L, Kolkata and Bakkāʾ have the more correct dual kānā. M1 has the singular kāna and
M2 did not correct it.
10 Bakkāʾ has the obviously erroneous ʿalayhi.
11 Kolkata has wa-jtarrū.
12 Kolkata has wa-nqaḏḏū.
13 Kolkata has wa-staʿiddā wa-ḍārrā with a clear ending ʾalif maqṣūra and superscript ʾalif.
14 A has the bizarre form yurāddanāni.
15 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.
16 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?
17 Kolkata has yaskunu.
18 Kolkata has urdudū.
19 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.
20 A and Bakkāʾ have the less correct ḥarrakū. B, L and Kolkata have ḥarrakūhu. M1 has ḥar-
rakūhu and M2 did not modify it.
21 Kolkata has only baʿda, without the suffix pronoun.
22 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
َ ْ ‫ ألقي‬26ّ‫[ َأر ُ َد‬9] ّ‫ض وإْن ترَ ُ َد‬
‫ت‬ ّ َ َ ‫ 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
ُ ‫[ وإن كان الساك‬11] ‫وذلك قولك ر ُ َدّا ور ُ ُدّوا‬
َ ْ ‫ ألقي‬30ٌ ‫ حاجز‬29‫ن ال ّذي قبل الأّول بينه و بين الألف‬
‫ت‬
‫[ عن الأصل‬13] ‫ في حال صاحبه‬31 ⸂‫ل واحٍد منهما ]يتحر ّك[←⸃ي َتحو ّل‬ ّ ‫[ عليه حركة الأّول لأ‬12]
ّ ‫نك‬
ّ ‫[ لأ‬14] ‫تحذف الألف‬
‫ن الحرف ال ّذي بعد‬ َ ‫ ولا‬32 ⸂‫ض‬ّ َ َ ‫ت ذلك في ر ُ َدّ وف ِ َر ّ و]×××ا[←⸃ع‬
َ ‫كما فعل‬
ْ ‫ فصار‬34ّ َِ‫ وإْن ت َْشم َئزَِ ّ َأْشم َئز‬33ّ ‫[ وا ِق ْشَع َ َر‬15] ‫ن‬
35 ⸂‫ت ⸃الألف‬ ّ َ ‫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 Kolkata has fa-ʾin.

24 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.
25 Kolkata clearly has the wrong form istaġnā, with an ending ʾalif maqṣūra and superscript
ʾalif.
26 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.
27 Kolkata has minhā, which is incorrect.
28 Kolkata has ḥaḏafta without the connecting wāw.
29 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.
30 Kolkata has ḥājizan.
31 A has yataḥarraku. B, L and Bakkāʾ have yataḥawwalu. M1 had yataḥarraku and M2 cor-
rected 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.
32 M1 had a different example, apparently with an ending ʾalif.
33 B is the only one to add wa-šmaʾazza after wa-qšaʿarra.
34 Kolkata clearly has wa-ʾin tašmayʾizz ʾašmayʾizz, with a yāʾ, which is probably only a
spelling issue of the hamza.
35 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’ read-
ing in M1, as was the case above in note 1? All the other versions collated here have this
al-ʾalif.
36 Kolkata has al-jarr, which makes no sense in the context here.
‫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
ّ ‫ لأ‬46‫[ لم تغي ّر‬1] [‫ظ‬62] ‫ف‬
‫ن‬ ٌ ‫ كان قبل الأّول أل‬45‫[ الألف مثله في غير الجزم وإن‬21] 44‫وثبات‬
ّ ‫ الحرف لأ‬47‫[ الوصل في ذا‬2] ‫ن المدغ َم ُ في َحتمل ذلك وتكون ألف‬
‫ن‬ ُ ‫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 ٌ ‫[ وإن كان قبل الأّول أل‬4] ‫وثبات الألف مثله في غير الجزم‬
ُ ‫ف ولم يكن في ذلك الحرف حر‬
[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 It is not clear who added this marginal gloss and who put it between brackets. If we con-
sider 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).
38 Kolkata has kāna l-ḥarf allaḏī, which only makes the expression clearer.
39 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 con-
sonant, 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.
40 Kolkata does not have al-ʾalif.
41 Kolkata has yataḥarrak.
42 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.
43 Kolkata has nanqadda.
44 According to Bakkāʾ, Mawṣil has banāt instead of ṯabāt.
45 A, B, L, Kolkata and Bakkāʾ have wa-ʾiḏā instead of this wa-ʾin. M1 has wa-ʾin and M2 did
not correct it.
46 Kolkata has yuġayyar.
47 Kolkata has hāḏā.
48 Kolkata has yataḥarraku.
49 Kolkata has wa-ḏālika qawluka.
50 Kolkata has tataġayyaru.
51 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 hap-
pens. So both forms can be regarded as equally possible here.
52 The ʾalif is unclear. To be checked directly on the manuscript.
53 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]
ِ ‫ يحر ّك الآ‬57‫ن منهم من جعل‬
‫خر كتحر يك ما قبله فإن‬ ّ ‫[ 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
ّ ‫ واجْت َرَ ّ وا ِْحمرَ َ ّ وضا َرّ لأ‬59ِ‫ض وف ِرِ ّ يا فتى و َأم ِّدِ واْطم َا ئ ِ ّ ِن وا ِْستعَ ِّد‬
‫ فتحة ً وألف ًا‬60‫[ قبلها‬12] ‫ن‬ ّ َ َ ‫[ وع‬11]
‫ إليك ولا‬63‫ضنا وم ُُّدني‬ ّ ‫ ور ُ ُدّنا ولا ي ُِشلـ ِّكم الل ّه ُ وع‬61‫فهـ]ـي[←⸃ـو⸂ أجدر أن يفُ ت َح‬
ّ َ َ ‫[ وع‬13] 62‫ضني‬
َ ‫ ل ِم‬66‫ت الخليل‬
ُ ‫[ وسأل‬15] ʘ ‫[ فتحوا أبد ًا‬14] ‫ت الهاء ُ والألف‬
ْ ‫ جاء‬65‫ضك ُْم فإذا‬
ّ َ َ َ‫ الل ّه ُ ول ِيع‬64‫ي ُِشل ِّك‬
ْ ‫ 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
‫ت‬ ّ ‫ذاك فقال لأ‬
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).
54 Without li-ʾannahu the text is less easy to read (less correct?) M1 is the only version not to
have it.
55 Kolkata has wa-huwa al-ʾawwal.
56 M1 had a different particle than all the other versions, with no implication for the meaning
or for the quality of the language.
57 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.
58 Bakkāʾ has an erroneous mā after maḍmūman.
59 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-ṭmāʾin-
ni 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 demonstra-
tion at this point, it is consistent with the teaching.
60 Kolkata has li-ʾanna mā qablahā.
61 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.
62 M1 had one more example, that M2 rejected by adding a superscript mark above it. This
additional example is found nowhere else.
63 Kolkata has wa-maddanī with a fatḥa on the šadda as if the verb was in the past tense,
not the imperative.
64 The lām and its vocalization is unclear. To be checked on the manuscript.
65 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.
66 According to Bakkāʾ, Mawṣil adds raḥimahu Allāh taʿālā. Kolkata adds rḥ.
67 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.
68 A is the only one to have the incorrect ruddan wa-ʾamiddan wa-ġullan with tanwīn.
69 L and Bakkāʾ are the only ones to add wa-ʾamiddahā after wa-ġullahā.
‫ت بالألف واللام‬ ّ ُ َ ‫ م ُُّده ُ وع‬70‫ضوا إذا قالوا‬

َ ‫ضه ُ فإن جئ‬ ّ ُ َ ‫[ م ُُّدوا وع‬17] ‫الهاء ُ مضمومة ً ضم ّوا كأّنهم قالوا‬
‫[ إذا كان‬19] ‫ن الفعل‬ َ ‫ كسر‬72‫ الخفيفة‬71‫[لف‬18]‫والأ‬
ّ ‫ت الأّول كل ّه لأن ّهكان في الأصل مجزوم ًا لأ‬
‫ل‬ ِ ِ ‫كس ِر َ وذلك قولك ا ِض ْر‬
َ ُ ‫[ ال َر ّج‬20] ‫ب‬ ُ ِ‫ لالتقاء الساكنين‬74⸉‫ ⸊صح⸉ ⸊فحرك‬73‫حر ّك‬
ُ ‫مجزوم ًا ]تـ[←⸃فـ⸂ـ‬
ّ ‫ لأ‬75‫[ ردّدتهَ إلى الأصل‬21] ‫ف الخفيفة‬
‫ن أصله أن‬ ُ ‫ف واللام والأل‬
ُ ‫ت الأل‬
ْ ‫ك فلماّ جاء‬ ِ ِ ‫وا ِض ْر‬
َ َ ‫ب اب ْن‬
‫ ذلك ج َر َى‬77‫ن نظا ئره من غير المضاع َف على غير‬
ّ ‫[ كما أ‬1] [‫و‬63] ‫ لغة أهل الحجاز‬76‫يكون مسّكناً على‬
ّ ‫[ 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
‫ن‬
‫[ 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 Kolkata has qāla.

71 A, B, L and Bakkāʾ have wa-bi-l-ʾalif. M1 has wa-l-ʾalif and M2 did not correct it. No inci-
dence on meaning.
72 Kolkata has bi-l-ʾalif wa-l-lām al-ḫafīfa which is a clear mistake.
73 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.
74 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.
75 A, B, L, Kolkata and Bakkāʾ have ʾaṣlihi. M1 has al-ʾaṣl and M2 did not correct it.
76 A and Bakkāʾ have fī. B, L and Kolkata have ʿalā. M1 has ʿalā and M2 did not correct it.
77 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 dif-
ference 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.
78 Kolkata has lam tabqa.
79 M1 had a different example, which has been erased by M2. All the versions collated here
have qāḍin.
80 M1 had a different wording. Al-Sīrāfī’s commentary (Šarḥ XIV, 62.9) has a combination of
the two readings: ʾiḏā ijtamaʿa sākināni.
81 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-l-
lām 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.
82 According to Bakkāʾ, Mawṣil adds raḥimahu Allāh taʿālā. Kolkata adds rḥ.
83 Kolkata has ʾiḏā.
148 druel
ِ ‫ الآ‬86‫[ ي ُت ْب ِعوا‬8] ‫ عربيتّ هُ ولم‬85‫ ممن ترُ تضى‬84‫[ بنو أسٍد وغير ُهم من بني تميم وسمعنا‬7] ‫الأّولون وهم‬
َ ‫خر‬
‫ ومنهم‬92‫[ قالوا اب ْن ُم ٌ واب ْن ٍِم واب ْنمَ ًا‬9] 91(90‫ كما‬89‫ل‬ ٍ ِ ‫ل كما )قالوا ا ِْمرُؤ ٌ وا ِْمر‬
ِ ‫ الآ‬88‫ و َأتبعوا‬87‫ئ‬
َ ‫خر َ الأّو‬ َ ‫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 ّ َ ‫[ ال‬11] ‫ض‬
َ ْ ‫طر‬ ّ َ ُ‫غ‬
‫[ مجراها‬12] ‫ ولـكن يجعلها في الفعل تجري‬98⸂‫ البت ّة ⸃من قال هلَ َ ُم ّا وهلَ ُم ّ ِي‬97‫مؤخروهلَ ُ َمّ مقدملا ي َكسر‬
ّ ‫[ ومن العرب من ي َكسر ذا أجمع على ك‬13] ʘ َ ‫في لغة أهل الحجاز بمنزلة ر ُو َي ْد‬
‫ل حال فيجعله بمنزلة‬
84 A, B, L, Kolkata and Bakkāʾ have samiʿnāhu. M1 has samiʿnā and M2 has not corrected it.
The form samiʿnāhu is probably better.
85 A, Kolkata and Bakkāʾ have turḍā. B and L have turtaḍā. M1 has turtaḍā and M2 has not cor-
rected it. Both forms have the same meaning and construction and it is not clear whether
one form is better than the other.
86 Kolkata has yutbiʿūhu, which makes no sense.
87 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).
88 A, B, L, Kolkata and Bakkāʾ have fa-ʾatbaʿū. M1 has wa-ʾatbaʿū and M2 has not corrected it.
89 The lām is unclear. To be checked directly on the manuscript.
90 A, B, L and Bakkāʾ have wa-kamā. Kolkata has only kamā.
91 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.
92 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.
93 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.
94 Kolkata has only one dot under the yāʾ, which is probably a typo, not to be read bi-jaʿlihi.
95 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.
96 M2 has completed the verse in the margin, but this addition is not found in the other
versions collated here.
97 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.
98 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-
ٌ ‫ بالألف واللام لأن ّه فع‬99‫ك وإن لم تجئ‬

ِ‫ل ح ُرّك لالتقاء الساكنين‬ ِ ِ ‫[ وا ِض ْر‬14] ‫ل‬
َ َ ‫ب اب ْن‬ ِ ِ ‫ا ِض ْر‬
َ ُ ‫ب ال َر ّج‬
‫ من‬101‫[ يا فتَ َى‬16] ّ‫ يقول هلَ ُ ِم‬100‫ل ولا يقولها في هلَ ُ َمّ ]و[لا‬
َ ُ ‫ب ال َر ّج‬
ِ ِ ‫ك واض ْر‬ ِ ِ ‫[ وكذلك ا ِض ْر‬15]
َ َ ‫ب اب ْن‬
َ ّ ‫[ 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
‫ف الفعل‬
‫ على أّنهم يقولون للنساء‬103‫[ وأهل الحجاز وغير ُهم مجتمِعون‬18] ʘ ٌ ّ‫ب وغ َن ِي‬
ٌ ْ‫كع‬
َ ‫ولم تقَ و قو ّته ومن ي َكسر‬
‫ف‬ ّ ‫ هاهنا لأمر ٍ ولا نهٍي وكذلك ك‬104⸂‫ن الدال لم ]××××[←⸃ت َسكن‬
ٍ ‫ل حر‬ ّ ‫[ وذلك لأ‬19] َ‫ا ُْرد ُْدن‬
‫[ لازم ٌ له‬1] [‫ظ‬63] ‫ن السكون‬
ّ ‫ ترى أ‬105‫يجزم ألا‬ ٍ ‫[ قبل نون النساء لا ي َسكن لأمر ٍ ولا لحر‬20]
َ ‫ف‬
‫[ )أخبرني‬2] َ‫ أن يرَ ْد ُْدن‬107ّ‫ وهّن يرَ ْد ُْدنَ وعلي‬106َ‫في حال النصب والرفع وذلك قولك ]×[←⸃ر َ⸂د َْدن‬
َ [3] ‫ وكذلك‬108(‫ت‬
‫يجري غير ُ المضاع َف قبل‬ َ ْ ‫أبو عبيدة أّنهم لا يقولون هلَ ُمّ ز َي ْد ًا إن ّما يقولون هلَ ُمّ َأن‬
َ ْ ‫[ و ي َض ْر ِ ب‬4] ‫ن‬
‫ن و ي َْذه َب ْنَ فلماّ كان هذا الحرف‬ َ ْ ‫ يح َر ّك في حال وذلك قولك ض َر َب‬109‫نون النساء لا‬
ّ ‫[ السكون حاجز ًا عنه ما سواه من الإعراب وتم‬5] ‫ل موضٍع وكان‬
‫كن فيه‬ ّ ‫يلزمه السكون في ك‬
‫ أو لحرف الجزم‬110ٍ ‫يجزم ]لـ[لأمر‬
ُ ‫[ من الفعل كرهوا أن يجعلوه بمنزلة ما‬6] ‫كن في غيره‬
ّ ‫ما لم يتم‬
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?
99 Kolkata has yajiʾ.
100 Unlike all the collated versions, M1 has wa-lā, which seems to be introducing a new argu-
ment. The suppression of this wāw makes the text smoother since what follows is not a
new argument but an example.
101 Kolkata has two dots below the yāʾ, which are probably a typo, not intended to be read
fatayya.
102 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.
103 B and L are the only ones to have mujmaʿūn. Kolkata is not vocalized.
104 M1 had a different word. All other versions have taskun. The original word may be readable
directly on the manuscript.
105 The ʾalif is unreadable. To be checked directly on the manuscript.
106 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.
107 Kolkata has ʿalā with an ending ʾalif maqṣūra and a superscript ʾalif, which is clearly a
typo.
108 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.
109 A and Bakkāʾ are the only ones to have wa-lā.
110 M1 has the definite li-l-ʾamr, which M2 corrected into li-ʾamrin. The difference is insignifi-
cant.
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 ّ ُ ‫ وم َد َْد‬113‫ت‬
ّ ‫ت لأ‬ ّ ُ ‫⸃ر َ⸂د َْد‬
‫سا‬ ّ ‫ أ‬116‫ وغيره‬115‫ وزعم الخليل‬ʘ ‫[ فتٍح‬10] ‫بمنزلته فيما فيه نون النساء يدل ّك على ذلك أن ّه في موضع‬
ً ‫ن نا‬
‫ 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 َ [12] ‫جميع المضاع َف‬
ُ ‫يجري كما ذكر‬
‫[ 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
ٍ ‫ب بن ُأمّ صاح‬
125⸂‫ب‬ ُ َ ‫ قعَ ْن‬124‫ قال الشاعر ⸃وهو‬ʘ ‫ على الأصل‬123‫وغيرهم على إدغامه أجر]×[وه‬
128‫ضننِ وُ ا‬
َ ‫ وإْن‬127‫جود ُ لَأق ْواٍم‬
ُ ‫ من خ ُل ُقِي َأن ِ ّي َأ‬126‫ت‬ َ ِ‫[ م َه ْل ًا َأعاذ‬18]
ِ ْ ‫ل قد ج َرّ ب‬
129‫[ ⸃وقال⸂ ومثله قول الراجز‬19]
111 B, L and Kolkata have fa-lam.

112 Kolkata has al-muḍāʿaf.
113 Again, M1 had a different example, probably wadadtu, which M2 corrected into radadtu,
which is the reading in all the other versions collated here.
114 B has yubnā.
115 According to Bakkāʾ, Mawṣil adds raḥimahu Allāh taʿālā.
116 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).
117 Kolkata vocalized ruddanā.
118 M1 probably had two examples here and M1 has erased the second one and replaced by a
large ending tāʾ of the first example.
119 Kolkata has the strange yajʿalūhu. We would have at least expected a marfūʿ form.
120 Kolkata has the obviously erroneous wa-lammā.
121 Kolkata is the only one to have the erroneous final nūn that was in M1 before M2 struck it
through.
122 M2 has erased the two ending letters. The original reading of M1 is not clear.
123 M2 has erased a letter. The original reading of M1 is not clear.
124 A, B, L and Bakkāʾ omit wa-huwa.
125 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.
126 Kolkata has an erroneous yāʾ instead of the bāʾ.
127 Kolkata has al-ʾaqwām.
128 Kolkata has ẓaninū.
129 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.
‫ل‬ ٍ َ ‫ الو َجَى م ِْن َأْظل‬130‫ي َْشكُو‬

ِ َ ‫ل و َأْظل‬
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]
ّ ‫ 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
ِ‫ن ما جاوز الا ِث ْن َي ْن‬
ُ ‫[ 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
ٍ ‫ 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
‫ل‬
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 َلا‬141‫[ له‬4] ‫ت فيها علامة ُ التأنيث وذلك قولك‬
ٍ ‫ث ب َنا‬ ْ ‫[ 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 ُّ ‫س‬
َ ‫ت ل ِب ْنٍ و‬ ِ ‫قو‬
ٍ ُ ‫س َأي ْن‬
ُ ْ ‫و َأْر ب َُع ن ِْسو َة ٍ وخَم‬
َ ‫[ كأن ّك قل‬7] َ ‫ت َأح َد َ ع َش َر‬
‫ت‬ َ ‫[ المذك ّر ُ الع َش َر َة َ فزاد عليها واحدًا قل‬6] ‫ فإذا جاوز‬143[‫تبلغ الع َش ْر َ]ة‬
130 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.
131 Kolkata has only hāḏā without the wāw.
132 Kolkata has taqaʿu.
133 Kolkata has li-l-ʿadad.
134 Kolkata has jāwazat.
135 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).
136 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 cor-
rect text. If M1 had noticed the error, why didn’t he erase the passage and rewrite over it?
137 Kolkata has ʾaw ḫamsatu.
138 M2 has added ḏā and a collation mark in the margin.
139 A, B, L, Bakkāʾ and Kolkata have wa-ʾin. M1 had fa-ʾiḏā which M2 corrected into fa-ʾin.
140 A, B and Bakkāʾ have the plural. L and Kolkata have a singular. M1 had a plural, which M2
corrected it into a singular.
141 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.
142 M1 is the only version that did not have hāḏā but only ḏā. M2 has added the hāʾ.
143 M1 had the form al-ʿašara which M2 corrected into al-ʿašr. It is consistent with the exam-
152 druel
ُ 146‫[ ]حرفـ[←⸃اسمـ⸂ان‬8] ‫ وهما‬145⸂‫ف‬

‫جعلا اسم ًا‬ ٌ ‫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
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.
144 Kolkata has the incorrect jamalan.
145 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 commen-
tary (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.
146 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.
147 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.
148 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’.
149 The addition of ʾaḥad by M2 makes the text easier to follow. All the other versions have it.
150 Kolkata has min.
151 Kolkata omits this ġayr, which is a clear misunderstanding of the text (or simply a
typo).
152 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.
153 M2 replaces the expression al-ʿidda lam tujāwiz by al-ʿadad lam yujāwiz, which is the read-
ing 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.
154 Kolkata has wa-ʾiḏā.
155 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.
156 Bakkāʾ has fa-zādū ʾaḥadan where the initial wāw and ʾalif in wāḥid were confused with
the plural mark in the verb.
‫ 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 ِ ‫ حرفا‬161‫ وهما‬160⸂َ َ‫ت ِإحْد َى ]××××[←⸃ت َم ْرة‬
ُ ‫ن‬ َ ‫ قل‬159⸂‫[ ِإحْد َى ع َش ْر َة َكأن ّـ]ك[←⸃ـما‬13]
َ ‫[ 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
‫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
َ ّ‫ إذا ثني‬168⸂‫ الا ِث ْن َي ْنِ عن حالـ]ـه؟[←⸃ـهما‬167‫[ لم تغي ِّروا‬18] َ ‫ ع َش َر‬166‫ له ا ِثنْ َْي‬165‫ن‬
َ ‫ت الواحد‬ ّ ‫ 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 The omission of the expression ka-ʾannamā qulta in M1 makes the text difficult to follow,
if not incorrect.
158 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.
159 M1 replaced ka-ʾannaka by ka-ʾannamā, which is the reading of all the other versions. Both
forms are found in our chapters.
160 Same case as presented in note 158. In his commentary, al-Sīrāfī (Šarḥ XIV, 119.5) has the
same example, tamra.
161 Kolkata has the erroneous form humāni.
162 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’.
163 A, B, L and Kolkata have wa-ʾin and Bakkāʾ has fa-ʾin. M1 has fa-ʾin, which M2 did not cor-
rect.
164 M2 has struck through these words, which are repeated in M1, with no apparent explana-
tion.
165 A, B, L, Bakkāʾ and Kolkata have wa-ʾinna. M1 has fa-ʾinna, which M2 did not correct.
166 A has the incorrect iṯnā.
167 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.
168 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.
169 Same correction as described above in notes 143, 147 and 155.
170 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.
171 A and Bakkāʾ have ḥarf ʾiʿrāb, without the article.
172 M2 has inserted a deletion mark above ḥāluhu. The text was actually easier to follow in
M1.
173 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 َ ّ‫ ثني‬180‫[ حيث‬1] [‫ظ‬71] ‫ عن حالهما‬179ِ‫الث ِّن ْت َي ْن‬
ّ ‫ت الواحدة َ إلّا أ‬
‫[ غير‬3] 181⸂‫ف ال ّذي بعد ِإحْد َى وث ِن ْت َي ْنِ ⸃على‬
ُ ‫سواء ٌ و بنُ ي الحر‬ ّ ‫[ ق‬2] ‫ن‬
َ ‫صة المذك ّر والمؤن ّث‬ ّ ‫الا ِث ْن َي ْنِ لأ‬
‫ل فإذا انتقل عن‬ ُ ‫[ اللف‬4] ‫ وقد يكون‬182‫بنائه والعدد ُ لم يجاوز الع َش ْر َ كما فعُ ل ذلك بالمذك ّر‬
ٍ ‫ظ له بناء ٌ في حا‬
َ ‫[ ز َب ِينة‬6] ‫ق َأف َقِ ّ ٌي وفي‬
ِ ُ ‫[ ذلك تغيير ُهم الاسم في الإضافة قالوا في الُأف‬5] ‫ فمن‬ʘ ‫تلك الحال تغي ّر بناؤه‬
ّ ‫ 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
‫ن‬
‫[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
ُ ‫[ واحدًا فوق ث ِن ْت َْي ع َش ِرة َ فالحر‬11] ُ ‫ما بين هذا العدد إلى ت ِْسعة َ ع َش َر َ وإذا زاد العدد‬
‫ف الأّول بمنزلته‬
174 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 construc-
tions, so that one cannot be said to be preferable to the other.
175 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.
176 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.
177 M2 had a missing ṯāʾ in the word iṯnatay.
178 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?
179 Kolkata has iṯnatayni.
180 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.
181 The omission of ʿalā in M1 is clearly a mistake. The sentence makes no sense without it.
All the other versions have it.
182 Kolkata has the obvious typo al-ḏikr.
183 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.
184 Kolkata has the erroneous ʾarāda.
185 Kolkata has al-ḥarf allaḏī al-ʾawwal.
186 A has lam yujāwiz. And Kolkata has lam yatajāwaz.
187 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.
188 Same correction as above, see note 135.
189 M1 did not have ʿabdan, which is found in all the other versions. Without it, the example
is still understandable, although less clear.
‫[ 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
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 Sīb-
awayhi, XIV. Ed. by Hudā Qirāʿa. Cairo: Dār al-Kutub wa-l-Waṯāʾiq al-Qawmiyya,
2010.
190 A has lam yujāwiz. Kolkata has lam yatajāwaz.

191 A has bayna ʾiḥdā wa-ṯalāṯīna, which makes no sense, instead of baʿda ʾiḥdā wa-ṯintayni.
192 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.
193 M1 did not supply the Ḥijāz form, which is found in all the other versions and added by
M2.
194 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.
195 M1 did not have this mā, the addition of which is maybe slightly better grammatically. All
the other versions have it.
196 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 druel
B Secondary Sources
Druel, Jean N. 2012. Numerals in Arabic grammatical theory: An impossible quest for con-
sistency? 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 gram-
mairien andalou du XIIe siècle”. Le manuscrit arabe et la codicologie, ed. by Ahmed-
Chouqi 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 parti-
cle 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 rea-
son 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 parti-
cles 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 al-

šarṭ) kullamā, although the principles of syntax would exclude that, inas-
much 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 par-
ticle 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.

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 apoco-
pate. 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 distinc-
tion 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 repre-
sented 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 Sīb-
awayhi 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 com-
mentary 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 com-
pletely 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 bi-
hā (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 par-
allel.
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 observa-
tions 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 condition-
als 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 con-
dition’, 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 Sīb-
awayhi 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 cat-
egory 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 mak-
ing 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
wa-fī wa-law ḥimārun li-ʾannaka law lam taḥmilha ʿalā ʾiḍmāri yakūnu fa-
fiʿlu 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’, operat-
ing ‘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 l-
ḫabari 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 ʾakram-
nāhu’ 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
We note that in both these passages ʾin and law are treated as having paral-
lel properties. Moreover, the properties are partly syntactic and partly seman-
tic. 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 spo-
ken 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 expla-
nation (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 occur-
rence.
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 wa-
naḥwahā 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 Ara-
bic. 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)
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 apoco-
pated 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 logi-
cal 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 dis-
cussion 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 occur-
rences 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 al-
naẓari 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.
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 bi-
ttibāʿin 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 wa-
law.)
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 sin-
gle 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, fa-
ʾinna 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 counter-
factual 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?].
3 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 fea-
tures 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 promi-
nent 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 his-
torical 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.), Christia-
niae (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-
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 nom-
inal 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 Peri-
patetic 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, where-
as in Ibn Sīnā the words ‘always’, ‘sometimes’, ‘never’, ‘sometimes not’ that stand at the begin-
ning 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 Quan-
tifizierung 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).
vances. The reason is that its author does not understand how condi-
tionals are affirmative or negative, and how they are universally or exis-
tentially 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 muṣan-
nifahu lam yaʿrif al-ʾījāba wa-l-salba fī l-šarṭiyyāti, wa-l-kulliyyata wa-l-
juzʾiyyata 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 osten-
sibly over times, but Ibn Sīnā’s examples make clear that he includes situations
in some sense ‘unquantified’ (muhmal to the logicians). The sentences in ques-
tion 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; fa-
l-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 com-
plicated: 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 substan-
tially 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.
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 fea-
tures 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 ‘con-
ditional’) consists of a statement (the consequent) which is made within the
scope of a quantifier over a set of times, situations, events or something simi-
lar. 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 some-

thing 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 parti-
cle 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 ana-
lyze 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.
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 sen-
tence, 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) fī-
mā 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) yata-
kallamu 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 quanti-
fier of the conditional, he points out the practical consequences of an ambi-
guity 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 sim-
ply 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).
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 speci-
fied, 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 situa-
tion 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-li-
kulli 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 ka-
qawlika: allaḏī ʾatānī fa-lahu dirhamun, yuṯbitu ʾanna l-dirhama staḥaq-
qahu […]).
Šarḥ III, 309.23–310.7
This ambiguity was known also to the logicians, at least in general terms. Al-
Fārābī (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-
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 quan-
tifier 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 par-
ticle, for the following reason:
And I asked him about ʾiḏā, and what prevents it from being used as con-
ditional. 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 fī-
mā 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-ka-
ḏālika 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
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 con-
tenant 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 state-
ment 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 implica-
tion.
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 proposi-
tion [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 mak-
ing 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
šarṭiyyatan wa-ka-ʾannahum yaʿnūna bi-l-šarṭiyyati mā yulḥaqu fīhi bi-

qaḍiyyatin 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 fa-
qulta ‘ʾin kānat al-šamsu ṭāliʿatan’ fa-ḥarrafta l-qaḍiyyata fa-ṣārat ġayra
qaḍiyyatin ḥīna zāla ʿanhā ʾan takūna ṣādiqatan ʾaw kāḏibatan? wa-ka-
ḏālika kāna qawluka ‘al-nahāru mawjūdun’ qawlan ṣādiqan ʾaw kāḏiban fa-
lammā 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 for-
mal 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’.
If the common formal structure between the two logics carries over to a com-
mon 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 state-
ment 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 anal-
ysis 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:
[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 muḥtā-
jin 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-
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
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 prac-
tical 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 concrete-
ness, 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 fa-
ʾinnahumā ʾiḏā ʾuḫrijā fi ʾiḥdā l-jihatayni ltaqiyā fa-nuḫrijuhumā fī jihati
bd fa-yaltaqiyāni ʿalā nuqṭati k ʾin ʾamkana dālika fa-taṣīru zāwiyatu aḥṭ al-
ḫārijatu 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 cer-
tain 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 state-
ments where the antecedent is implicit, having been stated in an earlier sen-
tence. Unlike in Ibn Sīnā’s examples, her examples introduce a new quantifica-
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 grammar-
ians 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 Mean-
ings [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) sup-
pressed (wa-mā ḏakarahu l-Ḫalīl min ḥaḏfi l-jawābi fī qawli llāh […] wa-
qad 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 con-
sequent does not depend on them (at least semantically). The double-act of
Sībawayhi and al-Ḫalīl is extraordinarily effective in finding interesting exam-
ples, 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-
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 impli-
cation.
In spite of choosing these interesting examples, al-Ḫalīl as reported by Sīb-
awayhi 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 noth-
ing 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 exam-
ples, 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-l-
ḥimāru 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 sec-
ond 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).
of correspondence’ ( yuslabu l-muwāfaqatu). Both the affirmative and the neg-

ative 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 cor-
respondence, 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 al-
Sīrāfī have the better examples but Ibn Sīnā has the better theory. If only they
could have pooled their efforts …
A Primary Sources
Euclid, Elementa = Codex Leidensis 339, 1: Euclidis Elementa ex interpretatione Al-
Hadschdschadschii 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 audi-
ence 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.
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-mašriqiy-
yīn. 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.
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-mufaṣ-
ṣal 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 clas-
sique (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 Ara-
bic 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 al-
ʾAstarābāḏī 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 syntax-
ique 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 conse-
quences: 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.
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 McCul-
lough, 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”. Ker-
van. 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 Mas-
sachusetts, 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 lexicog-
raphy. 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 formal-
ize rules and to describe Persian characteristics by means of Arabic terms and
analytical devices. One striking characteristic of Persian is that it readily gen-
erates 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 “pho-
netic 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 diph-
thongs 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).

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 clas-
sify compounds in Persian.
These two terms taqdīr and taḫfīf are both of Arabic origin; yet, their mean-
ing 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 time-
limits 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 Farhangnāma-
yi fārsī by Ṣadrī Afšār, which contains a large number of Classical lexica, does
not acknowledge the grammatical meaning of the word taqdīr ‘fate, appreci-
ation’ (III, 843). On the other hand, the traditionally organized Luġatnāma,
compiled by ʿAlī Akbar Dihḫudā in the 1930s, deals with its use in grammati-
cal 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 (Naḥ-
vīān 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 tan-
vīn 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 vocabu-
lary 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 Šams-
i 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 Per-
sian 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 cen-
tury, 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 connec-
tion with phonological and morphological contexts. The first case refers to the
obligatory insertion of a prosodic vowel called majhūla, muḫtalasa, or nīm-
fatḥa (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 inter-
pretation of the long vowel as a sequence of “a vowel followed by a homorganic
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 scan-
ning the verse. Actually, Šams-i Qays offers two solutions to the pronunciation
of words with two or three consecutive consonants either in middle or word-
final 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 Ara-
bic 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 phonolog-
ical 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 met-
rical 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 pho-
neme /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 sec-
tion on rhyme-science. Discussing rhyme-pairs such as hami na-gardānad/na-
gardānad/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 con-
tinuous 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 pre-
fix 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 See more examples of this dichotomy in Jeremiás (2016:65).

7 See Algar (2003).
8 Full titles of Sūdī, Gulistān, Būstān, Ḥāfiẓ are given in the bibliographical references.
9 It is thought that he may have translated Ibn al-Ḥājib’s Kāfiya into Turkish.
10 Cf. Versteegh (1993:99).
yaʿnī … dar aṣl (Sūdī, Ḥāfiẓ I, 50), which appear to signify “the speaker’s inten-
tion” (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 rep-
resent 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 fre-
quency. 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 classi-
cal 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 rep-
utation’: 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 voca-
tive 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 peo-
ple [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 underly-
ing 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 < guf-
tan ‘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 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’.
13 This form is used only in compounds.
14 In the English translation, the difference between gū and gūyanda disappears.
15 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).
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 qabīl-
i 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 “ ‘excuse-

accepting’: means ʿuẕr [in Arabic] and its underlying meaning is ‘accep-
tance 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’: recon-

structed at the underlying level as ‘the person who is bitten’” (Sūdī, Gulis-
tān 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 (sur-
face 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 recon-
structed 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 recon-
structed 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 inde-
terminacy 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 sub-
stituted 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’: vir-
tually 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 vir-
tual meaning of the phrase is ‘to consume, to lose’ ” (Sūdī, Ḥāfiẓ I, 75)
(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 basi-
cally 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 yād-
aš 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 com-
pound 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 Per-
sians 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 partici-
ples and derived (or secondary) adjectives (used occasionally as nouns),20 and
also derived nouns. Then, he gives the following classification, dividing com-
pounds 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 Ch. XIV: dar bayān-i vaṣf-i tarkībī ‘On the description of the compound’ (ff. 56r–57r). See
Jeremiás (2012: 112).
18 See more examples in Jeremiás (forthcoming).
19 Cf. Afšār (1388/2009), II:1827; Carter (1981:483): ṣifa mušabbaha bi-l-fāʿil ‘quasi-participial
adjective’.
20 If I understand this passage properly, the last two categories might be conceived as sub-
categories of the ism interpreted in terms of tašbīh ‘formal similarity’ (cf. Bohas et al.
1990:51–53).
21 See above, note 3.
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’ (taḫ-
fīf ), 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 taḫ-
fīfī). 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 basi-
cally 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 taḫ-
fīf 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 suḫan-
dān ‘eloquence-knowing’ is dān-i suḫan ‘knowledge of eloquence’; and because
22 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 gram-
marians are concerned with” (Versteegh 1997:59).
23 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 taqdīm-
i 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 adjec-
tive’ (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 tark-
ībī mī-bāšad va manẓūr-i taḫfīf tanhā buvad hamču gul-āb “The word saʿādat-
ḫāna ‘happiness-mansion’, which basically is ḫāna-i saʿādat ‘mansion of hap-
piness’ 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 ‘light-
ening’, 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 ‘black-
eyed’ and tar-dāman ‘of wet-rock, immortal, polluted’, which creates the mean-
ing ism-i mansūb by means of taqdīm, is produced only for the sake of ‘lighten-
ing’ (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 com-
mon in Sūdī’s commentaries, for instance, dīgar > digar, afġān > faġān, čūn >
ču, agar > ar, etc.
24 Cf. above, notes 13 and 14.

25 This passage indicates that the terms aṣl and taqdīr might have different meanings.
6 Summary
The eclectic ways of using taqdīr or taḫfīf do not point to a definite interpreta-
tion of the original theory, or to any interpretation which might have dealt with
these concepts systematically. Nevertheless, from the beginning, the Arabic lin-
guistic 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 grammatical-
ity 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. Ten-
tatively, 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 pro-
cess 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 lan-
guage. 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 contribu-
tion, one which reveals his proficiency in Classical doctrine.
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 Muḥam-
mad 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.
Bohas, Georges, Jean-Patrick Guillaume, and Djamel Eddine Kouloughli. 1990. The Ara-
bic linguistic tradition. London and New York: Routledge.
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 Mod-
ernism, 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 indige-
nous 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.
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 tra-
dition”. Historiographia Linguistica 21.271–296.
Versteegh, Kees. 1997. The Arabic linguistic tradition. London and New York: Routledge.
Versteegh, Kees. 2009. “Taqdīr”. Encyclopedia of Arabic language and linguistics, ed. by
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 al-
muṭlaq was examined. Levin’s article (1991), entitled “What is meant by al-
mafʿūl al-muṭlaq?”, is dedicated in its entirety to this issue, whereas Larcher
(1991:153 [= 2014:292f.]) discusses it succinctly in a single paragraph. Their find-
ings are virtually the same:1 The element mafʿūl in the phrase al-mafʿūl al-
muṭlaq is a personal passive participle. It means ‘that which is done’, designat-
ing 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. al-
mafʿūl fīhi, lit. ‘that in which the action is done’ (technically: the locative/tem-
poral 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.

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 + gen-
itive.
This was indeed the most common interpretation of the term al-mafʿūl al-
muṭlaq 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 lin-
guistics 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 sen-
tence”.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 l-
ʾarḍi 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 al-
mafʿūl 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 Ara-
bic 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 preposi-
tion, 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 com-
mon 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 schol-
ars. 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.
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 tak-
ing 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 l-
ʾanāsiyya ‘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 some-
thing 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 cate-
gory 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 heav-
ens …’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, al-
samāwāt 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 cre-
ated 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 Ibn Hišām, Šarḥ II, 158f.

10 Jurjānī, ʾAsrār 340f.
11 Ibn Hišām, Muġnī VI, 541ff.
12 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.
14 Ibn Hišām, Muġnī VI, 578f.
15 Ibn Hišām, Muġnī VI, 579.
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 above-
mentioned verse.17
Ibn Hišām also explains why “most grammarians” parsed such effective
objects as al-mafʿūl bihi: people can only produce actions (ʾafʿāl), not sub-
stances (ḏawāt), while God produces both. And since the grammarians illus-
trated 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 al-
samāwāt, 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 al-
muṭlaq 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, mak-
ing, 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 Suyūṭī, ʾAšbāh VII, 140.

18 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.
19 Cf. Kasher (2012a).
20 Ibn Hišām, Muġnī VI, 579.
table 1 The parsing of effective objects
ḍaraba zaydun ʿamran ḍarabtuhu ḍarban ḫalaqa llāhu l-samāwāti
effective – + +
maṣdar – + –
Constituents parsed as al-mafʿūl al-muṭlaq are normally [+maṣdar] and

[+effective], whereas those parsed as al-mafʿūl bihi are prototypically [-maṣ-
dar] 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 fea-
ture 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 suffi-
cient 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 observa-
tions. 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 for-
eign 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 grammar-
ians 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?) asser-
tion 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’ recog-
nition 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 Sībawayhi, Kitāb I, 15: li-ʾanna l-ṯawba laysa bi-ḥālin waqaʿa fīhā l-fiʿlu.
22 See e.g. Carter (1981:8f.): maʿnā fī l-luġa / maʿnāhu luġatan vs. maʿnā fī l-iṣṭilāḥ / maʿnāhu
iṣṭilāḥan. See also Kasher (2012b:157).
23 Ibn al-ʾAnbārī, ʾAsrār 88f. See also Peled (1994:141, 1999:68); Guillaume (1998:56 f.).
24 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.
25 Ibn Ḫālawayhi, ʾIʿrāb 70. See also Peled (1994:141, 1999:54 f.).
26 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).
27 The term tabyīn lit. ‘clarification’, is synonymous with tamyīz, in the technical sense (see
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 function-
ing 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 com-
paring it with the tamyīz. According to al-Fārisī, this analogy explains the indef-
initeness 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 lit-
eral 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 grammar-
ians, who apply the term al-mafūl bihi also to effective objects, notwithstand-
ing 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 gram-
marians 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 lit-
eral 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
e.g. Carter (1981:380)). This term would seem to constitute a better illustration of al-
Jurjānī’s argument.
28 Jurjānī, ʾAsrār 369f.
29 Jurjānī, Muqtaṣid I, 675. See also Ibn al-ʾAnbārī, ʾAsrār 199.
30 Jurjānī, Muqtaṣid I, 675f. Ibn al-Warrāq (ʿIlal 371) explains the indefiniteness of the ḥāl on
the ground of this very analogy.
31 See Kasher (2012a), and the references therein.
206 kasher
the terminological level, as well as to the term’s history: as is well known, gram-
marians 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 Further-
more, 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, al-

samāwāt is excluded from being parsed as al-mafʿūl al-muṭlaq by the majority
of the grammarians. Ibn Hišām, however, also mentions their positive argu-
ment for parsing it as al-mafʿūl bihi, namely the possibility to apply to it the pas-
sive participle of its verb, maḫlūq. This property cuts across the effective/non-
effective 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 Carter (1981:324ff.).
33 Carter (1981:344). See also Ibn Hišām, Šarḥ II, 158f. (see above). See Peled (1999:72, 85, n. 18).
34 As is well known, the syntactic position of al-mafʿūl al-muṭlaq is not always filled by a maṣ-
dar. 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-l-
sawṭi. 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).
35 As noted in Levin (1991:917).
A note about Ibn Hišām’s line of argumentation is in order here. His counter-
argument 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 counter-
argument 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 pas-
sive 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 fol-
lowing 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 expres-
sions 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 explana-
tion 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 alter-
native 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 dis-
agreement 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 ʿUkbarī, Lubāb I, 261f.

37 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 …).
38 Ibn al-Warrāq, ʿIlal 320.
39 The expression used is yaṣilu ʾilayhi. On this notion, see Kasher (2013).
40 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).
41 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).
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 al-
Bayṭār. 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 Ḫāl-
awayhi, 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 l-
labīb 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 al-
ʾAnṣārī, 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 al-
balāġa. Ed. by Hellmut Ritter. Istanbul: Government Press, 1954.
Jurjānī, Muqtaṣid = ʾAbū Bakr ʿAbd al-Qāhir ibn ʿAbd al-Raḥmān al-Jurjānī, Kitāb al-
muqtaṣid fī šarḥ al-ʾĪḍāḥ. Ed. by Kāẓim Baḥr al-Marjān. 2 vols. [Baghdad]: Dār al-
Rašīd, 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, al-
ʾAšbāh 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 Ara-
bic 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. 1994. “Writing the history of Arabic grammar”. Historiographia Lin-
guistica 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 lit-
erary theory in Islam: The case of majāz”. Zeitschrift für Geschichte der arabisch-
islamischen 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. Mar-
ogy, 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”.
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 dis-
course. 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 Univer-
sity Press.
Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press.
ization. Amsterdam: J. Benjamins.
Peled, Yishai. 1994. “Aspects of case assignment in Medieval Arabic grammatical the-
ory”. Wiener Zeitschrift für die Kunde des Morgenlandes 84.133–158.
Peled, Yishai. 1999. “Aspects of the use of grammatical terminology in Medieval Ara-
bic 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 Confer-
ence 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 phe-
nomenon 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 pri-
mary 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).

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 pro-
ceedings.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, pat-
tern, 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 exten-
sion 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 Hanadi Dayyeh’s (2015) contribution to the proceedings of FAL 2 is a good article. Irre-
spective 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 al-
kalām.
7 See e.g. Mubarrad, Muqtaḍab IV, 330.7–9.
8 See Baalbaki (1988:163.26–28).
9 See Baalbaki (1988:129.28f.).
10 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 par-
ticiple (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 pro-
nounced 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 utter-
ance is illogical and even designates an absurdity, the speaker feels and knows
that it does not express the sense intended by him. Hence, when pronounc-
ing 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 character-
ized 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.
them’ (Sībawayhi, Kitāb I, 89.4f.), i.e., the travelers become the guests of such-
and-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 ordi-
nary 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 al-
muḫāṭab 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 phe-
nomena 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 Sīb-
awayhi’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 ordi-
nary 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 nomina-
tive 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 construc-
tion 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 express-
ing 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 des-
ignate 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 exam-
ple, 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 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”.
14 See Sībawayhi, Kitāb II, 137.8–15; cf. Jurjānī, Muqtaṣid I, 333.7–10. See Levin (1997: 154–157,
§4.3).
15 This point is inferred from Sībawayhi, Kitāb I, 372.17–373.17.
16 For this definition see Levin (1997: 21, 151–157, §4).
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 con-
struction 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 character-
ized by phenomena of ittisāʿ al-kalām is grammatical correctness. Hence,
the lafẓ of these utterances does not need any corresponding taqdīr con-
struction.
(ii) The title of chapter 42 of the Kitāb17 is evidence that the syntactic con-
struction 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 utter-
ances 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 cer-
tain 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 explana-
tion 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 Chapter 42, Sībawayhi, Kitāb I, 88.9–90.3. This title has been quoted, translated, and dis-
cussed in 3.2 above.
18 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 denot-
ing 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 l-
saʿa, 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 l-
laylati 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 func-
tion is that of a direct object, as in muʿṭī zaydin, e.g. hāḏā muʿṭī zaydin
19 See Ibn al-Sarrāj, ʾUṣūl II, 255.11.

20 See al-Mubarrad, Muqtaḍab IV, 331.9–11.
21 See Levin (1997:156.3–19).
22 See Levin (1997:156.20–157.7).
23 These adverbs of time and place belong to the category called al-ẓurūf al-mutamakkina.
24 The maʿnā of this verse is ‘O you who stole in the night from the people of the dwellings
of the tribe’.
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 can-
not occur in this position, the grammarians hold that the Bedouin speak-
ers 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 l-
laylata 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. Sīb-
awayhi 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 al-
mafʿūl. 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 pred-
icate 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 exam-
25 See Sībawayhi, Kitāb I, 75.1–9.

26 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 l-
saʿa (ʾUṣūl I, 195.22–196.3). The term mafʿūl ʿalā l-saʿa frequently occurs in Ibn al-Sarrāj, as
shown in this article.
27 Sībawayhi, Kitāb I, 75.1–9.
29 Sībawayhi, Kitāb I, 75.f.
30 Sībawayhi, Kitāb I, 75.12f.
31 See Levin (1979:199.18–200.2).
32 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 any-
more, 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 l-
dāri 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 bi-
zaydin 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 suf-
fix:
a. “As for [an example of a bound pronoun suffix occurring as a]
[mafʿūl] ʿalā l-saʿa [there] is the utterance yawmu l-jumʿati ḍarab-
tuhu 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 qum-
tuhu 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 fa-
taqūlu yawmu l-jumʿati qumtuhu fī mawḍiʿ qumtu fīhi, wa-l-farsaḫu
sirtuhu wa-makānukum jalastuhu wa-ʾinnamā hāḏā ttisāʿun; Mubar-
rad, Muqtaḍab IV, 330.7–9)
33 Ibn al-Sarrāj, ʾUṣūl II, 294.9–14.

34 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’.
c. yawmu l-jumʿati sirtuhu ‘Friday, I walked in it’ (Sībawayhi, Kitāb I,

33.14). The maʿnā is yawmu l-jumʿati sirtu fīhi.
d. yawmu l-jumʿati ṣumtuhu ‘Friday, I fasted in it’ (Sībawayhi, Kitāb I,
33.14). The maʿnā is yawmu l-jumʿati ṣumtu fīhi.
e. 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’.
f. 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 muta-
makkina, 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 sa-
ʾaḏhabu 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 follow-
ing 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 gar-
ment 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 See Sībawayhi, Kitāb I, 75.20.

36 ʾamsi is virtually in the accusative.
38 See Sībawayhi, Kitāb I, 11.15f.
39 It is impossible to conceive of al-laylati as a ẓarf, because a ẓarf cannot occur in the posi-
tion 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 char-
acteristics 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 sup-
porting 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 wa-
layluhu 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 Sībawayhi, Kitāb I, 14.23–15.1; 15.12–15.

42 See Ibn al-Sarrāj, ʾUṣūl I, 171.4–7, 18–22.
43 It should be emphasized that Sībawayhi does not accept al-Ḫalīl’s explanation of the form
ʾamsi (Kitāb I, 254, 1f.).
44 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.
45 For this sense of qāma l-laylu see above, n. 32.
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’.
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-l-
Kūfiyyīn. Ed. by Gotthold Weil. Leiden: E.J. Brill, 1913.
Ibn Ḫaššāb, Murtajil = ʾAbū Muḥammad ʿAbdallāh ibn ʾAḥmad Ibn al-Ḫaššāb, al-Murta-
jil. 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 al-
Risāla, 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 al-
muqtaṣid fī šarḥ al-ʾĪḍāḥ. Ed. by Kāẓim Baḥr al-Murjān. 2 vols. Baghdad: Wizārat
al-Ṯaqāfa wa-l-ʾIʿlām, 1982.
muqtaḍab. Ed. by Muḥammad ʿAbd al-Ḫāliq ʿUḍayma. 4 vols. Cairo: Dār al-Taḥrīr,
1385–1388/1965–1968.
Derenbourg, Le livre de Sîbawaihi: Traité de grammaire arabe. 2 vols. Paris: Imprime-
rie 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
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: Inter-
pretation 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 Ori-
entalia Memoriae D.H. Baneth Dedicata, 193–210. Jerusalem: The Magnes Press.
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 Sīb-
awayhi’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 rāk-
iban ‘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 On this phenomenon see Bernards (2007) and Carter (2002).

4 All examples and translations in this paragraph are taken from Wright (1997:II, 112–115, § 44
c).

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 mušāfa-
hatan 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 imper-
missible examples] *ʾatānā surʿatan ‘he came to us quickly’ and *ʾatānā
rujlatan ‘he came to us on foot’ are not appropriate (wa-laysa kullu maṣ-
darin 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 exam-
ples, 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.
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 ‘com-
ing’, which the main verb jiʾtu conveys.9 Al-Mubarrad provides two theoretical
explanations for this distinction. His first explanation is identical to that of Sīb-
awayhi, 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 samāʿiy-
ya 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 exist-
ing 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 See Mubarrad, Muqtaḍab III, 234.5.

9 See Mubarrad, Muqtaḍab III, 234.7.
10 On this phenomenon see Levin (1991), and above, Kasher’s contribution to the present
volume.
11 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), al-
Šantamarī (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-ʾAstarā-
bāḏī (d. 684/1285 or 686/1288).
12 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 al-
Jurjānī. 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 rea-
son 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 al-
Mubarrad, 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
(d. between 337–340/948–950), al-Rummānī (d. 384/994), Ibn Jinnī (d. 392/1002), al-
Harawī (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), al-
Baṭalyawsī (d. 521/1127), al-Šarīf al-Kūfī (d. 539/1145), Ibn al-Ḫaššāb (d. 567/1172), Ibn al-
ʾAnbārī (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).
13 See Sīrāfī, Šarḥ II, 257, last two lines. Cf. Šantamarī, Nukat I, 531, last two lines.
14 See Sīrāfī, Šarḥ II, 258.2–7. Cf. Šantamarī, Nukat I, 532.
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 ver-
bal noun is mafʿūl muṭlaq.
Ibn Yaʿīš supports Sībawayhi’s opinion and rejects the views of both al-
Mubarrad 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 al-
Masāʾil 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 sup-
port 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 opin-
ions 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 func-
tion 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 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).
16 See Ibn Yaʿīš, Šarḥ II, 12f.
17 See Ibn Yaʿīš, Šarḥ II, 12.6–4 from the end.
18 See Fārisī, Manṯūra, 16 (al-masʾala 8).
19 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).
20 See, for example, Yamanī, Kašf I, 480.7f.; 483.7f.
21 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. al-
Mubarrad’s view in favor of qiyās] is not inconceivable” (wa-laysa ḏālika bi-
baʿīdin).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 al-
Kāfiya.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).
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 l-
ʿarabiyya. 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 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.
23 See Ibn Mālik, Šarḥ al-Kāfiya I, 233.11. For the whole discussion, see ibid. I, 233.6–11.
24 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.
25 See ʾAstarābāḏī, Šarḥ, 671f.
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 Muḥam-
mad 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 al-
ʿIlmiyya, 1422/2001.
Mubarrad, Muqtaḍab = ʾAbū l-ʿAbbās Muḥammad ibn Yazīd ibn ʿAbd al-ʾAkbar al-
Ṯumālī al-ʾAzdī al-Mubarrad, al-Muqtaḍab. Ed. by Muḥammad ʿAbd al-Ḫāliq ʿUḍay-
ma. 4 vols. Cairo: Muʾassasat Dār al-Taḥrīr, 1385–1388/1965–1968.
Šalawbīnī, Tawṭiʾa = ʾAbū ʿAlī ʿUmar ibn Muḥammad ibn ʿUmar ibn ʿAbdallāh al-Šalaw-
bīnī al-ʾIšbīlī, al-Tawṭiʾa. Ed. by Yūsuf ʾAḥmad al-Muṭawwiʿ. Cairo: Dār al-Turāṯ al-
ʿArabī, 1401/1981.
Šantamarī, Nukat = ʾAbū l-Ḥajjāj Yūsuf ibn Sulaymān al-ʾAndalusī al-Naḥwī al-Šanta-
marī, al-Nukat fī tafsīr Kitāb Sībawayhi. Ed. by Rašīd Balḥabīb. 3 vols. Al-Maḥmadiy-
ya: 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.
Derenbourg, Le livre de Sībawaihi, traité de grammaire arabe. 2 vols. Paris: Imprime-
rie 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: Maṭ-
baʿat 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-mufaḍ-
ḍal 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
II, 224–228. Leiden: E.J. Brill.
Carter, Michael G. 2002. “Patterns of reasoning: Sībawayhi’s analysis of the ḥāl”. Pro-
ceedings 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 tra-
ditional grammar include pronouns, proper nouns, demonstratives, relatives,
nouns with al-, and nouns followed by another definite. But, in this classifica-
tion, 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 nowa-
days 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 view-
point.
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 Stud-
ies in Japan (Nippon Oriento Gakkai) in Tokyo. However, any and all possible mistakes are
mine.

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 refer-
ent.
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 Ameri-
can 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 with-
out referring to the context in which they are uttered. Pronouns and demon-
stratives 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 ana-
phoric 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 pro-
noun 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 ele-
ments)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 sakaedani
table 1 ‘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 ḏālika tilka ḏānika tānika ʾulāʾika
table 2 ‘Ambiguous nouns’ (demonstratives and pronouns) in Sībawayhi
masculine masculine dual feminine feminine dual plural

singular nominative singular nominative
close objects hāḏā hāḏāni hāḏihi hātāni hāʾulāʾi
tilka
far objects ḏālika ḏānika tānika ʾulāʾika
tīka
masculine feminine dual masculine feminine

singular singular plural plural
huwa hiya 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 demon-
stratives 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 ambigu-
ous 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).
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 (dual) allatayyāni
We have seen above that Sībawayhi does not include relative pronouns into
the category of definite noun phrases, but he includes them with the demon-
stratives 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 rel-
atives,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 ‘Def-
inite 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

ii. nouns prefixed with al-
iii. 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). How-
ever, in the chapter ‘Diminutives of ambiguous nouns’ (Taḥqīr al-ʾasmāʾ al-
mubhama, 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 conso-
nants 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. Like-
wise, 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 diminu-
tives 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 ter-
minology of the Kufan school
ii. al-mubham, ambiguous nouns, meaning demonstratives6
iii. (individual) proper nouns
iv. nouns with the prefix al-
v. 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.
table 4 Diminutives of relatives
ambiguous nouns diminutives ambiguous nouns diminutives
ḏā ḏayyā ḏāka ḏayyāka
ḏihi (= tā) tayyā ḏī (= tī) tayyā
ḏālika ḏayyālika
ʾulāʾika ʾulayyālika ʾulā ʾulayyā
hāʾulāʾi (long) hāʾulayyālika hāʾulāʾi (short) hāʾulayyā
allaḏī allaḏayyā allatī allatayyā
allātī allatayyātu,a
allawayyāb
allaḏāni allaḏayyāni
allaḏīna allaḏīyīna allaḏūna allaḏīyūna
a Sībawayhi’s view.
b al- ʾAḫfaš’s view.
by definite noun phrases. They include: miṯl ‘something like …’, šibh ‘some-
thing 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 indefi-
nite 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 demon-
stratives. He does not refer to relatives in his five categories of definite noun
phrases, but says that allaḏī (masculine singular) and allatī (feminine singu-
lar) 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 pro-
nouns 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 commen-
tary 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):
ii. pronouns
iii. ambiguous nouns (i.e. demonstratives and relatives)
iv. nouns prefixed with al-
v. nouns followed by definite noun phrases
Al-Zamaḫšarī mentions the following proper nouns which identify a genus (ap.
Ibn Yaʿīš, Šarḥ I, 34.26f.):
ism kunya
Lion species: ʾUsāma ʾAbū l-Ḥāriṯ

Fox species: Ṯuʿāla ʾAbū l-Ḥuṣayn
Hyena species: Ḥaḍājir ʾUmm ʿĀmir
Scorpion species: Šabwa ʾUmm ʿIryaṭ
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 intro-
duction 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. How-
ever, 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 view-
point 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 view-
point 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 atten-
tion 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 mascu-
line 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 (ḥaya-
wān), 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 depen-
dent (see Tanaka 1981).
7 However, ambiguous nouns are treated just as demonstratives in the argument of the defi-
niteness hierarchy, which we will see in the following section.
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. Demon-
stratives 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 gen-
itives/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 pro-
nouns 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 rela-
tive 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 neces-
sary 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 al-
v. 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, rel-
atives 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 rel-
atives’ (Šarḥ 75–77, on lines 88–90 of the ʾAlfiyya). Both particle relatives and
nominal relatives need subordinate clauses. As for the nominal relatives (maw-
ṣūl 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 rel-
ative 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 rel-
ative’ (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 fol-
low nominal relatives.
8 Giolfo (2012: 144, n. 38) says “Arabic ṣila designates a sentence after a mawṣūl either ismī (rel-
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 al-
vi. 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 rela-
tive 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, al-
maḍrūbu.
In (6) the pronoun -hu is an example of a pronoun in the relative clause refer-
ring 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’.
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 *ista-
qarra bika ‘?He settled by you’ do not make sense, and therefore they are not complete
(tāmm).
10 Ibn Hišām refutes two opinions about this al- (Šuḏūr 148). One view regards this as a par-
ticle 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. Addition-
ally, al- has a pronoun referring to the relative. The second view regards this as a definite
article. He denies this opinion, too, because alas 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-.
3 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 fol-
lowed 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 def-
inite, 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 lat-
ter.
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 pro-
nouns are more definite than individual proper nouns. In the first place, a pro-
noun 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 definite-
ness.
Ibn Yaʿīš refutes this argument by stating that proper nouns are more defi-
nite 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 definite-
ness.
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-yuʿṭī-

nī
‘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-maġ-
ḍūbi ʿalayhim
‘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 exam-
ple 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 uniquely type
> Activated > familiar > > referential >
focus identifiable identifiable
⎧
{
that ⎫
}
⎨
{
this ⎬
}
{indefinite
⎩ this N ⎭
{it} {that N} {the N} {a N}
this N}
figure 2 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.
b. { } the same dog that bit Mary Ben last summer.
It’s
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 expres-
sion. 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 ‘famil-
iar’ 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 infor-
mation 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-
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 Given-
ness 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 al-
mubhama (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 deter-
mined 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.
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 al-
ʾAnṣārī, Šarḥ Qaṭr al-nadā wa-ball al-ṣadā. 1st ed. Saida and Beirut: al-Maktaba al-
ʿAṣriyya, 1994.
Ibn Hišām, Šuḏūr = Jamāl al-Dīn ʾAbū Muḥammad ʿAbdallāh ibn Yūsuf Ibn Hišām al-
ʾAnṣārī, 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.
muqtaḍab. 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 lin-
guistics. III. The development of a tradition: Continuity and change, ed. by Georgine
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 old-
est 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 reorga-
nization 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 com-
bination with it, in order to specify the time of its appearance, and then we
analyze the terminological distributions in connection with the opposition def-
inition/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).

254 sartori
2 Definition/Indefinition and Determination/Indetermination
Two pairs of notions should be effectively and logically distinguished on the fol-
lowing 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/indef-
inition, 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 defi-
nite 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 ‘indefi-
nite 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 determi-
nation? 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), Reck-
endorf, 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 def-
inition 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). Regard-
ing 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 Franco-
German 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 broad-
est 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, accord-
ing 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 ‘particulariza-
tion’, 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 sec-
ondary 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 techni-
cal 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 pur-
poses 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).

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). Never-
theless, 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 adjec-
tives (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 wa-
ḏamm 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 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 under-
standing 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).
11 Troupeau (1976:85) records only one occurrence each of ḫallaṣa and ḫallaṣa min, the for-
mer in the sense of ‘to clarify’, the latter in that of ‘to get free from’.
12 See Mubarrad, Muqtaḍab II, 567 and Zajjājī, Lāmāt 114.
13 The appearance, at that time, of these words is a striking manifestation of the introduc-
tion 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 quo-
tation 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 expres-
sion 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 adjec-
tive 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 equiv-
ocity 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 phe-
nomenon is present, but with a different term. He writes (Taʿrīfāt 73): “Clarifi-
cation denotes the fact of removing the ellipsis that happens in definite expres-
sions” (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 ‘Polysemy’ in linguistics (see Larcher 2011:307, n. 4), is what in logic is called ‘equivoc-
ity’.
15 In Ibn ʿUṣfūr, taḫlīṣ does not appear with the technical sense identified elsewhere, as evi-
denced 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).
16 This term is cited twice by Gätje (1970:235, 239), who translates it similarly as “Verdeut-
lichung oder Explikation”, i.e. ‘clarification’ or ‘explication’.
Know then that, with respect to indefinite expressions, the adjective con-
veys particularization and, with respect to definite expressions, clarifi-
cation. 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 bi-
l-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 fa-
ʾanta turīdu ʾan tubayyina li-l-muḫāṭab ʾannaka ʿanayta minhumā allaḏī
huwa ṭawīl fa-kāna ḏālika ʾizāla li-l-labs wa-tawḍīḥan wa-lā yakūnu taḫ-
ṣīṣan 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 yata-
ṣawwara 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 al-
Zamaḫšarī (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 apposi-
tion: 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 particulariza-
tion 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 wa-
li-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] particulariza-
tion concerns indefinite expressions and that clarification concerns definite
expressions” (qawluhu wa-fāʾidatuhu taḫṣīṣ ʾaw tawḍīḥ fa-l-taḫṣīṣ fī l-nakirāt wa-
l-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 indef-
inite 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 occur-
ing 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 qal-
lalta l-ištirāk wa-l-iḥtimāl wa-maʿnā l-tawḍīḥ ʿindahum rafʿ al-ištirāk al-
ḥāṣil 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 corrob-

oration. 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
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 conso-
nantal 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 li-
l-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 inher-
ent 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 zay-
dun 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 qual-
ification 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 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 indéter-
miné, ex. marartu bi-rajulin kātibin, décrire l’objet d’ un nom déterminé, ex. marartu bi-
zaydin 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).
19 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 apposi-
tion 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 partic-
ularizing 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 l-
ištirāk 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 ʾa-
lā 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 oth-
ers 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 else-
where, 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 defini-
tion, 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).

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 ʾilā-
hayni ṯnayni).21
On the basis of all of these sources, except for the special cases of Ibn al-
ʾAnbārī 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 attribu-
tive 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 spe-
cial 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 particular-
ization” (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 annex-
ation, 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 l-
muḍāf 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 indef-
inite one, as in ġulāmu mraʾatin ‘a woman’s servant’ […]. It [i.e. the ʾiḍāfa
lafẓiyya ‘formal annexation’] conveys neither definition nor particulariza-
tion (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 defini-
tion (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 wa-
taʿrīfan 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 annexa-

tion 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
expression, like ġulāmu rajulin ‘a man’s servant’ (wa-l-ʾiḍāfa al-maʿna-

wiyya 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: “annex-

ation is joining two nouns in such a way that it conveys definition or particu-
larization” (al-ʾiḍāfa hiya imtizāj ismayn ʿalā wajh yufīdu taʿrīfan ʾaw taḫṣīṣan).
Thus, it appears that within the framework of annexation, the terminolog-
ical pair is indeed taʿrīf/taḫṣīṣ, of which the former corresponds to the annex-
ation 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 qual-
ification. 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 ‘determi-
nation’, the last mentioned case conveying definition and determination at the
same time. At this point, no single term seems therefore to express the con-
cept 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 defini-

tion 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 connec-
tion with ʿaṭf al-bayān.
266 sartori
is a merchant’ ( fa-ḫaraja bi-qayd al-takmīl al-nasaq wa-l-badal […] wa-

l-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 com-
pleted 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, how-
ever, only to the framework of (broad) qualification, not to that of annexation.
This prompts us to distinguish, under taḫṣīṣ, that which is opposed and com-
plementary 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 inter-
section 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).
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/indef-
initeness. Under this pair, while no term seems to exist in the Arabic grammat-
ical metalanguage to signify ‘indetermination’, the second member of the pair,
‘determination’, seems to correspond to many Arabic terms. It is taḫṣīṣ ‘particu-
larization’ within the framework of indefiniteness, and taḫlīṣ-tawḍīḥ (and even,
but more marginally, ʾīḍāḥ) ‘specification’, ‘clarification’ within that of defini-
tion, but also taʿrīf ‘definition’ for the particular case of annexation within the
definite framework. Only one author, Ibn Hišām al-ʾAnṣārī, assigns a special sta-
tus 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 defini-
tion.24 Therefore, Arab grammarians felt that something else than the mere
opposition tankīr/taʿrīf was at work. Obviously without using terms equiva-
lent to our pair ‘indetermination/determination’, they came close to ‘deter-
mination’ 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’, under-
stood 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 expres-
sions 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 Note that an indefinite expression can be determinate (a tall man, where man is determi-
nate by a (quantification) and tall (qualification)) or indeterminate (man), while a definite
expression is necessarily determinate (the man).
25 See Morais Barbosa (1998).
268 sartori
tankīr taʿrīf
↗ al-rajul
rajul → taʿrīf
↘ rajul al-
annexation rajul → taḫṣīṣ2 → rajul madina
madīna
qualification rajul → taḫṣīṣ1 → rajul ṭawīl al-rajul → taḫlīṣ- → al-rajul al-

tawḍīḥ ʾīḍāḥ ṭawīl
explanatory rajul → → rajul tājir zayd → → zayd al-tājir
apposition
takmīl
figure 1 tankīr ‘indefiniteness’; taʿrīf ‘definition’; taḫṣīṣ ‘particularization’; taḫlīṣ ‘speci-
fication’; tawḍīḥ ‘clarification’; ʾīḍāḥ ‘elucidation’; takmīl ‘completion’
Determinate
↙ ↘
Indefinite Definite
kalb(un) al-kalb(u)
kalb(un) jamīl(un) al-kalb(u) al-jamīl(u)
kalb(u) ġulām(in) kalb(u) al-ġulām(i) / kalb(u) zayd(in)
figure 2 Schema of ‘determination’
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 Muḥam-
mad 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 al-
Kutub al-ʿIlmiyya, 2010.
Ibn al-Dahhān, Ġurra = ʾAbū Muḥammad Saʿīd ibn al-Mubārak ibn ʿAlī al-ʾAnṣārī al-
maʿrūf 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 al-
Ḥājib al-Miṣrī al-Dimašqī al-Mālikī, al-ʾImlāʾ ʿalā l-Kāfiya fī l-naḥw. Ed. by Manuel
Sartori. [unpublished], 2012.
Ibn al-Ḥājib, Kāfiya = Jamāl al-Dīn ʾAbū ʿAmr ʿUṯmān ibn ʿUmar ibn ʾAbī Bakr Ibn al-
Ḥājib 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, al-
Ṣāḥibī 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 al-
Ḥamīd 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 al-
Qurā, 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 al-
ʿAṭā 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 Muḥam-
mad Muʿawwaḍ and ʿĀdil ʾAḥmad ʿAbd al-Mawjūd. 2nd ed. 2 vols. Beirut: Dār al-
Kutub 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 al-
Jārburdī 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 al-
Jurjānī, Dalāʾil al-ʾiʿjāz. Ed. by Muḥammad al-Tunjī. 3rd ed. Beirut: Dār al-Kitāb al-
ʿArabī, 1999.
270 sartori
Jurjānī, Muqtaṣid = ʾAbū Bakr ʿAbd al-Qāhir ibn ʿAbd al-Raḥmān ibn Muḥammad al-
Jurjānī, 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 al-
maʿrūf bi-l-Ḫaṭīb al-Qazwīnī, al-ʾĪḍāḥ fī ʿulūm al-balāġa. al-maʿānī wa-l-bayān wa-l-
badīʿ. 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 al-
Jurjānī 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: Cam-
bridge 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 Moroc-
can, Egyptian, Syrian and Kuwaiti dialects. Washington, D.C.: Georgetown University
Press.
Buckley, Ronald Paul. 2004. Modern Literary Arabic: A reference grammar. Beirut: Li-
brairie 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
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, partic-
ules. 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 self-
instructing students in general. London: Wm. H. Allen & Co.
Gätje, Helmut. 1970. “Zum Begriff der Determination und Indetermination im Arabis-
chen”. 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: Mai-
sonneuve.
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 Anghe-
lescu, 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 Universitäts-
buchhandlung.
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: Con-
tinuity 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. [Avail-
able 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.
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 ele-
ment (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’ per-
ception of language.
Like many other terms, tawṭiʾa is neither defined nor explained by the gram-
marians. 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 dis-
cussions 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 ‘father-
in-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.

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 al-
Warrāq (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 li-
mā 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 appar-
ent 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 some-
one. 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 sheyhatovitch
3 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 al-
ʾAstarābāḏī 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 sup-
posed 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 men-
tions that, undoubtedly, in most cases both ḥāl and waṣf ‘qualifier’ are morpho-
logically 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 “pre-
pares 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 sig-
nificant 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 cat-
egorizations of ḥāl: (i) distinguishing between ḥāl denoting transient and per-
manent 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).
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), refer-
ring 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 exam-
ples 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’ la-

The 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 com-
bination 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 con-
ditional. 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 la-
ʾātiyannaka.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” (tan-
bīhan 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 con-
sidered 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 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 recon-
struction by the grammarians (see Versteegh 2009b).
10 ʾAstarābāḏī, Šarḥ IV, 315.3f.
11 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.
12 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.
13 See Jones (2004) for a discussion of the Qurʾānic verses condemning poetry and stating
that poetry was not a suitable register for the revelation.
To summarize, in wa-ʾin ʾaṭaʿtumūhum ʾinnakum la-mušrikūna not only the

apodosis of the conditional is omitted, but also omitted are the oath expres-
sion 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 gram-
marians’ 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 wa-
ḥikmatin ṯumma jāʾakum rasūlun muṣaddiqun li-mā maʿakum la-tuʾmi-
nunna bihi wa-la-tanṣurunnahu (Q. 3/81) ‘And when God took the cove-
nant 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) wa-
llāhi 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 pre-
cedes 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-l-
lāmu 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 inter-
prets (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 “every-
thing 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 la-
faʿaltu ‘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 al-
šarri muẓlimu ‘I swear that if we were to meet you,/ it would have been a
dark day of evil for you’.19

16 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 con-
tradiction 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” (al-
ʾasmāʾ llatī yujāzā bihā wa-takūnu bi-manzilati llaḏī), and mentions mā in this context.
17 Sībawayhi (Kitāb I, 404.8–14) explains that when a speaker recalls an oath made by some-
one 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 l-
nabiyyīna in (i).
18 See Qurṭubī ( Jāmiʿ IV, 124.12–126.12) for a detailed discussion on the verse, including gram-
matical analyses that evoke various interpretations of the text.
19 See Baġdādī, Ḫizāna X, 80.3–84.14 for a discussion of this verse and its context.
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 con-
ditional.
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 sen-
tences 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 la-
here 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 pre-
senting 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 con-
ceivably 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 combi-
nation 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 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).
21 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 Sīb-
awayhi viewed the aforementioned ʾan as redundant.24 This means that al-
Baġdādī (in agreement with al-ʾAstarābāḏī) interprets preparing the way and
redundancy as two contradictory functions. However, Ibn Hišām has a differ-
ent 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 inten-
tion 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 exam-
ples of lā with the same function are: fa-lā wa-rabbika lā yuʾminūna ḥattā
22 See Baġdādī (Ḫizāna X, 81.10–15) where he cites Ibn ʿUṣfūr’s commentary on Kitāb al-ʾĪḍāḥ
written by ʾAbū ʿAlī al-Fārisī.
23 See Baġdādī, Ḫizāna X, 82.8–15 for the grammarians’ criticisms of this formulation.
24 The relevant fragment appears in Ibn Hišām, Muġnī, 50.6–51.1.
25 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).
26 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 exam-
ple 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.
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 oppo-
site), 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 ‘prepa-
ration’, whereas Ibn Hišām (and, supposedly, also Sībawayhi) do not. This dis-
agreement 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 distin-
guishes 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 ‘prepar-
ing’ particles presented above. Types of clarification can definitely include indi-
cating to the addressee that the second clause is the apodosis of the oath (which
27 See Baġdādī, Ḫizāna XI, 221.5–224.6 for a discussion of this verse.

28 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 psycho-
logically 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 empha-
sis. 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 func-
tion 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 contri-
bution, 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 contribu-
tion is restricted to emphasis), but no other grammarian supports this. More-
over, 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 con-
sidered 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 dis-
cussed 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.
5 Tawṭiʾa in Discussions on Adjectival Phrases
Al-ʾAstarābāḏī (Šarḥ III, 435.10–19) starts his discussion of the grammatical gov-
ernment 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 theoret-
ically 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 afore-
mentioned 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) can-
not include a personal pronoun. The structures with noun in rafʿ are consid-
ered 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 con-
stituents 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 See al-Ġaḍḍāb (2008:100–109) for a detailed description of all these possibilities.

30 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.
31 See ʾAstarābāḏī, Šarḥ III, 439.3–19 for discussion.
32 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, accord-
ing 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 prin-
ciple 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). Cre-
ating 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 con-
stituent to which it assigns the rafʿ. For instance, in the sentence zay-
dun 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 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<mīzān (Ḫaṣāʾiṣ I, 50.5–6); the preference for bound pronouns over independent
ones (see Ḫaṣāʾiṣ II, 195.1–2, 9–11).
34 Al-ʾAstarābāḏī (Šarḥ II, 205.3–4) explains that in a pure annexation the speakers intend to
merge two words so that the first will receive definiteness or specification from the sec-
ond. The speakers thus omit “the marker of the completeness of the [first] word” (ʿalāmat
tamāmi l-kalimati), i.e. the tanwīn, the nūn of the sound plural masculine ending (which
the grammarians regard also as tanwīn), and sometimes even the feminine marker. Else-
where he notes (Šarḥ I, 396.8–13) that this merger is not complete: although the first noun
loses its tanwīn, it still takes a case-dependent ending.
originally assigned] rafʿ take the form of an object, because the governing
adjective is not co-referential with [the noun to which it assigns] naṣb; for
instance, in the aforementioned example ḍārib is not co-referential with
ʿamr. Thus, if [the adjective] is annexed [to the noun to which it origi-
nally assigned rafʿ], after [that noun was assigned] naṣb, this amounts to
annexing a word to something foreign (i.e., not co-referential). Therefore,
[the noun to which the adjective originally assigned rafʿ] takes naṣb as
a preparation for jarr (ḏālika li-ʾanna ʾiḍāfata l-ṣifati ʾilā marfūʿihā qabī-
ḥatun fī l-ẓāhiri, li-ʾanna l-ṣifata l-rāfiʿata li-l-ẓāhiri hiya l-marfūʿu bihā fī
l-maʿnā, kamā fī qawlika zaydun ḍāribun ġulāmuhu ʿamran, fa-l-ḍāribu
huwa ġulāmuhu, fa-kāna ka-ʾiḍāfati l-šayʾi ʾilā nafsihi llatī hiya mustaqba-
ḥatun fī l-maḥḍati wa-hiya ʾaṣlun li-ġayri l-maḥḍati fa-jaʿalū l-marfūʿa fī
ṣūrati l-mafʿūli, li-ʾanna l-ṣifata l-nāṣibata ġayru l-manṣūbi bihā fī l-maʿnā;
ʾa-lā tarā ʾanna l-ḍāriba ġayru ʿamrin fī l-miṯāli l-maḏkūri, fa-ʾiḏā ʾuḍī-
fat ʾilayhi baʿda naṣbihi kānat ka-ʾiḍāfati l-šayʾi ʾilā l-ʾajnabiyyi fa-nuṣiba
maʿmūlu l-ṣifati ʾiḏan li-ʾajli tawṭiʾati l-jarri).35
In the above excerpt al-ʾAstarābāḏī interprets the construction with naṣb as an

intermediate stage in the transformation36 from the heavy construction with
rafʿ into an annexation in that the construction with naṣb allows the second
constituent in the phrase to be presented not to be co-referential with the first.
This is possible because, usually, the constituent receiving the naṣb from an
adjective is not co-referential with the adjective (unlike the constituent in rafʿ,
which is co-referential with the governing adjective). Thus, the construction
with naṣb “prepares the way” for the construction with jarr, which is the final
goal of the transformation process.
This notwithstanding, the combination al-ḥasanu wajhihi is considered to
be absolutely unacceptable (because of the contradiction between the ‘heavy’
combination of the definite article and the bound pronoun, found in the first
and the second words respectively, and the purpose of annexation, which is
to achieve lightness), and ḥasanu wajhihi is not to be used except in poetry.
Thus, one would expect their counterparts with naṣb, al-ḥasanu wajhahu and
ḥasanun wajhahu to be unacceptable too—“since [a construction with naṣb]
35 ʾAstarābāḏī, Šarḥ III, 438.7–12.

36 I do not use the term ‘transformation’ in the exact same sense as the generative linguists,
as it is difficult to relate the phenomenon described here to any transformation rule rec-
ognized by them. The use of ‘transformation’ in the current context seems appropriate
because al-ʾAstarābāḏī speaks of two basic structures, of which others are created with
the purpose of lightening speech.
288 sheyhatovitch
is a preparation for the construction with jarr, and is not intended for its own
sake” (ʾiḏ huwa tamhīdun li-l-jarri wa-laysa maqṣūdan bi-ḏātihi).37 Nevertheless
al-ʾAstarābāḏī justifies the constructions with naṣb as follows:
However, [the Arabic speakers] permit [constructions such as al-ḥasanu

wajhahu and ḥasanun wajhahu] despite their ugliness even when there
is a choice,38 so that the naṣb in the constituent that was originally in
rafʿ would appear as clearly as possible, regardless of the possibility [to
transform the phrase in question into] an annexation, so that it would
be clear that the constituent in jarr [in the acceptable annexation con-
structions derived from the most basic structures, in which the second
constituent takes rafʿ] was originally in naṣb (lākinnahum jawwazūhumā
ʿalā qubḥin fī l-saʿati ʾayḍan li-yaẓhara l-naṣbu fī-mā kāna fāʿilan, sawāʾun
jāzat al-ʾiḍāfatu ʾaw lā, ġāyata l-ẓuhūri fa-yatabayyana fī l-majrūri ʾannahu
kāna qablahu manṣūban).39
Here al-ʾAstarābāḏī regards “preparing the way for the jarr” as the main reason
behind the existence of constructions such as al-ḥasanu wajhahu and ḥasa-
nun wajhahu. The idea is that using these constructions with naṣb facilitates
creating constructions with jarr (which are desirable due to their lightness),
by mentally preparing the speakers to ignore the co-reference of the two con-
stituents in question.
It must be noted that although al-ʾAstarābāḏī builds an elaborate argument
to justify the usage of constructions of the patterns al-ḥasanu wajhahu and
ḥasanun wajhahu outside the realm of poetry, his only relevant example is a
poetry line ascribed by some to ʿUmar ibn Lajaʾ:
ʾanʿatuhā ʾinnī min nuʿʿātihā / kūma l-ḏurā wādiqatan surrātihā

38 The sense of the term saʿa here seems to be different from the one recognized and
discussed by several scholars that is related to “the process by which a word is placed
beyond its proper boundaries, as an extension of its normal domain” and to the charac-
ter of the Arabic language that permits such processes; see Versteegh (1990:183), Baalbaki
(2008:279–280), Peled (2009:58–59), etc. Here al-saʿa is contrasted with poetical usages,
with the intention of saying that the abovementioned constructions are permitted also
when the speaker has the ability to choose (and not only in poetry, where the speaker
sometimes is forced to use anomalous structures for the sake of the rhyme and the
rhythm). (Ḥāl) al-saʿa is contrasted with ḍarūrat (al-šiʿr) e.g., in Ibn Jinnī, Ḫaṣāʾiṣ I, 329.5;
ʾAstarābāḏī, Šarḥ I, 522.6–7.
‘I describe [the camels], being one of their describers,/ as having tall

humps and dangling navels.’40
Wādiqatan surrātihā is analogous to ḥasanun wajhahu; although surrātihā by

itself may be interpreted as a noun in naṣb/ jarr (being a sound plural feminine
form, in which the markers of these cases are identical), the phrase cannot be
considered an annexation because of the tanwīn in wādiqatan.
6 Tawṭiʾa in Discussions on Doubly Transitive Verbs
According to a general principle held by the grammarians, pronouns must be

bound as long as this complies with other linguistic principles and results in no
ambiguity.41 However, al-ʾAstarābāḏī explains the use of an independent pro-
noun in the position of the second object of a doubly transitive verb, when a
bound pronoun causes no ambiguity, in terms of preparing the way to construc-
tions in which using a bound object would cause ambiguity.
This issue is raised in his discussion of a transformation called al-ʾiḫbār bi-
llaḏī of sentences with the doubly transitive verb ʾaʿṭaytu ‘I gave [something to
somebody]’.42 In general, al-ʾiḫbār bi-llaḏī transforms a chosen word in a given
sentence into a nominal predicate, while the rest of the sentence is turned into
an independent relative clause functioning as a subject. This transformation is
used by the grammarians to test various rules and principles of their theory.43
Al-ʾAstarābāḏī (Šarḥ II, 430.9–12) explains that if one is to “turn the second
direct object of ʾaʿṭaytu zaydan dirhaman ‘I gave Zayd a dirham’ into a predicate”
(ʾiḏā ʾaḫbarta ʿan al-mafʿūli l-ṯānī fī ʾaʿṭaytu zaydan dirhaman), the structure
allaḏī ʾaʿṭaytuhu zaydan dirhamun is preferable to allaḏī ʾaʿṭaytu zaydan ʾiyyāhu
dirhamun ‘The thing I gave Zayd is dirham’ (in the first structure, the second
direct object, which was separated from the verb due to the transformation, is
replaced by a bound pronoun, the -hu of ʾaʿṭaytuhu, while in the second struc-
ture, it is replaced by an independent pronoun ʾiyyāhu). The option with the
40 ʾAstarābāḏī, Šarḥ III, 438.19. The verse is discussed in Baġdādī, Ḫizāna VIII, 221.13–222.7.
41 For instance, Ibn al-Ḥājib says (ʾAstarābāḏī, Šarḥ II, 427.7) that “[the use of] an indepen-
dent [pronoun] is permitted only when a bound one is impossible” (lā yasūġu l-munfaṣilu
ʾillā li-taʿaḏḏuri l-muttaṣili). See Peled (2006:556f.) for a review of cases in which the use
of an independent accusatival pronoun is permitted.
42 I follow Baalbaki (2008:216) in using the term ‘transformation’ in this context.
43 See Baalbaki (2008:215–217) for a discussion on al-ʾiḫbār bi-llaḏī as “a tool of checking sys-
tem validity” in Arabic grammatical theory.
290 sheyhatovitch
bound pronoun is preferred on account of the abovementioned general prin-

ciple of the necessity of the use of bound pronouns.
Al-ʾAstarābāḏī (Šarḥ II, 430.12f.) mentions that there are grammarians who
accept the second structure—allaḏī ʾaʿṭaytu zaydan ʾiyyāhu dirhamun—“as a
preparation for the removal of ambiguity [in cases where using] a bound
pronoun [would cause ambiguity]” (tawṭiʾatan li-ʾizālati l-labsi bi-l-ittiṣāli). He
exemplifies such cases of ambiguity with the sentence ʾaʿṭaytu zaydan ʿamran
‘I gave ʿAmr to Zayd’, but does not develop an argument for the position of
those grammarians. However, such an argument can be surmised to be as
follows. If one were to transform the second object of this sentence into a
predicate while using a bound pronoun in the relative clause, the sentence
would be allaḏī ʾaʿṭaytuhu zaydan ʿamrun. Then, it would seem as if Zayd was
given to ʿAmr whereas the intention is the opposite. For verbs like ʾaʿṭā, it is
well known that when the context does not include any clues as to the iden-
tity of the object given and the person receiving, the first direct object is to
be interpreted as the receiving person, and the second direct object as the
given object.44 Thus, to make the intention clear, one must use an indepen-
dent pronoun and say allaḏī ʾaʿṭaytu zaydan ʾiyyāhu ʿamrun. The independent
pronoun allows positioning the constituent denoting the given object after
the constituent denoting the receiving person, thereby rendering the sentence
unequivocal.
According to al-ʾAstarābāḏī, those who produce sentences like allaḏī ʾaʿṭaytu
zaydan ʾiyyāhu dirhamun take into account the existence of structures of the
type allaḏī ʾaʿṭaytu zaydan ʾiyyāhu ʿamrun and “prepare the way” for the lat-
ter. This preparation introduces a certain standardization by which all the
sentences obtained from al-ʾiḫbār bi-llaḏī where the second object of ʾaʿṭaytu
becomes the predicate may be constructed in the same pattern.
Interestingly, in al-ʾiḫbār bi-llaḏī transformation of ʿalimtu zaydan qāʾiman ‘I
knew that Zayd was standing’, the option allaḏī ʿalimtu zaydan ʾiyyāhu qāʾimun
‘What I knew about Zayd is that he was standing’ (with the independent pro-
noun ʾiyyāhu replacing the second object in the relative clause) is preferable
to the option allaḏī ʿalimtuhu zaydan qāʾimun (with the bound pronoun -hu
replacing the second object).45 Al-ʾAstarābāḏī (Šarḥ II, 431.1–3) says that the
preference for an independent pronoun in this case can be explained either
44 For instance, Ibn Yaʿīš (Šarḥ VII, 64.2–4) states that one can change the basic word order
of the sentence ʾaʿṭaytu zaydan dirhaman ‘I gave Zayd a dirham’ and say ʾaʿṭaytu dirhaman
zaydan and zaydan ʾaʿṭaytu dirhaman, but cannot change the basic word order of the sen-
tence ʾaʿṭaytu zaydan ʿamran ‘I gave Zayd ʿAmr’.
45 ʾAstarābāḏī, Šarḥ II, 430.14–431.1.
by “preparing the way [for sentences of the same pattern in which the use of
a bound pronoun would result in ambiguity], as explained above” (li-l-tawṭiʾati
l-maḏkūrati), or by “considering the basic structure for the second object [of
doubly transitive verbs], whose governor, in principle, is not supposed to be
bound to it” (li-riʿāyati ʾaṣli l-mafʿūli l-ṯānī, ʾiḏ al-ʿāmilu fīhi, fī l-ʾaṣli, mā yajibu
nfiṣāluhu ʿanhu).
In this discussion al-ʾAstarābāḏī does not make completely clear the differ-
ence between the syntactic behavior of ʾaʿṭaytu and ʿalimtu, but we may spec-
ulate that the difference lies in the fact that the two direct objects of ʿalimtu
are co-referential (unlike those of ʾaʿṭaytu). It seems that in the case of co-
referential objects the risk of ambiguity is greater than when objects are not co-
referential: it is easier to confuse something with what is known about it than
to confuse a receiver with a given object. This explains why the effect of tawṭiʾa,
the preparation for removing ambiguity, is stronger in ʿalimtu than in ʾaʿṭaytu.
7 Conclusion
The grammarians’ discussions on ḥāl muwaṭṭiʾa and ‘preparing’ particles in oath

sentences present one constituent as preparing the way for another (or oth-
ers). The function of the ‘preparing’ constituents in the discussed cases may be
considered as pragmatic, as these constituents elucidate the relations between
other utterance constituents, thereby helping the addressee to decipher the
utterance. The ‘preparing’ constituents are not obligatory, as the addressee
would understand the speaker’s intention also without them (hence, the dis-
tinction between ‘preparing’ and redundant particles in some cases is unclear),
but nonetheless serve as signals that facilitate deciphering.
The other discussions mentioned in this paper (the six nouns, adjectival
phrases and al-ʾiḫbār bi-llaḏī transformations of sentences that include doubly
transitive verbs) interpret a linguistic phenomenon present in a certain con-
struction as preparing the way for another construction that can be derived
from the former. In these cases tawṭiʾa may be viewed as a standardization
tool: the speakers producing utterances with the singular form of the six nouns
as a governing noun in an annexation are mindful of the plural forms, and
assign those nouns case markers in the form of letters (as an analogy to plural
forms); the speakers producing phrases such as al-ḥasanu wajhahu/ḥasanun
wajhahu take into account the annexation structures that can be created from
these phrases (and use the naṣb case as a preparation for the jarr); the speak-
ers producing sentences such as allaḏī ʾaʿṭaytu zaydan ʾiyyāhu dirhamun take
into account the existence of cases such as allaḏī ʾaʿṭaytu zaydan ʾiyyāhu ʿamrun
292 sheyhatovitch
(and use an independent pronoun in the former as a preparation for the latter,
in which an independent pronoun must be used).
The concept of tawṭiʾa indicates that the Medieval Arab grammarians view
language as a system in which a speaker who pronounces a word thinks not
only of the next constituents in his utterance, but also of other constructions
that can be derived from that utterance. The speaker in some way prepares
himself not only for the rest of his utterance but also for those derived con-
structions. Put in modern linguistic terms, the principle of tawṭiʾa acts on both
syntagmatic and paradigmatic axes.
A Primary Sources
ʾAbū Ḥayyān, Baḥr = Muḥammad ibn Yūsuf ʾAbū Ḥayyān al-Ġarnāṭī al-ʾAndalusī, Tafsīr
al-baḥr al-muḥīṭ. Ed. by ʿĀdil ʾAḥmad ibn ʿAbd al-Mawjūd and ʿAlī Muḥammad
Maʿūḍ. 8 vols. Beirut: Dār al-Kutub al-ʿIlmiyya, 1993.
ʾAstarābāḏī, Šarḥ = Raḍī l-Dīn Muḥammad ibn Ḥasan al-ʾAstarābāḏī, Šarḥ al-Raḍī ʿalā
l-Kāfiya. Ed. by Yūsuf Ḥasan ʿUmar. 4 vols. Benghazi: Manšūrāt Jāmiʿat Qaryūnas,
1996.
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: Maktabat al-
Ḫānjī, 1983.
Ibn al-ʾAnbārī, ʾInṣāf = Kamāl al-Dīn ʾAbū l-Barakāt ʿAbd al-Raḥmān ibn Muḥam-
mad Ibn al-ʾAnbārī, al-ʾInṣāf fī masāʾil al-ḫilāf bayna l-naḥwiyyīna l-Baṣriyyīna wa-
l-Kūfiyyīna. 2 vols. Cairo: al-Maktaba al-Tijāriyya al-Kubrā, 1961.
Ibn Hišām, Muġnī = Jamāl al-Dīn ʾAbū Muḥammad ʿAbdallāh ibn Yūsuf ibn ʾAḥmad Ibn
Hišām, Muġnī l-labīb ʿan kutub al-ʾaʿārīb. Ed. by Māzin al-Mubārak and Muḥammad
ʿAlī Ḥamd Allāh. Damascus: Dār al-Fikr, 1985.
Ibn Jinnī, Ḫaṣāʾiṣ = ʾAbū l-Fatḥ ʿUṯmān Ibn Jinnī, al-Ḫaṣāʾiṣ. Ed. by Muḥammad ʿAlī al-
Najjār. 3 vols. Cairo: al-Hayʾa al-Miṣriyya al-ʿĀmma li-l-Kitāb, 1986.
Ibn al-Sarrāj, ʾUṣūl = ʾAbū Bakr ibn Sahl Ibn al-Sarrāj, al-ʾUṣūl fī l-naḥw. Ed. by ʿAbd al-
Ḥusayn al-Fatlī. 3 vols. Beirut: Muʾassasat al-Risāla, 1988.
Ibn ʿUṣfūr, Šarḥ al-Jumal = ʾAbū l-Ḥasan ʿAlī ibn Muʾmin Ibn ʿUṣfūr, Šarḥ Jumal al-Zajjājī.
Ed. by Ṣāḥib ʾAbū Janāḥ. 2 vols. Baghdad: Wizārat al-ʾAwqāf wa-l-Šuʾūn al-Dīniyya,
1980–1982.
Ibn al-Warrāq, ʿIlal = Muḥammad ibn ʿAbdallāh ibn al-ʿAbbās Ibn al-Warrāq, ʿIlal al-
naḥw. Ed. by Maḥmūd Jāsim Muḥammad al-Darwīš. Riyadh: Maktabat al-Rušd, 1999.
Ibn Yaʿīš, Šarḥ = Muwaffaq al-Dīn ʾAbū l-Baqāʾ Yaʿīš ibn ʿAlī Ibn Yaʿīš, Šarḥ al-Mufaṣṣal.
10 vols. Cairo: Maktabat al-Mutanabbī, n.d.
Jurjānī, Muqtaṣid = ʿAbd al-Qāhir ibn ʿAbd al-Raḥmān al-Jurjānī, Kitāb al-muqtaṣid fī
šarḥ al-ʾĪḍāḥ. Ed. by Kāẓim Baḥr al-Marjān. 2 vols. Baghdad: Dār al-Rašīd, 1982.
Qurṭubī, Jāmiʿ = ʾAbū ʿAbdallāh Muḥammad ibn ʾAḥmad al-Qurṭubī, al-Jāmiʿ li-ʾaḥkām
al-Qurʾān. 20 vols. Cairo: Dār al-Kātib al-ʿArabī li-l-Ṭibāʿa wa-l-Našr, 1967.
Rāzī, Tafsīr = Faḫr al-Dīn ʾAbū ʿAbdallāh Muḥammad ibn ʿUmar al-Rāzī, al-Tafsīr al-
kabīr. 32 vols. Egypt: al-Maṭbaʿa al-Bahiyya al-Miṣriyya, 1938.
Derenbourg, Le livre de Sibawaihi. 2 vols. Paris: Imprimerie Nationale, 1881–1889.
ʿUkbarī, Lubāb = ʾAbū l-Baqāʾ ʿAbdallāh ibn al-Ḥusayn al-ʿUkbarī, al-Lubāb fī ʿilal al-bināʾ
wa-l-ʾiʿrāb. Ed. by ʿAbdallāh al-Nabhān. 2 vols. Damascus: Dār al-Fikr, 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.
Ed. by Māzin al-Mubārak. Beirut: Dār al-Nafāʾis, 1973.
Zamaḫšarī, Kaššāf = ʾAbū l-Qāsim Maḥmūd ibn ʿUmar al-Zamaḫšarī, al-Kaššāf ʿan
ḥaqāʾiq al-tanzīl wa-ʿuyūn al-ʾaqāwīl fī wujūh al-taʾwīl. Ed. by ʿAbd al-Razzāq al-
Mahdī. 4 vols. Beirut: Dār al-Turāṯ al-ʿArabī, 2008.
B Secondary Sources
Baalbaki, Ramzi. 2008. The legacy of the Kitāb. Sībawayhi’s analytical methods within the
Bernards, Monique. 2007. “Ḥāl”. Encyclopedia of Arabic language and linguistics, ed. by
Mushira Eid et al., II, 224–228. Leiden: E.J. Brill.
Cachia, Pierre. 1973. The monitor: A dictionary of Arabic grammatical terms. Beirut:
Librairie du Liban.
Diem, Werner. 1998. Fa-waylun li-l-qāsiyati qulūbuhum: Studien zum arabischen adjek-
tivischen Satz. Wiesbaden: O. Harrassowitz.
Ġaḍḍāb, Faraj Muḥammad. 2008. Maʿnā l-fāʿiliyya wa-dalālatuhu l-marjiʿiyya min ḫilāl
Šarḥ al-Kāfiya li-l-ʾAstarābāḏī. Sfax: Maktabat ʿAlāʾ al-Dīn.
Jones, Alan. 2004. “Poetry and poets”. Encyclopaedia of the Qurʾān, ed. by Jane Dammen
McAuliffe, IV, 110–114. Leiden: E.J. Brill.
Larcher, Pierre. 2006. “Derivation”. Encyclopedia of Arabic language and linguistics, ed.
by Mushira Eid et al., I, 573–579. Leiden: E.J. Brill.
Levin, Aryeh. 1986. “The medieval Arabic term kalima and the modern linguistic term
morpheme: Similarities and differences”. Studies in Islamic history and civilization in
honour of Professor David Ayalon, ed. by Moshe Sharon, 423–446. Jerusalem: Cana.
Peled, Yishai. 2006. “Ḍamīr”. Encyclopedia of Arabic language and linguistics, ed. by
Mushira Eid et al., I, 555–559. Leiden: E.J. Brill.
294 sheyhatovitch
Peled, Yishai. 2009. Sentence types and word-order patterns in written Arabic: Medieval
and modern perspectives. Leiden: E.J. Brill.
Testen, David D. 1998. Parallels in Semitic linguistics. Leiden: E.J. Brill.
Versteegh, Kees. 1990. “Freedom of the speaker? The term ittisāʿ and related notions in
Arabic grammar”. 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, 281–293. Amsterdam and Philadelphia: J. Ben-
jamins.
Versteegh, Kees. 2007. “ʾIlġāʾ”. Encyclopedia of Arabic language and linguistics, ed. by
Mushira Eid et al., II, 307–308. Leiden: E.J. Brill.
Versteegh, Kees. 2009a. “Ṣila”. Encyclopedia of Arabic language and linguistics, ed. by
Mushira Eid et al., IV, 235–237. Leiden: E.J. Brill.
Versteegh, Kees. 2009b. “Taqdīr”. Encyclopedia of Arabic language and linguistics, ed. by
Mushira Eid et al., IV, 446–449. Leiden: E.J. Brill.
Wright, William. 1967. A grammar of the Arabic language. 3rd ed. 2 vols. Cambridge:
Cambridge University Press.
Malay Grammar between Arab and Western Model
Kees Versteegh
1 Writing Indigenous Grammars in the Islamic World1
Ever since the introduction of Islam outside the Arab world, the Arabic lan-
guage was inextricably linked with this movement, and as such, it became an
essential part of intellectual culture in all Islamic countries. Knowledge of Ara-
bic was required not only for the study of Islam and the Qurʾān, but also for a
range of other topics. For most students in the Islamic sciences the introductory
level of grammar sufficed, and they rarely went beyond the study of a limited
list of treatises, such as the ʾĀjurrūmiyya (Drewes 1971; van Bruinessen 1990; Ver-
steegh 2018). At the end of their curriculum stood the ʾAlfiyya, which served as
a summary of everything the students had learnt (Viain 2014). Only those who
were really interested in grammar studied the more advanced grammatical lit-
erature. In some Islamic countries this is still the case: Arabic continues to be
taught there in much the same way as it used to be taught, both within and
without the Arabic-speaking world.
In the acquisition of Arabic the indigenous languages in the Islamic coun-
tries have always played an important role. Proficiency in reading Arabic texts
was not gained from studying grammatical treatises as such, but from recitation
of the Qurʾān and from the study of Islamic texts in theology, fiqh, ḥadīṯ, and
exegesis. These texts were usually learnt by heart together with their translation
in the indigenous language, which served as an auxiliary tool for explaining the
meaning of the text. Some scholars opted for this auxiliary language when they
composed a new elementary treatise, paraphrasing an existing Arabic text and
commenting on it. In all of these activities, native speakers used their own lan-
guage as a tool to understand Arabic. Obviously, they had no need to learn their
own language, so that, generally speaking, there was no demand for a grammar
of this indigenous language.
1 I wish to thank Jan van der Putten (Hamburg) and Peter Riddell (Melbourne) for their help
in procuring publicatons that would have been difficult to get by otherwise. I am also grateful
to Harimurti Kridalaksana ‘Pak Hari’ (Jakarta), who in 1998 presented me with a facisimile of
Raja Ali Haji’s Bustān al-kātibīn.

296 versteegh
In the Islamic world at large, the Arabic grammatical tradition functioned in

much the same way as the Latin tradition did for the European linguists when
they started to study exotic languages without a grammatical tradition. Euro-
pean missionaries, who were among the first to encounter these languages,
never doubted that the proper grammatical framework for the description of
these languages was that of Latin grammar, as this was the approach to lan-
guage study in which they had received their initial training during their early
school years, and it was in fact the only framework at their disposal. Moreover,
most of the languages with which they came in contact did not have any tradi-
tion of grammatical analysis of their own. The same situation obtained in the
language communities the Arabs encountered, where languages such as Cop-
tic, Berber, Turkic, Persian, Urdu, and Malay were spoken, which did not have
a grammatical tradition of their own.2
Eventually, some language communities in the Islamic world developed a
tradition of grammatical analysis of their indigenous language, comparable to
the grammars of vernacular languages in Western Europe that grew out of the
study of Latin. The contexts in which grammars of the indigenous languages
emerged in the Islamic world varied, and so did the motives for initiating the
study of the vernacular language rather than Arabic or in addition to Arabic.
They had one thing in common, though: the description of the vernacular lan-
guage in these traditions followed the model of the Arabic tradition. It was not
the case that the writers of these grammars consciously selected an appropriate
model but, rather, that for them Arabic grammar constituted the model for any
language study. Thus, they accepted without questioning the universal validity
of Arabic grammatical theory. On the basis of their education, they just knew
that language consists of three parts of speech, that there are weak and strong
consonants, that there exist only three vowels, that combinations of two qui-
escent consonants are impossible, and that such general characteristics apply
to all languages. Differences between the vernacular language and Arabic were
acknowledged, but these, too, were explained within the framework of Arabic
grammatical theory.
In the present paper, I shall look at the circumstances in which vernacular
grammars appeared in the Islamic world. My focus will be the Malay tradi-
tion, which produced grammars of both Arabic and Malay, written in both
2 The case of Hebrew and Syriac is different because these languages continued to serve as reli-
gious language for the Jewish and Christian communities in the Arab world. They had their
own tradition of studying these languages, yet even for them the impact of the Arabic model
was so strong that it fundamentally changed their grammatical tradition.
malay grammar between arab and western model 297
languages. Malay was the main lingua franca in South and Southeast Asia, and
remained so after Islamization, serving as the vehicle for the introduction of
Islamic science and scholarship in Malaysia and Indonesia (Collins 1996). It did
not become a subject taught within the curriculum of Islamic education until
the modern period.
In the Indonesian archipelago, Malay served for most people as a practical
second language, without enjoying the same prestige as languages like Javanese
(Groeneboer 1998). Javanese had a venerable tradition of language and liter-
ature study (Arps 1997), dating back to the reign of the Indian kings in the
Javanese Hindu empires of Mataram (8th–10th centuries) and Majapahit (13th–
14th centuries). This type of scholarship was heavily Sanskritized and included
the study of both Sanskrit and Javanese. An important part of it was devoted
to lexicography, but it also included grammatical, and especially phonetic stud-
ies. In modern Javanese grammar, a considerable number of Sanskrit loanwords
testifies to the lasting influence of this tradition, e.g. kriya ‘verb’, aksara ‘letter’,
purusa ‘(grammatical) person’, dwilingga ‘reduplication’ (Ogloblin 1981:213 f.).
The Malay learned tradition dates from a later period. It originated in Ma-
lacca and East Sumatra at the Islamic royal courts of the Johor Sultanate (16th–
19th century) and its successor, the Sultanate of Riau (1824–1911). These Islamic
states did not possess a tradition of Sanskrit learning, so that Sanskrit influence
in the technical terminology of Malay is limited. A few terms derived from San-
skrit are still found in Malay grammar, including arti ‘meaning’, nama ‘noun’,
kata ‘word’, and the general terms bahasa ‘language’ and tata bahasa ‘grammar’
(Ogloblin 1981:212). But the general study of grammar as it was developed in the
Islamic curriculum, was based on the Arabic grammatical tradition. Malay was
the auxiliary language that served for the explanation of Arabic grammar, but
it was not studied for its own sake.
2 Grammar and Exegesis in Malay
In 1886, L.W.C. van den Berg, an employee of the Dutch colonial authorities,
who had been commissioned to investigate the state of education at the local
pesantrens, the Islamic elementary schools in the Dutch Indies, submitted a
report about the contents of their curriculum. Thanks to his detailed list of
the textbooks used in these schools, we are relatively well informed about
the curriculum at the time. For grammar, the most popular treatises were
Ibn ʾĀjurrūm’s (d. 723/1323) ʾĀjurrūmiyya and Ibn Mālik’s (d. 672/1272) ʾAlfiyya.
Each pesantren selected its own elementary treatises. Only very few students
progressed beyond this stage to more advanced linguistic literature, such as
298 versteegh
the commentaries on the elementary treatises. The most popular commen-

taries on the ʾĀjurrūmiyya were those by al-ʾAzharī (d. 905/1499), al-Kafrāwī
(d. 1207/1787), and a Meccan scholar, ʾAḥmad Zaynī Daḥlān (d. 1304/1886). In
the case of the ʾAlfiyya, the most frequently used commentaries were those
by Ibn Hišām (d. 761/1360), al-ʾAzharī, Ibn ʿAqīl (d. 769/1367), and al-ʾUšmūnī
(d. 872/1468). Several local scholars in Southeast Asia contributed to local
knowledge about (Arabic) grammar, by commenting on the elementary trea-
tises and composing new adaptations. Among these indigenous treatises were
the commentaries by the Banten scholar Muḥammad Nawawī (d. 1897; van Bru-
inessen 1990:236, n. 20) and the 19th century glosses on Daḥlān’s commentary
by Muḥammad Maʿṣūm ibn Sālim from Semarang (Drewes 1971:69), both writ-
ten in Arabic.
Teaching these texts involved an extensive translation effort: the students
had to learn the Arabic texts by heart together with their interlinear Malay
translation. The teaching method of the pesantren involved the translation or
borrowing of grammatical terms in the Malay version, so that, even though the
subject matter of these grammatical treatises was the Arabic, rather than the
Malay language, a general stock of Malay linguistic terms needed to be built
up.
In Qurʾānic exegesis, too, insofar as it involved language-related comments
on the text, a set of Malay terms was needed. It is difficult to estimate how old
this exegetical tradition in Southeast Asia is. Riddell (1990, 2017) analyzed the
Qurʾānic commentary of the 17th century Acehnese scholar ʿAbd al-Raʾūf al-
Singkilī (d. 1693), entitled Tarjumān al-mustafīd, which was mainly based on
the popular commentary on the Qurʾān by the two Jalāl al-Dīns, al-Maḥallī
(d. 864/1459) and al-Suyūṭī (d. 911/1505), commonly known as Tafsīr al-Jalālayn.
In his commentary, al-Singkilī concentrates on the Malay paraphrase of the
text, with additional notes on selected qirāʾāt, added by his student, Dāʾūd
Rūmī.3
ʿAbd al-Raʾūf himself left out most grammatical comments in his adapta-
tion of the Tafsīr al-Jalālayn (Riddell 1990:64, 2017:61–65), so that grammati-
cal issues mainly occur in the context of discussions of qirāʾāt. Variant read-
ings are sometimes distinguished by a difference in meaning, for instance in
Q. 20/102 yawma yunfaḫu fī l-ṣūri ‘the day the trump will be blown’, where ʾAbū
ʿAmr’s reading nanfuḫu (Makram and ʿUmar 1984:IV, 112) is presented as follows
3 Riddell (1990:51–53); on Dāʾūd’s sources in the qirāʾāt literature, see Riddell (2014b:69, 2017:85–
100). Riddell concludes that in the commentary the canonical reading is that of Ḥafṣ ʿan ʿĀṣim,
while the variants that are provided stem from the ʾAbū ʿAmr and Nāfiʿ readings.
“when it is read with nūn, its meaning is ‘the day We shall blow the trump’ …”
(adalah maʿnanya tatkala dibaca dengan nūn hari Kami tiup sangkakala; Rid-
dell 1990:174; also 168).
Most of the technical terms used in the commentary concern the writing of
the vowels and the realization of ʾalif and hamza. Since the Malay commen-
tary is reproduced in Jawi Arabic script, Arabic terms usually appear in their
original orthography, rather than the form they later adopted as loanwords
in Malay. From Riddell’s (1990:245–250) list of grammatical terms used in the
commentary it appears that, along with the Arabic vowel names fatḥa, ḍamma,
kasra, the term tanwīn, the terms for the two hamzas (waṣl hamza-nya and qaṭʿ
hamza-nya; Riddell 2017:316), and the pair tašdīd/taḫfīf, some general phonetic
terms derived from Arabic were integrated to such a degree that they could
serve as the point of departure for verbal derivation, such as menghadhafkan
‘to apocopate’ (< Arabic ḥaḏf ‘deletion, elision’), mentahkikkan ‘to realize [the
hamza]’ (< Arabic taḥqīq ‘realization’), mentashilkan ‘to pronounce [without
hamza]’ (< Arabic tashīl ‘facilitation’), mengwakafkan ‘to pause’ (< Arabic waqf
‘pause’), mengwasilkan ‘to link’ (< Arabic waṣl ‘connection, joining’).4 Many of
the derived words contain the causal affix meN- … -kan, but other affixes are
used as well, for instance ber- denoting possession, as in tiada beralif ‘not with
an ʾalif ’ (Riddell 1990:159). This productivity is similar to that of most Arabic
loanwords in Malay (Versteegh 2003).
Two Arabic syntactic terms occur in the portion of the text edited by Rid-
dell (1990), ḥarf jarr and ism mawṣūl, both in connection with the explanation
of Q. 19/24 fa-nādāhā min taḥtihā ‘and he called out to her from beneath it’
(Riddell 1990:132). In the reading min taḥtihā, the kasra after the second t is
explained as the effect of the preposition (min ḥarf jarr). But there is a second
reading, man taḥtahā ‘the one who was beneath it’ (Makram and ʿUmar 1984:IV,
39), which the commentator explains as follows: “ʾAbū ʿAmr reads this as man
taḥtahā with a fatḥa on the mīm, as a relative pronoun, hence the fatḥa on the
tāʾ” (ʾAbū ʿAmr membaca dia man taḥtahā dengan fatḥa mīmnya ism mawṣūl,
maka sebab itulah fatḥa tāʾnya).
Original Malay terms are less frequent than Arabic loanwords in this Qurʾā-
nic commentary. For the sukūn ʿAbd al-Raʾūf uses mati ‘dead’, and for the notion
of ‘replacing [a sound with another sound]’ the Malay term menukarkan ‘to
change, replace’ (Riddell 1990:126, 165). In some cases, a Malay term is used as
variant for an Arabic term, e.g. menyebutkan (from the Malay root verb sebut
4 The two terms mentioned last also occur as nominal borrowings, e.g. tatkala wasl dan wakaf
‘both when connected and in pause’ (Riddell 1990:159).
300 versteegh
‘to pronounce, mention’; Riddell 2017:280) and menthabitkan (< Arabic ṯābit
‘firm’; Riddell 2017:278) or mengithbatkan (< Arabic ʾiṯbāt ‘confirmation’; Rid-
dell 2017:304), which are all three used for the notion of realizing a sound as
against deleting (menghadhafkan) it in pronunciation.
For ‘lenghtened’ (mamdūd) and ‘shortened’ (maqṣūr) with respect to the
ending of words, ʿAbd al-Raʾūf uses lanjut ‘long, prolonged’ and singkat ‘brief,
short’, for instance, when he describes the difference between the two vari-
ant readings in Q. 18/98 fa-ʾiḏā jāʾa waʿdu rabbī jaʿalahu dakkāʾa/dakkan ‘When
my Lord’s promise comes, He will level it to the ground’ (Riddell 1990:123,
2017:322; Makram and ʿUmar 1984:IV, 18). The first variant, dakkāʾa, is “with
hamza together with kāf and a lengthened [vowel]” (dengan hamza serta lan-
jut kāf-nya), while the second variant, dakkan, is “with tanwīn ending on the kāf
together with a shortened [vowel]” (dengan tanwīn kāf serta singkat).5
A similar example is that of the readings involving different connected forms
of the first person singular pronoun ʾana/ʾanā in Q. 18/34 and Q. 18/39 (Riddell
2017:286, 288; Makram and ʿUmar III, 364). In the latter verse, this difference is
expressed by ʿAbd al-Raʾūf as dengan singkat/lanjut ʾalif ʾanā-nya “with short-
ened/lengthened ʾalif in ʾanā”. In Q. 19/10 rabbi jʿal lī ʾāyatan ‘my Lord, give me
a sign’, the sequence -ī ʾā- leads to different realizations (Makram and ʿUmar
1984:IV, 34). In the Tarjumān one of these is described as follows: “with a vow-
elless y together with a long” (dengan mati yāʾ-nya serta lanjut).6 According to
Sībawayhi, when the hamza is preceded by iy/uw it is changed into a glide, so
that a word like ḫaṭīʾa is realized as [xɑᵵɪjːa] (Al-Nassir 1993:84f.). The Malay
phrase seems to suggest that in Q. 19/10 ʿAbd al-Raʾūf selected the variant liyyā.
The parallels cited above suggest that lanjut refers here, too, to a lengthened
vowel after the yāʾ.
Yet another Malay term, pendek ‘short [both of time and place]’ is used
for a short vowel variant. In Q. 19/66 ʾa-ʾiḏāmā mittu la-sawfa ʾuḫraju ḥayyan
‘When I die, will I be brought out alive?’, Ḥafṣ is reported to have read ʾa-ʾiḏā
“with short [vowel] and non-gemination [of the hamza]” (dengan pendek dan
takhfif ). The combination of two glottal stops, both followed by a vowel, was
a much discussed issue in Qurʾānic recitation, as well as in Sībawayhi’s Kitāb
5 It is not entirely clear what ‘shortened’ and ‘lengthened’ refer to here. According to Arabic
terminology the difference between dakkāʾ and dakkā (not dakkan) would be described as a
distinction between mamdūd and maqṣūr, because the former always retains its [aː], while
the latter shortens it in pausa and in a closed syllable. Both terms refer to a potentiality, ‘liable
to be lengthened/shortened’. It is not certain whether ʿAbd al-Raʾūf understood this; he may
have confused the terminology, hence his choice of Malay equivalents.
6 E.g. Riddell (1990:128, 139, 152, 155); he translates this phrase with “an unvowelled long yāʾ”
(1990:189).
(Roman 1983:322–348). According to Al-Nassir (1993:81–90), the rules for this

combination were slightly different for those instances where the first hamza
is that of the interrogative particle ʾa-. The usual solution for this combination
is to realize the hamza as a hamza bayna bayna, i.e., a weak glottal fricative [ɦ],
and to lengthen the first vowel (i.e., ʾa-ʾanta → [ʔɑːɦɑntɐ]). In the present verse,
several alternatives to read ʾa-ʾiḏā are reported in the qirāʾāt literature (Makram
and ʿUmar 1984:IV, 52). The one attributed in the Tarjumān to Ḥafṣ probably
indicates a variant in which the hamza is lightened (i.e., realized as a hamza
bayna bayna) without lengthening of the vowel, [ʔɑɦiðaː], as against the vari-
ant in which the hamza is lightened and the vowel lengthened, mentioned by
Sībawayhi (Kitāb III, 551; Roman 1983:343, Al-Nassir 1993:88).
Two other terms, the adjectives panjang ‘long [both time and distance]’
and bentar ‘momentary’ are used to distinguish the variant readings of Q. 19/2
(Makram and ʿUmar 1984:IV, 28). In this verse two readings are reported: ḏikru
raḥmati rabbika ‘the mention of your Lord’s mercy’, with normal t (tāʾ yang
panjang), and ḏakara (ḏakkara) raḥmata rabbika, with tāʾ marbūṭa (tāʾ yang
bentar).7
According to Riddell (1990:33–37), ʿAbd al-Raʾūf al-Singkilī had hardly any
predecessors in exegesis. The fragment of a commentary on Sūrat al-kahf in
the Cambridge manuscript MS Or. Ii.6.45, which probably dates back to around
1600, does not seem to contain much grammatical terminology (Riddell 2014a,
2017:11–13), although it does refer to a few variant readings. According to Riddell
(2017:60f.), most information in this commentary goes back to ʾAbū Muḥam-
mad al-Ḥusayn ibn Masʿūd al-Baġāwī’s (d. between 510/1117 and 516/1122) com-
mentary Maʿālim al-tanzīl.
In the fragment edited by Riddell (2017:127–261), two variant readings are
mentioned. In Q. 18/34, the reading ṯumurun instead of ṯamrun is described
as “with ḍamma on the ṯāʾ and the mīm” (dengan ḍamma ṯāʾ-nya dan mīm-
nya, Riddell 2017:200). In Q. 18/47, the reading wa-tusayyaru l-jibālu ‘the moun-
tains are moved’ instead of nusayyiru l-jibāla ‘We move the mountains’ (Riddell
2017:206) is attributed by the anonymous author to Ibn Kaṯīr and Ibn ʿĀmir.
The interpretation of both readings is introduced by the phrase maknanya ‘its
meaning [is]’.
After the publication of the Tarjumān al-mustafīd it took three hundred
years for a new Malay commentary to appear. But the text of the Tarjumān
7 See Riddell (1990:126); elsewhere (1990:187), he translates these terms as ‘lengthened’ and
‘momentary’, but this is probably incorrect. Presumably, the distinction is between a perma-
nent (‘long’) and a temporary t, the latter being represented by the tāʾ marbūṭa, which may
have been called ‘temporary’ because it is not realized in pausa.
302 versteegh
shows that there was a living tradition of applying Arabic grammatical termi-
nology to the language of the Qurʾān. The study of Arabic texts was an integral
part of the curriculum in the pesantrens, where they were taught with the help
of a kind of ‘translationese’, as it is called by Riddell (2002). In this curriculum,
Arabic technical terms were freely used throughout the Malay translation of
the texts, but always applied to Arabic, never to Malay.
ʿAbd al-Raʾūf’s commentary also shows that in some cases Arabic terms were
replaced by indigenous ones. Malay terms were mainly used for matters con-
nected with reading and writing, which were learnt during the very first stages
of education. This affected, for instance, the choice of the names for the vowel
signs. We have seen above that in the terminology of the Tarjumān, the sukūn
is referred to by a Malay term, mati, while the vowel signs are indicated by
the Arabic terms ḍamma, fatḥa, and kasra. In the traditional teaching of Jawi
script, the vowel signs themselves, too, have received Malay names: (baris)
di depan or hadapan ‘(written) in front’, i.e. ḍamma; (baris) di atas ‘(written)
above’, i.e. fatḥa; and (baris) di bawah ‘(written) under’, i.e. kasra. These Malay
names do not derive from Arabic terminology, but are calques of the Persian
names for the vowel signs, pīš ‘front’, zebar ‘upper side’, zīr ‘under side’ (Her-
bert and Milner 1989:103). The origin of this terminology in primary instruction
in reading and writing is obvious and probably goes back to the early period of
Islamization in Southeast Asia, in which Persian missionaries played an impor-
tant role.
3 Modern Grammars of Malay
The linguistic situation in Southeast Asia bears some resemblance to that in

South Asia. In his sketch of the early linguistic contacts between the Mughals
and the indigenous population in India, Rahman (2011) points out that an
Indian contact language, called Hindi, Hindustani, or Urdu,8 served as an aux-
iliary language in teaching Persian and Arabic (Rahman 2002:210). Conse-
quently, it was not seen as a language in need of a grammatical description,
just like Persian was used in Iran for the explanation of Arabic grammatical
treatises without being the subject of grammatical analysis itself. Interest in
8 It was not until the end of the 19th century that a distinction was made between Urdu and
Hindi. The latter became the exclusive term for a language variety written with Devanagari
script and borrowing loanwords from Sanskrit, which was associated with the Hindu commu-
nity, whereas Urdu, written with nastaʿlīq script and with Perso-Arabic loanwords, became the
marker of identity for the Muslim community (Rahman 2011:25).
the language for its own sake did not emerge until the colonial period, when
British administrators were the first to propagate the language and write gram-
mar books for it (Rahman 2002:206). What was possibly the first Hindustani
grammar, Johan Josua Ketelaar’s (1659–1718) Instructie oft Onderwijsinghe der
Hindoustanse en Persiaanse talen [Instruction or teaching of the Hindustani
and Persian languages] (Utrecht, 1698), primarily aimed to serve the inter-
ests of the colonial authorities, who needed to know the language in order
to communicate with their Indian subjects (Bhatia 1987:43f.). This grammar
did not use the framework of either Sanskrit or Arabic grammar, but came
from an intellectual context dominated by European administrators and lin-
guists.
Just like Hindustani in Mughal India, schools in Southeast Asia employed
Malay as a language of instruction, rather than a subject in the curriculum.
Since students were supposed to know Malay already as a first language or as
a lingua franca, no grammatical instruction was deemed to be necessary. This
changed when the Malaysian archipelago came under the sway of European
colonial powers. Dutch and British administrators and traders needed gram-
matical manuals to learn the language spoken by the inhabitants of their colo-
nial empire. Even though these were familiar with Malay as a lingua franca, they
tended to hold Arabic or, in some cases, Indonesian languages like Javanese or
Sundanese, in higher esteem. For some time, Dutch colonial authorities con-
sidered using the more prestigious Javanese language as the main language of
communication with the indigenous population (Groeneboer 1998), but this
did not work well for all inhabitants of the empire. Eventually, the colonial
authorities switched to Malay and did their best to promote its use. Dutch was
never an option, because the authorities thought it best to bar the indigenous
population from more than minimal knowledge of Dutch, except for a small
elite.
Once the colonial administrators had abandoned their preference for Java-
nese and shifted to Malay, they insisted that a high form of the language should
serve as the language of the Dutch Indies. Maier (1993) attributes this policy
to the authorities’ idée fixe that the manifold manifestations of the language,
from pidginized Malay to the court Malay of Riau, resembled the continuum
between Dutch dialects and Modern Standard Dutch. Accordingly, just as this
was done in the Netherlands, they set out to find the purest form of the lan-
guage in Classical Malay texts and introduced this as the new standard lan-
guage in grammars and exercise books. This form of Malay was deemed to be
essential in dealing with the indigenous population, preferably in a Romanized
spelling in order to sever the link between the Dutch East Indies and Middle
Eastern Islam (Leow 2016:77f.).
304 versteegh
Thus, a novel tradition of Malay grammar books emerged, written by for-

eigners and aiming at foreign users, at first simple conversation manuals, later
scholarly grammars in the tradition of Western linguistics. These works were
written in Dutch and English because the intended audience consisted of
Dutch and British administrators and traders. One of the first grammars was
Frederi(c)k de Houtman’s (1571–1627) Spraeck ende woordboeck inde Maleysche
ende Madagaskarsche talen, met vele Arabische ende Turcsche woorde [Gram-
mar and dictionary of the Malay and Malagasy languages, with many Arabic
and Turkish words] (1603), which used the framework of Latin grammar as the
natural basis for the description of languages (Kaptein 2000).9
In practice, most colonial employees used a simplified form of the language
in their daily dealings with the Indonesians (the so-called Dienstmaleisch ‘Ser-
vice Malay’). Errington (2008) connects their attitude with a general policy of
subduing the natives and interprets all efforts by Europeans to study the lan-
guage accordingly. Rahman (2002:145ff.) presents a more nuanced view and
allows for the possibility that, at least in British India, some administrators
were genuinely fascinated by the indigenous language and its literature. What
Hindustani and Malay had in common is the fact that in either case the colo-
nial authorities were the ones who propagated the use of the language as the
standard, even though it was less prestigious locally than the current cultural
language, Persian and Javanese or Arabic, respectively.
When Malay became the new national language in independent Malaysia
and Indonesia, the Western grammars, especially the Dutch ones, provided a
powerful model for its grammatical description. But this is not the whole story.
Although Malay never developed a full-fledged indigenous tradition of gram-
matical description, at the end of the 19th century, a few scholars did make an
effort to develop a Malay grammatical tradition. They did not need to develop
all of its terminology from scratch, because to some extent, they could draw on
the Malay tradition of instruction in Arabic grammar and Qurʾānic exegesis.
The first Malay-written grammar of Malay was that of Raja Ali Haji (prob-
ably d. 1873), a scholar from the Sultanate of Riau. In his Bustān al-kātibīn
[Garden of writers] (1857), Raja Ali Haji provided a grammatical analysis of
Malay (see Hidayatullah 2012; Mustari 2012).10 His incomplete dictionary Kitab
9 Exceptionally, the Dutch reverend Johannes Roman (d. 1658) states in the introduction to
his sketch of Malay, Grondt ofte Kort Bericht van de Maleische Tale [Brief report on the
Malay language] (1653), that it might have been helpful to find a knowledgeable Ara-
bic consultant for his grammar, but unfortunately he could find no-one. Accordingly,
this grammar, too, is written entirely within the framework of Latin grammar (Kaptein
2000:335).
10 In quotations from the Bustān I use, with slight changes in the transliteration of the Ara-
pengetahuan bahasa [Book of the knowledge of the language], written in the

1850s, also features a large number of Arabic grammatical terms (van der Put-
ten 2002), as it opens with a compendium of Arabic grammar (Pengetahuan
1–27).11
There has been some debate about the intellectual affinities of Raja Ali Haji.
Some people claim that he was a typical Malay court intellectual or an Islamic
scholar, while others maintain that he was a modern writer, influenced by his
contacts with Dutch scholars. Lawrence (2006) analyzes the various views and
concludes that the essential trait of his intellectual activities was that he tai-
lored style and method to his audience. What is certainly novel in Raja Ali
Haji’s approach is that he believed instruction in Malay grammar was worth-
while. The reason for this is given at various places in his work, and it is also
clear from his correspondence with Dutch scholars: he believed in the value
of studying Malay language and culture and transmitting it to younger gen-
erations, who were no longer familiar with it (van der Putten 1995; van der
Putten and Al Azhar 1995). Leow (2016:87f.) stresses the originality of his lex-
icographical work, which stayed clear from both Dutch and Arabic models;
yet, in his grammatical work, he did follow the framework of Arabic gram-
mar.
It is not clear which audiences Raja Ali Haji targeted. He does address
non-Malay speakers wishing to learn the language, but he also seems to be
concerned by the language choice of native speakers. He mentions the fact
that the Malay of Johor is the original variety (Melayu yang asli), and that
people from elsewhere use a lesser quality of Malay, which is mocked by
the people of Johor (Bustān 25; van Ronkel 1901:540f.;). Since the last part of
his treatise deals with the appropriate style to be used in writing letters, his
main purpose may have been to help Malay speakers in improving their style
bic words, the edited and transliterated text from the edition by Hashim bin Musa (2005),
which also contains a facsimile of the Leiden ms. Kl 107. I add the page numbers from van
Ronkel’s (1901) translation. Note that the manuscript used by van Ronkel, Leiden no. 218,
which is identical to the lithograph edition of Pulau Penyengat (Riau) [1851], has one addi-
tional paragraph, about separate writing of words. This paragraph is missing in the other
manuscripts (see van Ronkel 1901:533).
11 The edition by Hashim bin Musa of the Pengetahuan has an introduction containing
eighteen chapters with the following topics: 1 ism; 2 fiʿl; 3. ḥarf maʿnā; 4. lafẓ/kalām; 5.
mubtadaʾ/ḫabar; 6. mafʿūl; 7. fiʿl/fāʿil; 8. mafʿūl bihi; 9. mafʿūl muṭlaq; 10. mafʿūl lahu; 11.
mafʿūl fīhi; 12. mafʿūl maʿahu; 13. ḥāl; 14. tamyīz; 15. taʾkīd; 16. badal; 17. ṣifa; 18. ʾiḍāfa. I
haven’t made a systematic comparison of the Bustān and the Pengetahuan, but the sim-
ilarities are obvious, some of the sections being virtually identical in both treatises. On
Raja Ali Haji’s method in compiling his dictionary see Leow (2016:87 f.).
306 versteegh
and to prevent them from producing faulty written Malay. Accordingly, Raja
Ali Haji states: “It is not my intention to translate an [Arabic] grammatical
work” (bukan maksudku menterjemahkan ʿilmu nahu; Bustān 24; van Ronkel
1901:540). In the introduction to the Bustān, he explains that without knowl-
edge of the rules and definitions of the Malay language people are bound to
make mistakes in writing. They might, for instance, write the wrong letters and
use incorrect words, rendering their writings worthless (Bustān 6; van Ronkel
1901:523).
In writing his treatise, Raja Ali Haji borrowed the structure and rules of Ara-
bic works on ṣarf, naḥw and luġa (Bustān 24; van Ronkel 1901:540.14f.). Since
he had received a traditional Islamic education, including instruction in Ara-
bic grammar, it is understandable that he used the Arabic model to describe the
structure and categories of Malay. Just like Muslim intellectuals anywhere, he
naturally adopted Arabic grammar as representative of the universal structure
of language in general.
Raja Ali Haji faithfully mentions any differences between Arabic and Malay
that crop up, but with the implied corollary that the underlying rules are the
same. In general, his respect for the Arabic model was such that he strove to
apply all Arabic grammatical categories and terminology to Malay, even when
it was clearly not useful to do so. He introduces the term ʾiʿrāb ‘declension’, for
instance, even though Malay nouns do not have declensional endings (Kaptein
2000:335). According to Raja Ali Haji (Bustān 20.14 f.; van Ronkel 1901:535), ʾiʿrāb
consists in “changes in the endings of speech due to the difference in govern-
ing words, overt or implicit” (ma taġayyara ʾawāḫir al-kalām li-ḫtilāf al-ʿawāmil
al-dāḫila ʿalayhā lafẓan ʾaw taqdīran), which is almost identical with the defini-
tion at the beginning of the ʾĀjurrūmiyya (Carter 1981:34) al-ʾiʿrāb taġyīr ʾaḥwāl
ʾawāḫir al-kalim li-ḫtilāf al-ʿawāmil al-dāḫila ʿalayhā lafẓan ʾaw taqdīran. But
then he states that his own grammar is only an extract, so that he will not deal
with all of its rules, thus cleverly avoiding the issue of the relevance of declen-
sional endings for Malay.
Instead, his discussion focuses on the correct spelling of the vowels. In the
heading of the chapter about ʾiʿrāb Raja Ali Haji equates this term with the
vowel signs, using the traditional Malay names, baris di atas /a/, baris di depan
/u/, and baris di bawah/i/(Bustān 20.32f.; van Ronkel 1901:535). He also uses the
term ḥarf yang mati ‘dead letter’ for a vowelless consonant (e.g. Bustān 24.3).12
The main issue discussed by him is that vowels may be indicated either by a
12 An interesting neologism is dihidupkan ‘to be revived’ for vocalizing a ‘dead’ consonant

(Bustān 25.22).
vowel sign, or by a vowel sign combined with one of the ḥurūf al-ʿilla.13 The
default spelling is the latter, but the ḥarf al-ʿilla may be left out if there is a cause
(ʿilla). In §9 of the Bustān (22–25; van Ronkel 1901:537–541) the causes leading
to the omission of the letter are discussed, the most important one being a lack
of ambiguity. The word kepada ‘for’, for instance, is written ⟨kpd⟩ rather than
⟨kpʾdʾ⟩ because this combination cannot be confused with another word.
One striking example of Raja Ali Haji’s reference to Arabic grammar is his
analysis of the expression ‘I ate the fish with/without its head’. This example
only makes sense within the framework of Arabic grammar (Carter 1981:291),
because its ambiguity hinges on the different meanings conveyed by the accu-
sative, genitive, and nominative case marking in the word for ‘head’, as in (1a,
b, c).
(1) a. ʾakaltu l-samak-a ḥattā raʾs-a-hu

eat.1s ART-fish-ACC till head-ACC-3ms
‘I ate the fish, including its head’
b. ʾakaltu l-samaka ḥattā raʾs-i-hi

eat.1s ART-fish-ACC till head-GEN-3ms
‘I ate the fish up to its head’
c. ʾakaltu l-samaka ḥattā raʾs-u-hu

eat.1s ART-fish-ACC till head-NOM-3ms
‘I ate the fish, even its head’
Raja Ali Haji (Bustān 33; van Ronkel 1901:551) does not cite these Arabic sen-
tences, but he gives the Malay sentence in (2).
(2) makan aku akan ikan hingga kepala-nya pun ku-makan juga
eat 1s OBJ fish till head-3ms EMPH 1s-eat also
‘I ate the fish, up to its head it was eaten by me’
According to him, this Malay sentence may imply either that the head is eaten
or not. He then states that in Arabic these different meanings are indicated by
ʾiʿrāb, which does not exist in Malay. In Malay, the ambiguity is lifted by adding
se- to hingga (and pun), as in (3), which implies that the head is eaten together
with the body of the fish.
13 Raja Ali Haji is aware of the double meaning of ʿilla, ‘illness’ or ‘cause’, but he prefers the
latter meaning (Bustān 23; van Ronkel 1901:539).
308 versteegh
(3) makan aku akan ikan sehingga kepala-nya pun sekali aku makan
eat 1s OBJ fish so.that head-3ms EMPH all 1s eat
‘I ate the fish, so that even its entire head was eaten by me’
Arabic influence is also obvious in Raja Ali Haji’s classification of the parts of
speech. He distinguishes three parts of speech, just like the Arabic grammari-
ans do (Bustān 25f.; van Ronkel 1901:541):
The words that are produced have to be one of three things. The first is
in Arabic ism, which means ‘name’, and the second is fiʿl, which means
‘action’, and the third ḥarf. By this we mean here the ḥarf that has a mean-
ing (bermula yang diperbuat perkataannya itu tiadalah sunyi ia daripada
tiga perkara. Pertama, pada bahasa ʿArab ism yakni nama, dan kedua, fiʿl
yakni perbuatan, dan ketiga ḥarf. Maka dikehendak ḥarf di sini ḥarf yang
ada baginya makna).
After the short §10, in which he introduces the three parts of speech, Raja Ali
Haji discusses the details of each part in separate sections (§ 11 ism, Bustān 26–
29; van Ronkel 1901:541–546; §12 fiʿl, Bustān 29–32; van Ronkel 1901:546–549;
§ 13 ḥarf, Bustān 32–38; van Ronkel 1901:549–558). In the rest of his treatise he
uses the Arabic terms.
The distinction between perkataan and kata kata, which is the topic of § 14
(Bustān 38f.; van Ronkel 1901:558) shows clearly that the concepts Raja Ali
Haji introduced into Malay grammar were derived from Arabic grammatical
treatises in the ʾĀjurrūmiyya tradition. In this tradition, the distinctive charac-
teristics of kalām are the following (Carter 1981:8–10): it must be a formal utter-
ance (lafẓ), composite (murakkab), informative (mufīd), and conventional (bi-
l-waḍʿ). This excludes, among other things, speech by someone who is sleeping
and self-evident statements of the type al-samāʾu fawqanā ‘the sky is above us’
and al-ʾarḍu taḥtanā ‘the earth is beneath us’. Two of these criteria are repeated
almost verbatim by Raja Ali Haji when he introduces the concept of perkataan,
which he calls “an utterance that conveys useful information” (lafaz yang mem-
beri faedah, Bustān 38.26f.; van Ronkel 1901:558:5). This excludes the category
of speech by someone who is sleeping and that of self-evident statements. The
author gives the Malay translation of the two Arabic sentences quoted above,
langit di atas kita and bumi di bawah kita, and declares that they fall outside the
definition of speech (perkataan), because they do not convey new information.
It is obvious that Raja Ali Haji regards the structure of Arabic grammar as
a universal scheme, valid for any language. In case there is a discrepancy be-
tween Malay and Arabic structure, he attributes this to the fact that Malay has
less constructions or categories than Arabic. In the case of negations, for in-
stance, Malay lacks the two, three, or even four negations in Arabic (Bustān 35;
van Ronkel 1901:554). The author wishes to make clear, however, that any mean-
ing expressed in Arabic can also be expressed in Malay, although by different
means.
The paragraph about the particles illustrates Raja Ali Haji’s awareness of the
differences between Malay and Arabic. He states that the function of the ḥarf
in Arabic is to indicate a meaning ( fāʾida) in another word, and then mentions
as its first function that of a ḥarf jarr, which is always followed by a noun in
the genitive (majrūr). He then continues (Bustān 32; van Ronkel 1901:549): “In
Malay, no mention is made of its majrūr, but only of its purpose and meaning”
( Jika pada bahasa Melayu tiadalah dibicarakan majrurnya itu melainkan kehen-
daknya dan maknanya jua, adanya). What follows is a list of Malay prepositions
and their use. No attempt is made to equate each Malay preposition with an
Arabic one, but, where applicable, the meaning is exemplified with an Arabic
technical term. Thus, for instance, tetapi ‘but’ is said to be li-l-istidrāk ‘for recti-
fication’, and hai, wai, and weh are called ḥurūf al-munādāt ‘vocative particles’
(Bustān 46f.; van Ronkel 1901:552f.).
The next sections §§15–28 of the Bustān (39–47; van Ronkel 1901:558–568)
deal with syntactic categories. The author’s precepts about syntactic construc-
tions in Malay are sometimes rather confusing because they have been copied
directly from Arabic grammar, for instance, when he states that ‘an indefinite
noun cannot serve as mubtadaʾ’, or that ‘the fāʿil has to precede the mafʿūl’.
Sometimes, he reproduces rules that must have been hard to understand out-
side the context of Arabic grammar, for instance the rule that with a single
predicate no pronoun is expressed. He explains this as follows (Bustān 40; van
Ronkel 1901:559f.):
In a ḫabar mufrad, such as si zayd saudaramu [ART Zayd friend-2S] ‘Zayd

is your friend’, the rule of the ḫabar is not to receive a pronoun, i.e. ia
‘he’. But according to al-Kisāʾī [d. 189/805] and al-Rummānī [d. 384/994],
and a group of others, a pronoun must be understood here, as in si zayd
saudaramu yaʿni ia ‘Zayd is your friend, i.e. he’ (adapun misal ḫabar yang
mufrad itu ‘si zayd saudaramu’ adalah hukum ḫabar ini tiadalah diberi
ḍamīr yakni ‘ia’; akan tetapi pada Syeikh Kisaʾi dan Shyeikh Rumani dan
beberapa jamaʿah, harus ditanggungkan ḍamīr, seperti katamu ‘si zayd
saudaramu yaʿni ia’).
The origin of this rather mystifying explanation is a masʾala discussed by the

Arabic grammarians, about whether a pure noun as ḫabar in sentences like
310 versteegh
ʿamr ġulāmuka ‘ʿAmr is your slave’ contains a pronoun referring to the mub-
tadaʾ. Ibn al-ʾAnbārī (ʾInṣāf 30f.) discusses this masʾala and he states that the
Kufans (i.e. al-Kisāʾī) and al-Rummānī held that this predicate indeed contains
a pronoun, since ġulāmuka is equivalent to ḫādimuka ‘your servant’, with an
active participle that, according to the consensus of the grammarians, does
contain a pronoun.
Sometimes, Raja Ali Haji uses the availability of alternative constructions
in Arabic grammar in order to account for a Malay construction. In a passage
about the mubtadaʾ in Malay, he states (Bustān 41; van Ronkel 1901:561):
Furthermore, it is also forbidden to place the predicate in front, for in-

stance in si zaid telah berdiri ia [ART Zayd PERF stand 3S] ‘Zaid stood up’;
here, telah berdiri is virtually a participle ( fāʿil muqaddar),14 i.e. it con-
tains a hidden pronoun; therefore, it becomes the predicate of si zayd
and may not be put in front. Thus, one should not say telah berdiri si
zayd. If something like telah berdiri si zayd is said, it does not belong to
the chapter of mubtadaʾ ḫabar, but to that of fiʿl with fāʿil (demikian lagi
tertegah pula didahulukan ḫabar seperti umpama ‘si zayd telah berdiri ia’,
maka ‘telah berdiri’ itu fāʿil yang muqaddar yakni mempunyai tersembunyi
pada ia itu, maka iaitu jadi ḫabar kepada ‘si zayd’, tiada harus didahulukan.
Maka jangan dikata ‘telah berdiri si zayd’, dan jika dikata juga seperti itu
yaʿni ‘telah berdiri si zayd’, maka bukanlah pada bab mubtadaʾ khabar ini,
tetapi adalah ia pada bab fiʿl dengan fāʿil)
At first, Raja Ali Haji reasons that the correct word order is the one in (4a),
whereas the one in (4b) is forbidden.
(4) a. si zaid telah berdiri ia

ART Zayd PERF stand 3S
b. telah berdiri si zaid
PERF stand ART Zayd
‘Zayd stood up’
But subsequently, by parsing the two sentences in a different way, he makes

clear that in fact both word orders are allowed in Malay, each with its own
explanation.
14 I am not entirely sure this is the correct interpretation, but it does not make sense to
translate fāʿil muqaddar as ‘virtual agent’, which is why I have retained the translation
‘participle’ that is given by van Ronkel.
While Raja Ali Haji is not the only grammarian to analyze Malay within
the framework of Arabic grammar, he is by far the most consistent one in
this enterprise. A number of later grammarians adopted a tripartite division
of the parts of speech that is somewhat similar to his classification, but using
loan translations of the Arabic grammatical terminology rather than the Ara-
bic terms themselves (Norrudin 2009). The Kitab permulaan pertuturan Melayu
[The book of the beginning of Malay conversation] (1911) by Abdullah bin Abd
Rahman, for instance, has nama, perbuatan, and diwal. In this tripartite divi-
sion, nama, the Sanskrit term for ‘noun’, could be a loan translation of Arabic
ism, while the term perbuatan ‘verb’, from the verb buat ‘to act, do’, is a calque
of the Arabic fiʿl. While these two terms are already used by Raja Ali Haji as
translation of the Arabic terms, the term diwal is a bit of a mystery. It indicates
the third part of speech, ḥarf, but it is not clear what its etymology is. Its lexical
meaning is ‘wall’, and it may be a Persian loanword, but it is unclear how this
term came to be used for the third part of speech. Other treatises have divisions
into more parts of speech, usually based on the framework of Western school
grammar, but with Malay terms.15
4 The Heritage of Arabic Grammar
When Romance and Germanic grammarians shifted from Latin to their ver-
nacular language in describing this vernacular language, they translated almost
their entire grammatical terminology from Latin. Compared to the European
approach, traditions that originated in the Islamic world used a relatively small
number of loan translations. Even when Arabic grammars were composed in
the vernacular language, their authors stuck to the Arabic technical lexicon.
In the Berber literary tradition in the Sous, for instance, while other subjects
were dealt with in elementary Berber treatises for beginners, according to van
den Boogert (1997:96) the only author to deal with (Arabic) grammar briefly
is ʾIbrāhīm ibn ʿAbdallāh al-Ṣanhājī, known as ʾAẓnag (d. 1005/1597). His versi-
fied text, usually referred to as Lmazġiy, contains 148 lines on ʿAqīdat al-naḥw
15 After Raja Ali Haji, the oldest traditional grammar is Pemimpin Johor [Leaders of Johor]
(1878) by Muhammad Ibrahim Munsyi (1840–1904). Later grammars include Kitab punca
bahasa [Elementary language book] (1928) by Abdullah bin Talib; and Jalan Bahasa
Melayu [The way/method of the Malay language] (1937) by Dato’ Hj. Muhammad Said bin
Sulaiman. For these grammars and their classification of the parts of speech see Norrudin
(2009); for the classification of the parts of speech in Malay linguistics see also Kridalak-
sana (1994).
312 versteegh
(van den Boogert 1997:101). Presumably, most of the grammatical terms are
either Arabic or derived from Arabic.16 The Berber language itself does not
seem to have been the subject of publications in Berber, except in the Arabic-
Berber glossaries, which were probably compiled for the benefit of those who
wrote elementary treatises or translated texts from Arabic into Berber (van den
Boogert 1997:88–90, 113–119).
The Persian tradition did not develop a grammatical analysis of the language
as such, but focused on the analysis of rhyme and meter in Persian poetry.
In doing so, Persian scholars did develop an innovative use of the existing
terminology, which consisted almost entirely of Arabic loanwords (Jeremiás
1993; 2000). When these terms were applied to Persian, the different structure
of the language led to considerable shifts in the meaning of the terminology.
When the terms zāʾid and ʾaṣl were applied to the structure of Persian words,
ʾaṣl came to denote the basic form of a word, with a full meaning, whereas
zāʾid was used for all inflectional and derivational morphemes attached to it
(Jeremiás 2000:332). Yet, crucially, the Arabic terms were retained rather than
being replaced with Persian equivalents.17 While the greater part of grammat-
ical terminology in Persian consists of Arabic loanwords, Persian terms are
used for the vowel signs, zebar ‘above’, zīr ‘below’ and pīš ‘in front’, rather than
fatḥa, kasra and ḍamma. We have seen above (p. 301) that presumably through
Persian missionary activity these vowel names also found their way to South-
east Asia. The reason for the introduction of Persian vowel names may have
been that Persian was the language of instruction in most forms of education,
and Arabic treatises were usually taught with the help of Persian translations
(Zadeh 2012). In primary education, where children learned how to write, the
use of Persian terms was bound to be helpful.
16 In modern times, a Kabylian linguist Mouloud Mammeri (1917–1989) introduced a host of

Berber neologisms in his grammar of Kabylian Berber, Tajerrumt n-Tamaziγt (1976). The
model of his grammatical description was that of Arabic grammar, but in the classifica-
tion of the parts of speech, for instance, only one of the three terms is an Arabic loanword,
isem ‘noun’, the other two are Kabylian, amyag ‘verb’ and tazelγa ‘particle’ (Elmedlaoui
2001:388).
17 The grammatical terminology in treatises about Persian remained almost entirely Arabic
until the 19th century, when several Iranian authors published grammatical sketches of
the language (Jeremiás 2003), for instance Īravānī in his Qawāʾid-i fārsiyat (1846) and Kir-
mānī in his Ṣarf wa-naḥw-i fārsī (1859). Pellò (2003a) studied in depth a third work, Mīrzā
Ḥābīb-i Iṣfahānī’s (d. 1897) Dabistān-i pārsī. These authors based themselves mainly on
the Arabic grammatical framework, while adapting it to the characteristics of Persian, but
they also introduced innovatory concepts, like the classification into nine ajzāʾ-i kalām
(Pellò 2003b).
In present-day Pakistan, Urdu has obtained an important status by virtue

of its being employed by both nationalists and Islamic fundamentalists as a
weapon against other vernacular languages, such as Punjabi and Sindhi. The
language was also regarded as a defense against what was seen as the perni-
cious influence of English. In Urdu grammars, the tradition of the early San-
skrit grammarians as well as the Western linguistic tradition were replaced by
an Arabic framework, hence the large number of Arabic loanwords in mod-
ern descriptions of Urdu. The Arabic grammatical terms almost always con-
trast with Sanskrit terms in descriptions of Hindi. Examples from Nardella
(2008) include such basic notions as ‘semantics’ (Urdu ilm-i maʿni, maʿnaviyāt,
Hindi arthatatv; Nardella 2008:1229), and terms for sentence constituents, such
as ‘subject’ (Urdu fāʿil, Hindi kartā; Nardella 2008:185), ‘object’ (Urdu mafʿūl,
Hindi karm; Nardella 2008:1100) and ‘predicate’ (Urdu musnad, ḫabar, Hindi
vidheya; Nardella 2008:1167). For neologisms, Urdu turns to Arabic, for instance
sābiqa ‘prefix’ and lāḥiqa ‘suffix’, where Hindi grammarians borrow from San-
skrit pūrvapratyaya/parpratyaya (Nardella 2008:1169, 1290).
With respect to terminology the Malay tradition is rather exceptional. As
a vernacular language, its primary role was that of an auxiliary language in
teaching the grammar of the dominant religious language. But at an early stage,
Malay scholars introduced Malay terms in their discussion of variant readings.
Malay terms must have been current in primary education as well, as the names
of the vowel signs show.
In modern Malay grammars, none of the traditional classifications were
taken over, although they retained some of the traditional terms, while gener-
ally adhering to the system of Western grammar. In their standard grammar of
modern Indonesian, Tata bahasa baku Bahasa Indonesia [New grammar of the
Indonesian language], Moeliono and Dardjowidjojo (1988) have retained only a
few traces of either the Dutch-influenced terminology or the Arabic-influenced
terminology of Raja Ali Haji. Their main source for grammatical terminology is
modern Western linguistics.
In the classification of parts of speech, for instance, they distinguish be-
tween verba, nomina, adjektiva, adverbia, and a general category of ‘function
words’ (kata tugas), which contains among other word classes preposisi, kon-
jungsi and partikel (1988:30). This division goes back to the system of Greco-
Latin school grammar, but the older terminology shines through in some of
the synonyms mentioned for the modern terms. For verba, nomina, adjektiva
the alternative names mentioned by the Tata bahasa baku Bahasa Indonesia
are a mix of loan translations from Arabic and Dutch. One alternative term for
‘verb’, for instance, is kata kerja lit. ‘word of work’, which is an obvious calque of
the Dutch werkwoord, while kata benda lit. ‘word of object’ translates Latin sub-
314 versteegh
stantivum ‘substantive’, and kata sifat lit. ‘word of property’ contains the Arabic
ṣifa ‘property; adjective’. The alternatives kata depan lit. ‘word in front’, and kata
sambung lit. ‘word of joining’ for preposisi and konjungsi are loan translations of
the Latin terms praepositio and coniunctio, possibly through the intermediary
of Dutch voorzetsel and voegwoord, while the form of preposisi and konjungsi
reflects the Dutch pronunciation of the Latin terms. Note that in the terminol-
ogy for sentence constituents Latin terms prevail as well in this grammar, for
instance predikat, subjek, and objek.
Only a few etymologically Arabic terms have been preserved in modern lin-
guistic terminology. The term for adjective, sifat, was mentioned above, but
there are a few more (Ogloblin 1981:221), e.g. huruf ‘letter’ (< ḥurūf, plural of
ḥarf ), istilah ‘term’ (< iṣṭilāḥ), makna ‘meaning’ (< maʿnā), kalimat ‘word’ (<
kalima), jamak ‘plural’ (< jamʿ). It should be added, however, that these terms
are not used generally, although they are mentioned in the dictionaries. Thus,
for instance, for ‘plural’, the Tata bahasa baku mentions both jamak and pluralis
as alternatives. Probably, the Arabic terms are more current in pesantren train-
ing, in connection with the teaching of Arabic grammar. For ‘prefix’ and ‘suffix’
the authors introduce neologisms, based on Arabic words: awalan ‘prefix’ (<
ʾawwal ‘first’), akhiran ‘suffix’ (< ʾāḫir ‘last’), along with the modern linguistic
terms prefiks and sufiks (Moeliono and Dardjowidjojo 1988:27).
Writing Malay grammars remained for some time the exclusive realm of
colonial administrators, who introduced their own model in the grammatical
description and analysis of the language. A great deal of grammatical study
devoted to Arabic was carried out in Malay. When Malay writers started to write
about their own language, this was at least in part a form of emancipation from
the dominance of the language of the colonial power. The need for grammat-
ical instruction in the auxiliary language may also have had something to do
with the presence of non-native speakers, who were insufficiently familiar with
the auxiliary language. In fact, the number of non-native speakers of Malay far
exceeded that of native speakers. Raja Ali Haji’s initiative must be regarded as a
quite exceptional enterprise, not only for Malay, but for most languages in the
Islamic world. This explains, no doubt, why it was a short-lived experiment.
While Raja Ali Haji is still held in high esteem in Malaysia and Indonesia, his
work did not exert any lasting influence on the study of Malay grammar.
A Primary Sources
ʿAbd al-Raʾūf, Tarjumān = ʿAbd al-Raʾūf al-Singkilī, Tarjumān al-mustafīd. Partial ed. in
Riddell (1990) and (2017).
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
Gotthold Weil. Leiden: E.J. Brill, 1913.
Raja Ali Haji, Bustān = Raja Ali Haji, Bustān al-kātibīn li-l-ṣibyān al-mutaʿallimīn. Litho-
graph, Pulau Penyengat (Riau), [1851]. Repr., Singapore, 1310/1892/ Ed. (in transcrip-
tion) by Hashim bin Musa. Kuala Lumpur: Yayasan Karyawan, 2005/Transl. into
Dutch by van Ronkel (1901).
Raja Ali Haji, Pengetahuan = Raja Ali Haji, Kitab pengetahuan bahasa. Ed. (in transcrip-
tion) by Hashim bin Musa. Kuala Lumpur: Yayasan Karyawan, 2010.
Sībawayhi, Kitāb = ʾAbū Bišr ʿAmr ibn ʿUṯmān Sībawayhi, al-Kitāb. Ed. by ʿAbd al-Salām
Muḥammad Hārūn. 5 vols. Cairo: vol. I Dār al-Qalam; vol. II Dār al-Kātib al-ʿArabī
li-l-Ṭibāʿa wa-l-Našr; vols. III–V al-Hayʾa al-Miṣriyya al-ʿĀmma li-l-Kitāb, 1966–1977.
B Secondary Sources
Al-Nassir, ʿAbd al-Munʿim ʿAbd al-ʾAmīr. 1993. Sibawayh the phonologist: A critical study
of the phonetic and phonological theory of Sibawayh as presented in his treatise Al-
Kitab. London and New York: Kegan Paul International.
Arps, Bernard. 1997. “Koning Salomo en het dwerghertje: Taalpolitiek, taalonderwijs
en de eerste grammatica’s van het Javaans [King Solomo and the mouse-deer:
Language policy, language education and the first grammars of Javanese]”. Kolo-
niale taalpolitiek in Oost en West: Nederlands-Indië, Suriname, Nederlandse Antillen,
Aruba [Colonial language policy in East and West: Dutch Indies, Suriname, Dutch
Antilles, Aruba], ed. by Kees Groeneboer, 85–105. Amsterdam: Amsterdam Univer-
sity Press.
Basri, M.A. Fawzi Mohd. 1981. “Kitab Pemimpin Johor: Suatu pengenalan [The Kitab
Pemimpin Johor: An introduction]”. Jurnal Persatuan Sejarah Malaysia 10.47–57.
Berg, Lodewijk Willem Christiaan van den. 1886. “Het Mohammedaansche godsdien-
stonderwijs op Java en Madoera en de daarbij gebruikte Arabische boeken [Moham-
medan religious education in Java and Madura and the Arabic books used in it]”.
Tijdschrift voor Indische Taal-, Land- en Volkenkunde 31.518–555.
Bhatia, Tej K. 1987. A history of the Hindi grammatical tradition: Hindi-Hindustani gram-
mar, grammarians, history and problems. Leiden: E.J. Brill.
Boogert, Nico van den. 1997. The Berber literary tradition of the Sous with an edition and
translation of The ocean of tears by Muhammad Awzal (d. 1749). Leiden: Nederlands
Instituut voor het Nabije Oosten.
316 versteegh
Bruinessen, Martin van. 1990. “Kitab kuning: Books in Arabic script used in the pesan-
tren milieu. Comments on a new collection in the KITLV library”. Bijdragen tot de
Taal-, Land- en Volkenkunde 146.226–269.
Bruinessen, Martin van. 1994. “Pesantren and kitab kuning: Continuity and change in
a tradition of religious learning”. Texts from the islands: Oral and written traditions
of Indonesia and the Malay world, ed. by Wolfgang Marschall, 121–146. Berne: The
University of Berne Institute of Ethnology.
Bruinessen, Martin van. 2008. “Traditionalist and Islamist pesantrens in contemporary
Indonesia”. The madrasa in Asia: Political activism and transnational linkages, ed. by
Farish A. Noor, Yoginder Sikand, and Martin van Bruinessen, 217–245. Amsterdam:
Amsterdam University Press.
Collins, James T. 1996. Malay, world language of the ages: A sketch of its history. Kuala
Lumpur: Dewan Bahasa dan Pustaka.
Drewes, Gerard W.J. 1971. “The study of Arabic grammar in Indonesia”. Acta orientalia
neerlandica: Proceedings of the Congress of the Dutch Oriental Society, ed. by Pieter
Willem Pestman, 61–70. Leiden: E.J. Brill.
Elmedlaoui, Mohamed. 2001. “A cross-cultural reading in a Kabyle Berber grammar
handbook (Mammeri’s Tajerrumt)”. Indigenous grammar across cultures, ed. by
Hannes Kniffka, 379–401. Frankfurt a. Main: P. Lang.
Errington, J. Joseph. 2008. Linguistics in a colonial world: A story of language, meaning,
and power. Oxford: Blackwell.
Groeneboer, Kees. 1998. Gateway to the West: The Dutch language in colonial Indonesia
1600–1950. A history of language policy. Amsterdam: Amsterdam University Press.
Herbert, Patricia and Anthony Crothers Milner. 1989. South East Asia languages and
literatures: A select guide. Honolulu: University of Hawai’i Press.
Hidayatullah, Moch. Syarif. 2012. “Bustān al-kātibīn: Pengaruh tata bahasa Arab dalam
tata bahasa Melayu [Bustān al-kātibīn: The influence of Arabic grammar in Malay
grammar]”. Manuskripta 2:1.53–77. [Also appeared as “Bustān al-kātibīn: Kitab tata
bahasa Melayu pertama karya anak negeri (Bustān al-kātibīn: The first Malay gram-
mar by an indigenous author)”, Thaqāfiyyāt 13:1 (2012) 1–34.]
Jeremiás, Éva M. 1993. “Tradition and innovation in the native grammatical literature
in Persian”. Histoire Epistémologie Langage 15:2.51–68.
Jeremiás, Éva M. 2000. “Arabic influence on Persian linguistics”. Handbuch für die
Geschichte der Sprach- und Kommunikationswissenschaft, ed. by Sylvain Auroux,
Konrad Koerner, Hans-Josef Niederehe, and Kees Versteegh, I, 329–333. Berlin and
New York: Mouton de Gruyter.
Kaptein, Nico. 2000. “Arabic influence on Malay linguistics”. Handbuch für die Ge-
schichte der Sprach- und Kommunikationswissenschaft, ed. by Sylvain Auroux, Kon-
rad Koerner, Hans-Josef Niederehe, and Kees Versteegh, I, 333–336. Berlin and New
York: Mouton de Gruyter.
Kridalaksana, Harimurti. 1991. “Bustanulkatibin dan Kitab pengetahuan bahasa: Sum-
bangan Raja Ali Haji dalam ilmu Bahasa Melayu [The Bustanulkatibin and the Kitab
pengetahuan bahasa: Raja Ali Haji’s contribution to Malay linguistics]”. Masa lam-
pau Bahasa Indonesia: Sebuah bunga rampai, ed. by Harimurti Kridalaksana, 349–
361. Yogyakarta: Kanisius.
Kridalaksana, Harimurti. 1994. Kelas kata dalam Bahasa Indonesia [Parts of speech in
Indonesian]. 2nd ed. Jakarta: Gramedia Pustaka Utama.
Lawrence, Kelvin. 2006. All things to all men: Reimagining the intellectual life of Raja Ali
Haji of Riau. M.A. thesis, National University of Singapore.
Leow, Rachel. 2016. Taming Babel: Language in the making of Malaysia. Cambridge:
Cambridge University Press.
Maier, Hendrik M.J. 1993. “From heteroglossia to polyglossia: The creation of Malay and
Dutch in the Indies”. Indonesia 56.37–66.
Makram, ʿAbd al-ʿĀl Sālim Makram and ʾAḥmad Muḫtār ʿUmar. 1984. Muʿjam al-qirāʾāt
al-Qurʾāniyya maʿa muqaddima fī l-qirāʾāt wa-ʾašhar al-qurrāʾ. 8 vols. Kuwait: Jāmiʿa
Kuwayt.
Moeliono, Anton M. and Soenjono Dardjowidjojo (eds.). 1988. Tata bahasa baku Bahasa
Indonesia [A standard grammar of Indonesian]. Jakarta: Departemen Pendidikan
dan Kebudayaan Republik Indonesia.
Mustari. 1998. “Raja Ali Haji dan pemikiran kebahasaannya: Studi terhadap Kitāb bus-
tān al-kātibīn li as-subyān al-mutaʾallimīn [Raja Ali Haji and his linguistic thinking: A
study of the Kitāb bustān al-kātibīn li as-subyān al-mutaʾallimīn]”. Al-Jāmiʿah 61.181–
198.
Nardella, Umberto. 2008. Glossary of Hindi/Urdu and English linguistic terminology. 2
vols. New Delhi: Star Publications.
Norrudin, Muhammad. 2009. Perbandingan tatabahasa Melayu [Comparing Malay
grammars]. Kuala Lumpur: Universiti Terbuka Malaysia, Fakulti Pendidikan dan
Bahasa.
Ogloblin, Aleksandr Konstantinovič. 1981. “Tradicionnoe jazykoznanie v Indonezii i
Malajzii [Traditional linguistics in Indonesia and Malaysia]”. Istorija lingvističeskych
učenij: Srednevekovyj vostok [History of linguistics: The Medieval East], ed. by Agnija
Vasil’evna Desnickaja and Solomon Davidovič Kacnel’son, 210–223. Leningrad:
Nauka.
Pellò, Stefano. 2003a. Mīrzā Ḥabīb-i Iṣfahānī, Dabistān-i pārsī: Una grammatica per-
siana del XIX secolo. Venice: Cafoscarina.
Pellò, Stefano. 2003b. La teoria della qāfiya nel Mīzān al-afkār di Muḥammad Saʿd Allāh-
i Murādābādī. Venice: Cafoscarina.
Putten, Jan van der. 1995. “Taalvorsers en hun informanten in Indië in de 19de eeuw:
318 versteegh
Von de Wall als politiek agent in Riau? [Language experts and their informants in
the Indies in the 19th century: Von de Wall as political agent in Riau?]”. Bijdragen tot
de Taal-, Land- en Volkenkunde 151.44–75.
Putten, Jan van der. 2002. “On sex, drugs and good manners: Raja Ali Haji as lexicogra-
pher”. Journal of Southeast Asian Studies 33.415–430.
Putten, Jan van der and Ali Al Azhar (eds.). 1995. Di dalam berkekalan persahabatan ‘In
everlasting friendship’: Letters from Raja Ali Haji. Leiden E.J. Brill.
Rahman, Tariq. 2002. Language, ideology and power: Language-learning among the
Muslims of Pakistan and North India. Oxford and Karachi: Oxford University Press.
Rahman, Tariq. 2011. From Hindi to Urdu: A social and political history. Oxford and
Karachi: Oxford University Press.
Riddell, Peter G. 1990. Transferring a tradition: ʿAbd al-Raʾūf Al-Singkilī’s rendering into
Malay of the Jalālayn commentary. Berkeley, Cal.: Centers for South and Southeast
Asian Studies.
Riddell, Peter G. 2002. “Literal translation, sacred scripture and Kitab Malay”. Studie
Islamika 9:1.1–26.
Riddell, Peter G. 2014a. “Camb. MS Or. Ii.6.45: The oldest surviving Qurʾanic commen-
tary from Southeast Asia”. Journal of Qurʾanic Studies 16:1.120–139.
Riddell, Peter G. 2014b. “Study of the variant readings of the Qurʾan in 17th century
Aceh, with particular reference to Suura al-Kahf ”. Cetusan minda sarjana: Sastera
dan budaya, ed. by Ampuan Haji Brahim bin Ampuan Haji Tengah, 53–69. Bandar
Seri Begawan: Dewan Bahasa dan Pustaka Brunei.
Riddell, Peter G. 2017. Malay court religion, culture and language: Interpreting the Qurʾān
in 17th century Aceh. Leiden: E.J. Brill.
Roman, André. 1983. Etude de la phonologie et de la morphologie de la koine arabe. 2 vols.
Aix-en-Provence: Université de Provence.
Ronkel, Philippus Samuel van. 1901. “De Maleische schriftleer en spraakkunst getiteld
Boestānoe’l kātibīna [The Malay primer of script and grammar, entitled Boestānoe’l
kātibīna]”. Tijdschrift voor Indische Taal-, Land- en Volkenkunde 44.512–581.
Versteegh, Kees. 2003. “The Arabic component of the Indonesian lexicon”. Rintisan
kajian leksikologi dan leksikografi, ed. by Lilie Suratminto and Munawar Holil, 216–
229. Jakarta: Fakultas Ilmu Pengetahuan Budaya Universitas Indonesia.
Versteegh, Kees. 2018. “Learning Arabic in the Islamic world”. The foundations of Arabic
linguistics. III. The development of a tradition: Continuity and change, ed. by Georgine
Viain, Marie. 2014. La taxinomie des traités de grammaire arabe médiévaux (IVe/Xe–
VIIIe/XIVe siècle), entre représentation de l’articulation conceptuelle de la théorie
et visée pratique: Enjeux théoriques, polémiques et pédagogiques des modélisations
formelles et sémantiques du marquage casuel. Thèse de doctorat, Université Paris-
3.
Index
ʾa- 301 affected object 199

ʿAbd al-Masīḥ, George 78–81 affective object 199, 205
ʿAbduh, Dāwūd 237 Afšār, Ṣadrī 183
Abdullah bin Abd Rahman 311 agent 198, 204, 286
Abdullah bin Talib 311 virtual 310
absolute object 4f. ʾaḫaff 47, 113
abstract representation 182 see ḫafīf
abstract structure 24 ʾaḫaṣṣ 247 f.
abstraction 19f., 23–26 ʾaḥdaṯa 208
ʾAbū ʿAmr 28, 298 ʾAḫfaš, ʾAbū l-Ḥasan Saʿīd ibn Masʿada al-
ʾAbū Ḥayyān 69, 94–96, 102, 200, 278 62, 134, 238 f.
ʾAbū Muslim 61 ʾaḫḫara 108
ʾAbū ʿUbayd 61 ʾāḫir 246
ʾAbū ʿUbayda 64, 73 ʾaḥkām see ḥukm
accusative 40, 113, 207, 222 ʾahl al-ṣināʿa 71
adverbial 225–232 ʾAḥmad ibn Naṣr, ʾAbū l-Ḥasan 134
locative 198 ʾAḥmad, Kabīr al-Dīn 134–136
of condition 225–232 Ahmar, May 106 f.
of state 225–232 ʿāʾid 71, 246
temporal 198 ʾaʿjamī 12
see naṣb ʿAjlānī, al- 63
accusative nominal 205 ʾajnabī 287
acrophony principle 125 ʾĀjurrūmiyya see Ibn ʾĀjurrūm
action 198, 202, 205, 207f., 227, 229 ʾakalūnī l-barāġīṯ 47
activated utterance 250f. akhiran 314
active voice 18 aksara 297
ʾāḏana 55 al- 233, 235, 237 f., 240, 245, 247–249, 254,
asymmetry, between first/second person 285
56 al- relative 243 f., 246
addressee 31f., 42f., 51f., 56, 241, 250f., 283f., ʿalam 260
291 ʿalāma 33 f., 38, 41, 45, 49, 57
see listener ʿalāma li-l-ʾiḍmār 46
adjectival phrase 285–289 ʿalāma li-l-jamʿ 45, 47
adjective 247, 257–260, 266, 276, 285, 287 ʿalāma muḫtaṣṣa 45
annexation of 286f. ʿalāma ẓāhira 47 f., 54
attributive 256, 263 ʿalāmat al-ġāʾib 49
in Persian 187f. ʿalāmat al-ʾiḍmār 36 f.
adjunction 21 ʿalāmat al-jamʿ 45
ʾaḍmara 37f., 43, 45, 50f. ʿalāmat al-mutamakkin 45
ʿAḍud al-Dawla 65 ʿalāmat al-nudba 45
adverb ʿalāmat al-taʾkīd 45
of place 214, 218, 220f., 243, 245 ʿalāmat al-taʾnīṯ 45, 47
of time 143, 213f., 218, 220f., 234, 245 ʿalāmat ʾiḍmār 46–48, 54 f.
ʾafāda 259, 262, 264f. ʿalāmat muḍmar 37, 41, 46, 48, 54
ʾafʿāl al-qulūb 91 ʿalāmat tamām al-kalām 286
ʾafʿāl see fiʿl Al-Ani, Salman 15
320 index
ʾAlfiyya see Ibn Malik apocopate 158, 176

ʾalḥaqa 242 apodosis 76, 79f., 168, 171, 174, 277–279
Aliane, Hassina 2, 8, 10–29 negative 282, 284
ʾalif 299 of law 281
ʾalif al-istifhām 109 of oath 278–281, 283
allaḏī 279f. apposition 249, 261
al-ʾiḫbār bi- 289–291 explanatory 255, 259, 261, 263, 266 f.
Al-Munajjid, Ṣ. 133 coordinated 265
Al-Nassir, Abd al-Munim 301 Arab, as ethnicity marker 7
alphabet 124f., 275 Arabic
teaching of 125 acquisition of 295
alphabetic arrangement 125–128 as linguistic marker 7
ʾAltōnjī, Muḥammad 78–81 teaching of 295
ʿamal 37, 40, 44, 89, 92f., 100f., 215, 222 ʿarabiyya 12
ʾaʿmala 101 Aramaic 89
ʾamāta l-fiʿl 42 Aristotle 22, 172
ambiguity 170, 259, 289–291 Arps, Bernard 297
in Jawi script 307 arrow 22, 26
Ambrosiana manuscript 2, 133–156 arthatatv 313
ʿāmil 5, 91, 93f., 97f., 116, 119, 208, 217f., 291, arti 298
306 article, definite 246, 254, 285 f.
ʿamila 89, 102 ʾAʿšā, al- 42
ʿamila fī 5, 102 ʾAsfarāyīnī, al- 67
ʾamkan 47 ʿĀṣī, Mišāl al- 78–81, 255, 261
ʾammā 71 ʿAskarī, ʾAbū Hilāl al- 74
ʾamr 72, 109f., 175, 225f. ʾaṣl 6, 14 f., 17 f., 20 f., 113, 116, 121, 123, 186 f.,
amyag 312 190, 193 f., 212–214, 216 f., 287, 312
ʾan 244, 280f., 285, 291f. riʿāyat al- 291
elision of 61 ʾasmāʾ al-sitta, al- see six nouns
ʾan al-muwaṭṭiʾa 279–282 ʾasmāʾ see ism
analogical extension 16, 228 ʾAsmar, Jirjis 78–81
analogy 23, 116, 228f. ʾAsmar, Rājī 78–81
anaphora 30, 242 ʾAstarābāḏī, al- 3, 227f., 230, 260, 264, 276–
anaphoric use 235, 240 278, 281–291
animal, domestic 241 ʿaṭf bayān 255, 261, 265 f.
animate 242 ʾaṯqal 113
ʾanna 82, 244 attribute 285
annexation 254, 256f., 263–268, 274f., 285f., attributive phrase, in Persian 193
288, 291 Aussant, Émilie 6
formal 264 autonomy, of realization 15
semantic 264 ʾaw, with subjunctive 61
anomalous structure 288 awalan 314
antecedent 52, 93, 163, 167, 171, 176, 178, 248 ʾawṣala 220
explicit 167 ʾawwal 113 f.
implicit 167, 176 ʾawwal al-ʿadad 114
of relative 235f. axiom 24
redundant 177 ʾayna 114, 177 f.
anteposing 116–119 ʿAynī, al- 67
see fronting Ayoub, Georgine 1, 4, 30–60
index 321
ʾayyun 168, 177f., 243f. Blachère, Régis 171

ʾaẓhara 40, 44 blame 260
ʾAzharī, ʾAbū Manṣūr Muḥammad ibn Bloomfield, Leonard 12 f.
ʾAḥmad al- 77, 126 Boethius 165
ʾAzharī, Ḫālid ibn ʿAbdallāh al- 66, 298 Bohas, Georges 12, 106 f.
ʾAẓnag, ʾIbrāhīm ibn ʿAbdallāh al-Ṣanhājī Boogert, Nico van den 311 f.
311 bound pronoun 220, 286, 289–291
Bruinessen, Martin van 298
Baalbaki, Ramzi 2f., 14, 16, 19, 61–88, 106f., Brunschvig, Robert 100
113, 125f., 182, 195, 199, 212f., 217, 281, Brustad, Kristen 255
286, 288f. burhān 176
bāb 14–17, 19–21, 23–25
bāb al-ʾiʿmāl 101 Cachia, Pierre 281
badal 34f., 249, 266, 305 Carter, Michael 5, 11 f., 18–20, 38, 62, 89–
Badīʿ Yaʿqūb, ʾImīl see Yaʿqūb, ʾImīl Badīʿ 105, 136, 179, 182, 203, 225, 255 f.,
135, 255, 261 266
Baġāwī, ʾAbū Muḥammad al-Ḥusayn ibn case 4, 34
Masʿūd al- 301 case marker 275, 283
Baġdādī, al- 280–283, 289 Caspari, Carl Paul 255
bahasa 298 category 22–24, 26
bahuvrīhi compound 187 definition of 22
Bakkāʾ, Muḥammad Kāẓim al- 134, 136 category construction 23f.
Bakrī, Muḥammad ibn ʾAbī Ġassān al- 63 Category Theory 8, 22–26
bal 80 cause 120
banā ʿalā 39f. center of attention 250
baris di atas 302, 306 Chatti, Saloua 173
baris di bawah 302, 306 circumstantial accusative 68, 76, 78
baris di depan 302, 306 circumstantial qualifier 204
baris hadapan 302 clarification 256, 258–260, 267, 283
Basrans 45, 80, 101, 237, 286 class 19f.
Baṣriyyūn 65 class membership 19f.
Baṭalyawsī, al- 228 clause, qualifying 256
bayān 108, 205, 266 clitic 36
Bayhaqī 183 clitic pronoun, in Persian 191
bayyana 205 cliticization 34, 56
Bedouin informants 102f. double 52 f.
Bedouin speech 14, 212, 215f., 219, 222, 228f., clitics 46
244 coalescence 19, 22, 25
bentar 301 cognate object 3
Benveniste, Émile 4, 30f. see absolute object
beralif 299 cognitive linguistics 4
Berber 296, 312 cognitive verb 283
Kabylian 312 co-limit 23, 25
Berg, L.W.C. van den 297 colonial administration
Bernards, Monique 225, 276 British 303
bināʾ 50, 123 Dutch 303
binding problem 23 communication, successful 110
bi-transitive 117 communicative purpose 115
see ditransitive completion 265–267
322 index
complexity 22 Ḍarīr, Muḥammad ibn Saʿdān al- see Ibn

compound 187f., 190f. Saʿdān
in Persian 6, 182f., 192–195 ḍarūra 278, 288
concept 24 ḏāt 202
conciseness 71, 83, 121 Dāʾūd Rūmī 298
concrete universal 23 Dayyeh, Hanadi 3, 106–122, 212 f.
conditional 280f. declension 306
conditional logic 6, 158f. deep level 18
conditional particle 6, 79, 157f., 164, 167f., deep structure 187
170 defective noun 75
conditional sentence 76, 91, 159, 166f., 170f., definite 114, 268
175, 179, 278f. definite article 246 254, 285 f.
conditionality 6, 157–181 definite expression 254, 258 f., 261, 263–265,
conjunction 164f., 235, 244 267
consequent 167, 178f. definiteness 3 f., 107, 113, 233–252, 286
consequent, implied 177 see definition
consonant 185 definiteness hierarchy see hierarchy, of defi-
as case markers 275, 291 niteness
constituent 16–18 definition 253–272
redundant 284 deictic 54, 234f., 243 f., 251
context 235, 237, 242, 250 deixis 235, 240
pragmatic 53 deletion 21, 189 f., 300
conversation manual 304 in Persian 191
Coptic 296 demonstrative 54, 233–238, 240, 242–245,
copulative phrase, in Persian 193 247–249, 251
co-referential 286f., 291 demonstrative pronoun 3 f., 171
correctness 217 Derenbourg, Hartwig 135, 139
correspondence 178f. description 260
between elements 19 determinate 261, 268
of structure 20, 24 determination 4, 253–272
of transformations 21 determinative phrase, in Persian 193
corroboration 260 determiner 254
counterfactual 158, 162 Devanagari script 302
curriculum, of grammar 61 Dévényi, Kinga 157, 164
creation, of language 278 deviation
from ordinary speech 215, 218
Daḥlān, ʾAḥmad Zaynī 298 from the basic structure 275
dalāla 266 Dichy, Joseph 5, 123–132
dalla ʾalā 276 dictionary 5, 123–132
Damāḏ see Rufayʿ ibn Salama Damāḏ Arabic-Farsi 127
ḍamīr 4, 30–60, 309 digital 129
ḍamīr ʿāʾid 246 French 129
ḍamīr lāʾiq 243 modern 129
ḏamm 257, 260 Persian 183
ḍamma 299, 301f., 312 semasiological 123
ḍarb 99 dictionary arrangement 123–132
Dardjowidjojo, Soenjono 313f. Diem, Werner 285
Ḍarīr, al- 228 Dienstmaleisch 304
Ḍarīr, Hišām ibn Muʿāwiya al- 63 Dieterici, Friedrich 96, 102, 183f.
index 323
diminutive equivalence class 20 f.

of ism mubham 237f. equivocity 258, 260 f.
of relative 239 Errington, Joseph 304
diphthong, in Persian 185 error
direct object 214, 286 in terminology 77 f.
direct speech 91 linguistic 74 f., 77
discourse function 30, 32 of ʾiʿrāb 72
discourse, reality of 34 spelling 306
distinction 261 Euclid 176
distributional analysis 8, 12–14, 16, 20, 25 event 164
distributionalism 26 evolution, of theory 7
ditransitive verb 56 exegesis
diwal 311 in Malay 298
ḍ-m-r 35 of the Qurʾān 28, 38, 300–303
domain, of quantifier 167f. existential import 172, 173
Donatus 102 explanatory apposition 255, 259, 261, 263,
doubly transitive verb 289–291 266 f.
doubly-transitive see ditransitive explication 258
Drewes, Gerard 298 explicit pronoun 36, 44
Druel, Jean 2, 133–156 expression
dual 275 definite 254, 258 f., 261, 263–265, 267
ḏū ṭāʾiyya 243 indefinite 254, 258–261, 263–265, 267
Dutch 303, 313 extension
Dutch Indies 297, 303 analogical 16, 228 f.
dwilingga 298 of category 204 f.
of construction 213, 218
economy 15, 17 of qualification 255
education of speech 213, 216, 219
in Persian 312 of technical terms 203 f.
Islamic 182, 295–297, 305f.
primary 312f. fa- 70, 164 f., 169, 278
effected object 199 conditional 79
effective object 5, 198–211 omission of 278
Ehresmann, Andrée 22 with subjunctive 61
Eilenberg, Samuel 22 faedah 308
elision 43 fāʾida 16, 258, 260 f., 282, 284, 309
of verb 73, 92 fāʾida lafẓiyya 284
Ellermann, David 23 fāʾida maʿnawiyya 284
ellipsis 39, 43, 258 fāʿil 32, 35, 55, 71, 107 f., 117, 198 f., 201 f., 204,
Elmedlaoui, Mohamed 312 208, 226, 288, 305, 309, 313
elucidation 267 fāʿil muḍmar 40
emergence problem 23 fāʿil muqaddar 310
emphasis 82, 283f. fāʿil, nāʾib ʿan 70
endophora 242 familiar 250
English 313 fann 100
enunciation 31f., 34, 53 farʿ 14f., 17, 20 f.
enunciative hypothesis 106f. Fārābī, al- 6, 157, 169
epistemological arrangement 128 Farḥāt, Jarmānūs 96–98, 101
equational sentence 90 Fārisī, al- 65, 205, 227, 229f., 253
324 index
Farrāʾ, al- 39, 64, 99f. geometrical proof 176

farraġa 96 Gille, Christiane 94, 102
faṣl 34 Giolfo, Manuela 1–9, 157–181, 244
fatḥa 299, 302, 312 Givenness Hierarchy 4
fayida 193 glossaries, Berber-Arabic 312
Fazārī, al- 66 glosses 134
feminine 113 Arabic in Latin manuscripts 102
fī 99 glottal stop 300
figurative sense 212 Goguyer, Antoine 95, 102, 261
fiʿl 18, 32, 35, 202f., 305, 308, 311 Goldblatt, Robert 23f.
fiʿl al-muḫāṭab 160 Goldenberg, Gideon 202
fiʿl al-šarṭ 6 government 5, 89, 102, 285
fiʿl māḍī 71 disrupted 283
fiʿl majhūl 83 of adjective 285
fiʿl maʿlūm 83 of participle 285
fiʿl muḍāriʿ 82 grammar
fiʿl muḍmar 39, 42, 160 Berber 312 f.
fiʿl muẓhar 39 Christian 102
fiʿl šakk wa-yaqīn 118 European 30
f-ʿ-l 23, 27 for beginners 61–88
Fleisch, Henri 255 Germanic 311
focus 250f. Greco-Latin 313
formalism 26f. Greek 30
freedom, of the speaker 3, 25, 111, 119 Hebrew 8
Friedmann, Yohanan 220 Hindi 313
function words 313 Hindustani 303
functor 22 Indian 30
furūʿ see farʿ Indonesian 313
fuṣḥā 14 Javanese 297
Latin 296, 304, 311
Ġaḍḍāb, Faraj Muḥammad al- 285 missionary 296
ġāʾib 4, 30–34, 37, 51, 53, 56f., 191 pedagogical 61–88
ʿalāmat al- 49 Persian 6, 8, 182–197, 302, 312
ġarīb 125 Romance 311
ġarīb al-Ḥadīṯ 77 Sanskrit 303
Gätje, Helmut 254, 258 Syriac 8
Gaudefroy-Demombynes, Maurice 171 Urdu 313
ġayr 239f., 248f. Western 304, 311
ġayr qaḍiyya 172 Greek grammar 30
ġayr wājib 32, 10 Groeneboer, Kees 303
geminated verb 142–150 group structure 21
Generative Linguistics 41, 287 Guillaume, Jean-Patrick 6, 106 f.
generative structure 20 Ġulām Ṯaʿlab see Zāhid, ʾAbū ʿUmar al-
genitive 113, 159 Gully, Adrian 200
see jarr Gundel, Jeanette 249 f.
genitive pronoun 45
genus 233–252, 257, 259 ḫabar 112, 118, 160, 305, 310, 313
see jins ḫabar mufrad 309
proper name for 241 ḥadaṯ 32
index 325
ḥadd 115 ḥarf jarr 208, 299, 214, 222, 309

ḥadd al-lafẓ 107 ḥarf jarr zāʾid 83
ḥaḏf 43, 177, 183, 190, 213f., 216, 299 ḥarf jawāb 76
ḥaḏf al-jawāb 177 ḥarf jazm 71
Hadj Salah, Abderahmane 21 ḥarf madd 17
ḫafīf 15 ḥarf maʿnā 305
Ḥāfiẓ 184, 187 ḥarf muʿtall 128
Ḥafṣ 300f. ḥarf nafy 70
Ḥafṣ ʿan ʿĀṣim 298 ḥarf naṣb 70
ḥāl 3, 33, 80, 204f., 225–232, 241, 276f., 305 ḥarf tafṣīl 71
ḥāl fī l-ḥaqīqa, al- 276 ḥarf tanfīs 80, 83
ḥāl maḥkiyya 277 ḥarf tawkīd 71, 82
ḥāl muqaddara 277 harf yang mati see mati
ḥāl muqārina 277 ḥarf-i nidā 189 f.
ḥāl muwaṭṭiʾa 276f., 291 Ḥarīrī, al- 228
Ḫalaf al-ʾAḥmar 64 ḥarrafa 172
Ḫalīl, al- 5f., 42, 45, 64, 67, 123–126, 128, 158, Harris, Zellig 2, 13f., 20 f., 24–27
168, 170–173, 175, 177f., 212, 222, 279– Hārūn, ʿAbd al-Salām 134f.
281 Hashim bin Musa 305
ḫallaṣa 257 Ḥassānīn, Fatḥī 237
Ḥāmiḍ, ʾAbū Mūsā al- 63 ḫaṭṭ 123
ḥaml 17 hayʾa 276
hamza 300f. ḥayawān 242
writing of 299 Haywood, John 124, 127f.
hamza bayna bayna 301 ḥaẕf 190, 192–194
Ḥamza, Muḥammad Fawzī 135 see ḥaḏf
Hamzé, Hassan 204 ḫazl 43
ḥaqīqa 276 heaviness 48, 113
ḥāl fī l- 276 of construction 286 f.
mafʿūl fī l- 208 Hebrew 296
see ʾiḍāfa; mafʿūl ḥaqīqī Hedberg, Nancy 249 f.
ḥaqq al-ṣadr 73 Heselwood, Barry 255
ḥaqq bi-l-ṣadāra, al- 107, 111, 114f. heuristic arrangement 128, 130
ḥaraka 17f., 185, 228 hierarchy 93
ḥarf 18, 109, 123f., 185, 187, 275, 308f., 311, 314 between notions 113
ḥarf al-ʿilla 307 between parts of speech 113
ḥarf al-istifhām 115 Givenness 249–251
ḥarf al-istiṯnāʾ 94 metaphor of 103
ḥarf al-jazāʾ 158, 168, 170f. of definiteness 233, 242, 247–251
ḥarf al-mujāzāt 158, 168 of pronouns 34, 52 f., 57
ḥarf al-munādāt 309 of qiyās 19 f., 24
ḥarf al-šarṭ 71, 158 hierarchy problem 23
ḥarf al-ziyāda 126 ḫiffa 113
ḥarf ʿaṭf 70f. hijāʾ 67
ḥarf ʾiḍrāb 80 ḥikāya 91
ḥarf imtināʿ li-mtināʿ 78f. ḥīna 170
ḥarf ʾiʿrāb 47 Hindi 302, 313
ḥarf istidrāk 80 Hindustani 302–304
ḥarf istiqbāl 70, 80, 83 Hodges, Wilfrid 5 f., 157–181
326 index
Holtz, Louis 102 Ibn Qutayba 64, 90, 100

Houtman, Frederi(c)k de 304 Ibn Saʿdān, Muḥammad al-Ḍarīr 63
ḫ-ṣ-ṣ 253 Ibn al-Sarrāj 3 f., 49, 56, 64, 116, 119–121, 206,
ḥudūd see ḥadd 212 f., 217, 219–221, 223, 227 f., 238–240,
ḥukm 15, 17 248, 283
Humbert, Geneviève 133–135, 141 Ibn Sīda 77
ḥurūf 314 Ibn Sīnā 5 f., 157–181
see ḥarf Ibn Šuqayr 63, 227
ḥurūf al-hijāʾ 67 Ibn al-Ṭarāwa 65
ḥurūf al-maʿānī 67f., 81 Ibn al-Warrāq 205, 208, 253, 275
huwa huwa 113 Ibn Yaʿīš 48, 206, 227, 229f., 240 f., 243, 247–
Ḫwārazmī, al- 101 249, 260, 264, 276 f., 284, 290
hypothetical logic 165, 178 ibtadaʾa 111
hypothetical sentence 165, 170 ibtidāʾ 15, 91, 107, 110–114, 118
hysteron-proteron 106, 111, 116, 119– ʾiḏ 83
121 ʾiḏā 71, 79, 163–167, 170–173
ʾiḍāfa 81, 218, 249, 255, 263, 266, 288,
ʿibāra 70 305
ʾibhām 242 ʾiḍāfa ḥaqīqiyya 263
Ibn ʾĀjurrūm 98, 206, 297, 306, 308 ʾiḍāfa lafẓiyya 264
Ibn al-ʾAnbārī, ʾAbū Bakr 64 ʾiḍāfa maḥḍa 263 f.
Ibn al-ʾAnbārī, ʾAbū l-Barakāt 17, 204, 217f., ʾiḍāfa maʿnawiyya 255, 263–265
228, 261, 263, 275, 310 ʾiḍāfat al-maʿnā 275
Ibn ʿAqīl 95f., 102, 235, 242f., 264, 298 ʾiḍāfat al-šayʾ ʾilā l-ʾajnabī 287
Ibn al-ʾAṯīr 77 ʾiḍāfat al-šayʾ ʾilā nafsihi 287
Ibn Bābašāḏ 228 ʾiḍāfat al-ṣifa 287
Ibn Barrī 228 ʾīḍāḥ 261 f., 265–267
Ibn al-Dahhān 227, 230, 265 ʾīḏān 282
Ibn Durayd 76, 126, 128, 204 identifiability 107
Ibn Fāris 35, 126–128, 130, 257f. identification 250 f.
Ibn al-Ḥājib 157, 186, 227, 229, 260, 264, 276, ʾiḍmār 4, 30–60, 160, 236, 258
286, 289 ʾiḍmār, ʿalāmat al- 36, 46–48, 54 f.
Ibn Ḫālawayhi 64, 204 ʾiḫbār ʿan ġāʾib 32
Ibn Ḫaldūn 74 ʾiḫbār bi-llaḏī 289–291
Ibn Ḫarūf 66, 141, 258 ʾihmāl 166
Ibn al-Ḫaššāb 219, 228 iḥtimāl 260
Ibn al-Ḫayyāṭ 64 iḫtiṣār 215
Ibn Hišām 3–5, 61–89, 92, 95–97, 99–101, iḫtiṣāṣ 73, 253
157, 200–203, 205–207, 245f., 249, 260– ʾījāb 166
263, 265, 267, 276f., 282f., 298 ʾījāz 71, 82, 121, 215
Ibn Jinnī 64, 66, 82, 125f., 228, 253, 256–258, ʾilġāʾ 281, 283
263, 266, 275, 286, 288 ʿilla 116, 120 f., 307
Ibn Kaysān 63f. ʿillat al-ʿilla 120
Ibn Mālik 66, 79, 95, 100, 102, 227, 230, ʿilm al-ʿarabiyya 14
243f., 255, 263, 266, 297 ʿilm al-ḥadd 8
Ibn Manẓūr 77 ʿilm al-istidlāl 8
Ibn Muʿṭī 228 ʿilm al-naḥw 120
Ibn al-Nadīm 62–64 ʿilm-i maʿānī 313
Ibn al-Nāẓim 227, 230 ʾiʿmāl 101
index 327
ʾimmā lā 43 interrogative 114, 175, 238, 245

Immediate Constituent analysis 11f. interrogative particle 115
imperative 80, 109, 175, 194 interrogative sentence 108, 110
imperative sentence 110 invariants 21
imperfect verb 74, 80 ʾiʿrāb 7, 61–88, 256, 306
implication 164f., 172 error of 69
implicit 38 Īravānī 183f., 192–194, 312
implicit element 43, 49f. irreality 157
implicit noun 40, 44 ʾišāra 241 f.
implicit pronoun 36 ʾIṣfahānī, Mirza Ḥabīb-i 312
implicit verb 44, 161 Islamization, of Southeast Asia 302
imposition, of language 260 ism 18, 20, 241, 305, 308
Imruʾ al-Qays 282f. ism al-ʿalam 45
imtizāj 265 ism al-ḥadaṯ 4
ʾin 6, 157–159, 161f., 164–166, 168, 170f., 173f., ism al-ʾišāra 241–243
178, 279, 282 ism al-jins 242
ʾin, virtual 162 ism al-mutakallim 55
inanimate 242 ism ḫāṣṣ 45
ʾināṯ, nūn al- 82 ism mawṣūl 71, 200, 299
inclusion 27 ism mubham 3, 54, 233–252
incompetence, linguistic 77 ism muḍmar 39 f.
indefinite 114, 268 ism mufrad 16
indefinite expression 254, 258–261, 263– ism-i fāʿil 189, 192–194
265, 267 ism-i mafʿūl 189, 192
indefinite noun 107 ism-i mansūb 192–194
indefiniteness see indefinition ʾisnād 17, 112
indefinition 205, 253–272 istaʿmala 228
independent pronoun 73, 289, 292 istiʿāra 204
indeterminate 261, 268 istidrāk 80, 309
indetermination 253–272 istīfāʾ 71, 82
Indo-European 183, 187 istifhām 18, 108 f., 175
Indonesia 297 ištiġāl 35, 95 f., 102
Indonesian 7, 313 istiġnāʾ 115
infinitive, in Persian 188 iṣṭilāḥ 204, 260, 314
infiṣāl 15, 291 istilah 314
initial noun 32 ištimāl 215
initial position 107, 114f. istiqbāl 81, 258
ʾinna 82, 93, 284 ištirāk 258, 260, 262
ʾinnamā 214 ʾitbāʿ 25
inqaṭaʿa 31 ʾiṯbāt 300
ʾinšāʾ 253 iʿtibār 17
insertion 284 ʾiṭlāq, ʿalā l- 200 f.
intended meaning 106, 110, 115, 120 ittasaʿa 212
intention ittisāʿ al-kalām 3, 212–224
communicative 27, 32 ittiṣāl 290
of the speaker 17f., 26, 32, 106, 108, 182, iżāfa, in Persian 182, 188, 190
187, 214–217, 219, 250, 278, 291 izālat al-labs 290
interlocutor 159, 215, 259 izālat ištirāk 258
see addressee; listener ʾiẓhār 38 f.
328 index
Jaʿd, Muḥammad ibn ʿUṯmān al- 63 ka- 73

Jāḥiẓ 61f. kāf al-ḫiṭāb 242
Jakobson, Roman 47f. Kaffawī, al- 101
jamʿ 113, 275, 314 Kafrāwī, al- 298
jamʿ taksīr 17f., 21 kalām 40, 50, 305
jamʿ, ʿalāmat al- 45, 47 kalām al-ʿArab 38, 120
jamāʿa 35 kalām al-Zanj wa-l-Rūm 61
jamād 242 kalima 314
jamak 314 kalimat 314
jāmiʿ 17, 20, 25 kalimāt al-šarṭ 157
jāmid 261 Kaptein, Nico 304, 306
Jārburdī, al- 264 karm 313
Jarīr 223 Karmānī, al- 64
Jarmī, al- 63, 134 kartā 313
jārr 112 Kasher, Almog 2, 4 f., 69, 182, 198–211
jarr 245, 174, 186, 208, 288f., 291 kas̱īr al-istiʿmāl 192
Javanese 297, 303f. kasra 299, 302, 312
jawāb 158, 160, 168, 174f., 177 kasra-yi iżāfa 194
jawāb al-ʾamr 76, 80 kata 298, 308
jawāb al-šarṭ 76, 79f. kata benda 313
jawāb al-ṭalab 80 kata depan 314
Jawharī, al- 64, 94, 127, 130 kata kerja 313
Jawi script 299 kata sambung 314
jayyid 56, 107 kata sifat 314
jazāʾ 168 kata tugas 313
jāzim 76, 80 kaṯrat al-istiʿmāl 43
jazm 72 kay 244
Jazūlī, al- 228 kayfa 115
Jeremiás, Éva 6, 182–197, 312 Ketelaar, Johan Josua 303
Jespersen, Otto 202 Khalidov, A.B. 133
Jihāmī, Jīrār 78–81 Kirmānī 312
jins 100, 259 Kisāʾī, al- 62f., 309 f.
Johor, Malay of 305 knowledge, transmission of 129
Johor, Sultanate of 297 Koningsveld, Pieter-Sjoerd van 102
Jones, Alan 278 konjungsi 314
jumal see jumla 66 Kouloughli, Djamel Eddine 106 f., 254,
jumla 66, 68f., 79 267
jumla fiʿliyya 69 Kratzer, Angelika 6, 164f., 167, 176
jumla ismiyya 69 Kridalaksana, Harimurti 295, 311
jumla kubrā 69 kriya 298
jumla ṣuġrā 69 Kufans 45, 101, 140, 238, 247, 281, 310
jumla ẓarfiyya 69 kull 170
jurative particle 73 kullamā 6, 157, 164 f., 167, 170, 172
Jurjānī, al- 4f., 106f., 200–202, 204f., 207, kulliyya 82, 166
216, 228, 253, 258f., 276 kunya 241
Jurjānī, ʿAli ibn Muḥammad al-Sayyid al-Šarīf
al- 258, 265 lā 282–284
jussive 80 la- 93, 277 f., 280–282
juzʾiyya 166 Labadī, Muḥammad 78–81
index 329
labs 290 lexicon 5

lafaz 308 mubawwab 67
lafẓ 123, 183, 192, 214, 216f., 204f., 308 li- 34, 208
lafẓ, fī l- 111, 215 lightening 286
lafẓī see fāʾida lafẓiyya; ʾiḍāfa lafẓiyya lightness 48, 113, 286, 288
laġw 178, 281 lingua franca, Malay as 297, 303
lāḥiqa 313 listener 108–110, 119
laḥn 62, 76 literal meaning, of linguistic terms 203–
lāʾiq, ḍamīr 243 206
lam 71 literal sense 214, 216
lām al-ibtidāʾ 93, 284 Littré, Émile 126
lām al-muʾḏina, al- 277 loanword
lām al-muwaṭṭiʾa li-l-qasam, al- 277f. Qurʾānic 89
lām-ʾalif 125 Arabic in Berber 312
lan 70 Arabic in Indonesian 313
language Arabic in Malay 299, 302, 311
and mathematics 21 Arabic in Persian 312
as behavior 18 Arabic in Urdu 313
as social behavior 12 Dutch in Indonesian 313
auxiliary 295, 297, 302, 313f. Persian in Malay 311
holistic approach to 26 Perso-Arabic in Urdu 302
indigenous 295f. Sanskrit in Hindi 302, 313
vernacular 296, 311 Sanskrit in Javanese 297
lanjut 300 Sanskrit in Malay 311
laqab 241 logic 162f., 257f.
Larcher, Pierre 198, 207, 253, 255, 276 conditional 6, 157, 159
Larkin, Margaret 200 Greek 157
Latin grammar 296 Peripatetic 165
latitude, of speech 213 logical subject 40 f.
law 39, 78f., 157–164, 244, 280–282 looking-up arrangement 127
virtual 162f. luġa 204
lawlā 281 luġawiyyūn 62
Lawrence, Kelvin 305 Luġda 63f.
Layṯ ibn Muẓaffar, al- 42, 126 Lumsden, Matthew 188
legal science 170, 253 Lyons, John 199, 205
Lentin, Jérôme 24
Leow, Rachel 303, 305 mā 73, 91, 177 f., 244, 279–281
letter 275 mā ḏā 94
level mā l-kāffa 284
deep 18 maʿānī l-Qurʾān 62
of analysis 15 mabdūʾ bihi 115
surface 18, 186, 189 mabnī 72, 83
underlying 182, 186, 188–190, 193, 214 mabnī ʿalā 40, 112 f.
Levin, Aryeh 3, 44, 46f., 52, 112, 174f., mabnī ʿalā l-sukūn 82
182, 187, 198f., 207, 212–224, 275, mabnī li-l-majhūl 17
278 Mac Lane, Saunders 22
Lewis, David 164, 165, 167 macrostructure 123
lexicography 5, 8, 123–132 maḍā 31
Persian 183f. mādda 126
330 index
madḥ 257, 260 Mammeri, Mouloud 312

māḍī 71f., 169 mamnūʿ min al-ṣarf 83
madrasa 102 maʿmūl 96
mafʿūl 108, 117, 198f., 206f., 214, 218, 287, maʿmūl al-ṣifa 287
305, 309 maʿmūl fīhi 93 f., 119
mafʿūl ʿalā l-saʿa 218–222 maʿnā 72, 193, 204, 214–221, 223, 299
mafʿūl al-natīja 199 maʿnā l-mujāzāt 158
mafʿūl bihi 198, 200–203, 205f., 208, 305 maʿnā l-šarṭ 160
mafʿūl bihi muḥdaṯ 199 maʿnā maqṣūd 193
mafʿūl bihi mutaʾaṯṯir 199 maʿnā, fī l- 215, 287
mafʿūl dūnahu 94 maʿnaviyāt 313
mafʿūl fīhi 99, 198, 305 maʿnawī see fāʾida maʿnawiyya; ʾiḍāfa
mafʿūl ḥaqīqī 198, 208 maʿnawiyya
mafʿūl lahu 208, 305 manfī 282
mafʿūl maʿahu 208, 305 mansūb, ism-i 192–194
mafʿūl muṭlaq 4, 97–99, 198–211, 227, manṣūb 34, 99, 206, 288
305 manuscript
mafʿūl ṯānī 118, 289, 291 Ambrosiana 133–156
maḥall 55, 68f., 71, 73, 76, 80 Kazan 133, 140
Maḥallī, Jalāl al-Dīn al- 298 Toledo 102
maḥḍ 287 manuscript transmission 129 f.
see ʾiḍāfa maḥḍa manuscripts 129 f.
maḥḏūf 187, 278 of the Kitāb 133–156
maḫfūḍ 8 manʿūt 78, 258, 266
maḥkī 277 manzila 16, 20, 39, 48, 55, 93, 103, 158–160,
mahmā 177 170, 178, 279 f.
Mahmoudian, Morteza 13f. mapping 23
Maier, Hendrik 303 maqāyīs 14
Majapahit, Empire of 297 maqāyīs al-luġa 20
majāz 200 maqṣūr 300
majhūl 184 Marāġī, Muḥammad ibn ʿAlī al- 63
majlis 129f. marfūʿ 34, 37, 44, 287
majrā 16, 51 maʿrifa 37, 50, 113 f., 235, 237, 254, 258–
majrūr 34, 45, 246, 288 265
makān 113, 115, 168 marked 48
makna 314 Markedness Theory 48
maknī 238 marker 47–49, 57
maktūb 186 of definition 254
Malay 6f., 295–318 of determination 254
Classical 303 of indeterminacy in Persian 190
negation in 309 of feminine 55
of Johor 305 Marogy, Amal 1, 8, 107, 111 f.
prepositions in 309 Maróth, Miklos 165
Malay grammar 6–8, 295–318 martaba 93, 103
Malay word order 310 maʿrūf 111
Malaysia 297 masadd 114
malfūẓ 186 masculine 113
Malḫ, Ḥasan 75 maṣdar 3, 41, 97 f., 187, 190, 199 f., 203, 205 f.,
mamdūd 300 218 f., 226, 229, 246
index 331
maṣdarī 82 Moeliono, Anton 313f.

Massignon, Louis 21 morpheme 47
matā 177f. absence of 48
Mataram, Empire of 297 morphism 22f.
matbūʿ 93, 261, 266 morphological arrangement 127
Mathematical Category Theory 2, 10f., 22f., Mosel, Ulrike 20
26f. Muʿāḏ al-Harrāʾ 61
mathematics 21, 26 muʾaḏḏin 278
mati 299f., 306 muʾaḫḫar 108
maʿṭūf 51, 73 muʿallim 76, 125
mawḍiʿ 12, 14f., 48, 50, 53, 114, 159, 200, 226, muʾannaṯ 113
228 muʾaṯṯir 263
mawḍūʿ 266 Mubarrad, al- 3, 54–56, 61, 64, 99, 116–119,
mawqiʿ 37, 114 121, 133 f., 140, 212 f., 218, 220, 226–228,
mawṣūf 51, 93 230, 237 f., 257
mawṣūl 93, 243 mubawwab 67
mawṣūl ḥarfī 244 mubham 3 f., 6, 54, 158, 170 f., 173, 233–252
mawṣūl ismī 244 mubtadaʾ 32, 34, 111–115, 160, 222, 305, 309 f.
mawṣūla 241 mubtadaʾ muḍmar 40, 44
Māzinī, al- 55, 61, 134 mubtadaʾ muẓhar 44
measure 23f. mubtadiʾ 65, 69, 72, 120
Melayu yang asli 305 muḍāf 263, 265
membership 24 muḍāf ʾilayhi 264
in category 22 muḏakkar 113
memory 129f. muḍāriʿ 71 f., 80
long term 250 muʾḏin see lām al-muʾḏina, al-
short-term 250 muḍmar 4, 30–60, 174, 242
menghadhafkan 299f. muḍmar, ʿalāmat 37, 41, 46, 48, 54
mengithbatkan 300 Mufaḍḍal ibn Salama, al- 64
mengwakafkan 299 mufassirūn 75
mengwasilkan 299 mufīd 308
mentahkikkan 299 mufradāt 66
mental representation 41f. Mughal Empire 302f.
mentashilkan 299 muḥaddaṯ 51
menthabitkan 300 muḥaddaṯ ʿanhu 32 f., 37, 46, 54
menukarkan 299 muḥaddaṯ bihi 32
menyebutkan 299 muḫaffaf 194
metaphor 204 muḥāl 106
mī-, in Persian 186, 191 Muhallabī, ʾAḥmad ibn Muḥammad al- 63
microstructure 124 Muḥammad ibn Qādim 63
mind, of the speaker 40f. Muḥammad Maʿṣūm ibn Sālim 298
missionaries Muhammad Said bin Sulaiman 311
European 296 muḥarraf 174
Persian 302 muḫaṣṣiṣ 253, 266
miṯāl 11, 15–26, 95 muḫāṭab 4, 30–34, 37, 43, 46, 53, 56, 109,
miṯāl jāmiʿ 17 215, 259
miṯl 239, 248 muhla 71
modal sentence 173 muhmal 166
model 24 muḫtalas 184
332 index
muḫtaṣar fī l-naḥw 63f. muwāfaqa 17, 178 f.

muḫtaṣar naḥw 63 muwaṭṭiʾ 5, 274–294
Mujāšiʿī, al- 66, 228 see ḥāl muwaṭṭiʾa; la- al-muwaṭṭiʾa
mūjaz 71 mużāf ilayh 193 f.
mukallam 31 muẓhar 36, 39, 45, 49
mukammal 266
mukammil 265f. nāba 160
mumahhid 277 Nāfiʿ 298
munādā 73, 83, 187, 189f. nafs 53
munāsaba 17 nafy al-jawāb 282
munfaṣil 33, 46, 171, 289 Naḥḥās, al- 64, 133
Munsyi, Muhammad Ibrahim 311 naḥw 12, 66, 70, 306
muqaddam 107f., 193 naḥwiyyūn 51, 62
muqaddar 17, 183, 190, 277f., 310 nahy 110
muqārin 277 nāʾib ʿan al-fāʿil 70
Muqātil 38f., 49 Naiki, Ryoichi 235
muqayyad 198 Naʿja, Suhā 75
muqsam 282 nakira 113 f., 237, 254, 258–266
muqsam ʿalayhi 282 nama 298, 308, 311
muʿrab 7 naqd al-ʾiʿrāb 74
murād 73, 259, 265 naqd al-šiʿr 74
Murādī, al- 69 nāqiṣ 174
murakkab 308 Nardella, Umberto 313
muʿrib 70f., 77f., 79 nasaq 266
musallaṭ 100 naṣb 72, 275, 286–289, 291
musamman 242 nāṣib 76, 80, 112, 287
Musayyib ibn ʿAlas 280 nastaʿlīq script 302
musnad 112, 114, 313 naʿt 78, 255, 257 f., 261 f., 266
musnad ʾilayhi 112, 114 naʿt sababī 285
mustaqarr 114 natural language 10
mustaqbaḥ 287 nawʿ 100, 229, 260
mustaqbal 79 Nawawī, Muḥammad 298
muštaqq 192, 276 naẓāʾir see naẓīr
mustatir 71 naẓīr 13–16, 19
mustawfā 71 negation 73
mutaʿaddin 201 in Malay 309
mutaʿallim 69, 80, 119, 166 negative particle 282f.
mutaʿalliq 71 Nevin, Bruce 13
mutadarrib 69, 71 nidāʾ 31, 187, 189
mutaḥarrik 184f. Nīl, ʿAlī Fūda 67, 69
mutakallim 4, 30–34, 37, 53, 55f., 114, 120, nīm-fatḥa 184
190 niyya 17 f., 26, 39 f., 49 f., 55
mutamakkin 16 see speaker, intention of
mutamakkin, ʿalāmat al- 45 nominal relative 244 f.
mutamakkin, ẓarf 218 nominal sentence 75, 95, 107
muṭlaq 198 nominalization 170
muṭṭarid 228 nominative 113, 215
muttaṣil 33, 56, 164 f., 171–173, 289 see rafʿ
muwaḍḍiḥ 261, 265 non-definiteness 255
index 333
non-observable entities 41f., 49 operator 5, 21, 91, 94, 96 f., 99, 101, 103, 119,
norm 16, 107, 110f., 213 208
noun 113 optative verb 40
ambiguous 239–245, 247–249, 251 oral transmission 129 f.
definite 233, 237f., 244, 247, 249 ordering principle, in dictionaries 5, 123–
derived 276 132
implicit 40, 44 overt sign 48
indefinite 248, 266, 309 Owens, Jonathan 8
proper 233, 235, 238, 240, 244f., 247f.,
251 Pakistan 313
underived 276f. panjang 301
noun phrase paradigmatic axis 292
definite 233, 235f., 238, 240, 244f., 247f., paradigmatic relations 13
251 parpratyaya 313
indefinite 234, 239 participation 23
non-referential 233f. by universality 25
referential 233f. in a structure 27
nudba, ʿalāmat al- 45 participle 3, 54, 222, 229f., 246, 276, 285
numerals 136, 151–155 active 143, 214, 226–228, 310
nūn al-ʾināṯ 82 in Persian 188 f., 192
nūn al-wiqāya 34 passive 198, 202 f., 207, 228, 243
nunation 266 redundant 291
see tanwīn relative 244 f.
of oath 277–279
oath 277–280 particularization 4, 253 f., 256–260, 262–
implicit 280 268
omission of 278 parts of speech 20, 188, 214, 296, 207, 311 f.
object 107f., 117, 287, 313 in Persian grammar 312
absolute 4f. passive verb 71, 219, 222
affected 199, 202 passive voice 18
affective 199, 205 past continuous, in Persian 191
cognate 3 Patras, Frédéric 25
direct 214, 286 pedagogical grammar 2, 61–88
effected 202 Peled, Yishai 203, 278, 281, 288 f.
effective 202 Pelletier, Francis 165
inner 202 Pellò, Stefano 312
of a preposition 208 pendek 300
of result 199, 202, 205 perbuatan 308, 311
second 289–291 perfect verb 71
object suffix, in Persian 191 performative 32
Occam’s Razor 14f. Peripatetic logic 165
Ogloblin, Aleksandr 297, 314 perkataan 308
omission 213f., 216, 227, 278f., 284, 286, permutation 124
307 permutative 265
of the ḥarf al-ʿilla in Jawi script 307 permutative arrangement 125–127
of the marker of completeness 286 Persian 182, 296, 302, 304, 312
of the oath 278 adjective in 187 f.
omitted verb 246 compounds in 6, 182 f., 192–195
operation 22, 24, 26 education in 312
334 index
grammar 6, 8, 182–197, 302 in reading 295

infinitive in 188 linguistic 74, 76 f.
iżāfa in 182, 188, 190 pronominalization 50–52
participle in 188f., 192 pronoun 3 f., 30–60, 233–236, 238, 240, 242,
parts of speech in 312 244 f., 247–249, 251
past continuous in 191 bound 33
pluperfect in 191 implicit 36
present stem in 188 independent 290
verbal noun in 188 nominative 37
word order in 191 personal 30
Persian poetry 184, 186f., 195, 312 resumptive 244
Persian prosody 185f. proper name 4, 47, 51, 241, 254, 260
Persian vowel names 312 proposition 171
personal pronoun 285 prosody, Persian 185 f.
pesantren 297f., 302 protasis 79, 171, 280
phonetic order 124–126 Pruvost, Jean 129
phonetic realization 48f. Punjabi 313
Piaget, Jean 21 purusa 297
pīš 302, 312 pūrvapratyaya 313
pity 260 Putten, Jan van der 295, 305
pluperfect, in Persian 191
plural 275 Qabāwa, Faḫr al-Dīn 78–81
pluralis 314 qabīḥ 52, 56, 159, 186
poetic license 278 qabuḥa 53
poetry 278, 288 qaddama 108
Old Arabic 212, 216 qaddara 278
Persian 184, 186f., 195, 312 qaḍiyya 166, 172
Polotsky, H.J. 214 qāʿida 67, 70
polysemy 258 Qālī, al- 126
possessive, determinative in Persian 190 qāsa 52
postposing 116–120 qasam 277–279
pragmatic context 53 qaṭʿ hamza-nya 299
pragmatic property 50 qawāʿid al-ʾiʿrāb 70
praise 260 qawāʿid see qāʿida
prayer 41 qayd, bi-lā 201
predicate 112f., 117, 313 Qazwīnī, al- 256
anteposing of 114 Qifṭī, al- 62
nominal 289 qirāʾāt 298, 301
predication 32, 40 qiyās 14–18, 19–23, 25 f., 116, 213, 226–230
predicative relation 285 qiyās muṭṭarid 228
prefiks 314 qiyāsī 227, 229 f.
prefix 313f. qualification 255, 257f., 260–263, 267
preparation 274–291 qualifier 93, 256, 276
preposisi 314 quantification 164f., 167, 176, 254
prepositional phrase 198, 243, 245 quantifier 167, 170
prepositions, in Malay 309 qubḥ, ʿalā 288
present stem, in Persian 188 Qurʾān 12, 278
preservation, of structure 18f. exegesis of 298, 304
proficiency language of 14
index 335
recitation of 300 activated 250

šawāhid from 65 mental 41 f.
suppression in 177 Rescher, Nicholas 165
variant readings of 301, 313 resemblance 23
Qurṭubī, al- 278, 280 Resnik, Michael 26
resumptive pronoun 244
-rā, in Persian 191 Retsö, Jan 255
Rabāḥī, ʾAbū ʿAbdallāh Muḥammad ibn rhyme 288
Yaḥyā al- 133, 135, 137f. in Persian 193
rābiṭat jawāb al-šarṭ 79 rhyming order 127, 130
radd ʾilā l-ʾaṣl 17, 19, 21 rhythm 284, 288
radicals 27, 124, 126 Riau, Sultanate of 297, 303f.
rafʿ 72, 76, 80, 112, 275, 285–288 riʿāyat al-ʾaṣl 291
rafʿ al-ʾiḍmār 258 Riddell, Peter 295, 298–302
rafʿ al-ištirāk 258, 260 Roman, Johannes 304
rāfiʿ 112, 287 Ronkel, Philippus van 305 f.
Rahman, Tariq 302–304 root, formal 123f., 127
Raja Ali Haji 7, 295, 304–311 rubba 92, 248
Rāzī, al- 278 rubbamā 91
Reckendorf, Hermann 255f., 267 Rufayʿ ibn Salama Damāḏ 61
recurrence 15 rules, of grammar 70
redundant 281–284 Rummānī, al- 64, 90 f., 94, 100, 102, 228,
redundant antecedent 177 309 f.
redundant constituent 284
redundant particle 291 sa- 80f., 83
reference 4 saʿa 288
generic 234f., 251 ʿalā l- 214, 218–222
homophonic 234f. ḥāl al- 288
non-specific 233f. saʿat al-kalām 212
of pronoun 49–52 sābiqa 313
specific 233f., 235 Sadan, Arik 3 f., 54, 225–232, 235
unspecific 235, 251 Saʿdī 184, 187
referent ṣadr 244
intended 250 ḥaqq al- 73
of pronoun 248 ṣaḥḥaḥa 91
of proper noun 247f. ṣāḥib al-ḥāl 276
referential cognition 249 ṣāḥib al-ṣifa 111
referentiality 8, 233f. Ṣāḥib ibn ʿAbbād, al- 126
reflexive 53 šāhid 64 f., 67, 82
regere 89, 103 Sakaedani, Haruko 3, 233–252
regularity 19f., 26 sākin 184 f.
relative Sakkākī, al- 8, 157
al- 243f., 246 Šalawbīnī, al- 227, 229
particle 244f. salb 166
relative clause 235f., 242–246, 263 sallaṭa 5, 89–105
relative pronoun 71, 94, 233, 235–238, 240– samāʿ 103
245, 249, 251, 279f. samaʿ, ʿalā l- 229
representation 250 samāʿī 227, 229 f.
abstract 182 Šams-i Qays 184–186, 193
336 index
Sanskrit 297, 302, 313 zero see zero sign

Šantamarī, al- 135, 227 ṣila 93, 243 f., 281
šāraka 257f. similarity 16 f., 19f.
ṣarf 190, 306 ṣināʿa 70, 72–74, 77, 82
Šarīf al-Kūfī, al- 228 ṣināʿatāni, al- 74
ṣarīḥ see ṣifa ṣarīḥa; waṣf ṣarīḥ Sindhi 313
šarṭ 79, 175 ṣinf 100
šarṭī 162, 164, 171f. singkat 300
šarṭiyya 166 Singkilī, ʿAbd al-Raʾūf al- 298–302
Sartori, Manuel 4, 253–272 Sinnū, ʾAhyaf 78–81
Saussure, Ferdinand de 12 Sinnū, Hiba 78–81
šawāhid see šāhid Sīrāfī, al- 5 f., 53, 55, 135, 144, 157–181, 227–
sawfa 80f. 230, 240, 247, 253
Šaybānī, ʾAbū ʿAmr al- 125 Širbīnī, al- 99, 206
Ṣaymarī, al- 227, 229 six nouns 274f., 291
Schoeler, Gregor 129 Sous 311
scholasticism 102 Southeast Asia 295–318
school, Islamic 297 spatial expressions 54
Schubert, Lenhart 165 speaker 100 f., 107, 110, 115, 117f., 288
script as arbiter 107, 111, 115, 119, 121
Devanagari 302 as learner 107, 118–121
Jawi 299 as originator 107, 120 f.
nastaʿlīq 302 choice of 119, 288
spelling errors in 306 freedom, of 3, 25, 111, 119
see ḫaṭṭ intention of 106, 108, 182, 187, 214–217,
semantic contribution 284 219, 250, 278, 291
semantic see maʿnawī intention of, see niyya
sentence role of 3, 292
imperative 110 tendency towards lightness 286
types of 69 species 234, 257, 259 f.
see jumla see nawʿ
set theory 22 animal 241
Sheyhatovitch, Beata 5, 274–294 specification 256f., 263, 267f., 286
shwa 185 specificity 247 f., 255
Sībawayhi 1–11, 14–19, 23, 25–27, 30–60, 64, speech participant 30
74, 79, 89, 91, 99f., 106–122, 133–158, speech situation 31–33, 40
161–163, 167, 169, 171, 173–175, 178, 195, spelling error, in Jawi script 306
204, 212–215, 217, 219–223, 225–232, š-r-k 258
235–237, 239f., 247, 253, 257, 279–284, state see ḥāl
286, 300f. structural analysis 24
šibh 239f., 248 Structuralism 24 f.
šibh jumla 81 Structuralist Linguistics 2, 8, 13
ṣifa 34, 51, 92f., 256 f., 259, 261, 305 student, of grammar 67–69, 70, 74f., 77, 81 f.
ṣifa ṣarīḥa 243 subject 107 f., 117
sifat 314 logical 32 f., 40 f.
ṣifat-i mušabbaha 192, 194 of nominal sentence 240
sign of active participle 214
absence of 47f. of passive verb 204, 214
overt 48 of verbal sentence 285
index 337
subjectivity taʾkīd 283, 305

in language 30 taʾkīd, ʿalāmat al- 45
subjunctive takmīl 265–267
after wa-, fa-, ʾaw 61 ṭalab 80
subordinate clause 244 Ṯaʿlab 64, 134, 140, 227
substance 202 Talḥūq, Muḥammad Beg al-Lubnānī 98
substantivum 313 Talmon, Rafael 64
Sūdī, Muḥammad 184, 186–192, 194f. tamakkun 113
suffix 313f. tamām al-kalām, ʿalāmat 286
sufiks 314 tamhīd 282, 288
Suhaylī, al- 199, 229 tāmm 245
suhūlat 194 tamṯīl 16, 38, 54
sukūn 82, 299 tamyīz 204 f., 276, 305
sukūt 15, 115 ṯanāʾ 257
Suleiman, Yasir 10–12 Tanaka, Nozomi 233, 235, 237, 240, 242
sulṭān 89 tanāzuʿ 96 f., 101
Sundanese 303 tanbīh 18
suppletive insertion 182 tanfīs 81
suppression 38 taʾnīṯ, ʿalāmat al- 45, 47
in the Qurʾān 177 tankīr 253–256, 258, 265–267
surface level 18, 186, 189 tanvīn 184
Suyūṭī, al- 134, 201, 298 tanwīn 46–48, 75, 184, 266, 286, 289, 299 f.
syllogism 15, 162, 172 tanwīn, absence of 48 f.
syntagmatic relation 13, 292 taqaddama 96
Syriac 89, 296 taqdīm 106–122, 192 f.
Syriac grammar 8 taqdīm wa-taʾḫīr 3
taqdīr 3, 6, 16, 54, 99, 173–177, 182–197, 208,
Tābirī, Hānī 78–81 212, 216–218, 227, 258, 278, 306
ta fāʿil 76 taqdīr, ʿadam-i 194
tāʾ marbūṭa 47, 301 taqdīr-i kalām 190
taʾaḫḫara 96 taʿqīb 70
tābiʿ 93, 256, 266 taqrīb 76, 80
ṯābit 300 taqṭīʿ 186
tabyīn 204, 283 taqyīd 201
taʿdiya 94 taraḥḥum 261
tafḍīl 262 taʿrīf 4, 253–256, 258, 263–267
tafsīr 160, 177 tarkīb 182, 187
Tahānawī, al- 101, 183 tarkīb-i vaṣfī 188
taḫfīf 6, 43, 182–197, 222, 299 tartīb 70 f.
taʾḫīr 18, 106–122 tartīb ṭabīʿī 174
taḥlīl 173f. tašābuh 16, 19
taḫlīṣ 253, 256–258, 262f., 265, 267 tasallaṭa 89–105
taḥliya 51 tasalluṭ al-ʿāmil 100 f., 103
taḥqīq 299 tasalluṭ al-fiʿl 99
taḥqīq, ʿalā l- 208 taṣarrafa 52
taḥqīr 17, 21, 237 taṣawwara 259
taḫṣīṣ 4, 253–268 tašbīh 192
tāʾ-i āḫir 191 tašdīd 184, 299
tāʾ-i ḫaṭāb 191 tashīl 299
338 index
taskīn 184 ʿUkbarī, al- 207 f., 228, 276

taslīṭ 101 ʿUmar ibn Lajaʾ 288
taslīṭ al-fiʿl 99f. ʾumma 235
tasmiya 116 underlying level 182, 186, 188–190, 193, 214
taṣrīf 98 uniquely identifiable 250
taswīf 81 universal, concrete 23
tata bahasa 298 universal principles 26
taṯniya 18, 275 universality 12
tawābiʿ see tābiʿ unmarked 48
tawḍīḥ 258–263, 265–267 Urdu 296, 302, 313
taʾwīl 218 ʾUšmūnī, al- 298
tawkīd 36, 52, 261–263, 281 ʾuṣūl al-fiqh 8
tawsīʿ 80 ʾuṣūl see ʾaṣl
tawṭiʾa 5, 274–294 utterance 31–33, 41, 106, 111, 114, 213, 216
tazelɣa 312 actual 43 f.
teachers, of grammar 75f.
teaching, of grammar 62, 66 Vanbremeersch, Jean-Paul 22
temporal expressions 54 variant readings, of the Qurʾān 298, 300
term vaṣf-i tarkībī 192, 194
definite 257 verb 32, 113
technical 203f. cognitive 283
terminology 54, 82, 253 description of 31
grammatical 70–72, 78–83, 100f., 203– elision of 73, 92
206, 218 geminated 142–150
of Malay grammar 298–311 imperfect 74, 80
of Persian grammar 312 implicit 44, 161
of pronouns 54–56 movement 234
of redundancy 284 of condition 6, 161
test, linguistic 289 of doubt and certainty 116–118
Testen, David 277, 279 optative 40
textbook 62 passive 71, 219, 222
thematization 112 perfect 71
theme 112, 240 transitive 94, 219
ṯiqal 113 vicarious 202
topic 111–114, 117 verbal noun 3, 97, 225–232, 246
topicality 107, 110 see maṣdar
topicalization 112 in Persian 188
transformation 21f., 26, 287, 289–291 verbal phrase, in Persian 190
correspondence of 21 verbal sentence, inverted 95
transitive verb 94, 219 Versteegh, Kees 1–9, 11, 38, 157, 159, 161, 182,
transitivity 94 185, 189, 193, 212, 216, 278, 281, 283, 288,
translationese 302 295–318
Troupeau, Gérard 35, 94, 102, 255, 257 Viain, Marie 295
truth conditions 163 vidheya 313
ṯulāṯī 21 vocative 31, 237
ṯumma 71 in Persian 187, 189 f.
Turkic 296 voegwoord 314
Turkish, Osmanli 186 Von Fintel, Kai 164f.
type identifiable 250 voorzetsel 314
index 339
vowel 184f. Yamanī, al- 227, 229

long 184f. yaʿnī 186 f.
prosodic 185 Yaʿqūb, ʾImīl Badīʿ 78–81, 135, 255, 261
vowel insertion 184f. Yāqūt 63f.
vowel names 302 Yazīdī, ʿAbdallāh ibn Yaḥyā ibn al-Mubārak
in Persian 312 al- 63
vowel signs, in Jawi script 306f. Yazīdī, Muḥammad ibn al-ʿAbbās al- 63
vowels Yazīdī, Yaḥyā ibn al-Mubārak al- 63
as case markers 275
in Persian 185 Zacharski, Ron 249 f.
writing of 299 Zadeh, Travis 312
Zāhid, ʾAbū ʿUmar Muḥammad ibn ʿAbd al-
wa- 208 Wāḥid al- 134
circumstantial 75 ẓāhir 47 f., 54, 287
with subjunctive 61 ẓāhir al-ṣināʿa 72
waḍʿ al-wāḍiʿ 260 ẓāhir see ʿalāma ẓāhira
waḍʿ, bi-l- 308 zāʾid 6, 174, 281 f., 284, 312
wāḥid 113 Zajjāj, al- 63, 92 f., 100, 278, 280
wājib 31f. Zajjājī, al- 3, 64–66, 93, 100, 227f., 257f., 275
waqaʿa 31, 225f., 235, 259 Zamaḫšarī, al- 66, 92, 127, 135, 137f., 164,
wāqiʿ 79 206, 227–229, 240, 243, 246 f., 253,
waqf 15, 299 259 f., 264, 277 f.
waṣala ʾilā 208 Zamalkānī, al- 227, 229
waṣf 187f., 256, 262, 276 zamān 113, 168
waṣf ṣarīḥ 246 zamān ḍayyiq 80
waṣl 299 zamān wāsiʿ 80
waṣl hamza-nya 299 ẓarf 81, 198, 216, 218–222, 243, 245
Waššāʾ, al- 63 ẓarf li-mā maḍā 83
waṣṣala 266 ẓarf li-mā yustaqbal 79, 83
Watson, Janet 255 ẓarf mustaqbal 71
waṭṭaʾa 276 ẓarf mutamakkin 218–220
Weiss, Josef 89 ẓarf zamān 99, 206, 222
Wensinck, Arent Jan 255 żarūrat-i vazn 190
werkwoord 313 Zawzanī 183
word order 3, 32, 107f., 110f., 114f., 290 zebar 302, 312
in Malay 310 zero sign 47–49
in Persian 191 zīr 302, 312
Wright, William 254–256, 267 ziyāda 174, 281, 283
wuqūʿ 161 Zubaydī, al- 64, 76
wuqūʿ al-šarṭ 161

The Foundations of Arabic Linguistics IV The Evolution of Theory PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Foundations of Arabic Linguistics IV The Evolution of Theory PDF

Uploaded by

Copyright:

Available Formats

The Foundations of Arabic Linguistics IV

Aaron D. Rubin and Ahmad Al-Jallad

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

Manuela E.B. Giolfo

Copyright 2019 by Koninklijke Brill NV, Leiden, The Netherlands.

This book is printed on acid-free paper and produced in a sustainable manner.

Notes on Contributors vii

Introduction: the Evolution of Theory in the Arabic Linguistic Tradition 1

Contribution to a Modern Reading of Sībawayhi 10

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

Grammar for Beginners and Ibn Hišām’s Approach to Issues of ʾiʿrāb 61

Sallaṭa/tasallaṭa, a Possible Parallel for ‘Govern’? 89

The Intriguing Issue of Dictionary Arrangement in Medieval Arabic

Can Ambrosiana X Sup 56 Improve Our Understanding of Sībawayhi’s

Conditionality: Syntax and Meaning in al-Sīrāfī and Ibn Sīnā 157

How to Parse Effective Objects according to Arab Grammarians? A Dissenting

The Phenomenon of ittisāʿ al-kalām in Old Arabic 212

Which Verbal Nouns Can Function as Adverbial Accusatives of State or

What is Definiteness in Arabic? Focusing on Proper Nouns for Genera and

Definition and Determination in Medieval Arabic Grammatical

The Concept of tawṭiʾa in the Medieval Arabic Grammatical Tradition 274

Malay Grammar between Arab and Western Model 295

Manuela E.B. Giolfo

in Arabic grammar and other fields. He teaches at various academic institutions

The Evolution of Theory in the Arabic Linguistic

Manuela E.B. Giolfo and Kees Versteegh

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

bic Linguistics (Sadan 2018), and the topic of (in)definiteness is an important

known to him, that of Arabic grammar. He even went so far as to introduce

Genoa and Batenburg, June 2018

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

a synthesis of class and order by abstracting universal properties of objects

2 The Arabic Grammatical Tradition: the Heritage

When we look at the landscape of Arabic linguistics, Sībawayhi’s Kitāb rises

3 Contemporary Distributional Analysis

The distributional approach in contemporary linguistics was proposed by Leo-

On constate qu’une analyse distributionnelle au sens strict du terme

These procedures also do not constitute a necessary laboratory schedule

4 The Distributional Method of the Arabic Grammatical Tradition,

The Arabic grammatical tradition is called by its authors ʿilm al-ʿarabiyya,

stituents in a compositional way. Besides similarity analysis at the form level of

one of the most striking features of Sībawayhi’s analysis is that it con-

Coalescence or fusion in the sense of merging two linguistic elements,

Rediscussing all concepts making up the sophisticated apparatus of Sībaway-

5 The qiyās is Hierarchized

Sībawayh seems to be aware of the notion of distribution and environ-

À travers toutes ses variétés, la langue arabe maintient l’ usage de types

Nevertheless, group structure is still insufficient since it defines a closed static

In modern mathematics, Category Theory studies mathematical structures. It

D’une façon générale ce que Z. Harris demande aujourd’hui aux math-

egory Theory constitutes an important epistemological turn regarding struc-

Le structuralisme en mathématique comme en linguistique, s’ il propose

The convergence between distributional analysis and Category Theory may

In mathematics, I claim, we do not have objects with an internal com-

Qiyās is a mathematical method which succeeds in describing language in a

in such a vision objects are not considered as members of a structure, but as

usage in Arab grammatical theory”. Approaches to Arabic linguistics, ed. by Everhard

Mosel, Ulrike. 1980. “Syntactic categories in Sībawayhi’s Kitāb”. Histoire, Epistémologie,

As early as 1946, the denominations of personal pronouns by the Arab gram-

1 Benveniste (1966: I, 252).

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

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

As the beginning of every enunciation is always the vocative, unless you