Studies in Semitic Philology 9004047433, 9789004047433

Studies in Semitic Philology

273 37 60MB

English Pages 612 [607] Year 1997

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Studies in Semitic Philology
 9004047433, 9789004047433

Table of contents :
Title Page......Page 2
Copyright Page......Page 4
Table of Contents......Page 5
Preface......Page 9
A. PHONOLOGICAL AND MORPHOLOGICAL STUDIES......Page 11
1. The Vowel i as an Auxiliary Vowel; an Investigation of the Relation between Accent and Vowel in Semitic Languages......Page 13
2. A Phonetic Law in the Judeo-Arabic Dialect of Baghdad: a > ∂ in Pretonic Closed Syllable......Page 104
A. The Development of the Diphthongs aṷ, ai......Page 108
B. The Contraction of the Diphthongs ṷa, ia > ō, ē......Page 116
4. Bi-Consonantal Nouns of Roots III w ('ab, 'aḫ, ḥam)......Page 134
5. A Case of Quantitative Ablaut in Semitic (Broken Plurals and Related Forms)......Page 141
6. On Two Cases of Consonant Change in Modern Arabic Dialects......Page 164
7. Hebrew štayim ('two'), Syriac štā ('six') and a Turkic Analogue......Page 168
8. Concerning the Border-Line between Consonant and Vowel......Page 170
9. The Hebrew Perfect Forms qāṭlā, qāṭlū......Page 175
10. The 3rd Pers. Sing. Fem. of the Perfect of Roots III y/w in Arabic
......Page 178
11. The Aramaic Nomen Agentis qātōl and Some Similar Phenomena of Arabic......Page 181
Supplement: The Development of Aramaic gālūṯ (ā)......Page 185
12. The Plural-Ending -ūt of Masculine Attributive Adjectives in Akkadian......Page 191
A. The Object-Pronouns Formed with īyā-......Page 192
B. The Relative-Pronoun allāḏī......Page 195
C. The Demonstrative Pronoun ḏālika and Related Forms......Page 201
14. The Forms of the Imperative (and Jussive) in the Semitic Languages......Page 205
15. The Semitic Causative-Prefix š/sa......Page 210
B. SYNTACTIC AND SEMANTIC STUDIES......Page 217
16. Genetic Aspects of the Genitive in the Semitic Languages......Page 219
Supplement: The Development of the Infinitivus Constructus of Hebrew and Related Phenomena......Page 235
17. The Syntactic Background of Semitic Nouns with Prefix ma- and of Participles with Prefix mu-......Page 242
18. The Infinitive in the Function of 'Psychological Predicate' in Syriac......Page 271
A. The Construction mā rā'ahū illā bi......Page 277
B. The Preposition ilā Implying the Concept 'suddenly'......Page 301
20. Some specific Forms of Hypotaxis in Ancient Arabic......Page 305
21. Syriac dalmā 'lest', 'perhaps' and Some Related Arabic Phenomena......Page 332
22. Arabic lākin (na) and Related Expressions......Page 348
23. The Origin of Arabic ba'da 'after'......Page 353
24. An Arabic Sentence-Type Expressing 'Inner Compulsion'......Page 357
A. Significant Use of the Possessive Pronoun......Page 367
B. Various Semantic Characteristics of the Concept 'to have'......Page 375
26. Arabic Parallels to the English Phrase I am Friends with Him......Page 384
27. Expressions Based on the Noun yawm- 'Day'......Page 400
Supplement: The Semantic Mechanism of the Development of Arabic yawma 'when' (Conjunction) to an 'that' (= German 'dass')......Page 438
28. Arabic aslama (Islām) and Related Terms......Page 444
29. The Origin of the Principle of Iṣmah: Muḥammmad's "Immunity from sin"......Page 465
30. The "Completion" or "Improvement" of a laudable Deed: An ancient Arab ethical Motif......Page 470
A. Akkadian qarābu 'to be offered as a Sacrifice' and Related Expressions......Page 475
B. The Arabic Type iḏā birağulin 'suddenly there was there a man'......Page 487
32. The Onomatopoetic Origin of Some Terms for the Concept 'Sudden'......Page 493
33. 'Satisfying' and 'Restraining': Arabic kafā (kfy) > kaffa......Page 501
A. Arabic ma'tam......Page 510
B. Arabic ayyim 'Widow'......Page 515
35. Akkadian kipru (um), Pl. kiprātu (m) and Ethiopic kanfar......Page 518
36. Aramaic msar, Neo-Hebraic māsar 'to surrender (some-one)'......Page 523
37. An Arabic Parallel to benedicere......Page 527
38. North-Semitic hayyīm/n 'Life' in the Light of Arabic......Page 535
39. The Biblical Concept "The Treasure of Life" and Its Survival in Mandaean and Christian Doctrines......Page 543
40. The Root hwy 'to be', a Proto-Semitic Verb......Page 550
41. Arabic 'ad'ana......Page 554
42. The Origin of Arabic ġayr 'other (than)'......Page 555
43. Hebrew raq 'only' < *raiq......Page 557
44. An Arabic Cognate of Ugaritic ġmr-m......Page 560
45. The Non-Technical Origin of the Arabic Rhetorical Term tawriyah......Page 562
46. The Arabic (and South-Semitic) Root baḥuta 'to be pure, unmixed'......Page 564
47. The Origin of Aramaic dmā 'to be like, to resemble'......Page 566
48. Verbs derived from the Noun *qaš/st 'bow'......Page 569
49. The Use of the perfect Tense in the Conditional Sentences of Arabic......Page 573
50. The Early-Arabic Term āfat- "Damage", Used in the Sense of "Sin"......Page 582
51. Alqamah's Warning of the Approach of an Enemy......Page 584
52. The Original Meaning of "... a Man leaves his Father and Mother ... " (Genesis 2,24)......Page 603
Addendum to Chapter 25......Page 606

Citation preview

STUDIES IN SEMITIC PHILOLOGY

STUDIES IN SEMITIC LANGUAGES AND LINGUISTICS EDITED BY

G.F. PUPER Emeritus Professor in Arabic Language and Literature in the University of Amsterdam

VI M. M. BRAVMANN

STUDIES IN SEMITIC PHILOLOGY

LEIDEN E. J. BRILL 1977

STUDIES IN SEMITIC PHILOLOGY BY

M. M. BRAVMANN

LEIDEN

E. 1. BRILL 1977

ISBN 900404743 3

Copyright 1977 by E. J. Brill, Leiden, Netherlands All rights reserved. No part of this book may be reproduced or translated in any form, by print, photoprint, microfilm, microfiche or any other means without written permission from the publisher PRINTED IN BELGIUM

CONTENTS Preface .

XI

A. PHONOLOGICAL AND MORPHOLOGICAL STUDIES

1. The Vowel i as an Auxiliary Vowel; an Investigation of the Relation between Accent and Vowel in Semitic Lan~q~

.

2. A Phonetic Law in the Judeo-Arabic Dialect of Baghdad: a > a in Pretonic Closed Syllable . 3. Some Aspects of the Development of Semitic Diphthongs: A. The Development of the Diphthongs ay, aj . B. The Contraction of the Diphthongs ya, ja > 0, e . 4. Bi-Consonantal Nouns of Roots III w ('ab, 'ab, ~am) 5. A Case of Quantitative Ablaut in Semitic (Broken Plurals and Related Forms) . 6. On Two Cases of Consonant Change in Modern Arabic Dialects . 7. Hebrew stayim ('two'), Syriac stii (,six') and a Turkic Analogue 8. Concerning the Border-Line between Consonant and Vowel . 9. The Hebrew Perfect Forms qiirlii, qiirlii . 10. The 3rd Pers. Sing. Fern. of the Perfect of Roots III y/w in Arabic . 11. The Aramaic Nomen Agentis qiitOi and Some Similar Phenomena of Arabic Supplement: The Development of Aramaic giiliij( ii). 12. The Plural-Ending -iit of Masculine Attributive Adjectives in Akkadian 13. The Origin of Some Arabic Pronouns. A. The Object-Pronouns Formed with iyiiB. The Relative-Pronoun alliigi . C. The Demonstrative Pronoun giilika and Related Forms 14. The Forms of the Imperative (and Jussive) in the Semitic Languages . 15. The Semitic Causative-Prefix s/sa .

3

94 98 98 106 124

131 154 158 160 165 168 171 175

181 182 182

185 191 195 200

VIII

CONTENTS

B. SYNTACTIC AND SEMANTIC STUDIES

16. Genetic Aspects of the Genitive in the Semitic Languages Supplement: The Development of the Infinitivus Constructus of Hebrew and Related Phenomena . 17. The Syntactic Background of Semitic Nouns with Prefix ma- and of Participles with Prefix mu- . 18. The Infinitive in the Function of 'Psychological Predicate' in Syriac 19. The Expression of Instantaneousness in Arabic (with Attention to the Corresponding Usage in Western Languages) . A. The Construction ma ra'ahu illa bi . B. The Preposition ita Implying the Concept 'suddenly' 20. Some specific Forms of Hypotaxis in Ancient Arabic. 21. Syriac dalma 'lest', 'perhaps' and Some Related Arabic Phenomena. 22. Arabic lakin (na) and Related Expressions . 23. The Origin of Arabic ba'da 'after' . 24. An Arabic Sentence-Type Expressing 'Inner Compulsion' 25. The Idea of 'Possession' in Linguistic Expression . A. Significant Use of the Possessive Pronoun . B. Various Semantic Characteristics of the Concept 'to have' 26. Arabic Parallels to the English Phrase I am Friends with Him 27. Expressions Based on the Noun yawm- 'Day' . Supplement: The Semantic Mechanism of the Development of Arabic yawma 'when' (Conjunction) to an 'that' (= German 'dass') 28. Arabic aslama (Islam) and Related Terms . 29. The Origin of the Principle of 'I~mah: Mul:mmmad's "Immunity from sin" . 30. The "Completion" or "Improvement" of a laudable Deed: An ancient Arab ethical Motif . 31. Semitic Instances of "Linguistic Taboo" . A. Akkadian qarabu 'to be offered as a Sacrifice' and Related Expressions . B. The Arabic Type ida biragulin 'suddenly there was there a man' .

209 225 232 261

267 267 291 295 322 338 343 347 357 357 365 374 390

428 434 455 460 465 465 477

CONTENTS

32. The Onomatopoetic Origin of Some Terms for the Concept 'Sudden' . 33. 'Satisfying' and 'Restraining': Arabic kafa (kfy) > kaffa 34. Arabic ma'tam 'Mourning Assembly' and Related Etyma A. Arabic ma'tam . B. Arabic ayyim 'Widow' . 35. Akkadian kipru (um), Pl. kipratu (m) and Ethiopic kanfar 36. Aramaic mesar, Neo-Hebraic masar 'to surrender (someone)' 37. An Arabic Parallel to benedicere . 38. North-Semitic hayyim/n 'Life' in the Light of Arabic 39. The Biblical Concept "The Treasure of Life" and Its Survival in Mandaean and Christian Doctrines 40. The Root hwy 'to be', a Proto-Semitic Verb 4l. Arabic 'ad'ana 42. The Origin of Arabic gayr 'other (than)' 43. Hebrew raq 'only' < *rajq . 44. An Arabic Cognate of Ugaritic gmr-m 45. The Non-Technical Origin of the Arabic Rhetorical Term tawriyah 46. The Arabic (and South-Semitic) Root ba~uta 'to be pure, unmixed' 47. The Origin of Aramaic dema 'to be like, to resemble' 48. Verbs derived from the Noun *qas/st 'bow' 49. The Use of the perfect Tense in the Conditional Sentences of Arabic 50. The Early-Arabic Term afat- "Damage", Used in the Sense of "Sin" 51. 'Alqamah's Warning of the Approach of an Enemy . 52. The Original Meaning of "... a Man leaves his Father and Mother ... " (Genesis 2,24) Addendum to Chapter 25 .

IX

483 491 500 500 505 508 513 517 525 533 540 544 545 547 550 552

554 556 559 563 572 574 593 596

PREFACE The present volume includes new studies as well as studies which were previously published in various periodicals. The latter are here generally presented in a revised and, in part, enlarged form. Furthermore, some essays which were originally published in German or Hebrew, are now presented in revised English versions. The researches as a whole deal with various aspects of Semitic philology. Whereas a major part of the papers discuss basic problems of the historical and comparative grammar of the Semitic languages, a considerable number are devoted to semantic topics. Among those essays dealing with semantic problems, there are various studies in which the development of certain religious or social concepts forms the center of the discussion. Some of these are to be considered as a supplement to my book The Spiritual Background of Early Islam; Studies in Ancient Concepts (Leiden, E. J. Brill, 1973);1 mention especially chapters 28-30 (p. 434-464) and ch;tpters 50-51 (p. 572-592). As to the copious study "The Vowel i as an Auxiliary Vowel", published as chapter 1 (p. 3-93), I am deeply grateful to my friends Dr. Mark Jupiter (Columbia University Library) and Dr. Kurt S. Maier (Leo Baeck Institute, New York) for having translated its original German text into English. The publication of these Studies was necessarily a complicated enterprise in various respects. It was facilitated through the financial support granted to me by the following institutions, to which I extend my heartfelt thanks: The Council for Research in the Humanities of Columbia University, and The American Academy for Jewish Research. Thanks are also due to Dr. F. Th. Dijkema, Oriental Editor of E. J. Brill, for the devotion he has shown in seeing the manuscript through the various stages of proofreading to its bookform. New York, September 1976.

M. M.

BRAVMANN

A

PHONOLOGICAL AND MORPHOLOGICAL STUDIES

1 THE VOWEL I AS AN AUXILIARY VOWEL An Investigation of the Relation between Accent and Vowel in Semitic Languages 1

In Semitic languages, we observe very frequently the substitution of i for an original a, a phenomenon called 'attenuation' ('Verdiinnung') in older terminology. The conditions for this substitution have not been accurately established so far. I would like to remark at the outset that I have observed i replacing not only a, but any other vowel, of any colouring, and that the replacement of a with i should be studied in connection with the replacement with i of other vowels. The replacement of a with i in the Hebrew language is well-known. H. Bauer and P. Leander, Historisehe Grammatik des Hebriiisehen, p. 193 V, describe it as follows: "a in geschlossener, druckloser Silbe ist in grosser Ausdehnung zu i geworden". Bergstrasser, Hebr. Gramm., I, §26 b (p. 146) is content with a similar statement, but adds that this transition has taken place under conditions which can no longer be determined. Brockelmann, Grundriss, I, §52 ga (p. 146), explains this sound shift, which also occurs in Aramaic, by having "the colouring (Farbung) of short a in closed syllable" be "determined by the adjacent consonants". The irregularity of this replacement of a with i can, according to Brockelmann, probably in part be explained by the assumption "dass man fUr schwebende Nuancen bestimmte Grenzwerte festlegte". Bauer-Leander, I.e., p.94x, follow Brockelmann here. Bauer-Leander and Bergstrasser differ as to the time when this shift of a to i took place. Whereas the former (see I.e.), in my opinion correctly, believe to be able to prove this sound-shift

1 This is a slightly revised version of an original German text published in Le monde oriental, vol. 32 (1938), p. 1-102 (with corrections Ibid., p. 134-135). The original German version was discussed by J. Cantineau in his article "La voyelle de secours i dans les langues semitiques" (in Semitica, Cahiers publies par l'Institut d'etudes semitiques de I'Universite de Paris, II, 1949, p.51-67). Obviously, Cantineau, in spite of his criticisms of my comments on his own views, in general accepted my theory of Hi as auxiliary vowel in Semitic" (cf. below p. 15, n. 12).

4

THE VOWEL I AS AN AUXILIARY VOWEL

already for the period of the Amarna-Ietters, the latter (l.c., §26 c; p. 147) calls it "a very recent linguistic phenomenon". In order to explain the nature of this phenomenon, which, in my opinion, has been occurring in the Semitic dialects from the earliest times to this day, I would like to discuss certain observations which I have made in connection with various statements of medieval Arab phoneticians (quoted and discussed in my study M aterialien und Untersuchungen zu den phonetischen Lehren der A raber, dissertation of Breslau University, Gottingen 1934). In this connection a discussion of the forms of the imperative of the 'basic stem' of the verb in Hebrew becomes necessary. In Hebrew, the forms of the imperative whose endings are constituted by vowels only, characteristically differ from the forms without an ending or with endings starting with a consonant: qitli (sing. fern.), qitlu (pI. masc.); but qetol (sing. masc.), qetolnii (pI. fern.); or also ziqnii (energeticus), ziqni, ziqnu, beside zeqan, zeqannii. Brockelmann, Grundriss, I, §258 B i ~ (p. 548), following Barth (ZDMG, v. 43, 183), who first advocated the now prevailing view, comments on this phenomenon: "1m Imperativ hat sich i in den mit vokalischen Afformativen gebildeten Personen auf Kosten von u ausgebreitet (falls hier nicht etwa uralte Ablautsverhaltnisse bewahrt sind, s. Bb) ... ". Bauer-Leander, p. 305 i, write: "Schliesslich sind beim u-Imp. die Kontextformen des F. Sg. und des M. PI. umgebildet worden, und zwar entweder nach dem a-Imp., also zu *qaJali, *qatalU, oder vielleicht nach dem noch lebendigen i-Imp., in diesem Falle zu *qitili, *qitilu".2 Bergstriisser, l.c., II, §141 (p.81) voices a similar view: "i in er~ter Silbe erkliirt sich teils als 'Verdunnung' von a (Gr., 26b), teils vielleicht als Rest des i-Imper. *qitil". Barth (l.c., p.182) recalls the views of older grammarians: "Dieses auffallige i transitiver Imperative ist bislang noch ganz unerkliirt. Man hielt es teils nach veralteter grammatischer Auffassung fUr einen Hilfsvokal oder fUr eine Umbildung aus dem regelmiissigen u, aus dem es 'abgeschwiicht' (Olshausen, §234 a; Ges.-Kautzsch, §46,2, Anm.2) oder 'verdunnt' (Stade, §96 a) oder 'zugespitzt' (Konig, §20,12) sein sollte". Concerning the 'auxiliary vowel' -theory, Barth remarks (Ibid., n. 3): "Ewald (§226 d) nach der Theorie der mittelalterlichen judo Grammatiker, dass aus qetol mit Suffix *qerli und daraus wegen der Auf2 As to details about further phonological developments which. in this view, finally resulted in the present forms, see Ibid., p. 306j.

THE VOWEL I AS AN AUXILIARY VOWEL

5

einanderfolge zweier Schwa mob. qitli wiirde. Diese mit Recht von Philippi bekampfte Auffassung widerlegt sich allein schon durch Formen wie q9r/:zi, '9lzi, wo das (0 » 9 verblieben ist".3 A study of similar phenomena in modern Arabic dialects proves that we are dealing here with (relatively recent) secondary changes of phonetic nature, which seems to be plausible from the outset. L. Bauer, Das palast. Arabisch, 4. Aufl. (1926), p. 23, gives the following imperative forms of the root catab (class. kataba) 'to write' in the dialect of the Fella]:lin: m. sg. iCtib, f. sg. icitbi, m. pI. iCitbu, f. pI. iCtibin. For the urban dialect, he lists iktib, iktibi, ikitbu, and (Ibid., n. 1), as variants of the latter two forms: ikitbi, ikitbu, referring also to Ibid., § 16,2 (p. 16), where he mentions similar shifts in the corresponding forms of the imperfect. In the dialect of Tripoli (Lybia) we also find the following forms (cf. Stumme, Marchen und Gedichte aus Tripolis, p. 228): sg. 2. m. sg. 2. f. pI.

ektib ektibi ektibu

ekitibi ekitibu

ekitbi ekitbu

ektbi ektbu

These variants are also found in all forms of the imperfect which terminate in a vowel, as, e.g., in the 3rd pers. pI.: jektibu, jekitibu, jekitbu, jektbu. It seems important to state that in the dialect of 'Oman, which has given up the prosthetic vowel in the imperative forms (as well as in other forms), the forms of the imperative are more or less identical with the corresponding forms of Hebrew (also here, as in other Arabic dialects, the corresponding forms of the imperfect display analogous changes); cf. Reinhardt, Ein arabischer Dialekt gesprochen in 'Oman und Zanzibar, §274 (p. 151):

3 Of all explanations mentioned so far, the one offered by medieval Jewish grammarians and mentioned by Barth among others seems to be the most plausible one: *q'{'li « *q'rolf, with reduction of the u (or 0, respectively) preceding the stressed syllable, a reduction which is quasi in agreement with the general phonetic rule). For it is the only one which derives these problematic forms not from residues of other forms, but explains them phonetically which should always be tried first. However, this 'phonetic' explanation is, apart from other reasons, unacceptable simply because the Hebrew forms cannot be separated from the corresponding modernArabic forms mentioned below; and the modern-Arabic dialects neither show a reduction of short u and i, as does the Hebrew and the Aramaic, nor do the final vowels in question ordinarily bear any special stress.

6

THE VOWEL I AS AN AUXILIARY VOWEL

sg. sg. pI. pI.

m. f.

m. f.

ktub kitbi kitbo 4 kitben

The only difference between the Hebrew and the 'Oman! forms is that in the dialect of 'Oman this development has also brought about a transformation of the plur. fern. form, which is not the case in Hebrew and in Palestinian-Arabic. In the Palest. dialect the form is ietibin. As concerns Hebrew, the forms to be mentioned are (rare) forms like §ema'an (Gen. 4,23), which occur along with the common forms k elobnii, sema'nii. The development of §ema'an is clear: After the disappearance of the terminal vowel in the original §ema'nii, the form ended in a double consonant. This was unacceptable, and the terminal cluster was made pronounceable by a gliding vowel: §ema'nii > *§ema'n > §ema'an. In the dialect of 'Oman (in contrast to Hebrew and Palestinian-Arabic) this gliding vowel in the terminal cluster became an actual vowel (,Positionslaut') forming an independent syllable; it could therefore attract the accent and form the basis for a change in the syllable structure of the word (cf. below p. II): *ktubna > *ktubn > *ktuben > kitben. We are dealing here with forms characterized by 'break-up' (,Aufsprengung') of clusters, forms which very frequently occur in modern Arabic, especially Maghrebian, dialects (cf. Stumme, l.c., p. 220 ff.;

Tunis. Gramm., p.5; Tunis. Miirchen und Gedichte, Introduction, p. XXXVI ff.). The 'break-up' ('Aufsprengung') is thus a phenomenon which did not first occur in modern dialects, but already appears in Ancient-Semitic. The forms of the Hebrew imperative must, of course, not be considered apart from those of the Arabic dialect of 'Oman with which they are entirely identical. The color of the inserted vowel in the dialect of 'Oman (as in other dialects) may sometimes be influenced by the surrounding consonants (and therefore is not always i), as, e.g., sing. m. qfil, sing. f. qujli, etc.; q~am, qa~mi, etc.; lbes, lebsi, etc. (cf. also the corresponding forms of the dialect of potar (.?:tar): Rhodokanakis, Der arabische Vulgiirdialekt im po/iir, II [Sudarab. Expedition d. Wiener Akad., Bd. X, 1911], §63 c, d). One cannot, however, maintain an absolute dependence of the shade of 4 The plural-form of the imperative in the dialect of Tlemsen is identical with this 'Oman! form; see Mar~ais, Le dialecte arabe parle a Tlemcen, p. 61: ketbu beside ekketbu, as also kleb beside ekleb.

THE VOWEL I AS AN AUXILIARY VOWEL

7

the inserted vowel on the specific nature of the surrounding vowels (as does Reinhard, I.e., §268); rather i must be considered as the quasi basic (or natural) shade of this vowel, which is essentially independent of surrounding sounds (cf. below p. 47ff.). The processes which led to these 'break-ups' of clusters of consonants may be assumed as follows: The basic imperative forms, with suffixes constituted by a vowel, were: *q!uli, *q!ulu (> Aramaic: qe!uli, qe!ulu, the same in Proto-Hebrew), or *qe!oli, *qe!olu, respectively, forms which underlie the Hebrew pausal forms qe!Oli, qe!olu, in the same way as the context-form s. m. qe!ol forms the basis of the corresponding pausal form qe!ol (cf. Arabic (u)ktubi, (u)ktubu, and modern-Arabic iktibi, iktibu, respectively). These forms end in two successive open syllables. Basically, these two open syllables received an equal amount of stress. From the outset neither could require a stronger stress than the other. At first, the two syllables may have been spoken with a 'gliding' ('hovering') accent Cschwebender Akzent') (cf. below p. 12). However, it is difficult to maintain a 'gliding' ('hovering') stress, especially on open syllables, that is, to speak two adjacent open syllables with the same amount of stress. The difficulty may in such cases be removed by speaking one of the two syllables with reduced stress. The terminal vowel of the (personal) affix could, however, not be weakened, or, worse, undergo disappearance; its formative or morphological function required its unequivocal and unattenuated retention. This retention could only be secured by means of a strong and unequivocal stress. However, this could only be done at the expense of the stress on the adjacent syllable. Since the first of the two syllables began with two consonants and thus necessarily had considerable weight, the difficulty of two accents colliding with one another could not be removed here by reducing the stress on this first syllable. The only possible solution was the creation of a distance between the two stresses by removing the stress to the initial cluster (complex of two consonants), that is: *q/ul[ (*qe!u/t) > *qlulz (*qe!ull); *qlu/u (*qe/u/u ) > *qlu/u (*qe!u/u). This necessarily resulted in the 'break-up' ('Aufsprengung') of the cluster by a newly created vowel; simultaneously, the impact of this abruptly retreating stress effected an elision of the vowel of the following syllable which was stressed originally: > qilli, qit!u. This holds true both of Hebrew and the modern Arabic dialect of ·Oman. In the case of the related imperative forms with prosthetic vowel (and the corresponding forms of the imperfect), like Tripolitan ekitbu

8

THE VOWEL I AS AN AUXILIARY VOWEL

(see above p. 5), the process was somewhat different. This form originated, without any influence of the accent, from the parallel form ektbu (which on its part originated from another variant: ektibu, or rather from the ancient form uktubu [< *iktubU] through elision of the middle vowel which appeared between two accents) through break-up CAufsprengung') of the inconvenient cluster of consonants: ektbu > ekitbu. Therefore, the auxiliary vowel created here remained unstressed in spite of its position in a closed syllable. 5 The vowel which breaks-up the cluster of consonants is the vowel i. We are dealing here with an entirely new auxiliary vowel, to be compared with the 'prosthetic' vowels. As I already suggested in my Materialien und Untersuchungen (see above p. 4), p. 96, in connection with statements of medieval Arab phoneticians, such original auxiliary vowels have always i-coloring (an affinity to the vowel i) and are probably always voiceless (whispered) vowels at the outset. As to the reasons for it, see below p. 47ff. and p. 56ff. There are, however, in Hebrew forms of the imperative with the suffixes -i and -u or other endings, in which the same syllable structure is present as in qi!li, qi!lu, in which, however, the i of the first syllable (e.g., qi!li) is replaced by 9 < o. We have, e.g., m91!si 'rule' (sing. fern.), '9lzi 'rejoice' (sing. fern.), m9s!su 'draw' (pI. masc.), beside mis!su. The cohortative-form of the imperative is q9t/ii, beside less common instances with i, like mi!sf{l ('sell !'). Brockelmann, Grundriss, I, §258 B, i, ~ (p. 548), Bauer-Leander, p. 3061, consider these forms as "the old forms", and· thus make them originate from a supposed 5 The form ektibu (which is the basis of ektbu) itself-in spite of the identity of its syllable structure with that of ancient-Arabic uktubu-is not to be considered identical with that latter form. Rather, the vowel i in the second syllable of ektibu (adjacent to the labial b i), corresponding to u in the old form (cf. also Man;;ais, ZA 27, p. 25), should be interpreted as an auxiliary vowel arisen in the course of the following development: uktuM (respectively *iktuM, with an original, or newly developed, prosthetic i-vowel, see my Materialien und Untersuchungen, p. 88 and n. I) > *Ektubu, > iktbu, and now, with insertion of the auxiliary vowel (instead of before the 2nd radical as in Ekitbu, see above) before the 3rd radical, > Ektibu. With this twofold possibility of the insertion of the auxiliary vowel (before the 2nd as well as before the 3rd radical) we probably have also to connect the (imperative-) form ekitibu (and the corresponding forms of the imperfect). In this context we should also pay attention to the following statement by M. Cohen, Le parler arabe des Juifs d'Alger, p. 183: ..... c'est Ie cas de jik'tbu; assez souvent on peut la (i.e., the explosion of the stops) percevoir comme une voyelle extn!mement reduite, et on a l'iIIusion d'une voyeIIe avant ou apn:s la 2" radicale, sans qu'on arrive it determiner sa place. (Ce doit etre un etat analogue que Stumme a constate it Tunis; Tun. gr., § 12 c; il distingue quatre cformes au pluriel de I'imparfait, dont la derniere jikitibu parait bien repondre au jik 't bu note ici)".

THE VOWEL I AS AN AUXILIARY VOWEL

9

ancient form of the imperative, *qutul, with retained vowel of the first syllable. 6 This seems unwarranted to me (see above p. 5, and cf. n. 3). Rather, the sing. m. form with vowelless first radical must be considered as the basic form (see above p. 6, and cf. below p. 198). Accordingly, the first syllable in these forms should be considered as based on an original cluster which was affected by a 'break-up' ('Aufsprengung') caused by an abrupt retreat of the accent. This auxiliary vowel (effected by the 'break-up') is not always i in the modern Arabic dialects, as we would expect. Here the auxiliary vowel can be changed by the influence of the adjacent consonants (see above p. 6 and p. 31 ff.). We must not assume that for the instances of Hebrew. Rather, here the vowel must be explained on purely psychological grounds: When, as a consequence of the abrupt retreat of the accent, it became necessary to interject a new vowel, one chose the nuance of the vowel of the verbal stem, which (in spite of its elision in consequence of the accentual shift) was strongly rooted in the linguistic consciousness. Accordingly - in disagreement with the actual development - the process has the appearance of a metathesis of the vowel concerned (that is: the vowel 0 in *qo!li (> q9!li) seems to be based on a transposition of the vowel 0 « u) in the primary form *q!oli < *q!uli). This process is also observable in Arabic dialects. One should, however, keep in mind that we are not dealing with a genuine phonetic process. Most comparable to these Hebrew imperative forms are the imperative forms of the Arabic dialect of Palmyra. Here the forms with u as the vowel of the stem (see J. Cantineau, Le dia/ecte arabe de Palmyre, I, p. 136) are the following: sing. masc. 'oktob, but sing. fern. 'okotbWe i , pI. masc. 'okotbo", pI. fern. 'okotbWen", from *'uktubi> *'uktbi > *'ukutbi > 'okotb"e i , etc. (analogous forms also in the type with i > e as vowel of the stem, in contrast to the type with a, where the original syllable structure remained intact). Here too we cannot speak of an influence by adjacent consonants; rather, the u-nuance of the vowel arising from the 'break-up' of the consonantal cluster (basically appearing as i) was determined by the prosthetic vowell! which in turn developed (instead of the elementary i) by influence of the original u of the stem (see my Materialien und Untersuchungen, p.88, and n. 1). In this manner, the original stemvowel, as in similar cases in Hebrew, was retained, but in a position 6 As a matter of fact, such a form cannot be considered the primary pattern of the Semitic imperative. See below p. 198.

10

THE VOWEL I AS AN AUXILIARY VOWEL

different from its original position. Naturally, this is also valid for the forms of the Hebrew imperative sing. masc. with object suffixes, such as q9!leni, q9/lenu, etc. These are not based on an imperative *qu/ul (without suffixes) with the stress on the first syllable-as BauerLeander, p. 399 t maintain-but on the form if/ol « *q/ul; cf. Arabic uqtul; with suffixes: uqtUlni, etc., here without connective vowel), thus corresponding in their development to the plural forms with suffixes: qi!luni, qi!lunu, etc. (Concerning *qtul as the basic form of the imperative [developed from taqtul] see below p. 198ff.). As a matter of fact, in the case of Hebrew verbs with the imperfectvowel u > 0, like jiq/ol, we sometimes find also infinitive-forms with suffixes which are characterized by an i-vowel (after the first radical), in addition to such that are characterized by an o-vowel; cf. BauerLeander, p. 343": "Wie sich im Imp. qi/ta neben dem gewohnlichen qo/ta findet und qi/li, qit/u die alten Formen q9!li, q9/iU fast vollig verdrangt haben (§ 41 I), so tritt auch im Inf. bisweilen i fUr 9 ein: §i~ri 'mein Zerbrechen' ... , bi3rjo 'sein Abfallen' ... , ii/no 'sein Anklagen' ... , mi!sram 'ihr (m. pI.) Verkaufen' ... (aber m9!srah 'ihr [f. sg.] V.' ... ). - Vielleicht hat das Dissimilationsgesetz §21 k mitgewirkt". (Cf. also Barth, Nom.-Bldg., p. 104). But such a displacement of old forms with 9 by such with i can in the case of the infinitive no more be assumed than in the case of the imperative-forms. Rather, also here the basis for the forms with suffixes is always the form *qtul> qe/ol, and not the form *qu!ulu, as B.-L., p. 343 a, maintain. This form, qe/ol « *q!ol), was transformed, in the manner described, through the addition of vocalic suffixes. Also Barth (I.e., p. 100) has assumed if/ol « *q/ol < *q/ul) as basis, also before suffixes beginning with vowels, so that the vowel of the first syllable in Q9/lo, etc., is identical with the vowel 0 « u) originally standing between the second and the third radical. Yet I cannot go along with Barth when he introduces a form *qerlo (developed on the basis of regular phonetic rules) as an intermediate stage between *qe/olo and q9/lo (cf. above p. 4 concerning a similar interpretation of corresponding forms of the imperative, which in this case was rejected by Barth himself). Rather, the vowel of the second syllable was elided (without previous reduction to §ewd) by the development of the new vowel (i) between the two components of the initial consonantal cluster: *qe/olo > qi!lo (etc.). This insertion of the auxiliary vowel i and the subsequent elision of the vowel following the second consonant was caused by the accentual shift described above (see p. 7). Correspond-

THE VOWEL I AS AN AUXILIARY VOWEL

11

ing Hebrew infinitive-fonns with vocalic suffixes belonging to basic infinitive-fonns of the pattern tflal (to be derived from *qlal, associated with the imperfect-pattern yiqlal) have of course developed in the same way. This refers to fonns as si!5lziih, that is: se!5alz [associated with the imperfect jiskalz] augmented by the suffix of the 3rd pers. sing. fern. (as against Bauer-Leander, p. 344 d, who derived fonns like si!5biih, etc., from primary fonns [without suffixes] of the type *qatalu). As for other (recorded) fonns with suffixes of the same verb, that is: s9!5bo (suffix of the 3rd pers. sing. masc.), and s9!5b e!5ii (suffix of the 2nd pers. sing. masc.), they are to be ascribed to the influence (analogy) of the (extremely frequently used) corresponding fonns of the pattern qe,ol « *qlul), i.e., q9t/O, q91Ie!5ii, etc. As against the usual derivation of these Hebrew infinitive-fonns from basic fonns like *qulul- and *qalal-, it should be recognized that the basic pattern of the so-called infinitivus constructus of Hebrew is *qlul (*qlO/), *qlal, fonns which should be considered as developed directly from the associated imperfect fonns: jiqlol « *jiqlul) and jiqlal (for the details of this development see below p. 225). I now maintain that, in most cases where we find an i-vowel in place of an original vowel of another hue, 7 we are dealing with a vowel similarly interjected ('eingesprungen') as in the above-described cases, i.e., with a vowel which elementarily evolved through 'break-up' ('Aufsprengung') of a cluster of consonants. For the sake of simplicity, I will illustrate this with an example from the modern Arabic dialects. In the dialect of Tripoli, Lybia (see Stumme, Miirchen und Gedichte aus Tripolis, p. 228 and p. 215, 23,2), the fonns of the 3rd pers. fern. sing. and pI. comm. gen. of the perfect, as in other dialects, are kitbet, kitbu, fonns which in their syllable structure are rather distinct from the fonns of the other persons, such as kteb; ktebt, etc. « *katab < kataba; katabta).8 How should we explain the fonns with i?9 7 An auxiliary vowel i could also have been inserted instead of another i-vowel (which had been previously elided or reduced). It is, of course, difficult to trace instances of this type. 8 Compare, e.g., also the paradigm of the perfect in the dialect of 1;,)ofiir, Rhodokanakis, II, §61 m (p. 165 a):

Sg.

3. m. f. 2. m. f. I.

gtal gitlet gtalt gtalti gtalt

Pluf.

gitlew gitll!yn g( i) taltu

g(i)talten gtalna

12

THE VOWEL I AS AN AUXILIARY VOWEL

Stumme, I.e., §23,2 (p. 215/16) speaks at first of a "schwer erkliirbare Vorliebe" for i and u (concerning u see below note 9), but then he presumes that these vowels have originated "von verbis mediae vocalis i und u", an opinion that cannot be substantiated. If the vocalism of the perfect forms with i and u (which hardly exist any longer in the Tripolitan and in most other modern dialects) had exercised such an influence, this influence should appear in all forms (persons) of the paradigm concerned. Also in the cases now before us we are dealing with a genuine phonetic development from the original forms. In katabat, katabu, in no way different from the imperative forms ktitbi, ktitbU, there existed a sequence of syllables none of which surpassed the others in sonority and thus of necessity had to be more heavily stressed than the others. Accordingly, such forms were spoken with 'hovering' accent ('schwebender Ton').lo Socin, Diwan aus Centralarabien, III, p.229, with reference to Wetzstein, ZDMG 22,182, has observed a 'hovering' accent (,schwebender Tonfall') also in the speech of the Beduins of Central Arabia, and has stated that such an accentuation These I)fiiri-forms of the 3rd pers. sg. agree completely with Aramaix qe,al, qi,lal (cf. ~Iso Ibid., § I b, p. 68 b). -Cf. also Landberg, Etudes, I, Jfa¢ramout, p. 162, n. I concerning the word ~ [s-n-f-t] 'Ia femme s'est echappee': "Prononce sinfet par les l:la *sami'n > *smi'en > sum'en (with u instead of i [through the influence of the labial mJ). An analogous development is represented by the corresponding form of the Qfaridialect: gitleyn (with a certain transformation of the ending, which cannot be discussed in the present context). Cf. above p. 6 concerning the same transformation of the fem.-pl.-form of the imperative, uktubna. 9 In the case of certain specific roots, we find u, instead of the characteristic vowel i, in the Tripolitan forms of this type, as, e.g., 3rd pers. pI.: nf:zUrgu (3rd pers. sg. masc.: n~rt!g) < classical in~araqii, 'they were burned'. However, this vowel cannot, with Stumme, I.c., p.215, be considered equally original as i, but it is dependent (see already above p. 6, and below p. 76ff.) on the shades of the adjacent consonants. See also Ibid., §23,3, where Stumme, in the case of instances of a different syllable structure, states the replacement of an original a by i, without, in the case of these instances, variant forms with u being found. As will, however, be learned from the following discussion, both types, in spite of their different syllable structure, involve the same process. 10 The 'hovering' accent is perhaps also present in the word 'asarah in Landberg, Etudes, II, Dalina, p. 85. Ibid., n. 5, Landberg remarks concerning this form: "Les trois accents furent observes".

THE VOWEL I AS AN AUXILIARY VOWEL

13

is characteristic of "many modern Semitic dialects", e.g. also of the neo-Aramaic dialect of Tur 'Abdin. In connection with this, I also find interesting what he states in Grundriss d. iran. Philol., I, 2, p. 270 (§ 133), about the accentuation of Kurdic: "1m Grossen und Ganzen scheinen die samtlichen kurdischen Dialekte in Bezug auf den schwebenden Ton iiberein zu stimmen. Damit hiingt zusammen, dass sogar innerhalb eines und desse1ben Dialektes ein Wort bald mit dieser, bald mit jener Betonung vorkommt, oder dass die Auffassung des Tones eines von zwei verschiedenen Personen zugleich gehorten Wortes eine verschiedene sein konnte". I therefore imagine that the difficulty of equally distributing the stress to all syllables often caused the middle syllable to receive an excessive stress (a stronger stress than the two others): *katabat, *katabU. A result of this was then frequently the weakening of the syllable which preceded this middle, stressed syllable. This weakening of this syllable, though not easily determined in individual cases with respect to its degree, could eventually lead to the complete disappearance of the vowel: *qatalat > *qatalat, *qetalat, or > *qtalat, respectively. I I With respect to the original scheme of the forms, this strong emphasis of the second syllable, and with it the reduction of the vowel of the first, was of course in no way justified, and the linguistic feeling ('Sprachgefiihl') was not bound to it. In rhythmic respect, and from the point of view of syllable structure, forms like *qatalat, *qtalat; *qatalu, *qtalu (just as the imperative forms of Hebrew discussed above [po 7]: *qe!uli, *qe!ulu; qe/6li, qe/6Iu, respectively) were hardly allowable. A consonantal cluster followed by a sequence of two open syllables was less to be tolerated than a sequence of three open syllables. The 11 The succession of three open syllables in this form could, of course, also lead to an accentuation like qataldt, qatalil; and, in consequence of this accentuation, a reduction, or a complete disappearance, of the middle vowel, could occur: qataldt > *qatelat, *qatlat; etc. Perhaps the form 'amdet, in the following note by Landberg, Hafjramout, p.663, n. I, is thus to be interpreted: "Les J:iaribites disaient 'amadat {non pas 'amadet), tandis que les J:ia4ramites et 'awlaqites disaient 'amdet. Les voyelles se sont en general mieux conservees en BeyJ:ian-J:Iarib. On voit done qu'on ne saurait dresser des paradigmes fixes pour les dialectes ni pour un seul dialecte non plus". Accordingly, 'amdet could have developed on the basis of an accentuation like *'amaddt. Of course, this cannot be considered a certainty. According to the passage in Landberg, Ibid., p. 162, n. I (quoted above p. 12, n. 8), the J:Ia4ramites use (of another root) precisely the form qitlet as a correspondence of the form qatalet of the 'Awaliq and the BeyJ:ianites. Accordingly, it is more likely that also 'amdet corresponds to a form *qitlet, and that that its a-vowel is to be ascribed to the influence of the laryngeal' (cf. below p. 44 n. 47). As for the rest, the general remarks by Landberg merit high consideration.

14

THE VOWEL I AS AN A UXILIAR Y VOWEL

difficulty of a succession of three syllables, with each given equal stress ('hovering accent') could be alleviated by stressing the first and third syllables, giving one the main and the other a slightly weakened stress ('Nebenton'), and leaving the middle unstressed. This often resulted in elision or reduction of the middle syllable's vowel: qatalat > qatlat, but the word as a whole acquired an acceptable syllable structure. In forms, however, such as *qtalat, *qatalat; *qtalu, *qatalu, two strongly stressed syllables clash with one another; the first cannot be weakened because of the particularly strong stress borne by its initial consonantal cluster (see above p. 7), and the terminal syllable -at cannot suffer weakening, because of its specific grammatical function. Both stressed syllables had therefore to be detached from each other clearly and distinctly. This could possibly be achieved by gemination of the consonant separating the two syllables (vowels), a procedure which we see realized in certain similar cases. Cf. W. Man;ais, Le dialecte arabe des DIlld Briihim de Saiaa (Departement d'Oran), Paris 1908, p. 73: "C'est assurement a l'influence de l'accent qu'il faut attribuer Ie redoublement de la derniere con sonne dans Ie petit groupe des pluriels fflla, representant des classiques afilah, fa'ilah (cf. infra, pluriel brise)". Cf. also Ibid., p. 134. Correspondingly, a perfect form *qtalu could have become *qtallu, just as the ProtoHebraic imperative forms *q/uli, *q/ulu, which (like the plural form *filah < 'afilah) already originally had no vowel between the first two radicals, could have developed into *q/ulli, *q/ul/u (see below p.26). In our case the difficulty was solved, as for the imperative forms discussed above, by the accent of the syllable following the cluster of two consonants being thrown back onto the cluster itself. This resulted in the 'break-up' of the cluster by the interjection of a newly created, strongly stressed vowel i and in the elision of the vowel of the syllable following the new strongly stressed syllable: *qtalat, *qtalu > qitlat, qitlu. Simultaneously, there exists the possibility that the vowel between the first two radicals does not fully disappear in consequence of its loss of stress, but may only be reduced. (That in certain cases i has without doubt appeared in place of such a reduced vowel, and not in place of vowellessness, shall be shown with respect to other instances to be discussed below p. 25). Yet a reduced vowel in these cases obviously means hardly more than complete vowellessness (qtalat, qtalu), since the i-vowel which emerges here (qUlat), is not based on already existing phonetic substance, but

THE VOWEL I AS AN AUXILIARY VOWEL

15

represents, as it were, an entirely new creation. Apparently the tonal quality of the syllable in question is eliminated and destroyed by the process which we call 'reduction'; its acoustic impression and, what is most important, its retention in linguistic consciousness is so weakened, that its phonetic substance is no longer able to carry the abruptly retreating stress (qtalat > qitlat). No further proof is needed for my assumption that the i of the 3rd pers. fern. sing. of the perfect in Aramaic, qitlal, kilbal, is to be explained in the same manner.12 The form is completely identical with the (modern Arabic) Tripolitan (Lybian) kUbat, Pfari: gitlet, etc., just as the corresponding 3rd pers. masc. sing. of the Aramaic perfect, k ela/] , qetal, etc., agrees with the Tripolitan kteb, pfari gtcil. Cf. below p. 3 ff. This i has not arisen from the a of the completed form *qat/at after the vowel of the 2nd syllable had disappeared (as one generally assumes), but the substitution of a by i in the first syllable and the disappearance of the vowel of the second syllable are simultaneous results of one and the same process. 13 Accordingly, in Aramaic the development is: *qatalal >*qe tdlal or *qtdlal (with the stress on the second syllable of the stem as in the mascul.-form qetal) > qitlal. According to Sievers, Metr. St., I, p. 235, line 4 and line 26, Aramaic qi!lal is based upon "schwacher (oder

12 Wolfdietrich Fischer, in Festgabe for Hans Wehr (Wiesbaden, Otto Harrassowitz, 1969), p. 184, in dealing with this Aramaic form-qif/al (3rd pers. sing. fern. of the perfect)-and its Arabic analogies, suggests the same interpretation of these forms as the one given here (without reference to the original German version of the present study [Le Monde oriental, vol. 32 (1938), p. 14ff.). Also Fischer's explanations (I.e., p. 182-184) of the Syriac form qe!lel (1st pers. sing. of the perfect), and of the 'IraqiArabic forms gikar, kitab, kubar, etc. (3rd pers. sing. masc. of the perfect), as well as his treatment of Aramaic noun-forms such as nesmii, de/srii, besrii, and their Arabic parallels (see Fischer, I.e., p. 182-184) agree with the interpretations given here; see below p. 18, p. 33, p. 43; etc. (in the original German version: p. 17 ff., 33 ff., 45 ff.; etc.; and see also J. Cantineau, Semitiea, II (1949), pp. 54-55, 56, 59, etc. [see above p. 3, n. I]). 13 In principle the replacement of a by i might also have taken place in a closed (not only in an open) syllable-as we shall observe in a number of instances, discussed below p. 59 ff. - but also then not through that type of development by which this shift is ordinarily explained. Essentially, also in these latter cases (i.e., a > i in originally closed syllable) the i arising instead of the original a is a newly developed vowel as described here (cf. below p.20ff.). However, in the instances where a vowel has been elided between the consonants which follow this (newly developed) i, one has, as a rule, to assume that the replacement of a by i took place at the moment when this vowel was elided (the emergence of the vowel i being connected with the vowel elision), so that it is indeed an a in an open syllable which has been replaced by i.

16

THE VOWEL I AS AN AUXILIARY VOWEL

anderswie) betontes qataldt", while he assumes Hebrew qiirlii to have developed from the same form when it was more intensely stressed. 14 - Lagarde, Ubersicht, p.9, notes: "Die Syrer sprechen qit/al aus, wie die Araber in Xgypten von verbis mediae i (Spitta, §94 a) misket 'isqet biblet wildet sagen, das heisst: in qit/al ist die i-Bildung eingedrungen, wie in so viele dritte Personen masculini generis, die u oder i haben sollten, die a-Bildung". If an influence of the i-perfect (qatila) had been the cause, it should have affected the forms of all persons of the paradigm. Moreover, only two of the cited EgyptianArabic examples are based upon a Classical-Arabic i-perfect. For these modern-Arabic instances, as already stressed, the same interpretation is valid as for Aramaic qit/al; and it makes no difference whether the specific root concerned is an original i-perfect (i.e., of course, with the i after the second radical) or an original a-perfect. Barth, ZDMG 57,801, derives Aramaic qit/al and Hebrew qiirlii from the same primary form, namely from a form *qdtallit. Possibly the pattern on which Aramaic qit/al is directly (of course, not ultimately) based, is still preserved in forms such as Judeo-Aramaic gezaral 'she circumcised' (with the stress on the second syllable), that is, forms which in the 'neutral' [intransitive] basic conjugation, as, e.g., seteqal, are in the process of becoming predominant (from this type of forms, through the disappearance of the vowel of the femininesuffix, there developed Biblical-Aramaic *'amaral > *'amarl > 'amara! 'she spoke', etc., see Brockelmann, Grundriss, I, §262 by). According to Brockelmann (l.c.) these latter forms might also be 'new formations' (forms developed in analogy to the perfect forms retaining the stress on the second syllable). Yet, this does not seem very probable (cf. J. Cantineau, Bulletin d'Etudes orientales de l'Institut franrais de Damas, 1(1931), p. 94, at the top). J. Cantineau, in an article "Elimination des syllables breves en hebreu et en arameen biblique" (see Ibid., II [1932], p. 132), also traces qit/al back to a form *qatalal (with the accent on the second syllable); but the reasons assumed by him for this development do not seem to me very likely. Specifically he assumes that the vowel of the second syllable disappeared as a result of the shifting of the stress from the penultima to the ultima (*qatalal > *qataldt). It seems improbable to me that, in a form *qatalal (> qit!aJ), the vowel of the middle, previously stressed syllable 14 For the development of the Hebrew pattern qii(lii (3rd pers. sing. fern. of the perfect) see below p. 165 ff.

THE VOWEL I AS AN AUXILIARY VOWEL

17

should be eliminated because of the shift of stress to the ultima. It was precisely the strong stress on the second syllable which caused the weakness of the stress on the first syllable. The shift of the accent from the second syllable to the ultima could only result in a further weakening of the stress on the vowel of the first syllable. The vowel of the middle syllable could certainly not disappear by such a process. The same development holds true in a large number of similar cases of North-Semitic in which one is used to assume 'attenuation' or transition of an original a in a closed syllable to i. (Bauer-Leander [see above p. 3] speak, rather strangely, of an a in an unstressed closed syllable; how could such a closed syllable be specially unstressed?). Some cases in point here are certain st. cstr.-forms of Hebrew, like diQre (st. cstr. of deQiirim) < *daQaraj, ziqne (st. cstr. of zeqenim) < *zaqinaj , :jUjqal (st. cstr. of :jegiiqiiJ < :jadaqat, where one ordinarily considers the i as having originated from a in forms like *daQraj, *zaqnaj, *$agqaJ. Brockelmann, Grundriss, I, §43 p, e, pp (p. 104) explains the disappearance of the second syllable in the lastmentioned hypothetical forms as follows: "Ebenso schwindet vor einer Nebendrucksilbe von zwei freien kurzen Vokalen der zweite, so dass der ihm vorhergehende gedeckt wird: *daQaraj> diQre 'Worte', *kanafaj> kanje 'Hugel', *:jadaqat > $igqal 'Gerechtigkeit'''. We find a similar situation also in the development of st. cstr.-forms of feminine nouns with i in the second consonant (thus in nouns of the pattern qatilat) in Hebrew, like niQlal (st. cstr. of neQelii 'corpse') < *nabilat, etc. But on the other hand, we also find st. cstr.-forms of nouns of this type whose vocalism and syllable structure agree with those of the respective st. absol.-forms, as, e.g., b ere/sal, st. cstr. to bere/sii 'pond', etc., which Brockelmann, Ibid., 88 (p. 105), considers as caused by the analogy of the st. absol.-forms ('Analogiebildungen'). However, we find the same syllable structure and vowel change from a to i (that is, the pattern qitlat) in certain cases not only in the st. cstr., but also in other situations. Beside the st. absol.-form :jegiiqii « $adaqat) we have not only the st. cstr.-form :jigqal, but also the forms with possessive suffixes like 'figqiili, etc. (which with respect to syllable structure and vocalism are in complete agreement with the st. cstr.form). Brockelmann (Ibid., yy [po 104/5]) derives this latter form, $Ujqiili, from *$adaqatf with "the main stress on the last syllable". From the fact that this form, *$adaqatf, "did not result (in accordance with Ibid., aa) in the form *$egaqt{" (as, e.g., *qata1als¢m> qetallsr;m 'he killed you'), but rather :jUjqiilf, Brockelmann concludes: "dass

18

THE VOWEL I AS AN AUXILIARY VOWEL

die Femininendung at hier einst einen Nebendruck gehabt haben muss, der sich aber vor dem unmittelbar folgenden Hauptdruck nicht halten konnte ... ". We must point out here that in Biblical Aramaic the usual form of original ~adaqat (that is, its st. absol.) is ~itjqii (Syriac, st. emphat., zegqelii); 15 furthermore that in Syriac the st. emphat. nesamlii goes with the st. absol. ne§mii, st. cstr. ne§mal (Hebrew nesiimii, st. cstr. nismal); also that Hebrew berelsii « *barikat) is paralleled by Arabic birkat. Also, cf. Syr. delsrii 'male', metrii 'rain', besrii 'meat', Mandaic girbii 'leprosy', dinbii 'tail',16 bisrii, mitrii, miliii 17 'simile', kenfii 'wing', Akkadian zibbatu 'tail', kippiiti 'the ends', i.e., "the furthermost limits of the sky and the earth" (kanfol hii'iif(:~), plur. of kappu < *kanpu (Delitzsch, Wtb., 340 a), ziqnu 'beard', zikru, mi!ru, also Ethiopic kanf and the like; beside forms like Arabic and Hebrew, respectively, gakar, ziilsiir; malar, mii!iir; basar, biisiir, garab, giiriil]; ganab, ziiniil]; malal, miisiil; gaqan, ziiqiin; kanaf, kiiniif (see N6ldeke, Mandiiische Grammatik, p. 107; Barth, Nominalbi/dung, p. 16 and p. XXXII). is The Hebrew forms under discussion (see aeove p. 17) must be explained in connection with

15 The corresponding (feminine) noun-form of, e.g., the root !jdq is in Arabic: !jadaqat, in Hebrew !jegiiqii (as opposed to the form without feminine-ending: !jidq, or Iffgl}q, respectively). The corresponding form (i.e., the fern. noun-form of the root !jdq) in Aramaic, !j;gqii, zetjqelii (Syriac, stat. emphat.), must therefore also be traced back to the basic form !jadaqat (cf. Barth, Nominalbi/dung, § 57 d, Anm.; p.88). Cf. also, e.g., Syriac zeljelii (pI. zelje; Nestor. [sing.] zelafla; Arabic zalaJat) 'testa' (Brockelmann, Lex. Syr.2, 198 b, s.v.), etc. Furthermore, cf., e.g., Syriac helalslii 'manner of walking', 'step', of which ludaeo-Aramaic hillselii '(legal) procedure' is to be considered a phonetic as well as a semantic derivative (see my book The Spiritual Background oj Early Islam, Leiden 1972, p. 137). The ordinary form under which the word is used in Hebrew ('Rabbinic') style, i.e., haliilsii, reflects the more original form h elalslii. It is obvious that the development of these noun-forms with i (!jigqa, zegqelii; zelf"fa; hi/Is elii) is identical with the development of the verb-form (3rd pers. sing. fern. of the perfect) qi!lal « *qatalat), as outlined above. 16 The u in Syriac dunba has probably originated via *dJ:!bii through the influence of the adjacent b (or, more exactly, of the labialized n, which together with the b forms, as it were, a 'cons on antic diphthong'); and very similarly, Syriac burk elii beside ludaeo-Aramaic birk elii 'blessing' « barakat) originated via *brk elii; cf. below p. 77. 17 One may also compare with it Arabic mill, sibh, bidl in their relation to malal, sabah, badal (see Lane, Arab, diction., s.v. badal). Note also Arabic 'ilr beside' alar

(cf. p. 22, n. 23). 18 Barth, Nominalbi/dung, p. XXXII, strangely, contrary to common (but unjustified) assumption, does not assume *basrii, *dalsrii as intermediate stage of Syriac *basarii > besrii, *dalsarii > delsrii, etc., but rather *baserii, *dalserii, etc., as if this syllable structure would especially favor the change a > i.

THE VOWEL I AS AN AUXILIARY VOWEL

19

these latter forms, for which the accentual rules set up by Brockelmann for the Hebrew forms can by no means be valid. This leads immediately to the conclusion that the accentual rules derived by Brockelmann from these forms have no secure claim on validity. Aramaic $igqii < $adaqat, also Arabic birkat < *barikat (concerning the forms qatil > qUI, see below p. 22) cannot have originated through any other causes and in any other way than have, say, modern Arabic kitbet, Syriac qetial < katabat, qatalat (see above p. 15-17). Accordingly, the development sequence was: $adaqat > Bibl.-Aram. *$egaqii > $idqii; also Arabic barakat (Hebrew beriilsii) > Judeo-Aram. *brakat 19 > birk elii (st. emphat.) 'blessing' (cf. also Mehri birket 'blessing' < barakat, beside instances like ragahet 'neck' = Arabic raqabat, sebeket 'face-net for women' = Arabic sabakat, etc.; see Bittner, Studien zur Mehri-Sprache, I, p. 41), *barikat > Arabic *barikat, or *brikat, respectively, > birkat 'pond'. In Hebrew, these shifts did not take place in the inaugmented (that is, absolute) forms. The basic cause of these shifts, however, the succession of several open syllables (which in Hebrew did not have an effect on the 'absolute' forms), appears reinforced when the words are augmented by suffixes or by words spoken with them in a single accentual unit (,unitary accent', 'Einheitsakzent'). In *dal]araj hamm{?l{?ls 'the words of the king' (as against del]iirim 'words') there was a sequence of a large number of syllables on which the accent had to be distributed evenly (in all probability, the shift already took place before the occurrence of the two phenomena which affected the entire phonetic structure of Hebrew: a) the farreaching transfer of the stress to the last syllable of the word, and b) the pretonic lengthening of a, as, e.g., in d"l]iirim). Once the second syllable became preponderant in stress, the stress could immediately be moved back for the purpose of the preservation of the thus endangered first syllable (the vowel of which had been weakened quantitatively as well as qualitatively). This abrupt retreat of the stress to the first syllable led automatically to the insertion of a new vowel, i, between the first two consonants as bearer of this stress (in place of the original vowel which had been weakened quantitatively as well as qualitatively). And at the same time, the abrupt retreat 19 Thus actualJy in the modern Arabic dialect of the 'Anazeh: brlika (see Landberg, Langue des Bedouins 'Anazeh, Leiden 1919, p.9,29; and cf. also Wetzstein, ZDMG 22, 184). Perhaps one may also compare 'Omani baTf!ke, as welJ as derege, noted by Reinhardt, Ein arabischer Dialekt gesprochen in 'Oman und Zanzibar, §29.

20

THE VOWEL I AS AN AUXILIARY VOWEL

of the stress to the first syllable caused the elision of the vowel of the second syllable, a (in the case of *nal]ilat: i): dil.J.re,20 $Ujqal , nil.J.lal. 21 One can recognize the 'contraction' qatal> qitl (via *if tal, *qtal), which was caused by the augmentation of the stem, already in some very old forms of numerals: thus in the feminine-form of the numeral 'one' in Arabic: ilJda, as against the masc. alJad. Brockelmann (Grundriss, I, §94 a, following Barth, see also below, p. 60) assumes here dissimilation of the original a in a closed syllable, in a form *alJdii < *alJada. That is: he assumes that the short vowel a, when followed in a subsequent syllable by a long vowel a, was dissimilated to i. (The same dissimilation he assumes in a number of other instances where a short a-vowel appearing from the outset in a closed syllable has been replaced by i when followed by a long (a)-vowel; see below p. 60). In my opinion, the development took the following course: In *alJ adii , which has three open syllables, which at first were spoken with a 'hovering' stress (,schwebender Akzent'), the second syllable was occasionally stressed more strongly and thus the vowel of the first syllable became reduced: > *a/Jddii or *elJdda; then the accent was suddenly and vehemently thrown back to the reduced vowel of the first syllable. This led to the insertion of a new vowel, i, as bearer of this strong stress on the first syllable, and simultaneously to the elision of the vowel of the second syllable: > {lJda (that is: the elision of the vowel of the second syllable, a, is a direct consequence of the strong stress on the first syllable). The existence of three equally 20 In the case of the st. cstr.-form *kanajaj > kanji!, *rnalalsaj > rnallsi! and the like, the first vowel probably received the stronger stress at first (in the final analysis for the same reasons, namely, because of the succession of a number of syllables spoken with a unitary stress) so that the second vowel was immediately elided: *kanajaj > *kanjaj. One must, however, consider it possible that also in the case of a development corresponding to that of dil.lri!, that is: *kanajaj > *kanajaj, the result could have been kanji!. The quality of the first vowel, which was reduced at first (*kanajaj) , could possibly have prevailed in the end. Furthermore, the quality of the newly arisen vowel can be influenced by adjacent sounds (cf. below p.32), although this latter possibility seems less probable to me in this Hebrew instance. C£., by the way, Syriac ken/ii, pI. st. cstr. ken/ai, with e < i; also Ethiop. kanf 21 The i in Arabic nisbat 'relationship', as against the equivalent nasab, and many other instances, must, in my opinion, be explained in the same manner as the i in Bibl.-Aram. !fi4Qii, as against !fadaqat (see above). Between the two words, which differ in their basic structure only to the extent that nisbat includes an additional formative element, the feminine-ending, a morphological connection must necessarily exist; that is: the simpler nasab is the basis for nisbat. The vowel change in the latter, enlarged form (-at probably served here originaJly to indicate the nomen unitatis) is thus a consequence of the stress movement caused by the cumulation of open syJlables.

THE VOWEL I AS AN AUXILIARY VOWEL

21

stressed syllables was highly undesirable in a numeral which was spoken in a single accentual unit with the word designating the numbered object. Relevant here is also the dual form of the numeral 'ten' in the Semitic languages, explained by Brockelmann in the same way as he explains 'ilJdii (see above): Common-Semitic *'isrii 'twenty', and, of course, also the Hebrew feminine form of the word 'ten' (which goes with the 'ones'): 'dre, which in my opinion, with respect to the vowel of the first syllable, is related to the corresponding form of Syriac: 'esrii. Brockelmann, Grundriss, I, §249 a]( (referring to Ibid., §76 c) attributes its e < i [instead of a] to the influence of the adjacent sibilant. The word designating 'ten' has several other characteristics: the masc. form used not in conjunction with the 'ones' is 'asr in Arabic, '?s?r in Hebrew; the corresponding feminine-form is 'asarat and 'asiirii, respectively; on the other hand, the masc. form used in conjunction with the 'ones' is 'asar in Arabic, 'aSar in Hebrew; the corresponding feminine form is 'asrat 22 in Arabic, and (as already stated above) '?Sre in Hebrew. Reckendorf, ZDMG 65,558, considers the forms with vowelless second consonant as the original ones and attributes the forms with a vowel after the second consonant to a complicated process of analogy which is difficult to understand. Brockelmann (loc. cit.) is not certain whether the vowel is original or has developed secondarily: "mit Svarabhakti oder iilterem, im Ms. synkopiertem Vokal". I consider the forms with a vowel between the second and third consonant as the more original ones. The forms with i, which cannot be separated from ilJdii (originating without doubt from *aIJadii with an original vowel between the second and third consonant), also Common-Semitic *'isrii 'twenty', Hebrew '?sr¢ 'ten' (femin.), and the fact that the latter is paralleled in Arabic by a form with a as the vowel of the first consonant, but with no vowel of the second ('asrat), certainly support my position. Thus, the basic form of Common-Semitic *'jsrii 'twenty' seems to me to be: *' aSarii. And analogously, as the basic form of Hebrew '?sr¢ and Arabic 'asrat(a), we should assume the form *'aSaraj (cf. the independent form 'asiirii < *'aSarat) and 'asarat (identical with the independent form), respectively. *'aSarii developed to *'asarii> *'iSrii (exactly as *aJ;adii> ilJdii); and likewise Hebrew *'aSaraj> *'asaraj (cf. 'e sara 22 Also 'asarata and 'asirata are attested as dialect forms (beside 'asrata), see Wright, I, p. 256 C.

22

THE VOWEL I AS AN AUXILIARY VOWEL

< 'asarat in the dialect of the Sliit, in Cantineau, Etudes sur quelques parlers de nomades arabes d'Orient, I, p. 105) developed to *'israj> 'dr?, whereas in Arabic 'asarat the first syllable received the stronger stress from the beginning, as a consequence of which the vowel of the second syllable was elided: 'dsarat> 'asrat. The question why this tendency to reduce the number of syllables was realized just in the feminine forms connected with the 'ones' is a complicated one and still requires further clarification. The same explanation applied to the development of qatil to qitl; that means: to the development of ancient Arabic kibd(un) beside kabid(un)23 'liver', kUmat beside kalimat 'word', etc. (cf. e.g., also Socin, Diwan, III, § 88 c) [there was also a different development of original qatil-forms: qatil> qatl, as, e.g., kabd(un) < kdbid(Ufr)]. While Zimmern, ZA V (1890), p.370ff., here talks of cases "in welchen der zu synkopierende Vokal auf die vorhergehende Silbe iibertragen wird" (taking into consideration also cases like qatul> qutl, whose u instead of the expected i must be similarly explained as in the cases mentioned above n. 16), Brockelmann, Grundriss, I, §68 co, assumes as an intermediate stage a form *qitil developed through vowel assimilation: qatil> *qitil> qUI. The occurrence of such a type of vowel assimilation, however, seems to me difficult to prove (see below p. 18). A 'transfer' to a preceding syllable of a vowel which primarily should have been syncopated (as in qatil> qat!) cannot be assumed either as an actually occurring phonetic process. The cases in point must be explained as those already mentioned in which the elided vowel which followed the newly arisen i was an a (see above p. 16). Kalimat > kUmat corresponds to the 3rd pers. sing. fern. of the perfect qatalat > qUlat (like modern Arabic kitbet ['she has written'] or Aramaic qi!lal).24 23 In the case of these formations (without feminine-ending) composed of two syllables, the forms actually to be considered in connection with the further development of these formations are, of course, the forms with inflectional endings. (As far as the modern dialects are concerned, the further structural development of these nominal formations has to be based, to a large extent, on the forms with possessive suffixes starting with a vowel). 24 That a form qUI so frequently originates from two-syllable qatil-forms, is probably due to the fact that the medial vowel i, which was very prone to elision (more than a medial a, cf. N61deke, Mand. Gramm., p. 106, §92, and n.3), very often received a strong stress and thus on its part frequently caused a spontaneous and violent retreat of the accent to the first syllable (see above p. 20). By the way, qitl can probably also be found originating from qatal, both in old and modern dialects. Cf., e.g., above p. 18, n. 17, and also certain forms in J.-J. Hess, Beduinennamen aus Zentralarabien (Sitz.-Ber. d. Heidelb. Ak., 1912), as, e.g., gimel 'camel stallion', p. 7;

THE VOWEL I AS AN AUXILIARY VOWEL

23

Here a special case of these developments must be mentioned. In Brockelmann, Grundriss, I; § 72, we find a chapter "Fernassimilation von sonantischen an konsonantische Vokale". Brockelmann cites here almost exclusively instances where in forms of the pattern qatl in which the third radical (consonant) is a j, the a of the first radical (consonant) is replaced by i, as already (see Ibid., a) in Ancient Arabic naMun 'swamp' > Nagdi ni~jun, aljat> iljat 'fat tail', Modern SyrianArabic nasjan > nisjan 'forgetful', etc.; also in Hebrew (see Ibid., b) regularly under the same circumstances, as Arabic jJabj, Aramaic !aQja > Hebrew $e~i 'gazel'; Arabic gadj, Aramaic gOt}ja > Hebrew gegi 'ram'; *qarjat > qirja 'city'; similarly also in the case with u: Arabic kuljat, Ethiopic kJialit, Syriac kulila > Hebrew kUja 'kidney'; similar cases also in Modern Syriac, as da/Sja> Felli!)'i di/Sja 'pure', etc. The same view as that of Brockelmann (see also ZSem. X, 1935, p. 334) is also advocated by Bauer-Leander, § 18 a. However, as will be detailed further below, even the 'non-contiguous assimilation' ('Fernassimilation') of regular ('sonantic') vowels to regular ('sonantic') vowels in a sense in which this phenomenon is usually assumed (especially by Brockelmann), is a rather doubtful matter. This holds even more true of the 'non-contiguous' assimilation of regular ('sonantic') vowels to semi-vowels ('consonantic' vowels) under the conditions existing in the instances under consideration. In my opinion, the reason for the change in question here is the unstable character of the semi-vowels ('consonantic' vowels), their proneness to becoming 'fluid', that means: actually vocalic (,sonantic'); cf. p. 88 ff. In Arabic ,aljat, the extremely short i, that is: j, if we consider it - for our present purpose-as vowel, could easily be extended (lengthened); thus ' aljat became *' aliat; jJabj (~abD or *jJabjun became *jJabi or *jJabiun, with ('sonantic') i '(instead of D, which could easily get a certain stress: *' aliat, *jJabi or *jJabiun; with possible simultaneous reduction of the vowel of the first' syllable: *'aliat, or *'alijat (cf., e.g., Tunisian lijii, Stumme, § 53; Palestinian lije, and similarly in other dialects), and also *rbi (jun), or *pabi (jun). Here again the accent laqab 'nick-name', '6ttlb. nygef, gain. lyguf, Aulad 'Aliy negeb, p.8. Thus, probably: gamal, laqab> *g"mal, *laqcib (or *gmal, *lqab, respectively) > *giml, *liqb, or (with velarisation of the i by the adjacent q [gJ) *lygb, and now (through 'segolisation') > gimel, nygef, or (on account of the terminal labial consonant) lyguf. To ascribe the i-vowels in these latter cases to the 'imalisation' of the original a seems out of the question to me. (As to gimel, one ought to pay attention to Landberg's remark [Dalina, p. 93, n. 4]: "Obs. la prononciation gamal, et non, comme dans Ie Nord et les dialectes de ba9ar partout gemal").

24

THE VOWEL I AS AN AUXILIARY VOWEL

could be drawn back, which caused a 'break-up' ('Aufsprengung') of the initial double consonant (or 'semi-closed' syllable); at the same time, through the strong stress on the initial syllable (caused by the vehement regression of the accent), the subsequent i-vowel, which developed from an original j, became j again: > Arabic 'iljat, Hebrew *pibj(un), which latter then, according to the usual phonetic rules of Hebrew, resulted in ~eQi. Cf., e.g., gadj(un) > 'OmanI gdi (Reinhardt §48) > Tunisian Zidi (Stumme, § 136). This also explains 'IraqI ~ibi 'boy' (Weissbach, ZDMG 58,936) from ~abij > *~ab{ > ~ibi. We should like to point to the fact that in certain modern Arabic dialects we find i instead of the original a also in open syllables. In Central Arabia, we find word formations like bSibe 'piece of wood' < basabat (see Socin, Diwan, III, p. 116, §90 b). Socin writes (Ibid., p. 118, 92 c): "In !e'il gehen die primaren fa'al-Formen ofter tiber, wenn eine Endung, sei es der Nunation, sei es des Duals oder des Femininums an sie herantritt. Nach N. b zu 77,19 sprach der Erkl. s{!l{!m 'Gummiacacie', im Vers steht selim, was dem Metrum zuliebe selimin gelesen werden muss; das Nomen unit. lautet l.l. slimeh; eben so begtreh 'Kuh' N. c zu 94, 4. Haufig ist von basab 'Holz' das Nomen unit. bSibe, z.B. in 107; auch bSube kommt daneben vor, vgl. N. b zu 29 A,12. Ebenso zelimeh 'Mann', N. c zu 13,4; N. c zu 84,1; sbice 'Netz' N. a zu 16,10; ~edibe 'Rticken des Schwertes' N. c zu 43,5; ~limt edded 'Brustwarze' N. a zu 5,5 ... jedenfalls ist der dritte Radical bei ihnen etwas gescharft und i infolge dieser Scharfung aus e entwickelt, vgl. § 177 c.-Als Dual von ~anac 'Kieferknochen' wurde ~nicen gehOrt (Notiz)". Cf. here also several wordforms, which for the most part belong to the Beduin dialect of the 'Otabe, in 1.-1. Hess, Die Farbezeiehnungen bei innerarabisehen Beduinenstiimmen, in Der Islam X (1920), p. 74-86, as, e.g., 'bbisi < ~abasij (I.e., p.83, ult.); furthermore (Ibid., p.75, n. 1): "1m 'otebischen Dialekt wird jede zweisilbige W ortform einsilbig, wenn ein vokalisch anlautendes Suffix antritt. Man sagt el-g~ab und r;larab, aber g~ubeh 'sein Rohr' und r;lrilbet 'sie schlug'. Dies Gesetz gilt auch bei Anfiigung der Nunation, so dass z.B. die Formen el-jaras, el-gimel, el-~amar mit Tenwin frysen,25 gmiien, ~muren lauten". In Socin, Diwan, III, § 177 c, we find the following statement of interest to us: 25 Compare with this efrusu 'sa jument', Landberg, Langue des Bedouins 'Anazeh, Leiden 1919, p. 14, I. 8, where the new vowel, through the influence of the labialised r in the group fr, appears as u, as in the also mentioned IJmuren, while in frysen the i which one would expect, has been determined in its shading as y (a 'velarised' i-vowel) by the influence of the velarised (quasi 'emphatic') r.

THE VOWEL I AS AN AUXILIARY VOWEL

25

"Wie bei diesem Wort (i.e., min < man),26 so ist auch sonst i nicht selten allmahlich durch e hindurch aus a entwickelt, vgl. § 86 b; man hat hierbei die FaIle auseinander zu halten, wo dies in betonter Silbe, und wo es in unbetonter stattfindet; natiirlich diirfen auch keine den Lautiibergang hindernden oder einen anderen Vokal begiinstigtmden Consonanten in der Nahe eines solchen e stehen. Man vgl. in dieser Beziehung, was iiber die Femininendung a-e-i in § 74 und was iiber die Nunation in in § 76 ff. auseinander gesetzt ist. Besonders geht auch aus II. Formen des Verbs wie barrig (vgl. § 129 b) hervor, dass das unbestimmte i in solchen Fallen eine einfache Nuance eines unbestimmten e ist. - Anderseits ist nicht zu leugnen, dass i gerade auch in betonten Silben eine sekundare-ich mochte sagen-Zuspitzung von e = urspriinglichem a ist. Hierbei ist auf die Conjunktion in, inn § 56 zu verweisen, besonders jedoch auf urspriingliche fa' alFormen mit der Femininendung (vgl. §92 c), wie zelimeh, oder mit Suffixen, wie gemili § 146 cff.,-beim Verbum auf Formen mit einem vocalischen Affix, wie nsfd({t (§ 130 b ff.)". The i in open syllables (possibly u in the neighbourhood of labial consonants) in the forms mentioned by Socin as well as by J.-J. Hess (see above p. 24)-that is: the i (possibly u) in forms like nsid({t « nasadat), zelimeh « zalamah), gemili « gamali), g~ubeh ('his reed', beside el-ga~ab), c}rubet ('she beat', beside c}arab) - i's, in my opinion, identical with the i which appears in closed syllables (instead of a or any other vowel) in the forms discussed at the beginning of this study (see especially above p. 3 ff.). bSibe < [tasabat (mentioned by Socin) is thus to be considered developed via *I;af'bat, with reduced vowel of the middle syllable as a consequence of its position between the strong stress of the first syllable and the stress of the terminal syllable. 26 This phonetic derivation of modern-Arabic min (and its lengthened variant min, especially used in pause; see Socin, Diwan, III, §68 b) from Old Arabic man is, in my opinion, correct. That the Hebrew mi 'who?' augmented by n (as man from ma+ n) could have gotten from Proto-Semitic into modern-Arabic (without leaving any trace in Old Arabic (see Brockelmann, Grundriss, I, p. 327; also Barth in Landberg, Festgabe [1909], p. 19, and Barth, Pronominalbildung, §57) seems to me utterly improbable. The many variants of pronunciation mentioned by Socin clearly illustrate the development of man to min: man, mun (with u under the influence of the labial m, which, however, was only possible after previous reduction of the original a, see p. 32 ff.); "bisweilen auch m.in mit einem Zwischenlaut zwischen i und y"; "das n fallt besonders vor anlautendem i aus, dann hart man mi".-If there is any connection between Hebrew mi and modern Arabic min, min, it is only conceivable on the assumption that in Hebrew an original form man, which occurs in all other Semitic languages, has, under the same conditions as in modern Arabic, developed into mi, which then spread and entirely supplanted man. But this cannot be proved.

26

THE VOWEL I AS AN AUXILIARY VOWEL

To obviate the complete disappearance of this middle syllable (or in order to make it pronounceable), the stress moved to it, after the vowel had already been completely (or almost completely) wiped-out (as described above p. 19). As bearer of this strong stress (in the middle syllable) an entirely new vowel, i, was interjected. Socin (Diwan, III, § 90 b, p. 116 and § 177 c, p. 210) assumes that the i (instead of an original a) in the forms under discussion (as, e.g., bsibe < basabat) has developed from a more original e, which in turn would be directly originating from the original a-vowel (or, as it were, would be a variant pronunciation of the original a). This would mean that we would be dealing with the so-called Imalah-pronunciation of a (a> a> e, and finally-as it were, imalah sadidah in the phonetic terminology of the Arabs- > i). Socin's reference to the pronunciation of the feminine-ending (i.e., a> a> e, through Imalah) indeed clarifies his conception of the i-vowel in bSibe « basabat) and other forms under discussion. In my opinion, the i-vowel in open syllables, in the instances mentioned by J.-J. Hess and by Socin, as, e.g., bSibe (instead of basabat), is a completely new vowel, an elementary creation, with no connection with the phonic substance of the original vowel (as will be proved in detail below, see especially p. 56-57). We stress once more that this vowel, which basically appears as i, may appear also in various nuances, caused by the neighbouring consonants, as, e.g., c}rnbet « c}arabat), where the u is caused by the contiguous labial b. Wetzstein, in his study Spraehliehes aus den Zeltlagern der syrisehen Wiiste (ZDMG 22, 184, note), points to the fact that words like barakat, jarasi, qatalaka in Beduin pronunciation sound like brakka, jrassi, qtallak; furthermore that words like qatala, balad, 'alij, if pronounced by a l:ta9ari (who in contrast to the Beduin stresses the first syllable), sound like qattal, belled, 'alii. But for the Arab - thus Wetzstein states - no gemination of the consonant takes place here. It is clear that we are dealing here with the pronunciation of vowels with strong stress, called by Sievers (Grundziige der Phonetik 5 , § 589 ff.) "stark geschnittener Silbenakzent" ("strongly cut syllable accent"), an accentuation which, at least for the German, easily creates the impression of a gemination of the consonant which follows the vowel thus affected. That such a strong stress can, however, also cause genuine gemination, is proved by certain forms observed by Wetzstein. In his above-mentioned study (I.e., p. 190, n. 1), Wetzstein lists

THE VOWEL I AS AN AUXILIARY VOWEL

27

paradigms of original forms of the nominal pattern fdalat in the pronunciation of the Beduins as well as of the settled people (l;Ia *tafarat > *rfaret, *tfaret> tufaret) and thus merely changes the syllable struc-

THE VOWEL I AS AN AUXILIARY VOWEL

33

ture in some way, or develops by shifting the accent to a complex (cluster) of consonants, becoming the bearer of this accent, can be influenced in its timbre by the adjacent consonants. As is shown below (p. 56), this newly created vowel is at first not pronounced with full voice, and without a definite timbre (being an 'indifferent' vowel, although it possesses an affinity with i). The timbre is sometimes more exactly determined by the adjacent sounds, which holds true, e.g., for the cases, mentioned above, with adjacent u-related consonants. 33 Thus the U 34 in forms like 3rd pers. sing. fern. se,ubet, blu$et, bul$et, etc., as against instances like 2nd pers. sing. masc. sarab( e) t, etc., also indicates that accentual shifts have taken place here. But the same holds also for ktfbet, kitbet, although here the timbre of the new vowel was not determined (or influenced) by the adjacent sounds. It thus becomes almost unnecessary to rectify the phonetic rule given by Meissner (p. XI): "Treffen zwei solche e's [namely < a, as in sene 'year', etc.] 35 zusammen, so wird der erste gem in £; oder i differenziert: s£;beb 'Grund'; kitel 'toten'. [Weissbach, too (ZDMG 58, 933), commenting on this passage of Meissner's, gives the same interpretation]. Zuweilen ist dieser Vokal sogar beim Verbum stehen geblieben, wenn die zweite Silbe a, nicht e hat: gikar 'erwiihnen', rikah 'reiten', sigal 'wegtragen"'. The last-mentioned instances themselves clearly invalidate the phonetic rule. The development here was, of course: *gakar > *gOkar, *gkar > gEkar. 36 Cf. (Meissner, Ibid., p. LU): kiteb, and also sarah, tufar, which latter, by analogy, are probably to be traced back to *sarab> *§'rab, *srab> sarah, or *fafar> *fjar, *ffar> fufar with a new vowel; the specific timbre of this new vowel was determined by the adjacent r as a (sarab) , by the f as u (tufar). To see an original form in sarab (with an original a after the s), is prohibited in view of kiteb and fufar. These 'Iraqi-Arabic forms of the 3rd pers. sing. masc. (appearing without object suffixes) presuppose a development from earlier forms characterized by an accentuation which we encounter in other Arabic dialects 33 As to the conclusions which may be drawn from this for the understanding of certain so-called vowel assimilations, see below p. 76 f. 34 A similar case (u through influence of a labial), among many others, is also, e.g., (a)'arubi 'Arabian', beside 'arabi: > *'ar"bi> *'arubi, Meissner, Ibid., Glossar, p. 133 b, S.V. 35 According to Weissbach, ZDMG 58, 933, this e < a should be more accurately be presented as q (imiilah; cf. above p. 30). 36 Cf. footnote 37.

34

THE VOWEL I AS AN AUXILIARY VOWEL

(especially in the Magrib, see above p. 11) as well as in the old Semitic languages: cf. Ethiop. qatala, Aram. if/al (against Cantineau, Etudes sur quelques parlers de nomades arabes d'Orient, I, p. 81).37 Forms of the same pattern as the form kiteb for kataba, which appear in the paradigm of the verb in 'Iraqi-Arabic, can already be found in Old Semitic, in its noun, although (as it seems). not in its verb. In Ethiopic, the form qital, as a designation of abstract concepts and verbal nouns, has already completely replaced the qatalforms which are still preserved in Arabic (see Barth, Nominalbi/dung, §70j71; Brockelmann, Grundriss, I, § 117 b), e.g., 'amaq 'depth', kabad 'weight'. We find qital rather frequently in Arabic as a designation of 'size' relations, as (figar 'smallness', 'infancy', kibar ('greatness'), 'advanced age', etc. In view of the constant replacement of the Arabic qatal-abstracts by qital in Ethiopic, there can be no doubt that i replaces a here. Brockelmann (I.c.) thinks, with some doubts, that the influence of the imperfect, e.g., ja'maq, was here effective. In my opinion, the i (> a, in Ethiopic) represents a newly created vowel which came into being as a result of accentual shifts. These qital-formations are especially frequent in roots ult. j, as, e.g., Arabicfidan 'ransom', /:Iiman 'taboo-area'. As to the development of such forms, one cannot set out here from primary forms like *fadajun (and similar forms) and connect the i (in fidan) with the terminal j of the root (Jdj), as in the instances of the pattern qatl with post-consonantic j (see p. 23f.) with which Brockelmann (I.c., c) compares them (in assuming in both cases an assimilation of the sonantic vowel to the consonantic vowel). Rather, one must start directly from forms with terminal vowel (in which the original third consonant j had already disappeared), as *fada, etc. In forms with a second syllable terminating in a sonant (vowel), the accentual balance could easily be lost, so that the stress could be shifted to this terminal (second) syllable. The subsequent development is easily understood in the light of the preceding discussion of other forms, and the final result was the insertion of a new vowel i between the first two consonants (in other words: in the first syllable). Such a development was 37 Concerning the priority of the various types of the accentuation of this form in the Semitic dialects (which has been discussed so often), Landberg, Dalina, p. 1306, n. I, says: "Si en arabe fa'al est plus ancien que fa' ai, qui encore aujourd'hui se trouvent I'un it cote de I'autre, cela echappe it mon jugement", and I cannot but agree with him in view of the 'hovering' stress ('schwebende Betonung'), which I assume, and in view of the secondary accentual variation caused by it.

THE VOWEL I AS AN AUXILIARY VOWEL

35

in the case of forms terminating in a vowel (like *fadii) from the outset more likely than in forms which terminate in a consonant. This assumption is in agreement with Socin's observation (Diwan, III, §96, in the section on the nouns fa/i/u'al and fa/i/u'iil from roots ult. f/y) concerning forms like ~ufa 'rock' < ~afa(n), ~ufa 'purity' < ~afa', that "der in offener Silbe stehende Vokal der ersten Silbe bei diesen Nomina umso eher reduziert wird, weil der Ton ziemlich entschieden auf die letzte Silbe rallt" (that the new vowel- as we interpret itappears here as u, and not in the 'primary' nuance i, is caused by the influence of the neighbouring labial f). One must also mention Goitein's observation (Jemenica, Sprichwarter und Redensarten aus Zentraljemen [Jerusalem 1934], p. XVJIl, § 11 a), which is in agreement with Socin's and Meissner's accentuation rules: "Dieser Hauptbetonungsregel [refers to Ibid., § 10] steht gegenfiber eine Neigung zur Vokalelision, besonders bei stimmlosen Konsonanten, unter diesen wohl besonders bei k, s, ~, t, f Dieselben Leute sagen bfsidu und bsidu, bU~mah und b~dmah 'Prozessache' (nicht von b~m), bdfabiih und bSdbiih, dgall 'weniger', kidr 'mehr'''. It seems to me especially significant that these variations in accentuation and syllable structure occur with the same persons. These variations definitely indicate that the 'hovering' stress (,schwebende Betonung') is also common in the Yemenite dialects (cf. also above p. 12 and below p. 70. 38 I would like to comment here on a theory of J. Cantineau, contained in his Etudes sur quelques parlers de nomades arabes d'Orient, I, published in Annales de I'Institut d'Etudes Orientales de la Faculte des Lettres de l'Universite d'Alger, 11,1936, p. 1_118. 39 It is ordinarily assumed that changes in the syllable structure in 38 I would therefore not agree with Goitein's judgment, Ibid. (which in some way coincides with Cantineau's theory discussed in detail below): "Hier hat also offen bar nicht eine primare Akzentverschiebung Vokalschwund zur Folge, sondern umgekehrt gibt die Neigung zur Aufgabe kurzer Vokale dem Affekt oder dem Satzrhythmus weite Moglichkeiten zur verschiedenen Betonung gleicher Formationen". I fail to understand what might cause a tendency in the Yemenite dialects to drop short vowels. It seems unexplainable that this tendency should affect just the strongly accentuated vowels (thus the first a in bdsabiih and not the second). One must therefore stick to the view that the accentual shift precedes the disappearance of a vowel and constitutes its cause. 39 Cantineau had previously given hints of similar ideas in his articles De la place de l'accent de mot en hebreu et en arameen biblique (Bulletin d'Etudes Orientales de l'Institut Franc;ais de Damas, I, 1931, p.81-98), and Elimination des syl/abes breves en hebreu et en arameen biblique (Ibid., II, 1932, p. 125-144).

36

THE VOWEL I AS AN AUXILIARY VOWEL

the Semitic languages, especially in the modern Arabic dialects, are primarily caused by accentual shifts. Cantineau (see l.c., p. 68) believes that the development of the syllable structure is mainly determined by the "rythme quantitatif de la phrase". In the modern Arabic dialects the accentual shift is, according to him, caused by a change in the syllable structure, contrary to the prevailing assumption, according to which the changes of syllable structure are caused by accentual shifts. Cantineau assumes formulas of quantitative rhythms of long and short syllables whose sequence and alternation are differently structured in each specific dialect. M. Cohen, in his review of Cantineau's work in Bull. de fa Soc. de Lingu. de Paris (BSL) , t. 38, fasc. 3 (Nr.114), 1937, p. 178) fully approves Cantineau's view, so much the more since Cantineau adopts the view which underlies his own exposition of the syllable structure in Maghrebian dialects (in Le parler arabe des Juifs d'Alger, 1912). He finds this correspondence so much more interesting since (according to his own observation) in the Arabic spoken to the West of Tunis the word stress does not figure very prominently, whereas it is very prominent, violent as it were ('quasi brutal'), in the dialects studied by Cantineau. I believe that the particular changes of syllables and vowels in the dialects investigated by Cantineau (for which parallels exist ill other dialects which have already been discussed and explained in the present study) can be most simply explained by assuming accentual shifts. The fact that the accent in the dialects studied by Cantineau is almost violent ('quasi brutal') can only corroborate my view. I thus follow Wetzstein's and Socin's view based on the observation of accentual facts and contested by Cantineau (see f.c., p. 67). Socin's observations concerning accentuation and vocalism, whose judgement, in Cantineau's opinion (see l.c., p. 65), was clouded by his theory concerning the accent, merit our full attention, in my opinion. Wetzstein's and Socin's observation of the 'hovering' stress (,schwebender Akzent') seems to me to be of basic importance also for the understanding of the accentual facts of Old Arabic and other Old Semitic languages. The origin of the quantitative formulas posited by Cantineau, on the other hand, appears unexplained to me, and their very existence unproved. 40 When Cantineau (f.c., p. 67) considers the accent-both 40 Cantineau, in his study Une afternance quantitative dans fes pronoms suffixes semitiques (Bulletin de fa Soc. de Ling. de Paris, vol. 38, fasc. 1 (1937), p. 148-164)

THE VOWEL I AS AN AUXILIARY VOWEL

37

in general and also in Old Arabic-as a secondary, subordinate ('accessoire') element of language and declares quantity to be the governing factor of language, he makes, I believe, an unacceptable basic assumption. For the existence of a word or a speech unit both elements in question, accent and duration (quantity), are equally necessary (see Sievers, Grundzuge der Phonetik 5 , § 568). The accent, on its part, has two sides: tone (musical accent) and stress (dynamic accent, or intensity); and here the question may be raised, both for other languages and the Semitic, which of the two types of accentuation predominates. As is usually assumed and as I also consider to be essentially correct, it is stress, although this will still have to be clarified in detail for the various periods of language development and for the various languages and dialects consonant with the progress and views of general phonetics. It is certain that in many cases the syllable structure determines the place of the accent. Thus, for instance, originally closed syllables of strong sonority in Old Arabic, and likewise this type of syllables (developed through elision of vowels) in the modern dialects, very frequently attract the 'stress' (cf. below). On the other hand, stress on its part may influence the syllable structure. It can, among other things, bring about a shortening of vowels, but also their lengthening. Thus one usually explains the lengthening of vowels in many nominal forms which, for the most part, occurred in pre-historic time. Furthermore, one seems to have observed, also in historic periods, a stronger stress on the noun (which brings about quantitative changes) than on the verb (see below p. 66). Therefore, Cantineau is probably going too far in denying to the accent any morphological role whatsoever, in contrast to what is the case in Indio-European. Although this latter fact may at present remain undecided, it is certain that in modern Arabic dialects the accent has transformed syllable structure to a very considerable extent.41 has expounded a certain rule of quantity which is based on the same type of theory. Unfortunately, I cannot go into detail here; however, I would like to remark that Cantineau obviously has not taken into account A. Fischer's study Die Quantitiit des arab. Pronominalsuffixes hu (hi) in: Oriental Studies dedicated to Paul Haupt, 1926, p. 390-402. The treatment of this suffix in Old Arabic, which was probably caused by a basic law of rhythm, is by no means in accord with Cantineau's assumption, but would rather favor a rule of quantity which is the opposite of his rule, if it is at all possible to formulate a unitary rule of this kind. 41 Thus also in some Old Semitic languages in which these developments have already come to a certain conclusion and in which a fixed accent has evolved, in contrast to the (probably) free accent of Proto-Semitic and Old Arabic.

38

THE VOWEL I AS AN AUXILIARY VOWEL

Cantineau, p. 68, l. 7 ff., cites forms like gmi31i and gamdlne, fjr6bat and fjardbt as especially conclusive evidence for the fact that the accent has no influence whatsoever on the syllable structure in these Bedouin dialects. He notes that the aforementioned forms had all the same type of accentuation, but in spite of this fact not all of them had lost the vowel of the initial syllable. For me, it is not only remarkable that in gmi3li, fjrJbat, in contrast to gamdlne, fjardbt, the short vowel of the first syllable has disappeared - a fact to which Cantineau has paid his exclusive attention - but it seems to be equally noteworthy that the vowel of the second, stressed open syllable of these words has the quality of 0, which is quite different from the original a,42 while in the latter forms which have retained the short vowel in their first syllable, the vowel of the second, stressed syllable appears in its original shade a. 43 A study of the entire material presented by Cantineau only reinforces the impression which one gains from the instances cited by him to show the contrast between the two groups: in the instances in which the vowel of the first syllable has disappeared, the vowel of the second, open syllable is very frequently 0, whereas in those cases which have retained the vowel of the first syllable, this 0 can hardly be encountered in the second closed syllable. Rather, 42 If we (see Ibid., p.60), beside gm61i with a (i.e., a), in closely related dialects (observed by Cantineau) find gmali with a, and gmali with ii (Imalah); beside kt6bat (i.e., ktabat) also find ktilbat with ii, furthermore ktalat, nzalat with a; beside (Ibid., p. 62) Ijr6qak also find Ijraqee, etc. - if we thus, in these instances, meet the original vowel nuance (or one close to it)-then also this situation has its parallel in various instances noted above, where forms with newly developed i (corresponding to, or identical with, a) coexist with forms of identical syllable structure in which, instead of i, the original a-vowel appears. A new vowel, i (or a) is only created in case that the original vowel has disappeared or has been excessively reduced. If this has not occurred, then, at the 'transposition' of the accent (" 'Umspringen' des Akzents") the original vowel (ordinarily a) can become the bearer of the accent. The usual situation is, however,-as one must conclude from the paradigms in Cantineau p.79 and p. 103-that the vowel bearing the 'transposed' accent (in the forms with endings beginning with vowels) appears with the nuance a (in Cantineau's notation: a)-in sharp contrast to the stressed a in the forms with closed syllables. 43 In instances like gama/ne < gama/una, the accent on the second (sonorous) syllable is indeed not inherited from the original (early) Arabic language. It has however, become stable on this syllable immediately with the disappearance of the Old Arabic declension-vowels (a process through which the second syllable of the stem of the word became closed), that means: in quite early time. In any case, the decisive point is that here, as in Ijarabta, the accentuated syllable is a closed, sonorous syllable. Also in case that such a shift of the stress had taken place in a very late epoch, it had scarcely caused a disappearance of the vowel of the pretonic syllable. For this type of stress-shift does not take place abruptly and with force as the stressshift in instances like gm61i « *garrfli < gama/i), etc.

THE VOWEL I AS AN AUXILIARY VOWEL

39

mostly the original a (or, with Imiilah, ii, e) is preserved in the second (closed) syllable. Cantineau writes (l.c., p. 48): "Comme dans la plupart des dialectes arabes, les timbres fondamenteaux peuvent subir des variations assez notables: un ancien i peut passer a e ou a 0; un ancien u a 0, 0, ii; un ancien a est susceptible de timbres varies: u, 0,9,0, ij, a,~, e, i, etc., et cela sous l'influence des consonnes voisines". I have already repeatedly indicated (see also the more detailed statements below p. 60 and p. 76ff.) that in general I consider it hardly permissible to ascribe changes of vowel quality which deviate considerably from the original quality of the sound to the influence of adjacent consonants or vowels (except in cases in which special conditions prevail which are out of the question here). While changes from i to e, or from u to 0, 0, ii, caused either by adjacent sounds or other factors, are entirely possible (also changes from a to ii, or other nuances close to a), I consider the change from a to i or o-to mention only vowels which are decidedly different from a- which Cantineau assumes as possible and which he ascribes to the influence of adjacent sounds, as highly questionable. There can be no doubt that the o-vowel in the forms recorded and discussed by Cantineau (see above) corresponds to the i-vowel in corresponding forms in related dialects recorded by other dialectologists. But my contention is that this o-vowel in Cantineau's notation not only corresponds, and is not only related, to the ivowel in the notation of analogous forms recorded by other dialectologists, but Cantineau's 0 (in the specific forms here under discussion) is to be considered as completely identical with the vowel-phoneme ordinarily conceived, and recorded, as i. I stressed this fact already in a note published in the Hebrew periodical Tarbiz (Tarbi~), vol. 15 (1944), p. 215. As I made clear in this note, the vowel recorded by Cantineau as 0 is in reality the 'indifferent' vowel ('Mittelzungenvokal') a, which (in the Arabic dialects concerned) has a strong affinity with i. It seems, whenever it develops, to arise (at any rate originally) as a more or less voiceless vowel-phoneme. As far as its affinity with i is concerned, it is important to note that the speakers of the respective dialects themselves identify it (that is: this vowel a) with i; they call it kasrah. It may be assumed that this phoneme, which in the instances under discussion (see above) emerged as a newly created vowel, developed in due course into an ordinary i-vowel (articulated with full voice). In any case, this vowel-phoneme, which

a,

40

THE VOWEL I AS AN AUXILIARY VOWEL

has been frequently observed in modern Arabic dialects, should be identified with the specific vowel i (Arabic kasrah, Hebrew bireq, etc.) appearing in numerous Old Semitic forms discussed in the present study (see above). This vowel i, in these Old Semitic forms, according to my interpretation, is to be considered as newly developed (for the purpose of the 'break-up' of consonant clusters, etc.), exactly as in many modern Arabic forms, which in their development are comparable to these Old Semitic forms. The fact that in the traditional writing systems of the various Old Semitic languages this vowel is expressed by the ordinary sign for the i-vowel, corresponds to the fact that the speakers of the modern Arabic dialects in which comparable forms occur, consider in their turn this vowel-phoneme as i (kasrah) , although for the foreign observer (the observing dialectologist) it appears to be the vowel a. The identity of the auxiliary vowel i (in Old Semitic forms as well as in comparable forms of modern dialects) with the 'indifferent' vowel (,Mittelzungenvokal') a became clear to me when observing that in certain forms of the Maghrebian dialects the vowel a (replacing an original a) clearly fulfils the same function as the vowel i in comparable forms in other dialects (recorded by different investigators); see, e.g., below p. 62 and p. 83, n. 83. Furthermore, I conceived the complete identity of i (as auxiliary vowel in Old Semitic as well as in the modern dialects), when (after the original German version of the present study had already been in process of publication) I had an opportunity to observe the vowel a in analogous forms in a dialect belonging to the dialect group studied and recorded by Cantineau in his Etudes sur quelques parlers de nomades arabes d'Orient (see above p. 35). Moreover, I noticed the identity of this modern Arabic vowel a with the Old Semitic 'auxiliary vowel' i by observing 'Iraqi-Arabic speech, especially by recognizing that this vowel a appears under entirely regular conditions (comparable to the appearance of the Old Semitic 'auxiliary vowel' i) in the Arabic dialect of Bagdadi Jews. In this connection I refer to my study "A phonetic rule in the Judeo-Arabic dialect of Bagdad: a > a in pretonic closed syllable"; see the present volume, below p. 94-97 (originally published in Der Islam, vol. 36, 1960, p. 99f.; with a supplementary note, Ibid., vol. 37, 1961, p. 263). (As far as the identification of this a in pretonic closed syllable in this dialect with the 'auxiliary vowel' i is concerned, I refer to below, p. 95, [see Der Islam, I.e., p. 100D.

THE VOWEL I AS AN AUXILIARY VOWEL

41

On the other hand, it is clear that whenever the ('indifferent') vowe1phoneme a (which possesses an affinity with i) emerged (under specific circumstances, see above), it was apt to develop in due course into a (regular) i-vowel (uttered with full voice). Cantineau's notation of this vowel a (which possesses an affinity with i and is, at the beginning, a more or less whispered [voiceless] vowel-phoneme) as 0, seems, as far as his notations in his Etudes sur quelques parlers de nomades arabes d'Orient are concerned, to be based on erroneous observation (see my note in Tarbiz [Tarbi..s-], vol. 15 (1944), p.215). The same vowel 0 is also very frequently found in Cantineau's book Le dialeete arabe de Palmyre (Beirut 1934), and it may be assumed that also here the notation 0 should be replaced by a (or i). It cannot be doubted that the notation i which in analogous forms has been used by investigators of other dialects, refers to the same vowel-phoneme, which according to current phonetic usage is rendered as a. The notation i of this vowel-phoneme cannot be considered erroneous, since the vowel in the forms concerned (if not exposed to the influence of neighbouring sounds, as, e.g., labials or emphatic consonants) has a clear affinity with i, and is also identified with i by the speakers of the dialects concerned, and moreover rendered as i in the traditional writing-systems of the Old Semitic languages (see above). The notation as 0 is erroneous and misleading. That the 'auxiliary vowel' i of the Old Semitic languages originally has arisen as a 'voiceless' (quasi whispered) vowel, was assumed by me from the outset (see below p.48ff.), although this peculiarity of this i is strongly confirmed through its obvious identity with the vowel a in certain modern dialects (see above). We must now still consider certain 'Iraqi-Arabic forms which deviate from the regular paradigm of the perfect in 'Iraqi-Arabic. We quote Meissner, I.e., p. XLI, n. 2: "Die Meidanaraber bilden die 2. Pers. Masc. und die 1. Pers. Sing., jedenfalls nach Analogie der Verba tert. infirm., auf eO)t: 'yrfe(i)t 'ich habe erfahren'; gry~e(i)tni 'du hast mich gekneipt'. Diese Formen Mrte ich auch in Kuweirisch zuweilen von kleinen Jungen. Rachid verwarf sie aber als meidanisch (s. auch Geseh., 42,20f.) und spricht: 'artif(e)t, gerti..s-(e)tni". Meissner's assumption that the long vowel preceding the final t was due to an analogy with the verbs tert. i has also been adopted by Barth (ZDMG 57,772) and Noldeke (Neue Beitriige zur semit. Spraehwiss., p. 239). In reality, the length of the vowel (which is probably actually due to the influence of the verbs tert. j) is only a secondary lengthening

42

THE VOWEL I AS AN AUXILIARY VOWEL

of an already existing short vowel in these fonns, as, e.g., just in the cited instances 'ardfet, gerd~etni (for further examples see above); and such a vowel appears also in some other Arabic dialects (in similar verb-fonns) as an auxiliary vowel for the purpose of breaking up a consonant cluster after the disappearance of the terminal vowel, as, e.g., glarl sarMet 'I fled' (cf. Brockelmann, Grundriss, I, p. 573; also Rhodokanakis, I.e., §61 h). Once such a vowel was present, it was tempting to connect it with the long vowel which precedes the personal ending t in the tert. i and to lengthen it. 44 These forms of the Maidan-Arabs are also interesting and important for another reason. We recognize in these fonns the same accentual shifts which have taken place in other 'Iraqi-Arabic dialects in the 3rd pers. sing. fern., ktfbet, and in the 3rd pers. pI. masc., ktibau (fonns with personal endings beginning with a vowel), and in the 3rd pers. pI. fern., ktiben (with a personal ending secondarily [not originally] beginning with a vowel: -en < -n < -na), and also have taken place in Arabic dialects outside 'Iraq (and also in certain city-dialects of 'Iraq) in the same fonns-with the accentual shift having become effective in the opposite direction-in kitbet, kitbu, kitben. In the fonns of the Maidan-Arabs 'yrfet, or *'yrfet (to mention for clarity's sake, the basic fonn without lengthening), 'I have learned', gry~etni, or *gry~etni, 'you have pinched me', these developments have also taken place in other forms (with a personal ending which, in difference from ktfbet, ktibau, has a secondarily developed initial vowel). *'irfet 'I have learned' is directly to be derived from the alternate fonn 'ardfet « 'araftu) by assuming an abrupt retreat of the stress, exactly as the 3rd pers. sing. fern. kitbet, used in the same dialect, has (as we assume) developed from *katdbet « kdtabat).45 44 Cf. also Brockelmann, Grundriss, J, p. 575: "Nach Abfall des a [namely, in the 3rd pers. pI. fern. katabna] wird im Neuarab. ein Hilfsvokal angenommen in ·omlin. ketben .... 1m Dialekt von Dalina wird der Hilfsvokal nach den III i wie beim M[a]s[culin] gedehnt ... ". 45 In the form grY:jetni, or *gry:jetni, respectively, "you have pinched me", instead of gerd:j(e)tni, we recognize a form (of the 2nd pers. sing. masc.) which in its developmental structure corresponds to the 3rd pers. sing. fem.-form (i)ktibet which exists beside the form kitbet. This form (i)ktibet (see above) is necessarily based on a form *kdt"bat (> ktibet). The form underlying *grnetni can only be its alternate form gerd:jetni, with the accent on the 2nd syllable. It seems, however, that the shift to gry.~et-, or *grnet-, respectively, has taken place especially in connection with the suffix -ni (and related suffixes). In the form gerd:jetni, an accentual shift unavoidably had to take place after the auxiliary vowel preceding the consonant t had achieved the value of a regular vowel (in other words: had become established). The enclitic

THE VOWEL I AS AN AUXILIARY VOWEL

43

While in Iraq-Arabic this transformation of the 1st pers. sing. of the perfect, that is: 'yrfet < *'yrfet, is (as it seems) restricted to a single, local dialect, in certain old Semitic dialects, namely in BiblicalAramaic and Syriac, this typical structure is the regular structure of the form concerned: qitiei ('I have killed'). This is an analogy to the fact that the 3rd pers. sg. fern. of Aramaic, qitlal, has its parallel in the 'Iraqi-Arabic dialects (as well as in other modern-Arabic dialects). Also in this form (1st pers. sg.), qitiei < *qetalt (with objectsuffixes still qetalteh, etc), the terminal consonant cluster has been broken up by an auxiliary vowel (see Brockelmann, Grundriss, I, p. 573). ludaeo-Aramaic, according to Brockelmann (Ibid.), "ersetzt diese Form meist durch eine Neubildung iftalii ... nach dem Muster von iftal, iftalt". Brockelmann does not say anything about the fate of the vowel of the second radical in the form *qatalt(u) which underlies qitiei. (The i in qitiei has, he assumes, originated from a in a form *qatiei (see his Syr. Gramm. [5th ed., 1938], §§ 74, 44), that is: i from a in closed syllable). Aramaic qitiei (1st pers. sg. of the perfect) is a case in which the existence of an i in closed syllable, replacing an original a, must, as in the already discussed modern-Arabic instance *'yrfet, necessarily be connected with the disappearance of a vowel which originally followed it; indeed it is the only possible explanation for the absence of this vowel. In a form *qataiei or *q etalel, with a weakly stressed vowel of the first syllable-cf. Drari sarddet 'I fled', 'Iraqi 'arafet 'I have learned' (the variant and basic form of *'irfet), ludaeo-Aramaic if!alii (which does not necessarily represent a new form, as Brockelmann, I.e., believes [cf. Cantineau, I.e., I, p. 95], indeed may, with considerable certainty be considered the immediate precursor of the form qi!leJ)-an abrupt retreat of the stress took place because of the awkwardness of its syllable structure; as the bearer of this abruptly retreating stress a new vowel, i (iJ), developed (between the first two radicals), and the subsequent vowel (between the second and third radical), which became deprived of its stress, was elided: qitlel, 46 as with the Maidan-Arabs *'y~ret (> -ni required placing a stress on the penultima. This, however, was only possible, if the main-stress was distanced to the first syllable in order to avoid the collision of two stressed syllables: *gara$etni (*ger~etni) > *gara$etni. At this stage the accentual shift to *grY$etni probably took place, in analogy with the development of the 3rd pers. sing. fern. in 'Iraqi-Arabic: *katabet> *katObet> ktibet. In relation to the material at our disposal, it is not easy to decide such questions with certainty. 46 Cf. above p. 15, n. 12.

44

THE VOWEL 1 AS AN AUXILIARY VOWEL

'yrfet) < 'arafet (exactly, as, e.g., in the imperative-forms [see above p. 7] qi!li < *qe!6li).47 Bauer and Leander, Grammatik des Biblisch-Aramiiischen, p. 102 p, trace back qi!lel to *qatlel « *qatalt), as does Brockelmann. For the disappearance of the vowel of the second syllable in qatalt they refer to Ibid., p.67/8 (§ 18 u) where the following rule is laid down: "Gingen der Drucksilbe zwei kurze, freie Vokale vorher, so wurde der zweite elidiert (wie im Hebr. vor Nebendrucksilbe, BL 26 w'): *hm;!arii (vgl. hebr. hii(jiir) > hm;!rii 'die Hoheit'; *mala!sifl (vgl. hebr. melii!sim) > mal!sin 'Konige'; *'al]adel (§ 12 x) > 'al]del 'ich habe getan'; *dahal]ii (vgl. arab. gahab) > dahJ.2ii ... ; basarii (vgl. hebr. biisiir) > *basrii> (6x) bisrii 'das Fleisch'; *saliqal (vgl. seliqii) > *salqal > silqal 'sie stieg hinauf .... Die verschiedene Behandlung der Vokale zeigt, dass die 1. Silbe einen Nebendruck hatte". It is not very clear to me what is meant by "die verschiedene Behandlung der Vokale" ("the varying treatment of the vowels"), which "presupposes a secondary stress (,Nebendruck') on the first syllable". Apart from this, in this formulation of the rule which, to my knowledge, was first laid down by Brockelmann (Grundriss, I, p. llO)-who, however, rightly does not mention qitlel in this connection - a proviso seems important to the effect, that the intermediate vowel in words like hm;!rii < *hm;!arii, mal!sin < *malalsin, etc., could only drop out, if not only the third syllable, but also the first one was stressed before the elision took place. That this stress on the first syllable was - in all cases-a secondary stress (,Nebendruck'), would become certain, if we knew for sure, that the medial vowel disappeared only after the general shift of the main stress to the ultima had already taken place. It is, of course, also conceivable that this elision took place earlier, perhaps, for instance, when in the words concerned the stress was distributed as follows: *hagara (and not *hd4ara) > hm;!rii; etc. What is important here, however, is the assumption that vowels between two stressed syllables were affected by elision. This condition (elision of an unstressed vowel standing between two stressed syllables) can under no circumstances apply to the case in question here (qitlel), although Bauer and Leander subject it to the same rule (and similarly 47 Naturally, here, too, as in other cases of an abrupt retreat of the accent, a vowel of the same quality as the original unaccentuated one which occupied this position (the initial syllable), that is: a, may serve as bearer of the accent; indeed, in a syllable with an initial laryngeal the new stressed vowel is ordinarily adapted to it, i.e., appears as a, as, e.g., Biblical-Aram. 'i1QdeJ, 'amreJ.

THE VOWEL I AS AN AUXILIARY VOWEL

45

the instances silqal, bisrii [see above p. 44] are not subject to this rule, although their basic forms may have been stressed in the manner postulated by Bauer and Leander). In a form *qatalel the second syllable necessarily bore the stress, as did its underlying form *qa!alt, from which it differed only by the auxiliary vowel which broke up the consonantal cluster (double consonant) in the terminal syllable. In *qatalt, the first syllable was unstressed since the syllable -talt, because of its particularly sonorous structure, received the main stress, and a second stress, even a slighter one, was impossible immediately preceding this strongly stressed syllable (-talt). Bauer and Leander, however, give no reason whatever, why the first syllable should have received a secondary stress (,Nebendruck') after the secondary vowel had appeared, that is: after the emergence of the form *qatdlel (or *q etdlel, if we assume that this secondary (auxiliary) vowel came into existence in the form *qetalt < *qatdlt, developed like qetal < *qatdl). Bauer and Leander set the elision of the medial vowel in all instances mentioned by them into the period in which the shift of the main-stress to the ultima had already materialized (as *saliqdl > *salqdl > silqdl, and similarly also *qataiel > *qatiel > qitiel). However, just in a form like *qataiel, where the accent could have reached the ending -el only by moving forward from the second syllable -tal-, its original position (as, e.g., *qatdlt> *qatdlel > *qataiel), this forward move must have necessarily brought about a weakening of the stress on the first syllable, so that a secondary stress on this syllable is out of the question. As a matter of fact, the development *qatdlel > qi!lel is unexplainable if one assumes an earlier stage *qataiel, in which the accentual shift to the ultima had already taken place. The form qitlel must have originated from *qatdlel or *q etdlel already before this shift of accent had taken place. Cantineau (Bulletin d'erudes orientales, II, p. 132), also starts, as I do, from a form *qatdlel; however, he relates the disappearance of the medial vowel directly to the shift of accent to the ultima, which he also does in the case of the 3rd pers. fern.: qi!ldl < qa!dlal (see my objections to it, above p. 16).48 48 Wright, Lectures on the compo gramm., 175/6, who equates the terminal -el with -al (see the details in his book), explains this form as follows: "qit/el, qet1el, which stands for qat/a!., by transposition from qatalt, which is the form used in vulg. Arab .... ".-Barth (ZDMG 57, 771) suspects with respect to the ending -el (which he reads -el) that it "auf einer Analogiebildung des starken Verbs nach den Wurzeln ult. i beruhe", and considers as the reason of this 'Analogiebildung' "dass sonst die

46

THE VOWEL I AS AN AUXILIARY VOWEL

This appearance of i (in an originally open syllable) has some connection with what Stumme, Miirchen und Gedichte aus Tripolis, § 23 (p. 215), states regarding the predilection for i in the various types of groupings of syllables with (seemingly) 'transposed' vowels (" 'umgesprungene' Vokale") in the Maghrebian dialects, especially in the Tunisian. I am quoting the following especially interesting passage (l.c., § 23,3; p. 216): "Wie in Tunis ursprungliches a nach der ursprunglichen Tonsilbe, sobald es betont wird, gem zu i oder einem dem i iihnlichen Laute verwandelt wird [v. Stumme, Tunis. Gramm., §2, p.4], so geschieht dies auch in Tripolis. Diese Vorliebe fUr i ubert6nt oft alle sonstigen in Frage kommenden Momente (als Vokalharmonie u.a.). Beispiele: maiinnii 'unsere Zisteme' (ohne Suffix: mazen), maklithii 'ihre Speise' (0. S. mlikta), dabblythum 'sie brachte sie hinein' (0. S. dabblet). - Nach ' und ~ ist das Substitut fUr jenes sekundiire i ein e (fj) : ref ethum 'sie hob sie empor', §ib~ethii 'sie sah sie'; nach wist's ein u, welches auch in der Nachbarschaft von Labialen hiiufig eintritt: kiswUtha 'ihr Anzug', rabbUtni 'sie zog mich auf". That a stress which is shifted to the penultima after the accession of a suffix (enclitic), transforms an original a of the feminineending into an i after it had been reduced by a weak stress and (respectively) also by a diminution of function, is a phenomenon which occurs in many dialects. Also Socin (see quotations above) mentions the feminine-ending -it in connection with the i 'attenuated' ('zugespitzt') from an original a. In mliklii, e.g., the vowel of the feminine-ending a( t) was reduced (affected by imiilah, which in this case is somehow identical with 'reduced') and the fullness of its sonority was also very considerably weakened by its terminal position. When, through the accession of a suffix (an enclitic), the syllable constituted by the feminine-ending (i.e., the penuItima) necessarily had to receive a strong stress, this 2. Pers. sing. fern. und die erste Pers. comm. sing. in der Form qa!al-t(i) zusammengetroffen waren". As also Brockelmann (l.c., p. 573, n. I) stresses, Barth's assumption is impossible from the outset, because of the "standige Kiirze des e". Such an analogy, caused by the desire to distinguish forms, would indeed be highly strange in this case. The Syriac form must by no means be separated from the corresponding analogous forms of the modern Arabic dialects (with a short auxiliary vowel preceding the terminal t) about whose origin no doubt is possible. Chr. Sarauw, Zeitschr. f Assyrial. 22 (1908), p. 57, derives qi!lel (on the basis of a peculiar and not very convincing analogy-process) from *qe{f;l(}l (cf. forms like Bibl.-Aram. 3rd pers. sing. fern. 'a m(}r(}l < *'amart < *'amarat, see above p. 16) which, on its part, presupposes *qa!alt.

THE VOWEL I AS AN AUXILIARY VOWEL

47

stress was not placed on the reduced (dissipated) vowel. This latter probably ordinarily had disappeared from consciousness, so that a new vowel had to develop as the recipient of this stress: maklii, but maklfthii. Of interest are related phenomena in 'Iraqi-Arabic. Here, the ending -at (but, indeed, here not the -at which serves as the feminine-ending of the noun, but -at as ending of the 3rd pers. sing. fern. of the perfect) becomes -it only if a suffix starting with a vowel is added, e.g., gqbbqt 'she threw away', but + -q (,him'): gqbbitq; akqlqt 'she has eaten', but + -q (,him'): akqlitq; on the other hand, + -nq ('us') akalgtnq (cf. Weissbach, ZDMG 58, 933, who considers this process [i for a] as a differentiation [dissimilation] of two a's in succession). Thus, this 'Iraqi-case represents a special phenomenon. One cannot maintain in this case that the vowel a of the affix of the 3rd pers. sing. fern., -at, when terminal and unstressed, had become so dissipated that a new vowel, i, had to take its place, when an enclitic was added and this afformative, -at, had to be strongly stressed (enclitics frequently throw the stress on the penultima). The quality of this vowel apparently always remained preserved. Forms as akalqtnq show this clearly. The fact that a was replaced by i if a suffix starting with a vowel (or represented by a vowel) was added, as in akqlftq, must be due to the change of the syllable structure brought about by the accession of such suffixes. In a form like akalqtnq the -t goes with the preceding vowel: akalqt-nq. This preceding vowel was thus protected by the following consonant and therefore not exposed to the danger of dissipation. However, when a suffix starting with a vowel was added, the -t of the ending was drawn to the acceding vowel: *akalq-tq. The vowel preceding the t was thus no longer protected, it had, as it were, obtained the status of a terminal vowel, and was therefore apt to become dissipated, as any unstressed terminal vowel is apt to do. However, since the suffix starting with a vowel as an enclitic required a strong stress on the preceding penultima (see above)-also, by analogy with forms like akalqtnq, in which the syllable -lat because of its sonority necessarily received the main stress - a new vowel, i, had to be interjected as carrier of this stress. I have already dealt with i as an 'auxiliary vowel' in the function of breaking up unpronounceable consonantal clusters in my Materialien und Untersuchungen (see above p. 4), p. 88, in connection with certain forms about which certain information has been handed down to us by medieval Arab phoneticians. I have drawn attention

48

THE VOWEL I AS AN AUXILIARY VOWEL

(see l.c.) to the role of i as an 'auxiliary vowel' in words which begin with two consonants (prosthetic vowel) [with reference to the statements of A. Fischer, Islamica, I (1926), p. 6 ff.; cf. also Landberg, Glossaire dalinois, II, 1923, p. 1599]; and I have also (Materialien, Ibid.) referred to its related function of breaking up consonantal clusters in the middle of a word. I have moreover concluded, from certain statements and definitions of medieval Arab scholars (see l.c., p. 82ff.), that the extremely frequent use of i as an 'auxiliary vowel' is due to the original voicelessness of such auxiliary vowels, which agrees with the phonic nature of the vowel i far more than with the phonic character of other vowels. I have, unfortunately, failed to mention in my earlier publication that the auxiliary vowel in these cases is identical with what appears in the modern dialects as 'interjected' vowel (" 'eingesprungener' Vokal") as a consequence of the process of breaking up clusters of consonants and of the abrupt shifting ('Umspringen', 'transposition') of the accent. This easily leads to conclusions as to the nature and origin of such vowels also in modern dialects. More details on that point further below. - The forms mentioned l.c. (Materialien, Ibid.), for whose newly interjected vowel 'whispered articulation' is recommended (as I understand the statements of the Arab phoneticians), are mainly taken from the Qur'an, as jabi~!jimuna, and also jab~~imuna, < jabt~imuna; jahiddi and jahaddi < jahtadi; etc., also ni'immii. The phenomenon is also mentioned by the grammarians, see ZamalJSari's Mufa~~al, §756, and Ibn Ya'g's Commentary, p. 1487,11 ff. There the root qtl, which contains the dental t as second radical, is used as an example; and for the perfect, the form qittala (instead of iqtatala) is cited, for the imperfect, the forms jaqattilu, jaqittilu, jiqittilu and finally jaqttilu 49 (instead of jaqtatilu). Of the coexisting forms jaba~!jimuna, jabi~~imuna, etc., Brockelmann, Grundriss, I, §68 b. c. (p.181) considers the forms with a as the more original ones and interprets the forms with i as due to 'harmonization' of the vowels (while noting that these are the only instances in which, as he assumes, this process has taken place in medial position, "im Innern von Verbalformen"). I have already pointed out (l.c.) that in these transformed forms of the 8th 'stem' of the verb the vowel i must be considered as the only 49 The interpretation of this form by Vollers (Volkssprache und Schriftsprache im alten A rabien, p. 113) as *jaqtetilu does not have the least support in the statements of the grammarians. Ibn Ya'iS expressly says: biddigiimi t-tii'i fi t-tii'i ma'a sukuni l-qiif Cf. Fleischer, Kleine Schriften, I, p. 131/2.

THE VOWEL I AS AN AUXILIARY VOWEL

49

(quasi) legitimate auxiliary vowel for breaking up consonant clusters in the forms, postulated by me as intermediate forms, *jahddi, *jab~~imuna, 50 which agree with the form jaqttilu, which has been actually handed down. The vowel i, as a matter of fact, is the only vowel attested (or handed down) in the also mentioned ni'imma (which thus has no variant with a). As to the vowel a, as far as it is attested in the case of certain forms (see above), it is to be considered due to the memory of the original scheme of these forms of the 8th 'stem', that is : jaqtatilu (cf. p. 9). In connection with my conclusions (which will be substantiated in more detail in the sequel), the following remark of Stumme, Miirchen und Gedichte aus Tripolis, p. 233, n. 1, seems of interest and important: "Auch im Klassischen - dies mage hier hinzubemerkt sein - gibt es wohl Beispiele fUr aufgesprengte Formen, bei denen also die Betonungsgesetze gleichfalls verletzt werden. Wenigstens vermute ich, dass ein fUr ntma ma stehendes ni'imma (s. Caspar iWright, Grammar of the Arabic Language, 3. edit., vol. I, p. 97) gelesen werden muss: nf'imma (und nicht ni'imma, wie der Buchstabe des Gesetzes verlangt)". While Stumme mentions this instance (nf'imma) as an old-Arabic example for the phenomenon of 'break-up', 'Aufsprengung' (a phenomenon otherwise known only from modern Arabic dialects), the medieval Arab phoneticians mention this instance (according to my interpretation of their statements) as an example for the whispered pronunciation of vowels (a phenomenon which, conversely, has been hardly observed with certainty in modern dialects, indeed has been contested; see below). If, according to Stumme, an interjected vowel ('eingesprungener Vokal'), although preceding a geminate, does not receive stress (which rather remains on the originally stressed syllable), one may, conversely, conclude, in view of the statement of the Arab phoneticians that this vowel (the second i in ni'imma) was reduced, i.e., whispered, 51 that the 'interjected' vowels 50 The speaker, indeed not even the listener, did not necessarily have to become fully aware of such construed transitional forms. One could imagine that at the very moment when in jabt~imuna the a between t and $ was dropped and thus t and $ moved together and became $$, a whispered vowel was interjected between band $$ (cf. Ibn Ya'iS, Ibid., line 16). However, such forms as *jab$$imuna may have very well existed temporally since we find such sequences of sounds in modern Arabic dialects; cf., e.g., bjub$$ani in S. D. Goitein, Jemenica (Jerusalem 1934), p.6, no. 25 (see below p. 6lff.); they are also expressly substantiated for the ancient period in the form jaqttilu (see above p. 48). slOne should be aware of the fact that in this context, that is, with reference to whispered vowels (voiceless vowels), the expressions 'reduction', 'reduced', etc.,

50

THE VOWEL I AS AN AUXILIARY VOWEL

('die eingesprungenen Vokale') of modern Arabic dialects, which originate through 'break-up' (,Aufsprengung') and similar processes and which ordinarily, contrary to the prevailing accentual rules, do not bear the stress, are not spoken with full voice, but rather whispered (at least when they first come into being). Reliable observations regarding the reduced, voiceless quality of these special vowels are non-existent so far. However, the existence of whispered vowels in the Moroccan-Arabic dialect has been asserted by H. Grimme (first briefly in the Festschrift in honour of E. Sachau, p. 141, then, at length, Zeitschr. for Semitistik, vol. III (1924), p. 1-16), without, however, taking into consideration the secondarily developed, 'interjected' vowels ('eingesprungene Vokale'). A. Fischer, ZDMG 71, 446, has raised strong objections to Grimme's statements. In spite of his opposition, however, (and of that of Priitorius, see Fischer I.e.), I am inclined to believe in the existence of whispered vowels in the dialects in question, and to assume whispered pronunciation especially for those secondary vowels which, contrary to the prevailing accentual rules of Arabic, bear no stress. In this connection, the observations of the medieval Arab phoneticians must not be disregarded, namely their observations that certain vowels (vowels which can be shown to represent secondary vowels developed through 'break-up' of consonantal complexes) are uttered with a reduced pronunciation, which, if I interpret their statements correctly (see my Materialien und Untersuchungen, p. 84 ff.) must necessarily be a whispered, voiceless pronunciation. The Arab philologists, especially "the Readers [of the Qur'an]" (al-Qurrii'), had an ear for such matters no less sharp, and probably sharper, than ours. When Fischer emphatically rejects Grimme's assertion and declares the vowels observed by him as whispered vowels to be murmured vowels, his argumentation does not carry much conviction, for the difference between a whispered voice and a murmured voice is not an absolute one, but one of degree (see Sievers, Grundziige der Phonetik 5 , § 84 fT.). 52 New and more exact never signify quantitative weakening (abbreviation). Rather, whispered vowels, which sometimes bear strong stress, indeed often owe their existence to such a stress, are to be considered, with respect to quantity, as completely equivalent to regular short vowels (short vowels articulated with full voice). They are only 'reduced' to the extent that they are not articulated with full voice. 52 Similar differences of opinion with respect to the character of reduced vowels, whether it concerns murmured, whispered or otherwise pronounced vowels, seem also to exist in Indo-European linguistics; cf. H. Hirt, Indogermanische Grammatik, IV (Heidelberg 1928), p. 350/1.

THE VOWEL I AS AN AUXILIARY VOWEL

51

observations, of experimental nature, if possible, as postulated by Grimme (I.e., p. 16), seem to me to be absolutely necessary. In view of the observations of the medieval Arab scholars and in view of the facts which do not permit any other satisfactory explanation, I, for one, am inclined to assume that the vowels which result from a 'break-up' (,Aufsprengung') and similar processes are originally whispered vowels. However: which are the processes that have produced forms like ni'irnrna, ja/Ji!!ifu, jahiddi, etc. « ni'rna rna,53 ja/Jta!ifu, jahtadi) and thus have caused the auxiliary vowels to arise? Brockelmann originally (i.e., ZDMG 59, 630) treated the forms concerning us here, in which the vowel between two identical or similar consonants was elided, as examples of 'haplological ellipsis of a syllable' ('haplologische Silbenellipse') together with all other cases of haplology (i.e., elision of sounds [consonants] within a sequence of two syllables beginning with identical or similar consonants). But after A. Fischer (ZDMG 60,249), with reference to Brugmann, contested the conception of these forms as instances of the phenomenon of 'haplology', in that he saw in them "Synkopen ... z[um] T[eil] mit nachfolgenden Assimilationen", Brockelmann excluded them in his Grundriss from the chapter on "Haplologische Silbenellipse" (Grundriss, I, p. 257ff., see also p. 259, n. I), by understanding as cases of haplology only those cases of elision of sounds in which a complete syllable was eliminated, i.e., consonant plus vowel. He therefore treated our examples in Grundriss, I, §96, in a special chapter, titled: "Dissimilatorischer Vokalschwund und dissimilatorische Metathesis". The demarcation of the concept of haplology in this manner seems unjustified to me. Even if the consequences of the process in both categories are different, the causes and even the process itself can only be identical. Examples such as that presented in the § on "Dissimil. Vokalschwund": ia(jgakkaru < jata(jakkaru cannot be separated from the case treated in the § on "haplology" (p. 261): taqattalu < tataqattalu. Brockelmann, ZDMG 59, 629, defines the concept of "haplological ellipsis of a syllable" in the following manner: "Unter haplologischer 53 Brockelmann (Ibid., b) derives the form nrimma from nrima ma. Already in my Materialien und Untersuchungen, p. 86, n. 2, I remarked that the statement by the indigenous phoneticians concerning the reduced character of the vowel preceding the geminate -mm- necessitates to consider this vowel as a secondary, auxiliary vowel, and that, accordingly, ni'imma must be derived from nima ma.

52

THE VOWEL I AS AN AUXILIARY VOWEL

Silbenellipse, auch syllabische Dissimilation, syllabische Superposition oder Silbenschichtung genannt, versteht man den Vorgang, dass zwei Silben, die, entweder unmittelbar aufeinander folgend oder durch eine Silbe ungleicher Lautung getrennt, den gleichen oder sehr ahnlichen konsonantischen Anlaut haben, oder von denen die zweite denselben Konsonanten im An- und im Auslaut hat, bei simultaner Assoziation in eine verschmolzen werden (s. Brugmann, Kurze vergl. Gramm. der indogerm. Sprachen, § 337)".54 The presupposition for "haplological ellipsis of a syllable" therefore is not the complete identity of two succeeding syllables, that is, identity of consonnant and vowel, but simply the identity (or similarity) of the (initial) consonants of the two syllables. Notwithstanding the fact that only the two (identical) consonants separated by a vowel are in conflict with each other, the effect of this conflict consists in any case in the elision of the vowel separating the two (identical or similar) consonants. In cases like ni'immii, iabi!tifu, etc., which Brockelmann incorrectly excluded from the phenomenon of haplology, we see retained precisely the two consonants whose succession was felt disturbing, while the vowel separating them was elided. It is interesting that these particular cases, which appear quite frequently in Arabic, occur also in Indo-European languages, even if, as it seems, relatively infrequently. I refer to Brugmann, Das Wesen der lautlichen Dissimilationen (Abhandl. der phil.-his!. Klasse der siichs. Gesch. der Wiss., 27. Band, Leipzig 1909, p. 152, n.1): "BeiHi.ufig eine Vermutung uber eine Klasse von Lautentziehungen, die man mit den Verkurzungen wie K€AUlV€ gibbiir, giltii', sikkiif. In the same light I see the Algerian word-forms mentioned by M. Cohen (see above), i.e., forms like baltib beside bltib. That means: in contrast to Cohen, I do not consider bttib as based on baltib, or tfkira based on tafkira. But I consider bltib and tfkira as developed

62

THE VOWEL I AS AN AUXILIARY VOWEL

from the more original forms batfib « ba!ib) and tafkira, respectively, and recognize in ba!!ib and tafkira forms developed from btfib and tjkira, respectively, through 'break-up'. of the consonant cluster by means of the 'auxiliary' vowel a. This a in bat!ib, tafkira, etc., corresponds to the vowel i which appears in analogous forms of other dialects according to the observation of (and manner of notation used by) many other dialectologists. I refer in this connection to my remarks above p.60ff. Accordingly, bat!ib « *ba!!ib) may, e.g., be compared with Lebanese niZiiir « original naggiir) mentioned above p. 61. But what are the causes for the disappearance of the vowel before a double consonant (thus in closed syllable) in cases where one cannot explain it as having been absorbed by adjacent consonants? In the disappearance of the vowel in cases such as gabbiir> *gbbiir (and also in the analogous instances cited by M. Cohen) one can think of the long vowel of the second syllable having been a partial factor (see also M. Cohen, l.c., p. 162, n. 1). But this factor, although to be kept in mind, as we shall immediately see, is not involved in many other cases where a in closed syllable is replaced by i, and also in those cases where it is involved, cannot have caused the reduction and disappearance of the vowel by itself alone. In my opinion we are dealing here with a particular type of gemination or with a particular type of vowel pronunciation before a double consonant, no matter whether the double consonant is composed of two identical or two different consonants. I am particularly thinking of that difference in accentuation of syllables or vowels which Sievers, Grundziige der Phonetik, § 589 ff. designates as pronunciation "mit stark bzw. schwach geschnittenem Silbenakzent", and which Jespersen, Phonetik (4th ed.), p. 202, esp. n. 1, calls pronunciation of vowels "mit festem bzw. losem Anschluss" (i.e., "with tight or loose contact with the consonants following them"). A vowel before a geminate or other types of double consonants demands, according t.O the usual interpretation and observation (cf. Sievers, Grundziige der Phonetik, § 560) exactly this exclusive pronunciation with 'strongly cut' (,stark geschnittener') syllable accent: "er muss im Moment der Verschluss- oder Engenbildung noch mit kraftigem Druck gebildet werden", or expressed differently: the vowel concerned is firmly joined to the double consonants following it. This is the case in the Germanic languages. However, there are languages which differ in this respect. Cf. Jespersen, l.c., p.204: "Die romanischen und slavischen Sprachen kennen im

THE VOWEL I AS AN AUXILIARY VOWEL

63

Gegensatz zu den germanischen nur (oder fast nur?) losen Anschluss; in der italienischen Aussprache von Dante ist der Ubergang zwischen a und n der gleiche wie in der deutschen Wendung da'n Telegramm ... ". Observe further (Ibid., n. 2): "Italienisches jatta zeigt auch losen Anschluss, wodurch der Eindruck von dem des schwedischen jatta mit festem Anschluss verschieden wird; das italienische wiirde sich als ja/tta bezeichnen, ebenso ha/nno,ja/nno, e/cco usw.". I now maintain, however, that in Semitic short vowels appearing before a sequence of two consonants were frequently pronounced with 'weakly cut accent' (,schwach geschnittener Silbenakzent'), therefore with decreasing stress (see Sievers, Ibid., and § 544), and that for this reason these vowels stood only in loose contact with the sequence of two consonants following them. Accordingly, in instances like gabbiir, qa!!ii' the syllable ga- or qa- (respectively) was sporadically (or even frequently) spoken with 'weakly cut' accent, and under these circumstances the second syllable began - in the interior of the wordwith a sequence of two consonants which was not directly preceded by a phoneme. The vowel of the open syllable ga-, not standing in an uninterrupted (tight) connection with the phonemes succeeding it, was thus naturally exposed, under certain conditions, to weakening and reduction. This possibility existed especially if the syllable following the sequence of two consonants had a long vowel, and, as a consequence of this, possessed strong stress, thus drawing to itself the sequence of two consonants (the geminate). But also in instances where there is no long vowel following the geminate, could this reduction take place readily, as, e.g., in the perfect of the 'intensive' stem: qattil 65 > qittil, as Hebrew qi!!el (qi!!al), Moroc.-Arab. gtbbil < qabbala (Socin, Zum arab. Dialekt von Marokko, p.33,20), birrig 'holte heraus' (p. 34, I), IJurikkibhii 'und steckte ihn an' (p. 36,11), $illii 'betete' (p. 34,2), niggiihii 'hat sie errettet' (p. 30,2), etc., beside barreg (p. 36,3), $allii (p. 34,2; 46,18), kabbarhii 'hat sie gross gezogen' (p. 30,2), etc. 66 Thus also, of course, in the Vth stem, as tebi(d)dlit 65 This form, qattil (with the a of the 1st radical preserved), is present in Aramaic qa!!el (Pa"el) < *qattal (cf. Arabic qattala and less frequent Hebrew forms as qiddas, etc.). The perfect form *qattil developed, as is well-known, through adaptation of the vowel a of the 2nd radical to the vowel of the 2nd radical of the imperf., a phenomenon which we also find frequently in Arabic dialects and also in Ethiopic languages (see H. J. Polotsky, Bulletin de la Soc. de Ling. de Paris, vol. 39 (1938), p. 174). 66 In his article Zur Bedeutung von Akzent und Vokal im Semitischen, in: ZDMG 64, 269-311, H. Torczyner has sought to show forms as Hebr. qi!!el (as he also does analogously with respect to certain other forms) as original (primary) variations

64

THE VOWEL I AS AN AUXILIARY VOWEL

< class. tabaddalat (p. 30,14), further also in the imperf. of this stem, as jitebiddil < class. jatabaddalu (p. 46,11), itelibbid < class. jatalabbadu (p. 34,18). Similarly also in the 'intensive' (or lInd) stem of the praridialect, cf. Rhodokanakis, II, § 65 c: "1m Perf. wechseln die Vokale mit a, i sowohl in der ersten wie in der zweiten Silbe ab". Ibid., d: "Ihr Vokal [i.e., the vowel of the first syllable of the stem of the imperf.] ist trotz des Akzentes zumeist i durch Assimilation, auch dort wo urspriinglich i der zweiten Silbe zu a geworden ist: jibillagah ... ". The substitution of a by i in these forms can, of course, not be interpreted as 'assimilation'. In the dialect of polar i seems to have replaced a also in the imperat., see Ibid., e. - Brockelmann, Grundriss, I, 68 c e, mentions similar instances from various dialects, as in the dialect of I:IaQram6t in the imperf.: jilibbis, jirikkib, jisillim; also in the imperat.: !ibbiq (from !bq II 'zusammenfalten', Landberg, Etudes, I, lfa¢ramout, p. 269, n. 2: "Une fois on disait !abbiq, une autre !ibbiq)"; furthermore (see Brockelmann, l.c.) Maltese mfittis (participle of the lInd stem); Negdi teyiggid 'Liebesweh' (infin. of the Vth stem), tifirric 'Zwietracht', tekillif'sich auflegen', and similar forms. There developed therefore, e.g., from ga-bbiir: *gabbar or *gebbiir (or *gbbiir, respectively), or from *qa-ddas: *qaddru or *qeddas (or *qddaS). The syllable of the reduced vowel could now receive for its protection a (,Nebenformen') of Arab. qat/ala, Aram. qa!!e! (see especially l.c., p. 280ff., and also ZDMG 66, 87, where he speaks of a "kritiklose Gleichsetzung resp. Gleichmachung scheinbar entsprechender Verbalformen"). He thus stands in opposition to the generally accepted interpretation, specifically confirmed by Barth, Nominalbildung, p. XXII-XIII, furthermore ZDMG 48 (1894), p. I ff., through penetrating and convincing arguments. Barth contrasts the forms of the perfect of the intensive stem and the causative stem: Arabic qat/ala, 'aqtala, Aramaic qa!!el, ' aq!el, Hebrew qi!!el, hiq!il with the infinitives belonging to these stems: Ethiopic qat/iiI, Aramaic qa!!iilii, ' aq!iilii, Hebrew qa!!61, however, Arabic: qittiil, 'iqtiil. In accordance herewith qattal( a) must be assumed as basic form, and the other forms are to be considered as based on secondary development. In connection with Torczyner's assertions I wish to refer to the statements of E. Konig, ZDMG 65, 715 ff., and Ibid., 66, 261 ff. In addition I wish to call attention to the forms of the Nijal of roots med. IJ/i and med. gemin., as niiq6m, niisal.! from ,J qym, ,J sbb, beside niq!al from ,J q{l. As a matter of fact, already before Torczyner, Zimmern, ZA V, p. 7 (in connection to certain ideas of Lagarde, Ubersicht iiber die ... Bildung der Nomina, p. 87) as well as Ungnad, Beitr. z. Assyriol. u. semit. Sprachwiss., VI,3, p. 57, had considered the i in Hebrew qi!!el and hiq!il as primary (hence not developed from a) and as proto-Semitic (with respect to Ungnad see Brockelmann's objections in Grundriss, I, § 257 B a, Anm. 2, and G e, Anm. 3). Apart from the necessity to consider the various respective forms in related languages in a relationship of historical dependence from each other (see Konig, l.c., 65, 716), this relationship shows itself in the coexistence of all of these forms in one and the same modern Arabic dialect.

THE VOWEL I AS AN AUXILIARY VOWEL

65

strong stress. However, it was not the old, reduced vowel which was the recipient of this stress, but rather a new vowel, i: > gibbiir, qiddas. 67 If the weakened vowel in question had already completely disappeared, the consonant cluster could be broken up in the simple already known manner: *gbbiir, *qddas> gibbiir,68 qiddas. Thus also, 67 Naturally, the difficulty of such an accumulation of consonants as in *gbbiir, *qUii', *&!!ib, etc., can also be solved by simplifying the geminated consonant con-

tained in this accumulation of consonants. Thus I would like to interpret forms as getii' found in Socin, l.c., line 14 and 19. Cf. also Ibid., p.40, line 2: I;tli 'until', Ibid., p. 34,2: I;era; beside I;tta < I;atta in other dialects (see also below p. 66). 68 There is probably no reason to trace (with Brockelmann, Grundriss, I, p.361, n. I) the i in this Biblical-Aramaic form, gibbar, to Hebrew influence (gibbOr). - The form gibbor < gibbar is also on this account remarkable, that in Hebrew it does not only distinguish itself through this i in place of a in the first syllable from other forms of the pattern qatta/, as gannaq, etc., but also by the fact that the a in the second

syllable is-in accordance with the phonetic rule-shifted to 0, whereas this is not the case in gannaq and other instances with preserved a in the first syllable. Cf. for the non-appearance of this shift of a to 0 in these forms the interpretations of Barth, Nomina/bi/dung, §33 c; Brockelmann, Grundriss, I, § 149 a (p. 361); and also of Bauer and Leander, Hebr. Gramm., p.478/9, who trace forms as gannaq (as also certain other formations) back to a Canaan.-Aram. 'blending' (,Sprachmischung'), hence explaining them as Aramaisms (a similar solution had already been suggested by Lagarde, Obersicht iiber die Bildung der Nomina, p. 89). Similar developments of identical formations in two different directions, and to be sure exactly also in the case of the 'basis' qattiil, are apparently also found in other Semitic languages. Cf. Rhodokanakis' remark (ignored by Bauer and Leander, I.c.) in his St. z. Lex. u. Gramm. d. A/tsiidarab., I, p. 16, n. 2: "Bittner macht mich auf die Nominalform qetel neben qetehe/ aufmerksam: rekeb 'Reiter', sereq 'Dieb' neben qedeher 'Koch' und vergleich zu jenen hebr. gannaq. Da neben diesem auch ein gibbOr besteht, hatten wir meines Erachtens auch im Soqotri eine verkiirzte qatta/- neben der regelrechten qatta/-Form ... ". In any case this parallel clearly shows that Bauer and Leander's theory of a Canan.-Aram. 'language blending' ('Sprachmischung'), which from the outset is improbable, is unnecessary for the explanation of these instances. Landsberger, OLZ 1926, column 975, n. 3, suggests to trace instances as gannaq to a basis qattal, from the outset separate from qatta/, which both, according to him, appear in Akkadian, and more specifically: qatta/ "als Adjektiv (Gewohnheit)", and qattii/ "als Substantiv (Beruf)"; and he assumes that in Hebrew the first type had prevailed, and the second in the remaining Semitic languages. However, one has to proceed in the instances of the type gannaq from the qattal type (clearly present in the other languages), of which the former (qatta/) can only be a secondary development. In my opinion, the fact that the long a in qatta/ in Hebrew is usually not shifted to b, as might be expected with a strongly accented a, is again to be explained by the difficulty of the distribution of the stress at the encounter of two stressed syllables (in connection with this cf. F. Pratorius, Ober den riickweichenden Akzent im Hebr., p. 2 ff.). In qatta/ both syllables are of equally strong sonority. Both have therefore to receive equally strong stress. This is naturally very difficult. Thus, under certain circumstances, the one must be weakened in favor of the other. If the first syllable is immediately pronounced with strong stress, then the second receives lesser stress. However, this need not signify that it is also quantitatively shortened, although this naturally can occur simultaneously. But in any case, the a-vowel cannot be shifted to b,

66

THE VOWEL I AS AN AUXILIARY VOWEL

e.g., 'aqtiil> 'a-qtiil> *'aqtiil > 'iqtiil, etc.; but in this instance undoubtedly through an abrupt retreat of the stress. In connection with this it is well to keep in mind what Rhodokanakis, po/iir, II, § 12 e, states: "Auffallend ist, dass sogar nach einem-offenbar schwacheren-Akzente, obwohl dieser unter anderen Umstanden (§ 1 s) sekundare Verdoppelung zur Folge haben kann, oft primare Gemination nicht gehOrt wird, ohne dass man aber nachfolgende Vokalverfluchtigung (§ c) dafiir verantwortlich machen konnte. So saqatet = saqqatat ... ". That Rhodokanakis was unable to notice a gemination in the instance observed by him, would be easily comprehensible if one assumes that the preceding vowel was uttered with 'weakly cut' accent ('schwach geschnittener Silbenakzent'), i.e., with decreasing stress; for the first component of a geminate with which a syllable begins medially in a word, becomes easily reduced (see since this transition applies only to strongly accentuated ii. In the instance, however, where the accent loss did not affect the second but rather the first syllable, the shift of ii to 0 did necessarily take place. The result of this development appears in gibbor < gibbiir (the latter form in Bibl.-Aram.). A question in itself naturally is whether the transition of a to i in gibbor (and thus also in Bibl.-Aram. gibbiir, Moroc.-Arab. gi!!ii' [see above p. 61] and similar forms) stands in relationship with the preponderance of the stress on the second syllable which caused the shift of ii to 0, hence stands in relationship with the weak accentuation of the first syllable, perhaps is of necessity even conditioned by it. As preliminary stage of the interjection of i in such instances we have determined lack of vowel or reduction of the vowel present in the underlying basic form as a consequence of its pronunciation with 'weakly cut' accent (,schwach geschnittener Silbenakzent'); cf. above p.63. Such an i decomposing a consonant cluster can develop in two ways: it may originate in consequence of the retreat onto this consonant cluster of the stress resting on the neighbouring syllable -leading to the weakening of the latter-it can also simply form itself without the participation of the accent through the mere dissolution of this unpronouncable consonant cluster. In the case of gibbor, etc., where precisely the shift of ii to 0 clearly shows that the second syllable was not affected by any lessening of stress, the latter process must have taken place. Hence *gbbiir became gibbiir through simple breakingup of the consonant cluster without the latter simultaneously receiving the stress.Also for the remaining instances where the shift of ii to 0 does not take place, as e.g., in the instances of the 3rd pers. of the perf. of 'concave' roots, as qiim (root lJ1Jm J, one need not think of a 'language blend' ('Sprachmischung'). For an understanding of the latter instances Landsberger's remarks (OLZ 1926, column 975) are worthy of notice: "qiim steht zu miiqom (aus maqiimJ in gleichem Verhiiltnis wie qii!a/ zu diil]iir ... ". One can explain the fact that qiim has not been shifted to *qom (cf. maqom < *maqiim) and also that qii!a/ has not been developed to *qii!ii/ (i.e., in the 'context'), by the fact that the verb (as a category) bears as a weaker stress in the sentence than the noun, an experience which is obviously valid in the area of IndoEuropean languages and which was also made in the Semitic area (cf., e.g., Brockelmann, Grundriss, I, p. 407; Bergstriisser, Phonogramme im neuaram. Dia/ekt von Ma'lU/a [Miinchen 1933, SBBA]) and that, therefore, the lengthening or qualitative vowel change, respectively, did not take piace in these instances.

THE VOWEL I AS AN AUXILIARY VOWEL

67

Feghali, Le parler arabe de Iq'ar'abida, p. 106) and then may disappear completely (see above p. 65, n. 67). It seems to me necessary to examine in detail the contemporary Arabic dialects with respect to these and related phonetic characteristics. I have postulated above (p. 63) for the Semitic languages (including the modern-Arabic dialects) a sequence of two consonants in the middle of a word which is uttered as if it stood initially, which therefore is not firmly joined to the vowel preceding it. Such a manner of articulation of a sequence of two consonants essentially depends on the accentuation of the preceding vowel with 'weakly cut' accent (,schwach geschnittener Silbenakzent'), an accentuation for which long vowels are best suited (see Sievers, I.e.). And indeed this type of articulation of a sequence of two consonants in the middle of a word has been established precisely for such phonetical combinations in the Arabic linguistic area by Feghali, Le parler arabe de Kfar'abida, p.107. Feghali polemisizes there against Mattson's assertion (l.e., p. 113) that (contrary to Brockelmann) a long vowel could preserve its quantity before a consonant cluster or a geminate: "Les nombreux exemples qu'il donne Ii I'appui de son opinion ne semblent pas la confirmer parce que dans aucun d'eux la voyelle longue ne se trouve reellement en syllabe fermee par deux consonnes ou par une con sonne geminee. L'erreur 69 de M. Mattsson provient de ce qu'i1 n'a pas prete attention Ii la coupe dialectale des syllables. Ainsi (pp. 112-113) Ie mot mlidne 'minaret' < cl. mtrjanatun doit etre coupe mli-dne; de meme biilfin 'posant, mettant' (masc. plur.) < c1. /:liittina ... et zarna 'notre voisin' < c1. zaru-nii ... doivent etre coupes /:Iii_Itin et za-rna; etc. Quant Ii sabt(!ssamak 'Ie marche aux poissons', il doit etre coupe sa-bt(!s-samak. Les deux consonnes b et t sont naturellement groupees dans la syllable fermee par -es-". In any case, Feghali's remarks show that within a word (i.e., medially) a sequence of two consonants after a vowel may set in in the same manner as a sequence of two consonants occurring initially. If this is per se possible, then it cannot fundamentally matter, whether the vowel preceding the consonant cluster is long or short. Such conditions appear to exist not only in the syllable formation of the Lebanese dialects, but also in other Arabic dialects. Landberg, 69 It is out of the question to speak of an actual error by Mattsson, since the vowels concerned are in fact found medially before a consonant cluster.

68

THE VOWEL I AS AN AUXILIARY VOWEL

Etudes, I, Ifmjramout, p. 376, n. 2, remarks in connection with the word gii/sin « giilisin), appearing Ibid. in the text: "Prie de prononcer ce mot lentement, il dit ga' -e/-sin .... Il avait donc Ie sentiment que I devait avoir une voyelle; ilia place ici avant,. et la consequence naturelle est qu'un hamza separe les deux voyelles. C'est ainsi qu'il pronon i, u in forms of the verb may have again quickly been excluded by force of 'Systemzwang' (see p. 74).

THE VOWEL I AS AN AUXILIARY VOWEL

73

follow prefixes. Cf., e.g., the perfect of the n-reflexive *naqtala (> Hebr. niq!al), as contrasted to the imperf. janqatilu. Brockelmann, Grundriss, I, § 42 d, has explained this vowel elision as a consequence of a 'Neuheitsdruck' resting on the modifying prefix, e.g. na- (see also Brockelmann, Ibid., b). 75 A psychologically conditioned stress of this type seems to me here to be improbable. These prefixes are proclitics,76 which therefore - as is usually the case with proclitics - in their merger with the verbal stem must have been precisely weakly stressed. But it is just here where the impetus lay for a further development which resulted in the elision of the vowel of the first stem-syllable. We begin with *naqatala. After the prefix had been fused with the stem to an inseparable unit so that it had completely lost its independent function, this new word unit became subject to the process characteristic of syllable complexes of such nature: the stress abruptly struck the unaccentuated initial syllable of relatively weak sonority, whereby the vowel of the following syllable was elided. Thus *naqatala became *naqtala. In the form *janaqatilu it was just this syllable na which was the first syllable following the proclitic ja; thus there resulted janqatilu; etc. In his study Die Femininendung t im Semitischen (see reference in note 75), p. 7, Brockelmann mentions a specific case of Arabic as involving vowel elision as a consequence of the 'Neuheitsdruck' with which, in his opinion, a modifying prefix (a proclitic) is pronounced (see above). This case is represented by instances like ljahlja < ljahUlja, faljakun > *falijakun. This case precisely makes it clear that in all instances which Brockelmann interprets as affected by an alleged 'Neuheitsdruck', we are dealing with the secondary stress on an originally unstressed syllable. The particles fa, lja, as all monosyllabic proclitics (prepositions and other particles), are basically unstressed (see Brockelmann, Grundriss, I, §42 b). If now, in the 75 Originally-in his paper Die Femininendung t im Semitischen (in lahresbericht der Schlesischen Gesellschaft for Vaterliindische Kultur, Breslau 1903)- Brockelmann had set up for the dropping of the vowel a law much more extensive, embracing numerous other instances. See Barth's objections in ZDMG 57, 628 ff., 801 ff., and also Brockelmann's remarks ibid., 795 ff. 76 These proclitics, at least in part, were originally, to be sure, independent words. Cf., e.g., the noun formation with ma-prefix. This ma- is, as is generally assumed, the original pronoun ma and the stem following it an original verbal form; i.e., we are dealing here with relative clauses (and also substantive clauses) transformed into nouns: maqtal < *ma-qatal (cf. below p. 232ff.).

74

THE VOWEL I AS AN AUXILIARY VOWEL

above-quoted combinations, we find fa and ya stressed, this must be ascribed to a (secondary) abrupt shift (retreat) of the stress onto these monosyllabic proclitics; and the elision of the vowel of the syllable following these proclitics must necessarily be connected with this abrupt shift (retreat) of the accent. The assumption of an elision of u before y, and of i before j, as a process of dissimilation (thus Barth, ZDMG 57, 633; Brockelmann, Grundriss, I, §95 b), cannot only not be corroborated by any analogous development in other forms, but this explanation-and this seems of decisive importancedoes not take into account forms like faltakun, yaltakun < falitakun, yalitakun, etc. Consequently, this completely clear Arabic case throws light on the afore-mentioned (p. 73) wide-spread Proto-Semitic (and Common-Semitic) instances. I have presented above (pp. 63-68) numerous examples of forms of the intensive stems of the verb from modern-Arabic dialects, in which the original a following the first radical is replaced by i. Among these are forms of the perfect, as well as of the imperfect, imperative and of the participles. The fact that in Hebrew this i instead of a before a geminate in the verb is found only in the perfect, however, not in the imperfect or in the imperative, etc.,77 is to be explained by the 'Systemzwang', which governs the morphology, especially of literary languages. We find in modern-Arabic dialects in one and the same form, e.g., in the perfect of the intensive stem of the dialect of Mor\.-.co, instances with i beside such with a. Here the process is therefore still in a state of flux. The difference of the new i from the original a is not yet clearly conceived in the linguistic consciousness. Gradually, however, such a difference is taken into account in the consciousness of the speaker, especially then if the shifting process has to some extent been completed. As soon as the difference is clearly 77 Brockelmann, Grundriss, I, p. 136, attempts to explain this through the 'tendency for dissimilation' (,Dissimilationstrieb') which in qarrel, j'qa!!el, etc., on account of the e in the second syllable, did not permit the i in the first syllable to arise, while the i in the perfect, as he assumes, had already appeared at the time when the form sounded exclusively qa!!al with a in the second syllable (cf. qiddas). The i of the second radical of the imperfect, however, had penetrated, according to Brockelmann, only then into the stem of the perfect (qi!!el) when the "tendency for dissimilation" had been no longer an effective force. For the merit of the dissimilation principle cf. the comments made here on p. 60. According to this, the principle would also have to be rejected if we had only forms of the type qiddas and no qi!!el-formations. The form qi!!el must, in my opinion, make clear that such a principle can by no means come into consideration here. This is completely proven by the modern Arabic dialects in which i (after the first radical of the root) is also found in the imperfect.

THE VOWEL I AS AN AUXILIARY VOWEL

75

conceived, without any kind of clear recognition still being present as to which vowel is semantically more justified, the gravitation towards uniformity, the so-called 'Systemzwang', begins to create homogeneous and systematic conditions. It admits in one and the same genre of forms only one, specific vowel, whether it be i or a. And, indeed, this usually will result in each of the two vowels being felt characteristic of a specific one of the genres of forms in which, as a consequence of phonetic changes, they coexisted at the beginning; and thus, if in the one category (genre of forms) a is carried through, i will generally prevail in the other. Thus we may probably explain the pervading i after the first radical in the perfect of the Hebrew intensive stem in its relationship to the constant a of the imperfect-imperative. Hebrew is a literary language whose phonetic development, as it is transmitted, in contrast to that of modern-Arabic (colloquial) dialects (and probably also in contrast to the original Hebrew vernacular dialects coexisting with the Hebrew literary language) already had come to a certain conclusion and become stabilized. One may probably presume that the 'Systemzwang' (the propensity for leveling) which diffuses the results of new phonetic developments in the whole area of the respective genres of forms or abrogates them again in their entirety, influences the language with special force exactly in those phases of linguistic history in which the language in its development has come to a certain conclusion and begins to become a literary language, which is scarcely any more affected by change through linguistic (more exactly: phonetic) processes, whereas the phonetic processes carrying through without regard to the form-system constitute the main factor precisely in more primitive but also livelier stages of linguistic development. The pervading replacement of the original a of the first radical in the perfect of the Hebrew intensive stem by i, which has the appearance of a phonetic law effective without exception, would according to this be the result of the 'Systemzwang'. Many so-called phonetic laws, which have not come into being through certain physical causes (e.g., constitutional changes of the speech organs) or other unconditionally and un exceptionally effective factors, owe their 'invariability' probably first to the effect of the 'Systemzwang' (the propensity for leveling). In this sense 'phonetic law' and 'analogy' would therefore not represent contrasting motives, but rather the regularity of the sound change itself would here be only a consequence of the analogy.

76

THE VOWEL I AS AN A UXILIARY VOWEL

The interjected (voiceless) auxiliary vowel is in the traditional writing-systems of the old-Semitic languages (as well as by the speakers, and frequently by observers, of modern-Arabic dialects) identified as i (although it may also be characterized by the phonetic symbol a; cf. above p.40). In several passages of the foregoing discussion I have already pointed out that this (originally whispered) auxiliary vowel, i (a), may also be determined in its coloration through the surrounding consonants. Along with this I have emphasized that in most of these cases in which one speaks of 'Umfarbung', 'colouring' of a vowel through neighbouring consonants (assimilation)-thus, e.g., a> u in the vicinity of a labial consonant-one should not imagine this transition as being directly from the earlier vowel nuance to the present, therefore, e.g., from a directly to u. Rather one must first of all presume the disappearance of the original vowel, which may have been caused through various factors, whether it may be through elision of the vowel as a consequence of its position between two accents or through an accentual shift, or, in closed sylla ble, as a consequence of its reduction through a 'weakly cut syllable accent' (,schwach geschnittener Silbenakzent') or through absorption by surrounding sounds of considerable sonority - in other words, on e has to assume the emergence of a consonant cluster as an intermediate stage. The new vowel, e.g., u (in the vicinity of labials or velars), which replaces another, original vowel, e.g., a, is however to be considered as an auxiliary vowel interjected for the breaking-up of this consonant cluster (or also as a bearer of a new accent resulting from a shift of the accent)-therefore with respect to origin and function to be considered as identical with the usual and a priori to be expected the vowel. i. Very often we find i and u coexisting (see above p. 24), e.g., in Central Arabia, bSilbe beside bSibe < basabat. As has already been shown, such newly developed vowels are a priori whispered (voiceless) and without an entirely definite nuance (though they posses an affinity to i) and are therefore apt of being determined in their colouring by neighbouring consonants (or sounds in general). It shall not be denied that consonants are able to affect the colouring of vowels already present (not newly developed). It is well enough known that, e.g., the emphatic sounds of Semitic (especially in Arabic) are capable of 'velarizing' contiguous vowels, that they may darken them, just as the laryngals may influence the vowels in their specific manner. One may perhaps even maintain that each vowel according to the consonant with which it stands in close

THE VOWEL I AS AN AUXILIARY VOWEL

77

contact exhibits another nuance, just as also, inversely, the consonants are influenced in their articulation by the vowels surrounding them. However, we are dealing in these latter cases with relatively minor changes. An a which follows a ! is articulated approximately as a; etc. The unconstrained linguistic sense for this reason does not yet perceive it as a special vowel in itself, as a phoneme diverging from a. It is, however, different in the cases we are dealing with. Here it is a question of more far-reaching shifts which make themselves clearly noticeable in the linguistic consciousness. I consider it, from a phonetico-physiological point of view, impossible that these strong changes could be caused by the mere vicinity of certain consonants, that, therefore, the u in bSube might have developed directly from the a in a form *bSabe through the influence of the neighbouring b (that is: a 'coloured' into u through influence of the b). That such a process cannot come into question, was clearly proven above (p. 29 ff.) in the case of the paradigm of the perfect in 'Iraqi-Arabic (as well as in related dialects), where in word-forms of now completely equal syllable structure and accentuation influence of contiguous consonants is only to be determined in the cases whose present syllable structure is based on a different one which was changed through accentual shifts, e.g., 3rd pers. sing. fern. serubet « saribdt); however, 2nd pers. sing. masc. sarab( e) t, fern. sarabti. But this is valid also for a rather large part of instances dealt with by Brockelmann, Grundriss, I, § 73 ff. (p. 194 ff.) in the chapter on "Assimilation von Vokalen und Konsonanten", to a lesser degree for the influences (discussed Ibid., § 74) of the laryngeals and emphatic sounds, however, all the more for the influences (mentioned Ibid.) of the velar sounds, and especially for the effects of the labials (treated § 75). The u in an instance as (see Ibid., 00) Egypt.-Arab. rnufta/:t 'key' < rnifta/:t might therefore have to be viewed just as the i of this rnifta/:t (originally from rnafta/:t; the rni-prefix is probably in all cases to be traced to the rna-prefix), probably therefore as a secondary vowel instead of an original i (standing in closed syllable) which, by the influence of a 'weakly cut syllable accent' ('schwach geschnittener Silbenakzent'), became reduced and weakened: therefore probably rnifta/:t > *rni-fta/:t, (*rni-fta/:t) > rnufta/:t, as in Old Arabic rnafta/:t> *rna-fta/:t > rnifta/:t (cf. above); etc. Thus, e.g., also Arab. barakat 'blessing' (see Brockelmann, Ibid., f; p. 201) > Aram. barakat, *brakat (in the modern Arabic dialect of the 'Anazeh we actually find braka, see above p. 19, n. 19) > Judeo-Aram. *birkal or birkegl,

78

THE VOWEL I AS AN AUXILIARY VOWEL

respectively, and simultaneously with this also Syriac *burkal or burk ela, with u instead of i through the influence of the labial. The Syriac form may possibly also be explained as a secondary development of the Judeo-Aramaic form. In the form birk ela the i between band r was namely apt to be absorbed by the sonorous r: > *brk ela, and in the latter form a vowel u determined by the labial could be interjected for the purpose of breaking-up the consonant cluster: > burk ela. Likewise, one cannot actually speak of an assimilation of ('sonantic') vowels to other ('sonantic') vowels in neighbouring syllables. Several of the examples cited by Brockelmann, Grundriss, I, § 68, were mentioned in the present study and explained differently. A 'vowel harmony' ('vowel agreement') , as the Uralo-Altaic languages indeed possess, is unknown to Semitic. 7 8 The same is naturally also valid for most of the instances which Brockelmann, Grundriss, I, § 94, has collected in the chapter on "Dissimilation von sonantischen Vokalen" (see especially § 94 a and f) and which I have in part touched upon above p. 60f. and p. 74, n.77. I do not, of course, want to deny that dissimilations of this type exist. I wish to call attention, e.g., to the instances cited by Stumme, Tunis, Miirchen und Gedichte, p. XXVII: zabethOlu < gabathu lahu, l;allhOlu < I;allahu lahu, zibtuhOlu < gibtuhu lahu, etc., "wo statt des dumpfen und unschonen it-it das besser klingende a-it eingetreten ist". And in analogy to these Tunisian-Arabic instances we interpret the a-vowel in Archaic-Hebrew lama as a dissimilation of it in the basic form (preserved in Class. Arabic) lahumit: lahumu> *lahuma > lama. (Analogously, certain other, similar Hebrew forms are to be interpreted). An aesthetic motive of this type, as it is easily comprehensible with such a sound-sequence, cannot, however, come into question with respect to the substitution of a by i in the vicinity of a (and similar substitutions). Furthermore, in the above-mentioned dissimilation process of Tunisian-Arabic (and similar instances in related languages) it is a question of a minimal shift of the original vowel nuance, while there is between a and i such a gap that it cannot be explained by a process of dissimilation. 78 Instances of specific nature which do not fall within the scope of this study, are the instances in which the vowel assimilation is mediated through laryngeals (see Brockelmann, Ibid., b ~). In any case, also here one does not deal with a phenomenon which can be designated as "vowel harmony".

THE VOWEL I AS AN AUXILIARY VOWEL

79

In connection with this investigation concerning i as an auxiliary vowel, the question should be examined as to how far under certain conditions, as for instance if word-forms are augmented by endings, the forms resulting from the accentual shifts described above become the regular forms of the respective words (that is: the forms of these words even if the specific conditions no longer exist, e.g., without suffixes) so that the original forms disappear. Reinhardt, Ein arabischer Dialekt, p. 25, gives the following rule: "Die einsilbigen Warter mit kurzen Vokalen wie: bit 'Schwester', 1um 'Mund', gin 'Ohr', bin 'Sohn', lid 'Hand', verlieren bei Antritt von i und u ihren Vokal: ldi-ida, btf 'meine Schwester', 1mi 'mein Mund', gni 'mein Ohr', bna 'sein Sohn', sma 'sein Name', aber sumhe 'ihr Name', binhum 'ihr Sohn'''. 1um 'mouth' and lid 'hand', forms which are used before suffixes beginning with consonants and without suffixes (thus used as stat. absol.) go back to classical Jam and jad. How could the a here develop into u or i? One usually ascribes this to the influence of the m or of the l, respectively.79 Such a type of influence of consonants upon contiguous vowels is, however, as shown (see above p. 76), only possible with a newly developed, (in the beginning) whispered (voiceless) vowel, which appears in place of a dropped or reduced vowel. That j, e.g., might change a following a simply into i would, in my opinion, be simply an impossible phonetic process. (One could possibly understand an 'Umlaut' from a to ii through the influence of a neighbouring D. The dropping or the reduction of a vowel is, however, possible only under certain conditions, e.g., in consequence of an accentual shift, which in our case actually seems to come into question. This view is corroborated by the other monosyllabic words mentioned by Reinhardt, which exhibit the same two forms of syllable structure, that of the stat. absol. and the structure before suffixes beginning with consonants on the one hand, and that before suffixes beginning with vowels on the other hand, namely gin, bin bit (like 1um, jid) and (with suffixes beginning with vowels) gn-, bn-, bt- (like 1m-, jd- [id-]). For in the case of gin, bin, bit, sum « 'ugn, ibn, 'ubt, ism) there is indeed no doubt that the initial vowel became lost through an accentual process. 79 Thus Brockelmann, Grundriss, I, §69 d~ (p.187). In jadi 'my hand'-I am not able to find the form in Rossler, MSOS III 21, 4 lines from below, as Brockelmann notes it - the assimilation of the a to j is supposed to have been impeded through the dissimilatory effect of the i. But also such a dissimilatory effect of i is in itself quite doubtful (cf. above p. 74, n. 77).

80

THE VOWEL I AS AN AUXILIARY VOWEL

Hence, I also trace back fum or Jum < Jam, lid < lad to an accentual process. The i in gin < 'ugn, bin < 'ibn,80 bit < ubt, or the u in sum < 'ism, represents a newly developed vowel, parallel to the newly developed vowel in 'Omani-words such as tbin < tibn 'straw', btin < baln 'stomach', :;dar < :;adr 'chest' (see Reinhardt, l.c., p.41), words based on Classical primary forms with the same syllable structure as the latter forms (i.e., gin, bin, bit), hence qatl, qitl, qutl. (,)gin, (' )bin, (' )bit, (' ) sum relate to 'ugn, 'ibn, 'ubt, 'ism just as tbin relates to tibn, b{in to baln, :;dar to :;adr. How should one now explain the relationship of Jum to Jam, lid to lad, which must be parallel to that of gin to ' ugn, of bin to ' ibn? The words Jam, lad are indeed originally two-radical words, while 'ugn, 'ubt, 'ism, 'ibn (the latter two only as a consequence of secondary development) are three-radical stems. The vowels now bearing the accent in blin, :;dar, etc.; gin, bit, etc., are originally unstressed auxiliary vowels on which the accent has shifted through a secondary process. From the beginning it is completely unclear, which factor could have thrown the accent away from the stem-vowel upon the secondary interjected vowel. This is no less unclear than the question as to the factor which in Jam and lad had caused the original vowel to drop or become reduced and then again has brought about the emergence of a new vowel. But all of these forms-also fum and lid-appear not only in the stat. absol., but also before suffixes beginning with consonants: 81 b!inhe for *batnhii 'her stomach', :;darha for *:;adrhii 'her chest', bithe for *ubthii 'her sister', fumhe for *Jamhii 'her mouth', lidhe for *ladhii 'her hand', etc.; and if in these latter primary forms (*'ubthii, *Jamhii, etc.) accentual changes have appeared, then this may be very readily understood.

80 Just as the (contemporaneous) colloquial Arabic form bin is to be considered developed from 'ibn via an intermediate stage *bn (see Socin, Diwan, III, p. !O8), I assume the ancient Arabic form 'ibn, Hebrew bin (ben) to have developed via the same intermediate stage -bn- (existing in ancient Arabic beside 'ibn) from an original *ban, which included the vowel a of the plural, Arabic banii/ina, Hebrew biinim. The vowel i in place of the original a is to be considered an auxiliary vowel developed through the break-up of the consonant cluster -bn-. (What I have noted concerning *ban, bin in my article "Effets de I'accentuation emphatique en semitique", in Mem. de fa Soc. de Ling. de Paris, vol. 23, fasc.6, 1935, p. 352f., is untenable, as well as my explanation of the u in 'uht. In the forms ribbOn < *rabbon, rabbiin; ribbi < rabbi, treated Ibid., the i is to be explained in agreement with above p.61). 81 These words representing degrees of kinship and parts of the body will appear, to be sure, rather seldom without possessive suffixes.

THE VOWEL I AS AN AUXILIARY VOWEL

81

In the 'Iraqi-dialects there are parallel forms to the forms which appear in the 'Omani-dialect (and in other dialects) in the stat. absol. and before suffixes beginning with consonants; indeed these parallel forms appear here almost exclusively before such suffixes and not in the stat. absol. Cf. Meissner, Neuarab. Geschichten, Introd., p. XXVII, 41 b: "Bei den Nominalformen mit ruhendem 2. Radikal rallt bei Antritt der vokalisch anlautenden Suffixe der Hilfsvokal der zweiten Silbe gem aus: darbi 'mein Weg'. Bei konsonantisch anlautenden Suffixen geht das natiirlich nicht, zuweilen erhalt der Hilfsvokal sogar den Akzent, oder der Akzent schwebt wenigstens: dar ( u )bnii 'unser Weg'; nef(e)shii 'ihre Seele'''. (Cf. also above p. 29). If this auxiliary vowel draws the accent to itself, then this conforms - since it stands before two consonants, therefore in a syllable of considerable sonority - on the one hand with the normal rules of accentuation, on the other hand with the rule that enclitics draw the accent to the pen ultima. In 'Omani *batnhii, *'ubthii, *!fadrhii there existed (in a parallel manner) medially consonant groups consisting of three components, i.e., consisting of the last two radicals of the stem and of the beginning consonant of the suffix. Between the first and second components an auxiliary vowel was interjected, > *batinhii, *$adirhii. This auxiliary vowel then drew the accent to itself, whereby, however (in contrast to the 'Iraqi instances), the original vowel of the stem from which the accent was removed, was elided: btinhe, $darha (0 for i through the influence of the r), bithe, binhe, ginhe, sumhe (u through the influence of the m). The forms resulting hereby before suffixes beginning with consonants were now also used without suffixes, thus becoming the usual forms of the respective words: b!in, $dar, bit, bin, etc. Only thus, in my opinion, may the remarkable fact be explained, that in 'Omani in the noun-forms involved the vowel between the 2nd and the 3rd radical which was inserted for the breaking-up of a consonant cluster, drew the accent to itself. How could one otherwise explain the phenomenon appearing in many Arabic dialects and also in Tigre (see E. Littmann, ZA 13, 144), that just these secondary vowels, which certainly at first served only the purpose of making the pronunciation easier, carry the main stress? 8 2 82 This is true also for dialects in which also original qatal-forms exhibit 'transposed' accent ('umgesprungener Akzent'), as, e.g., in the dialect of Tunis, see Stumme, Tunis. Miirchen und Gedichte, p. XXVII. Stumme (Ibid.) assumes as certain, "dass diese Art der Betonung vom Verb ausging", and compares ktib = kataba, etc. It is true,

82

THE VOWEL I AS AN AUXILIARY VOWEL

If to the last-mentioned word-forms suffixes beginning with vowels were joined, then in turn another syllable structure naturally had to appear, which relates to the forms with suffixes beginning with consonants as, e.g., among the imperative forms of Hebrew, the forms qi!li (sing. fern.), qi!lu (pI. masc.) relate to the forms qe!ol (sing. masc.), qe!olnii (pI. fern.) (corresponding forms exist even in 'Omani itself, e.g., $idro 'his chest', beside $d6r 'chest', $d6rha 'her chest', etc.; cf. above p. 6). The original syllable structure of these nouns indeed agrees with the syllable structure necessary for nounforms with suffixes beginning with vowels: $adr, ba!n, 'ubt, 'ibn, etc. However, the forms of the modern dialect are here not based on these primary forms, but rather on the new forms which developed within the dialect (in connection with suffixes beginning with consonants). These noun forms (developed in connection with such suffixes), to be sure, then completely forced out the old forms and were even used without suffixes. Thus. there developed, e.g., from *$d6ro 'his chest': $idro, with i as an auxiliary vowel; etc. (cf. Reinhardt, l.c., p. 25, 13 a, VII, where it should state: "zwischen 1. und 2. Radikal" instead of "2. und 3."). Analogously in the case of words originally beginning with a vowel, the forms developed before suffixes beginning with a consonant, which had become the normal forms of the words concerned, were used also before suffixes beginning with a vowel. However, it is difficult to maintain unweakened the structure of monosyllabic words of this type, if they are being connected with suffixes beginning with vowels, whereby the stem-syllable becomes an open syllable, hence *bito, *sumo, *bino. For this reason the structure of such words is already strengthened in Classical Arabic in certain cases through gemination of the 2nd radical (cf. N6ldeke, Zur Gramm. d. klass. he admits that the origin of the accentuation of this latter form (ktib) had not yet been clarified: "Woher diese Betonung der 3. s. pf. stammt, -d.h. ob hier etwas Uraltes, speziell Siidarabisches, zu Grunde liegt, -oder ob die Tunisier (und Maghrebiner iiberhaupt) erst spiiter in ihrer afrikanischen Heimat, vielleicht aus Analogie ... zu fdalt, fdaltil, fdalnii auch fa'al (jal) gebildet haben,-vermag ich nicht zu entscheiden". As I have attempted to show above n. 32, the accentuation ktib is certainly secondary and even in its turn needs clarification. The analogy of forms of other grammatical 'persons' can, however, not come into question here. It is true that we find in other dialects (see above p. 29) a reverse shift of the stress onto the initial syllable in this form of the perfect. It seems to me much more probable that also the form ktib has developed with the affixation of 'suffixes' beginning with consonants; the same form is, to be sure, then again transformed when 'suffixes' beginning with (or consisting in) vowels are affixed to it.

THE VOWEL I AS AN AUXILIARY VOWEL

83

Arab., p. 14, 15, and p. 15, n. 1 and 2; cf. further Classical Arabic minni, 'anni). The same is also observed in modern dialects (see Landberg, Etudes, II, Dalina, p. 131, n.6), and one may assert that the gemination arose mainly before vowel (in Classical Arabic, of course, a vowel always followed the consonants concerned) and then was generalized. Cf. Socin, Diwan, III, § 85 n (p. 109) with respect to jad: "... einmal sogar als jadd vor einem anlautenden Vokal" (cf. also N61deke, Neue Beitriige zur semit. Sprachwiss., p. 115, 175; Brockelmann, Grundriss, I, p. 333).83 The language, however, does not constantly make use of this means. But then it is difficult to preserve the structure of the word unweakened, and the vowel becomes lost. Thus developed in 'Omant bti 'my sister' < *biti, sma 'his name' < *sumo, etc., gno 'his ear' < *gino, but also lmi 'my mouth' < Jami (with J> 1 through dissimilation of J and m), and thus before all suffixes beginning with a vowel. Cf. also N6ldeke, I.e., p. 178; furthermore Socin, I.e., § 85 m (p. 109): "Die gew6hnliche Form fUr das Wort, welches 'Mund' bezeichnet, ist in Prosa ilm; so wenigstens vor einem anlautenden Vokal ... ". The thus developed initial consonantal cluster can naturally be dissolved by a prosthetic vowel. However, how shall we explain the development of the form lum, with u for a, of 'Omant, a form 83 Since the gemination is, however, secondary, and fum, lum 'mouth', without gemination of the m, is also found in modern dialects (see Landberg, Gloss. dalinois, I, p. 251; N6ldeke, I.c., p. 176), Brockelmann's assumption, Grundriss, I, §41 aa; p.67 (occasioned by Landberg's comment, Dalina, p.62, n. I) that the consonantdoubling in lumah 'seine Miindung' had been lost, is scarcely justified. Also Brockelmann's other examples for the loss of gemination are uncertain. Kabat 'sie stiirzte urn' is a formation of a root kblj (see Landberg, Ibid., 1103). In jikimnina 'er bedeckt uns' for jikenninna, the loss of the gemination is probably sufficiently explained through the shifting of the accent onto the preceding syllable which is no less sonorous, and certainly also through the tendency towards lessening the numerous n-sounds. In the case of mibnie 'gebaute' (partic. fern.) and Zejdie (see Brockelmann, I.c.) one should not proceed altogether from a consonant gemination (as also Landberg does [see I.c., p. 1394]), but rather from basic forms as mabnijat, Zajdijat. In such forms the accent in these dialects usually retreats, and consequently the long vowel is shortened; see Landberg, Ibid., and Arabica, V, 170. To be noticed is the i for a in mibnie: mabn4at > *m abn4e > mibnie. In the dialect of Beirut as well as in other dialects, especially in the passive participle of verbs III j, the prefix is constantly mi-, as mibni < mabnij; see Mattsson, it. phonol. sur Ie dial. de Beyrouth, p. 88. This i < a must stand in close connection with the retreat of the accent to the vowel, which has loose contact with the consonant following it. Cf. instances as 'i/jat < 'aljat, p. 23f., further cf. p. 59). One will therefore not be able to produce examples from the dialect of Da!ina for the loss of gemination of consonants in the interior of words in the linguistic area of Arabic (see Landberg, Dalina, p. 1395). However, this appears to have been observed with certainty in the dialect of Qotar, see above p. 66.

84

THE VOWEL I AS AN AUXILIARY VOWEL

essentially appearing before suffixes beginning with consonants ([um also in other dialects, cf. NOldeke, Ibid.)?84 If words of such types as we have discussed above, are changed in any manner with the affixation of 'suffixes', then the new forms resulting herefrom can become the regular forms of the words concerned. The speaker is by no means conscious of the secondary character of the forms resulting through the affixation of the respective 'suffixes'. For this reason the original form can be forgotten (even though it need not be forgotten). The new form can connect itself with differently constructed suffixes, which on their part may again change the syllable structure of the word in another manner. This may be applied, for instance, to the forms bt-, gn-, bn-, sm-, 8 5 which developed from bit, gin, bin, sum (developed in their turn from the basic forms 'ubt, 'ugn, 'ibn, 'ism through affixation of 'suffixes' with an initial consonant). These forms, bt-, gn-, bn-, sm-, which developed from bit, gin, bin, sum through affixation of 'suffixes' beginning with vowels, may have in turn been used before 'suffixes' beginning with consonants. Forms other than those previously developed through the affixation of 'suffixes' with initial consonants to the primary forms of the words could, of course, not here develop: *bthe naturally had to become bithe again just as the original *'ubthii became bithe. Such a process could not have apparent consequences. If now, however, Classical 84 Landberg, loc. cit., p. 251, considers the usual u of modern Arabic in this word as more primitive than the a of the normal form of ancient Arabic. However, it contradicts a basic principle of the historical investigation of language, to assign greater originality to later arising forms than to those recorded earlier, unless some special reasons might have to be considered. But this is here by no means the case. How could the a of the ancient Arabic form have developed from an original u? On the contrary, it suggests itself from the start to ascribe u < a to the influence of the labials (naturally under the conditions described here). As a matter of fact, the historical relationship between ancient Arabic Jam and modern Arabic Jum suggests establishing the same relationship for Aramaic, hence to trace (Old-)Aramaic pummii to a primary form *pam(mii). To be sure, the vowel after the first radical of the word is to be considered as a 'fossilized' inflection-vowel (indeed, we find in the various forms of the word in Semitic various declension-vowels thus 'fossilized', cf. below p. 92f.). However, one must trace ArabicJam(un), Aramaic pummii, the only instances where the enlargement of this root with m appears, to a single (common) basic form, and this latter could only have been a form with a as stem vowel. 85 One must also reckon with the possibility that these latter forms were able to arise first, hence earlier than oit, gin, etc., just as indeed jd-, fm- are earlier than jid, fum. For in *'uoto, *'ismo, etc., the initial vowel could probably immediately drop, as a consequence of a phonetic tendency of the 'Omiini dialect (see Brockelmann, Grundriss, I, § 43 i ): > oto, smo. But also on the basis of these latter forms, oit, sum must have again arisen before suffixes beginning with consonants and in the stat. absol.

THE VOWEL I AS AN AUXILIARY VOWEL

85

forms like jami, jadi (with a vocalic suffix) had become *jmi, *jdi, and the word-forms jm-, jd-, thus developed, were subsequently also used before suffixes beginning with consonants, then, of course, new forms developed. In *jdhe, *lmhe, the consonant cluster was broken-up by an auxiliary vowel which received the stress: jidhe and lumhe. In the latter word this new vowel was determined in its nuance by the following m to u (cf. above sum and Judeo-Aram. sum, Akkadian sumu; beside Hebrew sem < *sim-). Socin, I.e., §85 n (p. 109), commenting on the word for 'mouth', whose form before initial vowel he had observed to be ilm (see above), states: "Wenn das Wort fUr sich oder vor Konsonanten steht, hart man ilum ... ". In this latter form the u must, of course, be a newly developed vowel because the prosthetic vowel presupposes that there had once been no vowel between 1 and m (see Landberg, Glossaire Dalinois, p. 63). This is, however, also valid for forms with u which have no prosthetic vowel. I must still point-out an especially noteworthy category of the phenomenon here under discussion (i.e., 'break-up of consonant clusters'). Brockelmann, Grundriss, I, § 36 1, discusses instances from later Arabic and from the contemporaneous dialects where (according to his interpretation) "j, more seldom lj, develops from initial vowels articulated without glottal stop-probably at first in the interior of the sentence after a terminal vowel of a preceding word, as it frequently develops in the interior of words" ("[wo sich] aus dem leisen Einsatz im Anlaut j, seltener lj, [entwickelt] wahrscheinlich zunachst im Satzinnern nach auslautenden Vokalen, wie auch sonst im Wortinnern"). This type of origin of lj and i is indeed known. It is to be especially noted that lj and j originate as glide sounds between a prefix ending with a vowel and a stem beginning with a vowel (or also in the middle of a stem between vowels), subsequently become 'position' sounds, and after abstraction of the respective stem from the respective prefix can also remain as actual radicals of the words. Now Brockelmann (l.e.) mentions, inter alia, the instance '''oman. iii 'zu mir', aber jilne 'zu uns' (Reinhardt, § 170)" (Brockelmann indicates erroneously: § 178). Reinhardt, however, states the following: "ile (i/o) 'zu, hin' ist lautlich und in der Bedeutung mit I [< Ii] ziemlich identisch. Zu Anfang des Satzes bildet es mit den Suffixen iii, ilek, ilo, aber iilhe, jilne, jilkum usw.... ". It follows from the sentences cited by Reinhardt - in which he in each instance indicates the stressed syllable of the preposition - that in the form beginning with i, which appears when the preposition is used without suffixes and also when

86

THE VOWEL I AS AN AUXILIARY VOWEL

it is used with suffixes beginning with a vowel, the vowel following the I carries always the stress, hence ife (without suffixes), ilek, i16. It is therefore evident that the alternation of the initial sound between i and ji is directly dependent on the syllable structure and the accent. Consequently, the j in this case may not be considered as having developed for the purpose of fulfilling in initial vowels articulated without an initial glottal stop ('leiser Einsatz') the function which is otherwise fulfilled by an initial glottal stop (and, accordingly, may not be considered as an original glide sound after words ending in a vowel, as Brockelmann assumes). An instance of this type appears in Tunisian Arabic. The Classical word 'urj(u)n 'ear' sounds in this dialect, according to Stumme § l37 (p. 109), in the stat. absol.: urjen, and with possessive suffixes: ljurjni, ljurjnek, ljurjnu, urjenha, urjenna, urjeiikum, urjenhum. With suffixes beginning with vowel the word thus begins with ljU, in contrast to the form of the stat. absol. and to the forms with suffixes having initial consonants: these latter forms begin with u. Certain Classical-Arabic forms have very frequently developed, as is well-known, in the Tunisian dialect (see above p. 6ff.) as well as in other dialects into forms with so-called 'transposition' (of vowel and accent; 'umgesprungene Formen'); cf. Stumme, §47. In §48 a Stumme mentions 'umgesprungene Formen' of roots I lj, as u$db 'dirt', u{dr 'string, chord', u(ldf'sling', all of which are Classical fdal-forms; also ulid 'boy' (concerning this see below p. 87). Furthermore, Stumme adds here urjen 'ear', although it goes back to Classical 'urj(u)n, therefore to a stem I hamz. In words of root I lj having 'transposed' accent ('umgesprungener Akzent') we would have to expect the original initial syllable lja changed to lj through ejection of the vowel a. However, we find this lj to have become 'fluid', as u, since consonantal!! (the same is true of j) is only very difficult to maintain without vowel (sonant), and in unaccentuated state (see below) is usually dissolved (or: 'stretched') into u (or i, respectively). But this will usually be (see below) a ('sonantic') vowel of the shortest duration (a kind of 'glide'), which quantitatively may be scarcely distinguishable from consonantal lj (or j respectively). For our purpose we may thus postulate forms like *lj$db, *y!dr, *y(ldf and probably also *yrjen (more detailed explanations in the following). If to forms with what may be called 'transposition' ('umgesprungene Formen'), irrespective of the initial sounds of the underlying roots,

THE VOWEL I AS AN AUXILIARY VOWEL

87

suffixes are joined, then again differences develop in the syllable structure of these forms dependent on the initial sound of the suffixes which are added. When suffixes beginning with consonants are added, forms with 'transposed' accent ('umgesprungener Akzent') appear. If, however, suffixes beginning with vowels are added, then another type of syllable structure appears, which is closer to, or rather identical with, the original syllable structure (cf. the comments already made above p. 81). Thus in Tunisian (see Stumme, Ibid., p. 108), e.g. of qdilm 'heel': qddmi, qddmek, qddmu, beside qdilmha, qdilmna, qdilmkum, qdilmhum. With stems I lJ we have parallel conditions (see Stumme, Ibid., at the bottom). Of the stat. absol.-forms cited above (p. 86, line 24), the forms with suffixes beginning with a vowel are: lJClt/ji, lJa!ri, lJuldi. Thus also here with the affixation of 'suffixes' with initial vowel, again a vowel appears between the first and second radical, which brings the syllable structure of the forms close to the Classical syllable structure. It appears that also the vowel itself is in very many instances close to the vowel of the original Classical form. One may indeed assume that in many instances of forms with suffixes beginning with vowels, the original Classical form itself still continues to exist. In other instances, however, the vowel seems to be new and determined in its coloration by the surrounding consonants (possibly also by the vowel between the second and third radical of these forms with 'transposed' syllable structure ['umgesprungene Silbenstruktur']; cf. above p. 75; 49). This might be valid for instances as qddmi, but also for most of the instances with lJ as first radical, as lJa~bi, lJacjji, lJa!ri. Aside from the adjacent emphatic consonants, the memory of the original a may have determined the vowel as a. Of ulid (probably < lJalad) we have, however, the form with suffix: lJuldi. The u-vowel following here the consonant lJ is certainly a completely new vowel, hence theoretically the i (a) which is a priori to be expected and which naturally appears as u in contact with the preceding lJ. In my opinion this must be assumed in any case for lJwjni, etc. When a suffix beginning with a vowel was joined to the form with 'transposed' syllable structure: *lJg~n, it resulted in an abrupt retreat of the accent to the initial consonant cluster lJg-. As bearer of this accent a new vowel was interjected: u (for i through the influence of the lJ), > lJugni. We now have to assume a very similar process for the instance of the 'Omani: dialect from which we proceeded (see p. 85). As is shown from the mentioned observations concerning the accent conditions

88

THE VOWEL I AS AN AUXILIARY VOWEL

of the 'Oman! forms concerned, the accent of ile, which corresponds to Old Arabic ilii, rests on its second syllable: ile. And indeed this accentual condition is based on an accentual shift (peculiar to 'Omani), that is: on a 'transposition' ('Umspringen') of the accent, since we have to very likely assume the accentuation ilii for the Classical equivalent. This stress on the terminal syllable in this instance, which-as we may assume in general for the 'Oman! dialect-is probably based on an abrupt shift of the accent, results-as is always the case in these instances in this dialect (see above p. 81)-virtually in the loss of the vowel of the first syllable (thus the vowel with which the word begins), hence results in a form *Je, etc. Cf. btfn < batn, thus also gin < 'Uf/(u)n. I stress that this was the virtual development, because this was not the actual development. In this word, as exactly also in Tunisian Uf/en < 'Uf/(u)n, the vowel to be ejected was not, as usual, a 'sonantic' element (constituting a syllable) following an initial consonant, but rather it was both syllable ('sonantic' vowel) and 'syllable basis' (normally represented by a consonant) in one. In the Tunisian dialect-see Stumme, Tunis. Miirchen und Gedichte, p. XVIII, infra (for 'Oman! it does not become unequivocally clear on the basis of Reinhardt's statements)-indeed hamza (the glottal stop) has generally dropped in initial position, so that the initial vowels are pronounced without being preceded by an actual glottal stop-sound. If hamza (the glottal stop) were, however, to precede the initial vowel, then it could scarcely maintain itself with the disappearance of this vowel, and nothing would remain of the initial syllable, since hamza (the glottal stop) is of necessity bound to the existence of a vowel (a 'sonantic' element). Apart from that, one might probably assume that also in dialects in which initial vowels in general are preceded by a glottal stop (hamza), as soon as the accent through abrupt shifting is totally withdrawn from the initial syllable, its glottal stop disappears (and the initial vowel is pronounced with so-called 'leiser Einsatz'). The initial vowel was thus treated in the two instances before us as if it were constituting consonant + vowel ('sonant'). That is: It did not completely drop through the abrupt shifting of the accent. But rather there remained a hyper-short vowel, which corresponds to the initial consonant of words beginning with genuine consonants, and which factually equals, in its articulation and quantity, a consonantal vowel ('semi-vowel') which has become 'fluid' ('sonantic') as a consequence of the loss of the ('sonantic') vowel following it. These

THE VOWEL I AS AN AUXILIARY VOWEL

89

vowels ('sonants'), reduced through abrupt shifting of the accent (and likewise also!:! and j having become 'fluid' as a consequence of the same process), always stand in open syllable: Ue, ugen. If these syllables (through a change of the syllable structure) become closed at the affixation of 'suffixes' beginning with consonants, then a strong accent is thrown on them, i.e., on i or u (respectively). These 'sonants' (vowels) are, however, quantitatively so reduced, that they cannot become the syllabic bearer of this stress. However, since they cannot avoid the effect of this stress, their excessively slack articulationthey are, as it were, vowels with excessive 'weakly cut syllable accent' (,schwach geschnittener Silbenakzent') - is tightened by it. Yet their phonic substance is so slight that their tightening is only possible through even further quantitative reduction, up to a minimum, up to a 'glide' (see Sievers, Phonetik, §411). In doing so, however, the slack, unstressed phoneme, poor in sonority, strengthens itself. The consonantal j and!:! (even if quantitatively depressed to a minimum) possess greater strength than the vocalic ('sonantic') i and u mentioned, which, though quantitatively slightly more extensive, are based on a completely slack articulation. There has thus taken place, as it were, a condensation (or consolidation) of unstressed vowel substance, poor in sound, to a firm core of minimal quantity: *jlne, *!:!gne, etc. With it that well-known syllable structure-i.e., a consonant clusterdeveloped which finally had to be dissolved by the interjection of a new vowel: jUne, !:!ugne, etc. Both processes - that of the 'consonantisation' (consolidation) of i and u to j and !:!, and that of the breaking-up of the consonant cluster by a new vowel-need not probably be considered as having taken place not simultaneously, even if they represent two independent processes distinct from each other. It becomes hereby clear that the initial consonant j or !:! in jUne, !:!ugne, etc. (see above p. 85), is not a phoneme newly added to the form, as Brockelmann (see above p.86) has assumed. Rather it is precisely the vowel i (following D, or u (following!:!), which had been considered by him to be the original i of ile < ' ilti, or u of ' ugen < ' ugn (respectively), that is newly developed. As Mar~ais, Tlemcen, p. 20, n. 1 remarks: the preservation of the u in Tlemsenic uden (= Tunis. ugen) is "anormale, a moins qu'on ne suppose un primitif ljudn (comp. l'egyptien ljidn, Ie tripolitain !:!uden) passe ensuite par sursaut (cf. infra, p. 47) a ljuden, uden". Stumme (Tunis. Gramm., §48 a) actually lists the word (as already

90

THE VOWEL I AS AN AUXILIARY VOWEL

mentioned) together with the stems I y. According to this one might assume that in the Tunisian forms with suffixes beginning with consonants a basic form *ywjn (peculiar to these dialects) would continue to exist. One must then, however, also assume that jilne, etc., has as its basis a form without suffixes approximately like *jile, hence with an initial sound-sequence ji; and one has to assume further that this form *jile would without suffixes and in the forms with suffixes beginning with a vowel have become *jle> ile, in consequence of the shift of the stress onto the terminal syllable. Such a form as *jile is, however, not recorded for the 'Omant dialect (cf. below), nor could it in any way appear probable that such a j in initial position could have arisen as a glide after a word with terminal vowel in the middle of the sentence (as Brockelmann, l.c., p.46, assumes) and then have become a regular (stable) radical (consonant) of this word. The fact that we encounter the initial sound-sequences yu and ji (respectively) in the two words under discussion, whose initial sound was originally the (non-consonantal) vowel u or i (respectively), only when the words are provided with suffixes beginning with consonants, this fact itself must explain the peculiar initial soundsequences yu and ji. The sound-sequences yu and ji from original u and i must have developed in this specific word-structure (i.e., when the words were provided with suffixes beginning with consonants). The initial sound u or i which ugen and ile (respectively) have without suffixes and before suffixes beginning with vowels, must therefore go back directly to the (,sonantic') vowel u or i (respectively) and cannot hence be y or j which have become 'fluid' (> u, i, resp.). The preservation of u in wjen < 'wjn may be 'abnormal'; it is a fact which is thoroughly comprehensible. 86 This u was an essential semantic component of this word. To be sure, the initial y in Egyptian yidn (which Brockelmann, l.c., likewise mentions) and in Tripolitan yuden is apparently not bound to a specific type of suffixes (see above p.89). If we, however, with respect to this, take into account the 86 Beside ud~n is also found ud~n with a long u-vowel, and specifically in Morocco, see Fischer, in MSOS, I, 292. Now, it is known that u and i having developed from !! and j (which became 'fluid', son antic) may appear long as well as short. But we also have in the Moroccan dialect iilar 'trace' « 'alar), iimin 'reliable' « 'amin), see loco cit. (cf. also Fischer, Ibid., II, 276). To these latter instances, therefore, the instance ud~n is also to be added. One can natural1y not separate the lengthening of this original1y 'sonantic' vowel from the lengthening of vowels which developed from original 'consonantic' vowels, semi-vowels (which became 'fluid'), as, e.g., in ui~h « !!agh) 'face', flim 'orphan' « jatim).

THE VOWEL I AS AN AUXILIARY VOWEL

91

basic form 'Uffn and also the situation in the Tunisian dialect (see above), then we cannot but recognize that in the case of the Egyptian and Tripolitan form we are dealing with the expansion of a form developed under specific conditions (which are still apparent in the Tunisian dialect), with the basic form completely falling into disuse. The original (basic) forms of nouns can also be replaced by structures which indeed are developed in connection with pronominal suffixes, but without the break-up of consonant clusters (and the accentual shift preceding this break-up) being involved in this development. As it is well-known, there are in the modern-Arabic dialects wordforms as bU, bii 'father', bu 'brother' (cf. M. Cohen, Le parler arabe des Juifs d'Alger, p. 468; Stumme, Miirchen und Gedichte aus Tripolis, § 69, 6, p. 248; Mar~ais, Textes arabes de Tanger, p. 23/9; Landberg, Glossaire dalinois, I, p.9-1O; Rhodokanakis, pO/iir, II, §47 c), also *bii (in Mar~ais, I.e., p. 287); further cf. Tigre buje 'my brother' (see Brockelmann, Grundriss, I, 331). Rhodokanakis, I.e., has the disappearance of the initial vowel in these words procede from the stat. constr.-form, in referring to Ibid., § 2 c. For Arabic one can, however, not assume that in forms as *'abu, *'abu, etc., used in the stat. constr., the accent would be drawn away from the stem syllable to the terminal syllable. Such an accentual situation is found only in connection with personal suffixes. As a matter of fact, Rhodokanakis (Ibid., §2 c) himself cites as an example the form bah < 'abuhu (hence a form with a suffix). It is clear that the forms bu and bu, developed originally in connection with personal suffixes, could be abstracted from the forms with suffixes, and be used otherwise, also naturally in the stat. cstr. 87 We see thus also in the case of these nouns denoting degrees of kinship how significant the use of nouns in connection with suffixes is for the transformation of their original structure. 88 87 Indeed one must moreover reckon with the possibility that bu < 'abu could also develop from 'abu directly in its use as stat. cstr., namely in proper nouns. One need not assume that a form as be-nuljiis (thus in prari, Rhodokanakis, II, §47 c; but also in other dialects) does not directly go back to 'AbU-Nuljiis. Hence the form is not to be considered as transferred from forms with suffixes, but rather as developed by itself in its use as st. cstr., however, not through the effect of the stress, but rather in consequence of its complete fusion with the following word, resulting from its use in proper nouns, hence through loss of function. 88 Stumme, Miirchen u. Gedichte aus Tripolis, p. 249, § 33,6, has the forms bU and which, in my opinion, first appeared before pronominal suffixes, develop through 'transposition' ('Umspringen') of the accent. This could also appear as correct according to my explanation of the forms insofar as the 'transposition' ('Umspringen') of

au

92

THE VOWEL I AS AN AUXILIARY VOWEL

These remarks concerning the genesis of the forms bU and OU permit us now to determine unequivocally the basic form of one of the most common Semitic words, a basic form which still exists but has not yet been recognized for its linguistic significance. N6ldeke, Neue Beitrage zur semit. Sprachwiss., p. 171,89 writes: "F ... ? 'Mund'. Der gemeinsemit. Ausdruck fUr 'Mund' zeigt sehr wechselnde Formen. Fest ist nur das /oder p .... Was wir als Grundform vorauszusetzen haben, ist ganz unklar. Zu beachten, dass auch hamitische Sprachen den Mund mit einem Wort bezeichnen, das von den semitischen nicht wohl zu trennen ist: Agau, Saho, Adal, Somali a/; Bega yaj, ye/; Galla a/an. Man kann diese doch nicht vom athiop. 'a/ fernhalten und 'afohu usw. wieder nicht vom arab. /uhu usw. Die Annahme, dass alle jene hamit. 90 Sprachen den Ausdruck fUr diesen Begriff dem Athiop. entlehnt hatten, ware aber ganz unwahrscheinlich".91 Indeed, one cannot actually separate Ethiop. 'a/uhu (nom.-gen.), 'a/ahu (accus.) from Arabic/uhu,fihi,fahu. How is one, however, to explain the deviation in the form of the stem? This deviation becomes only comprehensible, in my opinion, if we assume the Ethiop. form as the oldest. From the structural point of view, I identify the forms' a/uhu, ,a/ahu with the forms ' abuhu, ' abahu. These latter forms developed in certain Arabic dialects to bUhu, biihu. And the form without suffixes, bU, or also bii, which is not rarely found, should be considered as the accent actually takes place very often at first before pronominal suffixes (of special structure); see above p. 80. If so, then one would also have to designate the accentuation of the basic forms of bu and OU before suffixes on the pen ultima, i.e., 'aMhu, 'aljilhu (an accentuation which is certainly the ancient one, see however also above p. 90), as a 'transposed' accentuation (" 'umgesprungene' Akzentuation"). However, this would probably overstep the customary application-sphere of the term and concept 'Umspringen' (with respect to accent and syllable structure) so that also bu and OU indeed should not be included in the chapter of "accent 'transposition'" (" 'Umspringen' des Akzents"). The forms buhu, ouhu (and thus also bu and OU abstracted from them) may be considered as the 'Schwundstufe' of 'abuhu, 'aljuhu. The forms bu, ou, compared with 'ab(un), 'alj(un) (just as also Arabic lu, Hebrew Pf < *pi, Akkadian pu 'mouth' compared with Ethiopic 'af, see below) are to be considered as 'Ablaut' -variations. This terminology - 'Ablaut' - might in a certain respect also be applicable to the normal instances with 'transposed' accent (" 'umgesprungener' Akzent"). 89 An essential supplement to Noldeke's compilation of the forms under which the word appears in the modern Arabic dialects (l.c., p. 174), see in Man;:ais et Guiga, Textes arabes de Takrouna, Paris 1925, p. XXVII, n. 2. 90 This is what it should correctly state. The text reads: "semitisclien". 91 According to Barth, Nominalbi/dung, p. 5 "die Grundform [ist] nicht zu ermitteln". In any case, we cannot accept his assumption that the form 1+ vowel is derived from the form lam; for the m of this latter form itself is, in agreement with general opinion, to be considered as constituting a secondary enlargement of the word.

93

THE VOWEL I AS AN AUXILIARY VOWEL

having arisen through abstraction from these forms with suffixes: buhu, biihu « ' abUhu, ' abiihu). In a similar manner originated I;u from ' al;uhu. The same has occurred, in my opinion, in Semitic, probably already in the primary language (in proto-Semitic), with 'ajuhu, *'ajihu, 'ajiihu. Through the effect of the accent abruptly thrown on the 2nd, most sonorous syllable, the initial vowel disappeared, so that the forms juhu, j{hu, JaM arose, which thus are present in Arabic. NOldeke, I.e., p. 172, moreover remarks in connection with the Arabic forms under discussion: "1m Altarabischen ist die vokalisch ausl[autende] Form noch sehr lebendig, aber fast nur im st. cstr. und auch dabei nahe ausschliesslich mit Suffixen. Das kommt nicht bloss daher, dass die Gliedernamen iiberhaupt vorwiegend mit Possessivsuffixen gebraucht werden, sondern die Hauptursache ist wohl, dass das Wort ohne solche gar zu kurz, vom allgemeinen Typus abweichend ist". Indeed, a basic formj ... would be too deviant from the general type. On the other hand, the form preserved in Ethiopic: ' aj < *, afiJ (as 'ab < *'a!Jy, etc.), with suffixes 'ajuhu « *'ajyuhu, as 'abuhu < *' abyuhu) , *, ajihu, ' ajiihu, is a form completely in harmony with other patterns of the formation of nouns.92 (For the details of the development of 'ab < *'aby, 'aj < *'ajy, etc., see below p. 124ff. and p. 130). Ethiopic has thus preserved for us the (approximate) basic form of one of the most frequently used nouns of Semitic.

92

With respect to the development present in

*. af!!uhii >

•afiihii;

*. abIJuhii

> •abiihu,

the following must be kept in mind: A fate common to diphthongic sound-sequences, whose structure is indeed sensitive and which can easily undergo changes through the influence of syntax-phonetic and other factors (tempo, affect, etc.), is the 'sonantization' ("becoming 'fluid''') of their consonantic components, whereby the diphthong -at least in instances such as the present ones-is contracted to a long vowel: yu > u. Thus also, e.g., jaqiimu < *jaCJ!!umu, jadinu < *jadjinu. Concerning this in more detail see the study Some Aspects of the Development of Semitic Diphthongs, below p. 100.

2 A PHONETIC LAW IN THE JUDEO-ARABIC DIALECT OF BAGHDAD a > ;} in Pretonic Closed Syllable

1

In the Judeo-Arabic dialect of Baghdad, I have observed the following phonetic law: The short vowel a in all closed syllables preceding the stressed syllable of a word is shifted into the neutral vowel a. I quote the following instances. Nouns like sqhr 'month', sqnq 'year', Ijaqqa 'leaf' « *ljagqa < Ijaraqa) have the dual forms sahr?n, san t?n , Ijaqt"fn. The numerals 'fifty' and 'seventy' are !Jams{n, sab'{n (as against !Jamsi 'five', sab'a 'seven'). The common Semitic pattern for nomina agentis qattal appears as qa tta I, e.g., rjabbG/:z 'slaughterer', naggag 'carpenter'. The plural of the nominal pattern qatla is qattat, e.g., ganmat (pi. of ganma 'a single lamb', nomen unitatis of ganam), samktll (pl. of samka 'fish', nom. unit. of samak). In forms with pronominal suffixes of the same pattern, the stem appears as qatlat-, e.g., Ijaq'a 'fall': yaq'itu 'his fall'. Further cases are the nomen unitatisforms with the ending -iiji of certain patterns, as sa' (a)g 'hair' (sa'gi 'my hair', etc.): sa'gaji 'a single hair', pl. sa'gajat; /:zabba 'a grain': /:zabbiiji 'a little grain'. Hebrew words like sabbGl 'Sabbath', mabbUl 'deluge', mazzal 'luck, destiny' are pronounced sabbGl, mabbUl, mazzal. I mention the following instances to exemplify the shift a > a in verb forms. Qtaltii « qataltu) (lst pers. sing. of the perf.), but with suffixes: qtattQk (suffix of the 2nd pers. sing.), qtaltQnu (sufT. 3rd pers. sing.), etc. Qatlat « qatalat) (3rd pers. sing. fern.), with suffixes: qatlitni, qatlitak, etc. Qtalna « qatalnii) (lst pers. pl.), with suffixes: qtalnak, qtalnaki, qtalnanu, etc. Siijaltu (3rd stem of the verb) 'I asked', but siijaltQnu 'I asked him', etc. Sqttqm (3rd pers. sing. of the 2nd stem) 'he reviled', sqtmat « sattamat) , sqtmu « sattamii), asqttam (l st pers. sing. of the imperf.); but sattqmtii 'I reviled', sattqmni 'he reviled me', satmatni 'she reviled me', satmQni 'they reviled me', sattamtQk I Originally published in Der Islam, vol. 36 (1960), p. 99-100, with a supplementary (corrective) note Ibid., vol. 37 (1961), p. 263.-A concise formulation of this phonetic rule was also given by me in Arabica, vol. 7 (1960), p. 198, n. 2.

A PHONETIC LAW IN THE JUDEO-ARABIC DIALECT

95

'I reviled thee', asattamu 'I revile him', etc. A further instance is tsalta/:! (5th stem) 'he stretched himself out', but tsa//abtu 'I stretched myself out', etc. We mention also some participle forms: maktub « maktub); man taka I (sing.) 'trusting', but mantakl{n (pl.); magtama' (also mastama'), but magtam'{n (also mastam'{n); etc. In forms of verb stems with doubled medial radical of roots med. Ij/j, the vowel a « a) is frequently assimilated to an adjoining Ij or j, respectively, as dOljljag « dOljljar) 'he looked for', but dUljljagtu 'I looked for'; sOljlja 'he did', but SUljljetii. 'I did'; zqjjqd 'he added', but zijjattu 'I added'; mqjjqt 'he killed', jmqjjat 'he kills', but mijjatu 'he killed him', jmajtQn > jmijtQn > jmitQn 'they kill'. A phenomenon in the Judaeo-Arabic dialect of Bagdad which is analogous to the alternation between a and a described above is the alternation ('Ablaut') between 0 « Olj), e « aj) and ii./u, iii, respectively, e.g., jom 'day': dual ju~n, jum~i 'daily'; noba 'time, turn': dual nubt?n, pI. nubiit; mota 'death', but with suffixes: mut;itu 'his death', etc.; sudii « sOljdii', fern. of asyad 'black') (beside sOda in certain combinations); bi¢& « bajrjii', fern. of abja{i 'white'), etc. Also bet « bajt) 'house'; bit?n 'two houses'; be{i « baj{i) 'eggs', be{ii « baj{iat-) 'a single egg': bi{it?n 'two eggs'. It should be noted that the neutral vowel (,central vowel', 'Mittelzungenvokal') a in this dialect (as in many other dialects) is characterized by a strong affinity to i. And in connection with this it may be stated that the native speakers themselves identify this vowel (a) with i (kasrah, or, with the Hebrew term used by the Bagdadi Jewish speakers, lJireq). For similar phonetic developments in other modern Arabic dialects and for the processes with underlie this shift, see our study above p. 59-60 (see especially also p. 40). Addendum

Jacob Mansour in his Hebrew article "The vowels in the JudaeoArabic dialect of Baghdad", published in Arabic and Islamic Studies (Bar-Ilan University, Ramat Gan, 1973), p.205-240, refers (p.231232) to the present study, as it was originally published in Der Islam, vol. 36/1961, p. 99-100; vol. 37/1962, p. 263. I quote Mansour's Hebrew remark (l.c., p. 232, line 21) here, in English translation: "The vowel a in Bravmann's examples indeed appears as a [literally: 'is heard as a'] when the accentuation moves [i.e., when the accent (or: stress) advances

96

A PHONETIC LAW IN THE JUDEO-ARABIC DIALECT

towards the end of the word], as, e.g., sana-santen. However, this a is nothing but an auxiliary vowel; for in this case there develops a cluster of three consonants at the beginning of the word-and according to § 21 [i.e., in Mansour's article] an auxiliary vowel is inserted between the first consonant and the subsequent two consonants. In phonemic respect, we thus deal with *snten; and in phonetic respect, with santen". While not paying attention in the present context to this kind of differentiation between 'phonetic' and 'phonemic', I must state that I indicated my opinion concerning the development of the vowel a in the instances concerned, in a footnote to the title of the original publication ("A phonetic law ... : a > a in pretonic closed syllable" [see Der Islam, vol. 36/1961, p.99, n. I]): "For comparable cases in other Arabic dialects [outside of Judeo-Bagdadi] and the processes which underlie this shift, cf. our study "Uber i als Hilfsvokal ['auxiliary vowel'] im Wortinnern", Le Monde Oriental, vol. 32/ 1938, p. 1-100 (especially p.63-65)". In the present reprint of my study concerning a in Judeo-Bagdadi, an equivalent statement, in a slightly different wording, appears at the end of the study; see above p. 95, lines 27-29. By referring to my earlier study ("Uber i als Hilfsvokal im Wortinnern") and especially to the pages 63-65 in it (cf. the present volume, p. I ff., and especially p. 59-62), I indicated that I consider the vowel a, instead of an original a, in the Judeo-Bagdadi instances under discussion, as an 'auxiliary vowel' taking the place of a vowel which disappeared - a fact not paid attention to by Mansour in his statement with respect to sana-santen, etc. (see above): " ... However, this a is nothing but an auxiliary vowel ... ". An important aspect of the appearance of a in place of a "in pretonic closed syllable" in the Judeo-Arabic dialect of Bagdad consists in the regularity of this phonetic phenomenon, a fact which almost transformed the basically phonetic phenomenon into a quasi morphological characteristic of certain form-systems which were regularly affected by this phonetic shift. This feature certainly represents a specific characteristic of the Judeo-Arabic dialect of Bagdad. I may also mention here that this phonetic peculiarity is a basic feature of A. S. Yahuda's phonetic notation (reflecting the Judeo-Bagdadi dialect) in his study Bagdadische Sprichworter, published in Studia Orientalia [Festschrift Th. Noldeke], 1906, vol. 1, p. 399-416. In his phonetic rendition of the Arabic texts, the 'auxiliary vowel' (,Hilfsvokal') taking the place of an original a-vowel is regularly expressed by i, which in this specific use is to be considered as related to, or phonetically

A PHONETIC LAW IN THE JUDEO-ARABIC DIALECT

97

identical with, the vowel a (the vowel a in Arabic [and Semitic generally]-especially where it serves as an 'auxiliary vowel' -has an affinity to i); cf .above p. 40ff. Indeed the material presented by Yahuda might already have served as basis for the phonetic rule of "a> a in pretonic closed syllable in Judeo-Bagdadi". Hassan el-Hajje in his Le parler arabe de Tripoli (Paris 1954), a study of the Syrian-Arabic dialect of Tripolis (Tarabulus as-Sam), presents (p. 28) instances of the occurrence of the vowel a where this vowel actually appears in pretonic closed syllable in the place of an original a-vowel. Of special interest in our present discussion is the paradigm of the '2nd stem' of the verb (ja"'ala) in this dialect, as presented by el-Hajje (I.e., p. 73): kalliim 'il a parle [a qqn.]' (3rd pers. sing. masc.), kallmet (3rd pers. sing. fern.), kallam't (2nd pers. sing. masc.), kallamti (2nd pers. sing. fern.), kallam't (1st pers. sing.), etc. As a matter of fact, the function of a, in all these cases, as an auxiliary vowel in pretonic closed syllable is not indicated by the author. However, the phonetic rule is not, according to the phonetic notations of the author, without exceptions in this dialect. Cf., e.g., I.e., p. 28, lines 1-2; and also p. 56, at the bottom of the page ("Ie participe passif"), where the forms of the passive participle of the "1st stem" which are mentioned, appear in the form maJul (not *maJul).

3 SOME ASPECTS OF THE DEVELOPMENT OF SEMITIC DIPHTHONGS 1 A. The Development of the Diphthongs all, ai The Hebrew language manifests two different models for nouns of the pattern qatl of the roots med. ylj, that is, for such forms as Proto-Semitic and Arabic mayt 'death' and bajt 'house'. The first is characterized by the monophthongisation of the Proto-Semitic diphthong, e.g., jom 'day', instead of jaym; Mq 'lap', instead of *J:zajq; and lei (Is. 21,11) 'night', instead of *lajl(aj), in Arabic lajt.2 In the second, the Proto-Semitic diphthong has been replaced by a disyllabic structure: bajil, mayt:l. There are two problems here to be solved. First, how the origin of the second of the above types is to be explained phonetically; 3 and, second, how it came about that the same basic form followed two different directions of development. The type bajil, mdyt:l is evidently the result of a phonetic process which is also to be found in other languages. In this connection the following quotation from K. Wulffs "Beobachtungen tiber die Aussprache des Malayischen im Sultanat Perak" (Acta Orientalia, vol. IV (1926), p. 289), is of interest: "Von anderen Dialekten ausgehend ist man sich sehr uneinig gewesen, ob a + i und a + u im Malayischen Diphthonge oder zweisilbige Verbindungen sind; holUindische Autoren gehen so weit, dass sie zwischen a und i ein y (j), zwischen 1 Slightly revised and enlarged version of a study published in Orientalia (Rome), N.S., vol. 8 (1939), p. 244-260; vol. 9 (1940), p. 45-60. 2 These forms existed as far back as Old Canaanitic; see Brockelmann, Grundriss, I, § 71 ey, Anm. I; N6ldeke, ZA 30, 168; and Bauer and Leander, Histor. Gramm., p. 202 k and I, p. 205 f . 3 It is generally accepted that the i and ~ in bajiJ and mdy~l, respectively, are parasitic vowels and identical with the ~ of the so-called segolates. Thus, Brockelmann, Grundriss, I, § 82 k£; Bauer and Leander, Histor. Gramm. des Hebr., p. 202j; Bergstrasser, Gesenius' Hebr. Gramm. 29 , I, § 23 i; loiion, Gramm. de l'hebreu biblique, §96A 1. Sievers, Metrische Studien, I, §203/4 is of the opinion that the presumed segolation did not develop naturally, but was artificially introduced and at a later date. Almost the same point of view is held by BrockeImann, Bauer and Leander, and Bergstrasser.

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

99

a und u ein w einschieben (bajik, dawun schreiben). Fiir den Perakdialekt ist diese Frage miissig: die Grenze zwischen Einsilbigkeit (also diphthongischer Natur) und Zweisilbigkeit solcher Verbindungen ist ja durchaus fliessend;. was einer fUr Diphthong erkHirt, wird ein anderer behaupten, seien zwei Silben; in diesem. Dialekt schwankt ausserdem die Aussprache je nach dem Tempo der Rede und der Betonung, so dass man in demselben Wort bald von einer, bald von zwei Silben reden konnte. Ich habe, urn ein Beispiel zu nennen, das Wort ..:;...Jb L sahut oder saut nicht nur als ein saUl, das mir durchaus zweisilbig vorkam (aber natiirlich keinen Kehlkopfverschluss hatte), sondern auch als saMt gehOrt, daneben aber die Ableitung Piaut so kurz gesprochen, dass man beim besten Willen keine zwei Silben hatte heraushOren konnen". The form of the Malayan bajik thoroughly corresponds to that of the Hebrew bajil and that of the Malayan dOlJun to that of the Hebrew *mayul, which Brockelmann designates as the precursor of may{!l (Dissimilation - Grundriss, I, § 93, I). The reference quoted now makes the mechanism of the origin of the forms bajiJ and malj{!l, perfectly clear: the i and u > {! in bajiJ and *maljul > mayel, respectively, are not, as is usually assumed, new vowels in the sense of parasitic vowels as with the segolates, nor are the j and Ij in bajil and *mOlJul identical with the j and Ij in the basic forms baft and mOlJt. The process of their development has been quite another. First, the j or Ij became the sonantic i or u, respectively, giving *bait and *maut (disyllabic). Thereupon, to bridge the hiatus between the a and i or a and u, the glide sounds, j or Ij, respectively, developed. Thus we finally come to the forms bajiJ and *mayuj. 4 As far as the type jom, /:leq is concerned, I am of the opinion that they show the normal treatment of Proto-Semitic diphthongs in accented syllables in the Hebrew language. 5 4 For a similar example, see the case of juam > jUljam 'day' in the dialect of elI;Iamma of Gabes in Tunis as cited by W. Mar aji) and a-u (> ayu), which were naturally not affected by the contraction.

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

101

Since the contracted forms originated as far back as the Old Canaanitic period (see above p. 98, n. 2), we are forced to the conclusion that the disyllabic forms just described, being even older, must certainly have existed in the earliest Canaanitic. Forms of the type baji! are found also in Biblical-Aramaic. Examples of this are qajit 'summer' (= Hebrew qaji~), and 'fgajin 'then'. In the Syriac language, too, there exist such forms as mayemjii 'she conjures' < maymejii, zaye'!ii 'trembling' < zay'e!ii, Neo-Syriac kayednii 'mule' < kaydenii, tayertii 'cow' 6 (vide Brockelmann, Grundriss, I, § 8218, Syr. Gramm., 5th ed., § 73, N6ldeke, Neusyr. Gramm. p. 23). However, these instances from the Syriac have not arisen as the result of phonetical factors in the sentence in which they occur, as in the cases described on p. 99, but rather developed almost compulsorily as the result of moments present in the words themselves. Whenever the reduced vowel (the seyii) was dropped in such Syriac forms as maffneM > *maffn/:lii, the resulting accumulation of consonants was split up by an auxiliary vowel, > maffen/:lii. In instances having diphthongs, however, such as maymejii > *maymjii, the consonantic complex was broken up by the sonantisation of the consonantic vowel y > u: *maymjii > *maumjii > *mayumjii > mayemjii (cf. Brockelmann, Syr. Gramm., 5th ed., §86).7 Besides the two patterns already treated - jom, /:leq, with monophthongisation of what was originally a diphthong, and baji!, maye!, with conversion of the original diphthong into two separate syllables - there exists a third type. This is represented by such forms as say ('), yay, gaj ('), /:laj and is characterized, in contrast to other monosyllables, by the fact that the diphthongs occur terminally.8 These diphthongs remain uncontracted, which is confirmed by examples from certain modern Arabic dialects which, as a rule, show contracted diphthongs. Thus, in the dialect of Tripoli (Stumme, Miirchen und Gedichte aus Tripolis, p. 212, § 18). For similar examples from the dialect of Takriina see Mar~ais et Guiga, Textes arabes de Takrouna, p. XXII, n. 1. Further examples will be found in the dialect of 'Oman (Reinhardt, Ein arab. Dialekt gesprochen in 'Oman Instead of Old Syriac tortii. Brockelmann, I.e., § 73, considers the e in mayemjii on the one hand as being identical with the e in ma4en~ii (v. above p. 98, n. 3), on the other hand (I.e., Anmerkung 1) to be the result of an alleged peculiarity of accentuation (I.e., §37). 8 This characteristic was disregarded by Sievers when he designated these forms as direct proof of the old monosyllability of the type baji!, mdyel, thus arriving at -bail, *mdyl as the originals (Sievers, Metr. Stud., I, §204,4, p. 286, above). 6

7

102

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

und Zanzibar, p.8, §5 and p.38, §21). As evidence may be cited sau 'evil', sei 'thing', r;lau 'brightness', ~ai 'alive', as compared with such forms as nom, bet with contracted diphthongs. It is thus evident

that monosyllabic words with terminal diphthongs form a group in themselves. To answer the question why this should be so in languages where monophthongisation is the rule, we must take into consideration the fact that monophthongisation presupposes extreme, consonantic shortness (glide character) of the second component of the diphthong. The Tagljid literature 9 expressly states, however, that the j and Ij in such words as sar and salj' are to be pronounced with madd (extension) when occurring at a pause. 10 "Madd consists in raising the middle of the tongue to the palate, as when pronouncing the j in lalj! and gaj!, and holding it in this position for a measurable length of time". The Ij, in salj', for example, is to be pronounced correspondingly.u As pointed out in my thesis (p. 134), the intention here must be to extend the consonantic j and Ij beyond the normal value of a consonant without the monosyllabity (i.e., the diphthongic character) being lost as happened in the case of baji!, malje!. From the instructions in the Tagljid literature as to the Old Arabic pausal sar etc. one may draw conclusions concerning the Hebrew and NeoArabic cases under discussion,12 in which aj and ay are not contracted as is otherwise usual. It can hardly be denied that the observation in the Tagljid literature on the extension of the j and Ij in sat, salj' is definitely to be considered together with that of modern dialectologists concerning the non-contraction of the terminal diphthong in monosyllabic words. 13 We have already mentioned such Biblical-Aramaic forms as qaji! 'summer' ~ = Hebrew qaji~), which correspond to the Hebrew type The literature on the art of reciting the Qur'an. The state in which the words are spoken without inflection vowels. I I Source literature in my M aterialien und Untersuchungen, p. 80. 12 For the rest, see Stumme, I.c., p.2l3, n. 2: "The diphthongs are often disrupted (,zerdehnt'), for example 58, 12 sei = 'thing"'. 13 This rule, however, is not without its exceptions. For instance, in Hebrew the word gaj (st. absol.) is also transmitted as ge (st. absol.), which can no more be considered as having intruded from the st. constr. than sor < *sayr (besides *mt'iyu!) (see p. 99). It takes its place along with the Tripolitan se, which Stumme (see above p. 101). has proven to exist in addition to the normal saj. The same lack of uniformity can be seen in the terminal sounds of polysyllabic words, for instance, the Hebrew siiljaj beside saqf 'field', Ethiopic sataj beside sate 'beverage', gabO 'side' beside ma(iay 'spring'. 9

10

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

lO3

bajil. This is the normal development of the original diphthong aj in an accented, closed syllable. In an unaccented closed syllable the diphthong is regularly contracted, e.g., in forms of the Status constructus (as bel, the constructus of bail{n. Open syllables, whether accented or unaccented, are generally considered to have retained their original diphthongs (vide Brockelmann, Grundriss, I, § 71 f.). Examples of the latter are 'ajnin 'eyes', bazaila 'you saw', baila 'house'. However, I wish to express my doubt that these are really the original, unchanged Proto-Semitic diphthongs. This conclusion is forced upon us after considering a certain peculiarity of all these cases, namely, that when a t 14 comes immediately after a diphthong, it is spirantic (fricative), as in baila (Syriac bajta), bazaila (Syr. bezajt), haili (Syr. 'ajti) 'he brought'. However, the spirantisation of the stops b, g, d, k, p, t in North Semitic depends on their being preceded by a sonant. 15 At the time the spirantisation occurred, therefore, the second component of the diphthong preceding the t could not have been a consonantic j. Rather its quantity must have exceeded that of a consonant, if only by a minimum amount. It follows necessarily that every j which formed the second component of the diphthong aj in an open syllable must have been sonantised before the process of spirantisation took place. The resulting hiatus was then bridged by a parasitic j (see above p. 99). Thus bajta> *baila > *bajila. Later the i was eliminated in accordance with the ordinary rules of Aramaic phonetics, without the disappearance of the spirantisation of the t: *bajila> baila. Biblical-Aramaic bail{l, etc., presuppose a former *bajila « bajta), etc., just as the Hebrew locatives, Mila, mt4!la, presuppose the basic forms *bajila, *maljula. With the latter this is obvious, since they are derived from forms of the Status absolutus, bajil, *miiljul.

What happens to the falling diphthongs aj and t1lj in Old Arabic? Are they completely retained? I believe I can demonstrate that at a certain period they were contracted to e and {) when occurring terminally in a word,16 while remaining the same when occurring 14 Other members of the group, b, g, d, k, p, t, besides t are not found following diphthongs in the texts. 15 A consonantic j or !! can not produce this effect any more than any other consonant. 16 The Status constructus (of the dual) does not fall into this category. Rather, logically enough, it is subject to the conditions governing medial sounds.

104

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

medially. The e and jj sounds thus formed, to be sure, later became a through a general phonetical transformation, in common with all other e and jj sounds of whatever origin. I explain in this manner, for example, Arabic 'ala 'on' as compared with the forms with suffixes as 'alajhi and with Hebrew 'ale < *'alaj; also Arabic 'ita 'to' as compared with 'itajhi, Hebrew 'ale < *'ilaj; also Arabic lada 'at' as compared with ladajhi; and also Arabic mata 'when' as compared with the Hebrew malaj. 1 7 The contraction of the diphthongs aj and ay, which in Old Arabic is limited to terminal sounds, is naturally based on the same supposition which I feel to be a preliminary condition of their contraction in general, namely, the consonantic shortness of their second component (see above p. 100). In certain modern Arabic dialects the rule of monophthongisation is valid not only for the terminal sound, but generally. However, the law does not apply in the case of monosyllabic words with terminal diphthongs, as say', sal', in consequence of the reasons presented above (p. 102). There are also several cases of retained terminal diphthongs in Old Arabic, of which only lay, kaj and ' aj are known to me. Thus, here, too, we are dealing exclusively with monosyllables. Accordingly, I assume in these cases, that the length of the second component of the diphthong is greater than that of a consonant, thereby preventing the contraction of the diphthong. The retention of terminal diphthongs, in monosyllabic words in Old Arabic and many dialects of modern Arabic, does not, therefore, conflict with the law. The law of the contraction of terminal diphthongs in Old Arabic has transformed to a large extent the inflection of the nominal and verbal forms of roots III fly. I base my conclusions on the hypothesis that they were originally, in Old North Arabic as well as in Ethiopic and Old South Arabic, 'strong' (regular) forms, without giving special proof of this or even considering such to be necessary. Thus, next to ramajtu must at one time have stood, instead of the present-day rama, 17 Brockelmann, Grundriss, I, §7l b, Anm.l (p.190), in order to explain the difference between 'alii, 'alajhi, etc., states that the prepositions under consideration have taken on the accusative ending, as, e.g., !Qljqa, tal)ta, and that aja, in the postulated *'alaja, in accordance with a law laid down by him (l.c., §39w), has been contracted to ii. This ii, however, would have had to be taken on also by the forms with suffixes, so that we should expect such structures as *'aliihu and *'aliika. In the case of the interrogative matii < *mataj it is in the nature of things impossible to employ such an accusative-a. Here it is particularly clear that the ii in question traces its origin back to the terminal diphthongs.

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

105

a *ramaja on the model of qatala; next to gazaljtu, instead of gaza, a *gazalja. Furthermore, the participia passivi of the derived conjugations must once have been: *murmaju(n) , *mugzaju(n) (or *mugzaljun), instead of murma (n), etc. It further stands to reason that the present-day jar{ia must be based on the forms *jar{ialju and *jar{ialja. Finally, such nominal forms as !ala(n) and nada(n) presuppose the basic forms *!alalju(n) , *!alalji(n) , *!alalja(n); *nadaju(n) and so forth. In my opinion, all these forms can go back only to structures without a terminal vowel, as *ramaj, *gazalj; *nadaj and *!alalj. According to the law stated, the terminal diphthongs must have been contracted to 0 or e, > *rame, *gazo, *nade, *!alo, etc. Finally, these 0 and e sounds were shifted to a. ls How do we arrive, however, at forms with terminal diphthongs, which later must have been subjected to monophthongisation? 19 In Old Arabic all roots, no matter what their structure, show special pause forms besides the context forms. The former are distinguished by the lack of the short terminal vowels, and have existed along with the context forms terminating in vowels since earliest times. 20 To the context form qatala belongs the pause form qatal. The pausal forms corresponding to the Proto-Arabic context forms, *ramaja, *gazalja, *murmaju, etc. must have been: *ramaj (as qatal to qatala), *gazalj, *murmaj, etc. Consistent with the law of contraction of terminal diphthongs in Old Arabic, these latter forms must have then become *rame, *gazo, *murme, etc., and their e and 0 must have later been converted to a as the result of a further process. The usual difference between the pausal forms and the context forms is that in the former the terminal vowel is missing. The stems themselves, however, remain identical. On the other hand, in the case of the 18 According to an Ancient Semitic law advanced by Brockelmann (Grundriss, I, §39w, §271 Bb), the forms would be explained as contractions: aya, aja; ayu, aju > Ii (compare p. 104, n. 17). At least the contraction of ayu, aju > Ii, is phonetically impossible. Since this law is supposed to apply to Proto-Semitic ('Ursemitisch'), one would furthermore have to assume that Ethiopic forms such as jer' aj, je/:ljay are new formations, a hypothesis which is otherwise completely without foundation. 19 Landberg (Etudes sur les dialectes de ['Arabie meridionale, II, Dalina, p.386) has constructed such forms as would have to be presupposed, assuming that, in the form fa'aja, already in earliest Arabic times the terminal vowel was dropped not only at a pause but generally. It is, however, inconceivable that the customary loss of the terminal short vowel in New Arabic had already taken place in Old Arabic especially in roots III semi-vocalis. 20 This is a specific characteristic of Old Arabic which must go back as far as Proto-Arabic and whose age can not be a matter of discussion.

106

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

roots III IiI:! there resulted a marked variation in the formation of the stem because of the above mentioned law of contraction, which affected only the pausal forms. For the context form *ramaja, etc. there was the pausal form *rame or, later, ramii. The context form possessed three radicals, whereas the pausal form had two radicals and a terminal long vowel. The discrepancy between the stem of the context forms and that of the pausal forms of roots III jil:! was not reconcilable with the accustomed identicalness of these stems in the case of all other roots. Those forms which occurred at the pause, accordingly, supplanted the others. Thus the context forms, also, of words of roots III IiI:! no longer had terminal inflection vowels. Terminal ay was contracted in an earlier period (> 0 > ii) than terminal aI (> e > ii). The contraction of terminal ay still belongs to that period of development of the Arabic language before the appearance of writing, whereas aI was not contracted until after its introduction. This is proven by the fact that the third person singular masculine of the perfect tense as well as those noun forms of roo!s III I:! on the model of qatal, are written with a terminal ' alif: Ty.., 5\1 ; the corresponding forms oflII I, on the other hand, with j :~J' ~..G. Orthographic differences of this type are generally to be interpreted on the basis of the phonetical development of the language. Barth (ZDMG 59, 159 sq. and 633 sq., also Sprachwiss. Unters., I, Leipzig 1907, p. 36 sq.; compare also Nominalhildung, § 232) has attempted to trace back the terminal ii in the above mentioned cases to an alleged primeval and common-Semitic e. Although unable at this point to go into the particulars of Barth's proof, I should like to mention that his explanation of the relation of 'ilii to 'ilajka, ramii to ramajta, and so forth-as a disruption (,Zerdehnung') of the monophthong e into the diphthong aj on the accession of a suffix -is phonetically impossible. (See esp. Sprachwiss. Untersuch., I, p. 37). Moreover, against Barth one should also take into account the statements of A. Fischer (ZDMG 59, 663 sq.). To be sure, equally as unacceptable as Barth's claim concerning an original e is also Fischer's opinion (l.c., p. 670 and n. 4) that the corresponding e vowels of North-Semitic and Ethiopic are to be traced back to primary ii in Arabic in the cases in question. B. The Contraction of the Diphthongs ya, ia > 6,

e

The various possibilities of the evolution of the Proto-Semitic falling diphthongs, ay, aj, have been presented. We now come to

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

107

the corresponding rising diphthongs, 'la and ja, for which we must ascertain whether they are liable to contraction and, if so, under what circumstances. As I have attempted to show (see above p. 100), it may be said that, in general, only true diphthongs are subject to contraction, i.e., those whose second component is of absolute consonantic shortness (glide character). In applying this to the contractibility of ja, 'la (> e, 0), I should like to stress that this rule states not only that one of the two components must possess glide character, but that this consonantic vowel must be the second component of the diphthong. The logical consequence is that we must presuppose the intermediate stages, ug and ig, respectively, for IJa > 0 and ja > e. As already indicated, the contraction all, aj > 0, e is absent in certain Semitic dialects 21 and in others takes place in a regular manner,22 without one knowing as yet the factors governing this contraction. In contrast, the transformation of the diphthongs 'la, ja > 0, e shows a complete lack of regularity in all Semitic languages, as is evident from the material presented below, p. 108ff. In the transition of 'la, ja to 0 and e, the immediate precursors of 0 and e are the falling diphthongs, ug, ig. In Semitic languages these never occur as primary phonetic complexes, but generally as developments from 'la, ja. They lend themselves poorly to articulation,23 for which reason, unless retrogradation to ja, 'la takes place, they are almost of necessity contracted to e, 0. 24 The development of the falling diphthongs, ug, ;g, from the rising diphthongs, 'la, ja, however, shows neither the conformation to rule of the contraction of the falling diphthongs 01j, aj> 0, e nor the compulsoriness of the contraction of the falling diphthongs ug, ;g > 0, e. Rather it depends on certain factors which are only active under the proper circumstances, similarly to the transformation of the diphthongs aj, 01j into the disyllabic combinations a - i > aji and a - u > O1jU. 25 This transformation takes place 1) on the basis of certain phonetic conditions given in the words themselves, making the sonantisation of 21 Especially in Old Arabic, except in the termination of the word (see above p. 103ff. 22 See above p. 100. 23 See Sievers, Grundzuge der Phonetik, 5th ed., §420. 24 A direct contraction of the rising diphthongs ya, ja > 0, e is hardly conceivable, as, with the phonetic arrangement present in these diphthongs, the individual components are brought sharply into relief. It is for this very reason that most students of Semi tics have been unwilling to trace 0 and e to ya and ja. (See below p. 113, n. 59). 25 See above p. 99.

108

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

the consonantic y and 1 necessary 26 and 2) by means of phonetic causes contained in the whole phrase, whose character is described above. 27 These two groups of factors are to be found in the transformation of ya > ua and ja > ia, also. Then, if the first, consonantic component in ya, 1a becomes sonantic (y, 1> u, i) by the operation of either of the two factors, the consonantisation of the second, sonantic component follows almost of necessity, at least in the cases we have studied, i.e., ya > ua> uq (> 0), ja > ia> iq (> e).28 We shall consider first those cases in which ua> uq (> 0) has developed from ya 29 because of conditions given in the word itself. The articulation of consonantic y immediately after labial consonants, such as m, b, f, is difficult. This fact can lead to its sonantisation. In this manner ya which follows a labial can develop to ua> uq>

o.

In this connection it may be well to cite the examples of diminutives in the dialect of Tripoli given by Stumme :30 bUoj (bboj, etc.) from bii 'father', fojjim from famm 'mouth', mUojjii (mmojjii, etc.) from and synonymic with ma 'water', beside other formations as ubej from bii 'brother', ubejjii from ubt 'sister', $yebii' from $bt/' 'finger' and syereb from sareb 'lip'. 31 It will be seen that, according to expectations, the diphthong aj in the form fu'ajjil generally results in e, but in certain cases in oj. This diphthong oj appears only when the word begins with a labial. To explain the sonantic 0 in this diphthong oj we can formulate the process phonetically as the result of a contraction between the originally present y and the a-vowel in the basic form *fyajjim < *fuyajjim. According to this, we have here in the Tripolitan dialect a case of the contraction of an originally rising diphthong ya to 0, which Brockelmann 32 failed to confirm for the Semitic languages. The question now arises as to the cause of this contraction. Obviously the articulation of y in contact with the initial labial is difficult, as See above p. 101. See above p. lOa. 28 ua, ia would have the status of hovering diphthongs, which are not popular; at any rate, they disappear in the course of time. A second possibility for their development to a structure more easily pronounced would be a transformation to disyllabics, as ai > ali, au > ayu (compare above p. 99). 29 ja is out of the question here. 30 Miirehen und Gedichte aus Tripolis, II, § 124, p. 261. 31 L.e., § 121, p. 260. 32 Grundriss, I, §69. 26

27

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

109

two such lip articulations coming one after the other are hardly compatible. Unless the position of the y is changed, 33 the difficulty can only be overcome by sonantisation (deconsonantisation, liquifaction) of the consonantic y, i.e., by giving it vowel quality, thus lengthening it to u. The difficulty in the pronunciation of *myajja, *fyajjim is not present in *muajja, *fuajjim. The intermediary of liquid (sonantic) u < y in the contraction of ya to 0, which I determined purely constructively, has been confirmed by the Tunisian dialect. In this dialect 34 the diminutives formed according to the pattern fu'ajjil (from fu'ajl) are, in the case of stems whose first radical is not a labial: lyejjir 'little ox' from lur, ryejjab 'the poor spirit' from rueb, byejiy! 'little thread' from bY!, etc. On the other hand, bllb 'gate' and far 'mouse', whose first consonants are labials, have as diminutives buejjib and bojjib 'little door' and fuejjir and fojjir 'little mouse'. The same holds true for bU 'father' with boj 'little father' and fumm 'mouth' with fuejjim, fojjim 'little mouth'.35 According to loco cit., §85 sub 2, there also exist fuojjim, fuojjim. Of interest is also bUyqet to bUqel 'water flask' in contrast to non-labially beginning words of the same structure, as fyireb to sllreb 'lip', zyimel to zllmel 'pathicus', qyyteb to qaleb 'mould, last'.36 This dissimilatory transformation of y into u provided the basis for the contraction of y + a to O. The 0 before the geminated jj was shortened to 0, the usual fate in Semitic languages of long vowels preceding geminates. A form which originated quite similarly is bebojjib 'beloved one' in a Bedouin dialect of Central-Arabia,37 a variant of bebaiiib 38 < bubajjib. There is no indication of the original existence of an etymologicaly which merged with the a to form O. It must be assumed that a glide sound was formed after the b, a process as common in 3J It appears to me that a similar situation (metathesis) is present in the SpanishArabic forms cited by Brockelmann, Grundriss, I, §99b in the chapter on epenthesis, viz. le/)ba < labljat 'lioness' 'aybar, plural of bor 'fallow land', ayba'a, plural of ba'a 'step' and Classical Arabic 'ayfaqa 'to put the arrow with the notch (juq) to the bow-string'. The reason for changing the position of the lj in these cases does not seem to be epenthesis. It seems more probable that we are here dealing with a skipping of the lj because of the difficulty of articulating it immediately after a labial (cf. I.e., Anmerkung, p. 279). 34 Stumme, Gramm. d. tunis. Arab., §85, 2. 35 Cf. Stumme, § 86. 36 Cf. Stumme, §89. 31 Socin, Diwan, I, n. 50, v. 11. 38 L.e., n. 21, V. 8.

110

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

this dialect as in others.39 This glide sound then merged with the following a to result in o. This presupposes that the !!, which originally arose from the b itself as a glide sound from the b to the vowel following it, later became a phoneme for itself with a proper value, independently confronting the b. Only as long as it remained a glide sound, articulated as a psychological and temporal unit with the b and not considered as an independent phoneme, could it succeed the labial b without giving rise to difficulty; as soon as it became an independent phoneme, !!, however, the articulation of the labial!! immediately sequent to the labial b must have presented difficulties. Therefore, the conson antic vowel, !!, then dissolved into the sonantic vowel, u. The 0 in this word can only be derived from ua> ug. This u, in turn, must have been derived from !!, as the result of the above-mentioned tendency to dissimilation. Thus we arrive at bebajjib> *bebUajjib> *beb!!ajjib > *bebuajjib > *bebojjib > bebojjib. We shall now discuss certain of the special phases of development of the above-mentioned words, beginning with labials. Stumme 40 cites such forms as bUagjil 'water pitchers', mUojja 'water', f'J~tah 'in its middle' instead of 1--1ja~tah = Ii !!asatihi, f'ebii 'in a Weba', f'ut 'napkins'. Stumme 41 remarks that the in these words, "der letzte Rest eines sich nach den Labialen verfluchtigenden, etymologisch also wirklich vorhandenen !! sei". The same view is held by Marc;ais.42 Brockelmann also 43 assumes that an etymological !:! was Oiiginally present and later reduced to a glide sound. He considers this supposed reduction of !! after a labial to be the consequence of a tendency to dissimilation. I fail to understand in what manner a reduction of !! to (that is a reduced !!) lessens the difficulty of pronouncing two lip articulations. In fact we still have two labials in contact with each other. The!! no more loses its character as a labial by reduction to than true consonants such as t, k, change their articulation when reduced to t, k, as in Arabic dialects. 44 I have, therefore, come to the conclusion that these forms have been derived from the forms bGgjil, mojja, jd~tah, which also exist in the dialect. The actual mechanism could be described as the secondary developU

U

U

39 Compare for instance bIJinnin < binnin in the same verse; cf. Socin, I.e., III, § 159 a; Stumme, Miirchen und Gedichte aus Tripolis, §9, p. 202; § 17, p. 212. 40 L.c., p. 211-212, § 16. 41 L.c., § 9, p. 202. 42 Le dialecte arabe des Ollld Briihim de Saida, p. 31. 43 Grundriss, I, § 85 c. E., p. 233. 44 See, for example, Feghali, Le parler arabe de Kfar'abida, p. xv.

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

111

ment of a glide sound between a labial and a vowel, a process common in this dialect. These latter forms for their part go back to the precursors *blpjgy/ (for *bayiigil), *mljajja (for *mUljajja), j-ljil!f!ilh (for ji Ijasa!ihi). Stumme 45 remarks concerning the forms bcigy/, mojja, jd!f!ilh, which have arisen from the words mentioned in the preceding paragraph: "In der stets vorhandenen Verdumpfung des a-Vokals ist hier noch Andeutung an ein geschwundenes Ij vorhanden". mojja, which occurs in many dialects, including eastern ones, must be distinguished from the other two examples in that it contains an 0, whereas the others contain an il. This 0 can be explained as a contraction of Ija > 0 in the manner already described. In the case of bcigy/ and jd!f!ilh, on the other hand, I find complete disappearance of the Ij as the result of the tendency to dissimilation. In contrast to this, the particular quality of original etymological a following the Ij does not seem to me to have any connection with the disappearance of the Ij. It seems far more likely that the nuance il developed even before the loss of the Ij. An appropriate comparison is provided by the Tunisian Ijilqt 'time', Ijil!f!fa 'commissioned'.46 It frequently occurs that the components of diphthongs mutually influence each other in their nuances, even without contraction, and this is readily understandable. "Der Ermittelung der Komponenten stellen sich aber oft ziemlich grosse subjective Schwierigkeiten entgegen. Einerseits Hiuscht die Kontrastwirkung der beiden Nachbarlaute iiber ihren wahren Charakter, anderseits treten in den Diphthongen oft Vokallaute auf, die in den betreffenden Sprachen als isolierte Vokale nicht vorkommen und daher umso leichter falsch eingeschatzt werden".47 If, however, as in our cases, one of the components has disappeared, the contrast effect mentioned by Sievers is no longer present. We can then visualize clearly the nuance of the remaining component, which has been influenced in its vowelpitch by the one which has disappeared. It is in this sense and only in this sense that it is permitted to state that an intimation of the missing Ij is still present in the dullness of the a in Mgy/ and jd!f!ah. However, it must be admitted that even without a preceding labial consonant, one finds the sonantisation of consonantic Ij and j U

45

46

L.e., § 16, 4, p. 212. Stumme, Tunis. Miirchen und Gedichte, p.

§69c. 47

Sievers, Phonetik, 5th ed., §415.

XXVI;

Brocke1mann, Grundriss, I,

112

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

in the rising diphthongs, Ija and ja, which, in turn, give rise to the falling diphthongs ug, ig > 0, e. In such cases, the moments mentioned above p.l08, must be considered as the factors in play. I refer to the following examples in modern Arabic dialects: dbiat, iiblad = Old Arabic 'abjarj; asuadu, asoat = Old Arabic 'asljad-all taken from the Bedouin dialects of the African interior. 48 Of interest are also the statements of Stumme 49 and the Tripolitan examples sirual 'trousers' < siryiil; Mui 'speak' (imperative of dayii), etc. The rule given by Stumme 50 is important in this connection, as are also the various Tunisian examples quoted by him. 51 According to him, 52 in such Tripolitan cases as gahlja, keslja and mafja, u and i can regularly take the place of Ij and j, respectively. These cases are identical with those occurring in Old Semitic, which will now be discussed. In my opinion, we must take into consideration that a contraction of post-consonantic Ija > 0, ja > e occurred in all phases and in all branches of the development of Old Semitic, as far as the nominal forms of the structure qa/i/utlat of roots III f/Ij are concerned (i.e., forms on the pattern ja'ljat, ji'ljat, jU'ljat, ja'jat, etc.). However, since this contraction depends on syntax-phonetic factors 53 which are not generally effectual, it has neither taken place in all these forms while in the oldest nor while in the more recent phases of development. For this reason we still find many 'strong', uncontracted forms in all Semitic languages. Compare Arab oa!ljat 'stride', gazljat 'raid' parallel to gaziit < *gazOt,54 nagyat 'rescue' parallel to nagiit, derived from and parallel to I;Iigazi nagot,55 etc.; Ethiopic janljat 'way' parallel to janot, qanjat 'service' parallel to qanet,56 etc.; Aramaic *~ajljalii > ~ajljelii > ~aj(j)ulii 'life, animal', ludaeo-Aramaic ga'jelii 'cry' parallel to $elOlii 'prayer;, m eniilii 'share' < *m enolii,57 etc. 58 48 Kampffmeyer, Mitt. d. Sem. f Orient. Spraehen, II, p.158/9. Compo other, Magribinic cases, Ibid., XII, p. 40. 49 Miirehen U. Gediehte aus Tripolis, § 15,3, p. 211. 50 Tunis. Gramm., §2 sub 5, p. 6. 51 L.e., § 55, 2, p. 49. 52 Miirehen U. Gediehte aus Tripolis, p. 251. 53 Described above p. 100. 54 See below, p. 114. 55 See below, p. 114. 56 See p. 113. 57 See below, p. 117. 58 The relationship of the forms with retained original !la and ja to those with

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

113

Examples of this contraction furnished by the Ethiopic language are: qanet 'service' parallel to its basic form qanjat, ramet 'blow, wound' parallel to ramjat, maset 'evening' parallel to masjat, Jatot 'desire' parallel to Jat/jat, 'atot 'income' parallel to 'alljat, 59 m~or 'lectica' parallel to ma.~/jar, 60 etc. Old South Arabic apparently also contained such forms ending in -ot, as testified to by the occurrence of !fIt as well as !f1/jt. 61 This would correspond to the occurrence in Ethiopic of accessory forms with -ot < -/jato Aramaic instances which come into this category are !felOlii 'prayer', Syr. m eMlii 'blow', Biblical Aramaic bazolii. 62 In North Arabic, the l:Iigaz dialect contains forms with ' alif at-taJbim, i.e., 0 or at least a sound close to 0 (written /j), which are to be mentioned in this connection. According to the records handed down to us by the grammarians,63 these forms are !falot, ua, ia (> 0, e, see p. 112) corresponds basically, at least with respect to the phonetic process which .has taken place, to the relationship of the Proto-Hebrew forms with aj, ay (> e, 0) to those with ai, au (> aji, ayu), see above p. 99. 59 Praetorius, .4'th. Gram., p.22 considers it possible that the 'weak' (contracted) forms are the older. Dillmann-Crichton, Eth. Gram., §40 (p. 38); Schrader, De linguae aethiopicae indole, etc., Giittingen 1860, p. II; Barth, Nomina/bi/dung, p. 390, p. 409, base their theories on 'strong' forms. However, Dillmann and Schrader and also Littmann, Deutsche Literaturzeitung, 1909, col. 3099, maintain that the contraction ya> 0, ja> e is only possible by way of the transposition ya> ay, ja> aj. The untenability of this view, phonetico-physiologically speaking, has already been shown by E. Koenig, Neue Studien iiber Schrift . ... des .4'thiop., p. 112. It must be admitted, however, that he considers the transition from ja, ya > e, 0 to be absolutely impossible and separates mesjal etc. from meset, which latter he derives from an independent basic form *mesait (as 'abajt 'magna'). The groundlessness of this requires no demonstration. 60 Dillmann, § 116. Brockelmann, Grundriss, I, § 198 b, considers the occurrence of ma$or instead of *m~iir to be due to a later formation based on the vowel of the perfect $ora. 61 Rhodokanakis, Studien zur Lexik. und Gramm. des A/tsiidarab., I, 63. 62 Barth, Nomina/bi/dung, §258 b, and Brockelmann, Syr. Gram., § 110 (until the 4th edition) consider this 0 to be contracted from ay, thereby assuming the form *$a/ayt{ii) < *$a/ayat{ii). The difficulty in this line of reasoning brought about by the fact that ay is retained in open syllables in Syriac is met by Brockelmann by postulating that the 0 comes from the status constructus form $e/01, presumably < *$alayt, by analogy. Bauer and Leander, Gramm. des Bib/.-Aram., p. 185 s', explain the 0 in /:lazo!ii (instead of the ii in qe$ii1, etc.) as borrowed from Canaanitic. Both Brockelmann and Bauer and Leander make use of the same principles which served them in the case of baji!, maye! as contrasted to jom, /:leq (see above p. 99, n. 5). It should further be mentioned that in § 110 of the 5th edition, and also in its foreword Brockelmann renounces this explanation. (See additional note, below p. 121). 63 Sibayajhi, ed. Derenbourg, II, p. 452.

114

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

zakot, /:lajot (in place of the ordinary Arabic pronunciation with _at).64 In addition to these, Qur'anic and the other writings and the rules governing the reading of Qur'anic literature 65 provide us with gadot, manot, nagot. 66 The corresponding forms in non-J:Iigazi Arabic contain an a instead of the 0 (written with' alif). Among the cases from Proto-Semitic or the earliest phases of the individual languages, I include the nouns ending in -at of the roots III jly, such as Arabic ~alat, /:lajat, nagat, etc. (the non-J:Iigazi variants of J:Iigazi, ~alOt, /:lajot, nagot, manot), gazat 'raid', Syriac m enala, ge'ala. The Arabic gazat has a parallel form gazyat (feminine of gaZlj); nagat (J:Iigazi nagot) has a parallel form nagyat. 67 Parallel to the Syriac ge'ala is the ludaeo-Aramaic ga'rla < *ga'jala, just as */:lajyala > /:lajy el a > /:laj (j)ula. I consider these last accessory forms with a post-consonantic rising diphthong, ya or ja, to be the basis of the forms with a monophthongic a in the endings -at, -ala. I place the origin of the forms ending in -at in the oldest epoch of Arabic or Aramaic, possibly even in Proto-Semitic. In the case of Aramaic, at any rate, we know with certainty that it possessed 0 and e sounds during historic time. These, however, can be proved to have been contractions of diphthongs in almost every instance. We find this also to be true of those of the other Semitic languages which have these 0 and e sounds. On the other hand, Old Arabic, except 64 Sarauw, ZA 21, 43 explains the a in these words as a contraction from aya in */:Iajayat, etc., which is phonetically as impossible as the contraction aya > a (see above p. IDS, n. 18, and below p. IDS). Barth, ZDMG 42, 344 and 44, 697, Nominalbildung, §258, considers these forms to have been borrowed from Aramaic (cf. below p. 122). The untenability of this assumption with respect to /:Iaiot is obvious (see Sarauw, l.c.). With regard to ~alot, ~alat, which Brockelmann, ZS 5, 14 also considers to be borrowed from Aramaic, see Landberg, Dalina, p. 1417, who declares it to be a word already existing in the ancient Semitic civilisation (cf.. above p. 113, line 7, concerning Old-South-Arabic). Even zakat can hardly have been borrowed from the Aramaic zalsula and then transformed on the pattern of ~alat (according to Brockelmann, l.c.). One might possibly explain the change of the ending -ut to -at (-at) on the basis of such a 'rhyme formation' CReimbildung'), but not the discrepancy between the short a in zakat and the long a in zalsula. (Cf. Aramaic !a'ula, /:Ianula, ba'ula > Arabic !agut, /:Iiinut, bagut). It is probable that to the genuine Arabic word, zakot (zakat), there has been assigned the meaning of the etymologically identical Aramaic word zalsula 'alms' (see Horovitz, Islam 8, 137). 65 The literature of the Taguid (i.e., the art of reciting the Qur'an) and the Qira'at. (See my thesis [v. above p. 102, n. 11], p. 36). 66 With regard to modern Arabic dialects, we can only refer to Landberg, Dalina, p. 296, n. 3. 67 /famasa, ed. Cairo 1927, p. 379, 1. 1.

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

115

for the above-mentioned I:Iigazi 68 and certain other forms, no longer presents any 0 and e sounds at all. 69 There is, however, no doubt that e and a sounds existed as independent phonemes in Proto-Semitic, whether long or short and whether original or the result of contraction. To be sure, the e and a-quality of these vowels did not remain preserved. Long e and a-phonemes (e, 0) were shifted into a; and this process started before the cleavage of Proto-Semitic into its various daughter-languages. This was brought about by a phonetic change which was the consequence of an alteration of the basis of articulation. The individual languages in turn took over from Proto-Semitic the peculiarity of possessing no e (e) and 0 (0) vowels. This characteristic was then most probably retained for a certain period of time, although the duration of this cannot be ascertained in any specific case. This peculiarity applies not only to those e and 0 vowels already existing at the time the change 0, e > a began regularly to take place (probably a quite late phase of Proto-Semitic), but also to all those arising subsequently in both Proto-Semitic and its daughter-languages. This remained the case until the latter languages recovered the ability to use a and e sounds (which arose by secondary processes) as independent phonemes. In the course of time, all the Semitic languages then received again into their phonetic system e and a sounds as independent phonemes. The exact moment of this process cannot be fixed for any individual language. At any rate, we know that in Old Arabic the transformation of e and 0 vowels in such cases as *mate « *mataj) > mata, *rame « *ramaj) > rama, *gazo « *gazGlj) > gaza took place during its development as a separate language (see above p. 104). In most of the older Semitic languages the ability to retain secondarily occurring e and a vowels as independent phonemes is present relatively early. 70 In Arabic, however, this becomes completely the case only in the later vulgar dialects, although even here it is imposible to establish just when it recovered these phonemes. 7 1 At any rate, e and 0 vowels must have recurred during the development of Old Arabic. However, up to the present, we have always Above, p. 113. We hereby disregard the secondary [miila pronunciation of original ii, which has not found any expression orthographically. 70 In Akkadian, this holds true only for the e. 71 The development of these dialects has, unfortunately, only recently become accessible to observation. 68

69

116

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

found them already changed to ii in this language, and their existence is to be surmised only on the basis of the reconstruction of their developmental sequence. There are, however, other cases, limited to certain dialects, in which we have positive proof of the occurrence of an 0 vowel as the result of contraction. We refer here to the already mentioned l:Iigazi forms with 'alif at-taf!Jim (see above p. 113). This 0 vowel was a living sound at the time of the Old Arab grammarians. We are unable to determine whether these forms with 0 survived in later dialects 72 or were changed to ii in a comparatively early phase. The l:Iigazi forms mentioned above are derived, similarly to the corresponding Ethiopic forms, from precursors with a vowelless second radical and!! as the third radical: *$al!!at etc. I consider these precursors to be identical with those of the ordinary Arabic forms, :jaliit, nagiit < $alot, nagot, etc. 73 I assume that *:jal!!at, etc., could have continued to exist along with the forms resulting from it by contraction ($a[ot > $aliit, etc.) just as with the Ethiopic jan!;!at, ramjat and the forms derived from them, janot, ramet. Then, at a later date, this surviving *:jal!!at, etc. would undergo the contraction to $alot, etc. just as once before in a previous phase. This can, perhaps, most clearly be represented diagrammatically: I. II. III.

Proto-Semitic or Early Old Arabic Later phase In time of Grammarians

*:jalyat *$alot :jaliit (Common Arabic)

*:ja/!!at :jalot (l;Iigazi dialect)

If, as has been shown, it is possible for one and the same root to exist simultaneously in both a contracted and an uncontracted form, it will be obvious that tliis is much more apt to be the case with words derived from different roots. By this is meant that one word at a given phase in the development of a language can be less evolved than another built around the same pattern. Many such uncontracted See above p. 114, n. 66. These forms, $aIOt, nagat, etc. represent the stage of the I:Iigazi dialect at the time of the grammarians. 72 73

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

117

forms are found in Arabic, as ba!llat 'step', gazllat 'raid', nagyat 'rescue', along with contracted forms, such as gaziit < *gazat, nagiit < *nagat, yafiit 'death' < *llafet < *yafiat (ancient contractions), $aiat, nagat « *$aillat, nagllat, with more recent I:Iigazi contractions), also in Aramaic and Syriac: /:tajll e!ii ('animal') « */:tajlla!ii, as a result of the normal phonetic shifting of the full vowel of the syllable preceding the accent into a reduced vowel; cf. the st. constr. /:tajlla!) > /:taj (f)u!ii, Jewish-Aramaic ga'nii 'outcry' « *ga'ja!ii) beside menii!ii, Syriac ge'ii!a « *ge'e!a with early Semitic contraction), $eli5!a « *$ailla!a, with a recent Aramaic contraction). The situation in Ethiopic in this connection forms a problem in itself. Apparently, neither Barth nor Brockelmann have been able to find any parallels in Ethiopic to the Arabic and Aramaic forms gazat, llafat; mena!a, ge'a!a, etc. (With regard to Hebrew, cf. Brockelmann, I.e., § 132 e). The termination -at actually does seem to be absent from Ethiopic. We find only 'strong' terminations, such as -llat, -jat or contractions of these, such as -at, -et. It will thus be evident that these -at and -et are not products of that ancient era in which a, e (themselves contractions) were changed to a. Rather they are products of a later date, in which those a and e sounds were retained which were contractions of lla, ja. It is unimportant whether we here assume that -at forms were at one time also present in Ethiopic, and were then, by a secondary process, replaced by 'strong' un contracted forms 74 which were later contracted, or whether we assume that there were present in Ethiopic during a considerable period of time III jly 'strong' nouns with feminine endings, surviving from earliest times. Brockelmann's 7S thesis that gazat is derived from *gazayat finds its counterpart in Sarauw's 76 that I:Iigazi bajat is derived from */:tajayat. However, aya > a is just as difficult phonetically as aya > a. aya cannot be contracted to either a or a, but a can be derived from lla. This is the explanation for I:Iigazi /:tajat < *bajllat. Since it cannot be denied, however, that a genetical relationship exists between I:Iigazi bajat and non-I:Iigazi baja!, Common Arabic gaza!, we must postulate for the latter two, also, a basic form corresponding to that of I:Iigazi baja!. In my opinion, lja is no more capable of resulting 74 Possibly as the result of an analogy-efTect of the many uncontracted forms handed down from Proto-Semitic. 7 S Grundriss, J, § 132 e. 76 ZA 21, 43.

118

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

in a than is aya, its contraction leading only to o. It must, therefore, be assumed that between the forms bajat, gazat and their basic forms *bajyat, gazyat,77 there existed the intermediate forms *bajot 78 and *gazot. The 0 of these latter then shifted into a. The same holds true for nouns of the roots III j, such as yafat, which arise from basic forms such as *yafjat by way of *yajet. Brockelmann's thesis that *fa'ayat, *fa'ajat are to be considered as the basic forms for these latter nouns is based on a mere assumption. 79 No one has ever succeeded in proving in the case of a single fa'at form that this supposed prototype actually existed. The postulation of such basic forms as *fa'yat, *fa'jat is certainly at least as well founded and, in fact, is strongly suggested by the Aramaic baj (j)uJa < bajyeJa 80 corresponding to Arabic bajat, bajot and to Ethiopic bajyat 81 with postconsonantic ya. Furthermore, in the case of the word $alat,82 many of the Arab grammarians do not designate *$alayat as the basic form of $alat, $alot, but rather *$alyat, the form assumed by me. An unquestionable example of the contraction ja > e exists furthermore in Neo-Syriac. This is found in the terminal syllable in such cases as lajle < *lajlja 'night', 'ure < Old Syriac 'urja 'manger', suse < susja 'horse'. These exist simultaneously with such forms as gegja 'ram'.83 The Ethiopic accusative and status constructus form with e also belongs here. An example of this is ba'ase, the accusative and status constructus form of ba'asi, in place of ba'asaja,84 which has been found in the Aksum inscriptions. 85 Still in existence. Still occurring in the 1:Iigazi dialect. 79 Barth, Nomina/hi/dung, p.91, n. 1, has attempted to derive the alleged basic form *ja'ayat from plural forms such as ~a/ayiit. The Arab grammarians are also of this opinion (see Lane, Diction., p.2234c, s.v. gadiit). This argument becomes void, however, when we take into consideration the fact that, in all forms constructed on the pattern qat/at, an a is inserted after the second radical in the plural. 80 Also signifying 'animal'. 81 In place of *I:lail!at, but having the usual a of the nomina verbi in Ethiopic instead of the a of Arabic. 82 According to Lane, s. V. 83 Noldeke, Neusyrische Grammatik, p. 55. 84 As would be expected, Dillmann, §143, considers this e to have come from 1a in the same manner as he considers the e in qanet to have come from ja in qanjat (compare above p. 113, n. 59). Praetorius (Gramm., p. Ill) considers it questionable altogether whether this e is derived from ia. Brockelmann (Grundriss, I, § 70 77

78

THE DEVELOPMENT OF SEMITIC DfPHTHONGS

119

This same contraction, ja> e, as it occurs in Neo-Syriac, serves also to explain the plural ending e of the status emphaticus in various Old Aramaic dialects, especially in East Aramaic. The corresponding ending in most West Aramaic dialects is ajja. In this latter ending, one can recognize the ending aj of the plural status constructus plus the sign of the status emphaticus, a. 86 This combination of aj + a must originally have resulted in *aja. In order to explain the ending ajja, we must take into consideration that in East Aramaic, in which it occurs, it is associated with biradical nouns or stems III j. Instances of the first are benajja 'sons', senajja 'years' and of the latter, qenajja, the plural of qanja 'reed', and qesajja, the plural of qasja 'hard'.87 !f we assume that the ending in question was originally *aja, we are forced to the conclusion that qenajja is derived from a basic form *qanajaja. This must then have become qenajja as the result of the haplological ellipsis of a syllable. 88 In this case the third radical of the stem was indissolubly fused with the ending, which thus appeared as ajja (in contrast to the original *aja). From then on only the first two radicals could be clearly distinguished from it. We have here to all intents and purposes biradical stems, with the ending ajja. This ending was then also apposed to those which are biradical in the singular. Examples of this are benajja, senajja, *fa/araja > fa/are. "After the loss of the a, aj became e". Why, however, should the sign of the accusative (and st. constr.) disappear if it follows j, when it is otherwise always retained? One might, of course, point to the pronominal suffix of the 1st person singular i instead of ija. That this is not an instance of a regular elimination or a regular contraction of unaccented terminal fa to i (sic Brockelmann, Grundriss, I, § 70b) is shown by the fact that the accusative a is retained in such cases as Arabic al-qiit/ija. Obviously, these cases are closely analogous to *ba'asia> ba'ase. (The relationship between the pronominal suffix i and its antecedent ija has not yet been cleared up and, unfortunately, cannot be discussed here). 85 Littmann, Aksum-Expedition, vol. IV, no. 9, 10, 11. 86 Brockelmann, Grundriss, I, § 242d. 87 Brockelmann, Grundriss, I, §242e. 88 Brockelmann, Grundriss, I, §97, 1,2, postulates basic forms such as *qenajajjii, with the ending ajja, whose j is supposed to have been doubled at the time of its apposition to the stem. He explains this doubling by the occurrence in ProtoAramaic of a phonetic process which has otherwise been found only in modern dialects, especially in modern Arabic (I.c., §40i). As examples of this in Old Aramaic he can bring only the West Syriac plural ending tl!!yala, as opposed to the East Syriac iiyala. However, it has not yet been settled which of these forms is the original. Furthermore, in ayala, in contrast to the modern cases, the consonantic vowel is preceded by a long sonant. Since haplology explains the duplication in the ending, ajja, quite sufficiently, Brockelmann's interpretation can certainly be dispensed with. (For further information on this type of haplology see above p. 51 ff.).

120

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

which, as a matter of fact, are in part originally derived from stems III j. Now, in the case of the nouns of 'strong' stems, the plural in Proto-Aramaic must have resembled *rnalkaja (or *rnalalsaja), the plural of rnalka 'king'. In accordance with the usual Aramaic phonetic laws, this *rnalkaja must have become *rnalkeja or *rnalkja. If, later, this terminal ja was contracted to e, we should have rnalke. Thus, e represents originally the ending of nouns of 'strong' stems and ajia that of nouns of stems III j and the biradicals. This is the situation which we find, for instance, in Syriac. In West Aramaic and particularly in Biblical-Aramaic, on the other hand, we find the ending ajja to be practically universal. This can readily be explained. As already indicated, the fusion between the third radical, i, and the ending results in a new ending, ajja, on a stem reduced to two radicals. Since one would no longer have been conscious of the developmental history of this ending, it could then have been joined to ordinary triradical stems. However, a last trace of the old state of affairs can still be found in Biblical-Aramaic, in that the ending aj of relative nouns is still followed by e (-aje) for phonetico-psychological reasons (tendency to dissimilation). In the dialect of Palmyra we even find this after several ordinary nouns of 'strong' stems. 89 It will, thus, be evident that in these cases of the ending e in West Aramaic we are not dealing with an innovation, but with remnants of an earlier condition. 90 In Mandaic, we find that the plural ending e (i) is written with K", (,91 just as every terminal e or i. 92 I assume that this manner of writing originates in the derivation of these terminal e sounds from the contraction of ja. In both Neo-Syriac and Early Aramaic, the ending ia was contracted to e, but in the case of Mandaic the phonemes from which this e is derived continued to be used when writing. Subsequently, the graphic expression of this particular e was applied to Compare Noldeke, Beitriige, p. 49. For other opinions on the endings, e, alia, I refer the reader to Praetorius, ZDMG 56, 685; Noldeke, Beitriige, p.49; Barth, ZDMG 58, 436; Brocke1mann, Grundriss, I, §242e and Cantineau, Bulletin d'Etudes orientales de l'Institut franc;ais de Damas, I (1931), p.92. Barth, proceeding from an ending i-a (instead of ai-a), refers to the Neo-Syriac contraction ia > e, but considers it inadmissible to assume the same contraction for Old Aramaic. His reason for this is not clear to me. In accordance with this, Barth has decided in favor of the common theory that the e in question is identical with the Akkadian masculine plural ending e. As for the rest, I am, in common with Noldeke, opposed to this and all similar opinions. 91 Rarely, with 'ain, see Noldeke, Beitriige, p. 50. 92 In contrast to medial e, i and initial e, i which appear as i and 'ain, respectively. 89

90

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

121

all terminal e sounds, no matter what their origin. Other terminal sequences ia also, of origin and functions other than the plural ending of the status emphaticus, may have been contracted to e, and thus contributed to the uniform writing of all terminal e's and i's as N", j'. At all events, it seems to me that any explanation of the fact that terminal e (i) is written as N", j' should be made in the light of the phonetical development of the language. ADDENDUM

As far as the Aramaic nouns ending in -ot, as, e.g., $eloJ{a) 'prayer', are concerned, Brockelmann, in later editions of his Syrische Grammatik, abandoned his original explanation of this structure (see above p. 113): -ayat > -ot (*$alayat > $e 101). Beginning with the 8th ed. (I refer here to the 10th ed., 1970, § 54, p. 33), Brockelmann traces $elola to a form *$elaljta (which presupposes a primary form *$aliiljat). That means: he assumes a contraction of the 'long diphthong' iilj to 0, that is: *$ elaljta > $elola. Such a form qefalt(a) < qatalat, is, as far as the word for 'prayer' is concerned, unattested. But the assumption that a sequence iilj (with a long vowel, a, preceding the semi-vowel) might have been contracted to 0 is basically unacceptable. An analogous phonetic sequence, with j in the place of Ij, is present in a characteristic Syriac form, that is: the feminine-form of the nomenrelationis with -aj (serving as adverb), which is characterized by the ending a'il (e.g., allaha'il 'divinely' < *allahiijt). That means: the 'long diphthong' -iii was not contracted, but rather was dissolved into two syllables (by the development of the semi-vowel j into the (long) vowel i). Another possible development of a 'long diphthong' (in an analogous syllable structure) would be the elimination of the semi-vowel (i.e., the consonantic element of the diphthong), thus: -aj- > -a-, and -alj- > -a-. As far as the 'long diphthong' -aj is concerned, this kind of development - that is: -aj- > -a- - is especially attested for certain ancient Iranian and Indian dialects (see, e.g., W. Eilers in A locust's leg; Studies in honor of H. Taqizadeh [London 1962], p. 80 ff.). However, also in Semitic dialects we find not rarely this development, that is: the development of aj and iilj to a; and also the 'short diphthongs' (i.e., the regular diphthongs), that is: aj and ay, may in Semitic dialects develop into the (short) vowel a (and more rarely into the long vowel [aD. (In the present context, we can, unfortunately, not discuss the details of the phonetic process effective

122

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

in this type of 'monophthongization', nor can we mention some of the relatively numerous Semitic instances for which this process is to be assumed). In any case, a contraction of the 'long diphthong' iiJj (> a), as suggested by Brockelmann (*$alayta> $elala; see above), is not attested and can not be assumed. Concerning this terminal syllable -at in Aramaic $elaj( a), (l:Iigazi-) Arabic $alat-, and in various other Aramaic and (l:Iigazi-)Arabic nouns (see our discussion above p. 113ff.), some scholars have voiced the opinion that the vowel a in this syllable should be conceived as based on a shift of an original vowel a into a. I refer especially to Ch. Rabin, Ancient West Arabian (London 1951), p. 104 ff., and to F. Altheim (in collaboration with R. Stiehl), Die Araber in der alten Welt, vol. 1 (Berlin 1964), p. 363 ff. Moreover, both of these authors assume that this alleged shift a > a originally took place only in one of the two languages concerned (Aramaic and Arabic) and that the occurrence of the same phenomenon in the other language is based on its transfer from the language in which it actually developed. Rabin, I.e., p. 106, assumes that the a> a-shift was transferred from the l:IigazI-Arabic dialect to West-Aramaic and to the Syriac dialect of Edessa (cf. Altheim, I.e., p.365). However, according to Altheim (Ibid.), a West-Aramaic pronunciation of 0 instead of a invaded the Edessean dialect of Syriac, as well as certain other Aramaic dialects, and also I:Iigazl-Arabic. As far as I:Iigazl-Arabic is concerned, Altheim states that the (alleged) West-Aramaic pronunciation of a as a does not only appear in certain Aramaic loan-words-he mentions as such: $alah, zakiih, (wjah - but also in two indigenous (Arabic) words, nagah (,deliverance, escape') and gadah ('morning'). He assumes that the pronunciation of these two latter words as nagah (nagat-) , gadah (gadot-) is based on the analogy of the words considered by him to be loan-words (" ... in iiusserlicher Anlehnung an eine Gruppe von Lehnworter"). We certainly cannot assume that Arabic /:tajat(in I:IigazI: /:tajat-, see above p. 114) represents a loan-word from Aramaic (where the noun appears as /:tajjii.la, see below p. 178). Furthermore, the theory of the transfer of the terminal syllable -at from one of the two language-branches concerned to the other - be it from Aramaic to l:Iigazi-Arabic (cf. also above p. 114, n. 64), or, inversely, from 'West-Arabian' to Aramaic-is unacceptable. Moreover, the vowel a or a (of the terminal syllables -ot or -at, respectively) in the words concerned cannot be based on an original a-vowel, but must-in accordance with the roots underlying the nouns concerned

THE DEVELOPMENT OF SEMITIC DIPHTHONGS

123

- be traced back to a diphthongic sequence. We doubtless deal here with the contraction of an original (post-consonantic) 'rising' diphthong Ija to jj (in part-as far as Arabic instances are concerned-secondarily shifted to a), as out-lined in our present discussion (see above p. 113).

4 BI-CONSONANTAL NOUNS OF ROOTS III W ('AB, 'AJj,

IfAM)

The so-called bi-consonantal (monosyllabic) nouns of Semitic to be discussed are the words for 'father', 'brother', 'sister', 'father-inlaw', 'mother-in-law', as represented by Hebrew and Arabic '00 'father'; 'ab 'brother', Akkadian 'abatu, Hebrew 'aMi « 'abat), Arabic 'uljt 'sister'; Hebrew ~am, Arabic ~am, ~amlj (~amw), ~amu 'father-in-law'; Arabic ~amat, Hebrew ~amoi « ~amat) 'mother-inlaw'. The original forms out of which these forms developed, are evidently not yet fully clarified. In An Introduction to the Comparative Grammar of the Semitic Languages, ed. by S. Moscati (1964), p. 83-84, the noun 'ab 'brother', e.g., is presented as an instance of a "bi-consonantal pattern, with a short vowel", built on the form qab. This conception is in agreement with Brockelmann's treatment of these nouns, in his Grundriss, I, p. 331 ff. For the sake of clarity, we quote some of Brockelmann's statements in extenso (bid., p. 331; § 115 b,a): "Die Verwandschaftsworter 'ab 'Vater', 'alj 'Bruder', ~am 'Schwiegervater' und bin 'Sohn' sind ohne Zweifel als von Hause aus zweiradikalig anzusehen. Bei den ersteren drei [i.e., ' ab, ' alj, ~am, with which we are dealing here] wird der Anschluss an die dreiradikaligen Bildungen dadurch erreicht, dass sie die urspriinglich langen, aber im freien Auslaut sonst verkiirzten Kasusendungen (s. §42 g) im St. cstr. und vor Suffixen lang erhalten. Zu 'eljuka (s. §66 dy) bildet das Ath. auch 'eljv, 'eljva, 'eljlla mit Labialisierung (s. §45 g,a) .... ". We also quote the following statement from Brockelmann's discussion of the word for 'sister' (Ibid., ~): "Wie der fehlende 3. Radikal beim Ms. durch. die langen Kasusendungen ersetzt wird, so beim Fern. zu ' alj durch Dehnung der Femininendung at > at (Wellhausen, Sk. u. Vorar. VI, 258), > 'abat 'Schwester' .... ". As we already pointed out in another context (see Le Monde oriental, vol. 32 (1938), p. 99, n. 3), lengthening of terminal vowels-and especially of declension vowels-cannot be considered a substitute for a third consonant. Although we do not know what the exact forms of these nouns were at the time when they first came into existence, we do assume that these nouns, as they now appear, are

BI-CONSONANT AL NOUNS OF ROOTS III W

125

directly based on more original tri-consonantal structures in which the third consonant was constituted by y (w). We regard this original y (w) contained in the long vowel u by which these nouns are characterized when used in the construct and when provided with possessive suffixes, as, e.g., (Arabic) 'abu, ' abuhu, etc. Accordingly, we derive 'aou, 'abuhu (etc.) from *'aoyu, *'abyuhu (etc.). These latter forms developed into 'abu, 'abuhu (etc.) by 'sonantisation' of the original consonant y (w). It may be assumed that this 'sonantisation' could easily occur under certain syntax-phonetical factors (cf. above p. 100). As soon as this development of the original forms of the nominative, *, abyu, *, aoyuhu, into 'abu, ' abuhu had taken place, the forms of the genitive and the accusative (originally: *, aby i, *'abyihi; *'abya, *'aoyahu) necessarily followed the analogy of these (secondarily developed) nominative forms, and thus appeared as ,abi, 'abihi; 'aba, ' aoahu. In this connection it should not be overlooked that in the case of the word for 'father-in-law': bam, Arabic possesses a second, tri-consonantal form: bamy (bamw), with suffixes: bamyuhu (bamwuhu), etc. These latter forms should be considered as the direct precursors of the more frequently occurring forms bam (u), bamuhu, etc. Brockelmann's explanation (see above p. 124) of the long vowel in ' abat 'sister', bamat 'mother-in-law' as lengthened from an original short vowel for the purpose of making these nouns join the pattern of tri-consonantal formations is no more acceptable than in the case of the masculine nouns discussed above. We derive ' aoat, bamat from the basic forms *'aluJat, bamyat (*'abwat, bamwat), and consider the original third consonant y (w) preserved in the plural-forms, e.g., Arabic 'aoayat ('sisters'), etc., which may be regarded as the normal plural-form of a noun fern. sing. of the pattern qat/at (hence *'abyat). Such a form' abayat as the original plural-form of the word for 'sister' is strongly suggested by the Old-Akkadian plural-form aouatu (pI. of abatu), as against the later form aooatu. The u in abuatu obviously represents the original third consonant y (w) of the singular *'aoyat (or is a 'sonantisation' thereof). If we derive· femine nouns like Akkadian abat, Aramaic ('a )baia, Hebrew 'aMi « *, aoat) 'sister', Arabic bamat, Hebrew bamoi « *bamat) from original forms with a third consonant y (w), that is: from the primary forms *' abyat, *bamyat, we have to show why the third consonant, y, is no longer preserved.

126

BI-CONSONANTAL NOUNS OF ROOTS III W

We identify the long vowel a in ' aoat, bamat (> Hebrew ' aMl, bamol) with the long vowel a in, e.g., Arabic bajat 'life' and analogous nominal forms of Semitic (see above p. 112-118). Forms as Arabic bajat are traced back by us to original forms like *bajyat; see above p. 114ff. With respect to this we should also take into consideration the Ethiopic term for 'life': bajyat (see above p. 118, n. 81). We assume that in basic forms of this type (i.e., of the pattern fa'yat) the original post-consonantic 'rising' diphthong ya was apt to be contracted into the monophthong 0: *bajyat> *bajot. We refer in connection herewith to the form bajot (= bajat) and analogous forms attested by the early Arab grammarians for the I:IigazI dialect of their time (pronounced [in the terminology of these grammarians] with alif at-tafMm, to be identified with 0). This I:IigazI form, bajot, derived by us directly from a basic form *bajyat, is to be considered developed by the recurrence in historical time of a process which already had taken place in proto-Arabic (and proto-Semitic) time. However, while in the I:IigazI forms observed by the grammarians the original vowel 0 (developed from the diphthong ya) remained preserved (at least until the time of the grammarians), in the analogous forms developed in earlier developmental stages (in protoSemitic or in, e.g., proto-Arabic) the vowel 0 was shifted into a: bajot> bajat. The phonetic system of proto-Arabic (exactly as that of proto-Semitic, or, in any case, of 'late' proto-Semitic) did obviously not include the vowels 0 and e (short or long). Whenever, during these developmental stages, (long) 0 or e vowels, i.e., 0 or e, developed through specific phonetic processes, they were shifted into a. In agreement with this development of Arabic bajat (and similar Semitic forms) from *bajyat (etc.) via the intermediate stage *bajot (etc.), we derive Akkadian aoat, Hebrew 'aMl « *'aoat), Aramaic 'abala (Syriac bala) 'sister', as well as the analogous Arabic bamat, Hebrew biimol (etc.) 'mother-in-law', from the primary forms *'abyat, *bamyat. These latter forms developed into *' abot, *bamot, forms in which the vowel 0 was finally shifted into a: ' abat, bamat. The original third consonant y is, however, to be considered integrally preserved in the plural-forms: Arabic ' abayat, Old-Akkadian abuat; and Arabic bamayat. In this context we should also not overlook the plural-forms of the word for 'brother' in Arabic: ' ioyat and 'ioyan, in which y as third consonant is in turn preserved (compare 'ioyan with, e.g., 'ibdan, pI. of 'abd; and 'iOljat with, e.g., filiat, pI. of fata(n) ).

BI-CONSONANTAL NOUNS OF ROOTS III W

127

In connection with our derivation of the North- and East-Semitic form of the word for 'sister': 'abat (> 'aMI in Hebrew) and of the wonl for 'mother-in-law': I)amat (> I)amal in Hebrew) from the primitive forms *'abyat, *I)amyat, with a post-consonantic diphthong ya, we refer also to certain other parallels, apart from I)ajat, etc. (see above). We think especially of Arabic nouns (of roots III y) like gazat 'raid', gadat 'morning'. Beside gazat (with pI. gazayat, cf. 'abayat), we have a synonymous noun: gazyat (pI. gazayat), which may be conceived as a so-called nomen unitatis of gazy. Analogously, beside gadat (pI. gadayat) 'morning', there exists a form gadJjat (pI. gadayat), and, indeed, also a form gudJjat. We consider it necessary to bring these coexisting synonymous noun forms derived from the same roots into a genetical connection. That is: Arabic gazat and gadat should be derived from Arabic gazwat and gadwat, respectively. As intermediate forms in this development we should assume the forms *gazot and *gadot, respectively. Concerning the development ya > 0 and the further development 0 > a see above p. 115. The development of Arabic forms like gazat, gadat (plural gazayat, gadayat) from gazyat, gadJjat is thus a clear parallel to our derivation of 'abat, I)amat from the primary forms *'abyat, *I)amyat. In connection with our thesis that the Semitic words for 'brother' and 'sister' are based on primary forms which included a third consonant, y, and more specifically: based on the forms *, aby and *, abyat, respectively, we should certainly pay attention to the SouthSemitic form of the word for 'sister': Arabic 'ubt> Ethiopic '(Jbt. Brockelmann (l.e., p. 332) describes the u of 'ubt as 'unexplained' ("des sen u aber noch nieht erkliirt ist"). We are of the opinion that this u clearly hints to the existence of y as a third consonant in the primary form of the word for 'sister'; in other words: hints to an original form *' abyat, a form to which we were also led by the North- and East-Semitic term for 'sister', i.e., ' abat. Accordingly, the vowel u in 'ubt is a reflex of the original third consonant y in the primary form *, abyat. An u-sound, clearly of secondary character, can also be observed in other nouns based on roots with a third radical y. We refer, in connection herewith, to Arabic gudJjat 'early part of the day, morning', beside gadJjat and gadat « gadJjat, see above p. 112). It should be recognized that gudJjat is developed from gadJjat. Moreover, we compare 'ubt 'sister' (with the secondary form of the feminine-ending: -t, like in bint 'daughter', beside ibnat) with fugat 'speech'. We assume

128

BI-CONSONANT AL NOUNS OF ROOTS III W

that lugat is developed from an original *laglJ at, based on laglJ, the infinitive of lagii, jalgft 'to speak'. Analogously, we should trace back 'ub(a)t (> 'ubt) to an original form *'ablJat (which, through a different kind of development, changed into *'abot> 'abiit). As far as the development of the vowel u in 'ubt < *' ablJat, fugat < *laglJat and gudlJat < gad1Jat is concerned, as well as with respect to the simultaneous disappearance of the original IJ in 'ubt < *, ablJat and lugat < *lagyat, we assume that these features are ultimately caused by a phonetic factor which is to be considered as of basic importance also with respect to the monophthongization (contraction) of the original 'rising' diphthongs lJa and fa> 0 and e, respectively, as in (e.g.) *'ablJat> *'abOt (> 'abiit), etc. (mentioned above p. 125). This phonetic factor consists in the fact that diphthongs are very sensitive structures. The sensitive character of diphthongic combinations (ay, aj; lJa, ja) can easily lead to the 'sonantization' of the consonants (,semi-vowels') IJ and j, so that the original diphthongic (monosyllabic) combinations are transformed into bisyllabic combinations. That is: ay and aj can easily develop into a-u, a-i, respectively. Such bisyllabic combinations as u-a and i-a may ultimately (re)develop into monosyllabic (diphthongic) combinations of a different type, that is: they may develop into the diphthongic combinations uq and iq, respectively (with q representing the consonantic element of the diphthong). These latter, diphthongic (monosyllabic) combinations uq and iq may again be contracted into the monophthongs o and e, respectively. An instance in point is *'ablJat> *'abuat > *, abuqt > *, abot (> ' abiit). We refer to the theory concerning this and related processes, as well as for a number of individual instances to above p. 114 ff. The easily occurring sonantization of the consonantal element in a diphthong means, with respect to the forms here under discussion, that original forms like *, ablJat, *laglJat, gad1Jat could develop into *' abuat, *laguat, *gaduat. It may be assumed that these latter, trisyllabic structures (based on original bisyllabic structures) were originally uttered with a 'hovering' stress, that means: the stress was equally distributed over all three syllables of which the words consisted. Through certain syntax-phonetic factors it can, however, easily occur that such an originally 'hovering' stress (equally distributed on all syllables of the word) abruptly concentrates (quasi: jumps) on a specific syllable (vowel), a process through which the other syllables of the word will be deprived of their amount of intensity (stress) to

BI-CONSONANTAL NOUNS OF ROOTS III W

129

a large extent, and, as a consequence hereof, their phonic strength may be utterly weakened; that is: they may be reduced to 'mumured' vowels of unclear vowel quality (or colour), or may disappear entirely. We thus assume that in forms like *'aI)uat « *'aI)yat), *gaduat « gadJjat), *faguat « *fagyat), the originally 'hovering' stress concentrated at first abruptly on the sonant u, a process which necessarily caused the weakening of the vowel a of the first syllable. Subsequently, to strengthen this weakened, or virtually destroyed, vowel of the first syllable-or, in other words: in order to preserve the first syllable (originally characterized by the vowel a}-the stress retreated abruptly (from the second syllable, i.e., from the vowel u < y) onto the first syllable, which, though extremely weakened or no longer existing, survived in the linguistic consciousness and, in any case, was indicated by the consonants (surrounding the original, now entirely weakened vowel). Since the original vowel (which, in the instances in point, was the vowel a) was entirely, or almost entirely, destroyed, a new vowel had to develop (in other words: to be interjected) to serve as the 'carrier' of this new stress. The colour (or: quality) of this new vowel - the colour of the original vowel had disappeared from the consciousness - was determined by the vowel of the subsequent, second syllable of the word (from which the stress had just been withdrawn by its abrupt retreat onto the first syllable). This vowel of the second syllable was u (developed from an original y). Accordingly, the vowel of the first syllable, that became the 'carrier' of the main stress of the word, was determined as u (see above). At the same time, by the abrupt retreat of the stress from the second syllable, its vowel itself underwent reduction. It, i.e., (in the instances in point) the vowel u, was either reduced into a 'glide', i.e., y (the consonantal variant of the 'sonant' u), or it disappeared entirely. The first type of development is represented by the form gudJjat « g-duat < gadyat); the second type is represented by the form fugat « f-guat < *fagyat). By the same kind of development, more specifically by the type of development represented by the form fugat « *fagyat), we derive the form 'ubt from *'aI)yat. The form 'ubt is different from fugat only insofar as its feminine-ending appears not in the (original) form at, but in the (secondary) form t; cf., e.g., bint 'daughter', beside (i )bnat, fern. to ibn (Hebrew bin/ben). We have now to discuss the masculine nouns 'abu(n) , 'aI)u(n) , bamu(n). As already described (above p.125), the long vowel u which these words exhibit before suffixes and in the status constructus,

130

BI-CONSONANT AL NOUNS OF ROOTS III W

is to be conceived as a contraction of a sound sequence yu. It is clear that in accordance herewith also the forms without suffixes must be traced back to the original forms *' aby, *' aby, bll1111j. Indeed, in the case of the word for 'father-in-law' this form is still attested: Arabic bll1111jun and bamlju r-raguli (see Lane, Arabic diction., s.v.), Egyptian-Arabic bam" (see Le Monde oriental, vol. 19, p. 112). Instances, which are to be added here, are, e.g., gadu(n) 'morning', 'alu 'above' (beside gadyun, gudyat-, and 'alyun, respectively). Originally, this type of nouns, that is: (al-), ab-, (al-), ab-, (al- )bam-, etc., must have ended in the status absolutus (without nunation) in a long vowel il (exactly as in the status constructus and before suffixes). That means: we would expect forms like *al-' aM instead of al-' abu (etc.). I consider here the long vowel shortened in accordance with the rule in Brockelmann, Grundriss, I, § 42 g, h, th;lt is: the rule concerning the 'anceps' -character of terminal long vowels: they may be shortened. Hence *al-' abil could (sporadically) develop into al-' abu. Medially, before suffixes, the length was necessarily maintained, as well as in the status constructus to which the rules concerning terminal sounds are not applied. The shortness of the vowel preceding a consonant, as in ' abun (instead of *, aMn) is accordingly to be identified with the shortness of the vowel in instances like ramal, fern. of ramii ('he threw'), qiirjin ('a judge'), form with 'nunation' to qiirji; etc. According to Brockelmann (Grundriss, I, §271, A, Anm.), gadun (developed from gadyun) has produced (through analogy) the forms gadin, gadan (instead of gadyin, gadyan). I apply the same principle of development to 'abi(n), 'aba(n); 'abihi, 'abiihu, etc., which, accordingly, are to be considered as formed on the basis of 'abu(n), ,abilhu. Consequently, the basic forms of this category of nouns are: *'aby, *'aby, bll1111j (which indeed is recorded)-and thus not *' abayun, etc., as the indigenous, Arab grammarians had assumed. Finally, we mention that also Ethiopic ' aj 'mouth' - with suffixes: 'ajilhu, 'ajiihu-is developed from an original form *'ajy. For the details of the development of' aj (as well as of the development of the corresponding forms of the other Semitic languages) we refer to above p. 92 ff.

5 A CASE OF QUANTITATIVE ABLAUT IN SEMITIC Broken Plurals and Related Forms

1

In the introduction to his Die Nominalbi/dung in den semitischen Sprachen, p. XII-XIII, J. Barth established a law called by him "law of compensation" ("Kompensationsgesetz"), which he considers fundamental to the formation of nouns in Semitic generally: "Statt einer Vokaldehnung (i.e., of the second syllable of the perfect stem), durch welche Abstracta und Infinitive zu Stande kommen (according to Barth's general theory of the derivation of nouns), kann auch die Form ungedehnt bleiben, muss dann aber notwendig die FemininEndung annehmen". Barth refers inter alia to the infinitives of the geminating conjugation in Arabic, taqtil: taqtilat, to infinitives of quadriliteral roots, such as qalqiil, dahdiiq: qalqalat, dahdaqat, and to Hebrew forms such as 'iisif: 'asejii « *'asifat), $iinif: $enejii « *$anifat), etc. The ancient native grammarians already recognised, with regard to some particular patterns, that nouns containing a long vowel in the second syllable exist beside others containing a short vowel and terminating in the feminine suffix. But Barth has been the first to trace this alternation through the whole field of Semitic noun-formation and to compare the alternative forms in each single case, calling them "lengthened forms" ("Dehnungsformen") and "compensatives", respectively. Barth's principle was accepted by Brockelmann and applied by him to noun-formation generally .in his Grundriss der vergleichenden Grammatik der semitischen Sprachen, I, § 131 (p. 344)ff. Barth assumes such 'compensation' not only where both corresponding forms exist side by side, but also in cases in which we know only the form called by him 'compensative', as e.g., those primitive nouns of the pattern qatalat which he derives from the perfect stem, as $adaqat, barakat, etc. In accordance with his law he comments on the 'com1 A revised version of a study published in Orientalia (Rome), N.S., vol. 22 (1953), p. 1-23. - The principle set forth in this study was originally pointed out by me -in connection with the forms taqtilat; qatalat, qatalt (cf. below p. 140)-in Kirjath Sepher (Hebrew periodical, publ. in Jerusalem), vol. 17 (1940), p. 373.

132

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

pensatives' as follows: "Es durfen also so1che Formen nieht nach ihrem ausseren Erscheinen als einfache Feminina eines Masc. ~adaq ... usw., angesehen werden); vielmehr sind sie von vornherein mit der fUr sie wesentliehen und charakteristischen Femininendung gebildet, und, was sehr wichtig ist, sie werden wie die Dehnungsinfinitive auch durch die vermehrten Konjugationen, aber stets mit dieser Femininendung versehen, hindurchgebildet". Barth evidently supposes that the lenghthening of the vowel and the addition of the feminine ending are equivalent modes of formation, applied simultaneously and independently of each other from the time when, according to Barth's theory, abstract nouns and infinitives arose from the perfect stem. 2 Thus, according to Barth, there exists no direct, genetic relation between the long vowel in forms without feminine ending and the short vowel in forms with such ending. 3 Brockelmann (Grundriss, I, §42 e; p.73), on the other hand, assumes that a noun, qatiil, was obtained from qatal-at by the elimination of the ending at, which was compensated for by the lengthening of the preceding vowel, > qatiil. 4 2 Barth consistently maintains that the lengthening of the second vowel is the main characteristic differentiating infinitives and nomina action is from the perfect, which undoubtedly has to be regarded as their base: qatiil, Akkadian qatiilu, Hebr. qiitOl, certainly originates in qatala; qattiil, Hebr. qattol, Arab. qittiil, Aram. qattiilii, in qattala; 'aqtiil, Arab. 'iqtiil, Aram. 'aqtiilii, in 'aqtala, etc. (cfr. Barth, I.e., p. XXIII). Yet nobody has so far succeeded in explaining how and why this lengthening came about. 3 Brockelmann (Grundriss, I, §42 e; p.73), failing to apprehend the above point of Barth's theory, writes: "Barth bezeichnet diese Feminina als Kompensativa, indem er die Femininendung als Ersatz flir die Dehnung ansieht.", as though Barth had derived the forms with short vowel and feminine ending from the forms with lengthened vowel without such ending. This misconception is again reflected in Brockelmann's own explanation of the relation between the two types; see the following note. 4 "Vielleicht aber war das historische Verhaltnis dieser Formenpaare das umgekehrte; die aus jetzt nicht mehr kontrollierbaren Grunden geschwundene Femininendung wird durch die Dehnung des Vokals ersetzt". To support this theory, Brockelmann refers to Streitberg's explanation (Indogerm. Forseh., 43, 305 ff.) of the lengthened grade (,Dehnstufe') in Indo-European (cf. Brugmann, Grundriss der vergl. Gramm. der idg. Sprachen, F, § 544, p. 496j, according to which the lengthening of a vowel is brought about by the loss of a short unstressed vowel in the following syllable. According to Brugmann. Streitberg's theory, which had been strongly contested by several Indo-Europeanists (v. Brugmann, Ihid.) , is not very convincing, though he describes it as noteworthy. Irrespective of the validity of this theory, the case under discussion cannot be compared with the I.-Eu. case, owing to basically different conditions. In the latter case, a short unstressed vowel at the end of the word is supposed to have been dropped; in the former, we would have to suppose the loss of an element of entirely different structure, namely of a complete suffix at, a process of phonetic diminution which Brockelmann himself regards as obscure. But even

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

133

In opposition to both Barth and Brockelmann, I derive the forms with short vowel and feminine ending (existing beside the forms with long vowel) from basic types with long vowel + feminine ending. The form taqtilat, existing beside taqtil, originates in the basic form

*taqtil-at.

To explain how the long vowel of the second syllable of a noun with feminine ending at was transformed into a short vowel, we refer to certain form-types in Ge'ez and Hebrew, for which this process has been fully established. Where the feminine ending (a)t is added to Ge'ez noun-forms of the type qatul, the long u is shortened to u (which, as well as i, is in Ethiopic transformed to a by a general process). Thus, e.g., nagast 'queen', the feminine of nagus 'king', must be derived from a basic *nagus-at. Hebrew forms as *gebirt 'mistress' (> gebhiiriit\ but with suffixes gebhirti, etc.), the feminine of gebhir 'master', go back to *gebirat (> gebhirii, which is still found as an alternative form), etc. The intermediate stages of this development are equally clear: Ge'ez *nagust > *nagust (> nagast), Hebrew *gebirt > *gebirt (> *gebhart > gebhiiriith). As is well-known, the oldest and most common feminine ending of the Semitic languages appears in two variants: at and t. It has long been accepted that these forms are genetically interrelated. Yet, formerly, opinions differed as to the application of the two variants, i.e., as to the different patterns of nouns in which each of the two forms is ordinarily used. Moreover, it has always remained doubtful which of them has to be regarded as primary, and what led to the development of the other. I consider, that each of the two variants is mainly associated with those types of nouns assigned to it by Brockelmann, Grundriss, I, p.405. Of the details stated by Brockelmann in this connection I would mention that t is particularly frequent in biliteral monosyllabic nouns and in triliteral nouns with a long vowel in the second syllable. Moreover, it can be proved that this situation had already existed in proto-Semitic. Brockelmann's theory of t having arisen out of at by the influence of stress 5 IS if such a process could be traced, there would still be no reason why the loss of a syllable should be compensated for by a lengthening of the preceding vowel. As far as I am aware, no similar process can be observed in the whole Semitic field. S See Brockelmann, Die Femininendung t im Semitischen, reprint from lahresbericht der Schlesischen Gesellschaft, Breslau 1903. - The details of these accentual conditions are still to be elucidated. In this connection, the conception of Semitic accentuation applied to similar problems in my study "The vowel i as an auxiliary vowel" (above p. 3-93) should be taken into consideration.

134

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

attacked by Barth, ZDMG 57, 628. Though some of his objections may be justified, I adhere on the whole to Brockelmann's opinion. To solve the question under discussion, we are not obliged to investigate this complicated accentual problem. For our purpose, it will be sufficient to state that in noun-patterns with a long vowel in the second syllable the ending t is most usual or even, in Ge'ez, universal (t being generally more frequent than at in this language as well as in Akkadian). Moreover, the shortening of a long vowel before a double consonant is a well-known phonetic peculiarity of Semitic, (almost) universal in Ge'ez and Arabic and apparently also in Hebrew (cf. below p. 135). Hence, the derivation of such forms as Ge'ez *nagust < *nagiisat, Hebrew *gebirt < *gebirat (Arabic *taqtilt < *taqtilat) presents no difficulty at all. Our actual objective, however, is to explain the form taqtilat (not *taqtilt) < *taqtilat. Indeed, if *taqtilt (instead of taqtilat) were the alternative form beside taqtil, no doubt could exist as to its development. But in old (North-) Arabic (in contradistinction to all other Semitic languages as well as to the modern Arabic dialects, which have arisen from the old language), forms such as *taqtzlt cannot be expected to exist. For classical Arabic has no ending t, except in some few monosyllables, as bint 'daughter', 'ubt 'sister', kiltani 'both' (fern.), lintani 'two' (fern.). Philippi (ZDMG 32, 82) supposes that in old Arabic the elimination of the vowel of the feminine ending occurred only in a few words of frequent use, Arabic having generally preserved the forms in their original vocalic integrity. According to this view, the variant of the feminine ending, t (without preceding vowel), is not proto-Semitic but an innovation of the daughter-languages of proto-Semitic. As against this, Brockelmann 6 maintains that t originates in protoSemitic, where it existed as a younger variant of at. According to his theory, the few Arabic monosyllables in t are the last remnants of the proto-Semitic situation (and not, as Philippi supposed, the beginnings of a new development). That would mean that in other (polysyllabic) nouns of Arabic the primary proto-Semitic at has been restituted. In my opinion, this restitution is due to the influence of the plural ending at, which was always used in words with the singular ending t as well as in those with at. The identity of the plural endings (e.g., 6

In the study mentioned in the preceding note.

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

135

barakiit of the sing. barakat, and taqtiliit of the sing. *taqtilt) brought about the transformation of *taqtilt into taqtilat. 7 A satisfactory connection between the alternative forms taqtilat: taqtil can only be established on the supposition that at is a substitute

for a previous t. The shortening of the long vowel in (e.g.) *taqtilt- > *taqtilt- is a consequence of the specific type of syllable structure by which *taqtiltis characterized. The long i-vowel (i) in *taqtilt- is found in strictly close contact with the subsequent sequence of two consonants, -It-, which (though normally in their turn followed by a[n inflectional] vowel) are themselves in closest contact with each other, in other words: they form a consonantal 'cluster'. Since it is difficult to preserve a long vowel preceding a sequence of two consonants (even in the case where this sequence of two consonants is followed by a vowel), the almost unavoidable consequence is the shortening of the long vowel: *taqtilt- > *taqtilt-. Cf., e.g., Hebrew geQirii « *geQiral, fern. of gel]ir 'master') 'lady, mistress', but gel]irti (Genesis, 16, v. 8) 'my mistress', geQirtiih (Ibid., v.4) 'her mistress' (in agreement with the form without terminal vowel: geQ?r?l < *geQart < *gel]irt < *geQirt [variant form of the more original form geQirii < gel]iralD. That an original long vowel cannot subsist (in the dialects concerned) in direct contact with a sequence of two consonants followed by a vowel is also clearly indicated by a variant development, which is characterized by the emergence of a 'hiatus' between the long vowel and the subsequent sequence of two consonants (see below p. 151). Of the numerous forms which may easily be explained in analogy with the case discussed (that is: taqtilat beside taqtil), I will mention a few only. There are, first of all, the forms of the pattern maqtulat 7 In the last remnants of the earlier situation, the monosyllables bint, 'uta, [intiini, kiltiini, etc., which are the feminine forms of biliteral nouns (or more exactly

-as I shall prove in another connection (cf. above p. 124-130)-of triliteral nouns which lost their third radical), the old ending has not been restituted, as t was felt to be the third consonant of the stem itself: an original ending was secondarily incorporated into the root. This is confirmed by the fact that later dialects formed diminutives or plurals from bases with t included into the root (cf. Brockelmann, Grundriss, I, § 225 A b, Anm. 2; p. 405). - (Native philologists either simply denied that t was an original feminine ending or described it as a quasi-phonetical substitute for the third radical II or j, cf. Philippi, ZDMG 32, p. 82, n. 2 and n. 3). - In this connection, we have to bear in mind that Old South-Arabic, this close relative of Classical Arabic, had t even in words other than monosyllables, as is clearly shown by the forms Ibtm < Ibntm 'brick', gflm < gfntm 'vine', with assimilation of the third radical n to the ending t (v. Brockelmann, I.c., §225 Ac; p. 405).

136

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

(Arab.), *maqtult (Hebr.-Aram.), such as Arab. ma'kulat, Hebr.Aram. *ma'kult (> Hebr. ma'akholiit\ Syr. mekhultii) 'food'. These forms may be said to originate in a feminine form of the synonymous passive participle ma'kul 'esum, edendum', 8 viz. ma'kUlat, which developed via *ma'kult> *ma'kult (Hebr.-Aram.) into ma'kulat (Arab.).9 In the same way, we explain Hebr. ,erefii 'animal torn by a wild beast', genebhii 'stolen thing', gezelii 'thing carried off by force', :;enefii 'skein', etc., as feminines of the passive participles *tarif, *ganib, *gazil, *:;anij (> :;iinij 'what is coiled around, turban'),10 thus deriving them from */arijat (comp. homomorphous, synonymous Arab. farisat), *ganibal (comp. synonymous Syr. genubhtii) , *gazilat, :;anijat. 11 We further explain 'asefii 'gathering' as a feminine to a verbal noun *'asij (> 'iisij) of identical meaning; §ereqii « *§ariqat) 'whistle' as original *§ariqat (comp. the pI. §eriqoth): *'asijat> *'asijt> *'asijt > *'asijat > 'asefii, etc. In the cases so far dealt with, there is no marked difference in function between the form with and without feminine ending. We shall now examine a case where the forms with short vowel + feminine ending have a grammatical function clearly distinct from that of the nouns with long vowel and without ending.

8 The pattern maqtiil in the function of a passive participle was once alive also in North-Semitic and is preserved in substantival use, e.g., Hebrew masliil 'track (what is trodden, beaten)" ma$piinay (Obadia 6) 'his secrets', etc. 9 It should be noted that the noun pattern maqtul appears in the great majority of cases with the feminine ending (v. Barth, Nominalbi/dung, § 168; Brockelmann, Grundriss, I, §202). We may assume that the masculine pattern maqtul-just like Arab. 'a'fun < *'a'funt (cf. Ethiop. 'a'jant) and numerous other cases (see below p. 148) - developed from the feminine form by suppression of the feminine ending, which had lost its function. 10 For qatil in the function of a passive participle in Hebrew comp., e.g., 'asir 'captive', ~am4 'what has been made sour', etc. 11 The character of these forms as passive participles (of roots of genuine activetransitive meaning) presupposes basic forms with a long vowel in the second syllable. This long vowel (qati/, qatiil, maqtii/) is an essential characteristic of the genuine passive sense in participles of (West-) Semitic languages (in contrast to ordinary adjectives, in which we may find a long vowel, a short vowel, or vowellessness). This conception of the forms in question is opposed to that of Barth, I.c., §92 a, p. 144 (see also Ibid., § 69 b ~), who explains them as infinitives or abstract nouns, and assumes the original meaning of ,ere/a, genebha to be 'Gegenstand des Zerreissens', 'Gegenstand des Diebstahls'. (The meaning 'theft', which is indeed found in genebha (compare also Syr. geniibhta), is due to the development-occurring also in other, similar concepts-of a concrete into an abstract noun).

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

137

The South-Semitic languages, i.e., Arabic and Ethiopic, have one common phenomenon, in which they differ from all other Semitic languages 12 and which, apart from other striking peculiarities, affords definite proof of their very close relationship. I am referring to the so-called 'broken' (or 'internal') plurals. In the present study, we are concerned with certain particular types of these plurals in which we can recognise peculiarities similar to those of the alternative forms discussed above. Current opinion, represented by Noldeke, Lagarde, Barth and Brockelmann, considers the broken plurals to be former singular abstracts without any genetic relation to the singulars now associated with them. These original abstracts are supposed to have adopted the function of (concrete) collectives, which were then attached to different types of singulars by various complicated processes. I cannot conceive, however, how, e.g., qatal ( a) t 'killing' (participle), plural of the singular participle, qiitil in Arabic, qatiili in Ethiopic, could have been derived from an abstract noun with the meaning 'act of killing' (cf. Barth, I.e., §214 b; p.420). Moreover, how could plurals like 'ujiin of 'ajn 'eye', kiliib of kalb 'dog' have arisen from abstract nouns? Even supposing, as Barth does, that the original relation of these plurals to their singulars has been overgrown by analogical formations, and that, as a result, it is often very difficult, sometimes even impossible, to recognise that original relation, this theory is insufficient to explain such extreme shifting of grammatical and semantic function. Barth points out (Ibid., p.418, n. 1) that Ewald, in his Gramm. erit. ling. arab.,13 as well as afterwards Dillmann in his Aethiop. Gramm. (§ 135-148), had started from a different conception-the same upon which the designation gam' at-taksir 'broken plural' of the native grammarians is based. According to this theory (v. Barth, 12 At any rate, only very sparse or uncertain vestiges of broken plurals in the South-Semitic sense are found in other languages. The fact that broken plurals are a (proto-) South-Semitic phenomenon does not preclude the possibility that a disposition for their formation was already present in proto-Semitic (as suggested, e.g., by Barth, l.c., and especially by D. H. Miiller, Actes du sixieme Congres international des Orientalistes (Leyden 1884), II, p. 446, and further by Brockelmann, l.c., § 228 e). The essential, characteristic and distinct types, at any rate the widespread patterns treated here, did not appear before the common period of Arabic and Ethiopic. There is no need for special proof that the phenomenon, in its peculiar aspects, has remained strange to North-Semitic languages. 13 Yet, from Ewald's words (Ibid., § 304) it would appear that even he maintained the original abstract meaning of the broken plurals.

138

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

Ibid., p.417), singular and broken plural are genetically related, the

plurals having arisen from the singulars by 'refraction', i.e., certain transformations of vowels and of the syllabic constitution. This view was contested by Barth (Ibid., p. 418), who maintained, inter alia, that there was no recognisable phonetic or formal principle by which the plurals might be derived from the singulars. I admit that in a great number of cases no definite genetic relation between the singular and the broken plural associated with it can be established. But this does not justify the assumption that the relation of the broken plural to its singular was ever anything but what it purports to be, viz., the relation of a plural to its singular. To derive the function of this plural from a previous different function, such as that of an abstract noun, is pure speculation. Even if we regard the broken plurals as collectives, thus differentiating them from the ordinary 'sound' plurals (formed from the singulars by endings), they have nothing whatever to do with abstracts; a collective, conveying the idea of a multitude as a totality, is no less concrete than an ordinary plural signifying a multitude with special regard for its single units. In actual fact, however, no kind of differentiation of function between the two categories of plural can be either observed or inferred in any way.14 Separately from the ordinary 'sound' plurals-with plural endings -and along with the so-called 'broken' plurals, we generally treat those Arabic plurals, known as collectives, which are formedespecially from participles and nomina agentis - by adding the feminine ending at, such as badijat 'nomads', /:!ar/irat 'residents', sabilat 'travellers', /:!amijat 'defenders', gammalat 'camel-drivers', raggalat 'pedestrians', muslimat 'Muslims' (=muslimun; cf., e.g., Diwan Hurjajl, ed. Wellhausen, No. 183,2; p. 32), muhagirat 'Emigrants' (= muhiigirun) (cf. Brockelmann, Grundriss, I, §228 b; Littmann, ZA 14, 89). With these plurals of obvious origin, patently derived from singulars by adding the feminine ending at, I compare the broken plural qatalat of Arabic or qatalt of Ethiopic, belonging to forms of the singular 14 Also in the oldest known South-Semitic documents-the Old South,Arabic inscriptions-the only ascertainable function of the 'broken plurals' is the plural function. This is expressly stated by Maria HOfner, Altsiidarabische Grammatik (Leipzig 1943), § 87, though she regards this function as due to secondary development and defends anew, with doubtful psychological arguments, the conception of the nature of the forms which is here rejected (see Ibid., § 86).

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

139

participle or nomen agentis, viz., qatiili in Ethiopic, qiitil (or, more rarely, qatil) in Arabic. To prove that this form, qatalt, qatalat, arose by adding at to an associated singular active participle, we cannot but connect it with the Ethiopic participle qatiil-i. 1 S Qatiil in the function of adjective or participle occurs, to a larger or lesser extent, in most Semitic languages (for particulars v. Brockelmann, Grundriss, I, § 131 a, b; p. 344/5). I find some instructive instances, in clearly transitive-active function, in C. Reinhardt, Ein arabischer Dialekt gesprochen in 'Omiin und Zanzibar (Berlin-Stuttgart 1894), p. 427, II. 25, 34, 35; p. 428, ll. 15, 18. These are most interesting as they appear with the feminine ending at in the function of a plural sign: gadafit lubuem 'Verschanzungen Ubersteigende', qsii'et lo/:l$Un 'Festungen Schleifende', qu~ii~t 16 elqane 'Lanzenbrecher', seriibt eddemm 'drinkers of blood', rekiibt efkjeliiji/ 'riders of steeds'. Thus the basic form qatiilat-with unchanged length of the vowel of the second syllable-which we assume for the plural, qatalt, qatalat, of the agent noun and the participle, actually exists even to-day. In those instances quoted from the 'Omani, in which an agent noun in the constructed state is followed by a noun (in the genitive) beginning with a vowel, we find a form with the feminine 15 According to Barth, Nomina/bi/dung, p.41, 59, 69, 356, supported by Hommel, ZDMG 44, 356, n. I, and Brockelmann, Grundriss, I, § 131 c (p. 346), qattili (with undoubled medial radical) is derived from an original infinitive qatti/, plus the Old Semitic ending of relative adjectives (nomina relativa), i. This would mean that the original signification of the word nagtisi 'king' (> nagtisi in Arabic, in Ge'ez replaced by niJgUs), rl1lJti$i 'runner', was 'belonging to reigning', 'belonging to running', respectively. In reality, however, the stem of the form qattil-i is itself a nomen agentis

or verbal adjective, to which the i of the relative nouns has been added, as it were, pleonastically, to stress the adjectival (nominal) character of the form. The addition of the nisbah ending to the participle is quite usual in the Arabic dialects of Tunis (v. Stumme, Gramm. des tunis. Arab., p.66) and Yemen (v. Marcel Cohen, Bull. Soc. Ling., 1935, 196). (In connection with these Arabic cases, both authors quoted have already drawn attention to the endings of the Ge'ez forms, without, however, exactly defining the structure and origin of the latter). Moreover, the addition of the ending of the nomina relationis is current in Tigre (v. Littmann, ZA 14, 92; Brockelmann, Grundriss, I, §228 b), in forms like ktitbaj 'scribe', 'writing', etc. Thus, this theory of the origin of the participle qattili is confirmed by actual, living instances of the mode of formation involved. In a similar manner the ending of the relative adjectives is frequently added in Ge'ez and Tigriiia, and always in Amharic, to the ordinal numbers, which take the form qtitil (as in Arabic), v. Brockelmann, Grundriss, I, § 249 f. The same enlargement by i occurs in the Arabic adjective form 'aqta/, denoting colours and striking physical qualities, as ' asyadi beside the ordinary 'as!lad 'black', etc., and sporadically in certain participles (cf. also the forms quoted by Reinhardt, Ein arab. Dia/ekt gesprochen in 'Oman und Zanzibar, § 110). 16 For the u < a in qU$ti$t v. below, p. 142, n. 18.

140

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

ending t < at (that is, the form q(e)tiilt), which-in analogy to the sequence assumed by us above, *taqtilat> *taqtilt > *taqtilt-is to be regarded as an intermediate form between qatiilat and qatalt. This very form qatalt has been handed down from proto-SouthSemitic to Ethiopic, where it is generally used as plural of its singular base, qatiil, which had meanwhile been enlarged by addition of the ending of the relative adjectives, i: > qatiili. The form found its way into Arabic, too. But here it appears as qatalat, owing to the general restitution in Arabic of the original form of the feminine ending at, a process which finds new and strong confirmation by the case under discussion. The same principle of formation occurs in certain other types of 'broken' plurals. In the second syllable of most of these patterns we likewise find a short vowel. Here, too, the short vowel can be proved to have developed from a primitive long vowel by the phonetic changes to which the addition of the ending at> t, now disappeared in some cases, has given rise. Moreover, these forms have 'a prefixed to the stem, as Ge'ez 'aqt(}lt, Arabic 'aqtul, 'aqtilat. Such 'a is also found in forms to which at had never been added, as Arabic and Ethiopic 'aqtiil beside Arabic qUiil, Ethiopic 'aqtul = Arabic qutUl, e.g., Ge'ez 'ahgur of hagar 'town', Tigre 'anfus (= Arabic nufus) of nafs 'soul', 'a'dug of 'adgi 'ass' (cf. Brockelmann. Grundriss, I, §234). As is clearly apparent from the last two patterns, 'aqtiil and 'aqtUl, the emergence of the initial 'a is connected with the elimination of a vowel in the first syllable of the stem. In Semitic, elimination of a vowel in the first syllable of the stem after a prefix is indeed very frequent. In proto-Semitic, it regularly occurred when a preformative was added to the stem (in which another prefix may previously have been incorporated), e.g., the imperfect *janaqatilu > janqati/u, cf. the perfect, proto-Semitic *naqtala (> Hebrew niqtal; cf. in Akkadian the imperat. napris and other forms) < *naqatala; furthermore, in specifically Arabic cases as ,:!aljakun, faltakun < *,:!alijakun, *falitakun, etc. (cf. above p. 73). It has indeed been maintained that 'a is a real prefix, fulfilling a definite function. If it were, the case under discussion would be completely identical with the last-mentioned cases, as far as the cause of elimination of the vowel is concerned. Barth, Nominalbi/dung, p. 421 (§ 264 d) regards 'a as a specific plural prefix, indicating the difference between these plurals and the abstract formations which he supposes to be their base. Nyberg, Wortbildung mit Priifixen in

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

141

den semitischen Sprachen (Le Monde oriental, 14/1919), p. 294, likewise considers 'a a prefix of definite significance, a sort of demonstrative element, which he asserts to be present in other nominal and verbal forms as well. Yet the existence of such a demonstrative 'a in Semitic has not been fully established. Nor is it clear why this element should have been added precisely to the plural, and not to the singular as well. 1 7 If the suggestion that 'a is a prefix of definite semantic value must be dismissed, it only remains to consider it a so-called prosthetic vowel, sprung up before an initial cluster of two consonants, which in turn owes its existence to the elimination of the vowel of the first consonant: 'aqtal (e.g., ,agmal, pI. of gamal 'camel') must therefore be derived from qUal (gimal), equivalent to it in meaning, by the intermediary of *qtal (*gmal). According to current opinion, Old Arabic has generally taken over and preserved the word-forms of proto-Semitic in their vocalic integrity. In this connection, we again refer to Brockelmann's theory (Grundriss, I, §43 a, p.82; §225 Ah, p. 408; v. above), according to which elimination of vowels has occurred in Old Arabic-or perhaps more exactly, in proto-Arabicas well. Processes like qUal> *qtal are quite common in modern Arabic dialects. In my study The vowel i as auxiliary vowel ... (see above p. 3ff.), I have attempted to trace the accentual processes which brought about the elimination of vowels in patterns of this type, especially in modern Arabic dialects. If, in our case, we wish to reconstruct such intermediate *qtal < qUal, we have to demonstrate that in proto-South-Semitic, or a period near to it-for it is in these periods that the forms under discussion originate-elimination of vowels occurred in forms of this particular syllable constitution, i.e., before a long vowel in the second syllable. In the same just mentioned study (see above p. 71), I refer to the pattern of the passive participle in Ge'ez, qatul < *qutUl from an original qatUl, as preserved in all other related languages. The short u (> a) in *qutul < qatul is ordinarily explained (v. Brockelmann, Grundriss, I, §68 d P) by the assimilating influence of the subsequent u. I have attempted to prove (see above p. 78) that the generally accepted principle of vowel assimilation and dissimilation (v. Brockelmann, I.e., §68, 73 ff., 94), without any further premises, is completely arbitrary and untenable. 17 H. Bauer's theory (ZDMG 66, 103) of the development of 'a was rejected, for good reasons, by Nyberg himself (v. I.e.).

142

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

Proceding from modern Arabic, I restrict the influence of vowels to such cases in which the assimilated vowel is a voiceless (whispered) i, arising spontaneously, by a sequence of processes, in place of an original vowel which has been reduced or completely eliminated. Only in the case of such secondary voiceless vowels of purely mechanical value and without psychical, semantic significance, do I admit a possibility of influence by the shade of other phonemes, consonants as well as vowels, according to the different tendencies of the various dialects. U (> a) in Ge'ez (q.1tul Ge'ez 'a'jant) must have become ' a'jun by dropping the feminine ending t. The spontaneous loss of this ending is to be explained by the plural function of ' aqtult being over-characterized. Even without the ending, the form was quite sufficiently marked as a plural; the t was devoid of any real grammatical function. We are faced here with the reverse of the tendency of specially emphasizing a grammatical function, which gave rise to this very ending t < at (v. above), viz., the striving for ease and minimum effort in speach. It is a principle largely accepted nowadays that under certain conditions words may be transformed and even mutilated by merely psychical factors, without actual phonetic causes being operative. 31 In the present case, however, we have not to do with the mutilation or loss of some accidental item, but with the spontaneous, sudden destruction of an originally significant element, an affix of definite grammatical function. We now turn to the Arabic plural 'aqtilat which, like the plural ,aqtulat, appears in Ethiopic as ' aqtalt, e.g., ' ajra!;t = Arabic ' ajri!;at 'young birds'. This form generally serves as plural of singulars with a long vowel in the second syllable, as qatil, qitiil, qutal (cf. Barth, I.e., p.467, §448), e.g., Arabic 'argifat, pI. of the sing. ragif 'loaf of bread', etc. 'Aqtilat is obviously a 'compensative', cf. Barth, I.e., p. 467. But unlike Barth (Ibid.), we do not regard it as an original abstract, associated with qatil because of their formal affinity, but as a derivative of the latter, formed by the plural sign at and developing on the lines indicated above: *ragif-at> *'argift > *'argift > ,argifat. 32 Finally we have to trace the origin of the pattern qatiililu, which serves as a plural of triliteral nouns with a long vowel after the first

See W. Horn, Sprachkorper und Sprachfunktion (Berlin 1921, 2nd ed. 1923). Unlike the plural 'aqtulat < *'aqtult (> 'aqtul), which is a broken plural additionally characterized by an external ending ('ujun + at > *'a'junt), 'aqtilat may be derived directly from a singular qatil frequently associated with it. In contradistinction to qutu!, qati! is rarely used as plural. - As the plural 'aqtilii' cannot be separated from 'aqtilat < *'aqtilt, it follows that the ending ii' has secondarily been substituted for (a)t. 31

32

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

149

or second radical, of singulars of quadriliteral stems and of triliteral stems with prefixes. 33 To trace this type of plural to its origin, we have to pay attention to the fact that it also occurs with the ending (a)t, and in Ethiopic perhaps more frequently with such an ending than without it. In addition, it should be noted that qatiililu serves as plural of triliteral singulars with a long vowel in the second syllable, e.g., Eth. kasiid 'neck': pI. kasiiJpd, qasis 'priest': pI. qasiiJpst, etc. We may therefore consider the basic form to be the singular qatiil, plus the feminine ending at as a plural sign: *qatiil-at,34 analogous to the basic forms of 'aqtil ( a) t and ' aqtul ( a) t, viz. qatilat and qutulat. *Qatiil-at developed into *qatiilt. This form did not produce what was to be expected from numerous analogies, *qatalt, with the second vowel shortened, but qatii'ilt; i.e., between the long a and the subsequent two consonants a new vowel, i, developed. The genesis of this new vowel may be illustrated by some quite analogous phonetic developments in certain modem dialects. Landberg, Etudes sur les dialectes de I'Arabie meridionale, I, lfat;lramOlU, p. 376, n. 2, remarks with regard to the word ~ ~, giilsin ( < giilisin), occurring in the text Ibid.: "Prie de prononcer ce mot lentement, il dit gii' -el-sin .... II avait donc Ie sentiment que J devait avoir une voyelle; ilia place ici avant, et la consequence naturelle 33 For literature on these formations v. Brockelmann, Grundriss, I, §237 a, Anm. In addition, there are Brockelmann's own theory, set out in Semitische Analogiebildungen, in ZDMG 67/1913, p. 100ff., and Rhodokanakis' remarks in WZKM 29/ 1915, p.60ff. (Brockelmann, I.e., likewise explains the i under discussion as an auxiliary vowel, but its origin as explained by him has nothing to do with the development suggested here). 34 Plurals of like phonetic development may, of course, have been formed originally also from singulars of the pattern qitiil, qutiil (which appear as proto-Semitic and protoArabic variants of qatiil, s. above, p. 142, n., and cf. Barth, Nominalbildwzg, p.60): *qitiil-at, *qutiil-at, as well as from forms with a long vowel other than ii in the second syllable, such as qatil, qulul: qalil-al, qulul-al (comp. the similar forms - - though of different grammatical function -discussed below: !arfialt, medient). The elimination of all possible formations except qatii'i1(a)t < *qatiil-at is due to formal analogy, which began to act when the connection with the respective singular formations had been obliterated by far-reaching phonetic development. In certain cases, the original connection between the plural and a singular base other than qatiil has been restored by secondary analogy (s. Brockelmann, Grundriss, I, §237 c, d) as, e.g., in Ethiop. kasiiljad, pI. to kiJsiid, in which case, however, we may as well and even with greater probability assume that we have to do with the vowel of the original plural formation, based directly on the singular (kasiid = Akkadian kisiidu): *kisiidat> *kisiidt > *kisii'jdt> *kiJsiiljiJdd> kiJsiiljiJd (comp. *ljii~iJdt > !lii~d 'one' (fern.), etc., cf. Brockelmann, I.e., § 56 e, a).

150

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

en est qu'un hamza separe les deux voyelles. C'est ainsi qu'il prononr;a lentement bii-re-ke et plus lentement encore bil -, mais vite biirke; v. p. 377, note 2". In this last-mentioned place Landberg remarks on the word occurring in the text Ibid., t)~ « gazi'-): "prononce gaz-'a et lentement ga'-ze-'a, ou Ie hamza est en vertu de l'accent sur une voyelle finale; cpo p. 376, note 2". Landberg assumes with regard to the e in ga' -el-sin that the vowel which by a regular accentual process of modern dialects had been dropped after the I, was reinstated before it, the speaker having a feeling that the I had to have a vowel. Moreover, Landberg considers the glottal stop to have arisen after the intercalation of the vowel, to separate it from the preceding vowel. The two other instances cited by Landberg, bii' -re-ke and ga' -ze-' a, likewise contain this glottal stop, but without a vowel following it as in ga'-el-sin. Now, the glottal stop in ga'-ze-'a and bii'-re-ke cannot be considered different from that in ga' -el-sin. It follows that the glottal stop in ga' -el-sin does not owe its existence to a hiatus between two vowels; on the contrary, the second vowel can only have developed after and because of the glottal stop.3 5 In these instances from the dialect of I:Ia~ram6t, certain phenomena characteristic of the beginning and the end of the word, respectively, occur in the middle of the word. The glottal stop in ga'-ze-'a (and bii' -re-ke) is, as Landberg himself pointed out, due to the same influence of the stress as ordinarily acts upon a fmal vowel. Indeed, in these southern dialects-and the same seems to apply to certain northern idioms (at any rate in Landberg's notations, as, e.g., in La langue des Bedouins 'Anazeh, Leiden 1919)-this secondary glottal stop after stressed final vowels, especially if they are long (cf. ' alif mamdudah in Old Arabic), is a common feature. What is more, in these dialects, a glottal stop after a stressed long vowel frequently appears even in the middle of the word, as shown by Landberg's texts from the dialect of Da!inah (and also by texts from other, including northern, dialects, contained in his Etudes). This peculiarity is found especially, but not at all exclusively, in poetry, e.g., ma'a'na (= ma'ana = class. rna'ana), Etudes, II, Dalina, p. 173, 6; nf'jeh (= class. nijah), p. 67, 7; leqi'jek (= class. laqijaka), p. 76, n. 5; and, most remarkably, in the word for 'hand', Old Arabic jad, in Da!inah, as in many modern dialects, id, as bif'deha, p. 64, 10; 65, 5, 9; bf'deha, 35 Landberg's hypothesis as to the supplementary intercalation before the I of a vowel which had previously been dropped after it, involves a process unintelligible from the psychological as well as from the phonetic point of view.

A CASE OF QUANTITATIVE ABLAUT IN SEMITIC

151

p. 65, n. 5 (all examples, except the first, from prose texts). But whilst in these latter cases, the glottal stop after a long vowel is followed by a syllable beginning with a simple consonant, in our cases quoted from the dialect of l:la zd and gt > st in this latter case with the change discussed above of qt > bt and qd > gd, although otherwise the two cases represent two separate types of dissimilation. , As explanation of this replacement of a in maSdUd by a in maidiid we state here our following observation: In the ludeo-Arabic dialect of Bagdad the vowel a in pretonic closed syllables is regularly changed into the neutral vowel a (or - differently described - replaced by a): maSdild > meidiid. Also the other words cited by us from this dialect (s. above) are instances of this change: mastam'{n < *mastam'fn (for mugtami'in) , zuyyastii < zayyastu < zayyagtu (I st pers. sing. of the perf. of the 2nd stem; cf. the 3rd pers. masc. sing., zay!!ag, with the original a-vowel preserved. See above p. 156.

7 HEBREW STAYIM ('TWO'), SYRIAC STA ('SIX') AND A TURKIC ANALOGUE 1 The vowel i has disappeared from Hebrew stayim ('two', fern.) < *sittayim (cf. Akkadian sWiin, Arabic lintiini).2 StaYim is to be compared with Syriac sta ('six', fern.; > 'e§tii coexisting with the latter) < Judeo-Aramaic and Mandaic sittii (cf. Arabic sitt-).3 We suggest that in both instances the vowel i, appearing between the unvoiced consonants sand t, has been absorbed by the s, thus an initial cluster st came about: I) *sittayim > *sttayim > stayim, 2) sittii > *sUii > stii. There exists a phonetic affinity between the vowel i and the consonant §. 4 Moreover, the articulation of s brings about the escape of a strong current of air, producing a hissing in which the vowel i (or u, see below), kindred to s, may easily be absorbed. A similar phenomenon may be detected in Turkic cfq- (= tsfq-)

1 Originally published in the Proceedings of the American Academy for Jewish Research, vo!' 21, 1952, p. 1-2. 2 Brockelmann, Grundriss, I, §249 a (p.485), Bauer-Leander, Histor. Gramm. des Hebr., p. 622 c, loiion, Grammaire de I'Hebreu biblique, § 100 c (p. 262), interpret C'l'I!Z1 as setayim with vocal sewa and plosive t. They explain this rather strange form, in' -a~cordance with Barth's suggestion in Oriental. Studien, Festschrift Th. Njjldeke, p.793, as a transformation of the original form *sittayim by an analogy with the form of the masculine, s'nayim. But none of these authors is able to explain the anomalous plosive t after vocal sewa. Kautzsch (Genesis-Kautzsch, Hebrew Grammar, 2nd Eng!. edition by Cowley, 1910, p.288, n. I) is certainly wrong, when he rejects *sittayim as the basic form of stayim by explaining it as "shortened from s'nalayim, which would be the regular feminine form of s'nayim". 3 Brockelmann, Syr. Gramm.' (1938), §166, Anm., p.60 (cf. also Grundriss, I, p. 486) gives the following explanation: "sta statt sitta zu sel nach ':wmsa zu ~ames", i.e., sta (instead of sitta) would be due to the analogy of [~almsa. Such an analogy is unacceptable, as the syllabic constitution of the original sitta is already completely analogous to that qf ~amsa, whereas sta and ~amsa absolutely differ with respect to their pattern (in contradistinction to Hebrew sissa: ~amissa, the latter of which is justly explained -cf. Brockelmann, Ibid. - as an analogy to the former). 4 The vowel i not only tends more to voicelessness (whisper-character) than any other vowel, but-and this is of special importance-is pronounced with approximately the same pitch as s or s. By this the absorption of i (or ii) into s and especially into s becomes a relatively simple process.

HEBR. STAYIM, SYR. STA AND A TURKIC ANALOGUE

159

< *tisi'q- ('to come out'). Brockelmann 5 comments on this case as follows: "Der Schwund des Vokals der Stammsilbe in *tfsi'q- > ci'q'herauskommen' (§ 161 a) zeigt wohl, dass hier der Druck vom Stamm auf die Ableitungssilbe iibergegangen ist (vgl. Benloews Principe du dernier determinant im Idg.). So schwindet auch in den lebenden Dialekten zuweilen ein i oder u in unbetonter Stammsilbe oder wird zur Uberkiirze reduziert, wie I"SUp (Jarring, Mater., 107,17), I"sedu (Eb., 187, 18), picanlerni (Eb., 89,26), kiCik 'klein' (Jarring, Lautl., 36, 12,4), kice 'Nacht' (Eb., 38, 17, I)." Attention ought to be paid to the fact that in all these instances of Turkic the i (or u) which has either been reduced 6 or completely disappeared (as in the above-mentioned Semitic cases), is followed by a s or c ( = IS) and preceded by a I (or at least another unvoiced plosive consonant like k or p). One ought not to explain the loss of i (or u) as the effect of an accentual process, but consider the vowel -e.g., in *Iisi'q- > ci'q--as having been absorbed by the following S. 5 Ostturkische Grammatik der islamischen Literatursprachen Mittelasiens (Leiden 1951), § 10 c, Anm. 3 (p. 65). Cf. also K. Menges, Der Islam, XXI, 157. 6 The specific modification of the 'reduction' of the vowel in the forms noted by Jarring ought perhaps to be more exactly ascertained (perhaps it is an unvoiced, whispered vowel ?). Possibly the reduced vowel is not a remnant of the original full vowel, but was subsequently intercalated to remove the initial cluster (e.g., kisi > Hi> kisi).

8 CONCERNING THE BORDER-LINE BETWEEN CONSONANT AND VOWEL 1 In some Hebrew forms containing a(n originally) geminated laryngal, the vowel ii (?, S?go/) precedes the laryngal, in contrast to analogous forms not containing a laryngal, where the corresponding consonant is preceded by the vowel a (palab) as primarily to be expected. Cf., e.g., 'iiMd, milJ!iibiim, piibiim, 'iiMy, jiJniibiim, kiibiisim (Is. 30,9), piiM; hi!!iihiiru (Num. 8,7), hiihiirim, biihiiriil (Lev. 14,56), hii'iiyon, etc. In view of the fact that in all these forms the laryngal preceded by ii « a) is followed by the vowel a (> ii, Qiime$), scholars as Brockelmann 2 and Bauer-Leander 3 came to the conclusion that the (original) vowel a before the laryngal is 'dissimilated' into ii (?): a-a> ii-a. Bergstrasser,4 who interprets these forms similarly (in his term 'Umlauf), takes into account an additional factor, i.e., the disappearance of the gemination of the laryngal. 5 These interpretations which have been widely accepted, seem to me doubtful for several reasons. Clear instances of such a kind of 'distant dissimilation' of vowels, that is: the dissimilation of two identical vowels separated by a consonant, are hardly found in the Semitic languages. 6 Apart from that, we cannot but ask: If indeed an esthetic (phonetico-psychological) feeling 7 caused the ancient Hebrew speaker (unconsciously) to dissimilate two successive identical vowels, why do we find this (assumed) dissimilation precisely in forms in which the two vowels Originally published in the Hebrew periodical U§anenu, vol. 15, 1947/48, p. 94-96. Grundriss, I, §41 dd. 3 Histor. Gramm. des Hebr., §21 n., p. 216. 4 Hebr. Gramm., §28, p. 152. 5 It is not conceivable how the vowel ii instead of an original a can serve as a substitute for a consonant-doubling which has disappeared. 6 See above p. 60 (= Le monde oriental, vol. 32, 1938, p. 63). 7 Such a case, i.e., a dissimilation based on such an esthetic feeling, is obviously represented by archaic Hebrew llima < *Iahuma < lahumu (14 > a) and analogous Tunisian-Arabic instances; see above p. 78 (= Le monde oriental, vol. 32, 1938, p. 82). But this is a case of quite a specific nature (see Ibid.). 1

2

THE BORDER-LINE BETWEEN CONSONANT AND VOWEL

161

are separated by a laryngal, and more exactly: only in case that the laryngal is (originally) geminated? Can it be assumed that in case the two identical vowels (a) are separated by a non-geminated laryngal, this 'esthetic' difficulty does not exist? On the contrary, it would seem that precisely a geminated (doubled) consonant separating the two identical vowels (a) is more apt to impede such a dissimilation than a non-geminated, simple consonant. But, as already stated, this type of dissimilation is non-existent in Semitic, and the Arabic word biM (e.g.), instead of *bihu, and similar instances hint at the inverse tendency, that is: the assimilation of the vowel following the laryngal to the vowel preceding it. Indeed, in the Hebrew forms under discussion we are faced by a phenomenon of dissimilation, but a dissimilation phenomenon quite different from the one assumed by the above-mentioned scholars. We are dealing here with a case of 'contiguous (juxtapositional) dissimilation', and the process that took place in these forms should be defined as a dissimilation caused by an 'accumulation of phonetic elements of identical nature'. In my view, the laryngals here involved are, in a phonetic-material respect, of a vocalic nature, while in a functional respect they are of a consonantic nature. That means: these laryngals are elements which, taken by themselves, cannot form a syllable, and herein they are different from regular vowels ('sonants'). More exactly: these geminated laryngals relate to the vowels a which are adjacent to them on both sides, in the same way as the consonants Ii (w) and j (y) relate to the 'sonants' (vowels) u and i. Accordingly, we should interpret the phonetic complexes a/:ll:la, ahha, a"a as phonetic sequences corresponding (in principle) to the sequences U'l'lU (UWwu), ijji (iyyi). These specific phonetic sequences containing laryngals may therefore be described as agga. This conception, which defines the laryngals as (some kind of) a-sounds in the function of consonants (g), is reinforced by the well-known observation concerning the affinity of the consonant 'Ajn with the vowel a in the contemporaneous, spoken Arabic dialects. One has, of course, also to pay attention to the hypershort a-vowels which develop in Hebrew as an excrescence from laryngals. The laryngals ' (Alef, Hamzah), h, I:z are different from' CAjin) mainly through their voicelessness. We understand therefore the process through which the a-vowel changed to ii in the instances under discussion. The accumulation of phonetic elements of the same nature is felt by the speaker as un-

162

THE BORDER-LINE BETWEEN CONSONANT AND VOWEL

bearable. This unbearable accumulation of identical phonetic elements is basically lessened by a single one being 'dissimilated' in relation to the others. I assume that the new sound which develops through this process of dissimilation is in all instances only very slightly different from the original phoneme. In our case, the original a-vowel (the initial sound of the sequence of phonetical elements of identical nature) is shifted into an ii-vowel: agga (ahha, ab/:za, a"a) is changed to iigga (iihha, iil},J;,a, ii"a). This interpretation takes account of the fact that this change of a to ii takes place precisely in cases with a geminated laryngal. In the following we deal with an interesting parallel from a contemporaneous Arabic dialect, a case apt to confirm our above interpretation. In the Arabic dialect of Da!ina (in South-Arabia) there occur forms like hinjllm instead of class. ajjam 'days', hinjllh instead of ijjahu 'him' (see Landberg, Etudes sur les dialeetes de I'Arabie meridionale, II, 27,6). Brockelmann 8 treats these forms (in my opinion without justification) in the section dealing with the phenomenon described by him as "Geminatendissimilation" ("Dem Dialekt von Da!ina eigentiimlich ist die Neigung, auch verdoppelte konsonantische Vokale zu dissimilieren ... "). Landberg, in his commentary on the texts in which these forms occur (I.e., p. 340-341, 344), adds other forms of this type: mifiiit < mijjit < majjit 'dead'; hifija < hajja ('allons', 'go ahead'); $ifijir or $ufijir < $iajjir < $i'ajjir < $Ugajjir (deminutiveform of $agir 'small'); qO$ifijir (variant-name of the month Sa'ban) < qU$ajjir (deminutive-form of qa$ir 'short'). Landberg (Ibid.) in turn sees in these forms instances of "Geminatendissimilation" and identifies them with forms like inda'a < idda'a, intafaqa < ittafaqa, etc. Against this conception two objections must be raised. On the one hand, in some of the forms concerned, Landberg's notation of the n-sound « j) is fi (a phonetic symbol not listed in the list of phonetic symbols given by him in· the introduction to his work). One must consequently assume that Landberg observed a sound clearly different from the regular (dental) n-sound. On the other hand-and this seems most important-this n (fi) does not appear in words in which the (underlying, original) sound is preceded by the vowel a. We do not find forms like *anjllm « ajjam). In all forms the vowel preceding the n is i. It is unimportant that in some of the forms this i is in turn rooted in a, as, e.g., hinjiim < *ijjam < ajjam. 8

Grundriss, I, §90 A b (p. 244).

THE BORDER-LINE BETWEEN CONSONANT AND VOWEL

163

According to Landberg (Ibid., p. 346), in this dialect diphthongic sequences like ajja, ajji; aYlja, aljlji develop very frequently into ija, iji; ulja, ulji, as, e.g., sajjid> sijid, majji/ > miji/, etc. On the other hand, this dialect preserves even actual diphthongic sequences like ijj, e.g., ijjehin < ajj(u)hunna 'which of them [fern.]?', thus sequences of the type which constitutes the basis of the process here under discussion. The process which changed ijja to ifija and ijji to ifiji, is clear. The sequence ijj constitutes a complex (cluster) of identical vocalic elements, and for that reason, its pronunciation is to a certain degree difficult. In the case under discussion the phonetic difficulty was solved through the transformation of the original specific modality of the first component of the geminate jj (contained in the sequence ijj) into another modality, (slightly) different from the original modality. The original sound, j, is modified (,dissimilated') by nasalization 9 (though its consonantal character remains preserved). Through this nasal modification it is sufficiently differentiated from the contiguous i-sounds, and the uninterrupted (hardly bearable) succession of several identical phonetical elements disappears. This new nasalized phoneme is approximately identical with a sound found in many languages, i.e., the 'palatal n', which is ordinarily written fi (and appears in several variants in various languages). Cf. also Klingenheben (Nachr. d. Gatling. Ges. d. Wiss., 1933, p. 380), who attests sounds of this character in the (African) language of the Vai-tribe (in Liberia). He expresses these sounds graphically by the symbols y", w" (i.e.: y [1] and w [Ij] sounds which are modified by nasalization). This form of graphic expression would also suit the Arabic sound assumed by us, which is rooted in an original j (y ). This phoneme, which I assume with respect to the instances (quoted above) from the Arabic dialect of DalIna, is conceivable without difficulty. As is well-known, sonantic vowels (i.e., vocalic sounds that constitute a syllable) may in various languages be 'nasalized'. However, exactly as there exist beside the ordinary sonantic (syllabic) vowels consonantic variants (i.e., vowels which quantitatively are reduced to a minimum, a 'glide', that is: j (y), Ij (w) beside i, u)-thus there may also develop nasalized vowels of consonantic quantity and function (in addition to the regular, nasalized vowels of sonantic, syllabic quantity and function).

9

In the terminology of medieval Arab phoneticians: gunnah.

164

THE BORDER-LINE BETWEEN CONSONANT AND VOWEL

Through these phenomena one becomes aware that the border-line between vowels (sonants) and consonants is sometimes so fluctuant that it is difficult to ascertain the exact point at which a vowel (sonant) becomes a consonant and vice versa. Moreover, sounds as y and i pronounced with nasalization (as observed, e.g., in the modem dialect of Da!ina) developed also in ancient Arabic through the reciprocal assimilation (iddigiim) of a n as terminal sound of a word and a i (y) or y (w) as the initial sound of the subsequent word. This process and the specific sound modification resulting from it are included in the rules governing the reading of the Qur'an.

9 THE HEBREW PERFECT FORMS: QAJELA, QAJELOl There are two forms of the perfect of the Hebrew basic stem of the verb, the so-called Qal, the vocalism of which has not been fully clarified. These are the 3rd pers. sing. fern., qiWlii, and the 3rd pers. pl. qiirlu. The same applies, of course, to the corresponding forms of the Nif'al, which are niqrlii, niqrlu, and of the Pu'al (and Hof'al), which are qutrlii and qutrlu (etc.). The main difficulty presented by these forms consists in the fact that the original a-vowel of their second syllable appears reduced, as §ewii. In An Introduction to the Comparative Grammar of the Semitic Languages, ed. by S. Moscati (1964), p.67 (under (g)), the reduced vowel of this type of forms is dealt with in a statement describing the evolution of §ewii in Hebrew in general: "[In Hebrew] short vowels in open unstressed syllables are reduced to a [i.e., §ewii] in accordance with the general Semitic tendency and in contrast to the instances listed under (d) where pre-tonic syllables frequently undergo lengthening; it is likely that these two opposed tendencies were. operative at different periods: e.g. *dabartm > dabiirtm; *qiibarn > qiibarn". This view that the §ewii in qiirlii, qiirlu (the latter form represented in the afore-quoted statement by qiibaru) has developed from an original a-vowel by way of 'reduction' represents the generally held opinion concerning these forms. We feel that it should be assumed that an original a-vowel in an open syllable directly preceding the (terminal) stressed syllable could in Hebrew not be reduced to §ewii. On the basis of the regular phonetic development, a short vowel a in an open syllable directly preceding the stressed syllable could undergo only one treatment, that is: (pre-tonic) lengthening. Indeed, instead of, e.g., the actually existing forms, niqrlii, niqrlu; qutrlii, qu!rlu, and (in the Qal) qiirlii, qiirlu, we would expect the forms *niq!iilii, *niq!iilu; *qu!!iilii, *qu!!iilft, respectively, and (in the Qal) *qe!iilii, *qe!iilu (cf. the forms with suffixes qe!iildl-hu, qe!iilu-hU). 1 Originally published in Journal of the American Oriental Society, vol. 91 (1971), p.429-430.

166

THE HEBREW PERFECT FORMS: QA'fELA, QA'fELU

We maintain that the fonns qiWlii, qiirli1 (niqrlii, niqrlii, etc.) arose by the analogy of the corresponding fonns of the 'intensive' stem, the so-called Pi'el, an extremely frequently used verbal theme, which was certainly able to exert considerable influence on the socalled Qal (as well as on the other verbal themes). It is obvious that the Pi'el-fonns qiWlii, qiWlii relate to the associated fonn of the 3rd pers. sing. masc.: qitlfl, in exactly the same way as the Qalfonns qiirlii, qiirlii relate to qiital. Moreover, in the case of the Pi'el, the sewii is indeed to be considered a phonetic development of an original short vowel, more specifically: of a vowel i (>~, *qittil> qi!lfl; cf. Bibl.-Aramaic qa!!al, Syriac and Targumic, etc., qa!!el). The 'reduction' (to sewii) of this original i represents a nonnal development of i in an open syllable preceding the stressed syllable of a word. The short vowel i before the stressed syllable of a word was frequently reduced to sewii, and not only in instances where it appeared in the middle syllable of a word (as in the instances under consideration), but also in instances where it appeared in the initial syllable of a word. We must therefore come to the conclusion that the rhythmical scheme of qiWlii, qiWlii (which are characterized by sewii in the middle syllable) was transferred to the corresponding fonns of the Qal. The correspondence between the fonns of the two stems is indeed remarkable. The fonns qii!elii, qiWIU (Qal) relate to the fonns qitrlii, qiWlii (Pi'el) in exactly the same way as the fonn qii!al (Qal) relates to the fonn qitt~l (Pi'el). The same interpretation applies, of course, to the corresponding fonns of the Nif'al, Pu'al and Hof'al: niqrlii, niqrlii; quWlii, qu!rlii (etc.). Moreover, we should pay attention to another fonnal correspondence between Qal and Pi'el. The pausal fonns associated with the (secondarily developed) context fonns, qiirlii, qii!elii, are qiit&lii, qiit&lii. On the one hand these fonns do not diverge in any way from the fonns which we would expect from the beginning and do not require any specific interpretation; on the other hand, these fonns (i.e., qiit&lii, qiit&li1) are again in complete agreement with the rhythmical scheme of the corresponding pause fonns of the Pi'el: qi!!~lii, qi!!~1i1 (which also reflect the original i-vowel which is the basis of the sewa in qiWlii, qiWlii). This rhythmical correspondence between the pausal fonns in both verb stems certainly supported the tendency to achieve an analogous correspondence between the context fonns associated with these pausal fonns of the two stems.

THE HEBREW PERFECT FORMS: QA'fELA, QATELU

167

The form of the 3rd pers. sing. fern. of the perfect in Hebrew: qiW Iii is not only remarkable through its peculiar pattern in general, but also through the particular feature that the feminine-ending appears here (in agreement with the ending of feminine nominal formations) as -ii « -ah). Hebrew and the Canaanite dialects in general differ hereby from, e.g., Arabic and Aramaic. On the other hand, this specific feature of Hebrew has an analogon in a language which genetically as well as structurally is very remote from Hebrew, i.e., in Soqotri (see Brockelmann, Grundriss, I, p. 571). It seems, however, to be of considerable interest that the same feature exists in modern ('N orth' -)Arabic, that is: in Yemenite dialects. We refer to our Hebrew article "Notes on the research of the Arabic dialects of Yemen" (Tarbiz [Tarbi~], vol. 13/1942, p.165-177), where we established that the corresponding form-with the ending -ah < -at-is the normal form of the 3rd pers. sing. fern. of the perfect in, e.g., the Yemenite Arabic dialect of Saddiih (a six-day trip from ~an'ah: ~an'ah, Damar, Yarim, Siiddah), thus, e.g., katabah (instead of katabat) 'she wrote', nazalah (instead of nazalat) 'she descended (stopped)'. Obviously, in all these languages, in Canaanite, modern Yemenite Arabic (as well as Soqotri), this feature of the 3rd pers. sing. fern. of the verb (in the perfect) is based on the transfer of the specific ending (ah < at) of feminine nouns, where this specific form of the ending certainly developed at first in pause.

10 THE 3RD PERS. SING. FEM. OF THE PERFECT OF ROOTS III Y/W IN ARABIC 1 W. Wright, Arabic Grammar, I, §25 (p. 26) gives the following rule: "The vowel of a syllable that terminates in a consonant, which we call a shut or compound syllable, is almost always short; as qui, not qui (Hebr. qUI). Generally speaking, it is only in pause, where the final vowels are suppressed, that the ancient Arabic admits of such syllables as in, un, an, etc.". In a Remark to this passage, Wright states: "Before a double consonant a long vowel is, however, not infrequent ... " (he refers to dabbal-, etc.). Ibid., § 166, Rem., on the basis of the rule quoted above, Wright gives the following explanation of the 3rd pers. sing. fern. of the perfect of the Arabic roots III y/w: " ... the masc. forms [are] gaza, rama ... , and not being able to say gazal and ramal (§ 25), they substituted gazal and ramal". In agreement with Wright's rule (l.c., I, §25, see above), Brockelmann, Grundriss, I, §41 k, P (p. 63) states: "Da bei jedem langer angehaltenen Vokal die Schallfiille allmahlich abnimmt, so duldet das Semit. in geschlossener Silbe urspriinglich nur kurze Vokale". And again in agreement with Wright, he applies (Ibid., §271, d; p. 619) this rule to the 3rd pers. sing. fern. of the perfect of roots III y/w in Arabic. Long vowels in Semitic (and Arabic) may, of course, easily be shortened before a sequence of two consonants, although there are numerous instances in which a long vowel in this position remains preserved. On the other hand, it is hard to understand why there should be any phonetic difficulty to pronounce a long vowel followed by a single consonant. And precisely Arabic pause-forms like acjcjallu/in (beside the context-form acj-cjallu/ina) are clear enough examples of the pronounceability of a long vowel followed by a (simple) consonant, and clearly prove that such combinations have existed in ancient Arabic. In modern Arabic dialects, such combinations (long vowels plus consonant in terminal position) are, of course, extremely frequent. 1

Originally published in Arabica, 1. 15 (1968), p. 317-322.

THE 3RD P. SING. FEM. OF THE PERFECT OF ROOTS III Y/W

169

In the fonns of the 3rd pers. sing. fern. of the perfect of roots III

w/y in Arabic, we indeed observe a short vowel before the tenninal

I,

e.g., gadal, ramat, although we would expect a long vowel in view of the fact that these fonns appear to be based on the corresponding masc.-fonns, i.e., gadii and ramii. We must, however, pay attention to the specific character of the tenninal vowel in fonns like gadii, ramii, etc. The tenninal vowel ii as it appears in these verb fonns, or also in nouns like dunyii,falii (with nunation !atan, with a short a), or in the preposition i/ii, etc., represents, according to indigenous grammatical tenninology, a socalled alif maq~rah, i.e., a(n originally) long vowel ii that can be shortened (in contradistinction to the so-called alif mamdudah which is 'protected' by the hamzah, the glottal-stop sound, as, e.g., /.!amrii'u, etc.; see, e.g., Wright, I.e., § 7, Rem. b). The frequent shortening of long tenninal vowels is also documented by certain fonns appearing in poetry as well as by traditions concerning certain Qur'anic readings (see Brockelmann, I.e., §42 h). Moreover, the possible shortening of long tenninal vowels is not only characteristic of Arabic, but it represents an early Semitic feature. Proto-Semitic long tenninal vowels are considered as 'anceps'. That means: one assumes that they could not only appear long (in agreement with their basic quantity), but that they could be easily shortened. We refer with respect to this specific character of protoSemitic long tenninal vowels to Brockelmann, I.e., §42 g (p. 74). In view of the facts described above, it can not be doubted that fonns of the 3rd pers. sing. masc. of the perfect of roots III w/y, as gadii and ramii were frequently pronounced gada and rama with a short tenninal a-vowel. And such forms like gada and rama, with a short tenninal a-vowel, could have served as basis for the corresponding fem.-fonns no less than the masc.-fonns characterized by the original long tenninal vowel, i.e., gadii and ramii. Consequently, it may be assumed that in early times the 3rd pers. sing. fern. of the perfect of roots III w/y in Arabic could appear in two variations. On the one hand, in agreement with the masc.-fonns ending in ii, like gadii, ramii, there may have existed fem.-forms of the type *gadiil, *ramiil (which agree with the corresponding fonns of Aramaic, as r emiil, fern. to remii, and have also analogies in certain modem North-African Arabic dialects). On the other hand, it may be assumed that a second type of fem.-fonns of roots III w/y developed. This type was based on masc.-fonns characterized by a

170

THE 3RD P. SING. FEM. OF THE PERFECT OF ROOTS III Y/W

shortened terminal vowel a. In other words: masc.-forms like gada, rama produced fem.-forms of the type gadat, ramat. It is perfectly clear that only one of the originally coexisting two types could survive, and that was the type with a shortened vowel: gadat, ramat. The same explanation applies, of course, to certain nominal forms of the same type of roots, which, when provided with nunation, display a short vowel instead of the long vowel to be expected, as, e.g., Jatan (Jata-n), beside al-fatii and Jatiihu, etc. Other well-known instances of the alleged rule of the shortening of long vowels in closed syllables in Arabic are the imperatives as well as the jussive-forms of roots II w/y in Arabic, like the imperatives qum (with affix: gUmi, etc.), bin (with affix: bini, etc.), and the jussiveforms yaqum, yabin (with affixes: yaqumu, yabinu, etc.). It is remarkable that yaqum, etc., is also reflected in Hebrew yiiqom « yaqum) , etc., whereas the Hebrew correspondence to Arabic qum is gUm, with a long vowel. The short vowels of these forms of the imperative and jussive, i.e., forms expressing a command, do not reflect a regular phonetic tendency of the languages involved (i.e., Arabic and Hebrew). Rather, the shortening of the original long vowel of these forms is to be considered a consequence of their specific function. The imperative (and jussive) is very frequently uttered with a considerable degree of energy, a mental attitude which quite naturally may lead to a short energetic word form. We refer to our study "The forms of the imperative (and jussive) in the Semitic languages", below p. 195ff.

11 THE ARAMAIC NOMEN AGENTIS QATDL AND SOME SIMILAR PHENOMENA OF ARABIC 1 Aramaic possesses, besides the ordinary (active) participles, certain additional forms of similar meaning, mainly appearing in substantival use and commonly called nomina agentis. As far as the so-called derived stems of the verb are concerned, these nomina agentis are formed from the participle of each verbal stem through the addition of the ending -an. In the basic stem of the verb, the ordinary participle of which has the form qatil ( qate!), the nomen agentis appears as qatOi. This form appears also (under the form qatu!) in a great many Arabic words borrowed from Aramaic, e.g., naqus from Syriac naqosa 'gong (used in churches) for calling to prayer', sa'ur from Syriac sa'ora 'inspector'; see Brockelmann, Grundriss, I, § 128 c (p. 344). The form qatu! appears also in a few indigenous noun-formations of Arabic, which, in agreement with S. Frankel (see 1. Barth, Die Nominalhildung in den semit. Sprachen, p. 177), are not to be regarded as genuine Arabic formations, but are to be considered as due to the analogy of the words borrowed from Aramaic. As far as Aramaic is concerned, in which I consider the form qatOi autochthonous, the u of the West Syriac form qatu! is to be regarded (as normally in such instances) as a secondary development of the original vowel 0 appearing in East Syriac (and also in ludaeo-Aramaic, cf., e.g., na!ora 'watchman'). I am here concerned with the problem of the origin of this nominal pattern, the oldest form of which in Aramaic can be ascertained as qatOl. Brockelmann (loc. cit.), who regards as the basic form of this pattern a form qatu!, with an original u in the second syllable, considers this form qatUi as a development ("Weiterbildung") of a more basic form, *qatu! (with a short u), described by him as an "Ablaut" variant of the ordinary participle qati/. 1 Originally published in Bulletin of the School of Oriental and African Studies (London University), vol. 34 (1971), p. 1-4. The additional note: "The development of Aramaic giilUJ{ii)" (below p. 175-180) is here published for the first time.

172

THE ARAMAIC NOMEN AGENTIS QATOL

As to the o-vowel of East Syriac (and Judaeo-Aramaic), which is to be considered as more original than the u-vowel of the form in West Syriac pronunciation, Brockelmann proposes to interpret this 0 (a long vowel) as arising through the influence of the imperfect, in which an original u appears as 0 (e.g., neqtol < neqtul, Arabic jaqtulu). Apart from the improbability of such an influence of the imperfect, we should assume that an influence of the short vowel 0 in, e.g., neqtol, would have produced a short vowel also in the nomen agentis, i.e., a form *qatol, the more so as, according to Brockelmann, the basic form of the pattern under discussion was *qatul (with a short u), a form which he describes as an "Ablaut" variant of qati/. In my opinion, the Aramaic pattern qatOi is based on a phonetic development of a modification of the ordinary active participle of the 'basic' stem of original roots III y. The historical Aramaic dialects no longer distinguish verbs of roots III y and verbs of roots III j. Both originally independent groups appear as roots III j (also called III aleph), whereas nominal forms based on the original Semitic roots III y, can, on the basis of the traceable phonetic development, still clearly be derived from original roots III y (e.g., ~elola 'prayer' which is to be derived from an original form *~alyata, see my remarks above p. 113ff., and especially p. 116ff.; cf. also forms like lJ,ajjula 'life' from *lJ,ajy ela < *lJ,ajyata, see below p. 178). In contrast to this situation in the historical Aramaic dialects, I assume that in protoAramaic the original roots III y were also preserved in the forms of the verbal system, and had not yet merged with the roots III j and thus disappeared as an independent group. It makes no difference which specific (original) root III y we use to describe the development of the form qatOi. We may, e.g., use the verb gela 'to uncover', also 'to go into exile, to emigrate' (cf. Arabic gala, jaglu of the root gly). The participle may appear augmented by the feminine-ending. In proto-Aramaic, the participle of the fern. sing. of a verbal root III y, as, e.g., gela (based on- an original gly), must have appeared as *galiyat- > *galeyat > *galyat-. Furthermore, a specific Aramaic modification of all nominal forms, including participles, is the so-called status emphaticus, which is formed by the affixation to the noun of the element -a (the original function of which, still widely preserved, was certainly nothing but that of a determinative article). Inasmuch as we should assume for proto-Aramaic the existence of a specific independent group of verbal

THE ARAMAIC NOMEN AGENTIS QATOL

173

forms of roots III !:!, we should also maintain that the status emphaticus of the masc. (sing.) active participle of such original roots III !:! (as, e.g., gl!:!) appeared in proto-Aramaic in the form *gali!:!a> *gar!:!a > *gal!:!a (and not in the form galja, as in the historical Aramaic dialects). In my study "Some aspects of the development of Semitic diphthongs" (see above p. 98-123) I described the development of the post-consonantal (,rising') diphthongs !:!a and ja into the monophthongs a and e, respectively, and gave a number of examples of this development (see p. 107ff.). According to this development of!:!a to a, the proto-Aramaic participial forms of original roots III !:!, *gal!:!at(fern.), and *gal!:!a (masc., augmented by the ending -a of the status emphaticus) could develop to *galat- and *gala, respectively. Of the contraction of a diphthong constituted by the status emphaticusending a and a preceding post-consonantal semi-vowel we have clear examples in Neo-Syriac, e.g., suse < susja 'horse', [ajle < *Iajlja 'night'; for references see above p. 118. These secondarily developed nominal patterns (as *galat-, *gala, etc.), in which the original patterns on which they are based, were no longer recognizable, could by force of analogy be applied to roots other than (original) roots III !:!, e.g., to the very similar roots III j (by which, through various developments, the roots III !:! were subsequently absorbed), as, e.g., the root rmj ('to throw'), but also to the vast group of the so-called 'sound' roots, possessing a 'solid' consonant as third radical, as, e.g., the root qt/. Although I consider the development, in proto-Aramaic, of the original fern. participle *giilyat- to *giilot- as entirely possible (cf. above p. 172ff.), the main influence on the structure of corresponding forms of other roots (as, e.g., q!l, or also rmj) was bound to be exerted by the masc. form of the participle in the status emphaticus, that is, by such forms as, e.g., *galo « *gal!:!a). I thus assume that the new pattern *galo ( < *gal!:!a) was transferred from its original use in roots III !:! to all other roots, e.g., to roots III j, like rmj (remaj, and also to ordinary 'sound' roots with a 'solid' consonant as third radical, like qt/, etc. This transfer, by analogy, of the pattern *gala to such roots as rmj and qt/ could not but produce forms of the masc. (active) participle (or nomen agentis) in the status emphaticus like ramaj-a, qatal-a (and also, subsequently, galaj-a). The speaker, of course, was not conscious of the fact that the model form, e.g., *gala, included already the status emphaticus-ending -a.

174

THE ARAMAIC NOMEN AGENTIS QATOL

I base my derivation of the Aramaic nomen agentis-form qiitOl on the possibility of the contraction of the 'rising' diphthong ya in postconsonantal (terminal) position and on the special circumstance that the participle which served as basis of the development, was followed by an affix. It seems to me that an analogous contraction of the 'rising' diphthong ja > e can be shown to have occurred in Arabic under similar circumstances. Brockelmann, Grundriss, I, §27l, C, b, Anm.2 (p.6l9), discusses the replacement, in the dialect of (the tribe) TaW', of the sequence ija with ii, in forms of the passive of the transitive-active verbs and in forms of the stative (neutral) verbs of roots III j, as rurjii instead of rurjija, and baqii instead of baqija, etc. He ascribes this terminal ii instead of ija to the analogy of the active, that is: to the analogy of forms like banii (root bnj), etc. However, he mentions in the same context also nominal forms (more exactly, forms of the active participle in the fern. sing.) like biidiit- instead of biidijat- (mentioned by Tabrizl in his commentary to lfamiisah, ed. Freytag, 77,2) and biiniitinstead of biinijat-, found in Diwiin Imri'i-' l-Qajs, ed. Ahlwardt, no. 29,2. It is self-understood that the phenomenon present in the perfect forms rurjii and baqii (see above) cannot be separated from the analogous phenomenon in the active participle augmented by the fem.-ending. If this may be assumed (as Brockelmann himself does), then the analogy of the perfect of active verbs (like banii) cannot explain forms like rurjii and baqii (used instead of rurjija and baqija). I describe the evolution of the long ii preceding the fem.-ending -t in biidiit-, biiniit- (instead of biidijat-, biinijat-) , etc., as follows: biinijat- > *biinjat- > *biinet- > biiniit-. The short vowel i of the original form biinijat- could easily be elided. Through the elision of this short i a post-consonantal 'rising' diphthong ja developed: *biinjat-. This post-consonantal diphthong ja could under specific circumstances be contracted to e. For the details of this process and for various instances in which it was operative, see above p. 112. However, with respect to this contraction of post-consonantal ja (in *biidjat-, *biinjat-) to e, it must be noted that (late) proto-Semitic and early Arabic did not possess the vowels e and 0 (short or long). Whenever (long) e- or a-vowels emerged by secondary development, they were shifted to ii (see above p.115). Accordingly, the forms *biidet- and *biinet- (from *biidjat-, *biinjat- < biidijat-, biinijat-, respectively) necessarily developed to biidiit- and biiniit-, respectively.

THE ARAMAIC NOMEN AGENTIS QATOL

175

Analogously, the perfect forms rurja (instead of rurjija) and baqa (instead of baqija) are to be explained as follows: baqija > *baqja > *baqe > baqa; rurjija > *rurjja > *rurje > rurja.

I may in this special context also refer to my interpretation of the form of the 3rd pers. sing. of the perfect of active (transitive) verbs of roots III jllj in Arabic, that is: bana (root bnj) and gada (root gdlj), etc. (beside the forms with [consonantal] affixes: banajtu, gadaljtu, etc.). These forms, bana, gada, etc., are to be explained on the basis of the following developmental sequence: bana < *bane < *banaj; gada < *gado < *gadalj. The forms *banaj and *gadalj in this developmental sequence represent originally the pause forms to the original context forms *banaja and *gadalja (cf. the forms with [consonantal] affixes: banajtu, gadaljtu, etc.). Terminal aj and alj are to be considered to have been contracted in early Arabic (cf., e.g., mata 'when?' corresponding to Hebrew malaD. The e- and o-vowels thus developed were shifted to a: *bane, *gado > bana, gada (cf. above). The forms bana and gada ultimately were not only used as pause forms, but also as context forms, superseding the original context forms *banaja, *gadalja. (For a detailed discussion of this development of bana, gada, see above p. 104-106). Supplement: The development of Aramaic gaJii!(a).

Above p. 172 I stated that in Proto-Aramaic the fern. sing.-form of the active participle of the verb ge la (based on a root gllj) must have appeared as *galiljat > *galljat, and that, in direct analogy to this form, the status emphaticus of the corresponding masculine form must have appeared as *galilja > *gallja (> *galo). I think that a certain term of historical Aramaic is directly developed from the just mentioned Proto-Aramaic form fern. sing. of the active participle: *giilljat < *galiljat, and that this postulated Proto-Aramaic form can thus be reconstructed on the basis of an actually existing Aramaic term. I have in mind the word galul 'exile, captivity' and also '(group, or: community, of) exiled persons'. I assume that the original meaning of this term (a form in the feminine singular) is 'exiled persons' (or: 'group of exiled persons') and that the abstract meaning: 'exile, captivity', is secondarily developed (cf. below p. 179-180). Accordingly, with respect to its original form as well as to its double meaning, Aramaic galUl is to be compared with Hebrew gola. Hebrew golii, which possesses the meaning '(group of) exiled persons' as well

176

THE ARAMAIC NOMEN AGENTIS QATOL

as the meaning 'exile, captivity', is, of course, the feminine sing.-form of the active participle of gala (of the original Semitic root gly) 'to go into exile, to emigrate'. Aramaic galu1 (a) has to be considered as based on a form of identical grammatical function. That is, it should be considered based on a Proto-Aramaic form *galiyat (a) > *galyat (a). The feminine-ending -at serves here (that is: in Hebrew gola, *gOlat, as well as in (Proto-)Aramaic *galiyat > *galyat) in the function of forming a plural (a 'collective'). I refer to instances such as Arabic Muslimat- = Muslimuna, Muhagirat- = Muhiigiruna, etc. With respect to this use of the ending -at we may, of course, (especially in the present context) also refer to Arabic galijat- (galijah) '(group of) exiled persons', which is to be equated with Hebrew gola. The Aramaic noun-form galu1 is characterized by a terminal soundsequence -ut, which is very wide-spread in various Semitic languages, and appears also as an actual ending, that is: as an affix characteristic of terms expressing 'abstract' concepts, an element which in this specific function is frequently affixed also to regular nominal patterns (based on three 'solid' radicals). Bauer and Leander, Grammatik des Bibl.-Aram., §51 g'll' (p. 197) state concerning the affix -ut (mainly) existing in Aramaic, Hebrew and Akkadian: "Es ist durch Anfiigung der Fem.-Endung-in der Form -1, §52 c-an SHimme auf -uy entstanden: *-uy-tu (§9 h) -ut- ... so auch einige nach dem Part. Akt. gebildete Abstrakta: giilu (det. giilulii) 'Gefangenschaft', v . .jgly 'auswandern' ... ; za!s:u 'sittliche Lauterkeit', v. .j zky 'rein sein'; salu 'Nachlassigkeit'; v. .jsly 'sorglos sein' ... ". With respect to this interpretation of the origin of the ending -ut, I wish to stress that Aramaic (and Hebrew) galu1, which is specifically

mentioned by Bauer and Leander, is distinguished from other nouns of the same pattern by the fact that it expresses not only an abstract meaning: 'exile, captivity', but also a concrete meaning: '(group of) exiled persons' (or also: 'emigrants'). And with respect to this double meaning, galu1 is completely analogous to Hebrew gola (see above p. 175). (Hebrew gOla in its turn has a parallel in Arabic galijat-, possessing only a concrete meaning: '(group of) exiled persons' [an abstract meaning, 'exile', does not seem to be attested for Arabic galijat-]). In view of the full semantic agreement between Aramaic galuJ(a) and Hebrew go la, I do not consider Aramaic galu (stat. absol.; stat. constr.: galu1; stat. emphat.: giilu1a) as an original 'abstractum' (in the meaning of 'exile, captivity'), "formed in analogy to the active participle (of the 'basic' conjugation, Qal)", as assumed

THE ARAMAIC NOMEN AGENTIS QATOL

177

by Bauer and Leander (see above), but I consider galli(Ja) to be based on an early Aramaic (proto-Aramaic) regular active participle of the (proto-Semitic and early-Aramaic) root gllj 'to be exiled, to emigrate', an active participle which is characterized as a plural by the affixation of the fem.-ending -at in its 'collective'-forming function (see above p. 176). Accordingly, the basis of the ending -lit in instances of this specific type cannot have been the phonetic sequence -ulJ-tu > lit, as assumed by Bauer and Leander. Rather, the basis of the terminal sequence -lit in this pattern must have been the phonetic sequence -ljat. It is moreover to be stressed (in agreement with my assumption above p. 173) that this phonetic sequence -lJat (which developed to -lit) appeared in post-consonantal position, that is: it followed a consonant without a vowel. That means: the form gallil (a) is directly based on a form *galljat(a) « *galiljat(a). The shift of the post-consonantal sequence -ljat to -lit in nouns of roots III lj (a shift on which the evolution of the Semitic 'abstractending' -lit is to be considered based) could certainly be caused by more than one phonetic process, the specific process operative in a specific case being dependent on the morpho-phonemic structure of a given word and also on the specific dialect or language concerned. In the case of a (proto-)Aramaic form like *galljata, on which the actually existing (historical) Aramaic form gallila should be considered based, the shift of -ljat to -lit may be described as follows: Due to the fact that the main stress of a word in Aramaic is on the ultima, moreover because of the regular shift in Aramaic of pretonic short vowels (in open syllables) into the 'reduced' vowel WlJa), a form like *gtillJala developed to *galljela. However, a 'reduced' vowel (selja) in a form such as this was (sporadically, depending on the speed of speech) apt to be eliminated. Thus, *galljela developed into *galljla. But, since a consonantal vowel lj in a consonantal cluster like this (-lljt-) had necessarily to develop into a regular vowel (a 'sonant')-and a regular vowel (a 'sonant') developed from a 'consonantal vowel' (under such circumstances) can (according to experience) appear long as well as short-the final result was the development of the (stat. emphat.-) form gaiUla (gallila < *galljata < *galilJata). It is rather obvious that the contraction (assumed by me) of the post-consonantal sequence -ljat into -lit-in proto-Aramaic forms such as *giillJat (> giillil) - took place when the nouns concerned were followed by vocalic affixes. We have to think especially of the case that the noun-forms proper were followed by the so-called status

178

THE ARAMAIC NOMEN AGENTIS QATOL

emphaticus-ending of Aramaic: -ii: *giilyalii > giilftlii (see above). The same development (-yat- > -ut-) could, of course, also take place when the nouns concerned were followed by pronominal suffixes (possessive suffixes) beginning with a vowel. With respect to the development of Aramaic giilulii ('group of exiled, or: captive, persons') from a proto-Aramaic (participial) form *giilyatii « *giiliyatii), it may be stressed that this phonetic development is entirely analogous to the development of the corresponding (historical) Aramaic form of roots III j (with which the original roots III y have merged), as, e.g., giililii ('captive woman') < *giiWlii < *giiljatii. Of course, the development of Aramaic giilulii from *giilyatii is analogous, e.g., to that of Aramaic ~aj(j)ulii 'life' (and also [in Syriac] 'animal(s)') from *~ajyatii (the status constructus preserves in this instance-when the word is used in the meaning 'animal(s)' (see Brockelmann, Lexicon Syriacum 2 , s. v.) - still the original phonetic sequence -yat: ~ajyal). I stress once more that the proto-Aramaic form *giilyat ( < *giiliyat), which I reconstructed on the basis of (historical) Aramaic giilul ( ii), represents an active participle of a root III y, augmented by the feminine-ending -at in its semantic function (cf. above p. 176) of forming 'collectives' (equivalent with plurals). That means: it is a structural type related to the type which I consider as underlying the nomina agentis of the pattern qiitOl, that is: *giiliy-ii > *giilyii > *giilug > *giilo (see above p. 173 and p. 175). As for giilul in its use in Hebrew (Biblical) contexts, its should be considered as borrowed from Aramaic. However, it should also be mentioned that Syriac possesses a noun giilyiijii, which expresses the meaning 'an exiled person' (see Brockelmann, Lexicon Syriacum 2 , s.v.), as well as a verb galyi 'to deport, to exile (someone)' (see Ibid.). The generally held opinion seems to be that these two words are derived from the noun giilul ( ii). This was obviously also Brockelmann's assumption, since he lists both words after giilul ( ii). On the other hand, J. Payne-Smith in her Compendious Syriac dictionary (p. 70) assumes that giilyiijii 'an exiled person' is derived from the quadriliteral verb galyi 'to deport, lead into captivity', while she derives this latter verb from the verb galli (Pa"el of gelii 'to go into exile'). In contrast to these interpretations, I am of the opinion that the form giily-, which forms the 'stem' of giilyiijii 'exiled person' is,

THE ARAMAIC NOMEN AGENTIS QATOL

179

exactly like the form gii/utii ( < *giilyatii, see above p. 177), based on an original (proto-Aramaic) form of the active participle of the root g/y (in later [historical] Aramaic: ge/ii [perfect]). That is to say: the form gii/y-, which is the basis of gii/yiijii, has developed from a protoAramaic participle *giiliy ('going into exile, emigrating', or: 'being deported'), a grammatical form being the basis of the original pattern from which the Aramaic nomen agentis-pattern qiitol should be derived (see above p.173). Accordingly, the 'collective' noun (or: nomen agentis in the plural) gii/ut ( ii), and the nomen agentis in the singular, gii/yiij( ii), are both primary forms; none of them is derived from the other. The form *gii/yat (> giilut (ii)) is the 'collective', formed of the participle-form in the singular *gii/y- « *giiUy-) by the ending -at. As for the form gii/yiij ( ii), it represents the participle in the singular: *gii/y- « *giiliy), augmented by the ending -iij of the nomen relationis (nisbah), to which normally - in accordance with regular Aramaic usage - the ending -ii of the status emphaticus is added. It is from this noun - that is: from giilyiij ( ii) - that the quadriliteral verb ga/yi should be considered derived. As to the affixation of the nomen relationis-ending -iij to the stem in gii/yiij ( ii), it may be considered as characterising the basic participle-form (*gii/y < giiliy) in its use as a nomen agentis (quasi a participle [adjective] used as a nomen substantivum). The form giilyiij(ii) is thus in its formation as well as in its specific semantic function completely identical with certain forms of (the modem-Abyssinian language) Tigre, as, e.g., yiirsiij 'heir', 'iiqbiij 'watchman' (see Littmann, Zeitschr.f Assyrio!. 14, p. 88; Brockelmann, Grundriss, II, §263 a [po 577]). Moreover, cf. the active participle of Old Ethiopic (Ga'az): qatiili, which is characterized by the nomen agentis-ending i (cf. Brockelmann, /.c.); cf., e.g., bayiiri, the Qur'anic word for 'apostle' (which is borrowed from Ethiopic). But also in other Semitic languages (especially in Arabic dialects), participles and certain adjectives may appear enlarged by the nisbah-ending (see above p. 139). As far as Aramaic (morphological) parallels to giilyiijii (based on the protoAramaic participle *giiliy) are concerned, we may refer, e.g., to Syriac yiiljiijii 'decens', based on yiile, yiiljii (participle in function of a predicate) 'decet' (see Brockelmann, Lexicon Syriacum 2 , p. 185 a). An important aspect of the above interpretation of Aramaic giilutii consists of the fact that this noun and its Hebrew parallel, gO/ii, possess not only the concrete meaning 'exiled persons' ('community of exiled persons')-a meaning which in my opinion represents the

180

THE ARAMAIC NOMEN AGENTIS QATOL

primary meaning of these terms - but both are used also in the sense of an abstract noun: 'exile (captivity)'. This double meaning of galula, gola is to be compared with the double meaning of Hebrew and Aramaic bajjim/n, which serves in the function of a nomen concretum: 'living ones' (to be considered as the original meaning) as well as in the function of a nomen abstractum: 'life'. For the semantic mechanism involved in this double meaning of bajjim/n-a mechanism which, mutatis mutandis, should be considered as also underlying the double meaning of galula and gola - I refer to the study below p. 525532.

12 THE PLURAL ENDING -UT- OF MASCULINE ATTRIBUTIVE ADJECTIVES IN AKKADIAN 1 The plural ending of feminine attributive adjectives in Akkadian, as in cognate languages, is identical with the (common Semitic) plural ending of feminine substantives, viz., -iit-. The corresponding masculine ending, however, appears in a form -ut-, different from the various plural endings of the masculine substantive which were in use during the successive epochs of Akkadian linguistic history, Brockelmann 2 explains -ut- as composed of the ancient nominative plural ending of substantive nouns, -u (as found in Old-Babylonian), and the feminine ending -t- in its collective-forming function. The existence of composite plural endings of this kind in Semitic languages can hardly be denied. 3 But any explanation of the Akkadian ending -ut- must take into account that this ending is confined to adjectives. 4 The Akkadian adjective appears in two different forms, according to its two different functions. The adjective in predicative function is inflected in exactly the same manner as the so-called permansive (stative), formed of verbal roots. In the plural, we find the endings -u for the masculine, -ii for the feminine. 5 These two endings are of identical structure, the only difference being that of vowel quality.

On the other hand, the sound of the feminine -a occurs also in the corresponding attributive ending -iit-. 6 By analogy a parallel relation-

ship was established between the predicative and the attributive form of the masculine: a new attributive ending -ut- arose as the counterpart to the predicative -u.

Originally published in Journal of Cuneiform Studies, vol. I (1947), p. 343. ZA 17, p. 257 and Grundriss der vergl. Gramm. der semit. Sprachen, I, p. 416. 3 See Brockelmann, ZA 18, p. 100. 4 They may, of course, be substantivized; cf. Ungnad, ZA 18, p. 6, n. 3. 3 These are identical with the oldest forms of the endings which mark the 3rd pers. plur. masc. and fern., respectively, of the West Semitic (genuinely verbal) perfect. 6 This -0- is, of course, a lengthening of the short -a- of the singular ending -at. 1

2

13 THE ORIGIN OF SOME ARABIC PRONOUNS

1

A. The Object Pronouns Formed with iyaMost Semitic languages possess independent (non-suffixal) personal pronouns in the function of the accusative (that is: object pronouns). Generally, these independent (non-suffixal) pronouns consist of a specific particle augmented by the various pronominal (possessive) suffixes. The particle used in Arabic as basis for these independent object pronouns is iyii- (thus iyii-ya 'me', iyii-ka 'thee', iyii-hu 'him', etc.). Various related languages possess equivalent and comparable, but structurally somewhat different forms for the independent object pronouns. In Ethiopic, the basis of the independent object pronouns appears in the form kiyii-; Aramaic uses yii1- (though the forms based on this particle are comparatively rarely used); Hebrew uses '81- (clearly very closely related to the Aramaic form). For details concerning these particles and the pronouns built with them, we refer to Brockelmann, Grundriss, I, 313-15; II, 324-6; Barth, Die Pronominalbi/dung in den semit. Sprachen, pp. 92-5; S. Moscati (and associates), An Introduction to the Comparative Grammar of the Semitic Languages, p. 108. It seems very probable that the majority of the independent object pronouns (accusative pronouns) found in the Semitic languages other than Arabic, are genetically (etymologically) related to the Arabic forms, built on iyii. On the other hand, to us it seems certain that the comparable forms outside Arabic are less original than the Arabic form iyii-; in other words it seems very probable that the above-mentioned Semitic forms outside Arabic represent modifications of the more original Arabic form. In any case, in attempting to explain the development of Arabic iyii- (plus pronominal suffixes), we certainly may restrict ourselves to Arabic, without paying attention to the comparable non-Arabic forms of the independent object pronouns (accusative pronouns).

1 Section A (concerning the particle iyii-) was originally published in Journal of Semitic Studies, vol. 16, 1971, p.50-52. Sections Band C (below p.185-194) are here published for the first time.

THE ORIGIN OF SOME ARABIC PRONOUNS

183

We assume that the independent forms of the object pronouns, in Arabic expressed by the particle iyii- plus (varying) pronominal suffixes, developed after verbs governing two objects; more specifically, that they developed in the case where a verb governing two objects annexed two pronominal objects, which, according to the original Semitic construction, could be expressed by two (consecutive) pronominal suffixes. It seems obvious that the expression of two pronominal objects by two pronominal suffixes (following each other directly) did, in certain instances, not give sufficient prominence to each one of these two pronominal objects. Thus a tendency must have made itself felt to detach the second pronominal suffix from its close connex ion with the preceding (first) suffix (which directly followed the governing verb) and to transform it into an independent word. We consider as the basis of this development instances in which the suffix directly following the governing verb was represented by the suffix of the first person singular. The pronominal suffix of the first person singular in Semitic appears in two forms. In the case where the suffix is added to a noun (that is, in the case of the possessive suffix or genitive suffix), its form is -I; in the case where it is added to a verb (that is, in the case of the object suffix or accusative suffix), it appears under the form -ni. It can, however, not be doubted that the original forms of these suffixes had been -iya and -niya, respectively. In Arabic, these original forms of the pronominal suffixes of the first person singular are still in use under certain phoneticomorphological circumstances. Cf., for example-as far as the possessive suffix is concerned-cases like 'a.yiiya 'my stick', fiya 'in me', 'alayya 'on me', etc. But also in the case of the object-suffix, the original form of the suffix, that is, the form with the terminal vowel a, is still in use in cases where the suffix is directly followed by the (vowelless) consonant of a subsequent word. Beside a'tiinil-kitiiba (a'tiini [a]l-kitiiba) 'he gave me the book', we still find the more original structure a'tiiniya l-kitiiba in use (cf. Brockelmann, Grundriss, I, 307). In cases where the second object of such a phrase is not represented by a noun-that is, in cases where the idea to be expressed is 'he gave it to me' (instead of 'he gave me the book')-the ordinary form of expression is: a'tiinihi (= a'tiini-hi for a more original form a'/a-ni-hu). It is, however, evident that, in the same way as the phrase a'/ani-l-kitaba represents a later development instead of the original form a'/aniya-l-kitaba, thus also a'/anihi represents a later develop-

184

THE ORIGIN OF SOME ARABIC PRONOUNS

ment of an original form a'/iiniyahU (which originally was the exclusively used form). It is on the basis of a structure like this, that is, a'/iiniyahu, that we explain the development of the independent object pronouns formed with iyii- (i.e., iyiika, iyiihu, etc.). That is, the independent object pronoun (e.g., iyiihu) should be considered as having developed from a phrase like a'/iiniyahu by the second pronominal suffix becoming detached from the first suffix and achieving the status of an independent word: a'/iiniyahu> a'/iini (i)yahu 'he gave it to me' (more literally: 'he gave [to] me if). This disengagement of the second pronominal suffix from its close connexion with the first suffix and its acquiring the status of an independent word are to be explained as follows. On the one hand, the original suffix-forms -iya (genitive suffix) and -niya (accusative suffix), when appearing at the absolute end of a word (as it were in pause), developed, in Arabic as well as in most of the related languages, to -i and -ni, respectively; the 'medial' form of the suffix, i.e., -niya, was thus to some degree at variance with its 'terminal' (and extremely frequently used) form, i.e., -ni. On the other hand, the very close connexion of two object pronouns (in the case of verbs governing two objects) was, in certain instances, felt as not giving sufficient prominence to either one of these pronominal objects. A quasi-inevitable consequence of these two factors was that the second pronominal suffix in phrases like a'/iiniyahu ('he gave it-or him-to me'), a'/iiniyaka, a'riin iyahuf'n , etc., split off from the original expression (based on a verb governing two objects) and acquired the status of an independent word: a'riini(i)yahu, a'riini(i)yaka, a'/iinj(i)yahum, etc. It seems obvious that in this process of the 'splitting off' of the second suffix, the phonetic sequence (n) iya (which corresponded to a more frequently used 'terminal' suffix-form -ni, see above) could not but be split into (n)i and iya-. That is, a'/iiniyahu, a'riiniyaka, a'riiniyahum, etc., were dissected into a'/iini iyahu, a'riini iyaka, a'riini iyahum, etc. The independent (object) pronouns iyahu, iyaka, iyahum, etc., which thus came into existence, could easily develop into iyiihu (or iyyiihu), iyiika (or iyyiika), iyiihum (or iyyiihum), etc. The 'strengthening' of the heterosyllabic diphthong iya to iyya (to iya) is a frequently observed process (cf. Brockelmann, Grundriss, I, 59); and the lengthening of the vowel a preceding the pronominal suffixes - iyyii-hu (iyii-hu) , iyyii-ka (iyii-ka) , iyyii-hum (iyii-hum) , etc. - is also easily understood. By the doubling of the middle y (or the leng-

THE ORIGIN OF SOME ARABIC PRONOUNS

185

thening of the initial i, respectively) as well as by the lengthening of the subsequent a vowel, the now independent element (serving as 'basis' of independent object pronouns) was strengthened in its physical nature and thus made more suitable to function as an independent morphological element (particle) characterizing (independent) pronominal objects. B. The relative pronoun

alla~ji:

The Arabic relative pronoun alla4i is distinguished from the relative pronouns in the related languages by its clearly composite character. In An Introduction to the Comparative Grammar of the Semitic Languages, by S. Moscati and associates (1964), p. 114, alla4i is analyzed as follows: " ... the element rj is preceded by the article al- and the infix _la_". This analysis of the relative pronoun basically agrees with Barth's analysis in his book Pronominalbildung, p. 157: " ... die ersten zwei Komponenten sind Demonstrative: al der arab. Artikel, la ursemit. Demonstrativ (§26) ... ". In the continuation (Ibid., p. 158), Barth remarks: "Wiihrend im Hebr. das entsprechende ha-llii-z? in demonstrativem Gebrauche blieb, ging alla4i im Arab .... in die Verwendung als Relativ tiber ... ". We also quote the following statement by Barth (preceding our last quotation): "Diese arab. Neusch6pfung des Relativs ist also aus drei Demonstrativen al-la-rje, -ti komponiert". This analysis of alla4i had been proposed for the first time by Landberg, in his Etudes, II, Dalina, p. 407 (see also the same author's Glossaire Dalinois, I (1920), p. 104). The element al- in allarji is, of course, identical with the definite article of Arabic, as in al-ragulu (> ar-ragulu) 'the man'. Diverging from Landberg's and Barth's opinion (see especially Barth, I.e., p. 158, n. 6), we do not regard the Arabic relative pronoun alla4i as a demonstrative (or, more accurately, as a compound consisting of three demonstratives), secondarily used as a relative pronoun (as German "der" and its many parallels in Semitic and other languages; cf. Barth, Ibid., o. 5). The Arabic relative pronoun alla4i is used in attributive (i.e., relative) clauses which qualify definite nouns, i.e., nouns provided with the definite article al-. In our opinion, it suggests itself therefore to interpret the element al- in alla4i in the light of the use of the article al- before simple, adjectival attributes, as in al-ragulu (a)l-basanu 'the handsome man', that is: to interpret it as brought into existence by the formal analogy of the preceding substantive noun, in other words:

186

THE ORIGIN OF SOME ARABIC PRONOUNS

to consider the use of the article al- in allcuji as a phenomenon of congruency. This specific origin of the article al- in the Arabic relative pronoun allcuji, on the one hand, and the fact that there does not exist in Arabic a form allcuji in the function of a demonstrative, on the other hand, preclude the identification of Arabic allarji and Hebrew hallaz~ (used as a demonstrative). The Arabic relative pronoun allcuji should be explained on the basis of a development which took place within the linguistic history of Arabic itself. As the basic element of the relative pronoun allcuji we may consider the element rji, which represents an original demonstrative secondarily used in the function of a relative pronoun (cf. di as relative pronoun in Aramaic as well as in certain modern Arabic dialects; see Landberg, Etudes, I, lfa4ramout, I, p.414-420). It remains for us, therefore, to explain the origin of the element -la-, which is, as it were, inserted between the element rji (a relative pronoun) and the element al- (basically the definite article). We see in the element -la- not an original (independent) element of a definite semantic function, as that of a demonstrative, but we assume that the form allcuji, which includes this -la-, is the result of a phonetic alteration, or mutilation, of a certain composite expression used, in an early phase of Arabic linguistic history, to characterize relative clauses. The element -la-, seemingly inserted between al- and rji, is to be considered as the remnant of an independent semantic clement which was mainly affected by this mutilation of the original composite expression. The substantival use of (originally attributively used) relative clauses is ordinarily achieved by the prefixation to the relative clause of a so-called correlative (serving, as it were, in the function of an antecedent noun). In most Semitic languages, the function of correlative is fulfilled by an interrogative pronoun or by a demonstrative pronoun. The use of demonstrative pronouns and interrogative pronouns in the function of correlatives is especially frequent in Aramaic dialects. Relative clauses without an antecedent substantive noun are in Syriac frequently introduced by expressions (consisting of a demonstrative and a relative pronoun) like the following: haw de (masc.), hay de (fern.) 'that one who (or: which)" hiinon de, (masc.), han en de (fem.) 'those who'; hiinii rje (masc.), hiirje ge (fern.) 'this one that', hallen cf' (pi.) 'those who' (see N6ldeke, KurzgeJ Syr. Gramm., p. 175, §236). Very frequently-certainly on the basis of the secondary extension of this usage-these expressions serve also to introduce ordinary

THE ORIGIN OF SOME ARABIC PRONOUNS

187

relative clauses, that is: relative clauses preceded by a substantive noun in the function of 'antecedent' (for instances of this type, see N6ldeke, I.e., p. 176; §236B). Also in certain modern Arabic dialects, we find-in complete analogy to the afore-mentioned Aramaic (Syriac) expressions-not infrequently the use of demonstratives in the function of correlatives, that is: as expressions which, in combination with a subsequent relative pronoun, serve to characterize substantivized relative clauses and which (secondarily) may also be used in the function of an ordinary relative pronoun characterizing non-substantivized relative clauses (Le., clauses preceded by a substantival antecedent). A characteristic expression of this type is the term halli. While in some dialects halli ('the one who', or: 'that which') seems to serve only in its original function as a genuine correlative, characterizing substantivized relative clauses, in other dialects, e.g., in a number of the dialects of Syria, halli serves also to characterize attributive relative clauses (preceded by a substantive noun), as, e.g., wttuzziir halli ... 'und die Kaufleute, welche .. .' (Bergstrasser, Zum arab. Dialekt von Damaskus, p. 54,4), bkull bet halli ... "in jedem Haus, welches .. .' (Ibid., p. 57,36; see W. Fischer, Die demonstrativen Bildungen der neuarabisehen Dialekte, 's-Gravenhage 1959, p. 51 ff.). Contrary to the opinion of Landberg, Etudes, II. Dalina, II, p. 407, n.2; p.4l2 and p.422, and Glossaire Dalinois, I, p. 104 (cf. also Fischer, Die demonstrativen Bildungen, p. 47ff.), halli is, in our opinion, to be interpreted as based on hii{jii (e)1Ii 'this [one] who (or: which) .. .'. The relative pronoun halli represents the result of a contraction (or mutilation) of an original hii{jii (e) iii, a contraction (mutilation) which involves the destruction of the consonant g (or d) and has various parallels in the development of original demonstrative elements in Aramaic as well as in modern Arabic dialects (see below). With reference to the phonetic development recognizable in the afore-mentioned modern-Arabic instance, we assume as basic form of the old-Arabic relative pronoun allarji the form al-hii{jii-gi. We are of the opinion that the basic form of the relative pronoun in Arabic was gi (as in Aramaic, where its basic form is di). In sentences in which the relative clause was used in a substantival function, the demonstrative pronoun of Arabic, hii{jii, was pre posed to this relative pronoun gi in function of a correlative: hii{jii gi. This form, hii{jii gi, originally characterizing substantivized relative clauses, was (in analogy to the modern-Arabic halli < hii{jii (e) lli) secondarily used as an ordinary relative pronoun characterizing attributive relative clauses, more

188

THE ORIGIN OF SOME ARABIC PRONOUNS

exactly: relative clauses qualifying definite nouns provided with the definite article, al- (in difference from· the use of relative clauses without a pronoun in the case of indefinite antecedents). Exactly as in the case of the adjective following a definite noun, as, e.g., al-ragulu I-I}asanu 'the handsome man', where the use of the article al- before the adjective constitutes a so-called phenomenon of grammatical congruency (see above p. 186), also the attribute constituted by a clause (a so-called relative clause) was brought into formal accord (congruency) with its antecedent by becoming characterized by the definite article al-. Thus, the Arabic relative pronoun (used in relative clauses qualifying definite antecedents, characterized by the article al-) finally appeared in the form al-hii4ii-rji. This form, al-hii4ii- gemiilattu, Aramaic *'et'aqtal> 'ettaqtal, Ethiopic *'ab'asa> 'abbasa (see Brockelmann, Grundriss, I, p. 158). In connection with our derivation of Arabic allarji from an original al-hii4ii-rji, we may, apart from the modern-Arabic relative pronoun halli < hii4ii lli, cite certain other pronominal expressions which, in accordance with the formerly widely held theory (especially represented by N6Ideke), involve similar phonetic changes. On the one hand, we may refer to hal- < hii4ii 1- (before substantive nouns), which is found in many modern-Arabic dialects, as well as to the modern-Arabic form of the relative pronoun, that is: elli (used in many dialects) from the old-Arabic allarji; and on the other hand, we may compare certain pronominal formations of Aramaic, as, e.g., the demonstrative of Syriac: hiinii, for an original *hii-rjenii (cf. JudeoAramaic hii4en); the interrogatives aynii, for an original *ay-denii (cf.

THE ORIGIN OF SOME ARABIC PRONOUNS

189

Judeo-Aramaic hayden), mana and man, for ma-c;rna and ma-gen, respectively. Contrary to this derivation of the forms involved from basic forms which included the ancient Semitic demonstrative (with initial g, or d, respectively), Landberg and W. Fischer (see references above p.187, and cf. also Fischer, I.e., p.46-47 and p.50) assume primary forms for these modem-Arabic and Aramaic pronouns which did not contain demonstratives of the g-stem. With respect to modernArabic elli (allegedly not derived from allarji), we refer specifically to Landberg, Glossaire Dalinois, I, p. 104. W. Fischer's view has been accepted by H. Grotzfeld, in his Laut- und Formenlehre des Damaszenisch-Arabischen (1964), p.47 (concerning the relative pronoun yalli, also-after vowels-iii, which he does not derive from allarji, see Ibid., p. 51). This view, held by Landberg, W. Fischer and Grotzfeld, seems to us unacceptable. The modem-Arabic relative pronoun elli cannot be explained but as a contraction of the old-Arabic relative pronoun allarji. Analogously, we should assume basic forms containing the (West-Semitic) demonstrative element g for all the other forms involved. This does not only apply to the modem-Arabic forms, as hal- < hiirja 1-, or halli < hiirja iii (or hiirja llarji), as we already stated above, but also to certain (ancient) Aramaic forms. J. Blau, e.g., although (in contrast to W. Fischer) deriving the just-mentioned modem-Arabic forms from primary forms with the g-element (see Orbis, vol. 9, 1960, p.201), is inclined (see Ibid.) to reject (in agreement with Fischer) the derivation of Syriac hana from *ha-gena. Indeed, also in S. Moscati's (and associates') book An Introduction to the Comparative Grammar 0/ the Semitic Languages (p. 112), Syriac han ( a) is not derived from a primary form with the d-element, but is interpreted as constituting a combination of the element ha with an element n. Concerning this interpretation of han ( a), we refer to our discussion below p. 414. With respect to our assumption (see above p. 187) that Arabic allarji is based- on an original relative pronoun which (apart from the definite article al-) contained a correlative, i.e., the demonstrative pronoun hiirja, we referred to the use of relative pronouns of this specific type (that is: compounds containing a correlative) in modernArabic dialects as well as in Syriac (see above p. 186). We may also mention that the use of a correlative preceding the relative pronouns has been observed in so-called Middle-Arabic. We refer to J. Blau, A Grammar o/Christian Arabic (Louvain 1967), p. 565f. It is especially to be noted that in Middle-Arabic correlatives are not only used in

190

THE ORIGIN OF SOME ARABIC PRONOUNS

the case of substantivized relative clauses (a case in which, in our opinion, the use of correlatives came into existence), but also in· the case of attributive relative clauses. We refer to the instances quoted by Blau, l.c. We have, however, the clear impression that in most of the instances quoted by Blau, in which the relative pronoun preceded by a correlative follows an antecedent substantive noun, the relative clause does not constitute a real attributive relative clause, but a so-called continuing relative clause (in German terminology (pI.) 'ankniipfende Relativsatze'), a type of relative clause which, in our opinion, can hardly be separated from the relative clause in substantival function. The use of relative clauses preceded by a correlative in the case of attributively used relative clauses should be considered an extension of the use of a correlative in the case of continuing relative clauses (which in their tum are related to relative clauses serving in substantival function; see above). Moreover, it seems to us that a correlative, supplementing a relative pronoun-a function to which the term 'correlative' ordinarily refers-should be strictly distinguished from a demonstrative pronoun preceding the antecedent noun of a relative pronoun. Blau (I.c.) treats instances of this latter type as instances containing a correlative, quoting them side by side with instances in which the demonstrative pronoun precedes the relative pronoun directly. We mean such instances as the following one (quoted by Blau, I.e., p. 565 B): fi !/iilika I-rna lla¢i fihi !-!ayr 'in the water in which the birds were', etc. We have in mind also those instances in modem literary Arabic (e.g., TaM I:Iusayn, alAyyiim, II, p. 15,3,7; p. 18,11; p.20,9) to which Blau refers in his book Syntax des pallistin. Bauerndialekts von Bir-Zet, 259/60, as well as in Orbis, 9 (1960), p.203, and in A Grammar of Christian Arabic, p. 565, n. 78. In all these instances, the demonstrative pronoun possesses a real deictic force and is designed to stress the substantive noun which it precedes. We revert to our interpretation of alla¢i as based on an original form *al-hii4ii-(/i in which the relative pronoun (*gi) was preceded by a demonstrative pronoun (hii4ii) in function of 'correlative'. It is possible that we deal with a relative pronoun of a similar type of composition in the case of an expression occurring in two (almost identical) Lil).yanic inscriptions. W. Caskel, Li~yan und Li~yanisch (Ko1n 1954), p. 63, refers to this expression as follows: "Als Relativpronomina dienen gu sowie man und mii. No. 30.2.3, 32.1 liegt eine Uberfiillung vor: hii ... hii4ii gu; hii ... hii4ii sollte geniigen". We

THE ORIGIN OF SOME ARABIC PRONOUNS

191

quote the passage in point (Ibid., p. 91 and p. 93): S w-lf ... w-'m-Sms lfyw hdqw h#mn hgh gngr l-'glbn 'bhm Ijrbrjgbt, which Caskel renders by: " ... S und I:I ... fund 'A]mm-Sams I:Iayau haben dargebracht dieses Standbild, welches ihr Vater Ijrl:H)ii-gabat dem 'Aglibon gelobt hatte". What appears to be an 'overfilling' could also be explained by assuming a correlative of the same type as assumed by us as contained in the basis of allagi that is: al-hiirja-gi (with hiirja serving as correlative). In the function of demonstrative pronoun we find in Libyanic the element gil/a. The function of relative pronoun is expressed by the element gil (see Caskel, Ibid., p. 63). As determinative article the element ha is in use (cf. Caskel, p. 68). Accordingly, hrjh g in hrjh gnrjr (following the nominal antecedent h-:jlmn) could be interpreted as a composition of the following elements: the article h (prefixed to the relative clause in analogy to the article of the antecedent: h-:j lmn), the demonstrative pronoun gh (originally functioning as a 'correlative') and the (basic) relative pronoun g (prefixed directly to the verb of the relative clause: ngr). Instead of interpreting the sentence by " ... have offered this statue, which their father had vowed to ... ", we could certainly also interpret: "... have offered the statue which their father had vowed to ... " (in its use as 'correlative', the original demonstrative had, of course, not preserved its original demonstrative meaning). The Arabic relative pronoun allarji represents in any case a secondaryenlargement of the element *gi which in an early period (we may say: in the proto-Arabic period) served by itself (in an unenlarged form) as a relative pronoun (in analogy to, e.g., Aramaic di). And this function of *gi as a relative pronoun was, of course, developed out of an original function as a demonstrative pronoun. For the process of the development of an original demonstrative pronoun (like Arabic *gi> Aramaic di) into a relative pronoun, we refer to our Studies in Arabic and General Syntax (Cairo, Institut Fran~ais d'Archeologie Orientale du Caire, 1953), §51 (p.63). C. The demonstrative pronoun galika and related forms

Arabic has a series of forms for the demonstrative pronouns which are characterized by the insertion of an element Ii. These forms are: galika (m.), tilka < *tilika (f.) (also, but rarely used, tiilika), and uliilika (pI., rarely used). The forms are used in the function of the 'far'-demonstrative, i.e., in the sense of 'that', 'those'. The same

192

THE ORIGIN OF SOME ARABIC PRONOUNS

element, Ii, appears also in the demonstrative adverb hunalika 'there' (based on the form hunaka 'here'). Barth, Pronominalbildung, p. 77-78, considers this element Ii as a demonstrative element to be identified with the element la in the Arabic relative pronoun alla4i, allati, etc. (concerning which see our remarks above p. 185). This opinion concerning the element Ii in rjalika, tilka, hunalika, etc., was held also by numerous other scholars (see, e.g., Brockelmann, Grundriss, I, p.318), and seems to be generally accepted. Barth (l.c., p. 78) compares with the demonstrative adverb of Arabic, hunalika 'there', the demonstrative (local) adverbs of Syriac, harka 'here' and hartamman 'there', and assumes that the consonant r in these words is based on an original I. This explanation of the r in these words as shifted from an original 1 is in conflict with Noldeke's interpretation of this r. Noldeke (WZKM 8, 265, n.4) recognized in the r of harka and hartamman a remnant of the imperative of the verbal root r'y (Arabic ra'a, Ethiopic ra'aya < *ra'iya) 'to see' (a root which in Aramaic generally disappeared as an independent verb). We accept Noldeke's interpretation of the r in Syriac harka and hartamman, forms which correspond to the Babylonian-Talmudic Aramaic ha/sa and hii!am (which do not contain the r-'infix'). What is more, we are of the opinion that the same element r, originally representing the imperative of the verb r'y 'to see', is to be taken as the basis of the syllable Ii in the Arabic local adverb huniilika 'there' as well as of the same syllable in the demonstrative pronouns rjiilika and tilka (talika). Moreover, we compare the combination -rka in Syriac harka with a certain linguistic usage of Old-South-Arabic. In Sabaean, we find an element r', which may be followed by an element k, thus appearing as r'k. We refer to N. Rhodokanakis, Gottingische Gelehrte Anzeigen, 176 (1916), p.27, and to A. F. L. Beeston, A Descriptive Grammar of Epigraphic South Arabian (London 1962), p.53. Rhodokanakis (l.c.) identifies this Sabaean r' with an element ra (also ra', with 'Ayn) known from certain modern Arabic dialects, which he (in agreement with Barth, Sprachwiss. Untersuchungen zum Semit., II, Leipzig 1911, p. 27ff.)-in contrast to its ordinary interpretation as the imperative of the verb ra'a 'to see' (cf. also W. Fischer, Die demonstrativen Bildungen, p. 186ff.)-interprets as a demonstrative (deictic) element. As against this, we interpret the Sabaean element as originally representing the imperative of the verbal root r'y 'to see'. Another question is the character of the element k by which the element r' may be

THE ORIGIN OF SOME ARABIC PRONOUNS

193

followed in Sabaean. Rhodokanakis defines it to be of a 'deictic' character and identifies it with Hebrew ki. We consider the Sabaean element k by which the element r' 'behold!' (imperative of r'y) may be followed, to be identical with the Aramaic particle kii 'here', which appears, e.g., in Judeo-Aramaic hiilsii 'here' and in the corresponding Syriac adverb hiirkii (cf. also Syriac rlsii 'hither', mekkii 'from here', Talmudic ikkii < il kii 'there is (here)" etc.). The same (Proto-Semitic) element kii appears also in Hebrew, and not only as the second element in combinations like kii-Isii 'thus', elsii 'how?', but also in an independent, unenlarged form, that is: ko, which ordinarily means 'thus' but is also used in the (more original) meaning 'here', and shows the regular shift of the (Proto-Semitic) long vowel ii to 0 in Hebrew. Accordingly, we reject the derivation of this Hebrew ko from a basic form 'as he, as it = thus', a derivation which (with reference to k e 'as') has been proposed in Koehler-Baumgartner's Lexicon in Veteris Testamenti Libras (1953), p. 423 b. We state that also the element kii in the (North-}Arabic demonstrative pronouns giika, aka, and also in giilika, tilka, as well as in the adverbs huniika and huniilika, must be identified with the Aramaic (and Proto-Semitic) element kii 'here'. Our identification of Classical Arabic enclitic ka (with short a) with the Aramaic (and Proto-Semitic) element kii 'here' (in Hebrew ko, beside *kii) leads us to the conclusion that the vowel a in the ClassicalArabic enclitic element ka is shortened from a long vowel ii. Likewise it must be assumed that also the Sabaean element k which may follow the element r' (,behold !'), is to be read: kii and to be interpreted as 'here', so that the combination r'k should be considered as originally expressing the meaning 'behold here!' (cf. French 'voici'). Sabaean r'k is thus identical with the combination -rkii appearing in Syriac hiirkii 'here' (which contrasts with hiirtammiin 'there' in the same way as in Talmudic-Aramaic hiilsii contrasts with hiiliim). Moreover, this original combination r' -kii 'behold here!' has to be considered as the basis of the element -lika which appears in Classical-Arabic giilika, tiilika (tilka) and huniilika. There is a further Arabic formation which we shouid interpret as containing the adverbial element ka < kii 'here'. This is the Arabic expression hii-ka 'there you have!', or: 'take!', with the feminine-form hii-ki, and the plural-form hii-kum. We identify Arabic hiika (etc.) 'there you have!', 'take!' with Judeo-Aramaic hiilsii 'here' (corresponding to hiirkii in Syriac). We thus assume that the original meaning of Arabic hiika « *hii-kii) 'take!' was just 'here'. It goes without

194

THE ORIGIN OF SOME ARABIC PRONOUNS

saying that an adverb of this meaning-especially when accompanied by a gesture-could easily be used in the sense of 'take!'. Moreover, it was only natural that, in adressing a single person, the original adverbial element ka « kii 'here') was apt to be identified with the (masc.) pronominal suffix -ka, and that, subsequently, in case that the command was adressed to a woman or to a number of persons, the forms hiiki and hiikum, respectively, developed. It seems interesting to note that the modern Arabic dialect of Tunis uses certain forms which are identical with the ancient Arabic forms hiika, hiikum, in a meaning different from that which these elements possess in ancient Arabic. According to H. Stumme, Grammatik des tunisischen Arabisch (Leipzig 1896), § 182,2, the dialect of Tunis uses hiik and hiikum in the meaning 'there you are' (sing. and pI., respectively: 'da bist du', 'da seid ihr'). We are of the opinion that this modern usage is directly based on the original (Proto-Semitic) meaning of the element hiikii, which is, in our opinion, the basis of ancient Arabic hiika (etc.) 'take!', that is: based on the meaning 'here', and not-as Brockelmann, Grundriss, I, p. 503, assumes-based on the ancient ('Classical') Arabic sense of hiika, hiikum, that is: 'take!'.

14 THE FORMS OF THE IMPERATIVE (AND JUSSIVE) IN THE SEMITIC LANGUAGES 1 The Hebrew system of moods shows a remarkable affinity with that of Arabic. Compare, e.g., the indicative and jussive of the basic and causative conjugations of Hebrew verbs mediae j/lj: jiiqum: jiiq(Jm; jiiqim: jiiq(Jm, with the completely analogous Arabic jaqiimu: jaqum; juqimu: juqim. A notable difference exists in the imperative of the basic conjugation: Hebrew qum (long vowel), Arabic qum (short vowel) besides qumi (fern.), qumu (masc. plur.). Owing to a mistaken phonetic rule, the short u of qum has always been compared with the short vowel of forms like ramat 'she threw' as against ramii 'he threw'; qiifjin 'a judge' as against al-qiifji 'the judge', etc. (cf. Brockelmann, Grundriss, I, §41 k~).2 Some years ago,3 I explained the short u of qum as resulting from the special psychological content of the imperative. The special intensity with which the imperative is very often uttered, is best expressed by a short word with final consonant. The same applies, of course, to the imperative of the causative conjugation, Arabic ' aqim, Hebrew hiiq(Jm, as well as to the jussive of the two conjugations, Arabic taqum, tuqim, Hebrew tiiq(Jm, tiiq(Jm, etc. (as against the forms of the indicative, taqumu, tuqimu; tiiqum, tiiqim).4 The same effect of the emphatic pronunciation of commands, viz., the development of forms with short vowel and final consonant, may be observed in various imperative forms of Hebrew and Arabic roots 1 This study (originally the text of a communication to the Congress of Jewish Studies held in Jerusalem in July 1947) was previously published in The Jewish Quarterly Review, vol. 42 (1951/52), p. 51-56. 2 For the relation of the short vowel (as in ramat) to the original long vowel (as in ramii) see above p. 168ff. 3 See my article "Des effets de I'accentuation emphatique sur la formation de la langue en semitique", Mem. Soc. Ling., t. XXIII, fasc. 6 (1935), p. 347. 4 The long vowel of the imperative of the Hebrew basic conjugation, qum, is proved to be secondary by comparison with the short vowel of the corresponding form of the causative conjugation, hiiql;'m, of the imperative of the Arabic basic conjugation, qum, the Hebrew and Arabic forms tiiqom and taqum, etc. Hebrew qi1m coincides with the ordinary imperative of modern Arabic dialects, qum, instead of ancient qum. In both cases, qum is to be explained as restitution in analogy to the forms with final vowel, qumi, qumu.

196

IMPERATIVE (AND JUSSIVE) IN THE SEMITIC LANGUAGES

tertiae jjy. Contrary to the accepted view, I explain in this way certain forms of the imperative of the intensive and causative conjugations in Hebrew, viz., $ay, gal; ha *irm. This accentuation is in fact known from living dialects. Arabic has a particular form of command, in which an apocopatejussive is preceded by the preposition Ii: lijaktub, litaktub 'let him write', 'let her write', etc. It has been recognized that this type of independent sentence developed from a subordinate clause by omission of the main clause. However, subordinate clauses introduced by the final conjunction Ii require the subjunctive: lijaktuba, litaktuba. Brockel-

IMPERATIVE (AND JUSSIVE) IN THE SEMITIC LANGUAGES

197

mann (Grundriss, II, § 348 c, Anm. 1) assumes that the original subjunctive was replaced by the apocopate when the original subordinate clause became an independent sentence, which would mean that the original form has been replaced by an entirely different one. As against this, I regard this apocopate as a form which developed organically from the original sUbjunctive by omission of the final vowel. Brockelmann seems also to assume that the function of the sentence-part introduced by Ii within the original, unabridged sentence was merely that of a final, i.e., an adverb clause, which is not an essential, indispensable part of the sentence as a whole. As against this, I hold that the abridged sentence lijaktub developed from a complete sentence such as 'uridu lijaktuba 'I want him to write'; the final clause introduced by Ii, i.e., an adverb clause, appears here in the function of an object, i.e., of an essential, indispensable sentence-part. 5 The sentence lijaktub is certainly to a great extent analogous to forms like French qu'i/ eerive, which was originally an ordinary object clause within a structure like je veux qu'i/ eerive. The omission of the governing verb is, of course, a result of the extraordinary intensity with which a command is very often uttered. Despite the omission of the word expressing the idea of volition, command, the mental element itself subsists; and moreover, as we learn from the Arabic example, it retains a special phonetic expression. The emphatic pronunciation of the element expressing volition passes to the object of this volition and produces in the Arabic case elimination of the final vowel of the verb: lijaktuba becomes lijaktub. Let us now consider the genesis of the ordinary form of the imperative in Semitic languages generally. It is known that in Indo-European languages, the imperative is on the whole identical with the simple verbal stem ('Tempusstamm'; see Brugmann, Kurze vergl. Gramm. der indogerm. Spraehen, § 730). In similar manner, Semitic languages show complete identity between the imperative and the stem of the imperfect. According to the view prevailing in the earlier period of Hebrew S The characteristic modal content of the governing (main) verb 'uridu, a verb expressing volition, demand, must be regarded as inherent also in the subordinate clause which forms the object of the demand. This modal content is closely allied to that of a final clause, the characteristic form of which was therefore adopted in the present case, too. The exact mechanism of this complicated structure still requires detailed investigation. For a more detailed description of the semantic mechanism which is involved here, I refer to my Studies in Arabic and General Syntax (Cairo 1953), p. 136ff.; moreover, see below p. 327-328.

198

IMPERATIVE (AND JUSSIVE) IN THE SEMITIC LANGUAGES

and Semitic Studies (down to Ewald, Stade, etc.), the imperative was derived from the imperfect, by omission of the personal prefix ta: taqtul> *qtul. This explanation was subsequently replaced by another, probably originating from N6ldeke, which was adopted by the new generation of scholars, such as Barth and Brockelmann, and is also found in the Hebrew grammars of Bauer-Leander and Bergstrasser. According to this view, the existing imperative forms arose from disyllabic primary forms, such as qiti!, qutul, qatal (see Brockelmann, Grndriss, II, § 258, B). It is asserted that these forms have been preserved integrally in the Akkadian imperative, e.g., kusud, and also the reduced vowel (Selda) of Hebrew-Aramaic forms, e.g., qet9l, is regarded as a remnant of the alleged original full vowel. Moreover, it is assumed that the imperfect arose from such imperative forms by addition of the personal prefixes-thus Brockelmann, Grundriss, II, § 70 a, § 74 c. The Hebrew imperative forms with a vowel after the first radical, such as qi !Ii, qi tlfl, are usually explained as direct descendants of these alleged primary forms; in my study "Uber i als Hilfsvokal" (Le Monde Oriental, vol. XXXII, 1938), p.2ff. (see this volume, above p. 4), I had already explained these forms as derived from primary forms beginning with two consonants without intervening vowel, viz., *q!uli, *q!ulU, etc. Contrary to the now prevalent explanation of the form of the imperative, I adhere to the earlier view, maintaining that an imperative

form like *qtul had developed already in the proto-Semitic period from forms like taqtul, taqtulu, by omission of the prefix tao I identify this primary form with a common sentence-type which, though expressing command, appears in the indicative, e.g., English 'you will go at once', 'you will not touch him'; German 'du kommst sofort', etc. In addition, we compare it with the above mentioned Arabic liiaktub and French qu'i/ ecrive, and draw the conclusion that these primary sentences are originally the object of an independent mental element expressing the idea of volition, demand, and bear, similarly to lijaktub < lijaktuba, the emphasis characteristic of the mental element expressing the idea of command. As we have seen before, the particular intensity with which a command is very often uttered produces an extremely short form ending in a consonant; the same intensity ousted already in the proto-Semitic period the personal prefix ta, which was functionally unnecessary in the specific situation attending the issue of a command, and thus arose a short form, perfectly adapted to the character of the imperative. '

IMPERATIVE (AND JUSSIVE) IN THE SEMITIC LANGUAGES

199

The combination of two consonants at the beginning of a word infringes the usual phonetic rules of the languages, but its pronunciation must be considered possible under the particular conditions of mental tension; the same conditions account for the omission of the personal prefix ta, which strikingly exceeds the limits of ordinary phonetic changes. The development of an auxiliary vowel in the initial consonantal group probably occurred not in the proto-Semitic period but in the period of separate languages, as may be inferred by comparing Arabic uqtul « *iqtul), igrib, etc., with Akkadian kU5Ud, Ethiopic qatal and Hebrew-Aramaic qet91.

15 THE SEMITIC CAUSATIVE-PREFIX S/SA The prevailing view concerning the three prefixes of the causative stem of the Semitic languages and of Egyptian, s/sa, ha, and 'a, is that we should· assume at least two separate original prefixes, that is: ha and s/sa; see, e.g., An Introduction to the Comparative Grammar of the Semitic Languages, ed. by S. Moscati, p. 125-126. As to the distribution of these two causative-prefixes, it seems noteworthy that the prefix Sjsa is used exclusively (i.e., without ha- or a-prefix used in addition) in those languages that are documented by literary monuments in very early historical periods: s/sa is the causative-prefix of Akkadian and Egyptian (s). It is moreover used in Ugaritic and, indeed, in all ancient South-Arabian dialects except Sabaean which uses h. However, precisely the fact that Sabaean uses h, whereas all the other Old-Semitic-Arabian dialects use s, seems to us to be of great interest. If dialects as closely related as the Old-South-Arabian languages use different, though similar and hence comparable, morphological elements, then it seems warranted to assume that one of them, and that is certainly h ( a), has developed from the other, that is, s( a). Moreover, in dealing with the causative-prefix we cannot overlook the forms of the personal pronouns-originally demonstratives and in some dialects still used as such-which, in some languages, are characterized by sis and, in other languages (more precisely, in the majority of the languages concerned), byh. In this regard it is notheworthy that the use of Sjs and h, respectively, with the pronouns parallels in all of the languages concerned, except Ugaritic, the use of Sjs and h, respectively, in the case of the causative-prefix. That is: Akkadian, Egyptian as well as Minaean-Qatabanian use pronouns characterized by a sibilant: Akkadian masc. SU, fern. si, Egyptian masc. SW, fern. sy, Qatabanian masc. SW, swt, fern. syt (with analogous forms in the corresponding plurals)-in contrast to Sabaean and all the remaining Semitic languages all of which use pronouns charac1

Originally published in Le Museon; Revue d'etudes orientales (Louvain), vol. 82

(1961), p. 517-522.

THE SEMITIC CAUSATIVE-PREFIX S/SA

201

terized by h (in place of the sibilant). Also in this instance one is, as a rule, inclined to assume two originally separate pronominal stems (see An Introduction to the Comparative Grammar of the Semitic Languages, ed. by S. Moscati, p. 104). Regarding this difference in the structure of the pronouns, the situation in the South-Arabian area is once again of considerable interest. We refer to the fact that dialects as closely related as Qatabanian, on the one hand, and Sabaean, on the other, diverge concerning these forms so decidedly from each other, while they agree, regarding the same forms, with certain other languages with which their relationship is far more remote. Thus, also in the case of the pronouns we are led to the conclusion that the Sabaean forms characterized by h have developed out of the Qatabanian forms characterized by s. Apart from these peculiarities, we have to draw attention to a specific fact which, to us, seems to playa decisive part in the solution of this problem. That fact is that in the languages using ha (or a) in the causative, the reflexive stem with t associated with the causative stem is based on a causative stem characterized by a sibilant. Thus Sabaean, although its causative has the h-prefix, nevertheless builds its reflexive-causative stem on a causative stem with s: stqtl. In this respect, Sabaean agrees with North-Arabic (,Classical' Arabic): aqtala ( < haqtala): istaqtala. Indeed, a reflexive stem based on a causative stem with h does not exist in any language. 2 The association, in Sabaean and 'Classical' Arabic, of the causativereflexive stem stqtl (istaqtala) with the causative hqt/ (in 'Classical' Arabic developed into' qtl, aqtala) is, we feel, the decisive criterion concerning the problem of the original structure of the causative-prefix in Semitic. The association of stqtl (istaqtala) with hqtl (haqtala> aqtala) shows that, originally, there cannot have existed various causative-prefixes-that is: ha, a and s/sa-but only a single prefix characterized by a sibilant: s/sa, which, through phonetic development, was changed into ha (which, in turn, in various languages and dialects, evolved into a). 2 The later Aramaic dialects, which form the causative stem as a{ el, form also a causative-reflexive stem, which is based on afel, that is i/ettafal < *i/et'afal. Since an analogous form (based on haf el) in the older dialects (Biblical-Aramaic and Egyptian-Aramaic) is missing, and since afel is a development from hafel (see below), i/ettafal < *i/et'afal should be considered as a relatively late secondary formation constituted in analogy to the reflexive stems associated with the 'basic' stem and the intensive stem (i/elpe'el and i/elpa"al).

202

THE SEMITIC CAUSATIVE-PREFIX SISA

In coming to this conclusion, we are guided by the specific character of the phonetic sequence st. We assume that the shift of the sibilant s (or s) to h was possible only in cases in which it was followed by a vowel, and that it could by no means have occurred where it was followed by the (voiceless) stop t. With respect to our .contention that the s in stqtl (istaqtala) could not undergo the shift to h (which did take effect in haqtala from s/saqtala), we draw attention to a certain pertinent phonetic development in Greek. In Greek, ancient (proto-Indo-European) s has to a large extent been shifted to h. E.g., Sanskrit saptd, Latin septem 'seven' appears in Greek as emu; etc. However, the s escaped this shift to h in all instances in which it was followed by an unvoiced stop, e.g., 110"'tat = Sanskrit aste 'he is seated'. The preservation of s in istaqtala can be compared directly with its preservation in Greek 110"'tat whereas its shifting to h in haqtala (from s/saqtala), or in the pronouns hu(')a, hi(')a from s/su(')a, s/si(')a, can be compared to E7nu (= Sanskrit sapta, Latin septem), etc. H should be assumed that, originally, the phonetic shifting of SIs to h in the causative-prefix did not affect all instances in which the prefix appeared, and that the exclusive use of ha instead of s/sa (in the languages in which the shift took effect) is to be ascribed, in part, to the force of analogy. The element ha, which, through phonetic development, had in a good many instances (but not in all) evolved from the element s/sa, was transferred to other instances by the force of analogy. This gradual expansion of the new prefix (by the force of analogy) explains the fact that we find the S;sa-prefix preserved in a number of so-called 'secondary' roots, i.e., in roots based on original ones consisting of only two 'solid' radicals in which the prefix of the verbal 'stem' (conjugation)-in our case: the element s/sa-has become part of the root itself constituting its third ('solid') radical. Cf., e.g., Arabic sabaqa 'to leave behind, to overtake' = Aramaic sel]aq 'to leave', which constitutes an original causative of the (Arabic) verb (in the 'basic' stem) baqiya 'to remain, to stay behind'; etc. In instances like this, where the original prefix s/sa had been incorporated into the root itself and was no longer felt as a prefix, the sibilant of the original prefix remained preserved, not becoming exposed to the analogy process which generally replaced sis by h in instances in which it had not been shifted to h by the phonetic process, which constitutes, of course, the basic cause for the replacement of s/sa by ha.

THE SEMITIC CAUSATIVE-PREFIX SISA

203

We mention here as an interesting, hitherto unrecognized, instance of this preservation of the original causative-prefix s/sa by its incorporation into the basic root and becoming part of it, the Arabic root sor, used in the 'basic' stem: sabara, as well as in the intensive stem: saboara, and possessing in both stems the meaning 'to overthrow (someone), to constrain (someone) to forced labour, to subdue, humble (him)'. The Arabic root has a parallel in Syriac. We quote Brockelmann, Lexicon Syriacum 2 , p. 771: "sel:zar (mand[aice] shr, ar[abice] sor 'subegit' ... ) 'timuit' ... Parel, i.e. sal:zl:zar] 1. 'anxit' ... , 2. 'domuit' ... ". We recognize in (Arabic) sabara an original causative of oarra 'to fall down, to prostrate oneself, a verb which has a parallel in (Old) Egyptian: or. The Egyptian verb appears also in the causative stem (formed with the prefix s): sor 'to overthrow (someone), to cause (him) to fall down, to subdue (him)', and it is with this Egyptain causative that we equate the above-mentioned Semitic etymon, Arabic sor, Syriac sl:zr. As mentioned above, the Arabic verb appears in the 'basic' stem as well as in the intensive stem without any appreciable difference in meaning. The action described by the verb is, of course, of such a nature that the use of the intensive stem (beside the 'basic' stem) is easily accounted for. With this Arabic saooara 'subegit' (synonymously used with sabara) we directly identify Syriac sal:z/:zar 'anxit, domuit'. The transitive verbal meanings 'anxit, domuit' easily suggest themselves to the speaker as constituting causatives of a meaning like 'timuit' and thus-since the so-called intensive stem in Aramaic, like in Akkadian, Arabic, etc., is also used in the function of causativecould easily produce a 'basic' stem expressing the meaning 'timuit'. The 'basic' stem is consequently in this instance not the basis of the so-called 'derived' stem (in the instance under consideration the intensive stem, in causative function), but, on the contrary, the socalled 'basic' stem, with respect to its specific meaning, is developed from the 'derived' stem. We stress once more that the developmental sequence of the Semitic causative-prefix is s/sa> ha > a. As to the development ha> a, one may be inclined to assume that the h of ha, in case it was preceded by an additional preformative ending in a vowel, disappeared, e.g. yu-haqtil> *yu-aqtil. The form without h which thus developed, could subsequently have superseded all other forms of the causative which originally began with h. 3 Another possibility is that the h of the 3 cr. H. Bauer and P. Leander, Grammatik des Biblisch-Aramiiischen (Halle 1927), § 17 r (p. 62).

204

THE SEMITIC CAUSATIVE-PREFIX S/SA

causative-prefix disappeared spontaneously: ha > a, and not in connection with any special conditions, as, e.g., in its position between two vowels, i.e., when fulfilling, as it were, the function of a 'glide'. It is this second development, the spontaneous disappearance of the h, which seems to us to be the correct interpretation. But how should the spontaneous disappearance of the h of the causative-prefix be accounted for? We feel that the disappearance of this h is a consequence of the functional weakness of this prefix. In other words: the weakness of the phonetic structure of the causative-prefix ha (> a) is a result of its purely formal function which was fully conveyed by an element of weaker phonetic structure, requiring a lesser amount of articulatory effort. In analogy to the development of the causative-prefix ha into a, we explain also the evolution of the original causative-prefix s/sa into ha (a form which constitutes the basis of the transition just interpreted by us), as well as the transition of the original sibilant sis into h in the case of the demonstrative/personal pronouns discussed above. As to the possibility of such a shift of s or s to h, it is not only proven by the shift, referred to above on p.202, of proto-IndoEuropean s to h in Greek (as well as Iranian and Armenian), and also by the related shift off to h in Spanish, but the shift of s to h is well-known from Semitic itself. In Mehri, a modern South-Arabian language, Semitic (Arabic) s appears generally shifted to h, e.g., hOba' 'seven' from sab' (cf. Greek brra, as against Sanskrit saptd, Latin septem); etc. It is with this (comparatively recent) general transition of s to h in Mehri that we identify the transition in early Semitic of sis in the causative-prefix as well as in the demonstrative/personal pronouns. It is evident that the general transition of a sibilant (or any other unvoiced spirant of a kindred nature, as f) into h is caused by a striving for a lesser amount of articulatory effort, in more general terms: by the striving for ease (or lessened effort), that influence to a considerable degree not only phonetic development but linguistic development in general (including syntactic development). If we now find the transition of a sibilant to h in early Semitic restricted to a few elements of formal function-i.e., the causative-prefix and the pronouns-we may safely assume that the lessening of the articulatory effort, which is the direct cause of the transition of the sibilant to h, was brought about by the 'functional weakness' by which the elements involved had been affected. This 'functional weakness' should be regarded as a

THE SEMITIC CAUSATIVE-PREFIX S/SA

205

consequence of the extremely frequent use of these elements as well as of the decidedly general, non-specific character of the concepts expressed by them (cf. above p.203-204 our remarks concerning the transition of the causative-prefix ha to a).

B

SYNTACTIC AND SEMANTIC STUDIES

16 GENETIC ASPECTS OF THE GENITIVE IN THE SEMITIC LANGUAGES 1 In many Semitic dialects, a relative pronoun which precedes a noun characterizes the latter as being in the genitive case. We find this construction especially frequently in Accadian, Aramaic and Ethiopic. 2 The pronominal particles used in this construction have been conceived as relative pronouns by most earlier grammarians. Philippi 3 has attempted to prove this interpretation as erroneous. On the basis of various considerations, and especially with a view to similar constructions in Arabic, he came to the conclusion that these pronominal elements were actually used in this construction as demonstratives. In this connection, he also concluded (I.e., p. 113) that the paraphrasing of the genitive by the relative pronoun plus Ie in later Hebrew and the similar formation of the possessive pronouns in Aramaic, must be considered as a separate phenomenon, different from this alleged construction with demonstratives. Brockelmann (Grundriss, I, p. 243) follows Philippi in the interpretation of these particles as demonstratives. He declares (Ibid., p. 244) the construction with a relative pronoun plus Ie as a secondary interpretation by the speaker of the original construction with a demonstrative. 4 Barth (I.e., p. 166) rejects the interpretation of the particles as demonstratives (at least in most instances where they occur) and insists on their interpretation as relatives. In Classical Arabic, the pronouns in question occur in expressions which may be defined as so-called free or independent genitives (genitives without an antecedent).5 Old-South-Arabic shows similar 1 Pages 209-225 were originally published in Journal of the American Oriental Society, vol 81 (1961), p. 386-394. 2 For details see C. Brockelmann, Grundriss der vergleichenden Grammatik der semitisehen Spraehen, II, p.243ff., and J. Barth, Die Pronominalbildung in den semitischen Spraehen, p. 166ff.-In the meantime, Ugaritic has been added to the dialects in which this phenomenon can be established. 3 F. W. M. Philippi, Wesen und Ursprung des Status eonstruetus im Hebriiischen, ein Beitrag zur Nominalflexion im Semitisehen iiberhaupt (Weimar 1871), p. 112ff. 4 " ... als selbstiindiges Possessiv mit Ie und Suff.... wird es [i.e., the demonstrative) schon als Relativ empfunden". 5 The forms of the pronouns are here declined like nouns (sing. gu/i/ii; plur.

210

GENETIC ASPECTS OF THE GENITIVE

instances. Here too, as in the non-Arabic languages, an antecedent ordinarily precedes the genitive. Barth considers the particles in these instances in Classical Arabic and Old-South-Arabic not as relative pronouns, but-with Brockelmann and Philippi-as demonstratives, in contrast to his interpretation of these pronouns in other dialects. Reckendorf 6 voices a similar view: "gu, eigentlich 'der von ... "'. 7 Philippi (l.c., p. 117) voices the opinion, obviously shared by the other authors mentioned above, that the pronoun considered by him as a demonstrative is in a status constructus-relationship to the following noun. He substantiates this by interpretations as the following (l.c., p. 114): Syriac mellf dalliihii ('the words of God'): 'Worte, die von Gott, die auf Gott beziiglichen', Ethiopic 'aklil zawarq ('the crown of gold'): 'Krone, die von Gold'; and contrasts this interpretation with the one voiced by Dillmann and others who interpret such expresions, as, e.g., the last-mentioned Ethiopic example, as follows: 'Krone, welche ist Gold'. Against this view one must point out the following facts. In Semitic, a demonstrative cannot be followed by a noun in the function of a genitive. An Old-Semitic primitive genitive presupposes a preceding noun in the status constructus. Consistent with this is the fact that the Old-Semitic primitive genitive cannot be used independently in substantivized function, i.e., as a so-called free genitive. Quite apart from this, one must point out the fact that the Arabic particle gil-as well as similar particles in similar constructions in related languages-has no determinative (or demonstrative) power. pu also appears in expressions which are indeterminate; they are characterized by the fact that the noun depending on the particle appears in them without an article. Aramaic, especially rjawu/i, 'ulu/i), in agreement with the varying syntactic functions in which the expres-

sions, composed of the pronoun and the noun following it, are used. 6 Arabische Syntax (Heidelberg 1921), p. 151 (§87 d). 7 The same interpretation, "der von ... ", is also given by Maria Horner, A ItsUdarabische Grammatik (Leipzig 1943), p.42 (§ 37 d). She calls it a "demonstrativrelative Bedeutung". This somewhat fuzzy concept of a demonstrative-relative meaning, whatever its merits, is in any case not warranted here. The demonstratives expressed by the rj-element and the relatives expressed by the same element (and undoubtedly to be derived from the former) normally appear with very distinct and unequivocal meanings and syntactical functions, and in the case at hand only one unequivocal meaning can be assumed. It might be noted that the sentence adduced by M. Hofner (p. 44, n. I) as an instance which "very clearly exemplifies the transition from demonstrative to relative" is, in our opinion, not a suitable example for the purpose. The pronoun 'Iy = Classical Arabic 'uN, casus obliquus of the plural, can be considered here only as a relative pronoun.

GENETIC ASPECTS OF THE GENITIVE

211

Biblical Aramaic and Egyptian Aramaic, also uses the genitival expres~ions with di after an indeterminate antecedent,8 e.g., '?$be'iin di yag fnii§ 'finger of a human hand' (Dan. 5,5), be zi lignin 'a house built of bricks' (Pap. Sachau, 30,9). This usage should not be explained, with Brockelmann and Bauer-Leander, by the loss of an original demonstrative meaning of the element di. The genitive particle di, de in Aramaic (and its counterparts in the related languages) has as little a determinative function as the same element when used to introduce regular relative clauses. We assume that the pronoun in the construction under consideration must be considered as a relative pronoun, not as a demonstrative pronoun, in all dialects, including Old-Arabic. Another problem is the essence and origin of this use of the relative pronoun as a nota genitivi. Barth (l.c., p. 168) declares this use of the relative pronoun as difficult to explain, and adopts, although with some reservations, the above-mentioned explanations proposed by Dillmann and others: "Dieser Gebrauch des Relativwortes ist befremdlich .... Der Ausgangspunkt konnen solche Hille gewesen sein, in welchen der Genitiv bloss erkliirend war oder den Stoff seines regierenden Nomens angab, wie 'Flammen, welche Feuer (sind), Krone, welche Gold (ist)', d.h. 'Flammen des Feuers, Krone des Goldes'. Von hier aus konnte der Gebrauch dann auch auf aIle Genitivverhiiltnisse ausgedehnt worden sein. Doch ist dies nicht sicher". The juxtaposition of a noun with another noun in the genitive simply implies some relation between the two nouns. The nature of this relationship follows from the nature of the relationships possible between them. 9 The syntactical relationship which obtains between the noun governing the genitive and the genitive must be described as that of a noun and an attribute which qualifies it. Whenever the specific kind of relationship between a substantival attribute in the genitive and its antecedent is not clearly and unequivocally expressed, one might, in each individual case, conceive of a different construction which would unequivocally and concretely express this relationship. We actually find many relationships expressed by the genitive also expressed by prepositional phrases. 8

See Brockelmann, l.c., p. 244, and Bauer and Leander, Grarnrnatik des Biblisch-

A rarniiischen, p. 313. 9

This the genitival constructions have in common with the nominal compounds.

212

GENETIC ASPECTS OF THE GENITIVE

We derive the genitive characterized by the relative pronoun from relative clauses whose subject is expressed by a relative pronoun, the predicate by a prepositional phrase. All prepositions could appear in these primary constructions of which many instances exist in the Semitic languages known to us. Since all prepositions could be used in constructions of this kind, the number of relationships expressed by such constructions was naturally very great. All these prepositions could subsequently be eliminated by a process described below. Through this elimination, one uniform type of a substantival term expressing relationship, i.e., a so-called genitive, came into existence. This type could subsequently be used to express all attributive ('genitival') relationships of substantive nouns to other ('governing') substantive nouns, including those of a non-adverbial character. Thus an absolute form of a noun expressing relationship had come into existence, comparable to the primary genitive. The new form was different from the primary genitive in the Semitic languages in this respect: its use was not dependent on a preceding noun in· the status constructus, and thus it could also be used in a 'substantivized' function (i.e., without another 'governing' noun). To illustrate the development just sketched, we quote some passages in which an Arabic expression for 'travelling allowance, provisions, food' appears in variants. I:lassiin b. Iiibit (ed. Hirschfeld), no. 8,5: wabagaltu gii ral)li ... 'and I have generously given up the contents of my saddle-bag (that is, my provisions) ... '. AI-I:lutai'ah (ed. Goldziher), no. 54,3: Jatan yabgulu gii qidrihi 'a man who liberally passes out the contents of his kettle'. Ibid., no. 37,3: al-gawiidu bigi l-faluri 'he who is generous with the contents of the leather cover (which is spread out on the ground and serves as table),. In the case of these three expressions, which are composed of gu and the name of some receptacle, the context leaves no doubt that they all mean 'food'. But food is kept or served in or on containers of the kind mentioned; one therefore may assume that mentally the nouns for a 'receptacle' function as local adverbial expressions, although the linguistic terms do not express this function. Beside the above-mentioned phrases with gu plus a noun in the genitive, as in gil r-rabli 'the contents of the saddle-bag', we find another expression for the same idea: maJi r-rabli, literally: 'what is in the saddle-bag'. We quote the following examples: MuJarjfj.aliyat (ed. Lyall), no. 68,15: barjulun limaJi rablihi gairu zummabi 'one who generously passes out what is in his saddle-bag, no

GENETIC ASPECTS OF THE GENITIVE

213

miser he'.I0 'Agani, 14,84,21: bafJilan bima fi ra/:zlihi ... 'one who stints with what is in his saddle-bag ... '.11 In this latter form of expression, the concept of 'contents of the saddle-bag' is expressed by a regular relative clause which contains a local prepositional phrase clearly marked as such. For the above-mentioned expressions (as gu r-ra/:zli) we assume prototypes of an analogous structure. Thus we assume that in expressions of this kind, a preposition with the meaning 'in' or 'on' has dropped out in an earlier developmental stage of the Arabic language. Similarly, for expressions li~ Aramaic baita di mafka 'the house of the king', Biblical Aramaic b?'fsur di Jarz?f 'in a fetter of iron' (Dan. 4,12), Syriac belsuf qeff gaglila 'in every city of Galilee' (Luke 5,17), Ethiopic 'aklil zawarq 'the crown of gold', the following clauses must be assumed as prototypes: 'the house which [belongs] to the king (di rmalka)" 'in a fetter which oj iron (di min parz?l)', 'in every city which in Galilee (delzaglifa)', 'the crown which oj gold (za'amwarq)'. Analogously, we recognize the Hebrew substantivized relative clause 'as?r 'af habbayil (1 Kings 4,6) '(he) who over the house' (i.e., 'superintendent of the palace') as the approximate prototype of the Accadian 'free' genitive sa bit keli 'superintendent of a prison'.12 The Hebrew expresion may also be compared with Soqotri de dim in qa'er 'the master of the house',13 literally: 'he (more exactly: this) who is of (min) the house': the introductory demonstrative serves here as correlativum and imparts to the expression the determination which in itself it lacks (cf. above p. 211). In spite of its vague character, the form which thus came into existence would give no occasion to misunderstandings. The function of the genitive in each case is indicated by the noun-concept which it qualifies, as well as by the content of the entire sentence. The linguistic expression of the relations indicated in the original constructions by various prepositions was felt as superfluous; these pre10 We cannot accept Lyall's interpretation who translates: "Ready to give freely all that he had in his possession, no niggard ... ", with the explanatory note: "Ra!;l seems here to have the sense of 'abode, habitation', rather than 'saddle or saddlebag ", . 11 In al-Bal1igurl's ' Ansiib al-' asriij; 5, 115, 5, we find the following version: babilan bimii fi riglihi ... , with the interpretation by the editor: "He is chary of using his foot (or his shoe; i.e., he does not like to walk far) ... ". Rigl is undoubtedly an error here, instead of ra!;l. 12 Quoted by Brockelmann, l.c., p. 247, § 168 c. 13 D.H. Muller, Die Mehri- und Soqo!ri-Sprache, II: Soqo!ri-Texte (Vienna 1905), p. 116,6.

214

GENETIC ASPECTS OF THE GENITIVE

positions were therefore eliminated. One has to assume that their elimination was preceded by a stage in which, although they no longer had any function, they still continued to be used (cf. below p. 215). In the construction under discussion, we deal with a development in which various prepositional expressions which expressed a relationship clearly and unequivocally, were replaced by one, uniform kind of expression which was indefinite, abstract and vague. Furthermore, we assume that an attenuation of the function of prepositions can also be observed in some cases in which a prepositional phrase is attributively added to a noun without a relative pronoun (asyndetically). We refer to the Hebrew and Syriac instances discussed by Philippi, l.c., p. 57 ff., and by Brockelmann, l.c., p. 233, as Hebrew simbal baqq~ir 'the joy of the harvest', or: ' ... in the harvest' (Is. 9,2), hare lzaggilbO a • 'the mountains of Gilboa', or: ' ... in Gilboa' (2 Sam. 1,21); Syriac b eri/sal benesse 'blessed among the women' (Luke 1,28), etc. Philippi assumes that originally an ordinary status constructus-relationship had been intended here, and that later, for the sake of clarification of the relationship of the two expressions, a preposition had been interjected; therefore, we were dealing here with a blending of two manners of expression. Brockelmann assumes that the status constructus can be used before a prepositional phrase in the same manner as before a simple noun in the genitive. 14 E. Konig, (ZDMG 53, 521 ff.) ascribes this use of the status constructus, which originally is only possible before genitives, to analogy. To these Hebrew and Aramaic instances, we add the following examples from Arabic. In Arabic, a noun qualified by a prepositional phrase with baina 'between' can precede this phrase in the status constructus. 15 While Brockelmann and Reckendorf want to trace back this usage to the original nominal meaning of the preposition baina, Noldeke declares this construction to be a combination of a governing noun in the status constructus with a ?arJ, i.e., a local prepositional phrase. This definition is identical with the one given by Brockelmann and Kautzsch for the above-mentioned Hebrew and 14 A similar view is held by E. Kautzsch in Gesenius' Hebrew Grammar 28, 2nd English ed., by A. E. Cowley (Oxford 1910), § 130 a (p. 421): "The construct state ... is frequently employed in rapid narrative as a connective form, even apart from the genitive relation; so especially before prepositions ... ". 15 We refer to examples given by Noldeke, Zur Grammatik des c1assischen Arabisch (Vienna 1897), §48; Reckendorf, Arab. Synt., § 130, 1; Brockelmann, I.e., p.360.

GENETIC ASPECTS OF THE GENITIVE

215

Aramaic construction. The view that an adverbial (prepositional) phrase in an attributive function can be governed as a genitive by a noun in the status constructus, seems untenable to us. In most instances, the noun which precedes the prepositional phrase with baina, expresses-in accord with the meaning of this preposition -ideas like 'relations' (qarii'inu bainihim, ijamiisah, 489, v.7), 'sincerity' (n~i~atu baininii, 'Alqamah 1,2), 'separation' (tafriqu baininii, Ibn Qutaibah, as-Si'r was-su' arii', 391, 8 ;furqatu baininii, al-M ubarrad, Kiimil, 733,11), 'hate' (san'u baininii, ijamiisah, 279, v.2), etc., thus ideas in which the concept of mutuality is implied. We quote the following sentence (ijamiisah, 301, v.2): 'asiyata qa!!a'nii qariiyina baininii 'in the evening when we severed the relationship between us'. We assume that in primary expressions like al-qariiyinu bainanii 'the relations between us', or as-sarru bainanii ('Alqamah 1,5) 'the enmity between us', the preposition baina was affected by attenuation of its function, and thus an immediate relationship between the antecedent and the noun governed by the preposition came into existence. Expressions developed which must be understood as: 'our relations', or 'our enmity', respectively. This latter form of expression appears of course also in Arabic as an independent construction (not containing the preposition baina 'between'), identical in its meaning with the form of expression in which baina possesses its full function. Similarily, we assume a loss of function of the prepositions in the above-mentioned Hebrew and Aramaic instances. Accordingly, in cases like sim~al baqq~ir and hiir~ Qaggilbii"', we assume that they originate from basic forms like ha.ssim~at- baqq~ir 'the joy in the harvest' and hfhiirim baggilbii"' 'the mountains in Gilboa', and that they are approximately equivalent to expressions with a regular genitive, as sim~al haqq~ir 'the joy of the harvest' and hiir~ haggilbo a ' 'the mountains of Gilboa'. We thus assume for these instances the same development which we presuppose for the genitive characterized by a relative pronoun. Whereas in this case the original prepositional phrase is introduced by a relative pronoun, the instances just discussed lack a relative pronoun. Both constructions were affected by the same occurrence: the loss of function of the prepositions. In the first of these types, characterized by a relative pronoun, the preposition itself finally disappeared. In the second type, the type without a relative pronoun, the preposition continued to exist linguistically. The preposition which had lost its function, was in the basic con-

216

GENETIC ASPECTS OF THE GENITIVE

struction not only related to the noun following it, but also to the one preceding it, since it introduced an essential qualification of the latter. It effected a connection between the two nouns. The loss of its function could thus lead to a moving together of the two nouns. In the process, certain unessential elements of the first noun disappeared. The closer connection of both nouns thus spontaneously produced in the first noun a 'status constructus' form. It suggests itself to compare this closer connection of two nouns developed in some Semitic languages with the status constructusrelationship which originated from Proto-Semitic, and to draw some conclusions as to the origin of the latter. We must consider the status constructus-form as a characteristic which goes beyond what one would expect as absolutely necessary for establishing a relationship between a governing noun and another noun added to it as an attribute (that is, in the genitive). The function of the second noun as an attribute of the first-a function which in Proto-Semitic as well as in some historical Semitic languages was indicated by the ending i-can, by itself, not explain the phonetic unity of these two nouns. The conjecture suggests itself that the status constructus in Proto-Semitic and Common-Semitic is related to the loss of function and disappearance of prepositions originally present in the primitive constructions. The assumption that the relationship of annexation (status constructus) is a substitute for a lost preposition which preceded the second noun, would also explain for us the fact that the Old-Semitic genitive cannot occur independently, without a governing noun. For a long time a genetic relationship has been assumed between the genitive-ending -i and the ending -/ (-iy) of nomina (adjectiva) relativa formed from nomina substantiva (mostly proper names). While Philippi (i.e., p. 194) derives the adjective-ending i (-iy) from the genitive-ending, Brockelmann (i.e., I, p. 460; II, p.239) considers the genitive-ending -i as originating from the former. He concludes from this presumed priority of the adjective formation and the secondary development of the genitive function which follows as a corollary, that the 'subordinate' attribute is younger than the 'coordinate'. The 'coordination' of the adjectivum relativum (nomen substantivum + -iy-), in contrast to the 'subordination' of the genitive (nomen substantivum + case-ending -i), consists in its case congruence with its antecedent (-iyu, -iyi, -iya). This purely formal difference does not entitle us to assume a contrast between 'coordination' and 'subordination' here. Case congruence between an attribu-

GENETIC ASPECTS OF THE GENITIVE

217

tive adjective and its noun antecedent is based on formal analogy. The essential point, however, is that both forms discussed here are identical in function and meaning; both are nouns which serve as attributes to other nouns. Whether we- to use the expression quoted above, hare f!aggilbO -say: 'mountains of Gilboa' or 'Gilboaish mountains', makes no difference. The adjective 'Gilboaish' implies a local reference-which might be illustrated by the preposition 'in' (,mountains situated in Gilboa')-in the same way as the genitive 'of Gilboa'. The nomen relation is-ending -IY- is as unable to express exactly the relationship of its noun to the antecedent as does the genitive, which in early Semitic is characterized by the ending -i. On the other hand, the genitive on its part, in the languages which have lost the ending -i, is equally well expressed by the status constructus of the antecedent (or by its substitute, the expression through a relative pronoun) as through the ending -i (in addition to the status constructus form of the antecedent). The explanation of the status constructus of the antecedent of a noun in the genitive must by all means take into account the instances of attributive clauses whose antecedent likewise appears in the form of the status constructus. This phenomenon which can be observed in almost the entire Semitic linguistic area has probably already existed in Proto-Semitic, and the individual languages must have inherited it from the primitive language. On the other hand, there can be no doubt that, in most cases, the construction was developed independently in the individual languages themselves, but very probably in their early stages of development. Contr'llry to prevailing opinion, we do not consider the use of the status constructus form in these cases simply as an extension of its use before a simple nomen substantivum in the genitive. Rather, we assume that the relationship of annexation (status constructus) developed in these cases independently. We ascribe the status constructus of the antecedent in these cases to the loss of function and disappearance of a certain element in the subsequent clause which had a relation to the antecedent. In many Semitic dialects, nouns which express a space or time concept, can precede a clause in the status constructus. Mostly-but not necessarily or exclusively-such a noun is used adverbially, that is, it is preceded by a preposition (like Hebrew be) or it is in the adverbial accusative. In Hebrew and Aramaic, 'Qs?r or de, respectively, can stand between a noun expressing time or space and the following Q

'

218

GENETIC ASPECTS OF THE GENITIVE

clause. 16 In an artificial, formally logical manner of interpretation, one ordinarily declares the clause following the noun in the status constructus as a substantive clause (equivalent to a nomen action is or an infinitive) in the function of a genitive. That is: one assumes that-in the same manner as in the Semitic languages a noun in the genitive is added to a governing noun in the status constructus-here a clause in the genitive has been subordinated to a governing noun in the status constructus. 17 We quote some instances of this construction: Arabic lailata $a/:lu 'in the night in which (or: when) they shouted' (Mufa#aliyat 1,5); 18 yauma laqau J)u'aibata 'on the day on which (or: when) they met with Ou'aibah' (Hzujail, ed. Kosegarten, 58,1); Hebrew beyom dibb?r Yhwh 'on the day on which (or: when) God spoke' (Ex. 6,28); kQI ye me hilhallalsnu 'ittam 'during all the days during which (or: when) we were conversant with them' (1 Sam. 25,15); qiryal /:lana J)awitj 'the city in which (or: where) David camped' (Is. 29,1); also-with inserted 'as?r-bimqom 'as?r higlu '010 'in the place whereto they have exiled him' (Jer. 22,12). On the assumption that the clauses dependent here on the nouns of time or space were from the outset substantive clauses in the genitive, these constructions would have to be interpreted as follows: 'in the night of: they shouted'; 'on the day of: God spoke'; 'during all the days of: we were conversant with them'; 'the city of: David camped'; 'in the place of: (that) they have exiled him'. Such phrases can only be considered as entirely inadequate and clumsy modes of expressing the ideas to which they point. It may

16 See Gesenius-Kautzsch, Hebrew Grammar, § 130 c, d; Brockelmann, I.e., p.620; Noldeke, Mandiiische Grammatik (Halle 1875), p. 451, n. 1. 17 Cf. Philippi, Der Status constructus im Hebriiischen, p. 77 ff.; Noldeke, M andiiische Grammatik, p. 451, n. I; Zur Grammatik des class. Arab., p. 106; A. Dillmann, Grammatik der aethiopischen Sprache (1st ed.; Leipzig 1857), p. 327 supra, p. 363 (§ 184), p.412 (§201); Gesenius-Kautzsch, Hebrew Grammar, § 130 c, d; Ungnad, ZeitschriJt fiir Assyriologie, 18, p. 60; Brockelmann, Grundriss, II, p. 532-38 (§343-347 a); Reckendorf, Arab. Synt., § 190,2. 18 Instead of 'in the night in which (or: when) they shouted' one may also say in English: 'in the night that they shouted', or even 'in the night they shouted'; or-to give a clearer example-: '1 recognized him at the moment that he came in', or: •... at the moment he came in'. These variations (especially the last one where the attributive clause is joined asyndetically) are identical with the Arabic construction inasmuch as the syntactic function of the antecedent within the attributive clause is linguistically not indicated. (The English mode of expression differs from the Semitic one only by the definite article of the antecedent: in the Semitic construction the determination of the antecedent is implied in the status constructus form).

GENETIC ASPECTS OF THE GENITIVE

219

be assumed that these modes of expression which seem to require such forced literal interpretations are not original, but are rather distortions of earlier and clearer constructions which were more in line with the ordinary usage of the Semitic languages. We assume that in the original constructions a (definite) noun (in the status absolutus) was followed by a relative clause-possibly introduced by a relative pronoun-in which the antecedent was referred to by a pronominal suffix dependent on a preposition (or, possibly, by a demonstrative adverb, like Hebrew sam 'there'). The adverbial phrase within the relative clause was identical in content with the antecedent; therefore, it was felt as superfluous and could be eliminated. Simultaneously with this loss of function and elimination of the adverbial phrase, a relationship of annexation developed between the antecedent and the relative clause. The status constructus form of the antecedent thus clearly manifests itself as a substitute for the eliminated adverbial phrase which had referred to it. 19 Through its status constructus-form, the preceding noun adds, to its original function as an antecedent, the function of the adverbial phrase in the following relative clause. This adverbial phrase, which denotes a different kind of relationship in each instance, is now, through this annexation, indicated in a general and abstract manner which, however, is felt to be sufficiently expressive of the intended idea. The syntactical relationship between the antecedent and the subsequent attributive clause in the instances under discussion now is not essentially different from the relationship between the antecedent (nomen regens) and its attribute in instances like (haS )simIJat- baqqi4ir (v. above p.215). In the latter case, the development which led to the status constructus form of the first noun has transformed the second noun into a genitive. We must assume that, also in the type 19 This syntactic type which may have developed already in Proto-Semitic and in the early developmental stages of the individual languages, very probably redeveloped independently in the later developmental periods of the individual languages. With regard to that possibility, it may be assumed that a relative pronoun between the antecedent (in the status constructus) and the attributive clause may have disappeared simultaneously with, or subsequent to, the development of the status constructus form of the antecedent and the elimination of the resumptive pronoun (or adverb) within the attributive sentence. This elimination of a relative pronoun under the described circumstances is to be compared with the elimination of prepositions between a noun (an antecedent) in the status constructus-or a relative pronoun-and a subsequent noun (an original prepositional phrase); see above p. 213 and p. 215. On the other hand, we refer to the lack (or elimination?) of a relative pronoun in the English construction referred to above n. 18.

220

GENETIC ASPECTS OF THE GENITIVE

now under discussion, the new linguistic structure implies the relation of a governing noun and a genitive. Essential, however, is the insight that a 'genitival' function of a clause dependent on a noun of space or time constitutes a secondary development. An expression like, e.g., Hebrew (Ruth 1,1) waihi bime sefot hassoPtim ... , literally: 'and it was in the days of the judging of the judges ... ', that is, ' ... in the days when the judges judged', is rendered in the Ethiopic translation as follows: wakona bamawii'ala yak"ennanu masiifant. This sentence exemplifies the construction under discussion: a noun of time in the status constructus followed by a clause which contains no prepositional phrase with a pronoun referring to the antecedent (or a temporal adverb of equal meaning and function). The expression of the idea as contained in the Ethiopic translation must be considered as earlier than its counterpart in the Hebrew original, where a noun of time in the status constructus is followed by a verbal noun (an infinitive) in the genitive. The Hebrew form represents the final stage of a development, while the Ethiopic variation must be considered as an intermediate stage. The infinitive, or verbal noun, sefot ('to judge, judging') implies a much more specific meaning than the one we associate with an ordinary genitive. This implied meaning has found complete linguistic expression in the basic construction which represents the preparatory stage of the Ethiopic version: 'in the days in which the judges judged'. What in the Hebrew version appears as a noun in the genitive, appears, in this basic construction, as a complete relative clause, with the function of an attribute of a noun ('in the days'). The transformation of a relative clause into a nominal structure in the genitive-by way of curtailing its linguistic expression, as extant in the Ethiopic transitional stagetook place in the same manner as we were able to establish for other, ordinary types of genitive (see above p.216). In these latter cases, the main component of the basic form was from the outset represented by a simple noun. In our present case in which we deal with a complete clause, the expression which we accept as a 'normal' genitive originated only after the transformation of a finite verb into an infinitive (verbal noun); '(in which = when) they judged' became a secondarily developed genitive: 'of judging' Wfo!}. This secondary development of a genitive may be compared with the transformation of the subject of '(in which) the judges judged', that is: 'the judges',2° into a genitive 20

Cf. the Ethiopic equivalent of the clause.

GENETIC ASPECTS OF THE GENITIVE

221

(genitivus subjectivus) dependent on sefot 'judging': '(in the days of) the judging of the judges'. The exact relationship between the secondary infinitive ('of judging') and its antecedent in the status constructus ('in the days') can be expressed by the following statement: 'and it was in the days in which the judging of the judges [took place]'. The case just discussed seems of importance to us for the understanding and the genetic derivation of the nominal relationship called genitive in general. Typically, we have derived this case (see above p. 216) from clause-like structures by way of elimination of a component, namely a preposition. In the specific instances just discussed, the development of a genitive from attributive clauses can be clearly recognized. In both cases-as can be especially clearly seen in the second case-the fonnation of the genitive was accompanied by the annexation between the attributive clause and its antecedent, that is, by the development of the status constructus-fonn of the latter. From this phenomenon, so frequent in many Semitic languages, we must not separate those relative clause-constructions in Accadian (Old-Babylonian) and Old South-Arabic in which nouns of whatever meaning and function precede the relative clause in the status constructus. A characteristic feature of these relative clauses is that they do not contain a pronoun referring to the antecedent (resumptive pronoun). Cases in point are the following Old-Babylonian instances (quoted from Ijammurabi's Code): awiit iqbu Iii uktin 'the word which he has spoken, he has not proven to be true' (§ 3); qisti sarrum iddinu 'a present which the king gave' (§ 34, 59); eqel usiikilu 'the field which he made graze' (§ 58, 75); kasap isqulu iliqi 'the silver which he has weighed out (i.e., paid), he should take' (§9,46); kasap i/qu 'the silver which he has taken' (§ 101, 11), etc. Ungnad (ZA 18, 59) assumes here primary constructions in which substantive clauses 21 are made dependent in the genitive-in the same manner as an ordinary noun-on a 'governing' noun in the status constructus. O. E. Ravn 22 agrees with Ungnad. He declares awiit iqbu as a primary construction which should be analyzed as follows: 'the word having to do with: he spoke' and considers this expression to be analogous with expressions like awiit misarim (ljammurabi's Code 21 'Substantivische Nebensatze', the counterpart of 'adjektivische Nebensatze' = relative clauses. 22 The so-called relative clauses in Arcadian, or the Accadian particle sa (Copenhagen 1941), p. 37.

222

GENETIC ASPECTS OF THE GENITIVE

XXVI, 64) 'word of righteousness', which, in his opinion, must be literally interpreted as 'the word having to do with righteousness', etc. This denies that the relative clauses which follow an antecedent in the status constructus are relative clauses. For the cases in which the antecedent is a noun of time or space, Brockelmann, in agreement with Philippi, Ungnad and others, assumed that the clause which qualifies the antecedent has the function of a noun in the genitive depending on the antecedent (cf. above p. 218). For the cases under discussion now, in which the antecedent is represented by nouns with other meanings, he rejects (l.c., p. 554) Ungnad's view of the qualifying clause as a genitive,23 and assumes that here "die enge auf dem Satzdruck beruhende Verbindung zwischen Leitwort und Relativsatz" has produced the same phonetic form (,Lautform') as in a case of a genuine genitive relationship, that is, the form of the status constructus, which he calls an 'allegro form' of the noun. The 'sentence stress' (,Satzdruck') which is here assumed to have brought about the close connection between the antecedent and the relative clause is an unclarified concept, unable to explain the status constructus form of the antecedent. In trying to explain this phenomenon in Accadian and Old-South-Arabic, one should not lose sight of the fact that in all Semitic languages an analogous phenomenon occurs, but only when the antecedent is a noun denoting time or space. This clearly indicates that no basic phonetic causes can be assumed here, rather that the origin of the status constructus is closely connected with the content of the dependent clauses. In cases like qiryal I}ana J}aw;g (see above p.218), the assumption is near at hand that the status constructus form of the antecedent is due to the fact that the relative clause lacks an adverbial phrase (bah 'in it' or sam 'there') which refers to the antecedent as generally required in the Semitic languages; or, to state it more exactly, that the annexation of the antecedent with the relative clause is a consequence of the elimination of the adverbial phrase which refers to the antecedent. The same naturally also holds true for similar cases in the Accadian language, in which especially asar 'in the place (where), and related expressions appear in the status constructus. 23 " ... die Annahme Ungnads ZA XVIII, 58, dass jeder asyndetische Relativsatz von seinem Leitwort im Genitiv abhiinge, [fiihrt] zu einer unvollziehbaren Vorstellung 00.' denn awat iqbu heisst ja weder 'das Wort dessen, der gesagt', ar. kalimatu man qala, noch 'das Wort: "er hat gesagt'" , kalimatu qala, sondem eben: 'das Wort, das er gesagt' . 00",

GENETIC ASPECTS OF THE GENITIVE

223

Since also in those instances in which the antecedent does not denote time or space, a pronoun referring to the antecedent is always missing, even where we would expect one (that is, when the verb of the clause governs an accusative object which is identical with the antecedent), then here, too, we are forced to the conclusion that the development of the status constructus is connected with the elimination of the resumptive pronoun (cf. p. 239, n. 14).24 Inasmuch as the just described and interpreted relative clauses of Akkadian are distinguished from the regular relative clause-type of Akkadian not only by the status constructus-form of the antecedent of the relative clause and by the lack of a 'resumptive' pronoun in the relative clause itself, but-in connection with these features-do also not contain a relative pronoun (sa), this may have to be explained by the possibility that this specific relative clause-type developed in a very early period of linguistic history, in which Akkadian did not yet (or not yet necessarily) use a relative pronoun as a characteristic of a relative clause. On the other hand, it is also possible that this specific relative clause-type may indeed have to be traced back to an original type which included the pronoun (sa). Indeed, the same process which caused the status constructus-form of the antecedent and the disappearance of the resumptive pronoun, may simultaneously have caused the elimination of the relative pronoun (sa). For the construction in which the noun in the status constructus preceding the relative clause denotes time or space, we have established that this secondarily developed construction has identical or comparable features with the relationship of annexation (originating in 24 Starting from cases in which a status constructus developed through elimination of a resumptive pronoun in the accusative or a prepositional phrase, the status constructus then necessarily became the normal form of the antecedent of relative clauses without a relative pronoun so that we find the status constructus also in cases where a resumptive pronoun never existed: bit imqutu 'the house that collapsed' (§ 232, 92). - The development described above may be assumed to have happened in a very early period of the linguistic history of Accadian, a period in which-in agreement with what was evidently the regular Proto-Semitic usage-relative clauses appeared normally in an asyndetic construction, i.e., were not (or not necessarily) introduced by a relative pronoun. On the other hand, even the regular relative-clause type of Accadian, introduced by the relative pronoun sa (and containing a pronominal reference to the antecedent [a so-called 'resumptive' pronoun]), may have undergone the development which resulted in the type awat iqbu. That is: the relative pronoun sa, together with the 'resumptive' pronoun, may have disappeared in consequence of the development of the relationship of 'annexation' between the antecedent and the relative clause (awatum [stat. absol.] > awat [stat. constr.], and iqbusa > iqbu [through loss of the 'resumptive' pronoun -sa); cf. above p. 219, footnote 19.

224

GENETIC ASPECTS OF THE GENITIVE

Proto-Semitic) between a governing noun in the status constructus and the dependent noun in the genitive. The difference between the latter and the former construction lies mainly in the fact that in the latter the genitive is represented by a noun, in the former by a verb (replaceable by a verbal noun). Also in the instances of Accadian and Old-South-Arabic in which the antecedent does not signify time or space, the subsequent relative clause can be interpreted as a genitive. As an expression like 'the city in which David camped' can be interpreted as 'the city of David's camping', so it must be considered as possible, to understand the statement 'the silver which he weighed' as 'the silver of his weighing', or 'the word which he has spoken' as 'the word of his speaking'. In both cases, the entity expressed by the governing noun (that is, the status constructus), has the (additional) function of a (logical) complement of the entity expressed by the governed noun (that is, the genitive). While in the former case we deal with an indirect, adverbial complement, it is in the latter case an accusative object of the governed noun. We indeed find analogies of the latter case (theoretically constituted by us) in actual language. In expressions like 'the topic of our discussion' or 'the man of her choice', the concepts 'topic' or 'man', respectively, are objects of 'discussion' or 'choice', respectively. These expressions can be restated in a more explicit way: 'the topic which we discuss', or, 'the man whom she has chosen'. We have thus established that in all cases in which the antecedent of an attributive relative clause appears in the status constructus-not only in those cases where this antecedent expresses time or space-the relative clause is on the way towards becoming a genitive, a genitive of quite a special nature, one more relationship in addition to the many others which can be expressed through this case. We contrast once more the main patterns of relative clauses and substitute constructions discussed above. We find, both in Semitic and non-Semitic languages, on the one hand, certain meanings expressed by a governing noun plus a nomen rectum in the genitive, as 'the city of David's camping' (or: '[David's] camping-site', a compound), also: 'the days of the judging of the judges'. On the other hand, we find the same meanings expressed by relative clauses of the following type: 'the city in which David has camped', 'the days on which the judges judged'. Similarly, we find the expression 'the topic of our discussion' along with 'the topic which we discuss'. In the Semitic languages we find for such meanings also expressions like

GENETIC ASPECTS OF THE GENITIVE

225

Hebrew qiryal bana !)awig, Ethiopic wakona bamawa'ala yaklVennanu masafant, Accadian kasap i/qu. These latter constructions have some features in common with the two constructions mentioned above: with the genitive construction they have in common that the noun which functions as antecedent appears in the status constructus form; with the relative clause construction they share the feature that the attribute which follows the antecedent contains a finite verb. This attribute clearly reveals its origin from a relative clause. This clause must, in accordance with common Semitic usage, have originally contained a pronominal reference to the antecedent. The disappearance of the latter and the rise of the status constructus form of the antecedent must be considered as the expression of one and the same psychological process. Certain as it seems that the primary constructions which underlie the above-discussed expressions contained no genitive, we cannot deny, on the other hand, that the secondarily developed constructions represent genitival constructions. This is to say: as in the types discussed in the first part of this study,25 so here, too, ordinary attributes, or more exactly, attributive clauses, have developed into genitives.

Supplement The Development of the Inflnitivus Construetus of Hebrew and Related Phenomena

In our preceding discussion (above p. 209ff.) we described various primary constructions in which the development of a genitive case may be observed. We noticed, e.g., the possibility that a finite verb (originally) serving as an attributive clause (relative clause) to a nominal antecedent (a nomen substantivum) develops into a clause dependent on the nominal antecedent in the genitive-the nominal antecedent hereby acquiring the form of the status constructus (see, e.g., above p. 216). This development of the original attributive clause into a genitive may, however, simultaneously lead to the development of the finite verb (constituting the central term in the original attributive clause) into an actual noun (more exactly: a nomen actionis, 25

See above p. 211 and p. 215.

226

GENETIC ASPECTS OF THE GENITIVE

an infinitive). We mention as a very characteristic instance of this type of development once more the Hebrew sentence quoted above p. 220 (Ruth 1, 1): wayhi bime s'iO! hassoje!im .... The infinitive (corresponding to a finite verb in the Ethiopic translation of the Biblical verse, see above) appears here in the characteristic form of the so-called infinitivus constructus of Hebrew, that is: in the form qetol (sejot). The identity of this Hebrew infinitive-form with the stem of the imperfect is generally assumed (see J. M. Sola-SoU:, L'injinitive semitique, Paris 1961, p. 72). Our contention is that this form of the Hebrew infinitive, that is: the infinitivus constructus qe!ol (and variant-forms: qe!al, etc.) developed directly from the 3rd pers. sing. masc. of the imperfect (yiq!ol, yispo!, etc.), and that this development took place precisely in the above-mentioned sentence-type (wayhi bime s'iO! hassojetim) as well as in related sentence-types. That means: we assume that Hebrew originally possessed a syntactic structure of the type *wayhi bime yispetu hassoptim, corresponding literally to the Ethiopic version of this Biblical phrase: wakona bamawii'ala yakWennanu masiifant (see above p.220); or-in a more elementary form, with the agens (subject) of the attributive clause (relative clause) in the singular-*wayhi bime yispo! hassoje!, which may be interpreted by: 'it was in the days when (or: in which) the judge was judging', or-in a certain respect more in agreement with the actual syntactic structure of the sentence (in which the word for 'days' appears in the status constructus)- 'it was in the days of the judging of the judge'. Cf., e.g., the Hebrew phrase quoted above (p. 218): qiryall:ziinii J)iiwiff 'the city in which David camped' (quasi: 'the city of David's camping'). We assume that this structure (reconstructed by us)-*wayhi bime yispo! hassoje!-in which a finite verb (yispot), i.e., an original relative (attributive) clause, depends on a noun in the status constructus (wayhi bime ... ), developed into the form *wayhi bime s'iO! hassoje! (in the actually existing Biblical verse it reads: wayhi bime 5'iO! ha55oj"!im). That is: the finite verb-in agreement with the noun in the status constructus preceding it-became transformed into a noun, an infinitive in the genitive. And also the noun following this newly developed noun (more exactly: infinitive)-a noun which originally had the function of subject (in the nominative) to the (primary) finite verb-acquired in due course the syntactic function of a genitive, a genitivus subjectivus (depending on the infinitive preceding it). The sentence based on an original syntactic entity of

GENETIC ASPECTS OF THE GENITIVE

227

the meaning: 'and it was in the days when (or: in which) the judge was judging' thus expressed (quasi) literally the meaning: 'and it was in the days of the judging of the judge'. This syntactic structure was the ultimate result of the development described above p. 220. Accordingly, the Hebrew infinitivus constructus of the form qetol (also qetal) may be considered as having developed out of a finite verb (in the imperfect), a finite verb occurring in subordinate clauses (especially relatives ['attributive'] clauses). The transformation of the original finite verb (of attributive function) into a noun (an infinitive) in the genitive caused the elimination of the 'personal' prefix y(O of (the 3rd pers. masc. of) the imperfect. Thus (e.g.,) yispot developed into *spot> sejot (yiqtol> *q!ol> qetol); yiskab developed into *skab > se/saQ ; etc. H should, of course, be assumed that the original 'nomen regens' ('antecedent') of this type of attributive (relative) clause (which at the end developed into a noun [an infinitive] in the genitive)- ... sejO! hassojet < ... yispo! hassoje!-appeared from the outset in the form of the status absolutus (quasi in the form [wayhi] bayyamim ... instead of [wayhi] bime ... ), and that it acquired the form of 'annexation' (status constructus) in agreement with the change in the inner structure of the original attributive clause following it (as described), that is: by the transformation of this attributive (relative) clause into a genitive clause. The development described above, which involves the development of a finite verb (an imperfect) into an infinitive (wayhi bime yispo! hassoje! > wayhi bime s'lO! hassojet)-this development was apt to affect basic constructions (containing finite verbs [imperfect forms]) of various types. We mention in this connection a very interesting infinitive-form of 'Neo-Hebraic' (the language of the Misna and related literary monuments), as, e.g., lega' 'to know' (perfect yiitja'), lereg 'to descend' (perfect yarad), lele/s 'to go' (perfect hiila's, imperfect yele/s), etc. Brockelmann, Grundriss, I, §286 f,~, Anm. 4 (p. 601), interprets the development of these infinitive-forms (always preceded by the preposition Ie 'to') as follows: "Die alten Inf. wie SfQfl ersetzt das Neuhebr. durch Analogiebildungen nach dem starken Verb (s. §267 c), wie lega' 'zu wissen', lereg 'herabzusteigen'''. In contradistinction to Brockelmann's explanation, we consider these infinitiveforms of roots primae radic. y (and other specific root-categories, as, e.g., roots primae radic. n) as having developed directly from the original imperfect-forms of the roots concerned. We assume that

228

GENETIC ASPECTS OF THE GENITIVE

in the original Hebrew construction (forming the basis of the combination Ie [Ii] plus infinitive), the preposition Ie (Ii) preceded the verbum finitum (imperfect) in the same manner as in (Classical) Arabic the preposition Ii can precede a verbum finitum, as, e.g., li-yaktuba (li-yaqtula). The basic meaning of Ii in this usage is 'for the purpose of', 'in order to', as, e.g. (see Reckendorf, Die syntakt. Verhiiltnisse des Arab., p. 731), lamma rjahabna Ii-na'bura I-gisra 'when we went off in order to cross the bridge', etc. But such clauses with Ii plus verbum finitum are also used after verbs expressing the meaning of ' to desire', or 'to command', as, e.g. (see Reckendorf, I.c.), uridu Ii-ansa 'I want to forget', or amarta man kana ma?liiman Iiya'tiyaka 'you commanded someone who was wronged, that he should come to you', etc. In these latter sentences, the clause constituted by Ii plus verbum finitum obviously fulfills (normally) the function of an object clause (depending on the verb of the main clause), not (or: no longer) the function of an adverbial clausea genuine 'purpose' -clause ('finalis')-as in the first instance quoted by us (lammii rjahabna Ii-na'bura I-gisra). This use of original 'purpose'clauses (adverbial clauses) in the (certainly) secondary function of object clauses was discussed by us in our Studies in Arabic and General Syntax (Cairo 1953), § 109-112 (p. 136-138). In the same context (see Ibid.) we dealt also with the use of certain phrases constituted by Ie (Ii) 'to' plus infinitive, phrases which in Hebrew and Aramaic (and sporadically in modern-Arabic dialects, see I.c., p. 137, at the bottomp. 138) appear in constructions which in their meaning and function correspond to the above-mentioned (Classical-)Arabic constructions with Ii plus verbum finitum. Our contention (in our earlier publication) was that the construction with Ie (Ii) plus infinitive (characteristic of Hebrew and Aramaic) is to be considered as having developed from a more original construction constituted by Ii (Ia) plus verbum finitum. This means: we asserted that Hebrew and Aramaic originally used the same type of construction as Classical Arabic for the expression of a 'purpose' -clause ('finalis') and secondarily for the expression of an object-clause after verbs of 'desiring', 'commanding', etc., that is: the preposition Ie (Ii) followed by the verbum finitum (in the imperfect). We may assume that this syntactic pattern has lived on (or has recurred) in the language forming the basis of the specific Neo-Hebraic dialect represented by the language of the Misna and related literature (cf. the instances quoted above p. 227). Accordingly, this original pattern of construction, in which a verbum

GENETIC ASPECTS OF THE GENITIVE

229

finitum depended on the preposition Ie (Ii) 'to'-a term normally followed by a nomen substantivum-was affected by the same transformation by which the construction (treated above p. 226ff.) *wayhi bime yispot haisojet > wayhi bime s,,/ot hassojet was affected. That means: the verbum finitum acquired the character of an infinitive (i.e., the character of a noun). A basic form like, e.g., le_yispot 'so that he may judge' thus developed into lispot '(in order) to judge'. That is: this transformation of the original verbum finitum (following Ie) into an infinitive caused the disappearance of the 'personal' prefix of the original finite verb-form (3rd pers. masc. of the imperfect); cf. above p. 227. However, in case that the verb used in this type of (original) 'purpose'-clauses was a root of the specific type mentioned above, that is: a root primae radic. y (j) - in this case, the original 'personal' prefix y( e), as, e.g., in yega' 'he knows', ydeq 'he sits', yelels 'he goes', in spite of losing its original character (by the finite verb becoming transformed into an infinitive), remained (phonetically) more or less preserved. That means: forms like Jeseq '(in order) to sit', lereg '(in order) to descend', lelels '(in order) to go', etc., are to be derived from *leydeq 'so that he may sit', *leyereg 'so that he may descend', *ryelels 'so that he may go', etc. Accordingly, the original 'personal' prefix ye « yiy) coalesced with the reduced vowel Wwa) of the preceding preposition into the phonetic sequence Ie (leseq, lereg, lelels, etc.). Moreover, also in forms based on roots primae radic. n (and analogous categories), as, e.g., litten 'to give' (perf. nalan, imperf. yitten), liqqaf:z 'to take' (perf. laqa/:l, imperf. yiqqab), the direct development out of r plus imperfect (3rd pers. sing. masc.) is still recognizable: *leyitten > litten, *leyiqqaf:z> liqqa/:l, etc. The letter y (yog) after the I (lameg), with which litten and liqqa/:z are spelled, cannot be considered a simple mater lectionis, but certainly constitutes the original consonant y of the basic form: *ryiqqa/:z > liyqqa/:z > /iqqa/:z (> liqqab), etc. Of course, as already indicated above, the same fusion of the preposition Ie with the 'personal' prefix of the 3rd pers. sing. masc. of the imperfect may have occurred in case that the verb concerned was constituted by a regular, 'sound' root, like Sp! 'to judge' (q!l). That means, we may assume the following development: *leyispot > *liyspot> lispo! (*leyiq!ol> *liyq!ol> liqtol), etc. It was only natural that from this ultimate form (constituting a 'purpose'-clause ['finalis']) an infinitive-form *spot> sefot (*qtol> qetol) was 'abstracted', a form which subsequently was generally used in the function of infinitive.

230

GENETIC ASPECTS OF THE GENITIVE

Naturally, the development of the infmitivus constructus of Hebrew, qe!ol, qe!al, from the imperfect, yiq!ol, yiq!al, may have taken place in syntactic constructions of various types (cf., e.g., above p. 226, line lOff.). An analogous development of infmitives may have frequently occurred in Proto-Semitic as well as in the various individual (historical) Semitic languages. We mention in this connection the infmitive-pattern of original roots primae radic. w (> y) in Hebrew, which is constituted by a monosyllabic stem augmented by the feminineending (a )t, as, e.g., s~l]~l '(to) sit' (perfect yiisal]; imperf. yesel]); «:rj~l '(to) give birth, bring forth' (perfect yiilag, yiilerjii; imperf. yelerj, telerj); da·al '(to) know' (perfect yiirja·; imperf. yerja·), etc. The regular (basic) vowel of the 'stem' of the infinitive (in most of these forms) is the vowel i; cf., e.g., the forms with suffixes, as sil]ti ('my sitting'), lirjti ('my giving birth'), etc. This infinitive-pattern is, as is generally assumed, of Proto-Semitic origin. This is clearly proven by certain common-Semitic nouns of this type, as, e.g., Arabic lidat, Ethiopic ladat, Hebrew lerjii (also l~rj~l [secondary development]), Akkadian littu (see Brockelmann, Grundriss, § 121 c [po 338]). (In this latter noun, as in others of the same pattern, the original abstract meaning of the noun [nomen verbi, infinitive] has partly-in some of the languages concerned-developed into a concrete meaning, a semantic change frequently affecting original infinitives). As far as Hebrew is concerned, an analogous infinitive-pattern is formed of roots primae radic. n, as tel '(to) give' < *titt < *tint (perfect niilan, imperf. yitten), with suffixes: tilti ('my giving'), etc. Of the same type is Hebrew (lii-) qal:wl '(to) take' (perfect liiqai} , imperf. yiqqai}), with suffixes: qai}ti ('my taking'), etc. Brockelmann (l.c.) states that the Common- and Proto-Semitic pattern lidat (etc.) represents 'theoretically' a pattern qUi! (augmented by the feminine-ending). As already hinted above, we assume that this infinitive-pattern, characterized by the stem-vowel i, developed, in Proto-Semitic as well as in early Hebrew (Proto-Hebraic), from the finite verb, more exactly from the imperfect (3rd pers. sing. masc.), of the roots concerned. We consider this development largely analogous to the development (described above) of the regular infinitivus constructus of Hebrew, as, e.g., s'iO! « *spof) < yispot. On the other hand, this infmitive-pattern of roots primae radic. w (> y) may be directly compared with the Neo-Hebraic infinitive-pattern of roots primae radic. w (> y) and primae radic. n, which developed from the forms of the imperfect preceded by the preposition Ie (see above

GENETIC ASPECTS OF THE GENITIVE

231

p. 227). Accordingly, we assume also here-in the case of instances like Iidat- (etc.)-that the development consists in the transformation of an original (finite) verb into a nomen substantivum (as, e.g., in the case described by us above: *wayhi bime yispo! hassoje! > wayhi bime '" SPO! (> s'iof) hassojef). Our derivation of these infinitives of roots primae radic. w (as Iidat-, etc.) from the imperfect is based on our assumption that the imperfect of these roots in Proto-Semitic did not include the first radical of the root (i.e., the consonant w), but was approximately identical with the forms of the imperfect of original roots primae radic. w as they appear in the (historical) West-Semitic languages. The affixation of the feminine-ending to this nominal pattern (as Udat-, etc.), which is to be assumed as developed from the finite verb (scil., the imperfect, which did not include the first radical w of the root), may be considered as having served the purpose of strengthening the phonemic structure of the form, which otherwise would have been 'monosyllabic' ('biconsonantal'). For the sake of completeness, we note that the infinitivus constructus of Hebrew with i-vowel after the first radical of the root, a pattern appearing in forms with pronominal suffixes, as, e.g., si/sb-iih ('her lying down' [e.g., Gen. 19, 33], of *s/sab [> se/sa !!], imperfect: yiska/z), etc., should be explained as based on the phonetical transformation of the original syllable structure of the infinitivus constructus, as it appears without pronominal suffixes (see above p. 227). That means: si/sbiih (e.g.,) is to be derived from *skabiih. The vowel i in such instances is an 'interjected' (newly created) vowel. For the details of this phonetical development see above p. 11. An analogous, but somewhat modified, interpretation is valid for the (more frequent) infinitive-forms with suffixes (associated with the basic patterns of the infinitivus constructus: *q!ol [> qe!o/j, *q!al [> qetalJ) in which (in the place of the vowel i) there appears a vowel 0 after the first radical of the root; see above, p. 11. Finally, we note that the process to which we ascribe the development of the infinitivus constructus of the 'basic' stem of the verb, that is: qe!ol « yiqlol), qelal « yiqla/) (see above p. 227), must be assumed as having also produced the infinitives of the so-called 'derived' stems (or: conjugations) of Hebrew, as qa!lel (infin. of the Pi'el) < yeqa!!el; haqtil (infin. of the Hif"il) < *yehaq!il (> yaq!il); etc. The same refers, of course, to de development of the infinitives of the 'basic' stem (or: conjugation) of specific root categories in Hebrew, e.g., of the so-called 'concave' roots, as qitm < yiiqitm; sim < yiiSim; bO < yiiQo, etc.

17 THE SYNTACTICBACKGROUND OF SEMITIC NOUNS WITH PREFIX MA- AND OF PARTICIPLES WITH PREFIX MU- 1 In our Studies in A rabic and General Syntax 2 we have discussed "The development of the 'psychological' (or: 'logical') subject-predicate relation".3 This type of syntactic relation, which in its basic form appears as a complex sentence, has been derived by us from certain types of the syntactic structures characterized by the phenomenon of 'extraposition' ('casus pendens', 'nominativus absolutus'), a phenomenon which, in agreement with Reckendorfs terminology, we call "isolation of the 'natural subject' ". These structures with "isolation of the 'natural subject' ", are, in contrast to the constructions with "'psychological' subject-predicate" derived from them, not single (unitary), complex sentences (expressing a single mental operation), but the "isolated 'natural subject''' (= INS) and its "natural predicate" (= NP) constitute each a sentence by itself (i.e., INS + NP express two mental operations); we may also say, INS + NP constitute, as it were, a "twin-sentence" -structure (see Studies, chapter I, p. lff.). Studies, § 52 ff. (p. 64 ff.), we have discussed a special type of the "psychological subject-predicate" relation. This type may be exemplified by the following Arabic sentences: 4 1. wamargi'uhii iljii na~nu nqalabnii nasifu I-baqli wal-Iabanu I-~aqinu (Zuhayr, Diwiin, ed. Ahlwardt, XIX, 12). 2. at~rifu agmiila n-nawii Siiginiyatan am-i I-~afaru l-a'lii MFalgin m~iruhii (Naqii'itj Garir wal-Farazdaq, ed. Bevan, p. 516, 10). 3. itii lliihi margi'ukum (Qur'iin, 5, 104; cf. also, e.g., 39,9). 4. i/ayya I-m~iru (Qur' iin, 31, 13). 1 Originally published under the title "The syntactic background of Arabic nouns with prefix ma- and related syntactic phenomena", in Israel Oriental Studies (Tel-Aviv University), II, 1972, p. 92-116. (For "participles with prefix mu-" see below p. 250 and p. 252). 2 Cairo 1953; in the sequel referred to as Studies. 3 Studies, ch. II, §34ff. (p. 37ff.). 4 Quoted Ibid., p. 64-65; see also p. 64, n. I and p. 65, n. I.

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

233

5. mawliduhu wamansa'uhu binii!:ziyati Manbigin biqaryatin minhii (Agiini, 15, 100, 1).

And cf. also the following Hebrew instance: 6. umo~ii hassusim as?r li-Slomo mim-Mi$rayim (1 Kings, X,28). All these examples are characterized by the fact that they include a nominal formation with the prefix ma-. As we stressed already in our original discussion,5 the use of this nominal formation (maqta/il) in these sentences is connected with their specific syntactic character, which (as we have clarified in detail, see ibid.) consists in the fact that they express a "'psychological' subject-predicate" relation, which in all the sentences can be traced back to a more original "isolated 'natural subject'-'natural predicate'" relation. We now go beyond this and maintain that the nominal formations of this specific type, maqta/il, originate in the specific syntactic structures represented by the above-quoted sentences (or related instances). Accordingly, the nomen action is (or infinitive) on the pattern maqtal, maqtil-the socalled m~dar mimi of the indigenous Arab grammarians-as well as the nomen loci on the same pattern, appeared at first in the syntactic function of a so-called "psychological subject" in (complex) sentences expressing the "psychological subject-predicate" relation. The above-quoted sentences, in which the noun on the pattern maqta/il appears in the function of "psychological subject", represent two specific varieties of the construction concerned. In the one construction the "psychological predicate" is represented by a noun in the nominative (the casus rectus); in the other it is represented by a prepositional phrase (or more generally defined, by a noun in a casus obliquus). The first type, in which the ("psychological") subject is constituted by a noun in the nominative, is represented by the following sentences: 1. wamargi'uhii itjii na!:znu nqalabnii nasifu l-baqli wal-labanu l-!:zaqinu. 2. (ata!frifu agmiila n-nawii .5iiginiyatan am-i) I-!:zafaru l-a'lii MFalgin m~iruhii.

Sentence no. 1 (which is part of a description of the horses of the poet's tribe) was interpreted by us (see Studies, p. 65) as follows: "and (as for) their (i.e., the horses') return, when we return home (from our expeditions)-[they return] to green fodder and milk poured into skins". We should pay attention to the fact that in this

5

Studies, p. 65 at the bottom; end of footnote 1 from p. 64.

234

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

interpretation the sentence appears analyzed as a construction with "isolation (i.e., extraposition) of the 'natural subject' (= 'casus pendens', 'nominativus absolutus')", in English commonly characterized by 'as for' (= Arabic amma).6 As a variant interpretation of sentence 1, which seemed to us more in agreement with the linguistic expression of the sentence (characterized by the nominative of its predicate), we suggested (Studies, p. 64), the following rendering: "and that to which they (i.e., the horses) return, when we return home, is green fodder and milk poured into skins". Analogously (see Ibid., p. 65) we suggested for sentence 2 (Naqa'icj, 516, 1) the following interpretations: (a) "(do you turn the roving-camels towards Saginiyah, or) is it to the upper well at Falg that they are going?", and (more literally) (b) "( ... ) is the upper well at Falg that which they are going to?". The first, seemingly less literal, type of interpretation, in which the predicate of the nouns margi' (in sentence 1), and m~ir (in sentence 2), respectively, is, in deviation from the Arabic original texts, introduced by the preposition 'to' (= Arabic ita), corresponds to the second type of construction, as represented by no. 3-6 of the sentences quoted by us above (p. 232), in which the ("psychological") predicate is indeed, in the original Arabic (and Hebrew) texts, introduced by a preposition: ita 'to, towards', bi 'in', and min 'from', respectively. We quote sentences no. 3-6 once more: 3. itii lliihi margi'ukum: 4. ilayya l-m~iru. 5. mawliduhu wamansa'uhU binabiyati Manbigin biqaryatin minhii. 6. (Hebrew) umo$ii hassiisim a Sfr liSlomo mimM~rayim.

These latter sentences may also be rendered in two distinct types of interpretation. In the one the ("psychological") predicate appears in the nominative; in the other the ("psychological") predicate appears preceded by the preposition 'to', 'in', or 'from', respectively (in agreement with the Arabic, or Hebrew, original texts): 3. (a) "that you return (or: your return) is to God", or (b) "the one to whom you return, is God". 4. (a) "that you go (or: your going) is to God", or (b) "the one to whom you (finally) go (or: come) is God". 5. (a) "his birth and growing-up was in the district of Manbig, in

6

See Studies, § 16, p. 17 fT., also §4, p. 4-5.

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

235

a village of it", or (b) "(the place) where he was born and grew up (is) a village of the district of Manbig ... ". A certain difficulty to apply these two types of interpreation (a and b) very strictly, we encounter with respect to sentence no. 6. A very exact interpretation, which preserves the intention of the original Hebrew text more or less accurately, seems only possible if we reduce the sentence to a construction characterized by the "'isolated natural subject'-'natural predicate'" relation (a "twin-sentence" construction which we consider as the basis of the unitary (complex) sentence construction expressing the so-called "psychological subject-predicate" relation). That is: we should analyze sentence no. 6 (I Kings, X, 28) as follows: "(As for) [the country] from which Solomon's horses came (out), [they came (out)] from Egypt". Or in the (apparently) less literal interpretation in which the ("psychological") predicate appears in the nominative (an interpretation which represents a possible secondary development): "The land from which Solomon's horses came (out), (is) Egypt". As we stated already (Studies, p. 64, n. I), the construction with the ("psychological") predicate in the nominative (represented in the original passages by sentences no. I and 2; see above p. 232) has to be derived from a more original construction, corresponding (at any rate formally) to the second type (represented by instance-sentences no. 3-6), in which the predicate of the nouns margi'; m~ir; mawlid and mansa'; as well as (Hebrew) mo$ii, is preceded by a preposition: ilii, bi, or min, respectively. That is: sentence no. I wamargi'uhii ... nasifu l-baqli wal-Iabanu l-!Jaqinu should be derived from wamargi'uhii ... itii nasifi I-baqli wal-Iabani l-!Jaqini, which may be interpreted by (a) "that they return is to green fodder and milk poured into skins", or (b) (in agreement with the more primitive construction with "isolation of the 'natural subject''') "(as for) that to which they return[they return] to green fodder and milk poured into skins". And analogously, we may assign the following quasi-basic structure to sentence no. 2 (see above p. 232): (at~rifu agmiila n-nawii Siiginiyatan am) itii I-!Jafari l-a'lii MFa/gin m~iruhii, which again may be interpreted by (a) "(do you turn the roving-camels towards Saginiyah, or) is it to the upper well at Falg that they are going?", or (b) " ... (as to) that to which they are going-[do they go] to the upper well at Falg?". These inferred basic forms of sentences no. I and 2 are confirmed by sentences no. 3: ilii lliihi margi'ukum; and no. 4: i/ayya l-m~iru

236

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

(as well as by sentences no. 5 and 6), although the word order obtaining in these sentences (no. 3 and 4), with the strongly stressed "psychol. predicate" preceding the "psychol. subject", should be traced back to a more original word order, as margi'ukum itii lliihi, etc. (cf., e.g., sentence no. 6: um8~ii hassusim as?r li.s18mo mim-Mi!;rayim; cf. above p. 235, and Studies, p. 47, line 4, and p. 52, line 11). The question is only, whether the nouns serving as ("psychological") subject in these sentences: margi', m~ir (and also-in the Hebrew instance-mo~ii), can be interpreted in one and the same way in both varieties of the construction, i.e., in the construction with the ("psychological") predicate in the nominative as well as in the construction with the ("psychological") predicate preceded by a preposition, as itii (or: bi and min). In other words: the question is whether these two types of construction should be considered as being different from the outset, or whether they can be considered as having been originally identical. Their having been originally identical would mean that the construction with the "psychological" predicate in the nominative is rooted iIf a primitive form in which the (predicate) noun, which now appears in the nominative, was governed by a preposition (or-with a more general definition-in which it did not appear in the casus rectus, but in a casus obliquus). The nouns which appear in these constructions in the function of ("psychological") subject, are-at least in the sentences quoted by us -nouns with the prefix ma-. As is well-known,7 such nouns may be used in a concrete meaning-a use in which they are ordinarily defmed as nomina loci-but also in an abstract meaning, that is: as nomina action is (infinitives). The use of nouns with prefix ma- in the function of nomina actionis is widespread in Arabic-where this morphological category is called m~dar mimi by the Arab grammarians-but this use of nomina action is characterized by the prefix ma- is well-known also from other Semitic languages (cf., e.g., Hebrew mo~ii in the instance no. 6 quoted above p. 233, and cf. also the Aramaic infinitives with prefix me-). With respect to one of the abovequoted instances (no. 1; Zuhayr, XIX, 12): wamargi'uhii Ujii na~nu nqalabnii nasifu l-baqli waMabanu l-~aqinu we stressed already 8 that the noun margi' uhii (in the nominative), in accordance with its actual, literal meaning, should be defined as a "participial formation", to 7

8

Cf. also above p. 233. Studies, p. 65.

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

237

be paraphrased by a relative clause: "that to which they return". That means: the nomen concretum which margi' here represents-in distinction from margi' in the function of a nomen abstractum (i.e., nomen action is, or infinitive)-can properly not be described as a nomen loci (which, of course, represents in turn a nomen concretum), but is of a different-we may also say: of a more general-character. As already indicated, we cannot define this noun but as a "participial" formation equivalent to a relative clause (more accurately: a relative clause implying an adverbial relation). 9 The main problem with respect to these sentences concerns the question which of the two functions expressed by the nouns with prefix ma- (i.e., margi'-, masir-, etc.)-that is: the function of nomen actionis (infinitive), or of nomen loci (or, more generally conceived: of a form of "participial" function)-should be considered as original; and how it should be explained that one of the two functions developed out of the other. Jussi Aro in his study "Der mG.$dar mimi und seine Funktion im Arabischen",l° p. 18, makes the following statement with respect to the development of the infinitives with prefix ma-: " ... Bei den meisten transitiven Verben aber, und bei den Verben der Bewegung kann man vielleicht Nybergs Formel ([Le Monde oriental, vol. 15 (1920)] S. 185): ma als Konjunktion = 'dass' anwenden: dass man eintritt ... madljal 'Eintritt' (als Ereignis) dass man schHigt ... matjrab 'Schlagen' Eine andere-und m.E. die historisch glaubwiirdigere-Moglichkeit ist, dass man diese Bedeutungen aus Ortsnomina, die bei dieser Form allen sem. Sprachen gemeinsam und sicher alt sind, ableitet. Dann wiirde eine Bedeutungsverschiebung etwa im folgenden Sinne vorliegen: art des Eintretens ->- Eintritt im abstrakten Sinne, besonders an einem bestimmten Zeitpunkt. art des Schlagens ->- Der Umstand, dass geschlagen wird". In our opinion, it should be recognized that a basic condition for the development of an abstract noun, a nomen actionis, from a concrete noun, a nomen loci (by us more generally conceived as a form of a 9 It certainly makes no good sense to say: "the place to which they (i.e., the horses) return, when we return home, is green fodder and milk poured into skins". 10 Studia Orientalia, XXVIII, No. II, Helsinki 1964.

238

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

certain "participial" function, as clearly present in sentences no. 1-2; see above p.234), was the appearance of such a concrete noun (or of its genetical basis: a relative clause) in the function of a "psychological subject" in sentences representing a so-called "'psychological' subject-predicate" construction. To fully understand and explain this development, it seems necessary to take as basis the original constructions out of which the constructions expressing the "'psychological' subject-predicate" relation are genetically derived. In Studies, §2 and 13, we have stated (see also § 34 ff.) that the '''psychological' subject-predicate" relation (which constitutes a complex nominal sentence) has its basis in an "'isolated natural subject'-'natural predicate'" construction, which is characterized by the "isolation" (or: "extraposition") of a nominal term at the beginning of the construction. This primary construction constitutes, according to our conception, a "twin-sentence" construction (consisting, as it were, of two independent syntactic terms, see Studies, loco cit.), whereas the" 'psychological' subject-predicate" construction which developed from it, constitutes a unitary (though "complex") sentence. Accordingly, a sentence like rnargi'uhu nasifu l-baqli 11 should be traced back to a basic structure like the following: rna yargi'u itayhi -[yargi'u] ita nasifi l-baqli "(as to) what it (i.e., the horse) returns to[it returns] to green fodder". This "twin-sentence"-construction (representing a construction with "isolation of the natural subject", that is: with casus pendens) could develop into a "unitary", complex sentence, in which the two independent terms of the primary "'isolated natural subject' -'natural predicate' "-construction (basically expressing two independent sentences) acquired actual subject and predicate function. This development-i.e., the fusion of the two originally independent terms of the "twin-sentence"-construction into a unitary (complex) sentence-led to certain simplifications in the basic linguistic expression and, simultaneously, could also cause a change in the function and meaning of the initial term of the construction (derived from the original "isolated natural subject"). These changes-we stress once more-concern the construction which we consider as primary, that is the construction: rna yargi'u itayhi-[yargi'u] ita nasifi l-baqli "(as to) what it (i.e., the horse) returns to-[it returns] to green fodder". The relative clause rna yargi'u itayhi (the first of the two 11 A sentence formed by us on the basis of instance sentence no. 1 (Zuhair, XIX, 12), see above p. 232.

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

239

independent syntactic terms which finally were fused into a unitary sentence) included a prepositional phrase ilayhi, which referred to the introductory relative pronoun ma (with the preposition itself, ilii, recurring in the second term of the structure). The so-called resumptive pronoun (or resumptive adverb) in relative clauses, as a consequence of the fact that it resumes the antecedent (which in the case of a relative clause functioning as a substantive noun may be represented by the relative pronoun), may be eliminated. This elimination frequently occurs in the case of the resumptive pronoun being an accusative pronoun (object-suffix), indicating an ordinary object. But this elimination may also affect a resumptive expression in which the pronoun (pronominal suffix) is dependent on a preposition. We refer to the numerous instances of this type discussed by us in our study "Genetic aspects of the genitive in the Semitic languages".12 We mention here the following examples: 13 laylata ~ii/:lii. "in the night in which (or: when) they shouted" (Mufatjrjaliyiit, no. I, 5); (Hebrew) beyom dibb?r Yhwh "on the day on which (or: when) God spoke" (Ex., XI, 28); kol yeme hilhalla/snii. ittam "during all the days during which (or: when) we were conversant with them" (1 Sam., XXV, IS); qiryal /:lana [)iiwid "the city in which (or: where) David camped" (Is., XXIX, I); bimqom as?r higlii. 010 "in the place whereto they have exiled him" (fer., XXII, 12); (Ethiopic) wakona bamawa'ala yakWennanii. masafant "and it was in the days in which (or: when) the Judges judged" (Ruth, I, I); (Hebrew) wayhi bime s"jO! hassop!im "and it was in the days of the judging of the Judges", a form of expression which, from the genetical point of view, should be considered as secondarily developed from the Ethiopic type of expression. It is a specific feature of these instances that the elimination of a resumptive adverb in the relative clauses caused the status constructusform of the antecedent of the relative clauses. 14 In any case, these lAOS, LXXXI (1961), p. 386-394; see above p. 209ff. Quoted above p. 218. 14 See above p.222. Also with respect to Akkadian clauses of the type awol iqbit 'the word he has spoken', or kasap ilqii 'the money he has taken', it should be assumed that the status constructus-form of the antecedent is connected with the disappearance of a resumptive pronoun in the accusative; see above p. 223. 12

13

240

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

instances show that the verb in a relative clause may lose the linguistic expression of its adverbial complement (expressed by an adverb or a preposition with pronominal suffix), without the actual meaning of the clause becoming affected by this elimination of a basically constituent term of the clause. The same type of elimination of a resumptive adverbial expression of a relative clause should be assumed in the case of the structures here under discussion. We may therefore draw the following conclusion. In case that the original "twin-sentence" structure rnii yargi'u ilayhi-fyargi'u] ilii nasifi l-baqli developed into a unitary (complex) sentence (see for the details above), then the adverbial complement of the (substantivized) relative clause, i.e., ilayhi (referring to the relative pronoun rnii), was bound to be eliminated. However, the meaning and function of this adverbial complement were not actually lost. On the one hand, the relative pronoun, rnii, in conjunction with the verb (which required a specific preposition, in the present instance itii), indicated this adverbial complement (ilayhi), of which the speaker remained conscious without expressing it. On the other hand, a point of decisive importance, the (more essential) nominal concept which was actually governed by this preposition, was clearly indicated (apart from its formal indication by the [introductory] relative pronoun rnii) by the nominal term which ultimately-after the development of the original "twin-sentence" construction into a "unitary" (complex) sentenceconstituted the predicate of this new, "unitary" sentence. Of course, after the development of the original "twin-sentence" construction (" 'isolated natural subject' -'natural predicate''') into a unitary (complex) sentence, the preposition ilii originally contained in the "natural predicate" of the original construction ( ... -ilii nasifi l-baqli "[it (i.e., the horse) returns] to green fodder") could be eliminated, and, in this specific case, through actual loss of function. The original construction, rnii yargi'u ilayhi-fyargi'u] ilii nasifi l-baqli "what it (= the horse) returns to-fit returns] to green fodder" developed into: "what it returns to (is) green fodder".15 This is a (unitary) complex sentence consisting of "psychological" subject and "psychological" predicate. A related, but formally different form of expression, in which the same sense seems to be expressed in a non-complex (non"cleft") sentence, would be: "it (= the horse) returns to green fodder" (in Arabic: yargi'u ilii nasifi l-baqli). 15

Cf. the actual text of the line (above p. 232): wamargi"uhii nasifu I-baqli.

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

241

However, one of the main purposes of our discussion is to explain the development of an original nomen loci (or, in our definition, of a participial formation, equivalent to a specific type of relative clause) into a nomen actionis (or infinitive), equivalent to a substantive clause (a clause introduced in English by the conjunction that = Latin ut, German dass). As we have seen above (p. 240), the resumptive adverbial expression ilayhi in the original relative clause ma yargi'u ilayhi was eliminated. The (primary) "'isolated natural subject'-'natural predicate'" construction ma yargi'u ilayhi-[yargi'u] ila nasifi I-baqli thus acquired the form ma yargi'u-[yargi'u] ila nasifi I-baqli. In case that this latter, "twin-sentence" construction (" 'isolated natural subject'-'natural predicate' ") developed into a "unitary" sentence, the primary form in which this "unitary" sentence (subject-predicate) appeared, could be expected to be the following: ma yargi'u, ila nasifi I-baqli (see above). This syntactic form was apt to develop into two different structures. In the first developmental variant of this syntactic form, the original relative clause ma yargi'u « ma yargi'u ilayhi) preserved (after the "twin-sentence" structure had become a "unitary" sentence) its original meaning-that is: "that to which it (= the horse) returns" (or: "what it returns to")- but the preposition ila, originally preceding the second term (i.e., the "natural predicate") of the original construction ("what it returns to-[it returns] to green fodder"), had to be eliminated: ma yargi'u (> margi'u[hii]) nasifu I-baqli "what it (= the horse) returns to (is) green fodder" (instead of ma yargi'u, ila nasifi I-baqli). In the second developmental variant of the unitary (complex) sentence ma yargi'u, ila nasifi I-baqli (developed from the original "twin-sentence" structure: ma yargi'u (ilayhij-[yargi'u] ila nasifi I-baqli), the preposition, ila, was not eliminated, but its presence (after the "twin-sentence" structure had developed into a "unitary", complex sentence) caused a decisive change in the character and meaning of the initial term of the construction (the "psychological subject", developed from the original "isolated natural subject"). What was originally a relative clause: "what the horse returns to " (expressing a concrete semantic entity), became a substantive clause (equivalent with a nomen actionis, i.e., an abstract entity): "that it (i.e., the horse) returns (= its return) (is) to green fodder". In both types of the development of the original relative clause

242

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

ma yargi'u « ma yargi'u ilayhi)-in the first one, in which the relative clause preserved its character, as well as in the second one, in which it was transformed into a substantive clause (equivalent with a nomen actionis or infinitive)-this initial clause (introduced by ma) could undergo a transformation of a morpho-phonetic character, a transformation caused by the fact that it served in the function of a noun and that its original clause-character was essentially diminished by the elimination of its original resumptive (adverbial) expression. That is: the original clause ma yargi'u (ilayhi) , losing its formal clause-character, developed into a morpho-phonetic unit, i.e., into a single word: margi'u-, which was conceived as a noun in the actual morphological sense. Thus, the ideas "what he returns to", and "that he returns" (or: "his return") were finally expressed by the expression margi'uhu (e.g., margi'uhu nasifu l-baqli "what he returns to, is green fodder", and margi'uhU ila nasifi l-baqli "his return is to green fodder", respectively). Apart from this formal transformation of what was originally a clause (containing a finite verb) into a noun (used as a concretum as well as an abstractum), we should emphasize the more basic (and important) transformation of the original relative clause, a con cretum , into a substantive clause (noun clause), an abstractum. The most important feature of this transformation of the relative clause into a substantive clause is, however, the specific semantic (and functional) shift by which the basic formal element of the relative clause-that is: the relative pronoun ma 'what', introducing it-was affected. This semantic (and functional) shift consists in the transformation of the relative pronoun, 'what', into a conjunction of the meaning 'that' (= Latin ut, German dass). We thus assume that it is by the process (described above) of the transformation-taking place in "'psychological' subject-predicate" -constructions-of relative clauses into substantive clauses that the relative pronoun ma has acquired the function of a conjunction characterizing substantive clauses (equivalent to infinitives), that is: ma in that function in which the Arab grammarians call it: ma al-m~dariyah. A similar development we have to assume in a somewhat different type,16 which is represented by the following sentence (Agani2, I, 149,6): innama tagannabtu tannuran min-a n-nari J:zamiyan. 17 Brocke116 17

Already discussed in Studies, p. 49. Quoted by Brockelmann, Grundriss, II, p. 575.

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

243

mann 18 considers the nominative as the original case of the sentencepart represented by tannuran and postulates the following development for the quoted sentence and other cases introduced by innamii: "Nun tritt aber eine Verschiebung ein; das ehemalige Subjekt des Nachsatzes wird zum Pdidikat eines auf ma folgenden Nomens, oder zum Subjekt oder Objekt eines ihm folgenden Verbums". As already stated by us in Studies, p. 49, we should (in contrast to Brockelmann's opinion) consider the accusative tannuran as original and reduce the sentence to the following primary construction: inna ma tagannabtu-tannuran min-a n-nari bamiyan, with ma tagannabtu in the function of "isol. nat. subj." (= protasis) and tannuran ... bamiyan in the function of "nat. pred." (= apodosis). This original construction should be paraphrased as follows: "(as regards) what (= ma) I avoid-[I avoid] a hot furnace". This structure (which constitutes a "twin-sentence"-construction) developed 19 into a unitary (complex nominal) sentence ("cleft" sentence, "psychol. subj.pred. "-construction). This unitary sentence is to be analyzed as follows: "that I avoid (or: my avoiding) is a hot furnace" (with "a hot furnace" representing an accusative). That means: what originally (in the primary "twin-sentence" -construction) was a relative clause ("what I avoid") developed into a substantive clause, equivalent with a nomen actionis (or infinitive). This development of a substantive clause from a relative clause, caused by the transformation of the original "twin-sentence"-construction into a unitary (complex nominal) sentence, entailed the transition of the relative pronoun ma 'what' into a particle of the meaning 'that' (= ut, German dass). This shift of an original relative clause into a substantive clause (accompanied by the simultaneous shift of the relative pronoun 'what' into a conjunction: 'that') agrees with the development observed by us in connection with the cases discussed above p. 241 ff. Whereas in these latter instances the primary rCIative clause contained an adverbial relation - "what he is returning to", "what he is arriving at (coming to)", or: "what he is coming (going out) from" -in the present instance it contains an ordinary object relation (expressed by the accusative): inna ma tagannabtu- tannuran min-a

18 19

Loc. cit. Cf. Studies, § 13.

244

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

n-nari I;zarniyan "(As to) what I avoid-[I avoid] a hot furnace". The

present instance, no less than the former instances (which comprise a relative clause containing an adverbial-local-relation), exemplifies in a clear and distinct manner the development of the relative pronoun (derived from an original interrogative pronoun) into a subordinating conjunction ('that' = German dass). But this sentence type, with the original relative clause containing a complement in the accusative, exactly as the instances containing an adverbial (local) relation, exemplifies also the development of nomina concreta as well as nomina abstracta (nomina actionis, infinitives) with prefix rna-. In the basic "twin-sentence"-construction (" 'isol. nat. subj.'-'nat. pred.''') of the sentence with innama discussed above, the (original) pronoun rna was independent, that is: it had not been fused with the particle inna preceding it. And this (independent) character of rna, as a word expressing a meaning by itself, remained also preserved in the "unitary" (complex nominal) sentence ("cleft" sentence, expressing a "psychol. subj.-pred." relation), which developed from the basic construction. But finally the sequence inna rna (two independent words) in sentences of this type developed into a single word: innama. 20

In connection with our discussion of the shift of the relative pronoun

rna 'what' into a subordinating conjunction of the meaning 'that' = German dass (see above p. 242), we should also mention Reckendorfs

interpretation of this shift: 21 "Es [i.e., the clause with rna 'that' from 'what'] ist eigentlich ein vom Verbum abhangiger innerer Objektsakkusativ (§ 48, 2. 3 h). Zur Veranschaulichung des Bedeutungsiibergangs: nasu rna kana !lahu $ana' a bihirn warna naggahurn (= wan-naga' a !lag! naggahurn) 'sie vergassen, was Gott mit ihnen getan hatte und dass er sie gerettet hatte' (= und die Rettung, die er sie gerettet hatte [= and they forgot the saving which he had saved them]), Tab. I, 2; 658, 16 ... ". This interpretation of the use of rna in the meaning 'that' (= dass) indeed agrees with the explanation suggested (among other explanations) for the use of Latin quod in the function of a subordinating 20 For the syntactic shift underlying this development of inna rna into innarna we refer to our discussion in Studies, p. 49,1. 14-p. 50, 1. 7. 21 Arabische Syntax, § 192,2, p.394. Cf. also Syntaktische Verhiiltnisse des Arahischen, p. 586 ff.

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

245

conjunction. 22 We doubt that a pronoun (and more specifically a relative pronoun) can be substituted for a so-called "inneres Objekt" (= maful mu!laq, "cognate object"), especially of the basic, abstract type, as represented by Reckendorfs example (of course, the "innere Objekt", "cognate object", represented by the noun t;larbat- in a sentence like t;larabahu t;larbatan 'he struck him [with] a stroke', may be referred to by a pronoun). For the sake of completeness we mention that the opinion that a relative pronoun can refer to an (unexpressed) "cognate object" is also held by Brockelmann. 2 3 We should, however, also quote Noldeke's reference to the relationship between the pronoun ma and the conjunction ma : 24 "Der Unterschied des pronominalen ma vom konjunktionalen (m~dariyah) ist nieht scharf. Ma mit seinem Satz kann ja den Inf. absol. (m~dar mu!laq) vertreten. In fani'ma ma anta 'inda I-hayri tus'aluhu wani'ma ma anta 'inda I-ba'si ta/:ltat;liru 'wie schon ist das, worum du in guter Zeit gebeten wirst, und wie schon, wie du beim Ernst zugegen bist' ['How beautiful is that which one asks from you in the time of peace, and how beautiful (is it) how you are present when the situation is getting serious'] Gamhara 137, 12 (wo ma ta/:ltat;liru = i/:ltit;laruka ist) wird der Araber kaum den Unterschied zweier Konstruktionen empfunden haben ... ". We consider the difference between the pronominal ma and the conjunctional ma as absolute and very distinct. Moreover, there can be no doubt that the passage quoted by Noldeke does not permit to draw any conclusion concerning the relationship between the pronominal and the conjunctional ma. What is more, Noldeke's interpretation of this passage cannot be accepted. As to the form of the line, as it is transmitted by the Arab tradition and accepted by Noldeke (as well as by R. Geyer in his edition of the poems of A'sa Bahilah, no. 4, 38; appendix to his edition of Diwan al-A'sa Maymun, p.268), it should be emendated. That is: instead of 'inda I-ba'si ta/:ltat;liru, we should read 'inda I-ba'si ya/:ltat;liru. In contradistinction to Noldeke's conception of the syntactic structure of the line, ma in the twice appearing phrase ni'ma ma anta does not relate to the words See 1.-8. Hofmann, Lateinische Syntax und Stylistik (1965), p.572 and p.574. Grundriss, II, §204 g, lines 2 and 5 (with reference to lasadda ma a~babta ahlaka 'du liebst deine Frau ja sehr'). 24 Neue Beitriige zur semit. Sprachwiss., p. 225-226. 22 23

246

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

following this phrase in the two coordinated sentences, that is: to the words 'inda I-bayri ... and 'inda I-ba'si ... , respectively, but rna in the phrase ni'ma rna (as in numerous instances of this expression) refers to the word directly following it, that is: to the pronoun anta. The line should be interpreted as follows: "How goodly you are when you are asked for the 'good' (i.e., for a benefaction), and how goodly you are when an inimical encounter arises !". We state that the preposition 'inda in 'inda I-bayri tus' aluM and in 'inda l-ba'si yal)tatJiru (as we must read) is used in the function of a conjunction. Although the predicate of the two conjunctional clauses introduced by the preposition 'inda is represented by a finite verb, the construction is related to the Indo-European construction of the type ab urbe condita (which itself has exact parallels in Arabic). This type of construction was at first described by us, on the basis of a number of examples, in the Hebrew periodical Tarbiz, vol. 15 (Jerusalem 1944), p.191-192, and vol. 17 (1945/46), p.49ff. (where also the passage quoted and translated above was for the first time interpreted; see Ibid., p. 51). We furthermore refer to our discussion of this specific construction, as well as of all related constructions, in our Studies in A rabic and General Syntax (Cairo 1953), § 29 (p. 30 ff., see especially p. 31, n. 3). Moreover, see below p. 295 ff. The instances of the" 'psychological' subject-predicate" -construction in which the "'psychological' subject" is represented by a noun with

prefix ma- (see sentences no. 1-6 quoted above p.232), have been traced back by us to "'isolated natural subject'-'natural predicate'" constructions in which the "isolated natural subject" is represented by a substantivized relative clause introduced by the relative pronoun rna. We have furthermore shown that these relative clauses with rna (in the function of an "isolated natural subject") could secondarily develop into substantive clauses (with rna in the meaning of 'that' = German dass) and into nomina actionis (corresponding to such substantive clauses). However; we might perhaps expect that even in the primary" 'isolated natural subject'-'natural predicate' "-construction ("twin-sentence" -construction) the function of "isolated natural subject" could also be represented by a (primary) substantive clause or a nomen action is corresponding to it. We might therefore presume the following structure having existed as a primary construction: "(as to the fact) that it (= the horse) returns-[it returns] to green fodder" (see instance-sentence no. 1, quoted above p.232), etc. We refer in

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

247

this connection to the following type of sentences: 25 (Tabari, Annales, I, 4; 1874, 15) nifiiruhii min-a l-an/:zii'i, literally: 'their fleeing was from the skins'; (Agiini, VI, 132, 7) inna masya ba'{likum fi iiliiri ba '{lin , literally: 'the going of some of you is in the tracks of the others'. As we have shown,26 sentences of this type imply "psychological subjectpredicate" constructions, and may therefore be interpreted as follows: 'it is from the skins that they flee', and 'it is in the tracks of a part of you that the rest go', respectively. We have furthermore shown that these sentences are most easily conceived in their character of "psychol. subj.-pred." constructions, if we reduce them to (primary) "'isol. nat. subj.'-'nat. pred.'" constructions ("twin-sentence"-constructions): '(as to) their fleeing-[they flee] from the skins', and '(as to) the going of some of you-[they go] in the tracks of the others', respectively. Nevertheless, we do not assume with respect to instances of this type that the (actually existing) "psych. subj.-pred." constructions are directly developed from such (primary) "'isol. nat. subj.''nat. pred.''' constructions in which the "isol. nat. subj." was represented by a nomen action is, as '(as to) their fleeing', or '(as to) their going', respectively (or by substantive clauses corresponding to such nomina actionis). We assume that the speaker is generally not just interested to know what refers to-or can be predicated of-the abstract concept 'their fleeing' ('that they flee'), or 'their going' ('that they go'), respectively, but that he is from the outset interested to know-and asks himself about-the concrete entities, expressed by relative clauses: 'that which they flee from', and 'that which they go in', respectively. And the same principle can be demonstrated on the basis of the sentence (discussed above p. 242) innamii tagannabtu tannuran min-a n-niiri lJiimiyan. Although an "'isol. nat. subj.' -'nat. pred.' "-construction of the type '(as to the fact) that I avoid-[I avoid] a hot furnace' can be formed and makes sense, nevertheless, it is far more natural for the speaker to search for what refers to-or can be predicated of-the concrete entity: 'that which I avoid', that is, to express himself as follows: '(as to) what I avoid-[I avoid] a hot furnace'. Of course, as we have already seen above p.246, through the development of the primary '''isol. nat. subj.'-'nat. pred."'-construction into a "psychol. subj.-pred."-construction, such concrete concepts as '(that) which one returns to', or '(that) which I am avoid25 26

cr. Studies, p. 50 fT. Loc. cit., p. 51.

248

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

ing', which represent relative clauses (used in substantival function), can develop into substantive clauses (noun clauses, clauses introduced in English by 'that' = German dass). Accordingly, we may assume as the primary construction of sentences of the type nifaruha min-a l-an~a' i (literally: 'their fleeing is from the skins', see above p. 247) the following construction: ma tanfiru minhu-[tanfiru] min-a l-an~a'i '(as to) that which they flee from-[they flee] from the skins'. The development of this (primary) "'isol. nat. subj.' -'nat. pred.'" -construction ("twin-sentence" -construction) into a "psychol, subj.-pred." -construction (unitary [complex nominal] sentence construction), could result in the following structure: 'that they flee (= their fleeing) (is) from the skins', a development which implies the transformation of the relative pronoun ma 'what (= that which), into a conjunction of the meaning 'that' (= German dass). Once this shift of the original relative clause into a substantive clause (equivalent with a nomen action is) had been achieved, this substantive clause, introduced by (the original relative pronoun) ma, could be replaced by any other equivalent expression, e.g., by the regular nomen action is (infinitive) of the verb concerned, as, e.g., in the instance on which our present discussion is based (that is: nifaruha min-a l-an~a' i), by the regular infinitive of nafara, that is: nifar. It should indeed be emphasized that such an infinitive like nifar- (of nafara 'to flee') can, of course, not be considered as developed in the specific syntactic construction here discussed, in difference from synonymous forms with prefix ma- « the pronoun ma), like, e.g., mafarr-, mafirr- (infinitive of farra, yafirru 'to flee'), with respect to which we may assume that they developed precisely in the sentence-type here under discussion. The difference between sentences like ma yargi'u (or: margi'uhu), ila nasifi l-baq/i 'that it (i.e., the horse) returns (or: its return) (is) to green fodder', or (inna) ma tagannabtu, tannuran ... ~amiyan 'that I am avoiding (or: my avoiding) (is) a hot furnace' (with 'that' in the function of conjunction), and sentences like nifaruha min-a l-an~a'i consists just in the feature that in the first type the substantive clause (developed from a relative clause)-or the nomen action is with prefix ma- based on such a substantive clause-is reflected in the morphological structure of the term which serves in the function of ("psychological") subject in the sentences concerned, while in the second type (as nifaruha min-a l-an~'i) such a direct morphological derivation of the actually used nomen actionis (equivalent with a substantive clause) from an original substantive clause, or relative clause, with rna

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

249

is not possible, since the original term has been replaced by a synonymous term of different origin. We may also mention that the idea expressed in the above-quoted (primary) structure (inna) mii tagannabtu tannuran min-a n-nari f:ziimiyan appears in its turn in such a form in which (in our opinion) the "psychological" subject is expressed by an infinitive which cannot be directly traced back to the primary structure with a substantivized relative clause introduced by the pronoun mii. We have in mind the following sentence: 27 ... walakin biIJari n-nara data t-talaffu'i ' ... but my avoiding (or: bewaring of) is a blazing fire' (with 'fire' in the accusative). The noun biIJari ('my avoiding') relates to ma tagannabtu[hu] (= ma biidartu[hU]) 'what I am avoiding') in the same way as nifaruhii, 'their fleeing' (in nifaruha min-a l-anba'i, see above) to ma tanfiru minha 'what they are fleeing from'. That is: here also the primary nomen concretum (expressed by a relative clause) has been transformed into a nomen abstractum (equivalent to a substantive clause [= noun clause]). But no doubt, this nomen abstractum in turn could secondarily develop the meaning of a nomen con cre tum , whereby the abstract concept of 'taking heed', 'avoiding', or 'fearing', was converted into a concrete concept: 'object of fear'. In case that this development of the abstractum into a concretum occurred, then, of course, the accusative of the ("psychological") predicate was transformed into the nominative, and the sentence should be paraphrased by 'the object of my dread is a blazing fire'. 28 A case of this type, that is: the concept 'object of dread' developed from the abstract concept 'fear, dread', seems to be represented by the use of Hebrew paba4 (properly 'fear, dread') in the sense of 'object of dread' (Ps., XXXI, 12; XXXVI, 2; etc.). Cf. also, e.g., German Sorge = 'solicitude', but also = 'object of one's solicitude'.29 A further interesting instance of this nature is represented by the frequently observed use of nouns basically expressing the abstract concept of 'seeking refuge' in the concrete sense of 'place of refuge'. 30 The nouns with prefix ma-, frequently appearing in the function of ("psychol.") subject in "psychol. subj.-pred."-constructions (see the Cf. our Studies, p. 47 n. 1. Cf. the variant tradition of the sentence, quoted by us Studies, p. 48, note: waliikin i)igiiri gai)mu niirin mulaffa'i, with the predicate noun, gai)mu, in the nominative, corresponding to the accusative an-niira in the other tradition (see above). 29 Cf. Studies, p. 66. 30 See for Arabic, as well as other, instances of this specific type, Studies, p. 66, n. 1. 27

28

250

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

instance sentences quoted above p. 232) should be derived from substantivized relative clauses introduced by the relative pronoun rna and (originally) comprising an adverbial complement with resumptive pronoun (referring to the initial pronoun ma). The specific verb form in which the verb of these (primary) relative clauses appeared, may have frequently been the imperfect (i.e., the form yaqtulu). We see the imperfect reflected in the various nouns appearing in our instance sentences (see above p. 232). An imperfect form (preceded by the pronoun ma) may especially be assumed as basis of nouns with prefix ma- ofroots med. y, as ma.yir- (of root $yr 'to go (or: come) [to]') (cf. above p. 232, instance sentences no. 2 and no. 4), or masir (of root syr 'to travel'). Also the stem-vowel i in margi'- (cf. Ibid., instance sentences no. 1 and no. 3) obviously reflects the vowel i of the imperfect yargi' u. That is: margi' - should directly be derived from ma yargi' u (ilayhi), in the same way as mayir and masir clearly reflect ma ya.yiru (ilayhi), and ma yasiru ( ... ). We defined the nouns of this and related patterns, as "participial formations" derived from relative clauses originally comprising an adverbial complement (which was subsequently eliminated; see for details above p. 236 ft). As to the morpho-phonetic development of this type of "participial formation", we identify it with the development on which in our theory 31 the ordinary participles of the "derived" verbal themes are based. The participles of the derived verbal themes are characterized by the prefix mu-, as Arabic/Akkadian muqattil ("intensive" conjugation); Arabic muqtil, Akkadian musaksid (causative conjugation). The vowel u of the prefix mu- in these forms indeed corresponds to the vowel of the personal prefixes of the imperfect forms with which these participles are associated: yuqattil (ukassid); yuqtil, usakSid. Nyberg,32 assumes that the prefix mu- of the participle forms represents an "Ablaut" variation of the nominal prefix ma-, in the same way as the vowel u of the personal prefixes of the imperfect of certain "derived" conjugations represents an "Ablaut" variation of the prefixvowel a (by which the imperfect of the "basic" conjugation, Arabic yaqtulu, and of certain other conjugations, is characterized). In our opinion 33 the participial forms muqattil, muqtil (musakSid), etc., are 31 This has been outlined by us in Compte-rendus du "Groupe linguistique d'etudes chamito-semitiques" (GLEeS, Paris, a la Sorbonne), mai 1940. 32 Le Monde Oriental, XIV (1920), p. 179. 33 See loc. cit. (above n. 31).

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

251

not derived from basic forms to which an element mu- ("Ablaut"variation of ma- < rna) was prefixed; but the basic forms from which the participles of the "derived" conjugations have developed, included the element rna (in the function of relative pronoun) itself. The participles of the "derived" conjugations have in our opinion developed from combinations of the particle rna (relative pronoun) with the respective forms of the 3rd pers. sing. masc. of the imperfect of these conjugations (see below p. 259). The development was as follows: rna yuqattil> muqattil; rna yuqtil « yu'aqtil) > muqtil (or: rna (y)usakSid > musaksid). That is: as a consequence of the fusion of the relative pronoun rna with the 3rd pers. sing. masc. of the imperfect, the diphthongic combination -ayu- (-aiu-) became contracted to -u-. More specifically: of the two sonantic elements of the sequence -ayu-, it was the second sonant, i.e., u, which maintained itself, while the first sonant, i.e., a, disappeared. There exists indeed a clearly attested case of the development of the combination ma-yu- to mu-. It is present in the medieval vulgar pronounciation 34 of rna yudrika (,what makes you know?') as mudrik (a), a form which survives in various modern dialects as madrik (etc.). Since some of the nouns with prefix ma- which appear in the instances discussed by us specifically, clearly reveal their derivation from the imperfect of the roots involved~as, e.g., m~ir- and margi'(see above p. 242)~ they, interestingly enough, suit the theory of the Arab grammarians, according to which the nouns with prefix main general are to be derived from the imperfect. 3 5 It is, however, a fact that many other nominal forms with the prefix ma- do not suit equally well such a derivation from the imperfect. From the outset, one should, of course, not disregard the possibility that also the perfect could have been used in the original relative clauses from which the nouns with the prefix ma- arose. But at the same time we must keep in mind that the larger part of the actually existing nouns with prefix ma- were certainly not directly developed from such relative clauses, but came into being through the analogy of a restricted number of 'nouns which are directly derived from relative clauses introduced by rna. Moreover, certain phonetical factors associated with roots of specific phonetic composition may have changed the morpho34 According to Gawaliqi (Morgenliindische Forschungen in honor of H. L. Fleischer, Leipzig 1875, p. 145). 35 Cf. Nyberg, Le Monde Oriental, XIV (1920), p. 178.

252

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

phonetic structure of certain nouns with prefix ma-, so that the original character of the specific verb form involved in a specific instance is no longer recognisable. Btlt in general, the imperfect should be considered the basis of the nominal patterns with prefix ma-, which in our opinion should (with respect to their original function) be characterized as "participial formations". The Semitic imperfect in general expresses the present tense, and is also frequently used in a quasi non-temporal function. And indeed a verb of such a character seems to be most suitable to serve as the basis of a "participial formation". We have one other specific group of participial forms clearly derived from relative clauses introduced by mii. This group is constituted by the regular participles of the so-called "derived" verbal themes, that is: the forms muqattil, muqtil, (Akkadian) musaksid, etc., which (see above p. 251) are to be derived from the original relative clauses ma yuqattil, ma yuqtil, ma (y )usakSid, etc. That is: these regular participles are all derived from (Proto-Semitic) relative clauses in which the verb was represented by the imperfect. From the outset, one could perhaps, also with respect to these regular participial formations, expect that the perfect too was used in the primary relative clauses (introduced by ma) from which they are doubtless derived. This is clearly not the case. In analogy hereto, it seems very likely that also this other type of "participial formations" (in our definition), represented by the nouns with prefix ma-, is exclusively based on original relative clauses with the verb in the imperfect. The main difference between the latter type (that is: the nominal formations with prefix ma-) and the regular participles of the so-called "derived" conjugations consists in the fact that the nouns with prefix ma- (by us defined as "participial formations") are in general associated with the "basic" conjugation (the "basic" stem of the verb). If it thus seems likely that the imperfect was the mainly (or exclusively) used verb form in the primary relative clauses from which the nouns with prefix ma- are derived, it seems also certain that the specific form of the verb used in these clauses was always the 3rd pers. masc. sing. H. Bauer 36 assumed that the verb form in the primary relative clauses which formed the basis of the nouns with prefix ma-, could be represented by the forms of various grammatical persons of the verb (e.g., by the 1st pers. sing. of the imperfect, by the 2nd pers. masc. sing. of the perfect, etc.). This seems to us an extremely unlikely 36

ZDMG, LXXI (1917), 407ff.; cf. Nyberg, lac. cit.

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

253

possibility. We may again refer to the regular participles of the "derived" conjugations, rnuqattil, rnuqtil, etc., which we consider based on the 3rd pers. masc. sing. (of the imperfect of the conjugations concerned). Here one could also perhaps from the outset expect that the primary relative clauses introduced by rna could contain the verb forms of the various grammatical persons. The use of the verb forms of the various grammatical persons would then, of course, be reflected in the actually existing participles, which developed by the fusion of the original relative pronoun rna and the finite verb following it in the primary relative clauses. But the actually existing participles of the derived conjugations can only be traced back to relative clauses containing a verb (more specifically: an imperfect form) of the 3rd pers. masc. sing. In analogy hereto, one should, also with respect to the nouns with prefix rna-, assume that they are based on original relative clauses containing a verb in the 3rd pers. sing. masc. Obviously, only such a verb form (that is: a 3rd pers. masc. sing., and most likely of the imperfect) was basically apt to develop (by its fusion with the preceding pronoun rna) into a compound of nominal character, that is: into a form of participial function. We mentioned above that H. Bauer assumed that the nouns with prefix rna- are based on the forms of the various grammatical persons of the verb (imperfect as well as perfect). Bauer restricted the form of the 3rd pers. masc. sing. of the imperfect as basis of nouns with prefix rna- to such instances in which the prefix appeared with an i-vowel (in place of the a-vowel), as in Arabic rnifta~- 'key' which he derived from rna yafta~u 'that which opens'. He obviously saw the consonantic vowel y of the imperfect-prefix ya- reflected in the vowel i of the prefix rni-. In our view, the fusion of the pronoun rna with the prefix of the 3rd pers. masc. sing. ya- resulted generally in rna-, with the consonantic vowel y of the original imperfect-prefix completely disappearing (in analogy to the fusion of rna yu- > rnu- (prefix of the participles of the "derived" conjugations; see above p. 251). Consequently, rna$iru- derives from rna y~iru; rnargi'u- from rna yargi'u; etc. As to the i-vowel in instances like rnifta~ (from rna yafta~u), it should be explained as developed from an a-vowel in a more original form *rnafta~ (cf., e.g., Hebrew/Aramaic yiq!ol from (Arabic) yaqtul). In connection with the development of nouns with prefix rna- we dealt in the present discussion with "psycho!. subject-predicate" constructions in which the (initially placed) "psycho!. subject" is repre-

254

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

sen ted by a relative clause introduced by the pronoun rna. The use in this construction of relative clauses of this specific type leads us to the problem of the character of the primary construction in which original interrogative pronouns, as Arabic (and Semitic) rna 'whatT, rnan 'whoT, acquired the function of relative pronouns. We quote Brockelmann's statement concerning this problem (Grundriss, II, § 370 a; p. 570): "Wie in den idg. Sprachen, so werden auch in den semit., namentlich im Arab., Fragesatze oft zu Relativsatzen verschoben. Ein Satz wie rnan yunaggikumrnin ?ulurnati l-barri wal-bai)ri tad'unahu uujarru'an (Sura 6, 63) ist urspriinglich 'wer wird euch retten aus den Finsternissen des Landes und des Wassers? Den werdet ihr anrufen in Demut'. Indem Frage und Folge zu einem Ganzen verschmelzen, wird die Frage zu einem Glied des Hauptsatzes. Daher wird nun das Fragepronomen ebenso behandelt wie die ehemaligen Demonstrativa ... ". We doubt that the two sentences following each other in this Qur'anic verse could ever merge so that a relative clause introduced by an original interrogative pronoun could develop. Another question is whether Brockelmann's interpretation of the Qur'anic passage is altogether acceptable. We refer to the Qur'an commentaries, as well as to the various translations. 3 7 In our opinion, the basic forms of relative clauses introduced by original interrogative pronouns are present in the type of construction dealt with in the present context. As basis of our derivation of the relative pronoun rna from the interrogative pronoun rna we may take the sentence discussed above (see p. 242 and p. 248): (inna) rna tagannabtu tannuran rnin-a n-nari /:zarniyan. We recognized this sentence as developed from a "psychol. subject-predicate" construction to be paraphrased by (a) 'That I avoid (or: my avoiding) (is) a hot furnace', or by (b) 'What I avoid is a hot furnace'. This structure was traced back by us to an original "isolated natural subject-natural predicate" construction rna tagannabtu-tannuran rnin-a n-nari /:zarniyan, to be interpreted as follows: '(as to) what I avoid-[I avoid] a hot furnace'. Now we claim that the (initially placed) relative clause in a "psychol. subject-predicate" construction of the type rna tagannabtu tannuran 37 See, e.g., Blachere's translation, Le Coran (Paris 1950), p. 677; Montet's translation, Le Coran (Paris 1949), p. 216.

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

255

barniyan (or: tannurun barnin) 'What I avoid is a hot furnace' may be

considered as directly derived from an original interrogative sentence, in which the pronoun rna served as an interrogative pronoun, rna 'what?'. The speaker, dealing with the problem of 'something to be avoided by him', is sometimes not aware (or conscious) of the concrete nature of this 'something': the 'something' does not arise in his consciousness immediately with the idea of 'avoiding'. He thus is seeking for this 'concrete object to be avoided', asking about it in an interrogation addressed to himself. This self-interrogation helps him to become aware of 'the concrete entity to be avoided', and he adds it in an answer sentence. 38 Interrogation and answer are thus constituted by the following linguistic structure: 'What do I avoid ?-[I avoid] a hot furnace', in Arabic: rna tagannabtu?-tannuran barniyan. That is: the linguistic expression of the answer is restricted to the concrete concept which did not immediately appear in the consciousness of the speaker and for which he was asking; the verb governing this concrete object did not find a linguistic expression in the answer. Subsequently, by the force of habit, this form of expression was frequently used by the speaker also when the object which 'he avoids' (i.e., the concept 'a hot furnace'), appeared in his consciousness immediately (i.e., together with the verb governing it: 'I avoid'). That means: this type of construction became mechanized. It is through this mechanization of the construction that the original interrogative sentence lost its interrogative character and joined the original answer sentence so that the two originally independent sentences came to form a single (complex) sentence, composed of subject and predicate. 39 The concrete result of this mechanization of the original interrogation-answer construction and of the fusion of the two sentences into a unitary sentence was the development of the original interrogative sentence into a relative clause, a process by which the original interrogative pronoun rna 'what?' acquired the character of a relative pronoun. That is: the original interrogationanswer 'What do I avoid?-[I avoid] a hot furnace' developed into 'What I avoid (is) a hot furnace'. We may exemplify this type of development by various other instances, representing a "psychol. subject-predicate" construction Cf. Studies, p. 4, I. 28-p. 5, I. 4. For further details concerning this fusion of this type of independent sentences, see Studies, § 13, p. 14. 38

39

256

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

of the specific type discussed in the present study. We mention the following sentence (already quoted in Studies, p. 47): fala yub'idanka llahu ya Tawba innamii liqii'u l-maniiyii diiri'an milia biisiri (Mubarrad, Kiimil, p. 105,4; Bul}turi, Ifamasah, p. 269,6). On the basis of its literal conception as a "psycho!. subject-predicate" construction, we may interpret this sentence as follows: " ... The hitting of Fate (is) an armoured one (armatum) and an armourless one (inermem)", which is in the more normal equivalent form of expression in English: "It is the armoured one as well as the armourless one (accus., pronominally expressed: 'him') that Fate hits". This "psycho!. subjectpredicate" construction may be traced back to the following form of a (primary) "'isolated natural subject'-'natural predicate'" construction: mii talqii l-maniiyii-diiri 'an milia biisiri '(as for) what Fate hits (or: [as to] that which Fate hits)-[it hits] the one armoured as well as the one armourless'. But the pronoun mii in this linguistic form can also be conceived as an interrogative pronoun, so that we deal again with an interrogative sentence (addressed by the speaker to himself) followed by its answer: 'What does Fate hit?-[it hits] the one armoured as well as the one armourless'. This, by the process described above, may have developed into a unitary assertive sentence (constituting a "psycho!. subject-predicate" construction) with ma talqii ... representing a relative clause: 'What Fate hits (is) the armoured one as well as the armourless one'. We may, of course, also construct an instance in which the initial clause, instead of being introduced by the pronoun ma 'what', is introduced by the ("personal") pronoun man 'who': man talqa I-maniiyii-diiri'an miJ/a biisiri, which may be interpreted as follows: 'Whom does Fate hit?-[it hits] the armoured one as well as the armourless one', a structure which, by the process described above, could develop into the following structure: '(those) whom Fate hits (or: those hit by Fate) are the armoured ones as well as the armourless ones' (or in a non-complex sentence, expressing a "psycho!. subject-predicate" relation: 'Fate hits the armoured ones as well as the armourless ones'). That means: the word man, which in the basic construction served in the function of an interrogative pronoun, has developed into a relative pronoun. To exemplify the development of the interrogative pronoun man 'whoT into a relative pronoun, we may also quote the following passage: 40 (1) wagadata nabnu ma' a n-nabiyi ganiibuhu bibi!abi M ak40

Ibn Hisam, Sirat Rasuli-lliih, ed. Wiistenfeld, p. 860, 2-3; already quoted in

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

257

kata wal-qana yatahazza'u (2) kanat iifabatuna !ida'i rabbina bil-I}aqqi minna I}asirun wamuqanna'u 'and on the morning (or: day) when we formed the wing of the Prophet in the plains of Mecca, while the lances were being swung - there responded to the herald of our Lord alike our armourless ones and our armoured ones', or more literally: 'there was our responding (nomen action is) to the herald of our Lordboth our armourless ones and our armoured ones' (that is: 'all of us, without exception, responded'). We may construct (on the basis of this "psychol. subject-predicate" construction) the following (primary) interrogation-answer structure, in which the interrogative pronoun man does not represent the syntactic function of an object (as in the earlier passage), but of a subject: man (or possibly ma) agaba !ida'i rabbina-[agaba] minna I}asirun wamuqanna'un 'Who responded to the herald of our Lord?-[There responded] both our armourless ones and our armoured ones'. This again could develop into '(Those) who responded to the herald of our Lord (were) both our armourless ones and our armoured ones (i.e., all of us responded)', a development by which the original interrogative pronoun man acquired the character of a relative pronoun. Our explanation of the development of interrogative pronouns, as ma 'what?', man 'who?', into relative pronouns has a parallel in the development of interrogative particles of the meaning 'why?' into particles conveying the meaning of English 'for' = German denn, as, e.g., French car from Late Latin quare. 41 In Studies, § 12, we mentioned this use of the interrogative adverb quare (> car) in place of a causal particle of the meaning 'for' = 'denn', and we interpreted this use as caused by a hitch in the framing of the intended statement (cf. Studies, §4). We characterized (see Ibid., § 12) this use of quare 'why?' in the meaning of 'for' as a parallel to the syntactic pattern which, in our view, is constituted by the construction with "isolation of the natural subject" (i.e., "extraposition", the use of a noun as casus pendens), dealt with Ibid., §.l ff. The construction involving "isolation of the natural subject" may be exemplified by the following clear instances: Zaydun ga' a abfthu and Zaydun ra' aytu abahu. These instances may be rendered by '(as to) Zayd, his father came' (= ammii Studies, p. 48. 41 Cf. Havers, Handbuch der erkliirenden Syntax (1931), p. 24f.; see also M. Leuman, Kleine Schriften (Zurich 1959), p. 289, whose view concerning this semantic shift can, however, not be accepted.

258

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

Zaydun, fagila abUhu), and by '(as to) Zayd, I saw his father' (= ammii Zaydun, fara' aytu abiihu), respectively. In Studies, § 4, we interpreted these syntactic patterns as basically constituting "twin-sentence" constructions consisting of (self-)interrogation and answer. That is: the noun "extraposed" at the beginning of the construction (i.e., the casus pendens, or "isolated natural subject")-in our instances: Zaydun-constitutes an interrogation, while the sentence following itin our instances gii' a abuhu and ra' aytu abiihu, respectively-constitutes the answer to this interrogation. The "extraposed" (or: "isolated") noun, i.e., Zaydun, constitutes an interrogation not characterized as such by an interrogative pronoun, but just by the intonation. Our definition of the "isolated natural subject" (the casus pendens) as an interrogation implies, of course, our identification of its intonation with the intonation characteristic of regular interrogative sentences. (The interrogative character of the "isolated natural subject", Zaydun, in the above-quoted sentences may be made explicit by the following interpretations of these sentences: '[What about] Zayd?His father came', and '[What about] Zayd?-I saw his father', respectively). We assumed as cause of the emergence of this type of syntactic patterns the same psychological condition as assumed by us for various other constructions, characterized by interrogative particles which subsequently lost their interrogative function (see above p. 257), that is: a hitch in the framing of the intended statement (cf. Studies, §4). It is obvious that in the function of an "isolated natural subject" (casus pendens) there may appear not only an ordinary (primary) noun (as, e.g., in the above-quoted instances, the noun Zaydun, referring to a person), but also a clause, as, e.g., a relative clause (used as a noun). We assumed the use of a substantivized relative clause in the function of an "isolated natural subject" for numerous instances quoted in the present context, as, e.g., the sentence innamii tagannabtu tannuran min-a n-niiri ~iimiyan (see above p.254), which we reduced to a (primary) form rna tagannabtu-tannuran min-a n-niiri biimiyan 'What I avoid (or: as to what I avoid)-[I avoid] a hot furnace'. In a somewhat more elaborate form (presenting our conception of this construction more clearly), this linguistic form could also be paraphrased as follows: '[What about] what I avoid?[I avoid] a hot furnace'. That is: the initial clause rna tagannabtu (by us conceived as a relative clause), in its being used in the function of an "isolated natural subject" (casus pendens), constitutes an inter-

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

259

rogative sentence, the interrogative character of which is just indicated by its intonation, not by a specific interrogative particle. The appearance of the construction rna tagannabtu tannuran rnin-a n-nari barniyan as a primary structure of the type just defined (a type which we saw represented in a number of instances quoted by us) should certainly be considered a quite natural phenomenon, a phenomenon no less acceptable than the analogous (but less complicated) type Zaydunga' a abuhu (,[What about] Zayd? - His father came'), or: Zaydunra'aytu abahu ('[What about] Zayd?-I saw his father'). The syntactical purpose of the construction rna tagannabtu-tannuran rnin-a n-nari barniyan in its interpretation as '(As to) what I avoid-[I avoid] a hot furnace' should be considered identical with the syntactical purpose of the same linguistic form in its alternative interpretation as a fully expressed interrogative sentence followed by its answer: 'What do I avoid?-[I avoid] a hot furnace'. As soon as the initial clause rna tagannabtu in the primary ("twinsentence") construction (rna tagannabtu-tannuran rnin-a n-nari barniyan) combined with the subsequent linguistic elements (tannuran ... ) into a unitary (complex) sentence (constituting a "psychol. subjectpredicate" relation), it represented the structure of a relative clause (,what I avoid', 'that which I avoid') under all circumstances. It constituted such a relative clause (serving as "psychol. subject") not only in the case that it constituted such a structure already in the primary (" 'isolated natural subject' -'natural predicate' ") construction, but also in the case that in the primary construction it constituted the structure of a regular interrogative sentence, rna serving in its original function of an interrogative pronoun (,What do I avoid T). By the process outlined above (p.255), the (initial) interrogative sentence, through its fusion with the answer-sentence, was transformed into a relative clause, and rna acquired the character of a relative pronoun. Above p. 251 (see also footnote 31) we interpreted the participles of the so-called "derived" conjugations ("stems") of Semitic-e.g., Arabic and Akkadian rnuqattil, Arabic rnuqtil, Akkadian rnusakSidas developed from original relative clauses of the type rna yuqattil, rna yuqtil, etc., that is, developed from clauses which were composed of the relative pronoun rna and a verb-form of the 3rd pers. sing. masc. Since the function of the verb of such a clause cannot have been but that of the so-called imperfect, which is the characteristic function of this verb-form in West-Semitic, we necessarily arrive at the conclusion that the preterite of Akkadian (which is in its forms identical

260

NOUNS WITH PREFIX MA-, PARTICIPLES WITH PREFIX MU-

with the West-Semitic imperfect), is based on a proto-Semitic imperfect. This conclusion is strengthened by the structure of the forms of the imperative. The various stems of the imperative in all conjugations of the verb in Akkadian are basically the same as in WestSemitic. In our remarks in Jewish Quarterly Review, vol. 42, 1951/52, p. 51-56 (see above p. 198), we have shown that the forms of the imperative in Semitic are to be considered as derived from the forms of the 2nd persons of the so-called "prefixing" tense by the elimination of the personal prefix: e.g., *qtul (> Akkadian qutul, kusud) < taqtul. Again, the basic function of the verb-forms from which (in protoSemitic) the imperative forms developed, must have been that of the West-Semitic imperfect.

18 THE INFINITIVE IN THE FUNCTION OF "PSYCHOLOGICAL PREDICATE" IN SYRIAC 1 One is wont to assume with regard to Syriac, just as with other Semitic languages, the existence of a special function of the infinitive as a "general object" or "inner object" (in German terminology: "inneres Objekt"). According to this conception, the infinitive may be dependent on, or serve as complement of, a finite verb derived from the same root. Regarding this conception of the Syriac construction we refer to N6ldeke, KurzgeJasste Syrische Grammatik, 2. Aufl., § 295, and to Brockelmann, Grundriss, II, § 203, who treats said construction within the framework of Semitic languages in general. We cite the following examples, on the basis of which we shall outline our own views regarding this Syriac construction. (I) c. Brockelmann, Syr. Chrestomathie (supplement to his Syrische Grammatik), p. 120, 15-16 (Jacob of Sarug concerning St. Simeon Stylites): qiiymii sefaddii 'af es!anii webii'ar kufniis wesiiQrin kense 'af !U/;!iinii demeqmag dammig 'a skeleton was standing on a column, and everybody was looking, and the multitudes were thinking concerning the blessed one that he was just sleeping'. (2) Kalila weDimna, ed. Fr. Schulthess (Berlin 1911), p. 19, 7ff.: ... wemar: kaif hiigan niifel niis bigii gesiine weyiiga' den me[kattas wen Iii memii[ mii'e[, ellii me!!ul liemii ziigeq dene[kattali 'Wenn einer auf diese Weise in die Hand des Feindes fallt und weiss, dass er, ob er kiimpfe oder nicht, sterben muss, so muss er doch urn des guten Namens willen kiimpfen' (Schulthess' translation, p. 21). (3) Spicilegium Syriacum (texts edited by W. Cureton, London 1855), p. I, line 24 ff.: .. .beram d~n do': deman deliappir nelial walmeWpiisu :fiiQe waiflii beryiinii me[qarraQ lurbii galiriirii Iii bayyiiQ weme[nakkaf, menul demaQsiimu maQsem behiillen lehaw demelita"al 'Nevertheless know that he who putteth his inquiries properly, and is willing to be convinced, and draweth near to the way of truth without obstinacy, needeth not be ashamed, because he will certainly give pleasure to him to whom the inquiry is directed' (Cureton's translation, p. 2). (4) Ibid., p.2, line 23 ff.: ... Iii g~r lie[es[ii dehaymiinulii i[ lehan neQnan 'e/~h, welii lulsliinii i[ lehan da'law nesberun, weme!!ul diif 'al Alliihii melpeliigu me[palgin, iif Iii qebleleh iJ behan dehi mefarreqii lehan men kul debliilii ' ... For they have not the foundation of faith to build upon, and they have no confidence upon which they may hope. And because they also doubt respecting God, they likewise have not within them that fear of Him which would liberate them from all fears' (Cureton's translation, p. 3).

I

Originally published in Le Museon (Louvain), vol. 89 (1971), p. 219-223.

262

INFINITIVE IN SYRIAC

(5) Ibid., p. 30, 12fT.: la~nayk dfn halsanna emar l"hon: dilaw Allaha a~a iflsul, "'Ia mehwa hewa wafla me'bat} 'e~it}, w'lsul meddem be$e~ylineh qa'em 'But as for thy children, say to them thus, That there does exist a God, the Father of all, who never was brought into being, neither was He made, and everything subsists by his will' (Cureton's translation, p. 50). (6) Ibid., p. 22, 20f.: ilaw dfn meddem d esarrira'il ilaw w'melq'rf AI/aha ... wehu hana, lawa me'bat} 'e~it} wafla mehwa hewa, ella ilaw men 'ti/am ... ' ... but there is that which really exists, and is called God ... and This same was not made, nor yet brought into being, but exists from eternity ... ' (Cureton's translation, p. 41). (7) Ibid., p. 26, 3: ... wemeddem deme'bat} 'ebi{!, b et}ukk ela dehaw dela 'ebi{! la nettesim lals ' ... and let not anything which has been made be put by thee in the place of Him who is not made' (Cureton's translation, p. 45). (8) Ibid., p. 11,7-8: ... kat} /;Iaywala /;Iranyala ... lawa maw/agu bal/;log mawleglin, ella af meqqas qassin men dalmawlat}u '... while other animals not only procreate, but also become too old to procreate' (Cureton's translation, p. 13). (9) Ibid., p. 8, 21-22: ... deha mekka tel/;1ezf $e~ula, daylen dela hawfn bit}ayn ella men keyana htiwfn lan, la me/;l~ /;Itiybinnan behen wafla mezka zalsfnnan behen ' ... for lo! from hence it is evident, that as to those things which are not done by our hands, but which we have from Nature, we are not indeed condemned by these; neither by these are we justified; .. .' (Cureton's translation, p. 10). (10) Ibid., p. 1, 19-20: maWfanf gfr mesta"elu-(h)u me§ta"elin, lawa mesa"elu mesa"elin 'For teachers are usually asked, and do not themselves ask' (Cureton's translation, p. 2). (11) Ibid., p. 9,11: w'sur/;lanr wemumf t}efagra megdas-(h)u gatjSin w'hawfn leh 'and bodily defects and faults happen and befall him by chance' (Cureton's translation, p. 11).

As we already mentioned above, the Syriac infinitive preceding a finite verb (or predicate of a participial nature) formed of the same root, constitutes, according to generally accepted theory, a complement, a so-called "general" or "inner" object of the finite verb (or of a predicate of participial nature). If we look, however, more carefully at the above-quoted instances, then we are struck by a certain (seemingly) formal feature of two of these examples, i.e., no. (10) and no. (11). In these two instances, the infinitive is characterized by the enclitic pronoun (h)u. Generally, this enclitic (h)u serves in Syriac to emphasize a single (nominal) term within a sentence. In other words: this (h)u characterizes the so-called" 'psychological' predicate" as against the remaining parts of the statement, which constitute its "'psychological' subject". The Syriac construction in which the "'psychological' predicate" is characterized by enclitic (h )u, constitutes a special type of the constructions implying a "'psychological' subject-predicate relation (we refer to our Studies in Arabic and General Syntax, Cairo 1953, §40ff., and especially to §46). In accordance with this conception of the two passages referred to-no. (10) and no. (11) of the instances quoted above-we analyze no. (10) strictly literally as follows: '[As to] teachers: it [is] being-asked-

INFINITIVE IN SYRIAC

263

they are asked; it is not (liiwii < lii-hewii = law < Iii hu) askingthey ask'. Along the same lines, we analyze instance no. (11): 'and [as to] bodily defects and faults: it [is] happening-by-chance-they happen by chance'. In a clearer and more adequate-though less literal-analysis, the two statements could be interpreted as follows: no. (10): '[as to] teachers: it is being-asked [what] "is done by them" (or: "happens to them"); it is not asking [what] "is done by them"'. And no. (II): 'and bodily defects and faults: it is happening-bychance [what] "they do" '. In the first, very literal analysis of the two passages, the two basic terms of each respective passage, that is: the" 'psychological' subject" and the "'psychological' predicate", are represented as two independent sentences. The pronoun (h)u 'it' in the first one of these two sentences (constituting a single statement) represents a symbolical (pronominal) expression of the "'psychological' subject". That is: (h)u is the subject (= '''psychological' subject") of 'being-asked', 'asking', and 'happening', respectively, expressions which, in turn, constitute the" 'psychological' predicate" of the respective statements. The "'psychological' subject" of each statement as a whole follows then in an explicit ("non-symbolical", "non-pronominal") form, as an independent, epexegetical sentence, which is added to the first sentence after a pause of minimum duration: 'they are asked', 'they ask', and 'they happen', respectively. The other instances of the syntactic type under discussion, in which the infinitive is not characterized by the enclitical pronoun (h)u (or by its negative equivalent: ltiwti < lii-hewti = law < lti-hu) should be viewed in the same manner as those characterized by (h)u (see above). That means: in all instances of this Syriac construction, both the infinitive and the finite verb (or an equivalent participial form) following the infinitive are independent terms which relate to each other as ("psychological") predicate and subject. The special feature of these instances without enclitic pronoun is that here the "'psychological' subject" has found no expression at all in the initial sentence of the basic "twin-sentence" construction, not even a "symbolical" expression by means of a pronoun (as in instances no. (10) and no. (11); see above). Accordingly, no. (1) of the instances quoted above (p. 261) should be analyzed as follows: "A skeleton (that is: St. Simeon Stylites) was standing on a column, and people were looking and thinking concerning the blessed one that sleeping (gerund) [it is]-[he] sleeps", or in a less literal, but completely adequate

264

INFINITIVE IN SYRIAC

interpretation: "[they] were thinking that It is sleeping [what] he does (i.e., 'he just sleeps')". Of course, in all of the afore-quoted instances (whether they do or do not contain the pronoun (h Ju) the original "twin-sentence" construction could develop into a single (unitary) sentence; such a development was in general inescapable. The final result was therefore a single (unitary) sentence involving a "'psychological' subject-predicate" relation. In the syntactic construction thus developing, the new term of the statement, which is unknown at the start of the utterance, that is: its ("psychological") predicate, is constituted by the infinitive (describing an action or process in a form in which it is detached from its "occurrence"), while the "occurrence" of the action or process, a concept in the mind of the speaker from the start of the utterance, constitutes the ("psychological") subject of the statement. While most of the instances of Syriac in which an infinitive accompanies a finite verb of the same root, are of the type of the abovequoted examples, in which the infinitive always precedes the finite verb, there are certain other instances in which the infinitive follows the finite verb. Characteristic instances of this specific type are the following examples (quoted by N6ldeke, I.e., §296; Brockelmann, I.e., p. 294, last line): qiim lals meqam 'steh doch auf', haymen mehaymanu 'glaube nur', dUals d elemar memar 'dein ist's zu reden'. Contrary to current opinion (represented, e.g., by N61deke and Brockelmann), we do not feel that the finite verb (imperative; or its paraphrase, as in diliils d elemar memar) and the infinitive formed here originally one unit represented by a (finite) verb and its complement (object). Rather, it is evident that the infinitive follows here the finite verb (imperative, or its paraphrase) after a pause of minimum duration. We do not feel that this special type is essentially different from the type discussed above. We hold that the sentences should be analyzed as follows: 'Rise!-Rising [is what you should do]', 'Believe!-Believing [is what you should do]', and 'Yours is to speak!-Speaking [is what you should do]', respectively. That is: the basic, "grammatical" construction represented by a formal imperative (or a paraphrase of an imperative), is repeated in a second (linguistically incomplete) sentence, in which the idea is represented in a structure involving a "'psychological' subject-predicate" relation. In this second sentence involving a "'psychological' subject-predicate" relation, the term representing the" 'psychological' subject", though present in the mind of the speaker, remains-exactly as in the type

INFINITIVE IN SYRIAC

265

discussed earlier-unexpressed, for it is implied in the situation. The function of the infinitive in stressing an action by indicating it in a form in which it is detached from its 'taking place', appears also very clearly in the following instance (quoted by Noldeke, I.e., § 297), in which two infinitives appear without the "occurrence" of the actions (expressed by the infinitives) having found any expression (by finite verbs, as in the instances quoted above): wal-Pawlos bazf2an mesar wef2azf2an mergam 'und den Paulus nahm man bald gefangen, bald steinigte man ihn' (Noldeke's translation). The two actions ('capturing' and 'stoning') described in this passage are emphasized, since they imply a kind of contrast. The "occurrence" (i.e., the" 'psychological' subject") of the two actions which appear in the function of '''psychological' predicate", is here implied in bazf2an ... wef2azf2an ... ('once ... , and then again ... '). This repeated temporal adverb clearly indicates the "occurrence" of two actions, so that special expressions for the "occurrence" of the actions, by means of finite verbs, are felt to be unnecessary. Sentences containing a "'psychological' ('logical') subject-predicate" relation constitute sentences with a nominal predicate ("nominal sentences"). In the present case this nominal predicate is represented by an infinitive. It is perfectly clear that the infinitive acting as a predicate cannot be considered an "inner object" (that is: a grammatical complement) of the finite verb which (normally) appears beside it. There can be no doubt that the infinitive in the function of "psychological predicate" is not only found in Syriac, but also in other Semitic languages, e.g., in Hebrew. I refer here to the following Hebrew instance (Deuter., 22,6-7): (6) ki yiqqiire qan-$ippor lefiin?lsii baddfrfls belsol 'e$ '0 'al-hii'iirf$ 'ffro/Jim '0 be$im wehii'em rof2f$fl 'al hii'ffrol:zim '0 'al habbe$im-lo-liqqal:z hii'em 'al habbiinim. (7) sallegl:z t esallal:z ' fl hii' em we' fl habbiinim tiqqal:z liils ... . I am quoting the current English interpretation (The New English Bible: The Old Testament [Oxford University Press, 1970], p.263): '(6) When you come to a bird's nest by the road ... , with fledglings or eggs in it and the mother-bird on the nest, do not take both mother and young. (7) Let the mother-bird go free, and take only the young; ... '. The infinitive sallegl:z (in v. 7) serves certainly as "psych. pred." (which implies that it is stressed). The specific function, which, accordingly, is expressed by [the infinitive] sallegl:z (in sallegl:z tesallal:z ... ), may somehow be indicated by the following German interpretation: ' ... Gehen lassen sollst du den Muttervogel, und nimm [nur] die jungen

266

INFINITIVE IN SYRIAC

Vogel' (with 'gehen lassen' stressed), or in English, by the [quasiartificial] paraphrasis: ' ... Letting [her] go free is what you should do with the mother-bird, and take [only] the young .. .'.

19 THE EXPRESSION OF INSTANTANEOUSNESS IN ARABIC (WITH ATTENTION TO THE CORRESPONDING USAGE IN WESTERN LANGUAGES) 1

A. The Construction rna ra'ahii illa bi Arabic possesses a specific syntactic construction which serves the description of an action which occurs while another earlier action (or process) has hardly been completed, and which thus implies an element of suddenness and surprise. For examples of this construction we refer to N61deke, Zur Grammatik des klass. Arabischen, § 73 (p. 76); Reckendorf, Arab. Syntax, §178,3 (p.359) and §262,8 (p.51O); Brockelmann, Grundriss, II, § 65 a. A. Spitaler has devoted a special study to this phenomenon, in his article "Ma ra'ahu ilia bi und Verwandtes", in Serta monacensia, Franz Babinger ... dargebracht (Leiden 1952), p. 171-183. One of the most important variations of this construction is the type: ma ra' ahu ilia bifulanin (or: bisai' in) 'suddenly this or that person (or thing) took him by surprise'. The noun which expresses the suddenly occurring entity is (seemingly) dependent on a verb which denotes fright or surprise and is very often introduced by the preposition bi. But the noun can also occur in the nominative. Very frequently the noun is followed by a complementary expression of noun-like or verbal character (which serves, as it were, as predicate to the preceding noun). We give the following examples, partly taken from the above-quoted works. Fama ra'a I-mar' ata ilia bi-Ifubaibin (TabarI, Annales, I, 1435,4), or with a variant: ... ilia Ifubaibun 'Suddenly I:Iubaib stood in front of the woman'. Lam yaru'ni ilia Rasulu-llahi ¢uban 'unexpectedly I perceived God's messenger in the morning' (Bubari, Sabib, ed. L. Krehl, III,34,11). Ma ra'ani ilia bamulatu ahliha wasta d-diyari 'unexpectedly, I perceived the 1 Originally published in Le Museon, revue d'etudes orientales (Louvain), vol. 84 (1971), p. 499-523; vol. 85 (1972), p. 527-529.

268

INSTANTANEOUSNESS IN ARABIC

beasts of burden of her people in the dwellings' (,Antarah, Mu'all., 11). Fama ra' a I-mar' ata walaha ~abiyun yadrugu ilia lfubaibun qad aglasa Nabiya 'ala fabirjihi 'suddenly the woman-who had a boy with herself who tripped along-noticed that I:Iubaib had placed (or: was placing) the boy on his thighs' (Agani, 1st ed., 4,42, 8). Falam yaru'-i I-qauma wahum fi ri~alihim ilia bir-rigali bi' aidihimi s-suyufu qad gasuhum (Tab. I, 1432,7), or, with a variant in which bir-rigali is replaced by the nominative (Ibn Hisam, Sirah, ed. Wiistenfeld, 638, -3): falam yaru' -i I-qauma ... ilia r-rigiilu ... 'suddenly the people, while in their quarters, became aware of the fact that the men, with swords in their hands, fell upon them'. Falamma mata fa'ala galika bihi lumma wa(la'ahu 'ala qari'ati t-tariqi wa'aqbala 'Abdu-llahi bnu Mas'udin fi rahtin min ahli I-'Iraqi 'ummarin falam yaru'hum ilia bil-ginazati 'ala ~ahri t-tariqi qad kadat-i I-ibilu tata' uha 'and when he (i.e., Abu parr) died, they (i.e., his wife and his servant) did this with him (that is, what he had ordered them to do), then they placed him on the middle of the road. [Thereupon] 'Abdallah b. Mas'ud came with a number of pilgrims from 'Iraq, when they suddenly came upon a corpse. The camels had almost stepped on it' (Ibn Hisam, 901,10-11). In a variation of the construction, we find instead of a verb denoting 'to frighten, to surprise' a verb denoting sensual or mental perception. The surprised person is naturally the subject of the verb in this construction. The suddenly appearing entity (or, the subject of the action which is perceived, respectively) is also introduced by bi here or appears in the accusative (as required by the governing verb). We quote a few examples of this construction. Fama yaS'ur-i I-Mungiru ilia bibni Mufarrigin qad uqima 'ala ra'sihi 'and al-Mungir was aware of nothing-when Ibn Mufarrig was placed in front of him' (Agani, 19,256). Falam yaS'uru ilia bil-kamini ya(lribuhum fi adbiirihim 'suddenly they were attacked from behind from the ambush (or: they were aware of nothing-when they were attacked from behind from the ambush),. Lam yagid ilia ragulan qad abaga bimankibihi 'suddenly he sensed someone seized him by the shoulder' (Hugailian poems, ed. Wellhausen, no. 202, Introduction, line 4). Falam tara illa I-qaum( a) 'adina min gawanibi I-wadi 'thereupon--quite unexpectedly-the enemies attacked from the sides of the Wadi' (Ibid., no. 175, introd., p. 26, 10). In other instances we find-as otherwise in regular sentences expressing "exception" -ilia wa used (instead of the simple ilia).

INSTANTANEOUSNESS IN ARABIC

269

We adduce the following examples: falam yafga' ijalidan ilia wassufunu gawani/:zu 'unexpectedly for Ijalid the ships listed' (Tab. I, 2038,2). Falam tas'ur Banu J)ubyana ilia wal-bailu dawa'isu 'the Banii Qubyan were unsuspecting-when the horses (with their riders) trotted one behind the other' (Naqa'itj, ed. Bevan, 95,4; cf. further examples in Reckendorf, Synt. Verhiiltnisse, p. 725; Brockelmann, Grundriss, II, p. 651; Spitaler, loc. cit., p. 179). We turn now to a discussion of the syntactical mechanism which underlies this construction. Noldeke, Zur Grammatik des klass. Arab., p. 76, Reckendorf, Arab. Synt., § 178, 3 (p. 359), Brockelmann, Grundriss, II, p. 124, see in the type in which the noun immediately following ilia is introduced with bi-as in ma ra'ani ilia bi-lfubaibin ... (Tab., I, 1435, 4)-a case of impersonal use of a verb. This explanation (as an impersonal use of a verb) is given by Brockelmann (I.e.) also for the type in which wa occurs between ilia and the following clause (beginning with a noun): he assumes that bi has been replaced here by a circumstantial clause introduced by wa. In accordance with this conception of the afore-mentioned types, Reckendorf (I.e., p. 359, n.3) describes the type in which ra'a is followed by a noun in the nominative, as the corresponding "personal" construction. Thus, in a sentence like (Antarah, Mu 'all. , 11) ma ra'ani ilia /:zamulatu ahliha wasta d-diyari, he considers /:zamulatu as the subject of ra'ani. Spitaler shares this view (I.e., p. 174ff.). The original meaning of sentences like ma ra'a I-mar' ata ilia lfubaibun (,suddenly the woman perceived I:!ubaib') or lam yaru'ha ilia $iya/:zu qaumin (Tab., III, 1709, infra: 'suddenly they heard the shouting of the people'), is, according to Spitaler (I.e., p. 175): 'Es war I:!ubaib, der [durch sein Erscheinen] die Frau erschreckte', and 'Es war das Geschrei der Leute, das sie [durch sein Ertonen] erschreckte'. Spitaler (I.e., p. 174) opposes the view that the use of this syntactical construction for the expression of surprising events is essentially connected with the double negation which occurs in it; rather, its essential feature-in his opinionis the verb ra'a (or: fagi'a) 'to frighten, to surprise'. However, if these latter verbs rather than the double negation contained in ma-ilIa form the essence of the construction, how are we to understand the instances in which verbs of perception appear in the place of verbs which denote fright or surprise? Further, Spitaier fails to explain the purpose of the double negation. Mara' a I-mar' ata ilia lfubaibun is, for Spitaier, "einfach das mittels rna-ilia negativ gewendete Gegenstiick zu einem positiven ra' a I-mar' ata lfubaibun,

270

INSTANTANEOUSNESS IN ARABIC

bzw. rifa lfubaibun-i I-mar' ala". Spitaler's identification of the construction with the double negation with the "positive" construction which lacks this double negation assumes that the latter type, that is, rii' a I-mar' ala lfubaibun, is used in the same sense as the former type, with the meaning of 'suddenly the woman saw I:Iubaib'. However, we never find this construction used in this manner. On the other hand, the interpretation of the expression which follows illii as the subject (or object, respectively) of rii' a or similar verbs becomes unacceptable in view of the construction in which illii is followed by wa, that is by a(n original) circumstantial clause. Spitaler (l.c., p. 180) remarks on this point: " ... ; ein solches (i.e., ein nominales Subjekt) kann aber nie durch einen Zustandssatz ersetzt werden. Infolgedessen bleibt nichts ubrig als wieder eine kontaminatorische Storung durch eine anders geartete, aber formal iihnliche syntaktische Erscheinung anzunehmen, in der der syndetische Zustandssatz berechtigt ist. Diese bietet sich ungezwungen in gewohnlichen Ausnahmesiitzen vom Typus mii min garil;in yugra/:tu fi sabili lliihi illii wa-lliihu yab' aluhu yauma I-qiyiimati, b. Hisam 586,4; ... Es ist ganz klar, dass die obigen Beispiele nach diesen Mustern gebildet sind". In the 'kontaminatorische Storung' assumed by Spitaler, we deal with a transfer of the adverbial clause introduced by wa into the role of the subject. Wa follows illii in sentences whose predicates have the meaning of 'become aware, observe' (see above), e.g., sa'ara. In Spitaler's opinion, the group of words introduced by illii forms the object of these verbs of perception. However, a circumstantial clause cannot possibly be either an object or a subject. Furthermore, an object dependent on sa'ara is usually introduced by bi (cf. below, p.290). Spitaler (p. 180f.) distinguishes two types in sentences with sa'ara in which illii wa appears. One type occurs rather frequently, e.g., '/qd al-farid, III, 360, -10: famii sa'artu illii wa' ana fi z-zuqiiqi 'suddenly I found myself in the street'. Spitaler assumes that bi has been omitted here because a "syndetic" circumstantial clause cannot be dependent on bi. He also remarks: "Es liegt hier einfach eine Kontamination mit dem freilich seinerseits sekundiiren oben besprochenen mii rii'ahu illii wa- vor" (concerning this latter construction see above p. 269). In the second type, sa'ara is followed by a nominal or pronominal complement which precedes the clause introduced with illii wa, as in Naq., 662,4: Iii tas'uruna bihim ilIii wahum mun/:tadiruna 'alaikum 'they will descend upon you unawares'. Spitaler (l.c., p. 180) assumes that this latter type has its

INSTANTANEOUSNESS IN ARABIC

271

ongm in a construction in which the clause with wa following ilia was dependent on hi. However, since a circumstantial clause with wa cannot be dependent on the preposition hi, he assumes that "das (grammatische oder logische) Subjekt des Zustandssatzes isoliert [ist], aber gleichzeitig mitsamt dem hi- vor das ilia gezogen worden". In answer to this statement, we would like to remark that the natural linguistic instinct cannot help considering the phrase la ta.S' uruna hihim (as it appears in the above quoted example) as the primary, natural expression for 'you are not becoming aware of them'. In this connection, Spitaler also discusses the phrase ma sa'ara /:latta ... , as, e.g., in (b. Qutaiba, 'Uyun, 197,14): fama sa'artu /:latta arani I-a'lama 'and suddenly he showed me the signs'. As in the case of the first construction with sa'ara, he assumes here also that it resulted from a "contamination" (blending) with constructions like ma nfagara l-fagru /:latta faraga 'it had not dawned yet, and he had finished'. Here, too, there is absolutely no reason to doubt that the construction (as, e.g., fama sa'artu /:latta arani I-a'iama) represents, from the genetical point of view, a primary, original structure, not having been affected by a "contamination" with any other construction. The construction appears already in very old examples, as in the following verse (from a poem ascribed to the Elder Muraqqis, 6th cent.; Mufatjtjaliyat, no. 52, 4):fama sa'ara I-/:Iayyu /:latta ra'au hayatja I-qawanisi fauqa I-gurar, in Lyall's translation: 'And the tribe [they beset] knew nought of their coming, until they saw the peaked helmets' sheen over the horses' blaze', which would have to be corrected to: 'over the foreheads of the horsemen' (for this meaning of gurrah, see Qais b. al-IJattm, no. 12, 7: gurratu garatin 'white forehead of a woman-client'). Since the construction ma sa'ara /:latta lacks a noun which could be construed as the object of sa'ara, as is the case also in the lastmentioned, more complicated one (ma sa'ara ilia wa), Spitaler again makes the assumption that we are dealing here with a secondary construction which resulted from a "contamination". He assumes that primary constructions containing ilia wa were blended with constructions like rna nfagara l-fagru /:latta faraga 'it had not dawned yet, and he had finished'. He adds the remark that this "contamination" (blending) could take place so much more easily since in expressions like the last-mentioned ones /:latta could be replaced by ilia wa, and refers to examples given by him in his article, p. 175, n. 1, as, e.g., (Tab. I, 1876,12, quoted from Reckendorf, Arab. Synt., p. 509, -4),

272

INSTANTANEOUSNESS IN ARABIC

rna !ala'a l-fagru illa wah urn wal-'aduwu fi :ja'idin wabidin, and ('Imad-ad-Din, al-Fatb al-Qussi, ed. Landberg, 80, 8) warna ntabahii illa wasufunu l-Firangi bihirn rnubdiqatun. Spitaler's assumption that the construction rna sa'artu balta expressing a surprising action has resulted from a "contamination" with expressions like rna nfagara l-fagru baltafaraga (Tab., II, 620, 19; see above) is difficult to reconcile with his opinion that constructions like rna !ala'a l-fagru ilia wahurn ... lack the element of surprise as contained in constructions with rna sa' ara, rna ra' ani (p. 179, n. 1). He interprets the first of the two examples quoted (Tab., 1,1876,12) as follows: 'Kaum war der Morgen angebrochen, da befanden sie und der Feind sich auf einem Plan'. However, with regard to the second example (al-Fatb al-Qussi, 80, 8), which is entirely analogous, he remarks: " ... Schon eine Wiedergabe mit 'kaum' - 'als' ware fehl am Platze". These examples, as well as others cited below prove that in constructions of this type balta and illa (wa) are equivalent and that all these constructions are to be considered as of a primary nature. Furthermore, we must stress that the element of surprise is decidedly the essential feature of these statements-also of the examples which supposedly lack this element, according to Spitaler. We might make this element more prominent by translating as follows: 'When it dawned-Io and behold!-they and the enemy were on the same plain', and 'And when they awoke-Io and behold!-the ships of the Franks had surrounded them'. These constructions are therefore to be considered basically identical with the constructions of the type rna ra'ahu illa ... (rna sa'ara illa ... ) from which we started (see above p. 267-271). A corollary of this analogy is also that the subject or object of ra'a and sa'ara in these latter constructions is not to be sought in the word group introduced by illa (as indeed has been assumed by Spitaler and others, see above p. 269). The sentences describing events which occur as a surprise bear a close affinity to sentences which denote a quick succession of two actions, or, more accurately, the quick succession of a principal one after a secondary one, of the type: 'Hardly had he entered, when ... '. Such sentences in Arabic employ constructions with rna ... illa, or rna ... balta (cf. the above-cited examples, and further examples below, p. 273), and their formal structure is identical with the sentences which describe surprising events. But they are also related to them in content. For quick succession of one event after another most frequently implies the element of surprise of a greater or lesser degree.

INSTANTANEOUSNESS IN ARABIC

273

Thus, an analysis of the sentences of the "hardly" type promises to shed light on those which contain this element of surprise expressly (ma ra'ahu ilia ... , etc.). As an example of a sentence which denotes quick succession of two events, we quote the following passage (Agani, 19, p. 284): " ... Zuhair had a scout from the Ginn, and he told him of their matter (that is, of the plans of the enemies regarding an attack) until it was morning. And he had a spot overshadowed by trees (mi~allat daul;.) in which he-out of fear of events (of fate)-used to tie up his horses so that they never left him. And when it was morning, one of the horses which had sensed the (approaching) riders and horses neighed. Thereupon Zuhair said: 'What is the matter with it (i.e., the horse)?'. Whereupon the scout replied: 'It senses the horses with their riders and it neighs towards them! Falam tu'ginhum bihim ilia wal-I;ailu dawasin (variant: dawa'isu) mal;.arjiru bil-qaumi gudayyatan. And hardly had it (i.e., the horse) warned them of their approach, when, lo-and-behold, the riders arrived riding on their horses, one after the other, on an early morning". In a more literal translation, this sentence says: "It (that is, the horse) had not (yet) informed them of their approach, [when nothing happened] but (that) the riders arrived riding on their horses, who came trotting". We quote another example (BaIaguri, Ansab al-asraf, vol. 5, p. 15, 19-20): inna 'Umara bna l-ffattabi I;ataba n-nasa yauma gum'atin fmjakara n-Nabiya (~l'm) wa'Aba Bakrin lumma qala inni ra'aitu ka'anna dikan naqarani wala arahu ilia I;.urjuru aga/i '''Umar b. all:Jattab addressed the people on Friday and mentioned the Prophet and Abu Bakr, then he said: [In a vision] I saw a cock peck me (that is, my eye); and hardly did (literally: do) I see him, when my end was at hand". In a literal translation: "... I do not see him (that is: not yet) [thereupon nothing happened] except the taking place of my end". In both examples quoted here the statement of an event which has taken place is followed by a sequence of two closely connected clauses: the first deals with the same event and states that it has not (yet) taken place: "It (that is, the horse which neighed [see our first quotation]) had not (that is: not yet) informed them", and "I do not see him (= the cock) (that is: not yet)". This is followed by another sentence which describes a major event immediately following the former. As already indicated, the first of these sentences can, less literally, be most simply rendered by "hardly ... ". The two examples of such a sequence with "hardly" cited here exhibit a peculiarity

274

INSTANTANEOUSNESS IN ARABIC

not usually found otherwise: they are preceded by a sentence in which the event which has not yet occurred, according to the literal expression, is designated as having already occurred, and must necessarily be assumed as having already occurred: "The horse neighed towards the approaching riders [hereby warning its master of the imminent danger]. And it had not yet warned him, when ... ". Or: " ... I saw a cock peck me. And I do not see him, when my end was at hand". This is an important feature of these two examples: the implications of this feature for our two examples will also hold true for other pertinent examples not preceded by such positive statements. In these two examples, the secondary (modifying) action described as not yet having occurred has undoubtedly already occurred before the principal action has taken place. We would therefore expect a positive, not a negative statement: "It (that is, the horse) had [just] informed them", and "I see him (= had (just) seen him [namely, the cock])". The problem is, therefore, why the secondary action which already occurred is stated as having not yet occurred; or, in other words, how the negation which occurs in many cases of constructions with "hardly" is to be explained. Although both sentences are independent and not subordinate clauses, they are, nevertheless, with respect to their content in a very close relationship, closer than the usual relationship between two ordinary independent, non-subordinate sentences following one another. Each sentence is unthinkable without the other. The action expressed in the first sentence which occurs just at the moment when the second action (expressed in the second sentence) takes place -or, more accurately, at a slightly earlier moment-is designated as having not yet occurred, in a kind of exaggeration and in order to underline the fact that it has occurred just now and at this moment. The instantaneous succession of the second action after the first ("hardly", "no sooner ... than") is underlined in these Arabic examples by the particle ilIii ("except", "if not") which introduces the second sentence. We stress that this particle introduces a second, independent sentence. This is to say that it does not, in any way, supplement the negation of the first sentence and has no direct syntactic connection with it (as must be apparent from our analysis of that first sentence). The negation which we postulate for the second sentence and the verb negated by it, both presupposed by the particle ilIii, are not expressed here, but only mentally supplied. And what is the purpose of the replacement of a positive statement ('the horses

INSTANTANEOUSNESS IN ARABIC

275

came trotting', and 'my end was at hand') by the (doubly) negative form [ma]-ilIa? IlIa-whose literal meaning, based on its component parts, is 'if not, unless' - implies a temporal connotation: "After the secondary event had 'just' taken place, nothing happened but [that] the event took place", that is: " ... nothing happened before the event took place". That is: after the first (secondary) event had taken place, the second (principal) event took place so promptly that no other event which could have conceivably intervened took place or could have taken place. The function of illa in the examples relevant here is identical with that of the less ambiguous ~atta 'until' which occurs in similar examples (cf. the examples mentioned above p.271272). As already indicated, we relate illa (and ~atta, respectively) in what we consider the second sentence of the construction, not to the negation which introduces the first sentence, but consider the negation (and the predicative expression negated by it) which illa (or ~atta) necessarily presupposes as preceding it more closely, but only mentally supplied, not expressed linguistically. This analysis of the construction assumes that illa (or ~atta) and the predicative complements which follow it constitute a new, independent sentence. In the above (p. 273) cited sentences (quoted there within their larger contexts): falam tu'c}inhum bihim ilia wal-bailu dawasin (dawa'isu) , and wala arahu ilia ~ur;/uru agali, the words introduced by illa thus constitute an independent communication in relation to the preceeding word-group introduced by a negation. The sentence introduced with ilia is given prominence in relation to the communication contained in the preceding sentence. The illa-clause must be considered as an independent sentence, independent of the preceding sentence (with simple negation), as the second sentence in the following English (or German) equivalents of this construction: "Hardly had I seen the cock (Kaum hatte ich den Hahn gesehen)-when my end was there (da trat mein Ende ein)", or: "I had just seen the cock (Ich hatte gerade den Hahn gesehen)-when my end was at hand (da trat me in Ende ein, or: als mein Ende eintrat)". We now revert to the types in which the surprise element is especially prominent, the types to which Spitaler has devoted a special study (see above, p.267). We consider first the patterns represented by the examples cited above (p. 270): fama sa' artu ilia wa' ana fi z-zuqaqi ('Iqd al-farid, III, 360), or: la tas'uruna bihim illa wahum mun~adiruna 'alaikum (Naqa'ir;/, 662,4), or: fama sa'artu ~atta arani l-a'lama (Ibn Qutaibah, 'Uyun, 197, 14). In these examples, illa or ~atta are not

276

INSTANTANEOUSNESS IN ARABIC

immediately followed by a nominal expression which - in accordance with Spitaler's analysis of the type ma ra' a I-mar' ata ilia Ijubaibuncould be considered as the subject of the verb involved. Spitaler therefore considers these types as secondary transformations of the latter type, through blending ("contamination") with other constructions (see p. 270). As we have already indicated, we consider these clauses, in contrast to Spitaler, as primary constructions (see above, p. 271). They can be interpreted, without depriving them of their surprise connotation (see the interpretations given above, p. 270-271), as follows: 'Hardly (or: just) had I noticed [something], when I found myself in the street'. 'Hardly will you catch sight of them, when they already descend upon you'. 'Hardly had I noticed [something], when he [already] showed me the flags'. We thus completely equate the essence and structure of these sentences with the essence and structure of those types which describe the instantaneous succession of two events. Accordingly, we may analyze these statements more literally as follows: 'I did not notice [anything] (or: I had not noticed [anything])-[Thereupon nothing happened] except that I found myself in the street', and: 'You will not catch sight of these-[Thereupon nothing will happen] except that they already will be descending upon you', and: 'And I did not notice [anything]-[Thereupon nothing happened] until he showed me the flags'. The same interpretation applies, of course, in those cases in which verbs of surprise are used, as falam yafga' ljalidan ilia was-sufunu gawanibu. Accordingly, we interpret here: 'Not (or: "not yet", or: "just") [something] had frightened Ijalid-[Thereupon nothing happened] except that the ships listed'. We go beyond that and interpret in the same way also such cases with an implied surprise element which Spitaler obviously considers as primary (not brought about by "contamination", as the above-mentioned examples). Thus, we interpret the above p. 268 quoted statements: fama ra' a I-mar' ata ... ilia Ijubaibun q(1d aglasa Nabiya 'ala fabiejihi ( Agani, 4, 42, 8), or: wa' aqbala 'Abdullahi bnu Mas'udin fi rahtin min ahli I-'Iraqi 'ummarin falam yaru'hum ilia bil-ganazati 'ala ~ahri t-tariqi ... (Ibn Hisam, 90 I, 11-12) as follows: 'Hardly (or: "just") had [something] frightened the woman-[Thereupon nothing happened] except [that] l:Iubaib had already seated the little boy on his thigh'; or, more literally: 'Not [yet] had [something] frightened the woman-[Thereupon nothing happened] except .. .'. Furthermore: "Abdallah b. Mas'ud came with a number of pilgrims from 'Iraq; and hardly (or: just) had something frightened them-

INSTANTANEOUSNESS IN ARABIC

277

[Thereupon nothing happened] except [that] a corpse appeared on the way in front of them ... '; or, more literally: '... and not had [something] frightened them-[Thereupon nothing happened] except [that] .. .'. The same interpretation holds in cases in which illa is followed by a noun only, without complement-not by a clause. (This noun may be in the nominative or may be introduced by the preposition bi). We quote the following instance (Tab., 1,1435, 1): Jama ra'a l-mar'ata illa bi-lfubaibin 'Hardly (or: just) had [something] frightened the woman-[Thereupon nothing happened] except that I:lubaid suddenly stood before her'. Or, more literally: 'Not [yet] had [something] frightened the woman-Thereupon ... '. We also quote the following example (Sil'idah b. Gu' ayyah, no. 2, 33; Hell, Neue Hwjailiten-Diwane, p. 19): ... walam yarta' lahum Jaza'un /:latta ra'auhum bilala s-sabyi wan-na'ami, which Hell renders as follows: " ... und man hatte von ihnen noch keinen Schrecken (d.h. Uberfall) erlebt, bis man sie (eines Tages) inmitten der Gefangenen und des Viehes sah". However, a more meaningfull translation is: " ... und plotzlich sah man sie (i.e.: die Feinde) inmitten der zu Gefangenen gewordenen und deportierten Frauen und des [erbeuteten] Viehes" (for content, cf. Niibigah, ed. Ahlwardt, no. 20, vv. 11-13 and vv. 15-16). And in literal interpretation: "... and not [yet] (or: hardly, or: just) had they become aware of a threatening danger and becomefrightened-[Thereupon nothing happened] until they saw them (that is, the enemies) in the middle of the women and the cattle". That is to say: Between their becoming aware of the threatening danger and the actual appearance of the enemies an infinitesimally small amount of time elapsed, which did not (or: hardly did) permit them to defend themselves. In this context we refer also to the following line (MuJar/4aliyiit, no. 90, 8): Jama Jazi'ii i4 biila!a l-qaumu ahlahum waliikin ra' au :firJan min-a l-mauti ashabii, which Lyall (Translation, p. 257) renders: "And they blenched not, what time our host dashed into theirs, but they looked full-face at Death unmixed, the dusky" (an interpretation adopted by Wolfdietrich Fischer, Farb- und Formbezeichnungen in der Sprache der altarabischen Dichtung, Wiesbaden 1965, p.373). We see in the line an instance of the construction under discussion and interpret it as follows: "Hardly had they become alarmed (or: frightened) at the moment when our host dashed into their crowd (or: hardly had they become aware of our host's dashing into their crowd), when they already met unmixed (i.e.: outright), dusky death". Or in a more literal analysis: "They

278

INSTANTANEOUSNESS IN ARABIC

had not [yet] become frightened (or: become aware of an imminent danger) at the moment of our host's dashing into their crowd (or: They had just become aware of a danger at the moment of our host's dashing into their crowd)-[Thereupon nothing occurred] before (more literally: except [that]) they met unmixed (i.e.: outright), dusky death". That is: "Extremely little time passed between the start of the attack and their being slain". We conceive waliikin here in the meaning of ilIii (an) 'except [that], but' (cf. below p. 342). We add here another sentence which in our opinion is of the type of construction here under discussion. Gai)i?, Kit. al-Bubalii' (ed. Taha al-J:lagiri, Cairo, l)a!:Jii'ir al-'Arab, 23, p. 79,11): lam yafga'ni qa!!u wa' ana iikulu tamran ilIii staffahu saffan wabasiihu baswan wazadii bihi zadwan. Ch. Pellat, in his translation of the book (Le livre des avares, p. 113), translates this sentence as follows: "11 n'est jamais venu une seule fois me surprendre devant un plat de dattes sans les ingurgiter, les engloutir et les faire disparaitre". Rescher's translation (in his Excerpte und Uebersetzungen aus den Schriften des Philologen und Dogmatikers Giibit aus Bacra ... , Teil X, Stuttgart 1931, p. 113) is similar: "Und findet er sich dann bei Gelegenheit einmal ganz unverhofft ein, wenn ich gerade Datteln esse, dann ... packt er gleich die Schiissel ... hebt sie von der Erde auf ... ". We interpret as follows: "Nothing at all had frightened me (or: I had not yet been aware of anything = I had just become aware of something), while (or: since) I was [engrossed in] eating dates-when (or: [then nothing occurred] but) he [snatched them away], swallowed them and gulped them down and ... ". As our preceding interpretations show, we analyze the act of being taken by sudden surprise-if expressed by the construction mii rii'ahu ilIii and similar expressions-as consisting of becoming aware of the approach of the unexpected action (or its agent) and its quick or instantaneous happening. The realization of the danger confirms, so to speak, the presentiment. It seems that in the event of being taken by surprise, phases can be distinguished. They are discernible in the examples discussed so far, although they appear there, so to speak, in a compressed form. In a more distinct manner the two phases of the event of being taken by surprise are indicated in the following passage (Sa'idah b. GU'ayyah, no. 1,46-52; Hell, Neue Hu4ailiten-Diwane, p.11-12): (46) bainii hum yauman ka4iilika rii'ahum r;labrun libiisuhumu l-badidu mu'allabu ... (52)fababat katibatuhum w~addaqa rau'ahum min kulli faggin giiratun la takrjibu

INSTANTANEOUSNESS IN ARABIC

279

"While they one day were thus (that is, unsuspecting and lulled in security), they became frightened by a(n approaching) group of ironclad warriors united in a close-knit group ... and their (that is: this) group of horsemen came ready to fight, and their fright was confirmed by an unmistakably real attack from every crevice". We come back to the above quotation (p.273) from al-BaIaguri's Ansiib al-asriif (vol. 5, p. 15,20): ... inni ra'aitu ka'anna dikan naqarani walii ariihu ilIii bufjuru agali " ... In a vision I saw a cock pick me (i.e., my eye); and hardly had I seen him (lit.: I do not see him [yetF -there, my end had arrived", or, in a more precise analysis: " ... and I do not see him [yet]-[thereupon nothing happened] except that my end occurred". The peculiarity of the expression bufjuru aga/i has not been taken into consideration in this interpretation. Quite literally, we would have to interpret: "... [thereupon nothing] except the occurring of my end". This striking nominal expression, instead of the complete (subordinate) clause, occurs here in a case whose peculiar characteristic is the extraordinary speed in the sequence of two actions. We can find the same nominal mode of expression, also in cases whose specific characteristic is to express the moment of surprise which, as we have seen (p. 272), implies in turn the moment of the immediate sequence of two actions one after the other. By way of example we cite the following passage CUmar b. AbI Rabfah, no. 176,5): lam yaru'hunna 'inda giika waqad gi'tu limi'iidihinna ilia dubuli. Freely interpreted this means: "Suddenly-[while they] were doing that and I had come to the rendez-vous with them-they were surprised by my entering". In a more literal analysis we can reproduce this quotation as follows: "Nothing had surprised them [yet]-[while they] were doing that and I had come to the rendezvous with them-(then nothing occurred] except my entering". The 2 Whereas the verb 'to see' (ra'ii) as used in our construction can rarely be found in the cIassicallanguage, it, or its equivalents, respectively, is very widespread in modern dialects. In the latter, it is regularly being used to characterize suddenly and unexpectedly occurring events. The (negated) verb used in this connection does never have an object; it is always being followed by iIIii plus a sentence whose noun (i.e., subject) does not have any direct relationship to the introductory verb 'to see', such as, e.g., in the following examples from the Judeo-Arabic dialect of Baghdad: yal:mii bhal/Jaki mii safnii allii lqahwa kalla $iigat $anta, tfagg-alebgi tasma' /Jassa "while we were engaged in this matter, there, suddenly, the Cafe became quiet as a mouse (literally: we had not yet seen anything [when nothing happened] except (i.e., before) the Cafe became still as a mouse); one could hear a pin drop". Further: $iig arjt;/ahag, mii-siifat allii gii hiirjiika a~~abi "it turned noon, when suddenly there came that boy" (literally: "she had not [yet] seen [anything, when nothing happened] except (i.e., before) that boy came".

280

INSTANTANEOUSNESS IN ARABIC

use of the verbal noun instead of the finite verb is also in this instance to be explained from the characteristic moment of this syntactic type, that is: the representation of the extraordinarily rapid succession of a (main) action after a (secondary) action that had preceded it. The verbal noun - instead of the finite verb -expressed by the second action divests this action of any verbal and temporal character (in other words, an event requiring a certain time). What appears suddenly and surprisingly, is, especially in the cited passages, above all else the concrete entity that occurs, i.e., agali 'the end of my life', or-in the case of the second quotation using dubuli-the person (that is to say: '1, the poet who is speaking') who is entering. The "occurring" (of the end of the life) and the "entering" of the person, respectively, are only secondary circumstances of lesser importance. The emphasis lies here especially on the genitives dependent on the verbal nouns (this holds true in particular for the first example in which, in other words, the main emphasis is on agali, whereas the preceding verbal noun ~utjur does not express any particularly essential concept). We find also elsewhere in cases in which mii rii'a ilIii ... , etc., is followed by a word group implying a sentence that, in most instances, a noun follows directly after ilIii. Cf., e.g., 'Umar b. Abi Rabfah, no. 6, 11: falam yaru'hii waqad natjat magiisidahii ilIii sawiidun warii'a I-baiti yastatiru "Then suddenly-she had just taken off her bodiceshe noticed a person hiding behind the tent".3 Or, more literally, "Nothing has frightened her [yet] (i.e., scarcely had something frightened her)-[thereupon nothing] except a person hiding behind a tent". Or, Ibid., no. 10, 11: falam yaru'hunna ilIii l-'isu tiilratan ya~milna bin-nali rukkiiban wa' akwiirii. Literally interpreted this means: "And nothing (i.e.: scarcely) had [something] frightened them-[thereupon nothing] except the gray camels coming up on the mountain-side, while they were carrying riders and saddles". Spitaler (l.e., p. 178) declares-with references to Reckendorf's conception (Synt. Verhiiltn., p. 520; Arab. Synt., p. 387) of the analogous word order in the construction with irjii l-mufiiga' ah followed by bi, a construction 3 Concerning warci'a I-baiti yastatiru in this context we may compare 'Umar b. Abi Rabi'ah, no. 38, 15: qalat lahunna abu mugaharatin qad ga'ana yamsi wama statara "She spoke to them: [He is] one who comes openly, there he is already going, not hiding himself", and also ibid., line 8: /au kana ya'tina mugaharatan fiman laraina ;gan /aqad suhira "If he would come to us openly among those whom you see, he would already be visible".

INSTANTANEOUSNESS IN ARABIC

281

which, however, we feel, is comparable with that now under discussion the word-order we are dealing with here of a noun (subject) followed by a finite verb (as in the instance 'Vmar, no. 6, 11) or by an adjectival predicate (participle) in the accusative (as in the instance 'Vmar, no. 10, II) is to be identified with the word-order of the construction (discussed by Reckendorf, Synt. Verhiiltn., § 169) of accus. cum verbo fin ito (ra' aitu Zaidan yalJrugu, or ra' aitu Zaidan biirigan). We cannot approve this identification. Rather, we note, consistent with our statements above, that what is first perceived by the eye at the suddenly and surprisingly appearing phenomenon is the phenomenon's agent or carrier represented by a noun which assumes a particularly close relationship with the term expressing the suddenness of the appearance, that is illii. If the construction we deal with here would be identical with the one of the accus. cum verbo finito customarily following the verbs of intellectual and sensual perception, we would have to expect that the noun following immediately after illii would be in the accusative (instead of being in the nominative or introduced by bi). Now as to the accusative of the adjectival predicate following the noun (in the nominative or introduced by bi), we refer to a construction of Arabic which we illustrate by the following examples. Qur. 11,5: h&jii ba'/i saiban 'My husband (here) is an old man'. Qur. 12, 65: h&jihi bicjii' atunii ruddat ilainii 'Voici que nos marchandises nous ont ete rendues' (Blachere's translation, p.475). Constructions in which there occurs a brief chronological interval between two actions or events happening in sequence one after the other point also in modern Western languages to certain peculiarities that recall the above discussed construction encountered in the Arabic language. The Arabic constructions remind us for instance of certain constructions in English and French. As far as the English language is concerned, we quote some pertinent examples from standard textbooks (Curme, English Syntax; Franz, Shakespeare-Grammatik, ed. 1924; Oxford English Dictionary), as well as from Henry Fielding's The History of Tom Jones; A Foundling (edited by W. Somerset Maugham, Philadelphia & Toronto 1948). In some of these instances, the part of the construction which expresses the second, principal event following immediately the first and less important one is introduced by before, till, and even when (cf. in Latin "cum inversum"). We quote from Fielding, p. 265: Jones had but little time given him to reflect on this letter, before a second was brought him from the same hand. Ibid., p. 341: ... ; but she had scarce patience to let the young ~that

282

INSTANTANEOUSNESS IN ARABIC

gentleman finish his speech before she interrupted him, saying: According to Curme, l.c., p.274, in older English we can also find till (instead of before or when): I had not been many hours on board till I was surprised with the firing of muskets. For the very frequent when we quote (from Fielding, p. 323): Mrs. Miller had not long left the room when Mr. Western returned. Furthermore we mention the following examples from Shakespeare (Love's labor lost, quoted according to Franz, l.c., p. 451), where the clause complementing a clause involving the use of scarce (= scarcely) is introduced by but: I scarce had paid the chairmen, and was coming up after her, but (= when) I met her on the stairs. There is no doubt that, in all the instances cited, the part of the constructions started by before, till, but, or when represents a new independent sentence. For instance, in the case of the last-mentioned example, the clause using but cannot be conceived as a complement (i.e., an adverbial phrase) of the clause with scarce that precedes it, as Franz (lac. cit.) does however seem to conceive it, since he interprets but in such cases by 'ohne dass'. We refer in this connection to our analysis of similar cases in the Arabic language, using ilia (see above, p. 275). In another mode of expression for the immediate sequence of two actions the idea is implied that the second (the principal) action occurred prior to the first one that precedes it in fact: no sooner ... than ., .. Analogies to this construction occurring in certain Western languages (see below) cim be found also in Arabic. In Arabic, we note a construction introduced by falam yakun (or: fama kana) '" . We cite the following examples: A. Siill;1iini's Rannat al-maJalil walmalani min riwayat al-Agani, vol. 1, (Beirut 1923), p. 250: ... faqala liba'rji l-badami l-wuqufi: man yagi'u bihii? Falam yakun bi'asra'a min an baragat waft yadiha 'uduha ... " ... and he (scil.: the caliph al-Wiiliq) said to some of his servants who stood [before him]: 'Who will bring her (i.e., the singing-girl, al-Faridah)?,. And [it] was no faster (i.e., his order was not uttered faster) than she came out (or: emerged) with her lute in her hand". al-Gal;1i~, Kit. al-Bubala', 266, 8fT.: ... wadabala $a~ibu s-sur!ati fa'aqbalat-i l-Gamilatu 'alaihi tu¢ii~ikuhu watula!ifuhU fama kana bi' asra' a min an qalat-i l-gariyatu l-qii¢i bil-biibi ... "... and the police-chief entered. Then al-Gamilah approached him, laughing with him and joking with him. And [it, or: that] was no faster than the girl-slave said (i.e., immediately thereafter the girl-slave said): 'The judge is at the gate' ... ". We find also a variant construction with walam yakun bi'asra'a an,

INSTANTANEOUSNESS IN ARABIC

283

in which, in other words, min ('than') is missing. We mention in this connection the following example from a late text (Ibn ad-Dawadari, Kanz ad-durar wa-giimf al-gurar, vol. 9, ed. Romer, Cairo 1959, p.61, 1 fT.): qiila: tu'linu ~wiitakum bit-takbiri watardumu hii4ihi l-bi'ra fi hii4ihi s-sii'ati! ... Falam yakun bi'asra'a an !ammu tilka l-bi'ra 'alii t}iilika r-raguli "He said: 'Raise your voices, calling out Alliih akbar, and shovel this cistern close, this moment!' ... and it had no sooner [been said than] they shovelled this cistern close over that man". The same text contains also other examples of the same type. However, this special construction without the use of min (= 'than'), which seems to us more peculiar than the one including min, can be proven by us to exist already in classical literature. Note the following passage (Tabari, I, 3385, 2fT.), in which in lieu of asra' there occurs its synonym ausak: ... falii tufiiriqunii (ya 'A izaru) /Jattii nadhaba bika itii amiri l-mu'minina fanubbirahu babaraka falam yakun bi' ausaka an gii' a 'A liyun fa' abbarahu babarahu " ... 'and you will not leave us [0 'Aizar], until we go with you to the Commander of the Faithful and inform him of your story'. And immediately thereafter (or: in this same moment; literally: and no sooner) did 'Ali arrive and they reported his story to him". Furthermore cf. 'Abdallah b. 'Abdalbakam, Sirat 'Umar b. 'Abdal'aziz (Cairo 1927), p. 164,8: wamii liyiibuka yii amira l-mu'minina, qiila: qami$un waridii'un wa'iziirun. Famii kiina bi'ausaka an gii'a 'Amru bnu Muhiigirin .... From Dozy, Supplement, II, 809 a, it follows that this construction using ausak (in lieu of asra') is used also by late authors (e.g., by Ibn ijaldiin). In these later authors we find by side with the construction famii kiina ausaka an the constructions famii kiina ausaka min an and fama kiina ausaka /Jattii. This Arabic construction may be compared with French constructions such as lin 'eut pas plus tot jete ce cri de revolte qu'it eut vers son pere un retour passionne "No sooner had he uttered this cry of revolt than he was seized by a return to his father" (or: "Hardly had he uttered this cry of revolt than ... "). See E. Gamillscheg, Histor. franz. Syntax, Tiibingen 1958, p. 679; and compare also Joachim Blass, Der Ausdruck der zeitlichen Unmittelbarkeit, mit besonderer Berucksichtigung des Franzosischen, Bern 1960 (Romanica Helvetica, vol. 68), p. 6. The corresponding constructions in English (No sooner ... than) are of course also to be compared (see below). However, whereas in the English and French constructions the verb corresponding to the comparative (sooner or plus tot = Arabic asra' or ausak) is variable

284

INSTANTANEOUSNESS IN ARABIC

and is usually a verb of concrete meaning, there appears in Arabic merely the verb 'to be', i.e., used impersonally as it were, its implied subject being represented by the content of the preceding sentence (see the examples mentioned above). It follows therefrom that falam yakun (or: fama kana) ... [min] acts as an adverb meaning "at the same moment" or "immediately", or "right away", etc. The phrase has the same function as the phrase fama kana illa an (or fama huwa illa an), as, e.g., (al-Gal:1i~, al-Bubala', 318, 15): lumma !fa~atfi d-dari: aljrigna ilaina saraban min ma'in wagairi ma'in fama kana illa an aqbala lalaluna w~ifatan bi'aidihinna !-!asatu wal-gamatu wal-aqda~u (a variant reads: fama kana gaira qalilin ~atta aqbalat ... ) "then she called in the house: 'Bring drinks out to us!', and at the same moment, thirty girl-servants appeared ... ". In English, apart from the construction (used exclusively in the modern language) in which the sequence of two clauses (introductory clause expressing secondary action plus subsequent clause expressing main action) is characterized by no sooner ... than ... , we frequently find in the older language a construction characterized by no sooner ... but ... (with but taking the place of than). We cite the following examples: He was therefore no sooner discharged out of the custody of physic, but he thought ... of fulfilling his engagement (Fielding, p. 104). There has no sooner anyone done me good service, but ... he cancels his interest in me by deep injury (Oxford English dictionary, vol. 10, col. 427 c, s.v. soon, 13 b); etc. Only the construction with no sooner ... than ... (involving than instead of but), exclusively used in the modern language, seems to be logical. The older English language has another, equivalent construction which, in our opinion, poses a similar problem. It involves the construction mentioned in the Oxford dictionary, vol. 10, col. 426 c, s.v. soon, 5 b, characterized by Not so soon ... (that) or but (that) .... For instance: The husbandman had not so soone throwne seed in his ground, but steppeth up the enimy (with but at the head of the second sentence). Furthermore: We were not so soone on land, the knight ... did instantly request me [etc.] (in this case, the second clause does not have any introductory particle). In another rather early example (from the year 1390) cited ibid., the second clause is introduced by that. Only one construction, namely one containing the particle as supplementing the particle so could be considered 'logical' (e.g., The husbandman had not so soon thrown seed in his ground as steppeth up the enemy). It would seem, however, that such a construction cannot be proven to exist.

INSTANTANEOUSNESS IN ARABIC

285

We must assume, both, that in the types of construction in which no sooner ... is followed by the particle but or when, the construction no sooner ... than is implied, and that in the construction with not so soon ... followed by but or even without a subsequent particle (see the above quoted examples), a construction not so soon ... as is implied. The problem consisting in the fact that the particle than or as, which we would by all means expect as a supplement of the phrases no sooner and not so soon, is missing in these constructions is to be considered in the light of the circumstance that the particle than in the type of construction no sooner ... than ... does not, as would be logical, belong to the same sentence, introduced by no sooner, which describes the secondary action, but to the new clause following it (which represents the principal action): no sooner had he ... than ... = scarcely had he ... , when ... (see the examples cited above). The nature of (independent) sentences introduced by but (and related particles) that express an action following with extreme speed a preceding action has been explained by us (see above p. 282) in connection with some kindred types. Accordingly, we analyze the constructions at hand on the basis of the examples quoted above as follows: The husbandman had not so soon thrown seed in his ground, [then-or: when-nothing occurred] but steppeth up the enemy. Also: There has no sooner anyone done me good service, [then ---{)r: when-nothing occurs] but he cancels his interest in me. Or: There has no sooner anyone done me good service, [then-or: whennothing occurs sooner] than he cancels his interest in me. N ow as to the variations of the constructions beginning with no sooner in which the second sentence is not introduced by but or when but by the comparative particle than, one may-upon superficial inspection-be inclined to consider the part of the construction introduced by than to be dependent on the overall-construction introduced by no sooner. Such a view is not tenable. Than introduces an independent sentence, in other words, it cannot be referred to a sentence-part of the preceding independent sentence. The independence of than from the preceding sentence is also evidenced in that in numerous variants of the construction (which are particularly frequent in the older language) such a than, as it seems to be required by the preceding no sooner, is missing altogether, or, rather, is replaced by but or when (see above). Similarly, in the older language, in equivalent constructions involving the use of not so soon (see above), we fail to find the as required by the particle so. The English constructions

286

INSTANTANEOUSNESS IN ARABIC

involving no sooner ... than (cf. also the constructions involving not so soon ... as) as well as the corresponding constructions in French consist-just as the construction with no sooner ... but (or when), and the equivalent constructions involving vix, hardly, kaum, d peine ... que (see the examples quoted above p. 272 and 282}-of two independent sentences. Among the diverse variations of the constructions discussed here, we must distinguish between those in which the first sentence contains a negation and those in which it does not. To illustrate the latter category, we select the following examples. English: I had hardly (or: just) arrived at home, when my friend came in. French: A peine etais-je arrive chez moi, (que) mon ami entra. German: Ich war kaum (or: gerade) zu Hause angekommen, da trat mein Freund ein. Whereas in this instance the first clause-in agreement with reality-notes that its action was completed prior to the occurrence of the action of the second clause, in many variant constructions the first clause expresses-in exaggerated contradiction to reality (see above)-that its action at the time of the occurrence of the action of the second clause had not yet occurred or been concluded. However, all constructions share the fact that the first clause contains an expression pointing to the fact that the moment of the occurrence of its action is governed by another event: hardly, just, not yet, earlier (than). A clause characterized by such particles necessarily refers to another event and requires, under the special circumstances we are dealing with, an adverbial complement. The existence of such a complement is suggested also by the use of the pluperfect-a relative tense-in the modern Western variants of these constructions. In other words, we must assume that the first (non-subordinated) sentence of the overallconstruction is followed by an adverbial complement existing mentally only, which, as far as its content is concerned-and not with regard to its syntactical function-is identical with the content of the second, independent sentence of the overall-construction: I had hardly -or: just-arrived at home, when he entered (or: I had just-or: scarcely-arrived at home at the moment of his entry) = I had not yet arrived when he entered (or: I had not yet arrived at the moment of his entry); No sooner had I arrived than he entered (or: I had not yet arrived before his entry). The adverbial specification forms in this instance a component-not a particularly underscored one-of the overall-statement occurring in these sentences-a component which -on account of the special circumstances we are dealing with here-

INSTANTANEOUSNESS IN ARABIC

287

did not find any linguistic expression. This first sentence (containing a linguistically suppressed adverbial complement) of the overallconstruction, is followed by a second independent sentence which, in its contents, is identical with the contents of the linguistically suppressed adverbial complement of the first sentence. The linguistically suppressed adverbial complement of the first sentence reappears in the second (non-subordinate) sentence of the overall-construction as an independent statement-and not as an adverbial complement of another sentence. In the mind of the speaker, already at the moment at which he starts to utter the sequence of the two sentences closely related to one another, all particulars of the two sentences (that means: also those of the second one) are present. In view of the fact that this involves the representation of the action occurring in the second sentence as happening immediately after (or following with extreme speed) the one described in the first sentence, he cannot help using the second action in order to determine the point in time of the occurrence of the first one (see our model sentences cited above p. 271-284). This inclusion of the contents of the second sentence in the first one influences only the form of the first sentence (note especially the types in which the first sentence is characterized by the expressions no sooner [... than ... ] and not so soon [... as ... ] suggesting adverbial complements). The content of the second sentence (although it may be supposed to be mentally present therein) does not appear in its linguistic expression. This content, on which the major' emphasis is placed, and which represents something completely new for the person being spoken to, follows the first sentence as an independent mental operation. The action reported in this new independent sentence is represented clearly and unambiguously as. the action following immediately the one mentioned in the first sentence; and this refers to all types of the construction, also to those in which -contrary ro reality-the event noted in the first sentence is represented as happening after the event stated in the second sentence (or in any event not preceding it, see above). In these latter types we have, in other words, a mixture of realistic and non-realistic representation (actually: a representation exaggerating reality). We noted above p. 275 that the Arabic type lam tu'ginhum bihim illti wa ... (see above p.273), and related types of Arabic and other languages, just as well as the type No sooner had I arrived at home, than my friend came in, consist in each case of two independent

288

INSTANTANEOUSNESS IN ARABIC

sentences. We concluded that, in spite of an impression to the contrary, the negation contained in the first sentence (lam, or ma, etc.) does not exert any effect on the particle illa ('but') in the second sentence; and likewise, in the second type, the comparative (no sooner) in the first sentence does not exercise any effect on the particle than and its dependence in the second sentence. A similar situation, i.e., an illusion that a term (e.g., a verb) in a preceding sentence seems to govern (in other words, seems to exert 'rectio' on) a subsequent sentence (in reality a new sentence), may result under other, widely varying conditions. We mention here a syntactical type in which this very situation occurs under quite different circumstances. We quote from Diwan Labid, no. I,ll (ed. t!alidi, p. 3, v. 3):falam taula dabla I-ma'i batta tamahharat wisabun laha min 'arma4in wabarimu. Brockelmann, in his translation of the line, had constructed batta and its dependence as governed by the introductory verb falam taula: "sie (i.e., the wild donkey and his wives) begniigten sich nicht eher, als bis sie im seichten Wasser umher sich tummelten, mit einem festgedrehten Giirtel aus 'Arma mulqii) zifru 'ayhalatin ... ], and with ['inda] l-fityati s-su'li qad haffat baqii'ibuhum ... corresponding to 'inda[ma] qad haffat baqa'ibu l-fityati s-su'li ... ). (18) wayawma Quraysin irj atawna bigam'ihim wafi'na 1- 'aduwa wa(ata I-mutalaqili 18 "On the day when the [Banii] Qurays came to us with their army, we trampled on the enemy with an extremely heavy pressure" (with wayawma Quraysin id atawnii bigam'ihim corresponding to wayawma id atawna Quraysun bigam'ihim ... ). der sein Gut frei weiss auszugeben". But also Reckendorf, Arab. Syntax, §60,1 (p. 115) has not recognized the specific syntactic function of the preposition ma'a in this line. He quotes the sentence in the following form: waylummi latjtjiiti s-.~ababi ma'isatan ma'a l-kuJri (thus ignoring the syntactic relation between ma'a l-kUJri and the subsequent words: yU'!iihu I-fatii l-mutlifu n-nadi), and translates this (incomplete) form of the sentence by: "Wie herrlich sind die Vergniigungen der Jugend, ein Leben in Hiille und Fiille". 16 Ibn Hisam, Sirat Rasuli-lliih (ed. Wiistenfeld), p. 27,9. 17 al-I:Iutay'ah. Diwiin (ed. Goldziher), no. 18, 3-4 (ZDMG 46, 491). The noun mulqii (in line (a) [v. 3]) serves here as nomen actionis (infinitive)-as against the interpretation of the scholion (yaqulu: ni'ma mmwji'u mulqii ri/:liili t/-t;layfi). With this nomen actionis, the subsequent term (in line (b) [v. 4]): a/~rityati s-su'Ji qad bajfat /:laqii'ibuhum is coordinated. Both terms (i.e., mulqii ... as well as al-jityati qad baffat /:laqii'ibuhum) are equally dependent on 'inda, which fulfils here the syntactic function of a conjunction. 18 I:Iassan b. Tabit, Diwiin, ed. Cairo 1929, p. 316, line I. As an analogous instance we should consider also the phrase (Ibid., line 13 [and Ibn Hisam, Sirat Rasuli-/liih, ed. Wiistenfeld, p. 175,13-14)): yawma baJmin it:! atal1'ka .. , "on the day the enemies came to you ... " (= yal1'ma it:! atal1'ka baJmun ... ).

SOME SPECIFIC FORMS OF HYPOTAXIS

301

(19) ni'ma l-fawarisu yawma gaysi Mu/:tarriqin la/:tiqu wahum yad"Una yala l)irari 19 "What excellent horsemen [were they] on the day [when]

they caught up with Mu1:tarriq's army, while they were shouting (the battle-cry): 'Oh, house of l)irar'" (with yawma gaysi Mu/:tarriqin la/:tiqu corresponding to yawma la/:tiqu gaysa Mu/:tarriqin). (20) hal tagkuruna layaliya t- Turki naqtuluhum 20 "Do you remember the days when we were killin-g the Turks?" (the text corresponds to hal tagkuruna l-layaliya [/:tina] naqtulu t-Turka). (21) (a) arani iga ma bittu bittu "ala tawan wa'inni iga a$ba/:ttu a$ba/:ttu giidiya (b) ila /:tufratin uhda ilayha muqimatin ya/:tallu ilayha sa'iqun min wara'iya 21 "(a) I see myself (i.e., the natural course of my life [thus]): During the night I am hungry; and in the morning (at dawn) I set out (for the purpose of pasturing, hunting, raiding, etc.)-(b) until I shall be carried (some day) to a pit which makes [me] 'stay' (i.e., makes me a permanent sedentary), [a pit] to which [as a goal] a driver behind me (that is: Fate) urges [me] on" (with ... ila /:tufratin uhda ilayha muqimatin equivalent with ... ila an uhdii ila /:tufratin muqimatin). (22) wala'in agaqnahu na"ma'a ba"da ubayrah the Quraysite died?" (with ba'da pubayrata I-Qurasiyi mata corresponding to ba'da an mata pubayratu I-Qurasiyu). (30) (a) inni la'a'gabu min qawlin gurirta bihi bulwin yamuddu ilayhi s-sam'u wal-ba:jaru (b) kal-bamri was-suhdi yagri fawqa ?ahirihi wama libii/inihi /a'mun wala babaru (c) wakas-sarabi sabihan bil-gadiri wa'in tabgi s-saraba fala 'aynun wala alaru 30 "(a) I am amazed that you have been allured by a word to which the ears and the eyp.s are attracted, (b) over whose outside as it were wine and honey are flowing, while in it there is no taste whatsoever-(c) [thus] being like the mirage that resembles a pool left by the rain-but if you try 'an "from" by a (verbal) noun: innii mana'niika 'an ( < 'an )-i s-sugudi. Once 'an> 'an had (in its use after verbs of "preventing, hindering, protecting (etc.)") acquired the meaning of "separation" ("from", "away from", "ofT'), it could, of course, be trans-

334

SYRIAC DALMA AND RELATED ARABIC PHENOMENA

ferred into other contexts and be used in all instances in which the concept of separation, that is, the concept "from" or "away from", was required. The Arabic language had thus acquired a new preposition, expressing the concept of separation, which was competing with, and superseding to a considerable extent, the ancient Semitic preposition min. Our derivation of the Arabic preposition expressing "separation", 'an, from the conjunction 'an (originally "that", German "dass") in its use in the meaning "lest" is strongly supported by the fact that in Mehri (see Bittner, Studien zur Laut- und Formenlehre der MehriSprache, IV, § 50, p. 36), the preposition men ("from"; in Arabic, etc., min) is used as a conjunction in the meaning of "lest" in (negated) purpose clauses (that is, in the meaning of Arabic Ii' alia), as well as after verbs of fearing or taking heed, and also of being unwilling and being unable, etc. (see, for examples, Bittner, lac. cit.). This development of men "from" into a particle of the meaning "lest", in all the various uses in which this concept may appear, is nothing but a reversal of the process by which the Arabic conjunction 'an, in its use in the meaning "lest", acquired the meaning "from", as represented by the (secondary) phonetic form 'an. The 'ayn instead of the original alif (of 'an), by which 'an in the meaning "lest" (> "from") is characterized, is an analogue to the 'ayn in la'alla (originally: "lest") from ltalla (see above, p.329). Moreover, we may also refer to the temporal particle of Soqotri, 'am (with 'ayn), which is based on an original yawma "when", and must be compared with the temporal particle of Ethiopic, 'amma (with alif), which in turn is derived from an original yawma-ma. This comparison of 'an > 'an with 'am > 'am is all the more reasonable inasmuch as the particle' an "that" (German "dass"), which is the basis of 'an "lest" (> "from"), is itself derived from an original yawma "when": yawma > yam> 'am> 'an. Concerning this origin of the particle' an, we refer to our remarks, below p. 391 ff. In the passage from Brockelmann, Grundriss, II, §456, quoted above, p. 322, two sentences with the interrogative expression lema (or Hebrew lama) are mentioned. Both sentences constitute rhetorical questions, equivalent to negative statements (see p. 322). Brockelmann quotes in the same context, as further examples of this type of a rhetorical question expressing doubt ("eine rhetorische Zweifelsfrage"), the following two instances in which, instead of the unquestionably interrogative expression lema, the form dalma appears: '(/Ij dalma la

SYRIAC DALMA AND RELATED ARABIC PHENOMENA

335

"oder etwa nicht?" ... [and] hmj dan$aQ 'edna dalma la sama' hmj daQra 'ajnf dalma la mestakkal "der das Ohr geschaffen, sollte nicht hOren, der die Augen gebildet, nicht sehn?" (Ps. XCIV. 9). Although these two sentences imply an interrogation, they cannot (contrary to Brockelmann's opinion) be considered to represent the primary form of expression from which dalma ("lest", "perhaps") is derived. The original interrogation (serving as a rhetorical question) was expressed by lema, not by dalma. The word dalma used in the two quotations is undoubtedly identical with the secondarily developed particle dalma described above. The word appears in these two quotations, as it frequently does, in the sense of "maybe, perhaps" in the context of interrogative sentences. As to the interrogative character of these sentences, it is to be considered indicated by the intonation. Ps. XCIV. 9, for instance, in the Syriac version, may be interpreted literally as follows: "the one who has formed the ear, does [he] perhaps not hear, the one who has created the eyes, does [he] perhaps not see?". This represents a paraphrasis of the Hebrew original, in which the interrogation is indicated by the interrogative particle h a • It cannot be doubted that in sentences of this type dalma serves as a particle emphasizing the interrogative character of the sentence. We may hereto refer again to Arabic la'alla as a parallel (see above, p. 328), which also, on the basis of its meaning "perhaps", seems to have acquired the special function of emphasizing the interrogative character of a sentence. This function of la'alla appears in a particular construction, in which the interrogation is dependent op a sentence of the meaning "I do not know [whether] ... ", that is: "I would like to know whether ... ". We have in mind the frequent use of la'alla in sentences introduced by la adri, of which Reckendorf, Synt. Verh., p. 511, quotes various instances, e.g., Tabari (Annales, I, 1845, 17): la adri la'allakum satukallifuni ma ... "Ich weiss nicht, vielleicht ladet ihr mir etwas auf, was ... ". An interpretation in which la'alla is conceived as implying the meaning "whether" ("if") seems to be more adequate: "I do not know whether you will impose on me something that ... ". In the present context we should also mention Syriac dam < dema (see Brockelmann, Grundriss, II, § 456), which is used synonymously with dalmii (though it is found far less frequently than the latter). We cannot accept the opinion of L. Costaz, Grammaire syriaque (2nd edn., Beirut 1964), p. 143 and p. 170, who considers dam as abbreviated from dalmii. We should rather assume that in the original

336

SYRIAC DALMA AND RELATED ARABIC PHENOMENA

interrogative sentence on which demii > dam is based, the interrogative pronoun mii, which primarily means "what?", was used in the sense of "wherefore, why?", as it frequently occurs-in Semitic and nonSemitic languages-with words of the meaning "what?" (cf. Greek .1). There is a further conjunction which we think should be interpreted in analogy to the conjunctions discussed above (p. 327 and p.328). This is the Aramaic conjunction introducing "contrary-tofact conditions", that is: Egypt.-Aram. hn Iw, Targum. ilia, Syriac ella, etc. (see Brockelmann, Grundriss, II, § 429a). Also Arabic, though ordinarily using law, uses sometimes in-law (see Brockelmann, loc. cit., §426a, Anm. p. 643); we find also the form ilIaw (see, for example, Bayhaqi, Kitiib al-Ma~iisin wa-I-masiiwi, ed. Schwally, p. 88,11). According to Brocke1mann (Ibid., p.643, with reference to the rarely occurring Arabic in-law), we deal here with a secondary addition of in to the particle law (Aramaic la) for distinguishing the "contraryto-fact conditions" from sentences in which law (la) serves (in independent, non-subordinate sentences) in the meaning of "0 that!, Would that!". We cannot accept the view that en (Arabic in) has been added to la (Arabic law) with the purpose of distinguishing this use of lU (law) as a conjunction from its use in independent sentences implying a wish. For us, in the "contrary-to-fact conditions" under discussion, which are introduced by ellu (Arabic in-law), it was originally the conditional conjunction en (in) alone which appeared in conjunctional function, characterizing the clause dependent on it as a conjunctional (conditional) clause or, more exactly, as the protasis of a "contraryto-fact condition". As to the particle lU (Arabic law) following it, it was originally in no close connexion with the conjunction en (in) preceding it, and also did not express any conjunctional function by itself, but served in the other meaning which Semitic la (Arabic law) expresses, that is: in the meaning of "0 that!, Would that!", thus characterizing non-subordinate, independent sentences expressing a wish. This original structure in which ella (in law) in our opinion developed may be illustrated by the following example (taken from N6ldeke's Syr. Gramm. 2 , p.300): ella Iflsul-zel]an lewiileh iNh-wiil Iii siil]qii-wiil leh "if all the time she had been with him, she would not have left him". We assume the following syntactic structure as the original structure of this sentence: "If [... ].-Would that she had been with him all the time!-She would not have left him". That is: We assume that the conditional protasis, characterized by en "if",

SYRIAC DALMA AND RELATED ARABIC PHENOMENA

337

was cut short with this word, while its actual content (though present in the mind of the speaker) did not become linguistically expressed. Before the actual content of this protasis became expressed, the speaker started with a new, independent (non-subordinate) sentence, introduced by the particle lu (= Arabic law) "Would that!", in which the content of the linguistically uncompleted protasis starting with en did find its full linguistic expression. The protasis of a "contrary-tofact condition" frequently implies a wish, and this wish can, of course, also be expressed in a direct form, by an independent sentence, starting with an expression of the meaning "0 that!, would that!", a meaning for which the ordinary Semitic expression is lu (Arabic law). (Note that in Akkadian lu is mainly used in independent sentences expressing a wish or command). The syntactic structures with ellu (Arabic in-law) may therefore be compared with the sentence-types discussed above, on p. 323 and p. 325, in which it is an object clause which remains linguistically uncompleted and is followed by an independent sentence in which the idea to be communicated finds its full expression. In the case of ellu, in analogy to these latter types, it is the passionate interest of the speaker in the fulfilment of his desire which causes him to pass from the form of the subordinate clause (a protasis) to the form of an independent statement, in which his desire finds a clear and emphatic expression. As in the other types discussed by us, and also in the present type with ellu (Arabic in-law), the fusion of the uncompleted protasis (indicated by en) with the directly following independent sentence (starting with Ill) was, of course, unavoidable, so that finally a new conjunction, characterizing "contrary-to-fact conditions", developed, namely ellu (= Arabic in-law> illaw).

22 ARABIC LAKIN(NA) AND RELATED EXPRESSIONS Brockelmann, Crundriss, II, p. 480 (§ 299 a), interprets the "adversative" particle of Arabic lakin (Iakinna) as originally expressing the meaning "nicht so" ("not thus"), and (Ibid., n. 1) rejects Reckendorfs derivation of lakin(na) from an original form la' akinna with the initial la representing the "emphasizing" la (lam at-ta' kid). In contrast I consider lakin based on an original phrase ilia kan( a) (i.e., ilia plus the auxiliary verb kana), expressing the meaning 'except', 'but', which, as is frequently the case with expressions indicating 'exception', acquired the sense of an "adversative" expression: 'but, however, yet' (cf., e.g., Aramaic ellii and also English but). This ProtoArabic ilia kan (a) (> lakin) should be identified with ella kan mentioned by Rhodokanakis (Der vulgiirarabische Dialekt im polar (~/ar), II, p.13I; §37b) as an equivalent, in the dialect of polar, of the simple ilia, as in the following sentence: ella kdn beyt eM Zeyd milifak "nur die vom Hause A. Z. k6nnten dir gleichen". I assume that in the basic Proto-Arabic form ilia kan (a), the initial vowel (or syllable) i ( I) was dropped and the long vowel a of kan ( a) was shortened to a, being ultimately replaced by i (cf., e.g., Arabic an 'that' (= Latin 'ut') > modern colloquial in and numerous analogous instances, cf. below p. 391). With respect to the disappearance of the initial syllable if in ilia kan (> lakin) we compare Biblical-Aramaic lahen "except, but, yet", which is generally considered as a compound consisting of la (i.e., the simple negation) and the conditional particle hen, i.e., considered as an expression in which the two components of the ordinary form of this expression, in-Ia> ilia (as appearing in numerous Aramaic dialects as well as in Arabic), are inversely arranged: la + (h) en (instead of hen + la) (cf., e.g., Koehler-Baumgartner, Lexicon in Veteris Testamenti libras, 1953, p. 1090). This interpretation of lahen "except, but" has been rejected by Brockelmann, Grundriss, II, § 299 d (p.482), who identified the particle of "exception" lahen with the Old-Aramaic "causal" particle Ihn ("therefore"). In contrast to both interpretations of lahen "except, but, yet", I consider laheri based on an expression ilia (h)en, that is: on a form in which the expression

ARABIC LAKIN(NA) AND RELATED EXPRESSIONS

339

of "exception", ilia (ella), was augmented by the conditional particle in (en, hen), thus constituting an expression the literal meaning of which was: "except if". I refer in this connection to the Syriac expression ella en ("except if"), frequently used in a function corning close to that of the simple ella. I also consider another particle of Classical Arabic as derived from the particle ilia "except", namely the particle known under the Arabic term lam at-ta'kid, i.e., "the particle la expressing 'emphasis', 'affirmation' ", which, among other uses, is very frequently used to emphasize a nominal predicate (and especially the predicate of sentences whose subject is preceded by inna), as, e.g., in kunta lamu'minan hiM "if you indeed believe in him" (Bubar!, Sabib, I, 234, II), inni la'indahum "I am with them" (Tabari, Annales, II, 279,1); etc. This emphasizing la (by Brockelmann, Grundriss, II, p. 110, interpreted as an interjection) should be traced back to a basic expression [la-or: ma-or: laysa- ... ] ilia. Accordingly, the afore-quoted sentences would have to be derived from basic forms like in [ma] kunta ilia mu'minan hiM "if you are [not] but believing in him"; etc. The negation (la, ma or laysa) may originally have been expressed, but could also, though present in the mind of the speaker, from the outset have been suppressed in the linguistic expression of the sentence (cf. the prari instances quoted below). In any case, the predicate of the sentence was originally preceded by the particle of "exception", ilia, which ultimately (still in quite early time) became "reduced" to la. This particle la, in agreement with the function of its basis [la, or: ma, or: laysa] ... ilia "[not (or: nothing)] but", served to emphasize the predicate of the sentence. In modern Arabic dialects we find instances of this type, in which ilia or la ("reduced" from ilia, but with a long vowel, a) serves to emphasize the predicate. I again refer to instances from the dialect of polar (quoted by Rhodokanakis, I.e.): 'Ali ella magalil "'Ali ist gewiss impotent", more literally: "'Ali is [not] but impotent"; furthermore (with ella abridged to la) TofU! la mill el-barim 'T ist nicht anders beschaffen als die (anderen) Weiber", more literally: "T. is [not] but like the [other] women"; Tofulla mill el-mhe 'T ist wie die Antilopen", more literally: 'T is [not] but like the antilopes". It is from such a particle la (with a long a-vowel), developed from ilia, that the ancient-Arabic ta'kid-particle la (with a short a-vowel) is to be considered derived. It is interesting to observe that 'ar, ar, er < gayr (to a far-reaching extent a synonym of ilia) is in turn used to emphasize

340

ARABIC LAKIN(NA) AND RELATED EXPRESSIONS

the predicate of a sentence. Cf. the instances quoted by Rhodokanakis, Ibid., §31 d (p. 131 b), as, e.g., w-ene er ma'es, which (in the literary Arabic translation of the prari-text) is rendered by wa-ana lama'aki, that is: by a sentence containing kim at-ta' kid (as translation of er). For instances of a similar type, containing illii, see also Landberg, Etudes sur les dialectes de l'Arabie meridionale, vol. II: Dalina, p.662; Brockelmann, Grundriss, II, §453 bE (p. 651); Ph. Mar~ais, Le parler arabe de Djidjel/i (Nord constantinois, Algerie) (Paris 1957?), p. 448-

449. Regarding Classical-Arabic liikin(na) , with which I started in the present discussion and which I interpreted as derived from ilIii kiin (a), I also refer to a modern particle mentioned by W. Mar~ais, Textes arabes de Takrouna, vol. 2, Glossaire, p. 3585: "... le!;' r;nn, lqr;nn, lijjlnn, particule adversative, synonyme de lqkun [< liikin] (cf. supra, p. 3582); peut lui etre substituee dans to us ses emplois: 'mais', 'toutefois', 'cependant'; n'apparait que munie des affixes personnels .... Souvent precede de u 'et' ... ". I derive this expression, in complete analogy to my derivation of liikin, from an original expression illii an(na). With respect to the fact that the modern, colloquial expression is frequently preceded by u 'and', I note that also the classical liikin(na) is frequently preceded by wa 'and' (cf. below p. 342). I now refer to a particle recorded by Socin, Diwan aus Centralarabien, vol. III (Glossar), p. 308 b: "[lii-ka'an] liitin, liitinne 'es ist wahrlich als ob ... '.", with Socin's comment: "und oft mit dieser sekundaren Verlangerung des affirmativen la; auch latin kommt indessen vor" ("and often with this secondary lengthening of the 'affirmative' la [liim at-ta' kid]; but also latin occurs"). This liitin should be derived from an original illii ka' an (na), and the short a in the equally occurring latin is to be considered shortened from the long a in liitin (which latter is not, as Socin assumes, secondarily lengthened). The case represents therefore a parallel to Classical la (liim at-ta'kid) according to my derivation « Iii < ilIii). I also interpret the particles of "exception" recorded and discussed by W. Mar~ais, in Textes arabes de Takrouna, vol. 2, Glossaire, p. 3524-3526, that is: kqnsi (kqns), kqn-mq, and also kqn u (etc.)in analogy to my interpretation of liikin(na) as well as of the "affirmative" la (liim at-ta' kid)-as an abridgement of illii kiin. I assume that these modern, colloquial expressions indicating "exception" are based on original forms of expression in which the verb kiin was preceded by illii, which in these instances was not only abbreviated (mutilated), but disappeared entirely.

ARABIC LAKIN(NA) AND RELATED EXPRESSIONS

341

Above, p.338, I derived lakin(na)-most frequently used as an "adversative" expression: "but, however, yet" (introducing a new sentence)-from an original expression illa kan ( a) of the meaning "but", "except [that]" (= illa), introducing a clause (or a noun) which forms part of a (preceding) negated clause: ma ... illa kan (a) ... ("not ... except [that] ... "). Concerning this original function of lakin « illa kan), I refer, e.g., to ella kan = "but", "except" in (e.g.) the dialect of potar (see above p. 338). The "adversative" use of lakin « illa kan) is, in accordance herewith, to be considered as developed secondarily. This semantic development is analogous to that of English but, which is used in the sense of "except" and related concepts (original meaning), and also in the function of an "adversative" particle, more or less identical with "however" (secondary meaning). I stress here once more (as already stated above p. 340) that Classical Arabic lakin (na) - in full agreement with a related particle in modem dialects (see ibid.)-can be preceded by wa, i.e., wa-lakin(na). In my opinion, the use of lakin (or also-preceded by wa-walakin), in its (original) meaning and function: "... not ... but (= except [that]) ... " (= illa) survives in Classical Arabic texts. I refer in this connection to a verse by Imru'u'l-Qays, which in the standard edition of the poet's Diwan by Ahlwardt (Six divans), no. 20, 52, is presented in the following form: wama gunibat bayli walakin tadakkarat marabi!aha min Barba'iya wa-Maysara. In various editions of the Diwan, which were published fairly recently, the reading gunibat, adopted by Ahlwardt, is (in agreement with de Slane's original edition, see below) replaced by gabunat. I mention the edition of Imru'u'l-Qays' D.iwan by Mul.tammad Abu'I-Fa 'og) justifies my assuming its existence also for proto-Semitic (or for proto-West-Semitic). With respect to this, it seems important to note that medieval and modern Arabic dialects possess an element 'ad which expresses the meaning "yet". The (modern) Arabic element 'ad is used as an adverb, though its being rooted in an original verbal expression of the same meaning Cad, based on 'ada "to return", root 'wd) is still clearly recognizable. I refer to the instances quoted by N61deke, Reitrage zur semi!. Sprachwiss., p.60, n.2, and to Brockelmann, Grundriss, II, p. 512. Because of the clear fact that the modern Arabic adverb 'ad "yet" originates in a verb, N61deke (l.c.) has declined to identify it with the Hebrew particle 'Off. In contrast to N61deke, I consider it pertinent to identify the Hebrew element and the modern Arabic form of expression, and to draw the conclusion that Hebrew' Off has developed from an original verbal expression of the same type as existent in later Arabic dialects, that is: from an original verb 'ad(a) (root 'wd) "to return" ("he returns doing it", "he does it again", "he is still doing it"). We may assume that this verbal construction of the concept "still, yet" (frequently appearing with negation: "not yet") was current in protoSemitic. Accordingly, Hebrew 'og has to be regarded as a survival of a proto-Semitic form of expression, although the original verbal construction, still appearing in Arabic (see Brockelmann, l.c.), is not preserved in Hebrew. As against this, other Semitic languages (apart from Hebrew) lost even this adverb Cad) as an independent element (note, however, Ethiopic 'adi "again"). Nevertheless, even in them its existence is still reflected, namely in the root b'd "to be distant, remote", and especially in the preposition ba'da "after", which is clearly based on an original *bi/a- 'ad identical with Hebrew be' og. In connection with the existence, in proto-Semitic and in early developmental stages of certain individual Semitic languages, of an adverb 'ad "yet", I mention another Semitic particle, sc. the particle with the meaning "until", represented by Hebrew 'af! "until" (also used in a local meaning: "up to"), in poetical contexts (and with suffixes) also 'age; Aramaic 'af!; Old-South-Arabic 'd, and 'dy; Akkadian adi. Brockelmann, Lexicon Syriacum 2 , p. SlOb, connects this word with the root 'dw (Ethiopic 'adawa) "transgressus est", and this etymology is also suggested in K6hler-Baumgartner's Lexicon

346

THE ORIGIN OF ARABIC BA'DA "AFTER"

in Veteris Testamenti libros (1953), p. 680a. I consider 'ad "usque ad" as originally identical with 'iid> Hebrew 'Or} "yet". The interconnection of the concepts "yet" and "until" may be illustrated by constructions and phrases of various types. I only mention here the identity in use of Hebrew 'og me'a! we '" "still a little and [such-andsuch happening occurred]" with the frequently occurring Syriac phrase 'a4 qallil which is used in the same sense (see, e.g., Julianos der Abtriinnige, ed. G. Hoffmann, p. 90,2; cf. Brockelmann, Lex. Syr.2, p. SlOb, s.v. '(4).

24 AN ARABIC SENTENCE-TYPE EXPRESSING "INNER COMPULSION" 1 The Qur'anic verse Surat yu.suj, v. 10, reads as follows: qiila qii' Uun minhum: Iii taqtulU Yusufa wa' alqUhu fi gayiibat-i l-gubbi yaltaqi!hu ba'fju s-sayyiirati in kuntum fii'Uina. We are especially interested in clarifying the sense of the phrase in kuntum fii'i!ina in the frame of the verse as a whole. Concerning this phrase we quote az-Zamabsari's commentary in his Kassiif (vol. 1, p. 643): in kuntum 'alii an tajalu ma yal:z~ulu bihi gara¢ukum fahii(jii huwa r-ra'yu. We also quote al-Bay (Arab.) tabi'a(hu). That this form, *tabiiya'a (hu) , was once a living linguistic form, was assumed by me precisely on the basis of these Arabic and Ethiopic terms. In other words, this form, *tabiiya'ahu, was reconstructed by me on the basis of these terms. F. Rundgren, in his study "Bay'a 'Huldigungseid'" (in his article "Zur arab. Wortkunde", II, published in Orientalia Suecana, 21 [1972], p. 53 ff.), rejected my above-mentioned derivation of these Arabic and Ethiopic terms (for the details of my derivation see Background, p. 213 ff.). My intention here is to deal once more with my derivation of Arabic tabi'a-and also of the quoted Ethiopic terms-

376

ARABIC PARALLELS TO THE ENGLISH PHRASE

from an original *tabaya'a(hu) and to adduce further arguments for this derivation, in addition to those presented in my earlier study. In the reconstruction of the genetic development of Arabic tabi'a ("to follow") I proceed from the verb in the plural tabiiya'u (root by', in the so-called "6th conjugation"), "they made an agreement with each other" (also, e.g., especially in the dual: tabiiya'a they bought from, and sold to, one another), and also "they pledged allegiance to each other" (and analogously in other plural-forms of this conjugation: tabiiya'na, etc.). I assume that this basic form, i.e., the so-called "6th conjugation" of the root by' (which primarily is used only in the plural or dual), was in the present instance (*tabiiya'a-hU > tabi'a-hu) secondarily and irregularly used in the singular (more exactly: with respect to a subject in the singular, and thus appearing in the form of the singular). That means: the concept primarily expressed by baya'a-hu"he (i.e., some person) made an agreement (concluded a contract) with him (i.e., with another person)", and more specifically: "he pledged his allegiance to him" -could also be expressed by the form *tabiiya'a-hu (which in due course developed into tabi'a-hu). This irregular use of the "6th conjugation" in the singular involved another irregularity, that is: its being followed by a complement in the accusative (differently expressed: its governing an object in the accusative). Through these "irregularities", the characteristic features of the "6th conjugation": the prefix ta and the long vowel following the first radical of the root, lost their function in this specific instance. In consequence hereof, the form of the "6th conjugation" was transformed into a form of the "1st (basic) conjugation", a process which involved the incorporation of the prefix ta into the root itself. Accordingly, *tabiiya'a-hu developed into tabi'a-hu, which-exactly as the form underlying it-originally expressed the meaning "he (i.e., someone) pledged allegiance to him (i.e., to someone else)", and finally: "he followed him" (see below p. 387). With respect to this irregular use in the singular (*tabiiya'a-hu > tabi'a-hu) of a verb-form basically requiring a subject in the plural or dual, as, e.g., tabiiya'u (cf. above), I referred already in my earlier study to certain nouns which represent plural-forms used as singulars. As far as Arabic is concerned, I referred to the use, in the contemporaneous dialects of Syria, 'Iraq and other regions, of the noun qarayeb « qara'ib, pI. of qarib) in the function of a singular: "kinsman, relative", with the plural expressed by a new form: qaraybin (beside

ARABIC PARALLELS TO THE ENGLISH PHRASE

377

which the old, original plural-form qarayeb < qara'ib = 'aqriba' is probably also, but less frequently, used in the function of a plural). The original singular-form qarib seems no longer (or only rarely) to be used in the meaning of "kinsman, relative" (sing.) in these dialects. I mention here another modern-Arabic instance of this type, known to me from 'Iraqi dialects (e.g., from Bagdad). In these dialects, the (original) "broken" plural giran, which in literary Arabic serves as the plural of the noun gar, is the (more or less) regular form of the singular of the concept "neighbour", while the original singular, gar, is obviously rarely used in the dialects concerned. What is more, a new plural-form, gwarin (a "broken" plural, directly based on the "singular" giran; cf., e.g., qarabin, pI. of qurban) has developed and is rather regularly used as the plural-form of giran (as it seems, also giran exactly like qarayeb, see above, can still be used in the function of a plural, although gwarin is doubtless the more widely and more regularly used plural-form of the concept "neighbour"). As a matter of fact, the strangeness of the use of the original plural-form qarayeb ("relatives, kinsfolk") in the function of a singular ("kinsman, 'a' relative") has been avoided by A. Barthelemy, in his Dictionnaire arabe-fran~ais, Dialectes de Syrie, p.645, s.v. qrb, by his assumption that qarayeb constitutes the plural of the nomen abstractum qarabe « qarabat-) "kinship, relation", and that, accordingly, the concrete concept "kinsman" (singular) is here expressed by a nomen abstractum (in the plural). The use of the singular qarabat- (a nomen abstractum) in the sense of "kinsman", a use which obviously is not recorded, would quasi correspond to the use of English acquaintance in the sense of a nomen concretum: "a person with whom one is acquainted", or to English relation in the sense of "kinsman". Barthelemy's assumption, which even from the outset is not plausible, is proved as unacceptable by the entirely corresponding use of the (original) plural giran in the meaning of the singular, "neighbour" (gar); in the case of giran even an artificial derivation from a nomen abstractum does not come into question. A further, very important instance of this type, in its turn to be compared with Arabic tabi'a(hu) < *tabaya'a(hu) (abstracted from the plural tabaya'u or the dual tabaya'a, see above p. 376)-a case to which I referred already in my earlier study (Background, p. 219)is represented by the use in English of the plural friends with respect to a single person in phrases like I am friends with him, also I stayed friends with him, or I made friends with him (and other, related phrases).

378

ARABIC PARALLELS TO THE ENGLISH PHRASE

This remarkable use of the plural friends (with respect to a single person) is also found in sentences not containing such 'a complement with with as in the quoted phrases. I refer, e.g., to the following sentence from an American magazine column (Newsweek, July 2, 1973; p. 27, at the end of the page): "What mattered was the dedication to detente and the startling perception by most Americans that Leonid Brezhnev, for all his Communist credentials, really wanted to be friends". As regards the English phrase I am friends with him, W. Havers, Handbuch der erkliirenden Syntax (Heidelberg 1931), p. 82-83 and p. 232 (with reference to other researchers), states that this use of friends with respect to a single person is to be explained as caused by a "Konstruktionsmischung" ("Kontamination", "blending"), in other words: by a confusion, between the unrecorded phrase I am friend with him and the phrase they are friends. Accordingly, the plural friends with respect to a single person in the above-quoted actually used phrases would not constitute an inherent semantic feature of the idea expressed by these phrases, but would have been caused (secondarily) by an accidental mental process of an entirely formal character. Havers' explanation not only does not take into account instances like he wanted to be friends (as in the above quotation from a recent magazine article), where (as far as the linguistic expression is concerned) friends is not followed by a complement with with, but it is basically unacceptable. It is clearly shown as incorrect by the Arabic parallels mentioned above, where the plural forms of semantically related concepts are generally used in the function of the singular-forms of these concepts and are not restricted (as in the case of English friends) to a certain type of phrase. It should be recognized that the use of the plural in the abovementioned Arabic instances ("relative[s]", "neighbour[s]") and in the English instance compared with them ("friend[s]") expresses the fact that these concepts imply basically the idea of a plurality of persons, even if only a single person is mentioned. In such concepts the notion of a mutual relationship between persons is an inherent semantic aspect, not just a formal, accidental aspect. If someone is a "kinsman" or a "neighbour" of someone else, it necessarily implies that the other one ("someone else") is in turn also his "kinsman" or "neighbour". And the same refers, at least basically or generally, also to the concept "friend". Accordingly, even if only a single person is referred to in these concepts, it may lead to this single person being referred to by the plural-form of the concept concerned.

ARABIC PARALLELS TO THE ENGLISH PHRASE

379

The expression of the semantic aspect of "reciprocity" ("mutual relationship"), which in my opinion is the actual function of the use of a plural-form with respect to a single person in the instances under discussion, is further clarified and illuminated through (see above p. 375ff.) the development of the Arabic verbal expression tabi'ahu ("he followed him"), which I derive from an original verbal phrase *tabaya'ahu "he (i.e., so-and-so) pledged his allegiance to him", "he joined him in allegiance" (and thus "he is his liege-man", or "his ally"), and finally (through a specific semantic development, see below p.382) "he followed him". This *tabaya'ahu constitutes a quasi "irregular" usage of the "6th conjugation" of Arabic (and SouthSemitic generally), a conjugation ("verbal stem") which normally and basically is used for an "agens" ("subject") constituted by a plurality of beings, describing a reciprocal activity (or relationship) of these beings, as, e.g., tabaya'u "they bought from, and sold to, one another" (see the passage quoted in Background, p. 219, line 19), or also "they pledged allegiance to one another". In the instance *tabiiya'ahu (> tabi'ahu), the "6th conjugation" is, quasi irregularly, used for an "agens" ("subject") constituted by a single being, and, in consequence hereof, is followed by a complement (an object) in the accusative: *tabaya'ahu "he pledged allegiance to him". As already stated above p. 376, the shift of this particular form of the "6th conjugation" into a form of the "basic conjugation" was hardly avoidable. Thus *tabaya'ahu developed into tabi'ahu. Accordingly, Arabic tabi'a "to follow" constitutes a so-called "secondary root". Therefore, Arabic phrases like, e.g., tabi'ani ("he followed me"), or tabi'tuhu ("I followed him") (etc.) are to be considered rooted in proto-Arabic expressions like *tabaya'ani "he pledged allegiance to me", or *tabiiya'tuhu "I pledged allegiance to him", respectively, phrases which with respect to their original structure may be regarded as analogous to English phrases like he is friends with me, and I am friends with him (etc.). The difference between the Arabic and the English phrases consists mainly in the fact that in the Arabic phrases the "reciprocal" term (referring to a single person) is expressed by a verb and in the English phrases by a noun. The form tabaya'u "they pledged mutual allegiance" (and corresponding plural-forms), constituting the basis of the just mentioned Arabic phrases, may be structurally identified with the English phrase they (or: we, etc.) are friends (of each other), which in its turn may be assumed as constituting the basis of the compared English phrases.

380

ARABIC PARALLELS TO THE ENGLISH PHRASE

The so-called "6th stem" ("6th conjugation") of Arabic (existing in South-Semitic generally) serves to express "reciprocal" activities, e.g., Arabic takiitabu "they corresponded", "they wrote to one another" (based on kiitabahu "he wrote to him", "he corresponded with him"), or taqiitalu "they fought one another", "they attempted to kill one another" (based on qiitalahu "he fought him", "he attempted to kill him"). But, whereas in most instances of this "conjugation", the reciprocity is quasi an "accidental" aspect of the activity concerned, in a case like tabiiya'u (which I consider the basis of tabi"ahu < *tabiiya'ahu) the "reciprocity" is an "inherent" aspect of the activity (see below p. 383). That means: The activity or attitude exerted or exhibited by one of the parties engaged in the "reciprocal" activity or attitude is necessarily exerted or exhibited by the other party. For that reason, the aspect of "reciprocity" is not only inherent in, e.g., tabiiya'u "they pledged allegiance to one another" (that is: in the form of speech which expresses the aspect of "reciprocity" by a special verb-stem for "reciprocal" actions), but is also inherent in the corresponding formally quasi non-"reciprocal" conjugation, biiya'ahu "he pledged allegiance to him" (that is, in the so-called "3rd conjugation", which is the basis of tabaya'u, the "6th conjugation": "they pledged allegiance to one another"). It is for this reason of implying the aspect of "reciprocity", that biiya'a-hu (without falling into disuse) was replaced in proto-Arabic (or, already in proto-SouthSemitic) by *tabiiya'a-hu (> tabi"a-hu), which thus originally expressed the same meaning as biiya'a-hu. Of course, although *tabiiya'a-hu (> tabi'a-hu) emerged in the singular (that is: in just this form, or in any form-i.e., grammatical person-of the singular), it could secondarily also be used in the plural, in any grammatical person. My assumption (stated in my original publication, see above p. 374) that bay'ah was already in proto-South-Semitic time used in the sense of "homage", "pledging one's allegiance (to someone else)" is not only based on my derivation of Arabic tabi'a-hu from *tabiiya'a-hu, but also (see l.c.) on the Ethiopic etymon for "man" (see above p. 375), which I derived from precisely this latter verb-form, in interpreting it in the (quasi secondary) meaning "he served him as his 'man' (in war)" ( < "he pledged his allegiance to him"). The semantic difficulty ("semantische Schwierigkeit"), claimed by Rundgren (l.c., p. 55, bottom), in the transition of a term of the meaning "(he is the) 'man' (of so-and-so)", "(he is his) 'man' ('Gefolgsmann') [in war]" into a term of the general meaning "man" (as counterpart to "woman")

ARABIC PARALLELS TO THE ENGLISH PHRASE

381

is not existent. My reference, in this context, to French hommage (hominatieum), Engl. man red (see Background, p. 216), which Rundgren (I.e., p. 56) is unable to accept, is certainly in place. It is completely clear that in the same way as a term which originally designates the concept "man" (as counterpart to "woman") can acquire the additional meaning "vassal, liege-man, 'Gefolgsmann', companion in war, confederate", conversely a noun of the original meaning "liege-man, vassal, confederate, companion in war", may ultimately be used in the sense of "man" ("l'homme male") (as counterpart to "woman"). In this respect we should pay attention, e.g., to Latin socius "companion" « original Indo-European *soqUjos) and its cognates in other Indo-European languages (see Walde, Latein. Etymo!. Worterb., 3rd ed., vol. 2 (1954), p. 551). In Anglo-Saxon, e.g., the term corresponding to socius, that is: seeg, is used in the meaning of "companion", "man"; see, e.g., Boisacq, Diction. etymol. de la langue greeque, 3rd ed. (1958), p.269, 1. 4; and cf. also 1.-B. Bessinger, A diction. of AngloSaxon poetry (1960), p. 57: "seeg, warrior, man". Also the relationship of socius to the verb sequor "to follow" (see Walde, I.e., and Ibid., p. 519) is of considerable interest in the present context (see below p. 385 concerning a certain semantic parallelism between sequor and its Indo-European parallels on the one hand, and Arabic tabi'a "to follow" on the other hand). In connection with the origin of the Ethiopic etymon for the concept "man" one may probably also refer to Latin vir which is related to vis "strength". It seems to be generally assumed that vir originally designated precisely the "man" in his quality as the "warrior", "the one who is the 'man', who stands in the service of, or: accompanies, or: 'follows' a lord in his warlike enterprises". The semantic development also of this Latin word accordingly is, to a certain degree, an analogon to the semantic development of the Ethiopic word, according to my interpretation. With respect to the Ethiopic word for "man", tabii'!, we should of course also not overlook the meanings of the other terms derived from the same root in Ethiopic: t{)bu< "virilis, fortis", and the verb tab 'im, 'in (and to hom, h-n, respectively). The semantic-functional shift "[on the day] when" > "when" > "if" (cf. German wenn = "if"), which is the basis of the morpho-phonemic shift of yawma to 'im, 'in (etc.), has recurred in (e.g.) the modern-Arabic dialect of 'Oman, where yom serves regularly in the function of a conditional conjunction, that is: in the function of Old Arabic 'in "if"; (see for details I.e., p. 242). With respect to the replacement of the original a-vowel by i, which is a characteristic feature of the development of 'im, 'in (etc.) from am « yawma), we refer to ibid. We also mention the Mehri-forms han, han, where the original a-vowel is preserved. Another instance, showing an analogous phonetic development of the primary form yawma, we recognize in the Arabic particle 'an, characterizing subordinate clauses (substantive clauses), which in modern dialects, as far as it is still in use, appears as in (n). The semantic and functional shift present in the development of yawma "when" (conjunction) into an "that" (= Latin ut, German dass) may be demonstrated by a sentence like sarrani an gi'ta "it gave me joy that you came", which (theoretically) may be traced back to a form sarrani yawma gi'ta "it gave me joy when you came". That means: yawma "when" > an "that" (= "dass") is based on the transition of an adverbial (temporal) clause into a clause used in the function of a grammatical case (in the quoted instance used in the function of a subject, thus quasi being in the nominative). For the details of

392

EXPRESSIONS BASED ON THE NOUN YA WM- "DAY"

this development, we refer to our above-mentioned article, p. 243-244, and especially to the "Appendix" below p. 428. The transition of "when" (conjunction) to "that" (= German "dass") which we claim for Arabic yawma (> an), is confirmed, e.g., by the use of the Akkadian conjunction inuma (enuma), which, although its original meaning is certainly "when", is also used in the function of "that" (see Chicago Assyrian Dictionary, vol. 7 (Ill), p. 159 a. In connection with this certainly remarkable development of the conjunction yawma "when" (introducing adverbial clauses) into the particle an "that" (= German "dass") characterizing clauses in function of a "grammatical case" (substantive clauses), we must also mention the development of this an into a conjunction expressing the meaning "lest" (and thus again characterizing adverbial clauses; see above p. 332). A further, quite striking development of an « yawma) we recognize in its development into a preposition. Indeed, we assume that the Arabic preposition 'an (with initial 'ayn), expressing the meaning "from", "away from", "off" (to a certain degree synonymous with the ancient [West-]Semitic preposition min) has developed from 'an (with initial alif) and precisely on the basis of its use in the meaning "lest" (for the details of this syntacticsemantic development, see above p. 331 ff.). For the development of this consonant' ('ayn) in 'an instead of the original alif (the glottal stop) in 'an, we may refer, e.g., to the particle la'alla "perhaps" (with 'ayn), which we derive from the original conjunction Itallii « Ii' an Iii) "lest" (see above p. 328). The forms of the conditional conjunction "if", in Arabic: in, in Aramaic: en, hen (beside the form of Hebrew: im) as well as the subordinating particle of Arabic: an "that" (= German "dass"), in relationship to the primary form underlying these forms, that is: yawma, clearly demonstrate the possibility (from the phonetic point of view) of the development, assumed by us, of the Akkadian prepositions ana and ina out of an original form yawma. One should bear in mind that we ascribe the development of Akkadian ana and ina from yawma to the Proto-Akkadian, or even to the Proto-Semitic period, when the word for "day": yawm- had not yet been affected by the specific phonetic rule of Akkadian concerning diphthongs, which changed the Proto-Semitic yawm- into um-. Of course, Akkadian ana is somewhat closer to the basic form yawma from which we derive it than ina. But the possibility of the shift of the original a into i in ina ( < ana or < *ama, see below) is

EXPRESSIONS BASED ON THE NOUN Y A WM- "DA Y"

393

clearly proved by the West-Semitic conditional conjunction 'im ( > 'in) < yawma, and also by the shift of Arabic 'an "that" ("ut") into 'in in modern-Arabic dialects. A special problem is posed by the replacement of m by n which the development yawma > *ama > ana and ina involves. The shift of terminal minto n is clearly attested by some variations of the conjunctions and quasi-prepositions which developed from the original yawma in various West-Semitic dialects, as, e.g., the temporal conjunction of Sbawri, 'en "when" ([conjunction] < Mehri 'am, Soqotri 'am), the subordinating conjunction of Arabic, 'an ("ut", "that"), and the Arabic and Aramaic form of the conditional conjunction, that is 'in (["if"] < Hebrew 'im "if" < yawma "when"); cf. above p. 391 and p. 392. Also the Arabic ending n as sign of the indefinite noun, which is certainly based on an original m, is a clear instance of terminal m > n (see Brockelmann, Grundriss, I, p. 136). For a similar instance of Akkadian see von Soden, Grundr. der akkad. Gramm., §45 c. But in the case of Akkadian ana and ina, which, like the afore-mentioned conjunctions, we consider based on an original yawma in a conjunctional use, the original terminal a of yawma is preserved, and the shift of minto n thus does not appear to be connected with a terminal position of the original m. Of course, beside ana and ina, there exist the forms an and in, and precisely in the early periods of Akkadian linguistic history; we refer to C (hicago) A(ssyrian) D(ictionary) , vol. I ("A"), pt. 2, p. 100 a, and vol. VII ("I-J"), p. 142 a; furthermore to von Soden, ZA 41, p. 134. We cannot entirely disregard the possibility that these forms, an and in, are directly developed from *'am and *'im (in which the terminal a of the original adverbial accusative yawma was lost, in analogy to the above-mentioned West-Semitic forms), and that the consonant n which developed from the terminal m, was subsequently transferred to the more primitive forms *ama and *ima: > ana and ina (concerning the existence of *ima, or ema, see below p. 398). On the other hand, it may be assumed that the n in ana and ina could even have developed when the original m was not in terminal position, but was followed by a vowel, a; that is, the n may have developed in the forms *ama and *ima. We recognize an analogy to the forms under discussion in the endings of the masculine plural and the dual (of nouns) in West-Semitic. The masc. pl.-ending of Hebrew and Phoenician, -im, appears in Aramaic as -in, and in Classical Arabic as -una (nomin.), -ina (genit./accus.), i.e., augmented

394

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

by a terminal vowel a (which, indeed, disappears in "pause"). It cannot be doubted, however, that the Hebrew ending -im in turn is to be traced to a more original form -ima with a terminal vowel a. This earlier pl.-ending -ima is not only made probable by certain Canaanite word-forms found in the Tell-el-'Amarna documents, and also by certain "East-Canaanite" (Amurrite) forms (see A. Goetze, Language, vol. 17 (1941), p. 130), but this earlier form of the ending, with a terminal a-vowel, is also clearly attested in Ugaritic, a language which, like Arabic, has preserved terminal (declension-)vowels, and in which the masc. pl.-ending, characterized by the consonant m, appears under the forms -uma (nomin.) and -ima (genit./accus.). The same apparently refers to the forms of the dual. Cf. C. H. Gordon, Ugaritic Textbook (1965), p.54; and also Goetze, I.e., p. 133 (and n.54). If we compare the masc. pl.-endings of Arabic -una (nomin.) and -ina (genit./accus.) with the corresponding endings of Ugaritic, -uma and -ima, and likewise compare the dual-endings of the two languages, we have a clear parallel to the shift of *ama and *ima into ana and ina in Akkadian, a shift which is implied in our derivation of the Akkadian prepositions from Proto-Semitic and Proto-Akkadian yawma. Since the pause-forms of the Arabic plural-endings -una and -ina are characterized by the apocopation of the terminal vowel, and thus appear as -un and -in, respectively, one might be tempted to assume that the shift of minto n in the primary forms of these endings came about when the original m was vowelless (i.e., in pause): -um, -im > -un, -in, and that the consonant n which thus developed, was transferred to the context-forms -uma, -ima, which thus became transformed into -una, -ina. Although such a possibility cannot entirely be disregarded, it seems far more likely that -una, -ina developed directly from forms in which the original m was followed by the (terminal) vowel a; and analogously we may also assume that Akkadian ana and ina are directly based on a more original form *ama (or *ima) < yawma. The shift of intervocalic minto n should be considered no less possible than the clearly attested shift of terminal minto n (cf. above p. 393). In connection with this shift, assumed by us, of intervocalic minto n, we have to deal with some further particles which we consider based on yawma, from which we derive Akkadian ana and ina. As already mentioned (see above p.391), we derive from yawma "on the day when" the West-Semitic conditional conjunction 'im, 'in,

EXPRESSIONS BASED ON THE NOUN Y A WM- "DAY"

395

also h-m and hen (in Mehri: han and hiin) ("on the day when" > "when" > "if"); we refer to our remarks in Le Museon, vol. 83 (1970), p. 241-245. A quite similar origin we ascribe (see l.c., p. 245247) to Arabic ammii, the particle characterizing the "isolated", "extraposed" noun ("as for ... ", "as concerns ... "), in maintaining that its original meaning was "if" and that it is based on an original combination yawma-mii (a morphological structure which we also consider surviving in Arabic immii [coexisting with the conditional conjunction in] as well as in the conditional conjunction of Ethiopic: amma [yawma-mii > ammii > immii], see l.c., p. 245). As to the use of the conditional conjunction ("if") in the sense of "as to ... ", "as concerns ... ", we may refer to various uses of this type in several languages. We may mention the use of iljii and kiin « inkiin) in certain North-African Arabic dialects in the sense of "Classical" Arabic ammii; the use of Mehri wa-hen (comprising the conditional conjunction) in the sense of Arabic wa-ammii; the use of the Amharic expressions (serving as conditional conjunction, "if") in the same function, that is: in the meaning "as for ... ", "as concerns ... " (for details and references concerning all these uses see l.c., p. 246). Moreover, we should also refer to the conditional conjunction of Turkish, ise, which again is also used to characterize the "extraposed" noun, the "casus pendens"; (see l.c., p. 246, and also our Studies in Arabic and General Syntax (Cairo 1953), p.69, at the bottom of the page). For a more detailed discussion concerning the parallelism between the "extraposed" noun (the "casus pendens") and the (conditional) protasis we refer to our Studies in Arabic and General Syntax, p. 19. It should, however, be stressed in the present context that the" conditional" character of the concept "as to ... ", " as for ... " (Arabic ammii) manifests itself also clearly in the frequent use of Arabic ammii in combination with conditional(-temporal) conjunctions, e.g., ammii ig ... (a quite frequently used combination), also ammii law ... , ammii /:lina ... , etc. (see for further details our abovequoted publications). The Arabic particle ammii and the West-Semitic conditional conjunction im, in (and variations) are both based on an original (Proto-)Semitic temporal conjunction yawma (which in the case of the form underlying ammii had been augmented by the enclitic -mii: yawma-mii). We reject the widely held opinion (for references see our article in Le Museon, vol. 83 (1970), p. 241) that the conditional conjunction of West-Semitic: im, in (and variations) is based on an original deictic (demonstrative) element. With respect

396

EXPRESSIONS BASED ON THE NOUN YA WM- "DAY"

to the enlargement of the original conjunction yawma "(on the day) when" by mii, that is: yawma-mii (the form on which we consider Arabic ammii based, see above), we should also mention the temporal conjunction of Ethiopic: amma "when", in which (in contrast to a different opinion) we recognize the same enlargement of the basic term yawma by mii (see our remarks in Le Museon, vol. 83 [1970], p.245). The Arabic particle ammii "as to ... ", "as concerns ... ", the origin and use of which we have tried to establish (see above), is also of importance to us with respect to a specific problem, that is: the problem of the shift of intervocalic minto n (see already above p. 393). With respect to the possibility of the shift of intervocalic minto n, we now draw attention to the fact that in the Soqotri-texts published by D. H. Miiller in Die Mehri- und Soqo!ri-Spraehe, III (Sudarabisehe Expedition, Band VII) there appears a particle enne in the sense of "but", "however", which Muller correctly identified with Arabic ammii. We have to refer here to the well-known fact that Arabic ammii "as for ... " itself may in certain contexts (where a contrast is involved) be interpreted in the sense of "however", "but", a meaning which is especially frequent in modern (literary) Arabic. With respect to Soqotri enne in the meaning of "but", "however" « ammii), we quote the following passages from Muller's Soqotri-texts. L.e., p. 47, no. 12,11: Wu-bereq bis e'bOyhe wu-eFie (= wa'ammii) dihe bebe rer si be-nh6fs "Und es beneideten ihn seine Bruder, sein Vater aher bewahrte die Sache" (more literally: " ... and as for his father [wa'ammii abuhu], he preserved the matter"). Furthermore, Ibid., p. 49, no. 12,24: Wu-ze'e§ wa-qele/:l bis ictebehor, wunne [= wu-enne = wa'ammii] 'ebehor wu-kiinoh qahddoh bfsi bis rfho "Und sie packten und warfen ihn in einen Brunnen; aber dieser Brunnen war leer und kein Wasser darinnen". This Soqotri particle enne < ammii represents a further important instance of the shift of intervocalic minto n. That the m shifted into n in this clear specific instance is doubled (see also the instances mentioned below p.413), is with respect to the possibility of the shift of intervocalic minto n in principle unessential. Having examined the phonetic development of ana and ina from yawma, we may turn to their semantic characteristics. We should keep in mind that the various functional uses of ana and ina are interrelated to a certain extent and overlap, as do the functions of corresponding prepositions and equivalent forms of expressions in other

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

397

(also non-Semitic) languages (cf. below p. 398ff.). Furthermore, although ana and ina are predominantly used as prepositions, we should not overlook the fact that there exists a far reaching parallelism between prepositional and conjunctional function. We may therefore assume that, in the same way as prepositions (terms governing substantive nouns) may be secondarily used in conjunctional function (Le., as terms introducing adverbial, especially temporal, clauses), so original conjunctions of temporal meaning could develop into prepositions of temporal meaning, and secondarily also of local meaning. We may refer to such use of the original Arabic conjunction yawma > yom "when" as a temporal preposition in the Arabic dialect of .oofiir, as well as in Mehri, S.gawri, and Soqotrl. Not only has yawma acquired the function of a preposition beside its more original function as a conjunction in these dialects, but the word itself (in its use as conjunction and preposition) has been weakened in its phonetic structure: .ofiid yam, yem, em; Mehri am; Soqotri 'am (with 'Ayn); S.gawri en (beside yem). Also the Ethiopic temporal particle am(m)a has, with reference to these South-Arabian particles, been explained as derived from the Common-Semitic word for "day"; we refer to Brockelmann, Grundriss, II, p. 535; Rhodokanakis, Der vulgararab. Dialekt im pofar, vol. II (Siidarab. Expedition, Band X, 1911), §30 a; p. 123-124 (cf. also our remark above p. 396). Among the South-Arabian (and Ethiopic) forms quoted, special attention should be paid to S.gawri en, which is strikingly similar to the form represented by Akkadian ana and ina. We mention a few expressions in which the South-Arabian particle, based on an original yawma, serves as "quasi-preposition" . .ofiiri (see Rhodokanakis, I.e., p. 123 b): ii-yam wage! el-'a$or "zur 'A~r­ Zeit"; yam q-$abelJ "des Morgens" (= fi $-$ublJ); yam el-leyla "eines Abends". Soqotri (see Brockelmann, Grundriss, II, p.535): ya-'am fey jom "und eines Tages" (Muller, II, 114,26). Rhodokanakis (I.e., II, p. 123 b, ult.) refers also to a "local" use of this quasi-preposition yam (derived by him from yawma): "Eine lokale Bestimmung gibt: yam gimir remid 'bei einer Stauung [des Tales] Remid' 105,20" (see Rhodokanakis, Ibid., [Glossary] p. 35 b: "gimir Enge, Stauung (Tal)"). The use of .ofari yam in this latter instance, where it is used as a local preposition ("at"), is interesting to us in view of our derivation of Akkadian ina and ana (preponderantly used in local meaning) from Proto-Semitic yawma. In connection with the conjunctional as well as prepositional func-

398

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

tion of the afore-mentioned South-Arabian particles derived from an original yawma, we also refer to the Akkadian particle ema, or ema, which is used (in local as well as temporal sense) both in conjunctional function (before clauses and clause equivalents) and in prepositional function (before nouns). We refer to Muss-Arnolt, A concise dictionary of the Assyrian language, vol. I, p. 54 b; Brockelmann, Grundriss, II, p.415, 538, 558; furthermore CAD, vol. 4 ("E"), p. 136. MussArnoh equates ema in its function with ina and ana. In CAD (l.c.) the following is stated with respect to ema: "It mostly has spatial connotations [i.e., the meaning 'wherever', etc.], but note: e-ma arfJi ina iteddusika "at every new moon when you [Moon God] renew yourself' ... ; and e-ma A.MES IZKIM.MES 'whenever there are (evil) signs and portents' ... ". We mention also the following instance quoted by Brockelmann (l.c., p.415): ema uzammaru "wahrend sie den Ruf anstimmten". We recognize also in ema a phonetic development of Proto-Semitic yawma "when", and compare it with the prepositions ina and ana, especially with ina. Inasmuch as the vowel of the stem in ema is realized as e, we may refer to enuma "when" « *ina-umi-ma; see our remarks in Le Museon, vol. 83 (1970), p. 248; cf. below p. 400). Probably, the vowel e of ema should be considered as short, and not as long, i.e., ema, as the word is ordinarily transcribed. Of considerable interest is the use of ema as a local and temporal preposition, e.g. (Brockelmann, l.c., p. 415): ema biibiinisa "an ihren Toren"; ema !ibbam "in meinem Herzen"; ema qabli u tabiizi "in Kampf und Schlacht". The particle ema is thus characterized by its having preserved the original function of yawma as a temporal conjunction ("when") as well as by its use in the function of a local and temporal preposition, that is, by its use in the characteristic function of ina and ana. Concerning the prepositions ina and ana, we assume that the original conjunction of the meaning "when" developed into a preposition of the meaning "at (the time 00" (exactly as in the Arabic instances referred to above p. 397), and that this temporal preposition was secondarily used in local function, just as many other prepositions may be used in a temporal as well as in a local meaning. The prepositions ina and ana, in their use as local prepositions, are ordinarily more or less restricted to specific, distinct uses-so that ina is mainly used in a strictly "local" sense, to indicate "rest at a place": i.e., "at, in"; whereas ana has mainly a "directional" meaning, indicating "the movement towards a certain object", which involves meanings as

EXPRESSIONS BASED ON THE NOUN Y A WM- "DAY"

399

"to, towards, for". It should be noted that there is no exact borderline between "strictly local" ("at, in") and "directional" ("to, towards, for"). Although, on the basis of the origin of these prepositions, as viewed by us, the "strictly local" sense, "at, in" is to be considered more primitive than the "directional" (and "dative") meaning, "to, towards, for", the secondary use of such a "strictly 'local'" preposition in a "directional" function must be considered entirely possible. Analogously, the use of a preposition of "directional" (and "dative") function (as, e.g., Hebrew/Aramaic Ie) in a "strictly 'local'" function is often observed in Semitic. This use of a "directional" indicator in a "locative" function is apparently a widespread phenomenon. We refer here only to the following statement by Omelian Pritsak (Studia Orientalia [Helsinki], XXVIII: 6 (1964), p. 7): "1m Alttiirkischen hat sich des 6fteren noch ein Zustand erhalten, in dem die lokativische Beziehung sowohl mit Hilfe des Lokativ-Suffixes als auch des Dativ-Suffixes ausgedriickt werden konnte. (Vgl. den Dativus-Lokativus im Mongolischen). . .. " (in the sequel Pritsak mentions an expression from an ancient Turkic inscription, biiriiyii (dative of biiri'" biri "the South"), used in the function of the "locative": "in the South", as well as (quasi) in the meaning of the "dative": "towards the South"). As far as Semitic and, more specifically, Akkadian itself are concerned, we may refer, e.g., to anumisu ( < ana umisu, see below p. 403), used synonymously with inumisu « ina umisu, see ibid.) "at that time", "then" (see CAD, vol. 1 ("A"), pt. 2, 146 b; and vol. VIII ("1-1"), p. 162). (Of course, the more literal meaning of the terms inumisu and anumisu is: "at (or: 'to') 'its' day", i.e., "at (or: 'to') the same day"; see below p. 422). As to the transition of the Proto-Semitic conjunction yawma into prepositional use (i.e., yawma > ana, ina), we should also pay attention to the fact that the West-Semitic local preposition bali, as well as its Akkadian counterpart ina, can be used before an infinitive or nomen action is, thus before a noun which corresponds in content to a subordinate clause dependent on a conjunction. In the same manner, the Proto-Semitic conjunction yawma may in turn have been used to introduce infinitives (corresponding to subordinate clauses). We may compare herewith the use in Akkadian of inuma in prepositional function before infinitives or abstract nouns (in the meaning "at the time of"), as inuma ereb /Star ana ekallim "when Istar entered the palace", or inuma kakki "at the time of the battle" (see CAD, vol. VII ("1-1"), p. 161). The prepositional use of inuma must be derived from

400

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

its use as a conjunction, "when" (see Ibid., p. 159 a). We stress that we consider iniima based on an original form ina-iimi-ma (literally: "on the day when (or: that] ... "), which through haplology (-mi-ma > rna) developed into iniima (see our discussion in Le Museon, vol. 83 (1970), p.248, and cf. below p.402). We are unable to accept von Soden's interpretation (GAG, § 113f; see also § 114c, and § 116b), who considers the element -iim in iniima as the loc.-adv.-ending -urn added to a basis inu (an interpretation which corresponds to his explanation of the element -urn in anumma; see below p. 423). An explanation of iniima similar to von Soden's seems to be suggested in CAD, vol. VII ("1-1"), p. 159a (cf. also p. 158b), where iniima and also iniimi (see Ibid., p. 161 b) as well as iniimisu (cf. Ibid., p. 162a) are compared to the conjunction inu (see Ibid., p. 152 b). These words, i.e., iniima as well as iniimi, can, in our opinion, not be derived from a basis inu(m) , but should be considered developed from the basic expressions *ina iimi-ma, and *ina iimi, respectively. But also iniima in the function of an adverb: "now; here is", also "then" (see CAD, vol. 7, p. 158 b) we trace back to the same basic form: *ina iimi-ma (cf. the adverbs of approximately identical meaning: anumma, enanna, which we likewise consider developed by haplology from *ana (ina)-iimi-ma; cf. below p.413). Similarly, iniisu "at that time", which in CAD, vol. III, p. 162 b, is derived from inu (conjunction), should be derived from inumisu < *ina umisu (ef. above p. 399). Adverbial expressions of this kind, of whose original, literal meaning the speakers are no longer aware, frequently undergo farreaching changes in their phonetic structure. As to the conjunction inu (enu, inurn, enum; see CAD, vol. VII, p. 152 b), we consider it developed from the above-mentioned conjunction i/eniima "when" « *ina iimi-ma) by the loss of the vowel a of the terminal (originally enclitic) -rna: i/enuma > i/eniim (> i/enu). A further parallel to the secondary prepositional use of the conjunctions yawma ( > ana, ina) and inuma ( < *ina iimi-ma) is the SouthArabic and Ethiopic use of conjunctions-e.g., those based on the original yawma (see above p. 391}-in the function of "quasi-prepositions" before nouns expressing concepts of "time", as "day", "night", etc. (see above p. 397). We observed such a use of the conjunction yawma-in its original, phonetically unimpaired form-even in the ancient, "Classical" Arabic language. We refer to the following line of poetry (Tabarl, Annales, II, p. 1558, 15): Iii ta/:Isibanna I-/:Iarba yawma rj-rju/:Iii kasarbika I-muzzii'a bil-biiridi "Don't regard fighting

EXPRESSIONS BASED ON THE NOUN Y A WM- "DAY"

401

in battle in the forenoon (quasi literally: 'on the day of the forenoon') like your drinking sweet wine mixed with cold water!". This use, in "Classical" Arabic, of yawma in the function of a temporal (quasi-) preposition is identical with the use of the unimpaired yawm in (e.g.) the modern J:Ia9rami dialect, as, e.g. (Sudarab. Expedition der Wiener Akad., vol. VIII, p. 19,19 [no. 9,2]), wa ydum a!;-!;obab yasir ... "und als es Morgen war, ging er fort ... " (or simply [without a verb: "war"]: "and in the morning he went away"). Cf. also (Ibid., p. 20,24 [no. 9,5]) youm at-'a!;r ... "Wie es Spatnachmittag war ... " (or: "in the late afternoon ... "). (Cf. above p. 397). In analogy to this use of nouns of "time" in the function of "quasiprepositions" in phrases serving as temporal adverbs, we interpret also the development of Akkadian a/ina from (Proto-Semitic) yawma, and of inuma from *ina umi-ma (see above). We assume that in the initial stage of development these terms (temporal conjunctions of the original meaning "[on the day] when") were used as temporal prepositions before nouns expressing concepts of "time" or before infinitives (cf. especially the prepositional use of inuma in the instances quoted above p. 399). However, in agreement with the current use of the same prepositional elements for the formation of both temporal and local adverbial expressions, we fully understand that forms like ana, ina < yawma could also be used in the function of local prepositions (before nouns of concrete meanings). We extend our derivation of Akkadian ana, ina from an original (Proto-Semitic) form yawma to a West-Semitic local preposition frequently used in temporal function, that is: Hebrew and Aramaic 'im (in Syriac: 'am) "with", a preposition which is also attested in OldSouth-Arabic and in Ugaritic: '-m. The initial sound 'Ayn in this word corresponds to the 'Ayn which characterizes one of the variations of the (modern) South-Arabic temporal conjunction and quasi-preposition derived from yawma, that is: Soqotri 'am (see above p. 397). Moreover, this 'Ayn appears also in various other particles which are developed from the original form yawma. We mention the Arabic preposition 'an "away from", which we interpret as derived from 'an in the meaning "lest" « "that" [= German "dass"] < yawma "when", see below p. 428); and also the mainly used word in Arabic for the concept of "perhaps", that is: la'alla, which we derive from an original Ii' alia « Ii' an la) "lest" (see above p. 328). It can hardly be doubted that Arabic ma'a "with" represents a phonetic variation of Hebrew/ Aramaic 'i/am "with" (derived by us from *'ama < yawma). We

402

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

obviously deal here with a case of metathesis, although the basis of this type of metathesis does not seem to have been satisfactorily clarified. The particle 'a/im (and the same refers to Arabic ma'a), though generally used as a local preposition, may also be used in conjunctional (temporal) function, as in the following Syriac instance (see Brockelmann, Grundriss, II, p.415): we'am dasma' qiilii hiinii "und sobald er diese Stimme hOrte" (quasi more literally: "and 'at' [or: upon] his hearing this voice"). Instances of this type may indeed correspond to those basic constructions in which an original yawma "[on the day] when" acquired the meaning "with", "at", at first used in a temporal function, and subsequently also in a local function. (Cf. also Syriac ma ge "as soon as"; see below p.411). Let us now turn our attention once more to the shift of intervocalic minto n (see above p. 393 ff.). Apart from previously cited examples we may refer to other Akkadian forms which in their turn seem to us to exemplify this shift, especially the shift of intervocalic geminated m (i.e., mm) into geminated n (i.e., nn) (cf. above p. 396). We have in mind instances like the adverb anumma (anummi, anummu) "now, here" (~ee CAD, vol. I ("A"), pt. 2, p. 147), and the adverb inanna (eninna, enanna, etc.) "just now" (see CAD, vol. VII ("I-J"), p. 142). The two adverbs appear frequently together; we refer to CAD, vol. 7, p. 144: "While in E[I] Am[arna], Bogh[azk6i] and R[as] S[hamra] inanna is often used beside anumma for emphasis, the Mari letters use anumma to introduce the first topic of a letter, and inanna for the following and always for the last topic". (Cf. also CAD, vol. I ("A"), pt. 2, p. 147 a; and vol. VII ("1-1"), p. 143 a, sub "d"). We assume that the adverbs anumma and inanna are both derived from primary forms in which the preposition ana or ina, respectively, preceded the word for "day": ftm-, which in turn was followed by the particle ma-that is: *a/ina ftmi-ma. In agreement herewith, the deVelopment was: *ana ftmi-ma> anumma; and analogously inanna may be derived from *ina ftmi-ma > *inamma (> inanna [see below]). Accordingly, we consider the forms underlying anumma and inanna as having been of the same structure as the form assumed by us as underlying the (more or less) synonymous adverb inftma (see above p.399), that is: *ina ftmi-ma (see above p.401). However, while in this latter instance the first of the two original syllables beginning with m has completely disappeared by the process of haplology (without any trace of the original syllable -mi- being left), in anumma

EXPRESSIONS BASED ON THE NOUN Y A WM- "DAY"

403

< *ana-iuni-ma the same phonetic process has led to a geminated consonant m. This second type of development represents an instance of a specific variation of the process of haplology (see above p. 51 fT.). As for our analogous derivation of inanna from *inamma < *ina-umima, we note that this derivation involves a further instance of the shift of intervocalic m(m) into n(n); cf. above p. 396 and below p. 413. In our attempt to interpret anumma and inanna "now" (or: "here") as based on a combination of the preposition a/ina and the noun um"day", followed by the (emphasizing) particle ma (*a/ina-umi-ma), we are not only guided by the analogous origin assumed by us for the term inuma « *ina umi-ma, see above p.400), but also by the obvious relationship between anumma "now" and the adverb anumisu "then, thereupon". We derive anumisu "then" from an original *ana umisu "at the day of 'it' [Le., of such-and-such previously mentioned fact]" (see below p.422). The original preposition an( a) in anumma, as well as in *an( a)-umisu, would again express "rest (at a place)" (more accurately: "'rest' in a certain time") and not "movement towards" (in agreement with its ordinary function; cf. above p. 398). Beside a more original temporal meaning of most of these adverbs, a secondary local connotation seems always to be present. Besides anumma and inanna, we may mention also the adverb (see CAD, vol. I ("A"), pt. 2, p. 149 a) anummis, also annummis and anaummis "(over) there, thither" (in Mari and Standard Babylonian). With respect to the origin of this adverb (as well as of the other adverbs mentioned), the variant form anaummis is especially noteworthy. As noted above (p. 402), anumma and inanna (which latter is, in our opinion, related to the former) frequently occur together in letters under specific circumstances. With respect to this use in letters we should compare them with Biblical-Aramaic ke'~n~l (Esra 4,10 f.; 7,12; also occurring in Egyptian-Aramaic), which, in its turn, introduces the actual topic of a letter (see Torrey, Journ. of Bibl. Lit., 16 (1897), 166f.; Lidzbarski, Ephemeris, 2, 229 f.). Obviously, all these particles, the Akkadian no less than the Aramaic, were originally used to mark the beginning of the actual topic of a letter as against the introductory formula of greeting: "[and] now [we come to the matter which concerns us]" (cf. Arabic ammii ba'du "as to [the matter which comes] thereupon"). It may be ascribed to secondary development if this formula, "[and] now", is also used when no such introductory formula of greeting actually precedes (as is frequently the case with Arabic ammii ba'du). But the Akkadian and Aramaic word-

404

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

forms for this "now" are themselves to be considered interrelated, since *'r:nfl in ke'r:nr:l (Esra 40,10; 7,11), *'el « *'r:nr:l) in k e'el (Esra 4,17), or *'an in ke'an (Dan. 2,23; 3,15; etc.), as well as Hebrew 'el "time" ( < *'r:nr:t), should in their turn be considered derived from an original yawm (in Aramaic *'r:nr:l > 'el augmented by the fem.ending; cf. below p. 405), and thus illustrate the shift of minto n, which is the basis of the relationship between anumma and inanna. (With respect to the preposition k e which precedes the afore-mentioned adverbial expressions, these expressions may be considered as originally more or less identical with the adverbial expressions of Hebrew: kayyam and kehayyam "today", "on this day"). As for Akkadian anumma and inanna, with which we started this paragraph, it seems to us that the usual Akkadian term for the introduction of direct speech, umma, and the equivalent Old Akkadian term enma (see von Soden, GAG, § 121 b) are related to them. The forms umma and enma seem to be shortened (mutilated) from more original terms like anumma and *inamma (> inanna, see above p.403) "now", or from other variations of these forms (which all included the noun for "day", Proto-Semitic yawm-). We shall now look at some Arabic nouns (and verbs) derived from the original (Proto-Semitic) noun yawm-, which are again characterized by the shift aw > a and m > n. We assume that the Arabic noun an- "moment, instant, time" has developed from an original yawm-. The frequently used phrase (fJal-ana "(and) now" (or: "[and] at this moment"), e.g., should be traced back to an original (fJal-yawma "(and) today", that is: "(and) now". Variants of Arabic an- "moment" < yawm- "day" we have to recognize in awan- "moment, time, period" and in a/ina(nJ "time" (see for this noun, e.g., Blachere, Dictionnaire arabe, vol. I, p. 271 b). This noun, a/ina(nJ "time" (a variant of an"moment"), must be considered the basis of the verbal root of Arabic: ana "to approach, come (esp. time)" ("its time came"). This Arabic verb has a parallel in Hebrew, that is: inna (Pi'el) "to cause to occur" (see the dictionaries). Accordingly, this Hebrew verb should in its turn be traced back to yawm (cf. 'ana, see below p.405). Arabic ana "to come (esp. time)" is also used in the meaning of "to mature, become ripe" (aniya), and occurs also in the "fifth stem": la' anna "to do sth. slowly, to take one's time, to temporize, to put off (doing) sth." (see Blachere, I.e., I, 271 b). From this verb (in the "fifth stem"), ta'anna ("to do sth. slowly", etc.) we should derive 'Iraqi-Arabic tina (imperf.: yatnO "to wait" (especially used in Bagdad). The same

EXPRESSIONS BASED ON THE NOUN Y A WM- "DAY"

405

derivation-that is: from an-, or a/ina(n) , "time"-should also be assumed for the verb ana in the sense "to enjoy peacefulness", "to act with gentleness". The expression for "the slow and easy (and, secondarily, gentle) movement" is derived from the term for "time" and acquires very naturally the connotation "to behave in a gentle way". We assume that the root hwn, which in its meaning is related to the root 'wn (see Blachere, Diet. arabe, I, p. 313 a, at the bottom), is nothing but a further variation of the same, more original root 'wn, based on the noun an- "time" (cf. awan- "moment, time"), which itself has, in our opinion, developed from yawm- "day". Furthermore, we should not overlook the most frequently used word for "time" in Arabic, i.e., /:Iin-, with its verbal derivation /:lana, ya/:linu (cf. also /:Iayn- "fate"). That means: Arabic /:Iin- "time" is related to an- "moment, time" and should, like an-, ultimately be traced back to yawm- "day". The i-phoneme in /:Iin- has parallels in various conjunctional and prepositional particles discussed by us above; and as far as the i-vowel appears here lengthened, as i, it corresponds to the long vowel, a, in an-. Furthermore, the laryngeal /:I which precedes the vowel in /:Iin-, can be compared with the 'Ayn which precedes the vowel a in various particles derived from yawm(a) , as Soqotri 'am "at the time of"; Hebrew, Aramaic (also Ugaritic and Old-South-Arabic) 'i/am "with"; etc. (cf. above p. 401). A further derivative of yawm-, in this case again augmented by the fem.-ending (cf. above p.404), we recognize in the noun 'ona, which in Talmudic Hebrew is a current term for "time, period, season" (and related meanings. see dictionaries), and also occurs in Biblical Hebrew (Ex. 21, lO), in the meaning "marital intercourse", the concept of the "fixed time" here being used as a kind of concealing, euphemistic metaphor. (With respect to its specific formal characteristics: fem.-ending and shift of the original m to n [as assumed by us], 'ona reminds us of the terms of Soqotri and S~awri for the concept "year"; see below p. 408). We mention another etymon, which is Arabic as well as Ethiopicthus certainly having developed in Common-South-Semitic (pre-Arabic) time-and which we again consider based on an original yawm- "day" (or: yawma "on the day of (or: when)", or: "in the time of (or: when)"). We mean Arabic 'am-, Ethiopic 'amat and also-but more rarely used-'am, "year". With respect to the specific meaning "year", we refer to Greek &pa which is used in the meaning "hour" as well as "season" and also "year" (and apparently also in the meaning

406

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

"day"; see Lidell & Scott, Greek-Engl. diction., p. 2035 b; cf. also used in the sense of "year" (see ibid., p. 2037 b). (Very probably English year and its cognates belong to the same etymon; see W. W. Skeat, Etymol. diction. of the English language, 1963, p. 728). With respect to the above-discussed various Semitic nouns in which the original specific, quasi "narrow", meaning "day" has become the basis for various other (as it were, less restricted) concepts of "time", as "time" (generally), "period", "season", or "year" (but also "moment"), we may refer to idioms like the sometimes used English phrase "in this day and age", where "day" likewise functions in the meaning of "time", "period". We are indeed of the opinion that Arabic (and Ethiopic) 'am "year" is based on an original expression (al-) yawma "today" (i.e., "at this time"). With respect to this we may mention the actual synonymity, in certain contexts, of al-yawma (literally: "today", and also "at the present time") and al-'am- (literally "[in] the [present] year"). We refer to the following passage (dealing with the disputes between the partisans of 'Ali b. Abi Tiilib and the tIiirigites), in which al-yawma and al-'am- correspond with each other (al-Dinawari, al-Abbar a!{iwal, ed. Cairo 1960, p. 207, 18ff.): ... Tumma takallama AbU Ayyitbin binabwi haria. Faqalit: "Ya Aba Ayyitbin, inna in baya'nakumit I-yawma bakkamtum gadan abara". Qala: "Fa'inna nansudukumit llaha an tu'aggilit jitnata 1-'ami mabafata ma na'ti bihi fi qabilin" " ... Then Abu Ayyub spoke similarly [in favor of 'Ali]. They (i.e., the tIarigites) answered: '0 Abu Ayyub, if we pay homage to you (or: if we follow you) today (al-yawma), then you will appoint tomorrow (i.e., in the near future) another one as arbiter'. He replied: 'And we adjure you by God that you not hasten to bring about the "conflict" (or: the civil war) "in this year" (fitnata 1- 'ami, quasi literally: 'the "conflict" of this year') [that is: today, or: at the present time]out of fear of what we might bring about "in the next" (or: in the future [or: tomorrow D' ". The expressions al-yawma and al-'am- in this quotation clearly refer to the same concept. The term al-'amin this context clearly shows that the basis of 'am is yawm. The shift of "day" into "time" (that is: "period", "year", or "time [generally]") certainly took place in the adverbial use of the basic term al-yawm-, that is: al-yawma "today". This adverbial use is still implied in the genitive al-'ami in the expression fitnata 1-'ami (in the above-quoted passage). The "extension" (as it were: in a strictly "literal" meaning of this

&po~

EXPRESSIONS BASED ON THE NOUN YA WM- "DAY"

407

word-in German: "Streckung") of the meaning "day" into the meaning "period of time" (and-subsequently-"year")-a semantic development which evidently may be assumed on the basis of the correspondence between al-yawma (literally: "today") and al-'iim(literally: "[of] the [present] year") in the above-quoted passageis only a specific instance of a type of "extension of meaning" ("Streckung") to be observed generally in this type of "temporal" concepts. Our quotation above itself contains one of the analogous ("temporal") concepts used in a meaning characterised by the specific "extension" of meaning by which al-yawma ("today") in its use in the meaning "in the present [period of] time" has been affected. We mean the term which in the quotation appears as the contrast of al-yawma, that is: gadan, which means literally: "tomorrow", but possesses in the context of the quotation certainly a meaning like "in the [near] future". This expression (or any variation of it) has, of course, this latter (secondary) meaning also in certain other contexts. We quote here only the following verse by the pre-Islamic poet 'Antarah (ed. Ahlwardt, Six Divans), no. 8,3: falii takfur-i n-nu'mii wa'a!ni bifarjlihii wala ta'manan mii yu/:zdi!u lliihu fi gadi "Don't be ungrateful for the benefaction and praise its goodness-and don't feel safe from what God may cause to happen 'tomorrow''', i.e., "in the future". Moreover, also the other contrast-concept of al-yawma "today", that is: bil-amsi "yesterday", has undergone an analogous "extension" of meaning. The adverbial expression bil-amsi "yesterday" is also used in the meaning of "formerly", or: "a short time ago". We refer here to the following passage (al-Balaguri, Ansiib al-asriif, vol. i, ed. Mul;1ammad I:Iamidullah, Cairo 1959, p.299,6): aruway'iyanii bil-amsi (sic legendum, see our remark in Der Islam, vol. 42 [1966], p. 65) "0 thou [that hast been] formerly (literally: yesterday) a little shepherd of ours". Moreover, cf. e.g. (Ibn Sa'd, Tabaqiit, vol. 3, 1; p. 170, I-in a different context [the speaker is here Bil~l, the Prophet's companion and mu'adtjin]): innamii ana Ifabasiyun kuntu bil-amsi 'abdan "I am just an Abyssinian who was formerly (literally: yesterday) a slave". However, whereas in the case of the terms gadan (or: fi gadi) and bil-amsi, the secondary meaning (a meaning developed by the "extension" by which the basic meaning was affected) remained restricted to their original adverbial use (that is: to the terms gadanlfi gadi and bil-amsi), the "extension" of meaning by which al-yawma was affected, did not remain restricted to the (original) adverbial use of the term (i.e., to al-yawma), but the noun

408

EXPRESSIONS BASED ON THE NOUN Y A WM- "DAY"

(al-)yawm- itself-regardless of the grammatical function in which

it was used, that is: as an independent term-expressed now (in addition to its original meaning "day") the meaning of "time", "period of time" (> "year"), although in this new meaning the original term, yawm-, underwent, quite understandably, a very farreaching phonetical development, we may also say: a mutilation. The term yawm- developed into 'am- (and other, secondary, phonetic and morphological variations). In this context we should also mention the terms for "year" in the modern South-Arabian languages Soqotri and Sbawri, which are: Soqotri 'enoh, Sbawri 'anut. These words have been identified with Arabic 'am, Ethiopic 'amat (see M. Bittner, Vorstudien zur Soqotrisprache [Sitz.-Ber. d. Wiener Akad., phil.-hist. Kl., Bd. 186,4; 1918], p. 52). Since we deal in the case of these nouns of Soqotri and Sbawri with a feminine-pattern, they may be considered as having directly developed from a form corresponding to Ethiopic 'amat. These nouns of Soqotri and Sbawri for the concept of "year" are of specific interest to us, since they represent an additional clear instance of the shift of an intervocalic minto n, a shift which we assume to have occurred in various lexical terms which we trace back to proto-Semitic yawm"day" (see above p. 393 and p. 396). We furthermore mention an etymon which, though not directly based on the word yawm-, can, in our opinion, be considered as rooted in a secondary root developed from yawm-. We mean the Arabic root dnw "to approach", "to come close to", perf. dana, imperf. yadnu. We assume that the perfect dana is based on the "fifth stem" of the verbal root 'ny, i.e., ta'anna, which expresses the meaning "to do sth. slowly", "to take one's time" (apart from certain other meanings, see above p. 404; for our derivation of the root 'ny from a noun of the meaning "time" developed from yawm-, see ibid.). The concept of "approaching" frequently implies the connotation of "moving slowly towards (a certain goal)"; and thus ta'anna "to do sth. slowly" could have been used in the meaning of "to approach". The form ta'anna could easily have developed into *tanna; and this secondary root, which was interpreted by the speakers as the "intensive" conjugation of a root *tnw, led to the formation of a "basic" conjugation *tana (in agreement with the non-"intensive" meaning of the expression). (We refer here to 'Iraqi-Arabic tina "to wait"; cf. above p.404). In the imperfect associated with the root, i.e., *yatnu, the unvoiced t could in contact with the voiced consonant n become voiced: > yadnu (perf. : dana).

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

409

It seems to us that in the context of those word-forms which are derived from the Common-Semitic word for "day": yawm-, the origin of the Arabic preposition 'inda "beside, with, at" should also be dealt with. This preposition, 'inda, which is functionally related to, but not entirely identical with, the Arabic preposition ma'a (in its turn agreeing in function and meaning with Hebrew/Aramaic 'im), has frequently been compared with the prepositional stem appearing in Hebrew 'immagi "with me", "in my company", which is synonymously used with 'immi ("with me"). Indeed, it is a peculiar fact that the prepositional stem 'immad- appears only in connection with the genitive-suffix of the 1st pers. sing., -i. It seems certain to us that the Hebrew prepositional stem, 'immadwith a vowel before the consonant d, is more original than Arabic 'inda, in which the d is not preceded by such a vowel. We recognize in the consonant d in Arabic 'ind( a) and in the phonetic sequence -ad in Hebrew 'immag- ('immagi) a remnant of the Proto- and CommonSemitic word for "hand", i.e., yad. We compare the original expression on which Arabic 'inda and Hebrew 'immag- are based, with the Akkadian prepositional terms ina idi and ana idi. These prepositional expressions, based on idu ("arm" = West-Semitic yad "hand, arm") in its specific, secondary meaning "side", are used in the sense of "at the side of", "with", "at"; see CAD, vol. 7 ("I/J"), p. 13 b (sub 2'); and p. 14 (sub 3'). We refer also to the various prepositional expressions based on the noun yad in Hebrew and Aramaic, which are used synonymously with the afore-mentioned Akkadian expressions (ina idi, ana idi), as, e.g., Hebrew fl yag "by the side of" (e.g., 2 Sam. 18,4); leyag (1 Sam. 19,3); and also 'al yag (2Sam. 15,2). Cf. also the Arabic preposition lada "at, by" (in the forms with pronominal suffixes: laday-) which in turn is doubtless based on li-yad- (with the -a, -ay either due to the analogy of ita, itay-; 'ala, 'alay-; or based on the dual-endings: yada-, yaday-). As to Arabic 'inda and Hebrew 'im(m)ag- (in 'immagi), we assume that they are based on certain early-Semitic forms which were quite similar to (or even identical with) the original forms underlying the afore-mentioned prepositional expressions of Akkadian, ina idi, ana idi "beside" (see below p. 411). We thus recognize in the element 'im in Hebrew *'im(m)ag- ('immagi "with me") and in the element 'inin Arabic 'inda an original preposition closely related to the Akkadian prepositions ina and ana (see above p. 393 and p. 396, respectively), and to the preposition 'im of Hebrew, Aramaic, Ugaritic and Old-

410

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

South-Arabic (see above p. 401). The Hebrew prepositional expression 'im(m)arj-, based on the noun yad "hand" in its secondary meaning "side", still shows the consonant m which we regard as the basis of the consonant n in Akkadian ina and ana, forms which we, in analogy to certain other prepositions in the related dialects, consider ultimately derived from the Proto-Semitic form yawma (see above p. 391 and p. 394). Cf. also the Akkadian particle ema "when, where, at" (see above p. 398). As to the n in Arabic 'inda, it can certainly be explained as a partial assimilation of an original (labial) m to the contiguous dental d; but it could also represent an early Semitic preposition in( a), identical with the Akkadian preposition ina/ana (see above p. 391). The specific form of the Arabic preposition 'inda, which, in distinction from Hebrew 'immarj-, is characterized by the lack of the vowel a between the nasal n and the dental d, could easily have developed by the influence of other Arabic prepositions, as qabla, ba'da, fawqa, ta/:lta. Our comparison of the Hebrew prepositional stem 'im(m)arj- and its Arabic correspondence 'inda with the prepositional expressions of Akkadian ina idi, ana idi means that we consider 'im(m)arj- as an original compound consisting of a preposition *'im (or *'ima) "at" and the noun yad- "hand" used in the meaning "side". It seems to us that Akkadian itself still possesses a reflex of the prepositional expression ina (or: ana) idi in which the more original m (as in Hebrew 'im(m)arj-), instead of the n of ina (ana) ... , is preserved. We believe that such a more original form of the prepositional expression is reflected in the Akkadian verb emedu, which has correspondences (with initial 'Ayn) in Arabic and Hebrew. The preponderant meaning of the verbal root 'md in Hebrew is "to stand". The meaning of the root in Arabic ramada) is "to strive after, to move (or: advance) towards (a certain goal)", etc., a meaning which considerably differs from the characteristic meaning of the root in Hebrew, but can without difficulty be brought into full agreement with the principal meaning of this verbal root in Akkadian, a language in which this verbal root is frequently used. As for Akkadian, we refer to CAD, vol. 4 ("E"), p. 138: "emedu ... 1. to lean against, to reach, to cling to, to come in contact, to stand (near) by, to land (said of a boat), to take cover, refuge. 2. to place, lean (something upon or against something), to load, to impose ... ". These meanings of Akkadian emedu lead us to assume that the verbal root 'md has the same origin as the preposition Hebrew 'im(m)arj-,

EXPRESSIONS BASED ON THE NOUN YA WM- "DAY"

411

Arabic 'inda. Hence, we regard the verbal root of Akkadian, emedu and its West-Semitic parallels (see above) derived from an original prepositional expression like *erna idi, *, irn ( a) yad- (or a similar basic form), expressing the meaning "beside" ("at the side of"), synonymously used with-and constituting the approximate basis of -the actually existing prepositional expressions of Akkadian, ina idi and ana idi "beside". Accordingly we assume that the original prepositional expression of the meaning "beside" became "verbalized", acquiring the meaning "to lean against" ("to come in contact with") and governing its complement in the accusative (instead of in the genitive). To illustrate this development, we may, e.g., refer to the following passage (quoted in CAD, vol. 4 ("E"), p. 138b, sub. 1.): i-rnid igara luIJurnma iptasas "she (Lamastu) leaned against the wall and smeared it with soot". In view of our derivation of ernedu "to lean against" from *erna idi (or similar basic form, > ina idi) "beside", we may assume that a sentence like (arnelurn) ernid igaram "(a man) leaned against the wall" is derived from a primitive structure like (arnelurn) *ema idi (> ina idi) igarim "(the man) is (or: was) beside the wall". The West-Semitic preposition which is represented by Hebrew 'irnrnatj-, Arabic 'inda (= Akkadian ina/ana idi, see above) shows us that the characteristic Akkadian preposition ina and its variant ana (both based on *a/ima < yawrna) were originally in Common-Semitic, and certainly also in Proto-Semitic, use. That means: they were used in these early stages of linguistic development beside the primary Semitic prepositions bali and la/i. They subsequently disappeared from the West-Semitic languages in their function as regular equi-, valents of bali and tali in exactly the same way as these latter disappeared from Akkadian (see above p. 390). Among the rather numerous expressions derived from the ancient Common-Semitic term for "day", all of which are characterized by considerable phonetic alterations, we would also include a temporal conjunction current in Syriac, that is rna tje used in the meaning of "when" and also (especially frequently) in the meaning of "as soon as", a term ordinarily followed by the perfect (for instances see Brockelmann, Grundriss, II, §407 d; p. 616). We suggest to consider rna tje shortened from yawma tje "on the day when ... ". (Cf. also Syriac 'am de; see above p. 402). Such an abridgement of yawrn- to rn- (followed by a vowel) is probably also to be observed in another expression. We have in mind the

412

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

word for "tomorrow" in Hebrew and Aramaic: mii~iir and me~ar, respectively, if it is indeed to be interpreted as based on a combination of the word for "day?': yawm- and the word for "(an)other": *ii~ar (or: a~er?). This etymology of the word (see the dictionaries), which seems to us to be the only correct one, may be corroborated by a frequently used Syriac expression for the concept "on the next day", that is: [eyawmii ~renii, literally: "on the other day" (see, e.g., Brockelmann's Syrische Chrestomathie [appendix to his Syrische Grammatik, 9th ed., 1962], p. 16, line 5). Above p. 402, we mentioned the Akkadian adverb inanna, eninna, enanna, etc., "now, just now" (CAD, vol. 7 ("I-J"), p. 141 a-b). Brockelmann, Grundriss, I, §107 d (p. 317) sees in the second (doubled) n of this expression a demonstrative element. He recognizes this element in certain other Semitic adverbs (also outside Akkadian), and compares it also with the Akkadian demonstrative pronoun "this": annu (anniu), fern. annitu, pI. masc. annuti, pI. fern. anniiti. W. von Soden, ZA 41, p.99, in turn interprets the element anna in inanna "now" as a pronominal element, clearly identifying it with the demonstrative pronoun annu; and in GAG, § 119 c, he ascribes (in agreement with this interpretation of the n-element) to the adverb inanna "now" the literal meaning "in diesem; jetzt". We have proposed, in the present study (see above p.402ff.), a quite different interpretation of this adverb, comparing it with an adverb of identical meaning, anumma (anummi, anummu) "now" (but also "here"). Moreover, we referred to the local adverb anummis, annummis, anaummis "(over) there, thither". We now maintain that it might be possible -even on the basis of this derivation of the temporal adverb inannato find a connection between the element -anna in inanna and the "near" demonstrative of Akkadian, annu ("this"), and even-since we interpret inanna as developed from a form *inamma-to discover a connection between inanna (and anumma) and the "far" demonstrative (of Assyrian), ammu < ammiu (fern. ammitu, pI. masc. ammuti, pI. fern. ammiiti). The difference between the initial vowel of inanna "here" and the initial vowel of anumma, anummi, anummu "now, here" and anummis, annummis, anaummis "(over) there, thither", is insignificant, since we derive the prepositions ina and ana, with which the initial syllable of these adverbs should be identified (cf. especiaIIy anaummis), from the same primary form ( < *ama < yawma), and assume that there is no strict borderline between the more specific meanings in which· these prepositions are mostly used (see above

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

413

p. 399). As to the specific meanings of the various adverbs under consideration, we note that inanna means "now" (or also "just now"), and that anurnrna and its variants obviously combine the temporal meaning, "now", with a local meaning, "here"; whereas anurnrnis (also annurnrnis, and anaurnrnis) is interpreted as "(over) there, thither" (see CA D, vol. I ("A"), pt. 2, p. 149 a), but can, in spite of this local meaning, hardly be separatifd from anurnisu "then", "thereupon" (which should be derived from ana urnisu "then", literally: "at its day"; see below p. 422). Accordingly (as already outlined above p. 402), it should be assumed that the adverbs anurnrna and inanna are based on an original expression a/ina-urni-rna, and that this original form (consisting in a sequence of syllables of similar type) was contracted, on the one hand, to a/inurnrna, and on the other hand (with a different treatment of the original sequence of two vowels: -au- < a-u-) to *a/inarnrna. (Cf. the analogous origin of the adverb inurna ("now" and also "here is") from *ina urn i-rna, see above p.398). Moreover-as we already stated in our earlier discussion of the adverb inanna (see above p. 403) -the sequence -rnrn- could be shifted into -nn-: *inarnrna « *inaurnrna < *ina-urni-rna) > inanna (enanna). The forms *inarnrna and inanna (which latter developed from the former), and certain other variants which may have existed-as, e.g., (the actually existing) anurnrna < *ana-urnrna < *ana-urni-rna (cf. anaurnrnis)-may have been conceived (quasi analyzed) by certain speakers in a way which necessarily led these speakers to interpret the elements following the introductory preposition in( aj or an( aj, i.e., the elements -ann ( aj or -arnrn ( a j, respectively, as constituting demonstrative (deictic) elements of the meaning "this" or "that", so that the literal (and quasi-original) adverbial meaning of the expressions was understood to be: "at it", or: "at this", or: "that" (an interpretation preferable to "in it", or: "in this", or: "that"). We assume that such an interpretation ("analysis") by certain speakers led to the rise of two new deictic stems, ann-, and arnrn-, which ultimately, in the forms anniu (> annu) and arnrniu (> arnrnu)-forms obviously enlarged by the nomen-relationisending (cf. below p. 4l6)--acquired the status of demonstrative pronouns expressing the meanings "this" and "that", respectively. We may base this secondary interpretation (by certain speakers) of the above-mentioned adverbial expressions-a process which led to the development of two new demonstratives-on the temporal as well as on the local use of the adverbs concerned (i.e., on the

414

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

meaning "now/then" as well as on the meaning "here/there"), though the local use seems to be somewhat more adequate, and may be assumed to have occurred more frequently. In connection with the above derivation of the Akkadian demonstrative pronouns annu and ammu, we may refer to certain adverbial formations expressing the meaning "here" or "there" (and alsothough more rarely occurring-the meaning "now"), which are evidently formed by the combination of a local preposition expressing "rest at a certain place" and a demonstrative pronoun "this" or "that" (or a personal pronoun: "it", coming close in its meaning to a demonstrative pronoun: "this" or "that"). For details see below p.416. For the sake of completeness, we mention here the interpretation of the Akkadian "near" demonstrative anniu > annu which is suggested in An Introduction to the Comparative Grammar of the Semitic Languages, by S. Moscati and associates (Wiesbaden 1964), p. 112. The authors refer to an element n affixed to various "stems" used in the function of "near" demonstratives. They mention Bibl.-Aram. denii, Ethiopic zantu, etc., and then state: "n combines also with hii in the Akkadian anniu, etc., and in the Syriac hiin(ii)". They thus assume a special relationship (quasi an identity of the underlying primary forms) between Syriac hiin(ii) and Akkadian anniu. Such a direct relationship (or any relationship whatever) between these two demonstrative pronouns cannot be assumed, since Syriac hiinii (masc.) is to be considered based on an original form *hiir,rnii. Cf. its fern. hiige and its plural hiillen < *hii-illen; cf. also the interrogative pronoun aynii < *ay-denii (beside aydii [fern.] and aylen [pl.] < *ay-illen); furthermore cf. the interrogative pronoun miinii < *mii-r,rnii (beside miin [which is either abridged from miinii or based on mii-den]). We may also mention Syriac yawmiinii (also yawmiin) "today", interpreted by Brockelmann, Grundriss, I, p. 317, as developed from yawm(ii) augmented by a "demonstrativeJ ' n, an element which he identifies with the Akkadian demonstrative pronoun here under consideration: annu (see for details ibid.). However, also yawmiin( ii) should be considered originating in a primary term containing the regular demonstrative pronoun, that is: *yawmii genii "this day" (> yawmiinii). (The variation yawmiin [also appearing in ludaeo-Aramaic] may be assumed as being abridged from yawmiinii or as originating in *yawmii gen [> yawmiin]; cf. miin "what?", beside miinii, see above). Accordingly, in view of the development of Syriac hiinii from

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

415

*hii-rrnii, a derivation of Akkadian anniu > annu analogous to that of Syriac hiinii (as assumed by Moscati and his associates, see above) does not come into question. Moreover, the structural parallelism between anniu > annu ("near" demonstrative) and ammiu > ammu (the "far" demonstrative of Assyrian) is unmistakable and must be taken into consideration in the developmental analysis of the two demonstrative stems on which these pronouns are based. Both these demonstrative (deictic) stems (amm- and ann-) should be considered as "abstracted" from certain adverbial expressions which by the speakers were conceived (quasi subconsciously analysed) in the sense of "at this" and "at that", respectively (see above p. 4l3). The temporal and local adverbs discussed above (p. 412ff.), which we consider based on a combination of the preposition a/ina and the noun yawm- > um- (or: *am- > *an-), could, in our opinion, even in their primary and complete form acquire the meaning and function of a demonstrative pronoun. That is: they could acquire this meaning and function without being dissected by the speakers into a local/ temporal preposition (ina or ana) and a second element which was felt to constitute a demonstrative pronoun (as described above). With respect to this we refer to anummu (fern. annummitu) "this, the aforementioned" (see CAD, vol. I ("A"), pt. 2, p. 149 a-b), which should be compared with adverbs like anumma, anummi, anummu "now, here" (see Ibid., p. 147 a-b). Moreover, we may mention the demonstrative pronoun animmu (annimmu; fern. animmitu) "this, the one in question" (see Ibid., p. 122 a). Concerning the origin of demonstrative pronouns like these, we assume that the original adverb anumma "now/here" ( < *ana-umi-ma, cf. above p. 402) was understood by certain speakers as expressing (or implying) the meaning of a demonstrative pronoun of the meaning "this"-i.e., "this [point] in space (or: in time)"and that this seeming demonstrative pronoun, originally referring to a definite point in space or time, was subsequently used as an actual demonstrative pronoun, no longer restricted to a specific local or temporal function (function of a "local/temporal adverb"). Indeed, these demonstrative elements of the meaning "this", "the afore-mentioned" -i.e., anummu, animmu (fern. anummitu, animmitu) -are, on the one hand, clearly related to certain adverbs (discussed above p.412) which are based on the element an-, an element to be identified with the preposition ana. But on the other hand, it seems impossible to separate anummu, animmu from the "far" demonstrative [of Assyrian) ammu « ammiu), from which in tum the "near"

416

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

demonstrative (of Akkadian generally) anna « anniu) cannot be separated with respect to origin and function. We also should stress a special feature of the demonstrative pronouns anumma, animma (forms which evidently originate in certain ("local/temporal") demonstrative adverbs). These demonstrative pronouns are characterised by the "nomen relationis" (nisbah )-ending (and the nominal inflection which goes herewith), exactly as this is the case with the demonstrative pronouns amma and anna ( < ammiu and anniu, respectively), which are of a related origin (see above p.4l3). This affixation of the "nomen re1ationis"-ending serves as a means of stressing the pronominal, "quasi 'nominal''', function of both type-variations. We maintained (see above p. 413) that the demonstrative pronouns amma and anna are based on original adverbs of the type *inamma, *anamma (anumma), inanna, adverbial expressions containing the prepositions ina and ana (respectively) and expressing the meaning of "here/there" and "now/then" (see above p. 403 and p. 413). With respect to the development of the demonstrative pronouns amma, anna, from such adverbial basic forms, we assumed that the element following the initial i{/a)na (functioning as a local preposition) was interpreted by certain speakers to mean "this" or "that", the adverb as a whole, as it were, implying the meaning "at this", or "at that", with i/ana expressing the concept "at" (see below). With respect to this assumed interpretation (by the speakers) as basis of the development of the demonstratives of the type amma, anna, we may refer to a number of instances where the concepts "here/there" and "now/ then" are indeed expressed by word combinations which originally -thus not through secondary interpretation by the speakers-mean "at this" or: "at that" (or: "at it"). We mention Hebrew biizf "here" (literally: "at this" [preferable to "in this"]), Gen. 30,21,22; Ex. 24,14; also used of time: "then", Ex. 2,13; 1 Sam. 21,10 (bazzf); mizzf ... mizzf "on this side ... on that side" (more literally: "from here ... "), Num. 22,24; etc. The same type of expression for these local and temporal adverbs is also found in the modern Semitic languages of Ethiopia, where the concepts "here", "there", "hither", "thither", "hence", and certain corresponding temporal concepts are expressed by the combination of a preposition of the meaning "in, at", or "towards", or "(away) from", respectively, with a demonstrative pronoun, "this" or "that", or a personal pronoun (pronominal suffix) "it". We mention Amharic

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

417

bazzih, "here", kazzih "from here", or "from now on", askayah "up to here", bazzya "there", kazzya "from there"; etc. (see M. Cohen, Traite de langue amharique, Paris 1936, p. 113). We also refer to Tigrina 'ab'u "there" (literally: "at it"), 'abzuy "here" (literally: "at this"), 'abtuy "there" (literally: "at that"), and similar combinations formed with the prepositions expressing the concepts "from" and "towards" (see W. Leslau, Documents tigrigna, Paris 1941, p. 136). Furthermore, we mention the expression b(j "there" (thus an expression totally identical with Hebrew biizf, see above) which is found in the early (N orth-)Arabic dialect called Iamiidic, known to us from epigraphical monuments. In this context it seems also of interest to note that expressions of the literal meaning "at this" or "at that" used in the sense of "here" and "there"; respectively, exist, for instance, in Telugu (a Dravidian language); see, e.g., the quotations (and their translations) given by Peri Bhaskararao in Foundations of Language, supplem. series, vol. 14 (The verb 'be' and its synonyms, part 5), p. 156. In our study "The Simultaneous Emergence in the Mind of Two Distinct (Interrelated) Aspects of a Notion", published in our Studies in Arabic and General Syntax (Cairo 1953), p. 139-150, we have shown that this manner of expressing the concept "there" (or "here") by a combination of the local preposition ba ("at") with a pronominal suffix ("it") may be assumed to have existed in an early stage of Akkadian as well as Ancient Ethiopic (G;}·;}z). Delitzsch, Wo lag das Paradies? (Leipzig 1881), p. 166, identified the Akkadian expression for the concept of "existence" ("there is", "there exists"), i.e., basu, with the synonymous expression of Ethiopic, bo (also, with a usual enlargement of this type of formation, botu; with negation: alba "there is not"), recognizing in these expressions a combination of the preposition bali with the pronoun (pronominal suffix) of the 3rd pers. sing. masc. This formal analysis of the terms under discussion is doubtless correct. As to the semantic interpretation of this type of expression, Delitzsch assumed that the original meaning of basu (= Ethiopic bO) is "in ihm ist", oder: "besteht das und das", "an ihm ist das und das" = "er ist (oder: er hat) das und das" (basing this interpretation on certain Hebrew phrases). We are unable to accept this interpretation of the development of these terms which express the idea of "existence", that is: the concept "there is" ("there exists"). In our afore-mentioned study (/.c., p. 140) we compared these

418

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

expressions of Akkadian and Ethiopic with the widespread modernArabic colloquial term for the concept "there is", i.e., fih, or bih, bo(h). We stated (see I.c.) that the meaning "there is" of these expressions of Akkadian, Ethiopic and the modern Arabic dialects should be considered as based on a more original, but apparently (as far as these languages are concerned) unrecorded use of these terms in the function of a local adverb, "there" (or: "here"), identical with the use of the correspondingly formed Hebrew adverb bazzf (see above p.416). We drew attention to the fact that a great many languages express the concept of "existence" by such an adverb or by expressions based on such an adverb, and also discussed the semantic mechanism underlying this use of the local adverb "there" or "here" for the indication of the idea of "existence" ("there is"). Accordingly, if Akkadian basu "there exists" means originally "there" (or: "here") < "at this", or "at it", it illuminates that type of expression which we consider as the basis of the development of the Akkadian demonstrative pronouns annu and ammu (see above p. 413ff.). As to the Ethiopic use of the preposition baji with variable pronominal suffixes for expressing the concept of "possession" ("to have"), that is: bo(tu) "he has", baya "I have", banii "we have", bomu "they have", etc., a usage which has been considered as more original than the use of the term bo for expressing the concept of "existence": "there is" (see for references our Studies in Arabic and General Syntax, p. 139), it should be recognized that this use of the preposition baji with variable suffixes for the concept of "to have" constitutes a secondary development of bO (with the original "invariable" suffix) in the meaning of "there is" (see for the details of this development Ibid., p. 144). Our detailed discussion concerning the conjunctional, prepositional and adverbial usages of certain particles based on Proto-Semitic yawm"day" has given us an opportunity to mention (see above p. 399, and p. 403) the temporal adverbs of Akkadian: inumisu (enumisu, in umisum) "at that time" (see CAD, vol. 7 ("1-1"), p. 146 b) and anumisu "then, thereupon" (see Ibid., vol. 1 ("A"), pt. 2, p. 146 a), as well as the local adverb anummis, annummis, anaummis "(over) there, thither" (see Ibid., p. 149). We hinted (see above p.403) that there exists a genetic relationship between the latter local adverb and the temporal adverb anumisu, based on the close relationship between local and temporal prepositions and adverbs. We consider the identification (suggested in CAD, vol. 7 ("1-1"), p. 153 b) of the elements

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

419

in- and an- in inumisu and anumisu, respectively, with deictic elements, in (or en) and an, unacceptable (cf. our remark in Le Museon, vol.83 (1970), p. 248). We assume (see also above p.399) that the temporal adverbs inumisu and anumisu include the preposition ina and ana, respectively, and the noun um- "day". Accordingly, if we connect the above-mentioned local adverb anummis, annummis, anaummis with the temporal adverb anumisu, we assume that the temporal adverb is original and that the local adverb is based on secondary development. An important question, however, is whether there exists indeed a relationship between the ending -is in the local adverb just mentioned and the ending (i)-su in the temporal adverbs anumisu and inumisu, and how such a relationship should be described. This question leads, of course, to the question of the origin of the adverbial-ending -is of Akkadian in general. F. Rundgren, Uber Bildungen mit s- und n-t- Demonstrativen im Semitischen, p. 157, deals very thoroughly with the origin of the adverbial-ending -is (called by him, in agreement with von Soden's terminology, "Terminativendung"). In discussing various interpretations of the ending -is (see especially I.e., p. 160ff.), he mentions 1. Halevy's interpretation of, e.g., "tiibis 'in guter Weise'" as being based on "tiibisu 'gut fUr ihn"'. Rundgren notes that he considers Halevy's interpretation as basically correct, "since [the adverbial ending] -is may be connected with the adverbial use of the demonstrative su" ("weil namlich -is mit der adverbialen Verwendung des demonstrativischen SU zusammenhangen durfte"). Rundgren (l.c., p. 161) mentions numerous instances of this allegedly "adverbial use of the 'demonstrative' su": "umussu 'taglich' < *umutsu; "arl;usu 'monatlich' < *arl;ut-su"; furthermore "inumi-su 'damals''', inumi-su-ma 'in diesen Tagen, damals''', "inu-su 'damals''', "istu inumi-su 'von da an'''; moreover "arki-si < arki-su, vgl. urkissi, 'spater, hernach''', "adi-su 'bis dahin''', "ina/ana mubbi-su 'daruber'''; etc. After comparing numerous other seemingly related formations of Akkadian and other Semitic languages, Rundgren once more raises the question of the original meaning and function of the adverbial ending -is. For the sake of clarity we shall q~ote a larger section of Rundgren's conclusions (l.c., p. 165): "Dies weist alles in diesel be Richtung hin, dass namlich die Terminativendung als eine Bildung auf dem demonstrativen SU zu betrachten ist. Es ist aber nicht leicht, genau den syntaktischen Ursprung dieser Bildung festzustellen. Vielleicht geht tiibis 'bene' eigentlich auf einen Satz tiib su 'gut dies',

420

EXPRESSIONS BASED ON THE NOUN YA WM- "DAY"

'indem es gut ist, in guter Weise> gut' zuruck. Moglich ist auch ein lab-Hi, des sen sit das auf das Subjekt hinweisende Suffix darsteIlt, 'sein gut' z.B. *epus lab-sit 'er machte sein Gutes', 'bene facit'. Dann erstarrte sit und die Kongruenz ward vergessen, wie in bar yawmeh, so dass sit als Adverbendung aufgefasst werden konnte. Wie es sich auch immer mit dem syntaktischen Ursprung verhaIten mag, -is Hisst sich jedenfalls nicht gut von dem adverbiellen SU trennen. Eine Stutze fUr diese Annahme finden wir im Westsemitischen. Hier entspricht ja hU (natiirlich nicht etymologisch) einem suo Die semitischen Sprachen soIlen, wie haufig angenommen wird, eine Lokativendung -u gekannt haben. Sie soIl in arab. ta/:ltu 'unterhalb', qablu 'vor' und im G;)';)Z la'(aJlit 'oberhalb', qadimu 'zuerst, fruher, einst', ta/:ltit 'unterhalb' usw. stecken (Brockelmann, Grundriss, I, S. 460 ff.). Nach Barth stellte dieses it den Nominativ einer ursprunglich diptotischen Deklination dar (ZDMG 46, S.691). Es sei aber bemerkt, dass eben in den athiopischen Sprachen die Suffixe, vor allem hit und ha, auch eine determinierende Funktion, und zwar als Artikel erhaIten haben. M.E. ist die Adverbialbildung mit u von dieser Funktion aus zu beurteilen ... . So haben wir es wohl doch in diesen westsemitischen Adverbien auf u mit dem alten Demonstrativ (und Suffix) hu zu tun -wie wohl auch im akk[adischen] Lok. -it, -u neben -um . ... ". In the continuation Rundgren refers also to G;)';)Z 'araqu "naked" (with the suffix of the 3rd pers. sing. masc.; and with other suffixes, dependent on the grammatical person to which the concept "naked" applies). He includes the so-called gerundium of Ethiopic, a verbal noun in the accusative, characterized by variable personal suffixes, as, e.g., nawimo, in which he recognizes an original sentence, "*nawim hu 'schlafend (ist) er"', used as a circumstantial complement (in the accusative) to a non-subordinate sentence, as "*hallawa nawima-hu 'er war, indem schlafend er (war), > nawimo". He then concludes (l.c., p. 168, at the bottom): "Da sich aber Gerundium und Adverb einander ziemlich nahe kommen, konnte sich ein tab-sit 'gut-er' > 'indem er, es gut (ist)' (als Bestimmung zum Verbum) zu labis 'bene' entwickeln" . Neither can we accept Rundgren's derivation of tabis "bene" from the artificial syntactic structure "lab-sit 'good-he' > 'while he-or: it-(is) good' (as complement of the verb)", nor can we approve his derivation of labis from a sentence like "*epus !ab-su 'bene facit' (more literally: 'er machte sein Gutes')". We also cannot agree with Rundgren's interpretation of the ad-

EXPRESSIONS BASED ON THE NOUN Y A WM- "DAY"

421

verbial ending -u, -u in Arabic and Ethiopic as corresponding to the Ethiopic suffixes -hu and -hii in their "determinative, article-like function". In our opinion, the Arabic adverbs qablu, ba'du, fawqu, tal:ztu, etc., should be derived from the basic forms qablahu, ba'dahu, fawqahu, tal:ztahu, etc. (The corresponding Ethiopic forms, as tal:ztu, lii'aiu, etc., constitute, of course, identical structures). As for the basic (original) meanings of these expressions, we assume them to have been: "before it", that is: "before [the occurrence oj] a certain matter just mentioned"; "after it", that is: "after the matter just mentioned"; "over (or: under) it", that is: "over (or: under) a certain ( other) thing (or: locality)", respectively. However, it should be recognized that the pronoun appearing in these expressions in suffixal (genitival) form lost its original, specific meaning, a fact which also caused the loss of the original prepositional function of the words qabl(a), fawq(a}, tal:zt(a} , etc. The secondarily developed particles qablu, ba'du, fawqu, tal:ztu, etc., are of a uniform, quasi non-composite character, in distinction from the primary formations on which they are based: qablahu, etc., which were felt as composed of a preposition and a pronoun (a term representing a noun). What developed are adverbs of the meaning "before", "afterwards", "above", "below", etc. In these secondary expressions, the exact position, with respect to time or space, of a certain event or a certain concrete object, is not, as in the primary expressions (qablahu, etc.), indicated in its exact relationship to another event or object, but is only indicated in a general, vague manner: "later" (ba'du), "earlier" (qablu), "above" (fawqu), "below" (tal:ztu). Of course, in actual reality (to be distinguished from the linguistic expression), concepts like "later", "earlier" ("before"), "above", "below" always refer to a specific event or place; however, this event or place, in these expressions, is not (or no longer) exactly indicated, but is implied in the local or temporal relation ("later", "earlier", "above", "below") of which it forms the basis. (Cf. also below p. 545 concerning an analogous origin of Arabic gayru in usages like laysa gayru "not otherwise", etc.). A phenomenon of the same type as the above-mentioned adverbial expressions with an original pronominal suffix of the 3rd pers. (sing.) masc., we have to recognize in certain adverbial expressions current in a number of modern-Arabic dialects, which are constituted by a noun of "time" (or a preposition, as 'inda) plus an original pronominal suffix of the 3rd pers. (sing.) fem., as (for instance) recorded by M. Feghali, Syntaxe des par/ers arabes actuels du Liban (Paris 1928),

422

EXPRESSIONS BASED ON THE NOUN YAWM- "DAY"

p. 466: wtiqta, sa'era, yawmera, 'enda, etc. "a ce moment-Ia; alors" (Hghali names this pronoun: "Ie pro nom neutre -a"). An identical use of 'indahii, in the meaning "then" (= Lebanese-Arabic 'enda), we recognize also in Classical-Arabic. We refer, e.g., to Tabari, Annales, I, p. 2337,3-5: ... lumma niidii qataltu Rustuma wa-rabbi l-Ka'bati ilayya, ja'aliiju biM wamii yul;tissuna s-sarira walii yarawnahu wakabbaru wataniidaw wanballa qalbu l"musrikina 'indahii wanhazamu ... (,indahii is to be rendered here by "then" or: "at that moment"). Akkadian inumi§U, or anumisu, "then" (see above p. 419) is to be viewed in the same manner as the afore-mentioned expressions. The primary term ina umisu (or ana umisu) means literally "in its day", i.e., "in the day of a certain, already known event", or: "of an event just mentioned". This term was bound to develop into an expression concerning which one was no longer conscious of the original meaning of the pronominal suffix, or of the composite character of the term altogether, a fact which finds its expression in the usual (quite correct) interpretation of the term by "then". It is conceivable that the terminal vowel of such a composite form (more exactly, the vowel of the original pronominal suffix) could easily disappear. Such a form with its terminal vowel apocopated, i.e., *anumis < an( a) umisu, may be assumed to be the basis of the form anummis, annummis, anaummis "(over) there, thither", a local adverb which (see above p.419) we consider based on the original temporal adverb an( a) umisu "then" (literally: "at its day"; see above p.403). A similar development is recognizable in Akkadian elis "above", which we consider derived from eli-su, literally: "over it" (cf. Arabic jawqu aslama), for which I assume the same basic meaning (see Spiritual Background, p. 13). Another important and frequently occurring phrase for the concept of "abandoning (giving up) one's life in battle" (= "defiance of death", "self-sacrifice"), which has been compared by me (Spiritual Background, p. 20 ff.) with aslama nafsahu, is sara nafsahu, literally: "he sold himself' (Le., "he 'sold' his life", a metaphoric expression for "he 'betrayed' (or: abandoned) himself (by sacrificing himself)"; see my earlier discussion, l.c.). I now wish to draw attention to the interesting fact that sara in this specific usage can also be followed by the object waghahu, instead of the more frequently used object nafsahu. The phrase occurs, e.g., in a speech reported to have been made by 'Ali: b. AbI Talib during the battle of Siffin (year 37/659): ... ayna ahlu !j-$abri wa-!ullabu l-I;ayri? ayna man yasri waghahu lillahi 'azza wagalla? falabat ilayhi 'i$abatun min-a /-Muslimina ... (Na~r b. Muzal;im al-MinqarI, Waq'at Siffin, ed. 'Abd-as-SaIam Mul;ammad Hartin, Cairo 1382A.H., p. 392,4) " ... 'Where are the men of endurance and the seekers of virtue? Where are those who will sell their faces (i.e., are ready to abandon their lives [by fighting heroically]) for the sake of God?' .... And there gathered unto him (Le., to 'Ali:) a large group of Muslims (in obeyance to his summons) ... ". That wagh "face", when used in a phrase expressing the idea of "defiance of death" ("self-sacrifice"), represents quasi a synonym of nafs "soul" or &ayat"life", also becomes strikingly clear from reports concerning an earlier event, namely the battle of Ul;ud (al-WaqidI, Kitab al-Magazi,

436

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

ed. v. Kremer, Calcutta 1856, p. 238,9; ed. M. Jones, London 1966, p. 240-241): ... wa/:laddalani 'Utbatu bnu Gabirata 'an Ya'quba bni 'Amri bni Qatadata, qala: labuta bayna yadayhi yawma'igin lalaluna ragulan kulluhum yaqulu: waghi duna waghika wanafsi duna nafsika, wa'alayka salamun gayra muwadda'in ... waqalu: inna Rasula-llahi (~l'm) lamma la/:lamahu l-qitalu wabuli~a ilayhi wagabba 'anhu Mu~'abu bnu 'Umayrin wa-AbU Duganata /:latta kalurat bihi l-gira/:latu, ga'ala Rasulu-llahi (~l'm) yaqulu: man ragulun yasri nafsahu? fawalaba F atun min-a l-An~ari bamsatun, minhum 'Umaratu bnu Ziyadi bni s-Sakani, faqatala /:latta ulbita, wafa'at Fatun min-a l-Muslimina faqatalu /:latta aghat;lu a'da'a llahi ... " ... 'Utbah b. Gabirah related to me, on the authority of Ya'qlib b. 'Amr b. Qatadah, as follows: On that day there rallied to him (i.e., to the Prophet) thirty men, all of whom said: 'My face in defence of your face, and my soul in defence of your soul (i.e., I give my life in defence of your life). And peace upon you, who is never to be taken leave of (i.e., who is never to depart from us through death on the battle-field)'. - And they related: The Prophet-when the fighting threatened him and the enemies came close to him, while Mu~'ab b. 'Umayr and Abli Duganah defended him, until his wounds increased-[then] the Prophet said: 'Who will sell his soul?'. Thereupon a group of five men of the An~ar jumped to him, among them 'Umarah b. Ziyad b. as-Sakan; and he (i.e., 'Umarah) fought until he was exhausted; then a group of Muslims gathered and fought until they overcame the enemies of God ... ". The two above-quoted passages closely resemble each other. In both passages, the leader-in the first one: 'Ali b. Abi Talib, in the second one: the Prophet-incites his men to fight unreservedly and heroically; and in both instances the Muslims obey the summons of their leader and rally around him (Jabat ilayhi andfa'at, respectively). Moreover, the actual summons to defy death in both passages is expressed by variations of one and the same phrase. In the utterance ascribed to the Prophet (as transmitted in al-Waqidi's Kitab aIM agazi) , the summons is expressed by the phrase man ragulun yasri nafsahu "Who is prepared to 'sell his sou!'?"; and in 'Ali's speech (as transmitted in Na~r b. Muzal).im's Waq'at Siffin), it is expressed by man yasri waghahu lillahi "Who is prepared to 'sell his "face'" for the sake of God?". But of special interest is the fact that the Muslims respond to the summons of the Prophet with the words (each one speaking for himself individually): waghi duna waghika wanafsi duna nafsika "My face in defence of your face, and my soul

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

437

in defence of your soul!" (for this utterance, although in Wiiqidi's report preceding the Prophet's speech, is to be taken as a response to his speech). Accordingly, it is clear that wagh "face" serves as an expression for "man's self" (or: "man's life") in the context of the idea of "surrendering oneself (or: one's life) to death", that is: in the context of the idea of "self-sacrifice in battle". As already mentioned above (p.435), I assume this particular use of wagh "face" to be also present in the Qur'iinic phrase as lama waghahu lilliiM (cf. Spiritual Background, p. 23), identifying this phrase, on the one hand, with 'Ali b. Abi: Tiilib's phrase man yasri waghahu lilliihi (see above), and, on the other hand, with the more current phrase aslama nafsahu, or (abbreviated) aslama, the original meaning of which ~on the basis of a number of texts~ I assume to be: "he surrendered (or: 'betrayed')~i.e., abandoned~his soul", i.e., "he abandoned (or: gave up) his life, by heroically fighting in battle". I referred, e.g., (see Spiritual Background, p. 13), to Ka'b b. Miilik's line concerning the Muslims killed in the battle at "the Ditch", yawm al- /jandaq (Ibn Hisiim, Sirat Rasuli-lliih, p. 705, 10): fadaribU bif,/arbi l-mu'liminafa'aslamu muhgiiti anfusihim lirabbi l-masriqi "they exerted themselves in striking the heroes and sacrificed (more literally: surrendered, abandoned ['betrayed']) 'the blood of their souls' for the Lord of the Universe"; i.e., "they sacrificed their lifes by fighting heroically in the gihiid, the fight for the Islamic religion". In this passage not only the original full construction of the verb as lama ~i.e., its being followed by an object ("he surrendered [sacrificed] his soul"~or: "they surrendered [sacrificed] their souls") is preserved, but at the same time the specific Islamic usage of the concept is clearly and unambiguously indicated by the complement lirabbi l-masriqi, which is equivalent with lilliihi: "they sacrificed their souls (i.e., their lifes) for the sake of God". Accordingly, this line by Ka'b b. Miilik, one of Mul}ammad's companions, throws light upon the origin of the specific (pregnant) usage of the verb aslama (and the nouns isliim and muslim). The sense of the phrase fa'aslamu muhgiiti anfusihim ... , as well as the semantic identity of the verbs as lama and sarii (literally: "he sold", i.e., "he gave away", "he betrayed") in contexts of this type, is further clarified by the following passage referring to I:Iusayn b. 'AU b. Abi: Tiilib (Tabari:, Annates, 11,1; p. 341): ( ... 'Amru bnu Qarii?ata l-An~iiriyu yuqiitilu duna Ifusaynin wahwa yaqulu:) (1) qad 'alimat

438

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

katibatu I-An$iiri * anni sa'a~mi ~awzata rj-rjimiiri (2) r!arba guliimin gayri niksin siiri * duna lfusaynin muhgati wadiiri "( ... 'Amr b. Qar~ah al-An~art was fighting in defence of I:Iusayn while saying:) (1) The squadron of the An~ar knows that I will guard (from encroachment) the 'territory' of the values most dearest to a man,' (2) by striking [with my sword] as a young man who is no coward strikes, and 'selling' (i.e., 'betraying', sacrificing) my blood and my house in defence of I:Iusayn". (These lines, referring to I:Iusayn b. 'Ali, are, of course, of the same specific type as the two passages quoted above, p. 435, referring to the Prophet and I:Iusayn's father, 'Ali b. Abi ralib, respectively). Furthermore, in connection with Ka'b b. Malik's phrase ... fa'aslamu muhgiiti alifusihim lirabbi I-masriqi (see above) we may also quote the sentence by which a later hero ('Ali's partisan 'Abdallah b. I:Iasim) admonishes his companions (Na~r b. Muzal).im al-Minqart, Waq'at Siffin, p. 356, ult.): ... fagudu bimuhagi anfusikum fi !ii'ati lliihi fi hiirjihi d-dunyii tu$ibU I-iibirata wal-manzila I-a 'Iii. Ka'b b. Malik's fa'aslamu muhgiiti anfusihim lirabbi I-masriqi "and they sacrificed the blood of their souls for the Lord of the Universe" may-with respect to the basic idea expressed in it-be considered a variation of 'Abdallah b. I:Iasim's sentence fagudu bimuhagi anfusikum fi !ii'ati lliihi "Give up (or: sacrifice) the blood of your souls in obedience to God" (i.e., for the sake of God)!". As already stressed, my observation that there exists a phrase man yasri waghahu lilliihi (clearly equivalent with the phrase man yasri nafsahit ... ), represents an important additional argument that the Qur'anic phrase aslama waghahu lilliihi (see above p.435 and p. 437) should be interpreted in the sense of "he has given up himself (or: sacrificed himself) [by fighting heroically] for the sake of God" (or: "is prepared to sacrifice himself for the sake of God"). This interpretation of the phrase aslama waghahu lilliihi is, of course, essentially different from its usual interpretation by "he surrendered (himself) to God" (or: "he resigned himself to the will of God"). In this latter interpretation of the phrase, aslama waghahu is followed by a syntactically quasi indispensable complement in the dative. In the interpretation presented here, the phrase aslama nafsahu or waghahu-expressing the meaning: "he gave himself up [to death, by fighting heroically]" -is a self-contained expression, and lilliihi ("for the sake of God") represents an additional, syntactically quasi dispensable, adverbial complement. With respect to the Qur'anic phrase aslama waghahu lilliihi, which

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

439

may be considered as a variant of aslama nafsahu lillahi, we may refer also to the following early-Islamic lines (by Zayd b. 'Amr b. Nufayl, apud Ibn Hisam, Sirat RasUli-llah, ed. Wustenfeld, p. 149,6-9): (6) aslamtu waghi liman aslamat lahu l-arrju ta&milu $abran liqala (7) da&aha falamma ra'aha stawat 'ala l-ma'i arsa 'alayha l-gibala (8) wa'aslamtu waghi liman aslamat lahu l-muznu ta&milu 'Q(jban zulala (9) hja hiya siqat ila baldatin a!a'at fa$abbat 'alayha sigala. G. Guillaume (in !lis Life of Mu&ammad, by Ibn Is&aq, p. 102) translates: "(6) I submit myself to Him to whom the earth which bears mighty rocks is subject. (7) He spread it out and when He saw it was settled upon the waters, He fixed the mountains on it. (8) I submit myself to Him to whom clouds which bear sweet waters are subject. (9) When they are borne along to a land they obediently pour copious rain upon it". This English interpretation agrees with Gustav Weil's earlier German translation (vol. I, p. Ill): "Ich unterwerfe mich dem, welchem sich die Erde unterwirft, welche schwere Lasten tragt, ... Ich unterwerfe mich dem, welchem sich die Wolken unterwerfen, wenn sie susses Wasser tragen ... ". In analogy to these translations, H. Ringgren, in his study Islam, aslama and muslim (Horae Soederblomianae [Uppsala], II, 1949), p. 23, renders the phrases concerned by: "I have surrendered my face to Him to whom the earth ... surrenders herself", etc. These three translations of aslamtu waghi lahu ... and aslamat lahu ... are obviously based on the traditional interpretation of aslama (and islam), in its terminological (religious) use, in the meaning of "to be resigned (and 'resignation', respectively) to the will of God". The context in which the expressions aslamtu (waghi) li and aslamat li (respectively) appear in the above-quoted lines, makes it clear that the verb aslama refers here to an active (heroic) human attitude: "Man gives himself up (intentionally)", more literally: "gives up, sacrifices-and also: is ready to sacrificehis 'face' (i.e., his life) [by fighting heroically in battle]-for the sake of God"; and, analogously, also "the earth and the clouds" -here appearing as quasi-animated entities- "sacrifice themselves (through quasi heroic activity)-for the sake of God" (cf. below p.444). In emphasizing the importance in early Islam of "the fight for the sake of God", in other words: of "defiance of death, self-sacrifice ('abandoning' one's life, i.e., islam) for the sake of God", I now refer to one more passage, to words ascribed to Abu 'Arfa', a partisan of 'Ali in the battle of ~iffin (Na~r b. MuzaiJim, Waq'at Siffin, p. 305,

440

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

3 ff.): ... ya ahla harjihi r-rayati: inna 'amala l-gannati karhun kulluhu walaqilun, wa'inna 'amala n-nari biffun kulluhu wababibun, wa'inna l-gannata la yadbuluha illa $-$abiruna, allarjina $abaru anfusahum 'ala fara'itji .llahi wa'amrihi; walaysa say'un mimma ftaratja llahu 'ala 1'ibadi asaddu mina l-gihadi, huwa afdalu I-a 'mali lawaban, fa'irja ra'aytumuni qad sadadtu fasuddu "0 people of this standard! The work that leads to Paradise, is all repulsive and hard; and the work that leads to Hell, is all easiness and beloved. But Paradise only those enter that are enduring, those who have 'tied their souls fast' on (i.e., are enduring with respect to) carrying out the precepts of God and His command. But there is nothing of those things which God has imposed on his servants, that is harder than fighting; and it is also the most excellent thing with respect to reward. ... And when you see that I have attacked, so attack! ... ". Not only is here (as in many other traditions) the gihiid "the (holy) war" -with which the concept islam should be compared-declared the most important work among the works imposed by God on His servants, but it is, in addition, interesting to note that the dedication to the precepts of God, which leads to Paradise, is described as being based on the same psychic attitude on which the fight in battle is considered dependent, that is: $abr an-nafs "tying one's soul fast", or "endurance" - that is: upon a psychic attitude which is practically identical with the one implied in the term islam (aslama nafsahu) in its original meaning, that is: "'abandoning' one's soul, i.e., one's life", "defiance of death, self-sacrifice (for the sake of God)". D. Z. H. Baneth, in his article "What did Muhammad mean when he called his religion 'Islam'?; the original meaning of as lama and its derivatives" (in Israel Oriental Studies, I, Tel-Aviv University, 1971, p. 183-190), also deals with the Qur'anic phrase aslama waghahu lillahi (see especially l.c., p. 187 ff.). He considers it possible that the term aslama in its use with respect to MUQammad's religion is an abridgement of aslama waghahu lillahi, as had been assumed by early Muslim authorities (see Baneth, l.c.). As to the original meaning of this phrase, he assumes it to be (see l.c., p. 188): "to direct one's face to Allah exclusively" . In introductory remarks to his interpretation of the term islam (aslama) and its possible connection with the phrase aslama waghahu lillahi (in the just mentioned specific meaning), Baneth mentions earlier explanations of the term islam, and also the one presented by me: "defiance of death, self-sacrifice [for the sake of God]". However,

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

441

Baneth rejects this explanation of aslama (islam) in its terminological use. As to his own interpretation of the term, Baneth states that according to H. Ringgren (in his study "Aslama, islam and muslim", Uppsala 1949, p. 13) "aslama means approximately 'to give oneself entirely to God' ". As a matter of fact, this interpretation of the term as lama had already been suggested by earlier scholars, e.g., by N6ldeke (in his Delectus veterum carminum arabicorum, Berlin 1890, p. 166 b [Glossary, s.v. salima]): "totum se dedit Deo, Islamum professus est". With respect to this interpretation of the term aslama [islam] (as suggested by Ringgren), Baneth (l.c.) remarks: " ... when you give something to somebody entirely, you give it to him exclUSively". In agreement herewith, Baneth assumes that the terms aslama, islam, muslim express an idea which is the direct opposite of Mu1).ammad's conception of polytheism, according to which "his countrymen ... served other gods in addition to Allah and acknowledged them as partners to him". He furthermore states that "the existence of the postulated meaning [i.e., 'to give oneself exclusively to God '] can be proved, if not for the fourth form of the verb [i.e., aslama], at least for its first form, as well as for a noun belonging to the same root". Baneth means herewith the finite verb salima and the noun salam, salim (also silm and salm). Baneth assumes that aslama in its Qur'anic usage is derived from the noun salam, the verbal noun of salima, to which he assigns the meaning of "to belong exclusively (to)" (while salam would mean "exclusive property of a person"). At the same time, he considers it likely that aslama in the sense assigned by him to salima existed before Mu1).ammad, but admits that it does not seem "to have left traces outside the Koran". As for the reason for this (alleged) disappearance from common-Arabic of aslama in the (presumed) sense of "to belong exclusively (to)", Baneth asks: "How did this come about? Most probably because both the noun salam and the verb derived from it [i.e., aslama] were peculiar to the dialect of Mecca and the Hedjaz, or ... to Western Arabian, at all events to an Arabic idiom that vanished soon after the great Islamic conquests, at least from written Arabic .... ". Baneth does not explain why aslama (Ii) in the (presumed) particular meaning (see l.c., p. 186) "to belong exclusively to", allegedly characteristic of I:ligazi Arabic, was only preserved in the specific application: "to belong exclusively to God". We could even expect aslama in the meaning "to belong exclusively to" (generally, not only

442

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

in the sense of "to belong exclusively to God") to be found in the Qur'an, but in any case, outside the Qur'an, in the certainly not small amount of literature deriving from the I:Iigaz which has come down to us. Baneth does also not give any proof for his assumption that the noun salam in the alleged meaning "exclusive property of", from which-in his opinion-aslama was derived, was peculiar to the dialect of Mecca or the I:Iigaz and not used in other Arabic-speaking areas (if indeed the meaning "exclusively property of" can be verified altogether). There is not the slightest indication that this peculiar "Arabic idiom [dialect]" to which, allegedly, not only aslama but also salam belonged (both terms used in the specific meaning claimed by Baneth), "vanished soon after the great Islamic conquests, at least from written Arabic". Furthermore, if the finite verb as lama ("fourth" conjugation of the verb) in the alleged meaning "to belong exclusively to" is to be considered derived from salam "exclusive property of", we cannot refrain from asking why such a finite verb had to be derived from a noun which itself was the verbal noun (infinitive) of an intransitive verb (in the "first", basic, conjugation of the verb), i.e., of salima, which in Baneth's opinion possessed in turn the meaning "to belong exclusively to". Since, again in Baneth's opinion, there existed such a verb in the "first", basic, conjugation, appearing in the pattern of an intransitive verb (in agreement with its meaning), there was not the slightest reason to derive a new verb in the so-called "fourth" conjugation from the infinitive (secondarily used as an adjective and as a nomen concretum) of this original intransitive verb. It seems indeed very doubtful whether salima and salam possess the specific meaning "to belong exclusively to", and "exclusive property of", respectively, which Baneth ascribes to them and which he considers to be secondarily expressed by aslama. According to the lexical tradition (see Lane's Arabic dictionary, s.v.), which agrees with the actual linguistic usage in the texts, the verb salima, frequently followed by a complement with Ii, has the meaning "something is secured to him (or: is safely his property)". We quote, e.g., the following passage (Na~r b. Muzal).im al-Minqari, Waq'at Siflin, ed. 'Abdas-Salam Mul).ammad Harlin, Cairo 1382A.H., p. 319,3ff.): ... qiima 'Ammiiru bnu Yiisirin hi-Siffina faqiila: 'Im¢u [ma'i] 'ibiida lliihi ilii qawmin yatlubUna-fimii yaz'umuna-bidami ?-?iilimi linafsihi, al/:ziikimi 'alii 'ibiidi lliihi bigayri mii fi kitiibi lliihi, innamii qatalahu $-$iili/:zuna l-munkiruna /il- 'udwiini, al-iimiruna bil-i/:zsiini. faqiilu hii'ulii'i llagina Iii yubiiluna igii salimat lahum dunyiihum [wa]law darasa hiigii

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

443

d-dinu: lima qataltumuhu? faqulnii: li'il}diilihi .. .' " ... 'Ammar b. Yasir rose at Siffin and said: 'Come on with me, oh servants of God, to people who-as they claim-seek revenge for the [spilt] blood of the one who acted wrongfully towards himself, the one who ruled his people not in accordance with the regulations of the book of God (i.e., 'Utman b. 'Affiin). It is the pious that killed him, those who detest enmity, the ones that recommend the doing of good. But [now] those who do not care-if just (their) worldly life is secured to them (igii salimat lahum dunyiihum), even if this [glorious] religion might be effaced-they [now] said: 'Why did you kill him?', and we answered: '[We killed him] because of his illegal actions (his deviation from approved traditional usage)' ... ". Cf. also the analogous phrase (Tabar!, Annales, 1,3; 1566,7): ... fahalummu falnattabi'hu wanu$addiqhu fataslama lanii dunyiinii wa'iibiratunii " ... Come on!, and let's follow him (i.e., the Prophet Mu\:.lammad) and believe in him so that this world as well as the hereafter be secured to us!" (and thus, in the sense of "to be secured to someone", frequently, in various contexts). The noun salam (an original infinitive, used secondarily as an adjective and nomen concretum) does not possess, as Baneth assumes (see also I.e., p. 187), the meaning "exclusive property (of one person)", but it denotes (among related meanings) "the one who is the secured and safe possession (of someone)", e.g., "the prisoner (and slave)". The use of salam as an adjective and nomen concretum, "someone who is the secured possession of a master" (also "captive, slave"), represents-as the indigenous lexicographers well recognised (see Lane's Dictionary, s.v.)-a secondary use of an original infinitive, nomen abstractum (a secondary use quite frequently occurring in concepts of this type). Baneth quotes the Qur'anic verse (Siirah 29, 30): ¢araba lliihu mala/an ragulan fihi surakii'u mutasiikisuna waragulan sa/aman liragulin hal yastawiyiini malalan. He interprets the words waragulan salaman liragulin by "and a man wholly belonging to one man". But the exact translation of these words-in spite of their being contrasted by ragulan fihi surakii'u mutasiikisuna-is: " ... and a man being the secured (or: safe) property of a man". Inasmuch as Baneth interprets the term aslama as implying the idea of "strict monotheism": "to belong exclusively to [Allah]" (see above p. 441), and considers as lama in this meaning abridged from aslama waghahu [lilliihi] , I refer especially to the passage in Ibn Hisam's Sirat Rasuli-lliih, p. 149,6-9, dealt with above p. 439, where the phrase

444

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

aslama waghahu lilliihi is not only used with a human being as its subject, but is also used of (more accurately: transferred to) "the earth" and "the clouds" as its subject, and is applied to their activities in the service of God: (6) aslamtu waghi liman aslamat lahu l-an;lu ... (8) wa'aslamtu waghi liman aslamat lahu l-muznu ... (see the full text above p. 439). In such a context, aslama waghahu li ... must by all means signify a devotion [to God] through "service" (which-with respect to a human being-in the early Arab [and Islamic] conception means: "sacrificing one's life through heroically fighting in battle", see above p. 439) and not a kind of theoretical attitude (as presented by the concept of "strict monotheism": "to belong exclusively to [Allah] "). Baneth sees a confirmation of his theory concerning the meaning of the Qur'anic phrase aslama waghahu lilliihi (see above p. 440) in another Qur'anic expression which is characterized by the noun wagh "face". He states (I.e., p. 188): "In my own opinion aslama waghahu lilliihi means 'to direct one's face to Allah exclusively', and should not be separated from waggaha waghahu, which occurs in Surah VI, verse 79: inni waggahtu waghi lilladi ja!ara s-samawiiti wal-art;la /:lanifan wamii ana min-a l-musrikina 'I turn my face to the Creator of Heaven and Earth as a l:Ianlf, and I am not of those who acknowledge partners to God'. The connexion between the turning of the face towards the Creator and monotheism is obvious here; ... ". Though the phrase wamii ana min-a l-musrikina is contained in this verse, it must be doubted that the phrase inni waggahtu waghi lilladi ... by itself stresses the idea of monotheism and that the idea implied in it is equivalent with the idea implied in the phrase aslama waghahu lilliihi. The phrase inni waggahtu waghi lilladi ... can hardly be separated from certain early-Islamic phrases, as, e.g., the following one, appearing in a poem by Abu Al)mad b. Gal)s, one of the muhiigirun, those Meccans who emigrated together with the Prophet to Medina (Ibn Hisam, Sirat Rasuli-lliih, ed. Wiistenfeld, p. 318, 6): ilii lliihi waghi war-Rasuli waman yuqim ilii lliihi yawman waghahu Iii yubayyabi "My face is directed to God and the Apostle, and whosoever directs some day his face towards God, will not be frustrated". On the one hand, this line bears great similarity to the Qur'anic verse adduced by Baneth; on the other hand, no special emphasis of the monotheistic idea ("belonging to God exclusively") can be intended by this line. We may also mention in this context Uial b. l:Iagl's lines (14-15), referring to his heroism in the battle of Siffin (see the full text below p.445): (14) man jatan

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

445

ya'bugu !-!ariqa ila llahi fakuntu !lagi abagtu !-!ariqa. (15) lJasira r-ra'si

There are early passages that make it clear that the early-Islamic ideal of "defiance of death (self-sacrifice) for the sake of God" was -obviously on the basis of a secondary ideological developmentconsidered to serve an ultimate purpose: the purpose of "coming near to God". This idea is most clearly expressed in two lines of poetry by Ka'b b. Malik (Ibn Hisam, Sirat Rasuli-!lah, p. 614,11-12), a passage which, though already discussed in Spiritual Background, p. 21, may be quoted here once more: (11) waqala lana Rasulu-llahi lamma badaw lana * garu 'ankumu hawla I-maniyati wa!ma'u (12) wakunu kaman yasri I-lJayata taqarruban * ila malikin yulJya ladayhi wayurga'u "and the Messenger of God said, when they had appeared before us: 'Throw off the terror of Fate and be eager (to attack), and be like one who defies death (literally: "sells his life"), so that you may come near to a king by whom men live and to whom they return'''. This connection of "defiance of death in battle" (expressed by aslama/ sara nafsahu, and aslama/sara waghahu) with the ultimate religious goal- "becoming united with God" -has also found a most clear expression in the reports concerning the battle of Siffin. We quote the following passage (Na::;r b. Muzal;iim, Waq'at Siffin, p. 443,6ff.): ... wakanu qad laqulu 'an-i l-birazi lJina 'arjtjathumu I-lJarbu, faqala l-Astaru: ya ahla 1-'Iraqi, ama min ragulin yasri nafsahu [lillahi]? fabaraga Ulalu bnu !Jaglin fanada bayna l-'askarayni: hal min mubiirizin? fada'a Mu'awiyatu !Jaglan faqala: dun aka r-ragula ... fabaraza kullu wa/:zidin minhumii itii $iibibihi fabadarahu s-saybu bita'natin fa/a'anahu I-guliimu ... " ... and they were slow in going forward for single combat when the heavy fighting had 'bitten' them; and al-Astar rb. al-1:Iari! an-Naba'i] said: '0 people of 'Iraq! Is there no one who is ready "to 'sell' his soul" (= "to 'sell' his life")?'. Thereupon U!al b. 1:Iagl went forward and shouted between the two armies: 'Is anyone ready for single combat?'. Then Mu'awiyah (on the Syrian side) called 1:Iagl (i.e., U!al's father) and said: 'Attack the man !' .... Then the two went towards each other, and the old man hastened to spear him (i.e., his son), while the young man at the same time speared him (i.e., his father) ... ". According to our report this incident became the basis of poems by 1:Iagl as well as by his son Uta!. We quote from U!al's poem the following three lines (l.c., p. 444, lines 13-15): (13) qala ahlu 1-'Iraqi ig 'a~uma I-ba!bu wanaqqa I-mubarizuna naqiqii (14) man fatan ya'bugu !-!ariqa itii Iliihi fakuntu /lagi abagtu !-!ariqii (15) IJiisira

446

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

r-ra'si Iii uridu siwii I-mawti arii kulla rna yarawna daqiqii (variant: arii I-a'?ama I-galila daqiqii) "The people of 'Iraq said, when the affair had become momentous and those who were used to fight in single combat, were grumbling: 'Who is the man that will take the road towards God?'. And I was the one that took that road, with uncovered head, while I was desiring nothing but death, and was considering the momentous, dangerous affair as insignificant ... ". These lines, ascribed to Utal, express the same idea as the Prophet's words in Ka 'b b. Malik's lines quoted above (p. 445): "Be like one who defies death, so that you may come near to a king by whom men live and to whom they return". Interesting for us in Utal's lines is the expression ~iisira r-ra'si "with uncovered head". That is: "Utal 'surrendered' (exposed) his face", i.e., aslama waghahu or sarii waghahu = "defied death". The fighters' exposing their uncovered heads, which we consider pregnantly expressed by the phrase as lama (or: sarii) waghahu [lilliihi] (and related expressions, see above p.434), is very clearly implied also in the following passage (from a speech by Malik b. al-l:Iarit al-Astar, a partisan of 'Ali b. Abi Talib [in Na~r b. Muzal).im al-Minqari, Waq'at Siffin, p. 251, 1-2]): yii ayyuhii n-niisu, gurjrju I-ab~iira, wa'at/t/u 'alii n-nawiigirji, . wastaqbilu I-qawma bihiimikum " ... Oh people, lower your eyes and bite on your teeth (i.e., be stubborn), and go towards the enemies with your [uncovered] heads !". This pregnant use of wagh "face" -in the last quotation more concretely expressed by hiim (pI. of hiimat- "skull", "head")we recognize also (in the plural wuguh) in the following passage (Tabari, Annales, 11,2; 1087,4-5; year 83A.H.): faqiila lanii Gabalatu [bnu Za~rin],' irjii ~amaltum 'alayhim fa~milu ~amlatan ~iidiqatan walii taruddu wuguhakum 'anhum ~attii tuwiiqi'u ~affahum " ... If you attack them, attack them violently and do not turn away your (uncovered) faces until you break into their ranks!". We may also refer to the following passage (Na~r b. Muzal).im al-Minqari, Waq'at Si/fin, p. 354,3): falaysa min waghin ya~milu 'alayhi ilIii ~abaru lahu "No 'face' (i.e., [enemy] warrior) was attacked by him (i.e., by Hasim b. 'Utbah) but withstood him". In this context we should also pay attention to some sentences from a speech (~u!bah) which, according to tradition, 'Utbah b. Guwayiyah, an adherent of 'Ali b. Abi Talib, delivered to those fighting in the battle of ~iffin on the side of 'Ali (Na~r b. Muzal).im, Waq'at SifJin, p. 264,3ff.): ... waqad kuntu atamannii s-sahiidata wa' ata'arrat/u lahii fi kulli ~inin fa' aha lliihu ilIii an yubligani hiirjii

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

447

I-yawma; alii wa'inni muta'arricjun sii'ati hiigihi lahii, waqad !ami'tu allii uf:zramahii. Famii tanta?iruna [yii] 'ibiida !liihi min gihiidi a'dii'i lIiihi? a!Jawfu I-mawti I-qiidimi 'alaykum, ag-giihibi bi'anfusikum Iii maf:ziilata, aw min cjarbati kaffin aw gabinin bis-sayfi? atastabdiluna d-dunyii bin-na?ari ilii waghi lIiihi 'azza wagalla, aw muriifaqati n-nabiyina wa~-~iidiqina was-suhadii'i ... (I. 9 ff.) .. , lumma qiila: yii i!Jwatah, inni qad bi'tu hiigihi d-diira bid-diiri !lati amiimahii, wahiiga waghi ilayhi (sic); Iii yabraf:z-i lliihu wuguhakum walii yaq!a'-i lliihu arf:ziimakum " ... I have always striven for martyrdom and I have exposed myself to it at any time. But God insisted on having me reach this day. But now, at this hour, I am indeed exposing myself to it, and I am desirous not to be deprived of it. And what do you expect, 0 servants of God, from your fighting with the enemies of God? [Are you kept back from fighting by] the fear of death? It will certainly overtake you, and will unavoidably deprive you of your 'souls' (i.e., of your lifes). [Or are you afraid] of a hand or a forehead being hit by the sword? Would you not like to exchange for this world the chance of looking into the face of God, or to be in the company of the Prophets and the truthful and the martyrs? ... (I. 9ff.) ... then he said: 0 brethren!, I am selling this world (literally: 'house') for the world (liter.: 'house') which is in front of it. And this my face [is directedJ to Him (i.e., to God). May God not abandon your faces, and may He not forsake you (more literally: forsake you, as one forsakes one's kin)". I have the clear impression that the phrase (in the quoted text, line 9 ff.) "My face [is directed] to Him (i.e:, to God)" implies the concept of "the fighter's face, which is uncovered while he. attacks the enemy". In this condition-while exposing his uncovered face to the enemy-the fighter quasi turns his face to God, and he anticipates (by this aspect of his fighting activity) his ultimate goal, namely: to look into God's face (an-na?aru ilii waghi lIiihi). For various phrases expressing the same concept, see the quotations above p. 445, and below p. 450 (Siirah 13, v. 22). The concept "to look into God's face" is, of course, a special (very important) aspect of the more general idea of "coming close to God", an ideal which is preferably achieved by the fighter's defiance of death in his fight against the enemies of God, as expressed by Ka 'b b. Malik in words ascribed by him to the Prophet; see Ka'b b. Malik's lines of poetry (Ibn HiSam, Sirat Rasuli-lliih, p. 614, 11-12), quoted above p. 445. I stress here once more that Ka'b b. Malik's expression (in line 12 of this passage, see above p.445, line 11): ... taqarruban

448

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

ilii malikin ... " ... [in order to come near to a king (i.e., to God)] ... " clarifies the meaning of the preposition ilii in Siirah 31,21: yuslimu waghahu ilii Uiihi. The preposition ilii "to, towards" in this phrase implies the concept of "coming near (to)" (for further details see Spiritual Background, p. 23). Therefore, the phrase yuslimu waghahu itii Uiihi expresses the idea "he is ready to sacrifice his face (that is: his life) [so that he may come near] to God". In this connection we mention also the Qur'anic expression-Jesus' words to his disciples, his "helpers", "allies" (Siirah 3,45, and 61, 14)-man an~iiri ilii Uiihi, which should be interpreted by: "Who are my helpers [on the way] to God (i.e., so that I may come near to God)?". The expression an~iiri ("my helpers") corresponds to the expression an~iiru Uiihi in the disciples' answer: nabnu an~iiru Uiihi "We are God's helpers" (see my discussion in Spiritual Background, p. 69-70). We revert once more to the question of the basic (original) intention of the terms Isliim, Muslim(una}. In assuming (see above p.437ff.) the original meanings of these terms to be "defiance of death, selfsacrifice (for the sake of God)" and "[those] who defy death, or: give up their lifes by heroically fighting in battle (for the sake of God)", respectively, I consider them to be paralleled by the name of a religious Arab movement or sect which preceded earliest Islam. I have in mind the Quraysite group known by the name of Ifums. The view concerning the name Ifums held by Western scholars, who in this respect follow the traditional Muslim view, is well stated by W. Montgomery Watt, in his article concerning the group in Encyclopedia of Islam 2 , vol. 2, p. 577 b: "lfums 'people observing rigorous religious taboos'. The word is the plural of abmas 'hard, strong (in fighting or religion)" but ... one of the Ifums is called abmasi ... ". I doubt that abmas "hard, strong" is used from the outset in a very general way so that the word could be applied both to "fighting" and to "religious observance". I am of the opinion that the adjective abmasu means basically nothing else but "courageous, hardy, heroic, in fighting, in battle", exactly as the abstract noun associated with abmasu, that is: bamiisah, is used in no other meaning but "hardihood, firmness, heroism (in fighting)". The use of the term abmas-of which the plural bums as name of the religious group is derived-in the general meaning "hard, strong", not only with respect to "hardihood in fighting" but also with respect to "rigorousness in religious observance" (if this meaning of abmas is indeed documented), should be considered as based on a secondary (not original) interpretation

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

449

due to the assumption that the strict adherence of the l:Iums to certain religious observances ("religious taboos") must necessarily have found an expression in the name of the group. Indeed, not only this name of a group (or a movement) preceding earliest Islam is a kind of parallel to the names used by Mul)ammad for his religion and its adherents (isliim and muslim(una), respectively), but also the specific name of a very important religious-political party which arose after the Prophet's death but still in an early period of Islamic history, is again to be viewed as a kind of analogon to the name Muslimun. I have in mind the name Ijawiirig, the so-called tFirigites. The ijarigites are also named suriit- (Siirun). Suriit- (siirun), sing. siiri (n), in this specific usage, is doubtless to be derived from the original phrase sarii nafsahu "he sold his soul", or: "he sold himself" (see, e.g., above p.435). The ordinary interpretation of this phrase is that "man sells his soul to God in exchange for God's favor" (see my Spiritual Background, p. 20 ff.). The Qur'anic passage which is specifically interpreted in this sense, is Siirah 2, v. 203: wamin-a n-niisi man yasri nafsahu btigii'a marqiiti lliihi. See, e.g., Torrey, The Commercial-Theological Terms in the Koran, p. 38: "The believer is [in Siirah 2, v. 203] represented as selling himself, the price being marqiit Alliih and the purchaser Allah". And also the designation suriit- (pI.; sing. siiri (n)) for the ijarigites is traditionally interpreted in the same sense (e.g., also by L. Massignon, Encye!. of Islami, vol. 14,1; 392; for other references cf. Spiritual Background, l.c.). As already stressed (see Spiritual Background, p. 21), the verb sarii in the above quoted Qur'anic passage, as well as its use in the term siiri(n), pI. suriit- (used for theijarigites), means nothing but "the 'selling' of one's soul (i.e., of one's life) by heroically fighting unto death", in other words: "the defiance of death (the self-sacrifice) in battle". The verb sara (sara nafsahu "he sold his soul") does not constitute a "commercial-theological" term, but is to be considered as an expression for the image of "self-betrayal" (i.e., the harsh, quasi inimical treatment of oneself), by which (see for various analogous terms Spiritual Background, l.c.) the idea of "self-sacrifice", "defiance of death" is viewed. As for the term sari(n) , pI. surat-, it is in the present context (i.e., concerning its use with respect to the ijarigites) important to note that sari(n) ("seller" = "heroic fighter") is also used for the non-Muslim (pagan) Turks (see Spiritual Background, p. 21). Of course, the Qur'anic passage Siirah 2, v. 103 (see

450

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

above) indeed implies also the idea that man by "'selling' his soul", i.e., by "sacrificing his life (in battle)" intends to gain God's favor (and God's reward). But mar¢at Allah "God's satisfaction (favor)" by no means represents (in this passage) "the price for man's 'selling' himself"; the expression thus does not constitute a quasi necessary complement of the verb, so that the Qur'anic verse (2,103) indeed would have to be interpreted by: "Among men there are such as sell their soul ( ... man yasri nafsahu) [to God] for (or: in exchange for) God's satisfaction [or: favor)". Rather, the term maNjati llahi in ... man yasri nafsahu btiga'a maNjati llahi is part of an adverbial phrase (ibtaga'a mar¢ati llahi) which expresses an additional, independent idea: "He who gives up (literally: sells, i.e., sacrifices) his life [in fighting heroically], [doing this] out of [his] desire for God's favor ... " = "he who defies death out of his desire for God's favor ... ". Cf. Siirah 13, v. 22: ... walladina ~abaru btiga'a waghi llahi " ... and they who endured [and sacrificed their lifes] out of [their] desire for God's face (= their desire to see God's face)", where alladina ~abaru corresponds to man yasri nafsahu in Siirah 2, v. 103. This interpretation of the term surat- (as a variant name for the Ijarigites) in the sense of "those who 'sell' -i.e., give up, sacrificetheir lifes [in battle] (for the sake of God)" -in analogy to the original meaning of the terms islam, muslim(una)-strengthens-indeed confirms-the assumption that the more regularly used name of the group-i.e., ijawarig-means originally nothing but "those who go out (or: go forth)" in the sense of "those who go out to the battlefield", in other words: "the fighters" and (at the same time) "those who defy death and surrender their lifes by fighting heroically". To this interpretation of the term ijawarig I am applying here my earlier statement (with respect to the term surat-; see Spiritual Background, p. 21): The transformation of this expression (lJawarig) into a name for a particular religious-political sect (party) of Islam is (exactly like the transformation of the term surat-) only a repetition of the semantic development which transformed the expression islam into a name for the religion of Mul).ammad. This parallelism in designation corresponds exactly to a parallelism of tendencies in original Islam and the Ijarigite-sect. That means: Of the various meanings suggested as original meaning of the term ijawarig ("ljarigites"), which are mentioned and discussed by W. Montgomery Watt in his book The Formative Period of Islamic Thought (Edinburgh University Press, 1973), p. 15ff. (and

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

451

see also Ibid., p. 327), only that one should be considered as correct which Watt lists as (no.) 4: "They are those who go out and take an active part in the jihiid, in contrast to those who 'sit still'; the two groups, and the concepts of khuruj, 'going out', and qu'ud, 'sitting still', are contrasted in the Qur'an (e.g., 9.83/4)". In connection with this use of qa'ada "to sit still" as contrast to baraga "to go out" = "to take part in the gihiid" (cf. also Qur'iin 9,47), we may also mention the identical use of aqiima "to stay", "to keep aloof from the gihiid" (with the noun muqim, "the one who stays [keeping aloof from the gihiid]") as contrast to baraga "to go out and take part in the gihiid". We quote the following line of poetry by the ijarigite Ma'dan al-Iyadt, referring to the "moderate" section among the ijarigites (al-Mubarrad's Kiimil, ed. Wright, p. 528, I): saliimun 'alii man biiya'a lliiha siiriyan walaysa 'alii 1-lJizbi l-muqimi saliimu "peace [be] to him who has made a pledge to God (or: a contract with God) to 'fight unto death' (literally: to 'sell [his soul]'); but there may be no peace to the group that 'stays'''. AI-lJizb al-muqim "the group that 'stays'" are those that do not "mobilize" (nafara, yanfiru) like the genuine ijarigites, but keep aloof from fighting. The expression (al-lJizb al- )muqim ("the group that 'stays''') appears here as the direct contrast of siiri(n) ("selling [one's soul]"), singular of suriit-, which is a characteristic term for the ijarigites and thus a terminological synonym of the term bawiirig. Accordingly, the terms al-bawiirig and al-lJizb al-muqim in turn should be considered as literal contrasts: al-bawiirig means "those that do not 'stay' (muqim), but 'go forth' to the battle-field (being ready to 'sell' their lifes [= suriit-]". For aqiima "to 'stay' peacefully in the background", in contrast to the ideal attitude of the ijawarig, i.e., their "going forth", we may also quote the following passage (Riwiiyiit al-Agiini, selections from Kitiib al-Agiini, by A. Sall:tant, vol. 2, p. 438, 9):faqiila li'a~lJiibihi: mii yalJillu lanii l-muqiimu 'alii mii narii (ay: gawran ?iihiran ... ) walii yasa'unii ~-~abru 'alayhi "and [the 'Ibadt-Ieader 'Abdallah b. YaQya al-Kindl] said to his companions: 'In view of what we see (i.e., patent iniquity), we are not permitted to 'stay' (i.e., not to 'go forth'); and we cannot be inactive in view of that". The verb aqiima-with the nominal term (al-lJizb) al-muqim ("[the group] that 'stays' ")-and the verb qa'ada-with the noun qa'ad- (plural), or qii'iduna-are, of course (in this specific use), strict synonyms; and both verbs (with their nouns) are expressions for a concept that is the direct semantic contrast of the concept expressed by the verbal expressions baraga

452

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

and sarii nafsahu, which are the bases of the nouns bawiirig and suriit(respectively) in their terminological use. As already hinted, I not only interpret the expression /jawiirig in its terminological use as based on the verb baraga in the specific sense "to go out to the battle-field", but at the same time assume that /jawiirig implies the sense of "those who fight [for the sake of God]" (that is: baraga = qiitala = giihada), so that the term comes close in its meaning to the specific meaning implied in the original meaning of the terms isliim, muslim(una}. With respect to this assumption, I quote some passages contained in al-Balaguri's Ansiib al-asriif, vol. 4A (ed. M. Schloessinger-M.J. Kister, Jerusalem 1971), p.138ff. E.g. (l.c., p. 142,11): baraga mawlan Ii-Bani l-lfiirili bni Ka'bin yuqiilu laM Abu Maryama wama'ahu mra'atiini Qatiimu wa-Ku/:wylatu wakiina awwala man abraga ma'ahu n-nisii'afa'iiba fJiilika AbU Biliilin Mirdiisu bnu Udayyata wakiina awwala man kariha buruga n-nisii'i faqiila qad qiitalat-i n-nisii'u ma'a Rasuli-lliihi (!fl'm) waqiitalna ma'a l-Muslimina bis-Sa'mi waliikinni arudduhumii faraddahumii "and a client of the Banu l-J:Iari! b. Ka'b by the name of Abu Maryam 'went forth', and with him were two women: Qatam and Kul:laylah; and he was the first to let women 'go forth' with himself. And Abu Bilal Mirdas b. Udayyah blamed him because of this; and he was the first to disapprove of the women's 'going forth' [to take part in battle], and he said: 'the women fought with the Prophet, and they fought with the Muslims in Syria; nevertheless I shall hold them back'; and he held the two of them back". This passage is of special importance, for here the term baraga varies in the passage itself with the term qiitala ("to fight"), which thus serves as an interpretation of baraga. j furthermore quote (l.c., p. 143,22): falammii waliya l-Mugiratu bnu Su'bata I-Kufata (Ogtama'a Siilimu bnu Rabi'ata [wa]l-Mustawridu bnu 'Ulliifata ... wagayruhum ila lfayyiina [bni ~abyiina] fi manzilihi liyatasiiwaru fiman yuwallunahu amrahum liyabrugu munkirina lil-gawri wa??ulmi ... "and when al-Mugirah b. Su'bah acceded to the governorship of Kufa, Salim b. Rabi'ah and al-Mustawrid b. 'Ullafah ... and others met with J:Iayyan [b. z:abyan] at his house, to take counsel concerning the person whom they should put in charge of their affairs for the purpose of their 'going forth' because of their rejection of oppression and iniquity ... " (cf. also Ibid., p. 147,6ff.). In the preceding discussion (see especially p.449ff.) I rejected the usual assumption that the phrase sarii nafsahu (and variations) in the Qur'anic passage discussed above p.449 (Surah 2, v. 203) and in

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

453

other passages (see my Spiritual Background, p. 20, and also above p. 438) expresses the idea of "selling one's soul to God", and that also the term suriit- for the J:larigites is based on this meaning of this phrase. In spite of this, I nevertheless assume that this idea (or image)~"selling one's soul to God"~indeed developed (secondarily) on the basis of the idea and form of expression of which the quoted Qur'anic passage is a characteristic example. In my opinion, the idea that "the Muslim 'sells his soul' (i.e., sacrifices his life in battle) and thus gains God's favor (and eternal life in Paradise)", led to the image that "man sells his soul to God" and, secondarily, that "God buys man's soul~and grants him, in exchange for it, eternal life". That means: the idea that "man sells his soul to God" or that "God buys man's soul" should be considered as based on a semantic development~more specifically: as based on a "semantic 'association'''. Since "man 'sells' (= sacrifices) his soul~for the sake of God", he quasi "sells his soul to God". The idea that "man sells his soul (or: his life on earth) to God" or that "God buys man's soul~and grants him, in exchange for it, eternal life" is found in the Qur'an as well as in post-Qur'anic literature. I quote the following Qur'anic passages. Siirah 4, v. 76: falyuqiitil fi sabili lliihi llarjina yasruna l-/:tayiita d-dunyii bil-iibirah ... "Let those then fight for the sake of God (more literally: on the path of God) who sell the life of this world for that of the next ... ". Furthermore, Siirah 9, v. 112: inna lliiha starii mina l-mu'minina anfusahum wa'amwiilahum bi'anna lahumu l-gannata ... "Behold, of the believers God has bought their souls and their possessions in exchange for paradise for them ... ". From extra-Qur'anic passages (occurring in later literature) I quote the following. Tabari', Annates, II, 884,2 (year 70 H.): ... alii fabi'u lliiha anfusakum !ii'i'ina wa'amwiilakum tadbulU l-gannata iiminina ... "... Indeed, sell God your souls and your possessions obediently, so that you may enter paradise safely! ... ". Ibid., II, 1378,15 (year 101 H.): wafawiirisin bii'u /-iliiha nufusahum min Yaskurin 'inda l-wagii fursiini (from a poem) "Many a horseman from [the tribe of] Yaskur sold God his soul in the din of battle". Also Ibid., 2011,3 (year 130 H.): qad bii'u lliiha 'azza wagalla anfusan tamutu bi' anfusin Iii tamutu "they sold God mortal souls in exchange for immortal souls". (Cf. also Waq'at Siffin, p. 264,9ff., quoted above p. 447). In connection with this (in my opinion secondarily developed) idea, one should also pay attention to the phrase (see above p.435,

454

ARABIC ASLAMA (ISLAM) AND RELATED TERMS

line 29) man yasri waghahu lilliihi (= man yasri nafsahu tilliihi), in which the term lilliihi is an additional complement (unnecessary for the construction of the verb sarii, yasri): "Who sells his face (= his soul) for (the sake of God? = "Who is ready to sacrifice his life in battle for the sake of God?". That the meaning of the phrase man yasri waghahu lilliihi cannot be "Who sells his 'face' (i.e., his life) to God ?", but only "Who 'sells' (Le., sacrifices) his life for the sake of God?", becomes completely clear from the corresponding phrase in an analogous context, quoted above p. 436, line 8: man yasri nafsahu, for which-since the phrase is not followed by the expression lilliihi-no other meaning can be assumed than "Who 'sells' his soul?" (in an absolute sense, that is: "Who sacrifices his life [in battle, by fighting heroically]?"). On the other hand, in a phrase like man yasri waghahu lilliihi, the term with ti (i.e., lilliihi "for [the sake of] God") could, on the basis of the above-mentioned semantic process ("semantic 'association' "), easily develop into a direct (syntactically quasi necessary) complement of the verb yasri (or its variations): "Who sells his soul to God?" (that is: with Ii '" expressing the "dative"). The possibility of the secondary development of the idea of "selling one's soul to God" out of the concept of "selling one's soul-for (the sake of) God" (in the sense of "sacrificing one's life [by heroically fighting] for the sake of God" is also to be recognized in two others of the above quoted passages, in which God is described not only as "buying the souls of the believers" but also as "buying their possessions" (see above p.453). The concept of "possessions" in addition to "souls" is understandable in connection with the idea of "sacrificing one's life"; that is: "sacrificing one's life and one's possessions for the sake of God" (cf., e.g., Siirah 4,97: ... al-mugiihidina fi sabiti lliihi bi'amwiilihim wa'anfusihim, and see also Siirah 9,41). In heroically fighting against his (and God's) enemies, man may have to sacrifice his life as well as his possessions (which latter he sacrifices in preparing and carrying out his fight). Cf., e.g., the poetic phrase quoted above p. 438, line 2 (Tabari, Annales, II, I; p. 341, line 12): ... siiri * duna Ifusaynin muhgati wadiiri " ... 'selling' (i.e., sacrificing), in defence of ( = for the sake of) l:Iusayn, my blood (= my soul) and my house". However, in distinction from "God's buying the Muslim warriors' souls", which-though not an original motif (see above)may still be considered a more or less rational idea, "God's buying their possessions" (see Siirah 9, v. 112, quoted above p. 453) makes hardly good sense, if taken as an original idea.

29 THE ORIGIN OF THE PRINCIPLE OF 'ISMAH: MUJ::IAMMAD'S "IMMUNITY FROM SIN"l A well-known principle of Islam is the so-called 'i$mah, the "immunity from sin" ("sinlessness") which is ascribed to the prophets. Mubammad, the prophet of Islam, was ma'$um, "immune from sin"; and the same also applies (to a larger or lesser extent) to the prophets preceding him. In Shi'ite Islam the 'i$mah applies also to the Imams. 2 In my opinion, this immunity of certain persons from sin is a genuine Arab, pre-Islamic popular motif, and the Islamic idea which applies this characteristic to Mubammad and the prophets preceding him, is based on this pre-Islamic concept. In the first place, I refer here to a statement which I consider as the oldest recorded instance of the idea of Mubammad's immunity from sin. This passage is contained in a speech, which, according to tradition, Abu Bakr, the Prophet's companion and successor (balifah) as head of the Islamic community, delivered on the day after the Prophet's death. The form used in this speech to express the idea of "immunity from sin" Ci$mah) can be identified with .the form in which the analogous pre-Islamic idea usually appears, and thus represents the intermediary between the latter and the ordinary Islamic idea of the "immunity from sin" characteristic of the Prophet. The sentences in Abu Bakr's speech which are of interest to us in this context (they represent the beginning of the speech), read as follows (Tabari, Annales, I, 1845, 17ff.): Ya ayyuha n-nasu, innama ana millukum wa'inni la adri la'allakum satukallifuni ma kana Rasulu -llahi ($I'm) yU!iqu. Inna llaha $!afa Mubammadan 'ala l-'alamina

1 Originally published in Le Museon; Revue d'etudes orientales (Louvain), vol. 88 (1975), p. 221-225. Cf. also chapter 50, below p. 572-573. 2 Concerning the historical development of the 'i$mah-principle in Islamic theology see Goldziher, Vorlesungen iiber den Islam, 1. Aufl., p. 220-223; Idem, in: Der Islam, m (1912), p. 238-245; and especially Tor Andrae, Die Person Muhammads in Lehre undGlaube seiner Gemeinde, Upsala 1918, Kapitel III (p. 124-174): "Die Unfehlbarkeit (' i$ma) des Propheten".

456

THE PRINCIPLE OF 'ISMAH

wa'a~amahu min-a I-iifiiti. Wa'innamii ana muttabi'un walastu bimubtadi'in fa'in-i staqamtu fatiibi'uni, wa'in zugtu faqawwimuni. Wa'inna Rasula-lliihi qubirja walaysa a~adun min hiirjihi l-ummati ya!lubuhu bima?lamati rjarbati sawtin famii dunahii. Alii wa'inna Ii say!iinan ya'tarini fa'irjii atiini fagtanibuni Iii u'al1ir fi as'iirikum wa'abSiirikum .... This should be interpreted as follows: "0, ye people! I am only like one of you, and I am not certain whether you will not burden me with what only the Prophet was able to do. God has elected Mui:lammad over all other human beings, and has protected him (= made him immune) from moral weaknesses (stumblings) ('a~amahu min-a I-iifiiti). And I am only a follower, and I am not an initiator. So, if I act in a strait way, then follow me! But if I deviate (from the truth, or: from. the strait path), then straiten me out! For the Messenger of God has been taken away [by God], while nobody in this community accuses him that he has committed against him a wrong which would necessitate as expiation a stroke with the whip, or even less. Indeed, I am possessed by a demon (literally: a Satan) who is used to attack me suddenly (and to seduce me to sin). Now, if he comes to me, keep clear of me !-so I will not even scratch your hairs and your faces (i.e., I shall not inflict upon you the slightest injury)

The finite verb which in this passage expresses the concept of "protecting, immunizing" is the verb 'a~ama, the characteristic verb used with respect to the Islamic principle of the "immunity from sin" ('i~mah). The noun which (in Abu Bakr's statement) appears as the object of this (transitive) verb 'a~ama, is the noun iifiit-, the plural of iifat-, the ordinary meaning of which is "mishap, distress, damage". In the context of the statement under discussion, I interpret the word iifiit- (pI. of iiJat-) in the sense of "moral weaknesses, stumblings". This meaning of iifiit- (pI.) seems to be clearly present in some of the passages quoted in the sequel, which I consider as testimonies for the pre-Islamic existence of the idea of "immunity from sin". An important evidence for the pre-Islamic existence of the idea of "immunity from sin" as well as for the meaning of iifiit- indicated above, I recognize in the following passage from a poem by an-Nabigah ag-pubyani in honor of the royal family of the tribe of Gassan (anNabigah, Diwiin, ed. Ahlwardt, no. 25,3-4): (3) humu l-mulUku wa'abnii'u I-muluki lahum farjlun 'ala n-niisi fi I-Ia'wii'i wan-ni'ami (4) a~­ liimu 'Adin wa'agsiidun mUfahharatun min-a I-ma'aqqati wal-iifiiti walalami "(3) They are the kings and the sons of kings; they surpass

THE PRINCIPLE OF 'ISMAH

457

the [other] people in unhappy as well as in happy situations. (4) They possess the forbearing (considerate) souls of the 'Ad (the ancient Arab tribe, which passed away long ago); and [they have] bodies (or: are persons(?), see below) purified of irreverence (towards their tribal brethren), (moral) weaknesses (stumblings, sins; afat-), and guiltiness". The fact that afat- appears in this passage beside ma'aqqat(= 'uquq, the contrast of birr) and alam (compare ilm, and also Hebrew asam) , does not leave any doubt that its meaning is here "(moral) weaknesses, stumblings, moral failures (sins)". Moreover, the expression mu!ahharat(un) "purified" as attribute of agsad "bodies" clearly implies the idea of an inherent freedom or immunity of the persons involved from sin and guilt. In a different but very clear form of expression, we notice the concept of "immunity" in the following passage from an elegy by Abu lJiras al-Hugali (on a certain lJiilid b. Zuhair [in: Neue HurjailitenDiwane, ed. Hell, p. 67; no. 10, 15]): ... (15) a!famma kana!fli s-sayfi yartabu lin-nada ba'idan min-a I-afati wal-buluqi I-wahmi, in Hell's interpretation (his translation, p. 34): "Die Nase hochreckend wie eine Schwertspitze findet er Freude am Wohltun, abhold den Ranken (i.e., afat-) und unvertraglichem Wesen". I suggest the following interpretation: "; .. [his] nose is stretched into the hight like the point of a sword (i.e., he is proud); joyfully he hastens to practice charity (hospitality): he is exempt (literally: far) from (moral) weaknesses (stumblings, ba'idun min-a I-afati) and immorality". The term ba'id min (literally: "far from" is used here as a pregnant expression for the concept "immune against", or: "completely exempt from (or: beyond)". Accordingly, one should not, in agreement with Hell, attempt to harmonize the coordinated concepts aI-afat- and al-buluq al-wahm by assigning to afat the meaning "intrigues" ("Ranke") and to the entirely clear expression al-buluq al-wahm the meaning "unconciliating (or: irritable) human nature" ("unvertragliches Wesen"). One must cling to the ordinary sense of buluq' "natural disposition, character, morality", and assume for afat- a meaning corresponding to the meaning of al-buluq al-wahm. The formula ba'idun min-a I-afati in Abu lJiras's line, which is to be interpreted in the sense of "protected (or: exempt, immune) from moral failures", is obviously frequently used to describe persons. We find this formula, e.g., in a poem by Durayd b. Simmah, in Abu Tammam's lfamasah (ed. Freytag), p. 379, v. 6, where Tabrizi (see ibid.) interprets it (in my opinion, incorrectly) as follows: yuridu

458

THE PRINCIPLE OF 'ISMAH

annahu Iii dii' a bihi wahwa salimu I-a'r;/ii' "he wants to say: that there is no disease in him, and that he is one of unhurt limbs", That the term al-iifiit- in such a context-i.e., when preceded by the expression ba'idun min-certainly means "the moral weakness, the inclination to sin", we learn from the fact that the term ba'idun min is also found in coordination with other expressions for "sin" (in analogy to the use of al-iifiit- in Niibigah no. 25,3, quoted above p. 456). We may refer, e.g., to Zuhayr's Mu'allaqah, v. 21 (= Diwiin, ed. Ahlwardt, no. 16,20): fa'a$bahtumii minhii 'ala bayri maw!inin ba'idayni fihii min 'uquqin wama'lami. Noldeke (Funf Mo'allaqat, III, p. 15) interprets this as follows: "So befandet Ihr [beiden] euch bald an der ersten Stelle, fern dabei von Unvertraglichkeit und Verschuldung", with the interpretation of 'uquq by "Unvertraglichkeit" ("incompatibility") not being quite adequate. 'Uquq is the "irreverent, sinful attitude towards relatives and kinsmen", and, in secondary usage, also "inimical, unjust behaviour" in general. Accordingly, we should interpret: "Thus both of you reached with respect to it (what is mentioned before) the most excellent position, hereby being far (i.e., exempt, immune) from irreverence and guilty behaviour". It should not be overlooked that the relation of the second hemistisch of Zuhayr's line -that is: ba'idaynifihii min 'uquqin wama'lami-to the first hemistich (Ja'a$bahtumii minhii 'alii bayri maw!inin) has not been entirely clarified, and that the meaning of this first hemistich (which has a parallel in Zuhayr's Diwiin [ed. Ahlwardt], no. 14,31) is per se not completely clear. As to the sense of maw!in in this type of context, we should pay attention to Zuhayr no. 1,39: wa'immii an yaqulu qad abaynii fasarru mawii!ini I-I:zasabi I-ibii'u. 3 The use of iifiit- (pI.) in the sense of "[disposition to] moral stumblings", also appears in an especially clear manner in passages like the following (al-Mutalammis, Diwiin, ed. Vollers, no. 2,49): al-gadru wal-iifiitu simatuhu "Treachury and moral transgressions are his natural disposition", i.e., "he cannot escape this sinful behaviour". (V oilers interprets: "Trug und Unheilstiften sind seine Natur").4 3 With this line in turn one must compare l:Iu\ay'ah, Diwiin (ed. Goldziher), no. 8, 6 (ZDMG 46, p. 210), which is probably influenced by it. 4 It is rather peculiar that the concept ba'idun min-a l-iifiiti, is ascribed to "a house (a tent)" (lmru'u'l-Qays, Diwiin, ed. Ahlwardt, no. 40, 14-15): (\4) wabaytin yafo/:lu l-misku fi /:Iagariitihi ba'idin min-a l-iifiiti gayri murawwaqi (l5)da!;altu 'alii baytjii'a gummin 'i:riimuhii ... "and many a house (a tent) entirely filled with musk-a (house) far (or: exempt) from the iifiit-its entrance not covered with a curtain-I entered

unto a "brilliantly white" woman, one with well-rounded (fleshy?) limbs ... ".

THE PRINCIPLE OF '/SMAH

459

Accordingly, the term iifiit- (pI.) is a pre-Islamic term for "moral shortcomings (transgressions), sins", Thus, the phrase 'a$amahu min-a I-iifiiti, in AbU Bakr's saying concerning the Prophet (see above p. 456), means: "he (i.e., God) has made him immune from moral stumblings, sins"; and the idea expressed in this phrase is in tum to be considered as based on a pre-Islamic idea, as present in the above-quoted preIslamic passages, especially in an-Nabigah's line in honour of the Gassanid kings (quoted above p.456). In this line it is said that "the bodies (agsiid) or" the persons concerned are 'purified' of sins" (mu{ahharat( an) min-a I-ma'aqqati wal-iifiiti wal-alami), which should be understood in the sense of "their being immune from sin". However, in distinction from AbU Bakr's statement concerning Mul).ammad's immunity against moral stumblings (al-iifiit-), in an-Nabigah's verse concerning the Gassanid kings it is not indicated that this "purification" ("immunity") is to be ascribed to God's interference. Most probably, what is meant-this does not, of course, become apparent from the text itself-is "an immunity (against sin) based on a natural disposition (quasi inherited from the encestors)". That such a natural, quasi biological disposition (not dependent on the will or intellect of the individual) is meant in an-Nabigah's line may be concluded from the fact that the freedom from sin (more literally: the state of "purity" (or "having been purified") is being ascribed to the "bodies" (agsiid) of the individuals described by an-Nabigah. 5 In the Islamic idea of Mul).ammad's (and other prophets') "immunity from sin" (,i$mah), this "immunity" is ascribed to an act of God. This idea may be considered as an easily understandable interpretation of the more primitive idea of an "immunity from sin" based on a natural (quasi biological) disposition. With respect to this, it is quite interesting that the idea of an "immunity against sin inherent to the body" obviously appears also in the Islamic discussions concerning the 'i$mah (of the prophets); cf. Tor Andrae, Die Person Muhammads in Lehre und Glaube seiner Gemeinde (Upsala 1918), p. 125. 5 J. Derenbourg, Journal Asiatique, 1868, p.334, interprets agsiid(uhum) in anNabigah's line as "persons": ..... leurs personnes sont pures de toute rebellion, de toute souillure (sic!, iifiit-) et de toute faute". That agsiid( uhum) is to be interpreted (in its literal meaning) as "(their) bodies", can be concluded from the fact that the word appears in the line beside the concept a~liim in the phrase a~liimu 'Adin: "the (prudent and mild) minds of the 'Ad" (cf. above p.456). The terms agsiid and a~liim in an-Nabigah's line clearly represent two contrasting concepts, supplementary to each other: "bodies and minds".

30 THE "COMPLETION" OR "IMPROVEMENT" OF A LAUDABLE DEED: AN ANCIENT ARAB ETHICAL MOTIP In several passages of the Qur'an we find the expression atamma llahu ni'matahu 'ala ... "God has 'completed' his benefaction to ... ". Siirah 5 (al-Ma'idah), v. 5: al-yawma akmaltu dinakum wa'atmamtu 'alaykum ni'mati waraqitu lakumu I-Islama dinan ... "Today I have 'perfected' to you your religion, and [thus] have 'completed' to you my benefaction, and have approved for you Islam as religion (i.e., by approving for you Islam as religion, I have completed my benefaction)". Siirah 12 (Yusuj), v. 6: wakarjalika yagtabika rabbuka wayu'allimuka min ta'wili l-alJiidiJi wayutimmu ni'matahu 'alayka wa'ala ali Ya'quba kama atammaha 'ala abawayka min qablu Ibrahima waIsl;aqa inna rabbaka 'alimun I;akimun "and thus your Lord will elect you and will teach you the interpretation of the riddles, and will complete His benefaction to you and to the family of Jacob, as He has completed it to your fathers before you, [namely, to] Abraham and Isaac; your Lord is all-knowing, wise". Siirah 2 (Baqarah), v. 145: ... wabsawni wali'utimma ni'mati 'alaykum wala'allakum tahtaduna "... and fear me, so that I may complete my benefaction to you. Perhaps you will walk the strait path". The idea that God "completes" ("perfects") the benefaction which He grants to men, is also found in the post-Qur'anic literature. In the following instance (words spoken by Muslim b. 'Uqbah, year 63 A.H.; in Tabari's Annales, II, 416,2) God's benefaction, as frequently occurs in Arabic literature, consists of His help in war and in His granting victory: fatimmu 'ala al;sani ma kuntum 'alayhi min-a !-{a'ati yutammim-i llahu lakum al;sana ma yunilukum min-a n-na$ri wal-falgi "Continue with your beautiful obedience (to God), then God will 'complete' the beautiful help and the victory which He grants you".

I

Originally published in Del' Islam, vol. 49 (1972), p. 273- 276.

THE "COMPLETION" OF A LAUDABLE DEED

461

Moreover, the idea of the "completion" of benefactions is also to be found in pre-Islamic sources. In Diwiin Zuhayr, ed. Ahlwardt, no. 15,37 (= ed. Landberg, with al-A'lam's commentary, in Primeurs arabes, p. 112) an ordinary human benefactor, a tribal chief lauded in the poem concerned, "completes" the benefactions which he grants: warji nfmatin tammamtahii ... "and to many a one to whom you have granted a benefaction, you have :completed' it ... ". With the verb atamma, tammama in this peculiar usage, we identify the verb rabba, which is used in a number of passages to express this type of idea. It appears, e.g., in a pre-Islamic passage, in a line of a poem by an-Nabigah ag-Qubyani in praise of king an-Nu'man b. Mungir of I:Iirah (an-Nabigah, ed. Ahlwardt, no. 3,20 = ed. Derenbourg, no. 7,20, lAs., 6e ser., t. 12, 1868, p.278): warabba 'alayhi lliihu a/:lsana fjun'ihi wakiina lahu 'alii l-bariyati nii!jirii. Derenbourg (loc. cit., p. 323) translates this line: "Puisse Dieu repandre sur lui ses meilleurs bienfaits et etre son allie contre tous ses ad versaires". We quote the interpretation of rabba given in the scholion to the line (Ibid., p.397): ay: atamma wa'afjla/:la, yuqiilu: fuliinun yarubbu fjfjani'ata 'inda fuliinin irjii kiina yutammimuhii wayu'akkidu "rabba: i.e., 'He completed and improved'; one says: 'Such-and-such "improves" ("ameliorates", "cultivates") (yarubbu) the benefaction which he grants someone', in case he 'completes' and strengthens it". In accordance with this definition of the term in the scholion, we interpret the line: "May God 'improve' to him His beautiful benefactions (that is: the ones He has granted him until now) [i.e., May He bring them to an end, strengthen them, add new benefactions to them]; and may He be his helper against [those among] men [who do not submit to him] (that is: May He grant him victory over them)". We assume that atamma lliihu ni'matahu 'alii in the Qur'anic passagesand yutammim( u) lliihu lakum a/:lsana mii yunilukum min-a n-nafjri wal-falgi in the quoted post-Qur'anic passage-imply the same idea, which in this pre-Islamic line is expressed by rabba lliihu a/:lsana fjun'ihi (= nfmatahu) 'alii ... (compare also a/:lsana fjun'ihi with a/:lsana mii yunilukum in the preceding passage). A tamma and rabba, which latter is in the scholia and lexica glossed by atamma (and also by afjla/:la) are to be considered as synonymous. We assume that both in the line by an-Nabigah and in the quoted Islamic passages, an ancient Arab idea primarily applied to humans has been transferred to the deity, with respect to the deity's relation to humans. We now quote the following Islamic passage where the subject

462

THE "COMPLETION" OF A LAUDABLE DEED

of the verb rabba in its special use in the meaning of a#al:1a "to improve", or atamma, tammama "to complete" (as in an-Nabigah's line)-is represented by a human being (al-Balaguri, Ansiib al-asriif, vol. 4 b, Jerusalem 1939, p. 16,8-14): kataba Yazidu i/ii bni" z-Zubayri yad'uhu ilii bay'atihi ... wakiina fimii kataba biM Yazidu: ... fa'U(jakkiruka lliiha fi nafsika fa'innaka riu sinnin min Quraysin waqad marjii laka salafun ~iili~un waqadamu ~idqin min-i gtihiidin wa'ibiidatin farbub ~iili~a mii marjii walii tubtil mii qaddamta min ~asanin fadbul fimii dabala fiM n-niisu ... fa'abii an yubiiyi'a ... "Yazid [b. MU'awiyah] wrote to ['Abdallah] b. az-Zubayr urging him to pay him homage [as caliph] ... and Yazid wrote inter alia: ' ... and I adjure you by God for your own sake: You are one of the elders [of the tribe of] Qurays, and you have to your credit a solid record and an excellent past of manly performance and of God's worship; so 'cultivate' (or: 'improve', 'complete') your solid record [in order to preserve it], and don't annul (or: destroy) your excellent performance of the past; and 'enter into that into which the people have entered' (i.e., pay homage to me as others have done) .. .'. But he refused to pay him homage ... ". The words Iii tubtil mii qaddamta "don't annul your excellent performance of the past" are obviously a variation of the preceding statement urbub ~iili~a mii marjii "'cultivate' (or: improve) your solid record" (more literally: "your past excellent actions"). Lii tub/if "don't annul" obviously implies the idea of "preserve ... "; and the "preservation" of a property is clearly the aim, and the result, of its "cultivation" and "amelioration". In Der Islam, vol. 36, 1961, p. 18ff., we stated that, in accordance with the genuine Arab conception, the actions and social (and manly) achievements of the individual or the group (i.e., tribe or clan) represent a capital or a property which, as any actual, material possession (or capital), is in constant need of "cultivation", indeed of "amelioration" (in Arabic i~/iib), in order to prevent its decline or loss (ifsiid). Inter alia, we have (I.e., p. 21) quoted a marginal gloss to a line of poetry in N aqii' irj Garir wal-Farazdaq (ed. A. A. Bevan), p. 947, 6 : ya~rilu I-magda: yagma'u I-magda wayaksibuhu, ay: yu~li~uhu, wa'i~­ lii~uhu an yarubbahu bifa'iilihi "'he ploughs the glory': he gathers glory and acquires it (cf. Der Islam, I.e., p. 12), that is to say: he ameliorates it by improving it (rabba = a~/a~a) through his deeds". Of course, the expression (occurring in a line of poetry) ya~rilu I-magda "he 'ploughs' [his] glory" (interpreted by yu~li~u and yarubbu) is itself a concrete expression of the idea that the "acquired" achievement

THE "COMPLETION" OF A LAUDABLE DEED

463

and the glory resulting from it must be continuously cultivated in order to grow to a considerable capital. While in the statement by Yazid b. Mu'awiyah (in the quotation from al-Balaguri's Ansiib al-asriif, see above p. 462) the verb rabba "to improve, to cultivate, to complete" is used with respect to the achievements performed by someone in the past (without attention to a person's having profited thereby), in the following utterance of the 'Abbasid caliph al-Mahdi (Tabari, Annales, III, 538,6ff.) the verb rabba is again (as in the line by an-Nabigah, and as tammama, atmama in other passages) used with respect to a benefaction: mii tawa~~ala ilayya abadun biwa~'ilatin walii tarjarra'a birjartatin hiya aqrabu min tarjkiri iyiiya yadan salafat minni ilayhi utbi'uhii ubtahii fa'ubsinu rabbahii ICanna man'a l-awiibiri yaq!a'u sukra l-awii'ili. De Goeje, in his Glossary to Tabari, Annales, s.v. rabba, interprets the expression ubsinu rabbahii as follows: "absana rabbahii 'tropice de beneficio curavit ut largum fieret'''. Accordingly, De Goeje understands ubsinu rabbahii in the sense of "I make it (i.e., the new benefaction) abundant", without assuming a specific relation to the earlier benefactions. We interpret the quoted passage as follows: "Never has someone entered into closer (bonds of) friendship with me than by reminding me of a benefaction which I have granted him in the past: [if someone reminds me of such an act of benefaction] I immediately let follow a similar one, and I improve (or: supplement, complete, preserve) it (i.e., the old benefaction) in a very thorough manner (not as De Goeje suggests: 'I make the [new] benefaction abundant'); for the refusal of subsequent [benefactions] cuts off the thanks for the earlier ones". The expressions utbi'uhii ubtahii and ubsinu rabbahii (an intensification of the expression arubbuhii) are here variations of the same idea. In the passage just quoted, the "completing" of the earlier benefactions (liter.: the first, al-awa'il) by the later ones (liter.: the last, al-awiibir) is ascribed to a human benefactor and the activity is expressly identified with the activity expressed by the verb rabba. In another passage we find the same idea applied to God as benefactor. This motif is found in a speech in which the Muslim armyleader an-Nu'man b. Muqarrin al-Muzani (appointed by 'Umar b. al-tJattab; year 21 A.H.; Tabari, Annales, I, 2623,4-7) demanded full devotion in battle from the Muslim warriors: ... wayaqulu: qatf. 'alimtum mii a'azzakumu lliihu biM min hiirjii d-dini wamii wa'adakum min-a ~-~uhuri waqad angaza lakum hawiidiya mii wa'adakum

464

THE "COMPLETION" OF A LAUDABLE DEED

wa!judurahu wa'innama baqiyat a *bal;zdati-hu > *bal;ztihu. After such a form like bal;zt- had developed, it suggested itself to conceive it as based on a root bl;zt, expressing the meaning of "being alone", or: "being unmixed", "being unique", etc., and to make it the basis of other, nominal and verbal, patterns. With respect to my assumption that the initial b of the South-Semitic root bl;zt (as, e.g., Ethiopic bal;zut "solus, unicus"; Arabic bal;zt- "pure, unmixed") is to be traced back to the preposition bi, and is certaily not based on a phonetic shift of an original w into b, I also mention the following forms of the Judaeo-Arabic dialect of Bagdad: (a )bbal;zdi "I alone", (a )bbal;zdu "he alone", (a )bbdl;zadnii "we alone", (a )bbdl;zdam "they alone", etc. (corresponding to "Classical" forms like wal;zdahu "he alone", etc.). Probably, analogous forms also exist in other dialects of 'Iraq and of other regions. It is evident that the initial geminated b (bb) in these 'Iraqi forms represents the preposition b(O, to which the initial radical w of waf:zd- has been assimilated: > bb. In any case, this 'Iraqi form represents a further basis for the etymology of the root bal;zuta (bal;zt-) as described above.

47 THE ORIGIN OF ARAMAIC IYMA "TO BE LIKE, TO RESEMBLE" The Hebrew and Aramaic verb dama (Hebrew), dema (Aramaic)quasi representing a root dmy-is in my opinion not based on an original (primary) root of Semitic. Beside the verb, there exist also some nominal forms of the same root, forms of the (basic) meaning "likeness": d emul(a) (Hebrew and Aramaic) and dumya (Aramaic). Arabic possesses a noun of the same meaning: dumyat-. It is generally assumed that this Arabic noun is a loan-word from Aramaic. According to Wellhausen, Prolegomena zur Geschichte Israels (6th ed., 1905), p. 388, also the Hebrew forms derived from this root-that is: the finite verb dama no less than the nominal form d emul-should be considered as having penetrated into Hebraic usage from Aramaic. Wellhausen's theory of the Aramaic origin of Hebrew dama "to be like", d emul "likeness" is certainly correct. However, Aramaic dema "to be like, to resemble" itself should not be regarded as a "primary" term of Aramaic (possibly rooted in Proto-Semitic); indeed, the verb can be shown as having developed from another term, which in turn is based on a different, original Proto- (and Common-)Semitic root. My assumption is that Aramaic dema "to be like" is based on a term received into Aramaic from Akkadian. The Akkadian term concerned is the verb-form (3rd pers. sing. masc.) iteme, the "reflexive" variation of the "basic" stem (designated as "1,2"-stem) of the verb ewu/emu (infinitive), which represents the Akkadian correspondence of the North-West-Semitic verb of the meaning "to be": Aramaic hewa, Hebrew haya (beside which there exist forms based on the root hwy). I thus assume that the (voiced) consonant d as initial radical of the Aramaic verb dema ("to be like") is rooted in the (unvoiced) consonant t, representing, in the underlying Akkadian form, not an (original) element of the root proper, but the characteristic formative of the reflexive stems (conjugations) in Semitic. In connection with my assumption that the original Akkadian form from which I consider Aramaic dema derived, is based on a root identical with the Aramaic/ Hebrew root hwy "to be", we must pay attention to the fact that in

THE ORIGIN OF ARAMAIC DEMA

557

Akkadian an original w was frequently shifted into m. Indeed, in the "conjugation" involved, the "I,2"-stem of the verb ewu/emu, only forms with m-as iteme, etc.-seem to be recorded. I refer to the (Chicago) Assyrian Dictionary (CAD), vol. 4 ("E"), p. 413, col. b (s.v. ewu/emu), and also to Ibid., p. 414, col. b. As to the meaning of the specific "conjugation" concerned, that is: iteme ("1,2"), it is to be noticed that it is identical with the meaning of the "basic" (non-reflexive) "conjugation" of the Akkadian verb involved (that is ["1,1"] ewu/emu), which is defined (see CAD, ibid., p.413, col. b) as "to change, turn into (intrans.)". Moreover, with respect to the construction of ewu/emu and iteme, I should mention that the complement of these (intransitive) verb-forms ("to change, to turn into something") appears obviously rarely in the accusative (see CAD, ibid., p.414, col. a, sub "1 a"), but extremely frequently (see Ibid., sub "1 b") as a kind of "adverbial complement" -augmented by the ending -is ("like") or (see Ibid., sub "1 c") preceded by the preposition kima or ki ("like"). I quote from CAD, ibid., p. 414 (sub "1 b") the following instances (with iteme, as well as with iwe, ime): i-we !iddis (as well as im-me fiddis) "(his enemy) turned into clay"; [ibri s]a arammu i-temi tiddis "my friend whom I love turned into clay"; (p. 414, col. a, ult.) alani asbUti karmes im-mu-u "the inhabited cities will turn into mounds". And also (again with iteme) (p.414, col. b, 1. 12) nidutam illikma i-te-me qaqqaris "(the temple of Nabu) became neglected and level with the ground"; furthermore (Ibid.,l. 19) Tiamat ... mabbutis i-te-mi (var. -me) usanni !ensa "Tiamat became raving mad, she lost her mind", to be compared to (Ibid., 1. 22): e-mu-u mabbutis. (For instances with kima, instead of -is, see Ibid., p.414, 1. 24-33; sub "1 c"). Of course, also the "transitive" variant of the root, expressed by the "causative" sumu ("to turn, change into" in its transitive use) appears in analogous constructions, as, e.g. (p. 415, col. a, 1. 17), u-se-me karmes "I turned (the enemy cities) into mounds", or with kima (p. 415, col. a, 1. 40): ... kima til abilbi u-se-mi (var. u-seme-su) "he (i.e., Assur) turned [his land] into mounds (left) by the flood". For the usage exemplified by the above quotations from CAD we may also refer to von Soden's Akkadisches Handworterbuch, vol. I, p. 266-267, where the transitive use of the verb ewu/emu, in its use as a "causative", characterized by the verbal-prefix s (sumu) , is characterized by the interpretation "machen zu, gleichmachen" (see l.c., p. 267 a, sub 3), that is: "to make equal (to)".

558

THE ORIGIN OF ARAMAIC DEMA

The primary meaning of the original root involved - that is: the Aramaic/Hebrew root hwy, Akkadian ewu > emu-is "to be", a meaning which may have been based on a more original meaning: "to become". We certainly can understand that such a basic meaning can in certain usages (and constructions), of which the above quotations from Akkadian texts are pertinent instances, lead to the meaning "to become (or: to be) equal to" (or: in a specific conjugation of the root concerned: "to make equal to"). In this context we may also refer to the West-Semitic (Aramaic, Hebrew and Arabic) root s/swy "to be equal to", which in turn represents a development Of a specific (quite analogous) use of the proto-Semitic root hwy ("to become, to be"). For the details of the derivation of s/swy I refer to above p. 540ff. The (Aramaic) synonym of the root swy-dmy, demii ("to be like", "to resemble")-is characterized by certain specific phonetic features. On the one hand, the root for "to be, to become" (> "to become like", "to become equal to"), on which I consider demii basedthat is: Akkadian ewu (= Aramaic/Hebrew hewii, hwy)-shows here the characteristic Akkadian shift w > m. On the other hand, the original t-prefix of the "reflexive stem" (iteme) survives in demii (representing in it the first radical of a [newly developed] root) as d; that is: the unvoiced consonant t, having lost its original formative function and having developed into a component of the verbal root itself, has become "voiced", a phonetic change which cannot be considered as actually striking. The penetration into Aramaic of an Akkadian term (a verb) with a "plain" meaning - that is: a term which does not signify a concept characteristic of the civilization involved-should certainly not be considered as strange or striking in any way. The contact between Aramaic and Akkadian did not consist merely in a more or less frequent contact between two peoples (or two civilizations), but Aramaic was received by a large section of the original speakers of Akkadian in a relatively early period as their every-day language. It can, however, not be doubted that the original language survived subconsciously during a long period and certainly influenced the newly acquired language to a considerable extent.

48 VERBS DERIVED FROM THE NOUN *QAS/ST "BOW" The Arabic verb qa$ada "to aim at", "to intend", "to move towards" has obviously no parallel in the related languages. In my opinion, it is, however, derived from a Common-Semitic (and ProtoSemitic) noun. I consider the Arabic verb qa$ada a denominative of the Common-Semitic noun for "bow", that is: Akkadian qastu; Hebrew q~S~l « *qast); Targumic qastii, Syriac qestii; Ethiopic qast. In Arabic, the noun appears in a somewhat modified form: qaws (pI. qusiy, qisiy). That means: the Arabic form of the noun for "bow" does, on the one hand, not include the consonant t (which is a basic root-element in all the other languages), and is, on the other hand, characterized by w as medial radical: qaws (although its most frequently used plural: qusiy, qisiy, is quasi based on a root with w as the third radical). My derivation of the Arabic verb qa$ada "to aim at", as indicated above, is based on the supposition-or, what is more, shows clearlythat also in Arabic the noun for "bow" existed originally in the form present in the other Semitic languages, that is: characterized by the consonant t as 3rd "radical", a "radical" which indeed, originally, may have constituted the feminine-ending (-a)t. G. Hofmann, ZDMG 32 (1878), p.762, derived Arabic qawsun from *qaswun, basing this derivation on the plural-form qusiyun. Brockelmann, Grundriss, I, § 71 b a (p. 190), assumed a direct phonetic relationship between the Arabic form qaws and the form of the other languages: qas/st. In his opinion, the Arabic form indicates that the form qas/st is based on an original form *qaws/st (*qaljs/st), with the terminal t, of course, representing the feminine-ending. (Concerning the elimination of the medial w, as alleged by him, he refers to Arabic lastu/a as against laysa). This interpretation, by Brockelmann, of the noun qas/st is from the outset very improbable, and certainly so in view of the existence of the Arabic verb qa$ada "to aim at", which, because of its derivation (see above) hints in its turn to the original existence of a nominal form *qast in Arabic. In view of the above described derivation of the Arabic verb qa$ada

560

VERBS DERIVED FROM THE NOUN *QASIST

"to aim at" from the noun *qast "bow", its second radical !j and its third radical d should be considered as based on an original sand t, respectively. As far as the !j instead of an original s is concerned, it can be easily explained as an "emphatisation" of the s in contact with the preceding emphatic consonant q, as, e.g., *yaqsitu > *yaq!jitu. With respect to the presence or the lack of consonantal "emphasis" in roots of this type, we may refer to the secondary variants of the primary root concerned (qst) in Syriac: kesa! "arcu jecit" (with k instead of q, and with ! instead of t; see Brockelmann, Lex. Syr. 2, p. 340 b), and. also (used in the same meaning) qesa! (with! instead of the original t; see Brockelmann, Ibid., p. 703 a); cf. also Syriac qassii!ii "archer" (with ! instead of t), as well as the synonymous kassii{ii (with k instead of q and! instead of t). In this context, we should also pay attention to the "fourth stem" (af'ala) of the root q!jd. I refer to Lane's Arabic dictionary, p.2531 (vol. 4), s.v. q!jd: "aq!jada: It (an arrow) hit its object, and killed on the spot. '" He pierced a man with a spear or shot him with an arrow, and did not miss him". This specific use of the root q!jd is quite interesting in view of its derivation from the noun for "bow", *qast-, as suggested above. Certainly, not only aq!jada (the "fourth stem"; af'ala), but also the basic qa!jada, with its infinitive qa!jd-, is used in this sense. We may refer in this connection to the following line of poetry ascribed to the poet I:I~itim at- Ta'i: kalaYJin hizabrin kiina ya/:tmi dimiirahu ramathu l-maniiyii qa!jdahii jataqa!!arii "he (the lamented hero) was used to defend his honour like a strong lion: Fate shot at him with its arrows (and its bow), not missing him; and he fell on the ground, lying prostrated". (See for this verse Na~r b. Muzal).im al-Minqari's Waq'at Sifjin [2nd ed., Cairo 1332 h.], p.247, 3-4; and cf. also ad-Dinawari's al-Abbiir a!-{iwiil red. Cairo 1960], p. 176,9). The accusative qa!jdahii serves here as a kind of "cognate object" (maf'ul mu{laq) to the finite verb ramat(hu}. Or should qa!jd in qa!jdahii [ramathu l-maniiyii qa!jdahii] indeed reflect an original Arabic qa!jd- [< *qast-] in the sense of "bow": "Fate shot at him with its bow"? A further Semitic verb which should be considered a denominative of the word for "bow" and be directly compared with Arabic qa!jada "to aim at", I recognize in the Akkadian verb kasiidu, for the various meanings and uses of which I refer to The (Chicago) Assyrian Dictionary, vol. 8 ("K"), p. 271 ff.: "1. to reach, to arrive (said of a moment in time), ... 2. to conquer a country, a city, to defeat an

VERBS DERIVED FROM THE NOUN *QASIST

561

enemy, to be victorious, to capture an enemy ... , to seize (said of diseases, evil spirits, misfortunes, etc.), ... , to get hold of (in various shades of meaning), ... ". That is: Akkadian kasiidu is to be interpreted as derived from qast(um) "bow". Accordingly, the original third radical t was (quasi by a spontaneous sound-shift) shifted into d (in analogy to the development of the consonant d of Arabic qa!jada, based in its turn on *qast- "bow"). And the original ("emphatic") q was shifted into k, a sound-shift also observed in forms derived from the word for "bow" itself in Syriac, see above p. 560, line 8 and line 13. Also this change of the initial q into k may be interpreted as constituting a spontaneous sound-shift. In conclusion I mention a further specific etymon which I consider derived from (a variation of) the specific root-form present in the Arabic verb qa!jada "to aim at" (itself to be derived from the ancient Semitic word for "bow": qas/st-). I have in mind the Aramaic noun of the meaning "truth", "the right thing", which appears in Bibl.-Aramaic as qeSO!, in Syriac (ordinarily) as qustii, in ludaeo-Aramaic as qu§rii, in Mandaic as kustii, and is also represented in Hebrew, as qeSO! (loan-word ?). In Arabic it appears as qis! (usually regarded as a loan-word from Aramaic). To be mentioned is also the ludaeo-Aramaic (and late Hebrew) verbal root q§r and k§r "to arrange, to trim" and also "to adorn" (afel and pa'el). The various consonantal shifts (as they appear in the various forms mentioned)-! instead of t, k instead of q-are analogous to the consonantal shifts observed in various secondary forms of the original noun for "bow" (cf. above p. 560). The meaning "that which is right", "the truth" may of course be considered as a derivation from the verbal concept "to aim at": Arabic qa!jada (and also the verbal concepts "to arrange, to trim, to adorn" may be derived from a concept like "to aim at"). With respect to the change of meaning here involved ("the truth", "the right thing" from "to aim at"), we may (inter alia) refer to the analogous semantic shift involved in the development of the Arabic noun !jawiib "that which is right", derived from !jiiba (also a!jiiba) "to hit [the target]" (and also "to be right"). We may, however, go farther and interpret as a variation of the specific verbal root represented by Arabic qa!jada "to aim at" (derived from Proto-Semitic qas/st- "bow") also the Arabic, Hebrew and Aramaic root !jdq (in Syriac zdq) "to speak the truth", "to be right" (Arabic), "to be just" (Hebrew and Aramaic). Arabic !jidq "truth", Hebrew

562

VERBS DERIVED FROM THE NOUN *QASIST

«justice», e.g., may thus be considered as (quasi) a very early secondary variation of Arabic qis! (= Aramais qu§{ii). With respect to this suggestion to consider the root ~dq as derived from the root q~d (as existing in Arabic), I am referring also to the fact that Freytag in his Lexicon Arabico-Latinum, vol. 3, p. 457 a, gives for qa~ada (beside the meaning "intendit" and certain other variant meanings) also the following meanings (based on the information transmitted by medieval Arab lexicographers): (sub "3") "medio modo se habuit ... " and (sub "4") "iustitia usus fUi!" (with which one may compare [Ibid., p. 442 b]: qasa!a "iustitiam exercuit"). As concerns the structural (phonetic) relationship of the root ~dq (