You are on page 1of 755

SEMITIC LANGUAGES OUTLINE OF A COMPARATIVE GRAMMAR

ORIENTALIA LOVANIENSIA ANALECTA • 80

SEMITIC LANGUAGES OUTLINE OF A COMPARATIVE GRAMMAR

BY

EDWARD LIPINSKI

UITGEVERIJ PEETERS en DEPARTEMENT OOSTERSE STUDIES LEUVEN 1997

C I P Koninklijke Bibliotheek Albert I, Brussel L I P I N S K I Edward Semitic Languages: Outline of a Comparative Grammar. — Leuven: Peeters, 1997. — 756 p.: ill., 24 cm. — (Orientalia Lovaniensia Analecta: 80). © 1997, Peeters Publishers & Department of Oriental Studies Bondgenotenlaan 153, B-3000 Leuven/Louvain (Belgium) A l l rights reserved, including the rights to translate or to reproduce this book or parts thereof in any form. D . 1997/0602/48 I S B N 90-6831-939-6 (Peeters, Leuven)

To MALGORZATA

CONTENTS

PREFACE ABBREVIATIONS AND SYMBOLS I . SEMITIC LANGUAGES 1. Definition 2. Afro-Asiatic A. Egyptian a) Old Egyptian b) Middle Egyptian c) Late Egyptian d) Demotic e) Coptic B. Cushitic a) Bedja b) Agaw c) East Cushitic d) West and South Cushitic C. Libyco-Berber D. Chadic 3. Proto-Semitic . .

17 21 23 23 24 25 25 26 27 27 29 29 31 32 32 33 34 39 41 47 50 50 52 53 53 53 54 55

4. Classification of Semitic Languages 5. North Semitic A. Palaeosyrian B. Amorite C. Ugaritic 6. East Semitic A. Old Akkadian B. Assyro-Babylonian C. Late Babylonian

8

CONTENTS

7. West Semitic A. Canaanite a) Old Canaanite b) Hebrew c) Phoenician d) Ammonite e) Moabite f) Edomite B. Aramaic a) Early Aramaic b) Official or Imperial Aramaic c) Standard Literary Aramaic d) Middle Aramaic e) Western Late Aramaic e) Eastern Late Aramaic f) Neo-Aramaic C. Arabic a) Pre-Islamic North and East Arabian b) Pre-Classical Arabic c) Classical Arabic d) Neo-Arabic or Middle Arabic e) Modern Arabic 8. South Semitic A. South Arabian a) Sabaic b) Minaic c) Qatabanic d) Hadramitic e) Modern South Arabian B. Ethiopic a) North Ethiopic Ge'ez Tigre Tigrinya b) South Ethiopic Amharic Argobba Harari

56 56 57 57 58 60 60 61 61 61 63 63 63 65 66 69 70 71 72 75 75 77 78 78 79 80 80 80 80 81 83 83 83 84 84 84 84 84

,

j

i

CONTENTS

9

Gurage Gafat 9. Language and Script A. Cuneiform Script B. Alphabetic Script C. Transcription and Transliteration II. PHONOLOGY 1. Basic Assumptions A. Linguistic Analysis B. Consonantal Sounds C. Vowels D. Intonation E. Phonemes F. Voiced and Unvoiced Sounds G. Emphatic Sounds H. Proto-Semitic Phonemes 2. Labials 3. Dental Plosives 4. Interdentals 5. Dental Fricatives 6. Prepalatal and Palatal 7. Laterals 8. Liquids and Nasal 9. Velar Plosives 10. Laryngals, Pharyngal and Velar Fricatives 11. Synopsis of the Consonantal System 12. Vowels 13. Diphthongs

85 85 86 86 87 93 95 96 96 99 100 102 103 104 105 106 109 116 117 122 126 129 132 137 141 150 152 166

10

CONTENTS

14. Geminated or Long Consonants 15. Syllable 16. Word Accent 17. Sentence Stress or Pitch 18. Conditioned Sound Changes

173 178 181 184 186

A. Assimilation 186 a) Assimilation between Consonants 187 b) Assimilation between Vowels 190 c) Assimilation between a Consonant and a Vowel . 190 B. Dissimilation C. Metathesis D. Haplology E. Prosthesis F. Anaptyxis G. Sandhi H. Elision I . Hypercorrection IĪL MORPHOLOGY 1. The Root Morpheme 2. The Noun A. Noun Stems or Patterns a) Simple Patterns b) Patterns with Diphthongs c) Patterns Extended by Gemination d) Patterns Extended by Reduplication e) Patterns with Preformatives and Infixes . Preformatives '-/'Preformative yaPreformatives w-lm-lnPreformative tInfix -tPreformative šf) Patterns with Afformatives Afformative -ān .191 192 193 194 195 196 196 199 201 201 209 209 210 212 213 214 215 215 216 216 219 220 221 221 221

.

.

.

CONTENTS

11

Afformatives -iy/-ay/-āwī/-yal-iyya Afformatives in -t Other Afformatives Afformative -ayim/n of Place Names . . . . g) Nominal Compounds B. Gender C. Number a) Dual b) Plural External Plural Plural by Reduplication Internal Plural c) Paucative d) Collective Nouns e) Singulative D. Case Inflection a) Diptotic "Ergative" Declension b) Use in Proper Names c) "Classical" Triptotic Declension d) "Adverbial" Cases e) Historical Survey of Case Inflection E. The "States" of the Noun a) Construct State b) Predicate State c) Determinate State d) Indeterminate State e) Paradigms F. Adjectives G. Numerals a) Cardinals b) Ordinals c) Fractionals d) Multiplicatives e) Distributives f) Verbal Derivatives 3. Pronouns A. Independent Personal Pronouns B. Suffixed Personal Pronouns . .

223 225 226 228 228 229 235 236 238 238 244 245 251 251 252 253 254 258 259 260 262 265 265 266 267 272 274 278 280 280 292 294 295 296 296 297 298 306

I
!

12

CONTENTS

C. Reflexive Pronoun D. Independent Possessive Pronouns E. Demonstrative Pronouns F. Determinative-Relative Pronouns G. Interrogative and Indefinite Pronouns 4. Verbs A. Preliminaries B. Tenses and Aspects a) Fully Developed System b) Simplified Systems c) Transitivity — Intransitivity d) Modern Languages C. Moods D. Actor Affixes a) Suffix-Conjugation b) Imperative c) Prefix-Conjugation Set I Set I I E. Stems and Voices a) Basic Stem b) Stem with Geminated Second Radical Consonant. c) Stem with Lengthened First Vowel d) "Causative" Stem e) Stem with rc-Prefix f) Stems with i-Affix g) Frequentative Stems h) Reduplicated Biconsonantal Stems i) Stems with Geminated or Reduplicated Last Radical. j ) Other Stems k) Verbs with Four Radical Consonants 1) Passive Voice m) Recapitulation of Stems F. Infinitive and Participle a) Infinitive b) Participle c) Neo-Aramaic Verbal System d) Participial Tense Forms in Other Languages

311 312 315 324 328 331 331 335 335 340 343 346 351 359 359 366 368 369 376 378 378 382 385 387 393 395 402 405 406 407 407 408 409

.

415 415 419 421 . 424

CONTENTS

13

G. Particular Types of Verbs a) "Weak" Verbs b) Biconsonantal Verbs c) Verbs with Pharyngals, Laryngals, Velar Fricatives . H. Verbs with Pronominal Suffixes 5. Adverbs A. Adverbs of Nominal Origin B. Adverbs of Place and Negatives C. Adverbs of Time 6. Prepositions A. Primary Prepositions B. Prepositions of Nominal Origin C. Compound Prepositions 7. Connective and Deictic Particles A. Conjunctions B. Presentatives C. Subordinate Conjunctions D. Copulae E. Expression of Possession IV. SYNTAX 1. Classes of Sentences A. Minor Clauses B. Major Clauses C. Nominal Clauses D. Verbal Clauses E. Concord of Subject and Predicate 2. Nominal Phrases A. Attribute B. Apposition C. Genitival or Subjoining Relation

425 425 436 445 450 453 453 454 458 459 460 465 469 470 470 472 474 475 480 481 483 483 484 484 487 491 494 494 496 497

14 3. Verbal Phrases A. Accusative B. Infinitive 4. Clauses

CONTENTS

504 504 508 511 511 515 519 521 527 533 535 536 543 545 554 557 564 567 568 570 . . . . 575 593 593 597 598 600 601 605 610

A. Particular Types of Main Clauses B. Parallel Clauses C. Subordinate Clauses a) Relative Clauses b) Temporal/Causal Clauses c) Final/Consecutive Clauses d) Substantival Clauses e) Conditional Clauses V. LEXICON 1. Etymology 2. Derivatives 3. Languages in Contact 4. Internal Change 5. Proper Names A. Anthroponomy B. Toponymy GLOSSARY OF SELECTED LINGUISTIC TERMS BIBLIOGRAPHY 1. Semitic Languages in General 2. North Semitic 3. East Semitic 4. West Semitic A. "Canaanite" B. Aramaic C. Arabic

Syriac Argobba Assyro-Babylonian. Ethiopic 6. South Arabian B. Numidic. Cushitic 8. Old Akkadian . Neo-Aramaic. Late Babyloninan. Mandaic. Bantu Bedja Chadic Coptic East and West Cushitic Egyptian Gafat Ge'ez Greek Gurage Harari Hausa Hebrew Hittite Latin Libyco-Berber. Tuareg 617 617 619 622 624 627 629 633 639 681 681 681 684 684 685 696 702 702 713 713 714 714 714 715 716 717 720 721 724 725 725 731 731 731 . Anthroponomy and Toponymy GENERAL INDEX INDEX OF WORDS AND FORMS Agaw Amharic Ammonite Amorite Arabic Aramaic. . Chadic 9. .CONTENTS 15 5. South Semitic A. Languages in Contact 10. Libyco-Berber 7.

Epigraphic South Arabian. MAPS.16 Moabite North Arabian Oromo Palaeosyrian Persian Phoenician and Punic Rendille Semitic. Common Somali South Arabian. Modern Sumerian Tigre Tigrinya Ugaritic CONTENTS 734 735 735 736 738 738 740 740 742 742 744 745 746 748 749 753 TABLES. A N D TEXT FIGURES .

from the point of view of comparative linguistics. viz. directed towards . it was felt in different quarters that it is important to draw the attention of the students to cer­ tain tendencies discernible in modern dialects and to clearly bring out the main common features of Semitic syntax. and it abstained from a systematic treatment of the syntax and of semantic problems. I decided finally to acquiesce to a long-standing suggestion and to undertake the task of publishing the results of my research and teaching in the form of a textbook. the usefulness of an outline of a com­ parative grammar of the Semitic languages is self-evident since the last original work of this kind was published twenty-five years ago by B. but it is nev­ ertheless intended primarily as an introductory work. Syriac. and Ge'ez. but paid little attention to other Semitic languages. In addition. where he gives some paradigms showing the connections between Semitic and other Hamito-Semitic languages. Zimmern published his Vergleichende Grammatik der semitischen Sprachen. Brockelmann's famous Grundriss and its epigones seem to neglect this type of comparisons. although C. Akkadian. the Afro-Asiatic language families "cannot be studied. This work was based mainly on the so-called classical Semitic languages. The scope of the present Outline is thus larger. Yet. both ancient and modern. the material has increased considerably during the last decades and the need for a synthesis taking the new information into account was growing steadily. which resulted in the first comparative grammar of the Semitic languages ever published. that the one of earlier comparative grammars of the Semitic languages. Classical Arabic. (Moscow 1972). BBeAeirae B cpaBHHTejiLHoe royneHHe CCMHTCKHX JOHKOB. Diakonoff rightly stressed in 1988. comparative Semitics without a broader Afro-Asiatic or HamitoSemitic background is — in some areas at least — methodologically questionable. Grande. the present book owes a similar approach to itself.M. In fact. Finally. Besides. in isolation from each other". in a certain sense. the right approach was already outlined in 1898 when H. However.M. Biblical Hebrew.PREFACE Having taught the introduction to the Semitic languages and their comparative grammar for more than a quarter of a century. year by year. Designed to come out in the centenary of the completion of Zim­ mern's work. as I.

It might also be useful to stress at the outset that the present work is intended as a compendious and up-to-date analysis of the nature and structure of the Semitic languages. Z. Johnstone. Harris. that the views exposed in this book differ some­ times from the opinions expressed by the above-mentioned Semitists and by other scholars. von Soden. and thus to corroborate our views by quoting litera­ ture in extensive notes. I . we do not attempt to apply the latter's arsenals of techni­ cal vocabulary to the Semitic languages. of students of linguistics. It is a comparative analysis of a lan­ guage family. at the end of the vol­ ume. Kutscher. especially from spoken languages and dialects. M . To avoid an excessive overloading of the text. Its aim is to underline the common characteristics and trends of the languages and dialects that compose the Semitic language "family" by applying the comparative method of historical linguistics. on the one hand. aimed at lessening the possibility of misinterpreta­ tion. it is inevitable that incon­ sistencies w i l l appear in the transliteration and the spelling of Afro-Asi­ atic words and phrases. J. Beeston. Diakonoff. W. R.F. Cantineau. W. Littmann. For a more detailed presentation and analysis of linguistic data. No Semitist can be assumed today to be at home in all the Semitic idioms. W. we must ask the user's indulgence. The object it has in view is not a mere juxtaposition of forms belonging to various languages.S. but rather to present as clearly . Segert. i f some part of the evidence is not to be veiled. I. a selective list of which is given in the bibliography. Gelb.Y. on the other. Fischer. how­ ever. references are given. as a rule. E. T . Nevertheless.L. especially of A. and. of students of one or several Semitic languages. It is clear. the advanced students should rather refer to specific grammars. Macuch. and the present work relies to a great extent on publications of other scholars. The selection of linguistic facts and the degree of their condensation may also be subject to discussion and to criticism. only when they cannot be found easily in current grammars of the particular languages. For such occasional lack of uniformity and for certain redundancies.J. not a comparative study of the views expressed by com­ peting linguistic schools: Semitics is more wonderful than linguistics! Consequently. but a comparison and an explanation of the changes they incurred. M . E. S. seen in both a diachronic and a synchronic perspectives which must be used together. we deemed it unwise to explain here at full length why the preference was given to certain theories to the exclusion of others. of course. Leslau.18 PREFACE an audience consisting. In view of the great variety and intricacy of the material presented.

vowels. nominal and verbal phrases. Etymology. to the sequence in which words are arranged in a sentence. both fixed and free orders are found mingled in widely varying proportions in a great number of Semitic languages. derivatives. Hebrew. the pronouns. The presentation of the basic assumptions is followed by a synchronic and diachronic description of the consonants. the sentence stress. particular types of main clauses. as well as Arabic. After a preliminary section deal­ ing with the problem of the Semitic root. Part One is introductory. Old Akkadian. and Chadic. In fact. Egyptian. Phoenician.PREFACE 19 as possible the fundamental insights about the wide world represented by the history and the present reality of the concerned language family. by a selective bibliography. The last section of Part One deals with the problems of language and script. the coordinative and deictic parti­ cles are examined from a diachronic and synchronic point of view. coordinate. and the conditioned sound changes are examined in this part as well. Libyco-Berber. and Epigraphic South Arabian. Part Two is devoted to phonology. inter­ nal change. the prepositions. the nouns. the single languages of which are briefly described. Neo-Aramaic. with questions such as classes of sentences. languages in contact. It situates the Semitic languages in the wider context of Afro-Asiatic or Hamito-Semitic language family. Cushitic. includes such languages of antiquity as Palaeosyrian. Questions related to the sylla­ ble. i. Part Four treats of the main features of Semitic syntax. The Semitic group. the verbs. Part Three concerns the morphology. Aramaic.e. I have prof­ ited in particular from a number of questions raised by my Kurdish students and from the constructive comments of those who have fol­ lowed my seminars in the Department of Epigraphy at the Yarmouk University. and the con­ temporary languages of Ethiopia and Eritrea. proper names — these are the main questions examined in this part. Assyro-BabyIonian. the word accent. . It is a pleasure to acknowledge my gratitude to the many classes which have inspired the successive drafts of this grammar. the five main branches of which are Semitic. by a general index. Part Five aims at presenting some fundamental insights about lexico­ graphical analysis. and diphthongs. the adverbs. It is followed by a glossary of linguistic terms used in Semitics. parallel. and subordinate clauses. and by an index of words and forms. Diachronic factors come here distinctly to the fore in relation to word order.

whose skilful care is apparent over again in the way this book is printed and edited. . I cannot let go unexpressed my deep appreciation for the work realized by Peeters Publishers and the Orientaliste typography. Last but not least. I must thank my wife Malgorzata for help­ ing me to bring this work to a happy end. Malha for the great care and professional skill which she exercised in preparing the text for printing.20 PREFACE I also wish to express my sincere thanks to Mrs F. Further.

. E. POSENER. Judg. Is. Lev. Coll.. Cl..F. = feminine Gen.. masc. KevelaerNeukirchen-Vluyn 1978. Jon. Leipzig 1915.. II Chr.S. gen. SANMARTÍN. I En.Ar. I Sam. = masculine M. Ras Ibn Hani and Other Places {KTU: second. Assyro-Babylonian circa.. = millisecond(s) Ms. Neh.ABBREVIATIONS AND SYMBOLS The Books of the Bible Gen. Masc. Hag. compare Classical Arabic Colloquial corrected.... Miinster 1995.. Zech.... ca.-Bab.A. Arab.. Ez. = Hebrew KTU = M.S. El Amarna Tablets 359-379 (AOAT 8).. Princes et pays d'Asie et de Nubie. ARM Ass. DN E = = = = = = = = = = = = = = . Die El-Amarna-Tafeln (VAB 2).. Bruxelles 1940. about Consonant confer... Paris 1950 ff. Soph. cor. EA = The El-Amarna tablets numbered according to J. Other Abbreviations accusative Amorite Arabic Aramaic Archives royales de Mari. 2nd ed. fern.. mss.. Qoh. II Sam.. Cant. = = genitive Hebr. for example e.. enlarged edition). Jer. m. Act.. Amor. I Chr.J. Esd. acc. corrects in divine name Egyptian execration texts published by G.Ar. Mai. = Modern Arabic M. f. Mich. Prov.. Am. etymologically lit.. Esth. = manuscript(s) Ace. II Kings.g. Joel. Ruth. Ob. A.... Hab.. Job. KNUDTZON.. Sir. Deut..O.. RAINEY.. C cf..A. Aram. = exempli gratia. The Cuneiform Alphabetic Texts from Ugarit. = literally. Lam. I Kings. Dan. = Modern South Arabian msec. Hos..... DIETRICH . LORETZ . Jos.. = Epigraphic South Arabian Fern.A.. Nb. Ps. Ex. Nah..

Lists. = Texts from Qumrān grot 1. etc. Accounts. : the colon indicates length in linguistics.Akk. it is generally replaced by the macron in the present Outline'. Paris 1905-68. = Ugaritic v = vowel v = long vowel vs. Oxford 1982. enclose phonetic approximations or reconstructed parts of a text. Jeru­ salem 1989. signifies that the preceding form has developed from the following one. PORTEN . II. = Old Akkadian O. = plural PN = personal name Pr. Oxford 1973. = nominative n. = versus. = person Plur. nom. in cuneiform texts. Hebrew and Moabite Inscriptions. e.. Literature. LUGAL small capital letters indicate logograms. = singular TAD = B. // parallel with. symbols. III.Syr.. syllables. Letters.22 ABBREVIATIONS AND SYMBOLS Norn.Bab. enclose words not found in the original. "god".s. indicates form or vocalization supposed. 3. ? dubious reading or interpretation. Sing. Jerusalem 1986. d ki uru 1 // t] () * < > . Textbook of Syrian Semitic Inscriptions I. II. words. when placed between two letters.A. I I I . signifies that the preceding form develops or has developed into the following one. GIBSON. pers.. "country". Aramaic Inscriptions. but needed in the translation. determinative URU. etc. abbreviation of the determinative DINGIR. Contracts.g. 2nd ed. = Palaeosyrian 1Q. against Symbols. 2. 3Q..C. = Proto-Semitic P. Jerusalem 1993. because of a new reading. postpositional determinative KT. Phoenician Inscriptions. sing. = Old Babylonian Pers. appellations. Determinatives enclose phonemic transcriptions. Textbook of Aramaic Documents from Ancient Egypt I. + joins lexemes or morphemes forming one word. Ugar.. RES = Repertoire d'Épigraphie Sémitique. sumerograms. Oxford 1975. 2Q. hyphen used to connect the elements of certain compound words. as well as cuneiform and hieroglyphic "syllabic" graphemes pertaining to one word.L. TSSI = J. / indicates alternative forms.-Sem. etc.. but not attested as such in texts. in cuneiform texts. ! to be especially noticed. plur. = new series O. "city' . YARDENI. in cuneiform texts.

C. considering that this is the only language family represented in both Africa and Asia. and Aramaic. 617-626). London 1844. belong to the same large language family. J.H. but lately better known as Afro-Asiatic. i. but his intuition connecting the languages of this group with another branch of Afro-Asiatic. and was then followed by J. Benfey in his sole work on Semitic linguistics: Ueber das Verhaeltniss der aegyptischen Sprache zum semitischen Sprachstamm (Leipzig 1844). where he expresses the opinion that also Berber and "Ethiopic". where Miiller describes the concerned group of languages. Prichard's Researches as to the Physical History of Man (vol. 10.I SEMITIC LANGUAGES 1. Ibn Quraysh is rightly regarded as one of the forerunners of com­ parative Semitic linguistics. in his work known as Rìsāla. Cushitic in his terminology. Newman who had appended a note on Hausa in the third edition of J.e. at least in some particular cases. proposed . did not yield fruit before the 19th century.1. and the entire language family was named "Hamito-Semitic" in 1876 by Fr. The designation "Cushitic" was introduced by 1858.D. As for Hausa.F. They are spoken nowadays by more than two hundred million people and they constitute the only language family the history of which can be followed for four thousand five hundred years. instead. The existence of a relationship between Berber in North Africa and Semitic was perceived already in the second half of the 9th century A. They form part of a larger language group often called Hamito-Semitic. A broader interrelationship was first recognized by Th. it was related to this group in the very same year by T. they do not stand isolated among the languages of the world. p. from Tiaret (Algeria).L. Miiller in his Grundriss der Sprachwissenschaft (Wien 1876-88). Schlcezer in J. p. the best known of the Chadic languages. by Judah ibn Quraysh. Greenberg. Eichhorn's Repertorium fuer biblische und morgenlaendische Literatur (vol. based an Arabic.N. I V . 161) because they were spoken by peoples included in Gen.21-31 among the sons of Sem. Schon in the latter's Gram­ mar of the Hausa Language (London 1862). The "Semitic" languages were so named in 1781 by A. However. Hebrew. DEFINITION 1. V I I I .G.

gain. "to cause to teach". etc. Afro-Asiatic would more or less correspond to the group of Indo-European languages. Swahilipatiliz-ana. "to cause to shut").g.g. the reciprocal verb suffix -án(e. "teacher". M . A F R O . The pertinent observations are restricted here to the prerequisites necessary for an understanding and a reconstruction of Semitic linguistic history. are limited to Mesopotamia. of the verb "eat". For example. 1. 2. A reference to these languages will be made only occasionally. besides the Semitic family which will be described below.3. dating back to the third millennium B. as first shown by I . The links of Afro-Asi­ atic with the great Bantu linguistic stock of Central Africa seem to be more precise. Whereas the relation between the various Semitic languages can be compared with that of. as indicated e. say. "to be striking one another". predicative). issued in 1963.A S I A T I C 1. the various Germanic or Romance or Slavic languages.g. Moscow 1965) who reached the important conclusion that Afro-Asiatic belonged originally to an ergative language type. but these are scarcely sufficient to warrant assumption of any genetic connection. anyhow. Kwena murút-i.g. and in the Horn of northeastern Africa. e.g. The languages that belong to the Afro-Asiatic group are classified in four main families. and the causative suffix -is-1 -is. while . there is a structural analogy between Afro-Asiatic and the Cau­ casian languages. and Egypt.C. Kwena hu-rút-ís-á. Diakonoff (Semito-Hamitic Languages.. On the other hand. can lead to meanings like " w i n . parts of the body are often used as prepositions and the extended metaphoric use of words. although there are many features of semantics and idiom which are common to African languages and to Semitic. character­ ized by the opposition of a casus agens (nominative.(e. Sotho ho-op-án-á. Swahili fung-is-a. instrumental. A more detailed approach is unnecessary. The latter have a few points of contact with Afro-Asiatic. this topic is outside the scope of the present study. loca­ tive) to a casus patiens (accusative. "to vex one an other"). North Syria. by the noun prefix mu. since comparative Egypto-Semitic linguistics is still in its infancy. but their oldest written attestations. in North Africa. use".2. The languages in question are spoken nowadays in Western Asia.(e.24 SEMITIC LANGUAGES to call it Afro-Asiatic in his work The Languages of Africa. built on the stem -rut-).

dual. represented by inscriptions only partly understandable in the present state of our knowledge. these two branches were very distinct.2. therefore. a) Old Egyptian 2. Only Egyptian and some Semitic languages have records from very ancient times. Among the similarities is the phonological system. 1000. and Libyco-Berber.7) give some insights into the latest phase of a number of grammatical categories in a language that underwent important changes in the course of time. which is a Semitic language. The morphologies of Semitic and Egyptian were characterized by consonantal roots which are combined with vowel patterns and affixes. cannot serve comparative purposes. .1. and plural). which were incanta­ tions for the well-being of the dead king. Even in the third millennium B. Both possess two genders (masculine and feminine) and three numbers (singular. The Egyptian language was the speech of the Nile valley from the earliest historical times until some time after A. the royal decrees.D. Only Coptic dialects (§2. Despite these analogies. A. and a prefixed form in ś/s which corresponds to the Semitic causative stem. while its limits are not com­ pensated by any living tradition. Egyptian has a suffix verb form. the practical use of Egyptian in morpho-syntactic analysis and in comparative Afro-Asiatic studies in general is limited. namely the old perfective or "pseudo-participle". show peculiarities of their own. although next to nothing is known about the vowels of the older stages of Egypt­ ian. Egyptian 2. There is also the intrinsic default of the hieroglyphic writing system that lacks any indication of vowels and geminations.. This results partly from the current Egypto­ logical research that too often postulates syntactic principles unheard in language study and. There are also some affinities in the vocabulary.EGYPTIAN 25 none of the other African members of the Afro-Asiatic group is known from sources earlier than the 19th century. The main sources for our knowledge of the language of the Old Kingdom are the biographic texts. and the Pyramid texts discovered on the walls of chambers inside the pyramids of the kings of the Fifth and Sixth Dynasties. inde­ pendently from loanwords. which is related to the Semitic stative. however.C. These texts. except ancient Ethiopic or Ge'ez.

C. Libyco-Berber. " I paddle the canoe" ("applicative"). viz. the only ones of whose religious ideas there is definite knowledge. Egyptian religion presents the same basic characteristics as the Nilotic religion of the Dhinka and Shilluk tribes of southern Sudan. Besides the old perfective. Middle Egyptian survived in later times for many monumental inscriptions and for some literary compositions. Therefore.g. "to think out. . it is unlikely that it could have diverged from common Afro-Asiatic before the latter had developed its verbal system. a kiia ġáày. "not to know". Chadic. but the influence of the spoken language is reflected by the occasional use of forms which were to become standard only in Late Egyptian. hm.C. b) Middle Egyptian 2. like Nilotic languages. negative m vs.3. "wine". which are peculiar to Egyptian and are not paralleled in the other Afro-Asiatic languages. Old Nubian negative morpheme m. such vocalic differ­ ences cannot be expressed in hieroglyphic script. "to think". The suffixconjugations of Old Egyptian remained the major verb forms in use through Middle Egyptian. In another domain.26 SEMITIC L A N G U A G E S including very archaic linguistic features. Egyptian has a special verb for "notknowing": rh. It developed from Old Egyptian and was based on the language spoken towards the end of the third millennium B. vs. Old and Middle Egyptian dispense. Old Nubian ki-. Anyuk a kiì'o ki ġàày. with any equivalent of a definite or indefinite article. and Cushitic. it is extremely difficult to ascertain that the Egyptian conjugation system had developed under a Nilotic influence. because of the lack of vocalization in Egyptian. vs. Old Nubian orpa-gir. Since Egyptian is linked by evident lexical and morphological isoglosses with Semitic. as a rule. vs. " I am paddling a canoe" ("qualitative"). to plan". îrp. e. as k3i. We know at least that.. Now. Middle Egyptian is the classical stage of ancient Egyptian. just as Egyptian vocabulary comprises words alien to Afro-Asiatic but related to Old Nubian. the "qualitative" (indefinite) and the "applicative" (definite). but an important feature of several Nilotic languages consists in showing definiteness by the use of verbal forms involving an internal vowel change. along with Nilotic languages. being used for all pur­ poses from that time until the mid-second millennium B. Old Egyptian has a series of suffix-conjugations. However. For example. "to make wine". it stands to reason that Egyptian has lost the prefix-conjugation in prehistoric times under the influence of a Macro-Sudanic adstratum or substratum of the Nile valley. "to know".

There are also some hesitations in the transcription of Semitic voiced consonants. Demotic is written from right to left. ì-ś-k-3-n = 'šqln). but they are often com­ bined with "alphabetic" signs marking just one consonant (Fig. Fairly accurate deductions may be made about the phonetic value of the consonants and of the vowels thanks to cuneiform texts. w.y vs. which may be related to these of Coptic. i.6. d) Demotic 2. for instance.EGYPTIAN 27 2. like contemporary Semitic alphabetic scripts. for instance in the name of Byblos. This system involves the use of certain hieroglyphic or hieratic signs indicating a consonant followed by a weak consonant or a semi-vowel (3. and numerous foreign words appeared. 1). def­ inite and indefinite articles were used. Egyptian db' vs. "papyrus plant") and that the alleged " d " is the phonetic equivalent of Semitic s (e. and determinatives. . These "syllabic" signs are thought by some to represent Semitic syllables.4. Late Egyptian shows striking differences when compared with Old and Middle Egyptian: the old verb forms were being replaced. Hebrew sūp. but they rep­ resented them also by signs which Egyptologists transcribe with "r" and " n " . particularly the Amarna letters from the 14th century B. and continuing into Roman times. c) Late Egyptian 2. In the Ptolemaic age it first distinguishes / from r. word signs.g. are found in Demotic texts.g. The signs comprise phonograms.C. a special system known as "group-writing" was devised for the transcription of foreign names and words. Demotic was the ordinary language used for official acts and other documents. many phonetic changes occurred. since the Egyptians did not distinguish between r and / in their script. y).D.g. Definite traces of dialect distinctions. they usually used the sign 3 to transcribe these two Semitic phonemes in Middle Egyptian texts (e. "finger"). It appears also. Egyptian twf. Besides.C. and a single Demotic sign is often in origin a liga­ ture of several hieroglyphs. At least as early as the Middle Kingdom. particularly Semitic. that the Egyptian phoneme interpreted as " t " by Egyptologists corresponds then to Semitic s (e. Semitic 'sb'. beginning in the 8th/7th century B. down to the 5th century A. Gbl in Semitic but K-p-n or K-b-n in ancient Egyptian.5.

27. p.28 SEMITIC LANGUAGES SIGN TRANS­ LITERATION OBJECT D E P I C T E D It i ì Egyptian vulture 1 WW i y r w flowering reed j ( i ) two reed-flowers oblique strokes forearm quail chick foot stool homed viper owl water mouth reed shelter in Held* wick of twisted flax placenta (?) 1 * a J • f b P m n r ro h h h s i © animal's belly with teats bolt ((a) folded cloth pool hill-slope c I s k g t i A basket with handle stand for jar loaf tethering rope hand snake d 4 Í 1 Fig. 3rd ed. The uniconsonantal signs in the Egyptian hieroglyphic script according to A. 1. Egyptian Grammar. Gardiner.. . London 1957.

and reflexive . with its religious and ecclesiastical phraseology borrowed from Greek. still records that in his own day Copts in Upper Egypt spoke scarcely anything but Coptic. ') and of velar fricatives (h. As early as the 2nd century A.8. as verbal aspects. passive. magic spells. which have most likely disappeared like in several Semitic languages. although Demotic was also widely used. During the Roman and Byzantine periods Greek was the most common written language in Egypt. Coptic literature is almost entirely religious and consists mainly of translations from Greek. written in Greek letters supplemented by seven characters taken from Demotic. Aside from the rather slight difference of linguistic structure between Demotic and Coptic. the causative. In fact. as yet.D. ġ. from the Greek {Ai)gyptos. texts were written in Egyptian but in Greek letters. They are generally characterized in phonology by palatal consonants (c. Coptic dialects became pro­ gressively restricted after the Arab conquest of Egypt (A. The Bohairic dialect is still the liturgical language of the Coptic Church.7. 640). bom in Cairo (1365-1442). B. 19). There is. Cushitic 2. But it is generally assumed that Coptic died out as a spoken language during the 16th century. little agreement concerning the identification and classification of these languages that are spoken from the Red Sea littoral to the area south of the Horn of northeastern Africa (cf. by globalized emphatics (p = p\ t = t\ c = c\ q = k'). there is a marked change in vocabulary and general tone due to the shift from paganism to Christianity. breaking with the hieroglyphic and Demotic tradi­ tions. and the like. although a Coptic native speaker is attested at Asyût in 1672/3. the "ergative" and the nonactive cases of the noun inflection. n. The language was the one or the other of the Egyptian dialects as they were then spoken and are known as Coptic. These were not only horoscopes. The Arabic writer Maqrizi. Fig. Cushitic preserves some archaic Afro-Asiatic fea­ tures in morpho-syntax. s).CUSHITIC 29 e) Coptic 2. followed by a Christian litera­ ture.D. while a few men in the village of Zainīya (northeast of Karnak) could understand usual Coptic liturgical texts as late as 1936. but the pronunciation is based on the values of the letters in Modem Greek. and by the absence or the limited use of pharyngals (h. ġ). "Egypt". but also Christian translations of the Bible. The Cushitic family comprises about seventy mostly littleexplored languages.

but it frequently suffixes the characteristic mor­ phemes.C. Cushitic consists of five main groups of languages.30 SEMITIC LANGUAGES stems of the verb.. a still imperfectly understood language which is attested from the 3rd cen­ tury B. Cushitic is not-related to the Macro-Sudanic languages which were used and written in northern Sudan: Meroitic. The following diagram presents the main sub-groups: Cushitic North Cushitic (Bedja) Central Cushitic (Agaw) West Cushitic (Omotic) East Cushitic South Cushitic Bilin Khamtanga Qemant Qwara Awngi "Highland" "Lowland" Northern (Saho-Afar) Oromo Southern Galaboid Ba'iso Sam Rendille Bani Somali . just as it uses postpositions rather than prepositions. to the 5th century A. Instead. that might be further subdivided.D. and which is continued by the modern Nubian dialects of the Nile valley and of the Kordofan hills. The pronominal elements and the basic vocabulary often show close rela­ tionship to Semitic. and Old Nubian which is known from Christian writings dating from the end of the 8th century to the 14th cen­ tury.

Remnants of these western Bedja are to be recognized in the Bedyat of Ennedi. presents striking morphological analo­ gies with Semitic verbal stems. atkehan. " I don't collect") and seems to go back to a volitive form. called also (To) Bedawi. is presumably of Bisharin descent. king of Aksum in the mid-4th century A.D. (§8. preterite.D. about 100 km northeast of Kassala. They are called Bouyaeixoi in the Greek inscriptions of Ezana.CUSHITIC 31 a) Bedja 2. The Bedja tribes of eastern Sudan are essentially nomad pastoralists that belong to two main tribal confederacies: the Bisharin and Abdada. "to lengthen"). 550 by Cosmos Indicopleutes who had travelled throughout the Red Sea trading area. A. and Beni 'Amar in the south. In any case.D. many non-Islamic beliefs persist among the Bedja people until our days.H. the Bisherla. which has a present meaning in negative clauses (e. and the intensive or "pluriactional" doubling of a radical (dir. Thus.(gumad.9. is relatively recent. k-ādbil. conditional 'īdbil.g. present 'adanbīl < *'adabbīl.(kehan "to love". They probably inhabited what is now called the Bayuda desert. past 'adbfl. "to kill each other"). . 790) have been found in some places. " I am collecting". with the causative prefix s. sugumād. the conjugation of the finite verb parallels the Semitic imperfective. the reflexive/passive affix -t. Early Moslem monuments discovered in the area should be linked rather with the Arabs of the Beni Omayya tribe who had begun to cross the Red Sea as early as the 8th century. Bedja has lost the Afro-Asiatic pharyngals and the emphatic consonants. Amarar. about 200 km north of Omdurman. whose royal clan. and Bcyá in the "Christian Topography" written about A. " I collected".11). In earlier times the Bedja speakers extended much further to the west across the Nile. Circular stone graves with flat tops are presumably those of Bedja. mdedar. either Moslems or no. "to be loved"). The Bedja of the Sudan are probably the Medju of ancient Egypt and certainly the Blemmyes who used to raid Upper Egypt in the Roman period. Bedja or North Cushitic is spoken on the Red Sea littoral of the Sudan and in the hinterland. and probably jussive. 153 = ca. and elsewhere. for Maqrizi (1365-1442) wrote of them as mostly heathen. The Islamism of the Bedja. "to be long". the Hadendowa. Moslem tomb­ stones dating from the 8th to the 11th centuries (the earliest is dated from A. to the latitude of Kassala in the south. though fervid in some tribes such as the Haden­ dowa. Moreover. "to k i l l " . while early Moslem stone-built towerlike tombs occur at Maman.g. in the north. Their lan­ guage. e. As for phonology. " I may collect".

as well as their labialized counterparts. Awngi has a developed suf­ fix-conjugation with a clear distinction between the main verb and the verb of subordinate clauses. which claim to be of Jewish descent. The Falashas read the Bible in Ge'ez and speak Amharic. spoken south and west of the lake. But they have almost entirely forgotten their former lan­ guage with the exception of some outlying communities living in Qwara before the Falasha emigration to Israel. At the request of James Bruce. . The Agaw people are believed to have once occupied most of highland Ethiopia. and some Falasha prayer texts in Qwara. The Agaw dialects of the Qemant-Qwara group possess the voiced velar fricative ġ. "he knows" (vs. including the Grace after Meals.11.32 SEMITIC L A N G U A G E S b) Agaw 2. yig e. although Awngi seems to be in less dan­ ger of disappearing than the others. The Agaw dialects are receding nowadays before Amharic and Tigrinya. north of Lake Tana. East Cushitic comprises a number of languages spoken in the Horn of Africa and divided into "Highland" and "Lowland" East Cushitic. "he brings" (vs. It is noticeable for having preserved five basic verbs which belong to the prefix-conjugation. and also the velar nasal ng (/rj/). spoken in Eritrea around Keren. These languages are spoken in Eritrea and in northwestern Ethiopia. dating from the 19th century. Agaw or Central Cushitic is constituted of a number of closely related languages. Qemant and the Qwara or Falasha dialects. Semitic y-wq% yaqe. once spoke two Agaw dialects. that are not necessarily intelligible to speakers of another Agaw idiom. Otherwise. The Agaw dialects which are still living include Bilin. and it is still custom­ ary among them to recite certain blessings in Agaw.10. have been preserved. Semitic y-wqy). The Falashas. corresponding to the Khamta and Khamir varieties of Agaw reported earlier in the northeastern part of the Amharic area (Wello province). Semitic y-qūm). he becomes" (vs. "he is. parallel to the Semitic imperfective and perfective: yinte. among them into Falasha. yaġe. the text of the Song of Songs has been translated in 1769-72 from Amharic into three Agaw dialects. Their present scattered distribution must be the result of the Semitic expansion in this area (§8. w c) East Cushitic 2. Khamtanga. Semitic y-wg'). yage. in a region where Semitic influence has been rel­ atively strong. like Awngi. "he remains" (vs. "he comes" (vs.9). and Southern Agaw or Awngi. Semitic y-'ty).

the best represented . Qabenna. the languages of the Galaboid sub-group. Oromo is thus. with the publication of the first Oromo periodical in Ethiopian script (§9. Hadiyya. and the Somali dialects spoken by about five million people in Somalia. and the so-called "Sam" languages. the Saho-Afar in Eritrea and in the Djibouti Republic. and it was once used also in northern Somalia. in eastern Ethiopia. formerly called Galla or Galbnna. Hausa (§2. constitutes a family of some forty related languages spoken by about two million people in south­ western Ethiopia. onwards. is its main representative nowadays. and in northern Kenya. the total of Rendille and Boni speakers amounts only to a few thousand. Among the Omotic dialects. This sub-family of East Cushitic is a compact group with seven or eight languages and several dialects spoken by some two million people. spoken in Kenya. Alaba. east of the lower Tana river. There is a large body of Somali oral literature. The name Somali first occurs in a praise song of Yeshaq I of Abyssinia (1412-1427). after Arabic. there is no reason to believe that their ancestors arrived from Arabia. whose name is derived from a common root *sam ("nose"). Tembaro. Instead. It is spoken by some twenty million people living in Ethiopia and in northern Kenya. and perhaps Burġi. although the Arabic peninsula was the origin of an increasing immigration. also called Kafa group or Omotic — because it is spoken in the vicinity of the Omo river —.D. now called Highland East Cushitic. as well as the source of the Islamization of Somalia. including alliterative poetry. the African language with the largest number of speakers. 2° Other linguistically important Lowland East Cushitic languages are the Konso in Ethiopia. Boni.CUSHITIC 33 1° The main "Lowland" language is Oromo. spoken by about one mil­ lion people. attested mainly in Kenya. was the main substratum language of South Ethiopic.12.7). which are considered by some scholars as a distinct branch of Afro-Asiatic. West Cushitic.2). It comprises Rendille. 3° Eastern Sidamo. the other languages of this group are Kambata. Darasa. Ba'iso which is spo­ ken on an island of Lake Abaya. but it became a "written" language only in 1975. Sidamo proper. The latter sub-group is important for comparative linguistics because of its prefixconjugation with an aspectual distinction between perfective and imperfective. probably from the 8th century A. and Swahili (§1.16). d) West and South Cushitic 2. Contrary to a Somali tradition. east of Lake Turkana (former Rudolf).

"follow!") and the jussive (-Ikdm-) (§38. Most verbal roots are monosyllabic and belong to the types C vC or C vC C x 2 A 2 T 2. the Iraqw. syntactical. C . Djandjero. and the Dahalo.34 SEMITIC LANGUAGES is the Walamo dialect cluster with more than one million speakers. There is no doubt. Beside a stative con­ jugation (e.e. and lexical nature with Semitic. and by the Guanches of the Canary Islands (Fig. Spe­ cial attention was paid also to Moca. that indicate the aspect.g. hnin. in Tanzania.14. the imperative (e. as Mbugu and Dahalo. It shows many correspondences of a phono­ logical. atom. which is used also in subordinate clauses. These languages — except Iraqw — are little known and some of them.g. but these affinities can readily be explained within the general framework of Afro-Asiatic languages. Libyco-Berber 2. The Libyco-Berber language is spoken by some twenty million people from the Siwa Oasis in Egypt to the Atlantic and from the Mediterranean southwards into the Sahara. In the singular. the active subject case is characterized by the w-prefix. by the Tuareg of the Sahara. are influenced by Bantu languages.13. for instance. has only suffixed nominal and verbal formations. but it preserves the aspectual nature of the conjugation very well. 2). From the comparative point of view.1-7). as a rule. however. Libyco-Berber preserves the features of an ergative language type to a greater extent than Semitic and its declension system is based on the opposition of an active subject case (casus agens) to a pred­ icative or non-active case (casus patiens). except a few short Berber sentences in an Arabic manuscript from the 12th century and a number of Berber words and proper names quoted in works of Arab mediaeval writ­ ers. Libyco-Berber dialects were formerly spoken in all of North Africa except Egypt. but no written sources are available before some Shleuh manuscripts from the 16th or 17th century written in Arabic script. viz. morphological. South Cushitic comprises languages spoken in Kenya and in Tanzania. "he is gracious") and two non-aspectual tenses. i. like the Mbugu. that the pronouns are Cushitic and that the conjugation belongs to the common Cushitic suffix inflection. the imperfective and the perfective. viz. and Kafa.2). while the a-prefix is marking the predicative or non-active case (§32. Libyco-Berber has two verbal prefixforms. Madji. Considerable interest in the spoken Berber lan­ guages and their origins had developed by the middle of the 19th century.^ . Kafa.

Fig. 2. Geographical distribution of Libyco-Berber .

e. and affixes may be added in all dialects to the verbal root in order to express the causative. "cause to f a l l " . Tamazight uz. A l l these stems occur also in Semitic languages. viz. one must reckon also with possible loanwords from Songhai. e. and someone speaking Berber is an amaziġ (plur. "be freed". "be crushed".g. and in the city of Agades in the A i r oasis of the Sahara (Niger). -lākkdm-\ causative s-stem. e. agentless passive f/w-stem. e. etc. since it is attested as a North African personal name in Roman times. reflexive or reciprocal. However. in classical sources. "go often out". Ibn Khaldun (13321406) considers Mazigh as a forefather of the Berbers. Despite numerous lexical variations (e. exceptionally morpho­ logical or syntactical. the basic stem of -Ikdm-.g. which are instead numerous in other Berber dialects. -S9rt9k-. reflexive / reciprocal m/rt-stem. Kabyle and Tachaouit or Chaouia in Kabylia and in the Aurès (Algeria). uššdn in Kabyle but âhdggi in Tuareg) and important phonetic changes (e. i-maziġ-drì). although Tuareg and some eastern idioms appear to be its most archaic forms of speech. "heart". In Tuareg. .g. The term ta-maziġ-t is used nowadays in Moroccan and Saharan dialects to designate the Berber language in general. Tarifit or Rifan in northern Morocco.g. or passive meaning of the verb.36 SEMITIC LANGUAGES whether the action is considered as a lasting process or as a concluded action. e. -tdffdġ-. Tuareg is important also because it has but few borrowings from Arabic. the borrowings from Arabic are mainly lexical. -Ikām-. frequentative. Tarifit wr).g. in the Niger valley farther south. Libyco-Berber is still essen­ tially one language. the numerous dialects of which show but relatively slight differences. "fox". Tachelhit ul. -ttwadddz-. frequentative i-stem.g. "to follow": Aspect perfective imperfective Positive -Ikam-lākkdmNegative -Ikem-hkkdm- A vowel lengthening characterizes in Tuareg the intensive stem.g. Tamazight in the Middle Atlas region. Zenaga in southwestern Mauritania. e. like in some Semitic and Cushitic languages. an important isolated language spoken in Tombouctou (Mali). except the last one which is paralleled by the Egyptian "pseudo-passive": intensive stem. while some Libyco-Berber tribes are called Mazices or MG^IKSC. Tuareg dwl. Tachelhit or Shleuh in the south of the country and in Mauritania. m3trdg-. The word maziġ has a long history.g.

"she-goats".. b > b. "laughing"). original pharyngalization can disappear (e. 11 > â. e. although it undoubtedly presents some advan­ tages. The modem Berber dialects reflect the ancient loss of original gutturals. as well as g > z. There is also a large corpus of Libyco-Berber proper names quoted in Punic.g. 3). As a rule. z > š (>ẁ). and various assimilations may occur (e. 2nd century B. nbbn nšqr' corresponding approximately to *i-nbabdn n-u-šqura\ "the cutters of wood (were). the uninterrupted continuity of the LibycoBerber idioms appears to be accepted nowadays by all reputable schol­ ars in the field. r > r. but they have more pharyngalized emphatics than Common Semitic. d > t.15. s > z are quite frequent. "king".g. a secondary pharyngal may be inserted before t (e. The orthography of Tuareg in Latin characters. as indicated by monuments and inscriptions ranging over the whole of North Africa.rr> g.g. Nevertheless.g. Tarifit ydšša < *yikía. g > ġ. does of course not reflect the dialectal richness of the language. "he ate"). "olive oil".).C.of the case prefixes which have thus to be supplied. d > d. Greek. t > f.) and of the Roman Empire (Fig. e. not even the ini­ tial w-. it is not easy to connect the phonological.g. Besides. the tifīnaġ. k > g. The Berber-speaking Tuaregs have a writing of their own. "mule"). The Numidic noun gld. "Phoenician." (Dougga. officially adopted in Niger and in Mali. and lexical elements of this antique documenta­ tion with the modern Berber forms of speech. 11 > ġ. k > š. Its origin may go back to the 7th-6th centuries B. also more palatalized and fricativized consonants.. q > ġ. mor­ phological. "young goat". k > ġ. Kabyle ta-bdġliht < ta-bdġlit. s > s. t > t. d > d. Tachaouit ti-ġdtt-dn. However. Tachaouit dddhhast < *ta-dahhākit. syntactical. B. . /. Punic". l> z.LIBYCO-BERBER 37 2. gives a small idea of the problems facing the linguists. a-. z > z. and Latin sources. Most of the ancient inscriptions (about 1200) date however from the times of the Numidian kingdoms (3rd-1st cent.. they do not indicate vowels. to compare with Hebrew gddi and Arabic ġady. a plural apparently related to Greek (poiviK-. The changes d > d.C.C. pronounced nowadays idgid in Tarifit because of the phonetic changes g > z and // > g. Tachaouit dz-zdht < Arabic dz-zdyt.

gt y • • * •• •• • : .38 SEMITIC LANGUAGES PHONETIC VALUE . b m f P t t d d d t n V OLDER FORMS E3 PRESENT-DAY FORMS Consonants bt mt nb Clusters 0 3 u cj X <=< >o X Î X B 6 VA H nd -rnd t nu E E I nt m 3 3 H > mi r < -1 f ii = Tt G © O n 1 II It r s z z s Š a o — • CD o 0 XX # rt St zt « r- T 30Q 6G St Mw £ +9 :i: t í z y k g i w ġ h q h zt í ?i i 1/ nk ng gt ir n v^: II = • • T « .1.

The Mubi aspectual opposition between bēni. the formation of intensive or "pluriactional" verbs by internal consonant gemination. Hausa yazo. There are also some highly probable etymological connexions between Chadic and Afro-Asiatic. Bidiya. while the Old Akkadian corresponding verb is muātu. The dialect differences are not sufficiently serious to interfere with mutual intelligibility. The importance of Hausa cannot be underestimated. Niger. As result of Islamic influence. by adding a suffix -n and by inserting a vowel -a-. in northern Nigeria. Chadic 2.prefix forming nouns of place. 4. composed mainly in the dialect of Kano which became the standard literary lan­ guage. Distinctive Afro-Asiatic features that can be shown to exist also in Chadic are the affixed morpheme t with the triple function of feminine / diminutive / singulative (e. sin means "brother" like in ancient Egyptian. while náàsb. a large group that has only recently been described in a satisfactory way. so called from the name of Lake Chad. northern Cameroon. spoken in northern Cameroon and in the Chad Republic. mutum means "man". seem to be more archaic and to provide more parallels to Afro-Asiatic. western and central Chad. and of agent. and technology. Migàma.16. The Hausa speakers constitute the single^most numerous group in northern Nigeria and in southern Niger. to Semitic nasāpu. The chief idiom of this family is Hausa. "she came"). mutu means "to die" in Hausa. of instrument. among other ways. In both languages.CHADIC 39 D . Kwang.e. "he came". The Chadic languages. crafts. The language has become the general lingua franca in northern Nigeria and the number of people speaking Hausa as a secondary language is con­ siderable. but in general East Chadic languages. are spoken in Western and Central Africa. with metathesis. i. and binnāa. in the case of Hausa. "to breathe". and an asymmetrical conjugational system involving suffixed feminine and plural markers in addition to pronominal prefixes. particularly in the spheres of religion. numerous Arabic words have been borrowed. "he is building" (Mubi). is undoubtedly related to the conju­ gation of the Semitic verb bny. and. For instance. In East Chadic (Migama). feminine. the -n/t/n gen­ der-number marking pattern in the deictic system (masculine. Hausa is written traditionally in an orthography based on the Arabic alphabet. a recent subdivision of which is presented in Fig. "he built". and to Cushitic nēfso (Boni). the m.g. cor­ responds to Egyptian nsp. tazo. and an original Hausa literature does exist. Kera. . They form the most variegated branch of Afro-Asiatic with some 125 dif­ ferent languages. plural). as Mubi. the formation of noun plurals.

Goemai... Boghom........) (Nancere. 4.. Ngizim. Mubi. Tumak.) (Ron......) (Tera.. Kulere....40 Branch SEMITIC LANGUAGES Subbranch Group Language (Hausa. Bana) (Mandara..) (Mokulu) (Sokoro. Ga'anda.. Buduma. Gisiga...) (Bata.) (Bura.. Kwang) (Dangla. Newman (1977).) (Angas. Pa'a. Gude. Lamang.) (Zaar.. Zime..) (Sukur) (Daba.) (Musgu) (Gidar) (Somrai. Kanakuru..) (Higi. (Kotoko..) Matakam Sukur CHADIC (Matakam..) Fig.) (Kera.... Diagram of Chadic languages according to P..) (Bade.) . Barain... Hina.) FAMILY Kotoko Musgu Gidar Somrai Nancere Kera Dangla Mokulu Sokoro MASA Masa (Masa. Margi. Gabri. Gwandara) (Bole.) (Warji..

Thus Libyco-Berber is certainly closer to the Semitic branch than Egypt/ ian or Cushitic. have a much larger layer of common elements in their phonol­ ogy.1. are obviously the most distant from the other branches. syntax. strongly support the family-tree theory which regards the dividing process that affects a homogeneous language — in this case the Proto-Semitic — as the main impelling power from which new idioms originated. The inter­ relations between the five branches of Afro-Asiatic may therefore be represented schematically in the following way: Proto-Afro-Asiatic Semitic Berber Cushitic Egyptian Chadic 3. while Chadic languages. The Semitic languages. morphology. and vocabulary than the Afro-Asiatic group as a whole. although their number amounts to about seventy. This theory does not exclude. with identical morphemes indicating either the active subject or the predicate-object. both in the singular and in the plural. Also the system of conjugation in Libyco-Berber and in Semitic is built upon a "nominal" and a "verbal" bases. as far as known presently. easily recognizable in ancient and in modern forms of speech. Very characteristic of Libyco-Berber and of Semitic are the preserved features of the ergative language type. They also share certain common features in their evolution. with the aspectual opposition of accomplished to unaccomplished. maintained despite lapse of time and spreading over new areas. PROTO-SEMITIC 3. The five branches of Afro-Asiatic are not really parallel to each other. but this relationship can best be explained in the general frame of the whole language family. These common elements and parallel developments.PROTO-SEMITIC 41 2. . because closer relations can be established between some of them. These two branches of Afro-Asiatic are thus closely related to each other.17.

Arabia. more manageable owing to the fact that the Latin language is historically documented.42 SEMITIC L A N G U A G E S however. the concept of Proto-Semitic would seem comparable to that of Latin with regard to the Romance languages. while the Semitic languages of Africa are grouped in an apparently peripheral area of Semitic and their appearance in the Horn of Africa. the sedentary or halfsedentary protopopulation of North Syria and Mesopotamia was most likely non-Semitic. The problems of the latter group are. however. geographical names. In other words. reflect an old and inherited linguistic tradition of the specific areas. As for Arabia. of a new common language. this region could hardly have supported sufficient population for such large waves of emigration before the domestication of the dromedary in the second millennium B. ProtoSemitic is something more than a conventional name given to the whole of elements shared by the family of languages under consideration. neither the wave-theory nor its variant. Since the Semitic languages go apparently back to a common ori­ gin. In view of the relatively limited geographical dispersion of the ancient core of Semitic languages and of the great measure of affinity between them. more alike than are those which are not so widely separated. 3.C. In fact. midst Cushitic languages. Various regions have been taken into account: Syria. as appears from the large number of non-Semitic geographical names in Palaeosyrian and in Old Akkadian texts. Now. and Africa. In any case. The problem of the original homeland of the Semites cannot be examined historically without considering the linguistic relations between the five branches of the Afro-Asiatic language family. can be given to this question without considering the Afro-Asiatic linguistic interrelations. with the exception of newly founded settlements. however. No definitive answer.3.. the peripheral hypothesis. 3.2. the question of the location of the speakers of this Proto-Semitic language has been often considered of importance. is most likely due to an ancient conquest and emigration. concrete applications of the wave-theory that attributes com­ mon linguistic evolutions to the spreading of linguistic changes by con­ tacts between dialects. that may lead to the emerging of a new local koine. The . correspond to the global evi­ dence with distant Semitic areas. while Proto-Semitic is a linguistic prerequi­ site the existence of which in prehistoric times is necessary for an under­ standing of the mutual relations and parallel developments of the histor­ ically documented Semitic languages. as Akkadian and Ethiopic.

5500-3500 B. the most numerous isoglosses and lexicostatistical convergences are precisely those linking Semitic with Libyco-Berber. 1000 B. on the other. and reached Western Asia. also of PreSumerian substratum. so that erosion took place as in other moist temperate or subtropical regions.C. followed by Egyptian.PROTO-SEMITIC 43 main service that comparative linguistics can render to the investigation of this prehistoric problem is not simply asserting the common origin of the languages in question. Semitic and LibycoBerber around e.g. a major faunal break. closer lan­ guage contacts with Libyco-Berber and with Cushitic. but defining the degree of their divergence and relating it to two variables: time and separation. where written documents of the third millennium B.C. and there is ample evidence of Neolithic culture with rock drawings showing animals that no longer live there. the longer the time the greater the divergence. preserve noticeable traces of Pre-Semitic and. A worsening of environmental condi­ tions is indicated in North Africa ca. and desertion. The Palestinian tumuli. seem to be the less impor­ tant.C. Time is a variable related to divergence in the sense that. on the one hand.C. 3300-3050 B. under like circumstances.) may testify to the arrival of these new population groups. Separation is a variable in the sense that parts of an original language community will tend to diverge faster i f they become completely separated as. for a certain time.C. Now. while the isoglosses and the lexicostatistical factors connecting Semitic and Egyptian.C. while Proto-Semitic maintained. with disappearance of vegetation. and the Egyptian finds in southern Palestine from the Early Bronze period I (ca. and Semitic and Chadic. in Mesopotamia. This implies that the speakers of Proto-Semitic were still dwelling in Africa in the 5th millennium B. were the earliest to separate from the common trunk. belonging to the culture of semi-nomadic groups during much of the fourth and third . the conclusion from purely linguistic evidence seems inescapable that the Proto-Chadic languages. 3500 B.. Settlement was undoubtedly widespread in the Sahara at that time. The collapse of the Ghassulian culture in Palestine around 3300 B. when the Sahara's climate was much wetter. and there was a proper system of rivers and vegetation consisting of grass with trees. but to common linguistic tradition. not to cultural contact and borrowing. desertification. in the Neolithic Sub-pluvial (ca.C.). This might have been the period when the speakers of Proto-Semitic passed through the Nile delta from the West to the East. Although the available data and the very incomplete lexicostatisti­ cal studies must be regarded as preliminary. The similarities in language between peo­ ples living so far away from each other are due. say.

.3) and their congeners would appear to have followed as far as Syria before 2500 B.C. 5. especially Algeria. Fur­ ther subdivisions of course exist. journeying along the Fertile Crescent through Palestine and Syria. while the Ethiopians would have crossed over to the Horn of Africa when drier conditions prevailed in South Arabia ca. to occupy the original language area of the speakers of Afro-Asiatic. seem to confirm this hypothesis. but they are generally too ephemeral to be helpful in this context.2).2. wave after wave of Semitic migrations would seem to have set forth. 5).). 6. Semitic speakers settled among Cushitic pastoralists whose presence in the region probably goes back to 3500-3000 B. were the Akkadians who. Thus. but enough evidence is available to establish the fact that the Afro-Asians belonged basically to the long-headed or dolichocephalic Mediterranean peoples widespread in distribution in Late Neolithic and Chalcolithic times. from North Africa. reached Northern Babylonia ca. and their modem descen­ dants — through frequently mixed with negroes — are found among the .. 5.C. 3500 B.44 SEMITIC LANGUAGES millennia B. dated to the Amratian period (ca.C.4. Since only the most primitive type of raft was needed to cross the Straits of Bab elMandeb or to make the short voyage across the Hanish Islands. fol­ lowing the collapse of the Early Bronze culture in Palestine. since a very similar type of sepulture characterizes pre-historic North Africa. Their African origins may even be confirmed by a relationship of Afro-Asiatic with Bantu lan­ guages (§1. 1500-500 B. The Southern Semites would seem to have reached the moister highlands of the Yemen and Hadramawt after 2000 B. skeletal evidence seems to indicate that the same Neolithic peoples from North-Africa entered the Iberian peninsula and moved into the Egyptian upper valley of the Nile in predynastic times. Although the discussion of these problems lies outside the scope of the present work. and founded the first Semitic Empire at Kish (§4. instead. a rela­ tively early date for the beginning of the last mentioned migration would not be surprising. and crossing over into Mesopotamia. 3.C. 3000 B. (Fig. The data are not so abundant as might be wished.2.2) which form the central group of the large Niger-Congo family and whose homeland probably lies in the Nigeria-Cameroon area.C. The earliest of these migrants. and it is a typical feature of the old Libyco-Berber tradition. However.C. and those who went farthest to the East. The Libyco-Berbers continued. They are well represented by the Naqāda cranial series. it is useful to add that any linguistic mapping a Afro-Asiatic speakers should be complemented by an anthropological approach. The Amorites (§4.C.1-2.

158. p. longer faces. there was no drastic change in the main anthropolog­ ical type during the transition from the Chalcolithic to the Early Bronze age. of the contemporary Bedja population and the main Berber type. Schyiek pradzìejdw w środkowym Sudanie. The spread of the earliest pastoralists in Africa. like those of the Libyans in historical times. coming from Lower Egypt where the latter's origins begin to be investigated. are statuettes of bearded men wearing phallic sheaths. 8000-1200 B. Poznan 1992.C. ca. striking similarities link the physical characteristics of the predynastic Egyptians. Characteristic artefacts of the Amratian period. according to L.. The predynastic pop­ ulation of Lower Egypt differed from that of Upper Egypt in having broader heads. Krzyzaniak. The Amratian culture seems to have been absorbed by the Gerzean one. suggesting connexions with prehistoric Libyco-Berbers.PROTO-SEMITIC 45 Fig. 5. The subsequent racial history of Egypt was to be that of a gradual replacement of the Upper Egyptian or "Cushitic" type by that of prehistoric Lower Egypt. Summing up. I n Palestine. and narrower noses. and of the Palestinian skeletons of the Early Bronze . instead. speakers of Cushitic languages in the Horn of Africa and the Bedja people in the desert between the Nile and the Red Sea.

Areas densely settled in the Chalcolithic period were either totally or partially deserted. judging from the great similarity of the Semitic languages and from their close relationship to Libyco-Berber. However. represented nowadays by the Libyco-Berber dialects. in our hypothesis. migration signifies that whole tribes permanently displace them­ selves and spread over a new territory. viz. and migration. This indicates in turn that the Early Bronze age culture introduced by the Semitic population groups lacked the receptivity required to be modified in a very significant way through linguistic interference. the new migrants — Semites.65 m. with a stature of a little less than 1.55 for women. The substratum generally modifies the gaining language through interference. the Semitic tongues of the new territories followed together with other cul­ tural features a path of historical development more or less divergent from that of the Afro-Asiatic language of the original homeland. 3. This large span of time seems to be sufficient for explaining the differences between Semitic . thus caus­ ing the spreading language to differentiate itself from the language of the original linguistic homeland. infiltration. thus delineated. These are the circumstances obviously reflected in the settlement of Semites in Western Asia where Semitic idioms replaced the substratum languages of the regions where today Arabic. Linguistic expansion can take place by diffusion. Archaeological evidence from Palestine probably provides the correct interpretation of this fact. The spreading of Afro-Asiatic. implies a determi­ nate type of linguistic expansion in Western Asia. Now. and the new sites were usually situated in different spots. — seem to have brought an end to the Chalcolithic settlements in Palestine. the influence of the lin­ guistic substratum on Semitic must have been limited except in Mesopotamia where the Sumerian adstratum played an important role. and Hebrew are spoken. Whereas diffu­ sion necessitates no permanent displacement of language carriers and infiltration implies a movement of but a restricted number of individu­ als.5. with a projecting occiput and the chin prominent. The lat­ ter. Interference varies in degree and kind chiefly in proportion to non-linguistic cultural receptivity or hostility. for men and about 1. The dolicocephalic features are best preserved nowadays among the Bedouin Arabs.46 SEMITIC LANGUAGES age: dolichocephalic type. developed independently from Semitic during a period of 5500 years or more. the location of the new Early Bronze I set­ tlements shows a great shift from the preceding Chalcolithic pattern. Neo-Aramaic. i f we except the borrowings from Punic and Arabic. Thus.

CLASSIFICATION 47 and Libyco-Berber. and the South Semitic with Arabic and Ethiopic. and South Arabian and Ethiopic. 4. while Southwest Semitic split into Arabic. Ugaritic. by neigh­ bouring forms of speech which belonged to completely different lan­ guage families. on the one side. 48.C.2). while other languages have been identified as late as the 20th cen­ tury A. before 3000 B. Amorite. was between East Semitic or Akkadian and West Semitic. 4. at Ebla. respectively in Asia and in Africa. the Northwest Semitic with Canaanite.. and Arabic.C. reveals the existence of a group of dialects belonging to Semitic languages of the third millennium B. The distinct Semitic tongues are ranging from important lan­ guages with large literatures to language forms used over a limited terri­ tory and either entirely unwritten or possessing but a few preserved doc­ uments.1. probably Kish (§5. there is no way at .28. especially i f we take into account the fact that the two groups were affected. It is convenient to call "Palaeosyrian" those dialects that are attested by documents found in Syria. The language may be linked to some extent with the writing system brought from Mesopotamia and thus partly represent the written Semitic of the place from which the script was taken ca. 2400 B. as well as in the Kish area of Central Mesopotamia. CLASSIFICATION OF SEMITIC LANGUAGES 4. West Semitic was believed to have split into a northern and a southern branch.. but before 2000 B. Northwest Semitic further divided into Canaanite and Aramaic..2. until a short time ago. At a later date. Tell Beydar..5). This conception can no more be sustained because of the discovery of languages that do not fit into any of those branches. Amorite. that were related to Old Akkadian and slightly less to Amorite. and Aramaic. This classification was based on the view that the first division which Semitic underwent. on the other. and Mari. The discovery of new types of Semitic speeches in Northern Syria. It was usual.D. Some are attested only in the third or the second millennium B.C. while some "literary" and lexical texts are duplicated at Fāra and at Tell Abū Salābīkh (Iraq). and in view of doubts risen with regard to the classification of Ugaritic.C.C. to group all languages into three great branches: the East Semitic represented by Akkadian. Unfortunately. although the language shows a certain mixture (§41.

C.C. South Arabian. thus blurring clearly cut linguistic divisions. but take also into account. It appears also that Palaeosyrian and Old Akkadian texts contain many proper names in which occurs an ending -a that qualifies the predicate state of the noun and that is attested also in some Amorite names. Ethiopic. Semitic languages were spoken in Mesopotamia. varying in length in the various regions. Palaeosyrian dialects share certain linguistic features with Ugaritic. borrowed from the Sumerians or Proto-Sumerians. that obviously preserve some common archaic elements. the historically attested documentation. while some North Arabian lan­ guages used the prefixed article han-. In conclusion. all spoken dialects were of equal prestige. of course. For a time. and to the chronological and perhaps partly local vicinity of the written languages.48 SEMITIC LANGUAGES present to check this hypothesis. Besides. and the interpretation of other forms of speech as mere dialects of these literary languages cannot be sustained any more. There is also no clear cut between Northwest and Southwest Semitic in the first millennium B. For instance. The differences between the Semitic forms of speech obviously increased with the time. attested normally in Canaanite lan­ guages of the first millennium B. classifications based on important literary languages. A classification based on these standard languages does not reflect. it is due to the use of the same type of script. the variety of spoken dialects. One can assume therefore that this feature reflects an even older common stage of Semitic languages. The resulting picture shows therefore that there was no clear cut between East and West Semitic in the third millennium B.C. a subdivision of the Semitic language family should be based on the wide geographic distribution of the speeches. Syria- . As for the greater affinity between Palaeosyrian and Old Akkadian.3. but does not belong to the living languages of the texts. i f feasible. some Early Aramaic dialects probably possessed the internal or "broken" plural. Hebrew. and Ethiopic.4. 4. the differences of which often increase with the time and in proportion as the geographical dis­ tances grow. as Arabic. certain local dialects augmented their prestige and with their grammatical codifica­ tion came some measure of petrifaction allowing for clearly cut linguis­ tic features. In ancient times. regularly found only in the South Semitic area. 4. Therefore. But with the formation of literary languages in cultural and political centres. and the epigraphical documentation transmits fuller information on dialectal varieties than has since been available. and Syriac.

as Old Assyrian. and NeoAssyrian that together cover a span of 1500 years. (Palaeosyrian. Neo-Arabic). In any case. an East Semitic group with Old Akkadian. This survey does not aim at giving a detailed description of all the Semitic languages.e.5. 4. i. Middle Assyrian. Safaitic. The terms "dialect" and "language" are taken here in their rough definition. Standard Arabic. Although the latter has an introductory char­ acter. Old Akkadian). therefore. and the "discovery" of a new Semitic language merely expresses the scholars' conviction that a type of speech appears sufficiently distinct from others so as deserve a name of its own. when some entirely unwritten forms of Semitic speech have been described and analyzed. 4. and Late Babylonian. Ara­ maic. it also adduces evidence from other ancient and modern Semitic languages and dialects. to which belong written languages of the third and second mil­ lennia B. Beyond this area they have spread only as a result of later and historically known developments. and North Arabian languages (Thamúdic. a West Semitic group with Canaanite (Hebrew. slightly corrected in view of some chronological considerations. Moabite. Phoenician. no exact definition of "language" and "dialect" is feasible. to describe the Semitic languages and dialects roughly in the same geographic order. the lack of any up-to-date introductory work demands a summary presentation of the current knowledge in this field in order to clarify the concepts and the terminology adopted in the present comparative study. Amorite. and emphasizes the position of the great literary or standard languages. Lihyānite.CLASSIFICATION 49 Palestine. or conquest. but also historical stages of the languages considered. the distinct forms of speech being called "dialects" when the differences are relatively small.C. and a South Semitic group with South Arabian and Ethiopian languages. colonization. both ancient and modern. does not aim at exhaustiveness. until the present times. the present survey will distinguish a North Semitic grouping. that can­ not simply derive from the preceding stages of Babylonian. Ammonite). However. when we encounter the earliest written manifestations of a Semitic language (Palaeosyrian. Their position in the Semitic family has therefore to be briefly characterized. Assyro-Babylonian. Therefore.C. Arabia.6. Ugaritic). In this approach. . since the linguistic material of the present survey extends in time over some 4500 years: from the mid-third millennium B.. migration. not only geographically different forms of speech may be called "dialects". It is convenient. and Ethiopia.

However. however. that may reveal the existence of unknown dialects or even of new related lan­ guages. as was the case at Tell Mardikh/Ebla and at Tell Beydar. syntax. besides syllabic signs and auxiliary marks aimed at helping the understanding of the writing. in phonol­ ogy. by the tablets from Tell Beydar. 6). by the Pre-Sargonic and post-Ur-III texts from Mari. North Semitic is represented nowadays by Palaeosyrian (but cf. the spoken languages may differ to various extents from a written koine and. in Syria. and vocabulary. and Ugaritic. there are common features in the writing system. Common scribal traditions and cultural elements are revealed by these documents and by texts from the area of Kish. near Hassake (Syria). as well (Fig. 15 km east of Babylon. NORTH SEMITIC 5. They are known to us only through written records and can­ not be subjected to strict phonetic analysis. .1. especially with the written culture of Sumer. Palaeosyrian is represented by the "Eblaite" texts from Tell Mardikh/Ebla dating from the 24th century B. It would be pre­ mature. Palaeosyrian 5. the sources so far discovered — in particular the proper names — contain elements surviving from an older Semitic language that should also be studied and evaluated.2). §4. Amorite. 7). Palaeosyrian cuneiform script is of Sumerian or even of Pre-Sumerian origin and it uses Sumerian logograms or word signs. Further research and more discoveries are needed to establish how many written Semitic languages or dialects of the mid-third millennium should be distinguished in the area under consideration. in any case. morphology. It is impossible to consider the texts from different sites as written in one language spoken by a single people in the whole area extending from North Syria to Babylonia. However.2. according to the "short" chronology (Fig.C. Besides.C. their corpus expands steadily by the discovery of more written documents. going back to the mid-third millennium B.C. to term that cultural entity "Kish civilization" and to contrast it too sharply with the Sumerian culture. A . in Mesopotamia. These are languages spoken and written in Upper Mesopotamia and Northern Syria in the third and second mil­ lennia B.50 SEMITIC L A N G U A G E S 5.

G.Fig. 6. Ebla Tablet TM.1377 Obverse (Courtesy Missione Archeologica in Siria). .75.

These forms of Semitic speech are mainly known by the numerous proper names — with specific grammatical forms — which appear in var­ ious cuneiform texts. . The geographical area of the speak­ ers of Amorite dialects and the relation of these speech forms to Palaeosyrian suggest however to classify Amorite among the North Semitic tongues and to consider "East Canaanite" as an inappropriate designation of the language under consideration. and by certain linguistic peculiarities occurring sporadically in Old Babylonian texts.52 SEMITIC L A N G U A G E S Fig. Tell Beydar Tablet 2629-T-2 (Courtesy Euro-Syrian Excavations at Tell Beydar). by some loanwords borrowed by Old Babylonian scribes. B. 7. Amorite 5. Some Amorite names are found also in Middle Egyptian execration texts from the 19th and 18th cen­ turies B. Amorite is the name given nowadays to a group of North Semitic dialects spoken in North Syria and Upper Mesopotamia between the mid­ dle of the third millennium and the second half of the second millennium B.C. Amorite was once called "East Canaanite" and is often consid­ ered as a Northwest Semitic language. in particular those from Mari.3.C.

Ugaritic is the name given to the Semitic language discovered in 1929 at Ras Shamra. 2265-2210. I f the Early Dynastic I I I or Pre-Sargonic texts from the Kish area (§5. by the various branches of Assyro-Babylonian (roughly 1900-600 B. A few tablets in alphabetic cuneiform script were also found at other sites. there are letters and administrative-economic documents that reflect a somewhat younger stage of the language. the capital of the Semitic Empire of Sar­ gon of Agade (ca. Yet.C. attested roughly from 2400 to 2000 B. Ugaritic was written in an alphabetic cuneiform script using 30 simple signs which. this script was in several respects imperfect. Akkadian did use logograms or word signs. Like in the case of Palaeosyrian.2. on the coast of north­ western Syria. 13th. and the beginning of the 12th centuries B. its .1. the latter will generally be called "Assyro-Babylon­ ian" in this Outline. Next to mythological and epic compositions.. and by the Late Babylonian that cannot be derived from the preceding stages of Babylonian without admitting at least considerable interference from another Semitic language. A. "Akka­ dian" is the most diffused global appellation of these forms of speech. Old Akkadian may be dated between 2350 and 2000 B. according to a "short" chronology. on the other. However..2) are considered as written in an earlier dialect of the same language as the one used in the Semitic documents of the Empire created by Sargon of Akkad. but was written mainly in syllabograms that also indicated vowels. The texts discovered at Ras Shamra and at Ras Ibn Hani. and the Assyrian and Babylonian dialects of the second and first millennia B. it comes from Akkad or Agade. East Semitic is represented by Old Akkadian.EAST SEMITIC 53 C.C. E A S T SEMITIC 6. according to a "short" chronology). date from the 14th. present single consonantal sounds. on the whole.C. south­ west of Ugarit. owing to its Sumerian or PreSumerian and thus non-Semitic origin.4. Ugaritic 5.C. notably in Palestine.C. on the one hand. the site of ancient Ugarit. to underline the distinction between Old Akkadian. Old Akkadian 6.). 6.

Babylonian. and from Mari. between the 8th and the 6th centuries B. literary compositions. was also used as a literary language in Assyria. and literary texts preserved makes Assyro-Babylonian one of the princi­ pal sources for ancient Semitic. 1900-1500 B. Within the Babylonian dialect one can dis­ tinguish the following periods: Old Babylonian (ca. which originated in the Old Babylonian or Middle Babylonian periods.4. 6.C. Because of the cultural prestige of Baby­ lonian.). The huge number of private letters. that obviously was a local dialect of northern Babylonia that owed its prestige and literary character to the fact of being spoken in the power centre of the Kish dynasties and of the Akkadian Empire. The outstanding case of this is the Amarna correspondence.). 1000-600 B. for purposes of State correspondence and for official documents in areas where East Semitic was not spoken. Assyro-Babylonian 6. and between the earlier Old Babylonian and the later Old Babylonian has to be pointed out. The dialect of these literary texts has been termed Standard Babylonian (Fig. various local forms of Assyro-Babylonian were used in the neighbouring countries and served in the second millennium B. Its written use. however.D. Thus.).C. contrary to the Ebla practice. and Neo-BabyIonian (ca. There are several sub-dialects in the Old Babylonian period. Besides. contracts..C. On the other side. the dialect of the southern part of Mesopotamia. 8). the existence of dialectal differences between North Baby­ lonian and South Babylonian. chiefly from Syria-Palestine. . The Babylonians themselves were calling it "Akkadian". B. and certain speech elements are not omitted in writing. from the Diyala region. Middle Babylonian (ca.C.54 SEMITIC L A N G U A G E S writing is of Sumerian or non-Semitic origin and has the same general characteristics. continued until the 1st century A.3. continued to be copied in later times.C. generally conserving their original wording. however. public documents. but cuneiform signs are generally used with their normal Sumerian value. as it happens frequently at Ebla and at Mari. AssyroBabylonian died out as a spoken language and was replaced by Aramaic in its homeland. In addition. 1500-1000 B. there are provincial dialects from Susa (Elam). there seems to be no convincing way of deriving the earliest attested Assyrian or Babylonian texts from Old Akkadian. By a gradual process.

EAST SEMITIC 55 Fig. and Neo-Assyrian (ca.C.5. The various linguistic stages of Assyrian. espe­ cially in the northwestern regions of the Assyrian Empire and in the wording of contracts. onwards. Late Babylonian 6.). and Arsacid periods from ca. which was Aramaicized in its final phase. 6. . with records strongly influenced by Babylonian.C. 1000-600 B. can be divided into Old Assyrian (ca. Middle Babylonian fragment of the Gilgamesh Epic from Megiddo (Courtesy Israel Department of Antiquities and Museums). 19001700 B. but written in the same dialect.C. 1500-1000 B. C.C. the dialect of the north­ ern part of Mesopotamia. 8. Middle Assyrian (ca. Seleucid.).6. with texts principally from commercial settlements in Ana­ tolia. 600 B.). Late Babylonian is the written language of South Mesopotamia in the Persian.

Amorite and Ugaritic have often been considered as older forms of speech of Canaanite despite the fact that they are morphologically and syntactically more distinct from Hebrew than the North Arabian languages. The name Canaanite. West Semitic was traditionally divided into two groups. and translated.2. while the North Ara­ bian forms of speech w i l l be viewed as the third main family of the West Semitic languages of Syria-Palestine and Northern Arabia.6) and the occasional transmutation of the stative into an Aramaic perfect (§38. Moabite. the existence of Late Babylonian tablets belonging to this genre does not prove that Babylonian subsisted as a vernacular language at that time. although there were certainly educated people having a fairly good knowledge of the literary idiom. Since people resorted in the Near East to professional scribes to have even their private letters written.C. the ancient appellation of southern Syria and Palestine. the older stages of the Canaanite languages. Canaanite 7. read. and syntax. For this reason.10). instead. In recent times. as the use of iprus-forms in the volitive functions of the Aramaic imperfect (§54. Ammonite. W E S T SEMITIC 7. are classified. coined from the toponym Canaan. w i l l be used in the present work to designate. The latter does not seem to have borrowed an impor­ tant part of its lexicon from Aramaic. since their type of speech reveals a too far-reaching linguistic change in phonetics. as Hebrew.1. Amorite and Ugaritic have been classified here as North Semitic tongues. mor­ phology. 7. A.C. namely the Canaanite and the Aramaic. but certain texts can hardly be con­ sidered as written in a truly Babylonian dialect. . and Edomite.56 SEMITIC LANGUAGES while Aramaic and the practically unknown Chaldaean dialect were the spoken idioms which by a gradual process influenced the written lan­ guage. The Hebrew language is the only one in this group that survived the Antiquity. with Hebrew and Syriac as the main lit­ erary languages. Phoenician. as a rule. known from sources of the second millennium B. The stages of the first millennium B.

C. are reflected to a certain extent in the Old Babylonian tablets from Hazor. Also the pseudo-hieroglyphic inscriptions of Byblos are most likely composed in a Canaanite dialect. the Dead Sea scrolls.5. They are attested directly by a number of short inscriptions found in Palestine (Proto-Canaanite) and in the Sinai peninsula (Proto-Sinaitic). the Chronicles.C. but they cannot be consid­ ered as deciphered.C. and it represents the earliest purely alphabetic form of writing. The last mentioned works are written in the so-called Mishnaic Hebrew. it comprised two main dialects — the Israelite in the north and the Judahite in the south — but the biblical text retained but a few traces of dialects that can instead be identified in the epigraphical material. but cannot be further defined with any certainty.C. 7. The Dead Sea scrolls have revealed some linguistic features that are parallel also to the particular Samaritan .4. In the first millennium B. and Esther. The Amarna correspondence of the 14th century B.D.3. the documents discovered in the Judaean Desert. Also some of the documents discovered in the Judaean Desert are written in this idiom and its influ­ ence can be detected already in the later books of the Bible..g.C. This material can be supplemented by the Canaanite words and forms occurring in eight texts found at Kāmid elLoz (Lebanon) and in a few scattered documents. by the Semitic loan­ words in ancient Egyptian. 1000 B. I f the inscriptions on Phoenician arrowheads and the Gezer calendar are added to this group. at least in some parts of Judaea.CANAANITE LANGUAGES 57 a) Old Canaanite 7. Old Canaanite forms of speech of the second millennium B. Hebrew is the Canaanite form of speech used inland from ca. some superimposed upon datable Egyptian objects.. the latter can be dated between the mid-second millennium B. the Mishnah. and by the few words in Egyptian texts put into the mouth of Semites. onwards.C. Qohelet. Besides the Bible. e. and the 10th century B. b) Hebrew 7. which existed pre­ viously for hundreds of years as a vernacular but became a new literary language only in the late first century A.C. The whole series is variously dated by scholars from 1800 B. provides a large number of Canaanite glosses and linguistic peculiarities in its Babylonian cuneiform text. Also this material is unmistakably Canaanite. and the Tosefta belong to the period when Hebrew was still a spoken language. onwards.

retained as the lan­ guage of liturgy and revived as literary language from the 14th century on.D. In fact. although Samaritan Hebrew. and whose vocalization was adjusted to the system of Aaron BenAsher. Sidon. its phonologi­ cal and grammatical interpretation by the various Schools of "Masoretes" or traditionalists. 9). but it remained a written language that served for every written purpose and even flourished in poetry and literature. although the consonantal text of the Hebrew Bible is gen­ erally speaking reliable from the linguistic point of view.58 SEMITIC L A N G U A G E S tradition of Hebrew. Phoenician is the Canaanite form of speech used in the first millennium B. Petersburg B 19 which was written in 1009 A. since most of the Jewish immigrants who arrived in Pales­ tine from eastern Europe prior to World War I I were native speakers of Yiddish. Since 1881 Hebrew again became a spoken idiom and it is nowadays the language of modern Israel. in the .D. this mixed idiom cannot be employed as a trustworthy basis for the study of spoken and literary Mishnaic Hebrew used in the earlier period. This later form of Mishnaic Hebrew was influenced by Biblical Hebrew and by Aramaic. A c) Phoenician 7. Instead. developed under the influence of Aramaic and of the Arabic vernacular.C. the recent massive immigration of Jews from Russia brings about a Slavic impact on some aspects of spoken Hebrew. As a matter of fact. There was a certain impact of Yiddish on the early stage of mod­ ern Hebrew. especially that from Tiberias. exhibits innovative elements as well..D. The same must be said about the "Masoretic" Hebrew of the 9th-10th centuries A. Vocalized quotations of Hebrew words and sentences in the present Outline are generally based on the reading of the Tiberian Masoretes as preserved in the Ms. Mishnaic Hebrew ceased to be spoken around 200 A. As a result. (Fig. Tyre. is condi­ tioned by their knowledge of the language spoken more than a thousand years before them and by the reliability of oral traditions underlying the reading of the Bible in Jewish communities whose vernaculars were mainly Aramaic or Arabic dialects.6. though Elijah Levita (1468/9-1549) pointed already out that the Masoretic vowels and accents do not belong to the original text but had originated in post-talmudic times.D. known as ivrti. that serves as the main base for the grammatical investigation of Biblical Hebrew. its vowel points and accents are almost iden­ tical with those of the Aleppo Codex pointed by Aaron Ben-Asher him­ self in the first half of the 10th century A. St. in the coastal cities of Byblos.

. 2. 9. Page from the Aleppo Codex with the text of I Chron.CANAANITE LANGUAGES 59 Fig.4.26-3.

was a Canaanite form of speech. modern Amman. Neo-Punic continued to be spoken in North Africa until the 5th century A.D. As far as our information goes. Moabite. represented by a small corpus of inscriptions dated from the 9th to the end of the 6th century B. the Phoenician speech of West Mediterranean countries is called Neo-Punic and it is attested also in Latin script (Latino-Punic inscriptions).60 SEMITIC LANGUAGES neighbouring towns. and in the various settlements and colonies estab­ lished in Anatolia. In Carthage. It was probably more different from Hebrew than can be guessed from the unvocalized Aramaic script of the inscriptions. and on the Atlantic coast of Spain and of Morocco. Punic inscription from Carthage.8. In its latest stage. a Tyrian foundation. perhaps down to the 11th century A.7. called Punic (Fig. at Suit.. documented down to the first centuries A.C. Ammonite. 10).D. that was also used in the Numidian king­ doms of North Africa.. The epigraphical material attests the existence of different dialects in the Phoenician homeland and overseas. the language developed a distinct form. Fig. e) Moabite 7. used east of the lower Jordan valley around Rabbath-Ammon. in Libya. represented by two inscriptions and a few seals dated from the 9th through the 6th century B. 10. along the Mediterranean shores..D.C. d) Ammonite 7.D.. was a Canaanite idiom spoken east . but Phoenician died out as a spoken language in the Levant at latest in the 3rd or 4th centuries A.

Edomite.ARAMAIC 61 of the Dead Sea. 800 B.C. and some of its dialects survive until the present day. the two Samalian inscriptions from Zincirli (8th century B. and both do not use the deter­ minative-relative zy. except for the Tell Fekherye statue and the Tell Halaf pedestal inscrip­ tion. Assyria. Although the ninth-century B.C.11. Its earliest written attestations go back to the 9th century B.). Moabite inscriptions present the earliest "Hebrew" characters of the alphabetic script. 850 B. on. From the 8th century B. palaeography and morphology reveal some specifically Edomite features.C. attested by a few inscriptions and seals dated from the 9th through the 4th century B.) and on the stele found at Tell el-Qādi (ca.9.10. 850 B. . a) Early Aramaic 7.C. and even in the juridical and economic documents on clay tablets from Upper Mesopotamia and Assyria. Early Aramaic is represented by an increasing number of inscrip­ tions from Syria.C. North Israel. and northern Transjordan dating from the 9th through the 7th century B.) seems to testify to the use of internal or "broken" plurals.) apparently retain the case endings in the plural and have no emphatic state.. The latter is also unat­ tested in the Deir 'Alia plaster inscription (ca. f) Edomite 7. Aramaic forms a widespread linguistic group that could be clas­ sified also as North or East Semitic.C. While the Tell Fekherye inscription (ca. was the Canaanite idiom of southern Transjordan and eastern Negev. 11). The morphological variations point instead to the existence of several dialects that represent different levels of the evolution of the language. their language cannot be regarded as an Hebrew dialect.C. Several historical stages and contemporaneous dialects have to be distinguished. (Fig. Despite our very poor knowledge of the language.C.C. Aramaic 7. There are no impor­ tant differences in the script and the spelling of the various documents. B. a standard form of the language prevails in the inscriptions.

m n s c S T 0 > ? e > > * ) > 0 ¥ o 0 ? © ( à r » P s • q r V > ? 1 r ? V . > f . 3. and northern Transjordan in the 9th and 8th centuries B . 5. ca. 4. 800. 8. Panamuwa I (Zincirli). Sefire. Deir 'Alia. / % r tr ? w K <?? % V t % w s I f / / / Fig. 6. mid-8th century. ca. beginning of the 8th century. early 8th century. . 7. Tell Fekherye. I I . late 8th century.62 SEMITIC LANGUAGES 1 2 3 4 5 6 7 8 9 < b f f f ft â -\ A 9 A g d h A Í T «*- A A « 1 M s 2 \ H X ft w M % f a h IS © M © ® t v I % t 1 J? y / V & . Bar-Rakkāb (Zincirli). 9. Zakkūr (Tell Afis). mid-8th century. : I . Panamuwa IT (Zincirli). 2. mid-9th century. 730. Cilicia. Karatepe. late 9th century. Alphabetic scripts of Syria. Kilamuwa (Zincirli). C .

c) Standard Literary Aramaic 7.C. Beginning with the 8th century B.14. The documents and the Bar Kokhba letters discovered in the Judaean Desert represent the Palestinian Aramaic of Judaea. Besides the texts in Standard Literary Aramaic and in a faulty Official Aramaic that survived in non-Aramaic speaking regions of the former Persian Empire. Middle Aramaic is the name generally given to the Aramaic dialects attested from the 3rd century B.C. they already contain elements of Middle Aramaic on the one hand. of the Aramaic documents found in Egypt.13. and subsisted alongside the Official Aramaic of the Achaemenian period. and of Arabic on the other. known as Onqelos and Jonathan.ARAMAIC 63 b) Official or Imperial Aramaic 7. and the narrative in the Aramaic portions of Ezra are the earliest exam­ ples of this form of speech that is further used in the Book of Daniel. Pakistan.C. in the Targums to the Pentateuch and to the Prophets.C.15. in Afghanistan. and at Persepolis. there are a number of epigraphic dialects from this period. Documents written in Nabataean were also discovered among the scrolls of the Judaean Desert. at a much later date.12. and in the Cau­ casus. Official or Imperial Aramaic is the language of the Aramaic doc­ uments of the Persian Empire. Aramaic became the lingua franca of the Near East and it served later as the offi­ cial language of the Achaemenian administration until the end of the 4th century B. as well as of the Aramaic letters and documents quoted in the Book of Ezra. in the "Scroll of Antiochus". 7. Standard Literary Aramaic is the literary dialect that emerged in the 7th century B. and. but some authors apply this qualification also to earlier texts. Although they are basically written in Official Aramaic. The Story of Ahiqar. the Bar Punesh fragments. Turkmenistan. 7.16. perhaps the scattered phrases of the story from the tomb at Sheikh el-Fadl. in the Wadi Dāliyeh (Samaria). It is the language of various inscriptions on stone. to the 3rd century A.D. d) Middle Aramaic 7. in the literary Aramaic compositions discovered at Qumrān. in Megillat Ta'anit. like the Nabataean inscriptions .

12. 12).C. Copenhagen). are detected in some of these inscriptions. are written in a West Aramaic idiom based on Official Aramaic (Fig. dating from the 1st century B. Egypt.. Nabataean Aramaic was the written language of the Arab population whose main centre was Petra. to the 4th century A. D . the language of which was also influenced by an East Aramaic dialect. Traces of Arabic. 7. North Arabia. . There are also a few inscriptions written in Nabataean Arabic (§7. The Nabataean use of the Aramaic language and script contin­ ued a North Arabian tradition attested already in the 5th century B. historically attested from the beginning of the 4th cen­ tury B.D. which was the language of a substantial part of the population of Palmyra.D. The last dated Nabataean Aramaic text dates from 356 A. From the 2nd century B.D. dated A . The Palmyrene inscriptions. found at Tell el-Maskhūta (Egypt). by the inscriptions of the oasis of Tayma' and somewhat later by the inscription of Qaynû.C. through the 3rd century A. 214 (Courtesy Ny Carlsberg Glyptotek. Negev.38).64 SEMITIC LANGUAGES and graffiti from Transjordan. and Italy.C. Palmyrene inscription from Malkū's tomb. Fig.17. Greece. king of Qedar.C.

20. They date from the 2nd and 3rd centuries A..22. 7. are written with Aramaic logograms. mod­ ern Urfa. words writ­ ten in Aramaic but read in Middle Iranian. ca. The earliest Syriac inscriptions from the region of Edessa. and this may also be the case of the inscriptions found at Toprak-kale. from the 1st century B. and considered by their editors as Khwarezmian (Middle Iranian). are the precursors of the ideograms used later in the Pahlavi texts of the Sassanid dynasty (226642 A. 7. Despite the contrary opinion of some authors. i. and the inscrip­ tion of the Herakles statue from 150/1 A.D. but their language occupies an intermediate position between West and East Aramaic.C.). The Jewish Palestinian Aramaic of the Byzantine period is often called Galilean Aramaic since most of the material comes from Galilee. also the ca. It is a period with abundant written material..D. perhaps in the Chaldaean dialect. e) Western Late Aramaic 7.C. show the influence of East Aramaic. West Aramaic consists primarily of material known from Palestine. 7.ARAMAIC 65 7. and are all of pagan ori­ gin. phonology. in Uzbekistan.D.23. The most important witnesses of this use of Aramaic logograms are the Avroman parchment from 52/3 A. From the 3rd century A.18. found in southern Iraq and written in cuneiform script on a clay tablet. 2000 ostraca of Nisa (Turkmenistan). 7. The material consists of a variety of dedicatory and memorial inscriptions. The Uruk Incantation text from the 3rd or 2nd century B.D.D. Also the Aramaic texts of Hatra..19. The inscriptions from Ashur and other sites in the area of Upper Tigris.e. but the dialect is best . reflect a closely related form of speech and are written in the North Mesopotamian variant of the Ara­ maic script. morphology. but this appellation may be too restrictive. is composed in East Aramaic.21. positive distinctions between East and West Aramaic can be made on ground of vocabulary. 100 km south-west of Mosul. Their script resembles that of the contemporary cursive Palmyrene inscriptions. and syntax. The Aramaic logograms in Parthian inscriptions. on. go back to the lst-3rd centuries A. and their language is closely related to Syriac. all dating from the Late Parthian period.D.

The preserved sources date from the 5th-8th centuries A. but the differences are limited to some phonetic features.D.D.D. Christian Palestinian Aramaic. Eastern Late Aramaic is represented by the literary languages Syriac. was spoken by converted Jews living in Judaea and in Transjordan at least from the 3rd-4th centuries A. of Leviticus Rabba.27. as well as by the Ara­ maic logograms in Pahlavi and other Middle Iranian dialects. f) Eastern Late Aramaic 7. when it was used only in the liturgy. is rep­ resented by the Targum to the Pentateuch.. there are two different vocalization systems and three main Syriac styles of writing: the Estrangeìā. as best represented by the so-called Neofìti I Targum from the Vatican Library and by fragments from the Cairo Geniza.26.D. Besides some epigraphic finds. One can distinguish West­ ern and Eastern Syriac. The sources exhibit a dialect closely related to Samaritan Aramaic (§7. such as the Palestinian Talmud. originally the dialect of Edessa.25. the Aramaic hymns preserved in the liturgy.D. with a large literature in both poetry and prose. when the language was spoken.23). and of other Midrashim. 7. as well as of translations of other Greek religious texts..66 SEMITIC LANGUAGES known from literary works. Instead.. Its oldest literary works go back to the 2nd century A. until the Arabization of Palestine. on. the Aramaic parts of Genesis Rabba. and the language is used down to the pre­ sent day. written in an offshoot of the Palaeo-Hebrew script and spoken by Samaritans till about the 10th century A. Syriac. such as the Melchite liturgy.D. and from the Palestinian Targums. sometimes called Palestinian Syriac because of its script. this dialect is best represented by fragments of Bible translations from Greek. and such works as Memar Marqah and the Asatir. occupies an intermediate position between East and West Aramaic. and Jewish Babylonian Aramaic. and from the 11th-13th centuries A.. a formal script which resembles that of the Syriac inscriptions of the lst-3rd centuries A. Mandaic. 7. It is the best documented of the Aramaic languages.24) and to Galilean Aramaic (§7. Traces of Mishnaic Hebrew influence are found in this dialect.24. 7. . primarily of a religious Christian nature. although Syriac was generally replaced by Neo-Arabic as a spoken idiom from the 8th century A.D. Samaritan Aramaic.

A. . > -V V ^ *A. %x •X. X X Jt L Is A A- Fig.ARAMAIC 67 Estrangela 1 2 3 4 1 Serto 2 3 4 1 Nestorian 2 3 4 Trans­ Name cription of the Letters 1 J3 T -13. XX a JG s ua • q r X X š t i. K 1 °J o \ V o ) z 1 3 b g d <^ alaf bēt gāmal dālat hē waw zên hēt tēt yod kāf lāmad mīm nūn semkat 'ē P pē sādē qof rēš šīn taw • o . » T * \ \ 2*. Syriac Scripts. 13. J m n fix JS3 Ca SO & CD s J J5 43. k 1 V Sri 1 c o \ i \ JO \ 33.1 \ A c a % «• h w z h t y .

date from the 4th-6th centuries A .i6Û AIMJ»^ JAlìLĠ*S> ^\a±rju& y à * . D . A 4 ± .^ . and a form of colloquial Mandaic has been recorded. D . a t a p oyttrf«b/ F r o m the Canonical Prayerbook of the Mandaeans. a large number of inscribed "magic" bowls. Since Mandaic uses matres lectionis more than any other Aramaic dialect and does not follow any traditional orthography. Mandate is the language of the Gnostic sect of the Mandaeans.o*j> 6 ûii^*-V à**-*^*" «' . 7. have been discovered in southern Iraq and Iran. yxJu*»^ aí*A^U.«f^ft „ JAA*.D. like Jacob of Edessa (7th century A. MtH. E . and then moved to southern Iraq and Iran where its adepts have still been identified in the 20th century. a $ A . D r o w e r (Leiden 1959).af-dà" ±**ât*AP JJ^iẃr y aìaj» tyx. whether at the beginning. Aiio^r -Hit ^f*-»* a«ftáai. 13). . . u* A ^AíAAn^ 2à±mỳo Ju^^i »<. around Harran. HA*fA&^JJy oJ*j_^ a . another cursive variation used in the East. The sect flourished for a time in Upper Mesopotamia.**í<4*« 41 ^t-rf y»íi—*| t^i±~>y uûû&~. whose origins are obscure. and the Nestorian. The works of Syriac grammarians.AjyiijfAJ'^J «^ —^ é*. S .. known at present. y^jL^l» aiSu-»y d <4 a / ^ViiAJ» < . have exerted an influence on both Arabic and Hebrew grammatical traditions. The majority of the Syriac letters have different forms depending upon their position in a word. a developed cursive ordinarily used by the Jacobites in the West.. middle or end. and their script represents a South Mesopotamian variant of the Aramaic script-type. and whether they stand alone or are joined to others (Fig.28. i . it has been of great importance for establishing the phonol­ ogy and the precise morphology of East Aramaic. and their major literary works may also have been written in that period. .. a .).68 SEMITIC LANGUAGES the Serto. They date from the 5th7th centuries A . Besides. ed.± A t £ L < » v JẃUierf t*. in Mandaic script and language. ^ A i í í 9 4 f > o^tofdírT tSttl4^v».i A i S d . The earliest Mandaic texts.

north of Damascus.D. Most useful is the Frahang i Pahlavīk. a village in the Diyarbakrr province. Neo-Aramaic dialects are spoken nowadays by about half a mil­ lion people living in various regions of the Near East or emigrated to other parts of the world. Western Aramaic is exposed to strong phonetic.Middle Iran­ ian glossary that might go back at least to the 7th century A. They are divided into three main groups. The language is reminiscent in many respects of the ancient Aramaic dialects of Palestine (§7. Jewish Babylonian Aramaic is known primarily from the Baby­ lonian Talmud. g) Neo-Aramaic 7. Characteristic of this Western form of spoken Aramaic are the changes ā > 6 and p >f.29. the use of the y-prefix in the 3rd person of the imperfect. they show a tendency to use the pharyngal h and have developed a conjugation based on participles. Gubb 'Adīn. Tūroyo comprises the dialects spoken by Christians in the Tur 'Abdīn area.23-25). the Geonic texts. A closely related idiom was spoken at Mlahso. These dialects are the surving remains of the once widespread Aramaic languages.31. 7.30. near Mardin.ARAMAIC 69 7.32. etc. but some of them indicate changes due either to the influence of Late Eastern Aramaic or to errors made by the scribes who no longer knew the Aramaic language. in southeastern Turkey. Western Neo-Aramaic is still used by Christians and Moslems in the three villages of Ma'lūla. and the Jewish Babylonian incanta­ tions of the "magic" bowls from the Nippur region. and lexical influences of vernacular Arabic. These dialects occupy an intermediate position between Western and Eastern Neo-Aramaic. The large emigration of the local population . a kind of Aramaic . Differences have been detected in the lan­ guage of these texts spread over eight centuries. about 60 km. date from the 3rd through the 11th century A. for which good manuscripts should be used.33. Like Eastern Neo-Aramaic (§7.D. the Book of Commandments by ' Anan ben Dawid. but they exhibit the unconditioned change ā > o like Western Neo-Aramaic. and Bah'ā. These various sources. The Aramaic logograms in Pahlavi and other Middle Iranian dialects are mostly derived from Official Aramaic. 7. grammatical. the early Karaite leader.34). preserved by religious minorities in mountainous retreat areas. 7.

It gave rise to a spoken koine that coexists nowadays with the dialects. and above all by their way of life. . Neo-Aramaic dialects are used in Kurdistan. used for printing periodicals. but lacking in Syr­ iac.70 SEMITIC L A N G U A G E S resulted in the creation of scattered Turoyo-speaking communities in Western Europe.. 7. C . Benjamin of Tudela.D. most of the Jews have emi­ grated to Israel. Eastern Neo-Aramaic. is the continuation of the eastern branch of Late Aramaic. In this Outline. in the neighbour­ hood of Lake Urmia. While populations of merchants and farmers were settled in towns and oases. who visited Kurdistan in the mid-12th century A.29). The Christians write in the Nestorian type of Syriac script. While some of them bear Aramaic names. The fairly uniform standard written language of these publications is based on the Urmi dialect. and Turkey. Nowadays. and Jacobites. and near Mosul. and true nomads.34. They are spoken both by Jews and by Christians of different denominations: Nestorians. near the common borders of Iraq. but there are no documents extant in this form of speech since it was not used as a literary vehicle. Arabic 7. Chaldaeans. in Iraq. as at rule. It is assumed therefore that Eastern Neo-Aramaic developed from a language similar to Mandaic and to Jewish Babylonian Aramaic. point to the Eastern Neo-Aramaic. The earliest attestations of Arabic are a number of proper names borne by leaders of Arab tribes mentioned in Neo-Assyrian texts. Georgia. Various North Arabian populations have to be distinguished. references to Neo-Aramaic. as well as innovations shared by Mandaic (§7. others have names that belong to a group of dialects now called Proto-Arabic or Ancient North Arabian. in Iran. There are archaic elements retained in Neo-Aramaic which are absent from Classical Syriac (§7. reports that the Jews living there were speaking Aramaic. while the emigration of Christians to the United States and to Armenia. made without further specification. Iran.28) and by Jewish Babylonian Aramaic (§7. semi-nomadic breeders of sheep and goats were living in precari­ ous shelters in the vicinity of sedentary settlements.27). differing by their language and their script. called also "Modern Syriac" or "Assyr­ ian". however. and pamphlets. books. and Russia had already started as a result of World War I .35.

C.D. from the language of the socalled "Dedānite" inscriptions which antedate the period when Dedān was the residence of a Persian governor in the 5th century B. Dedān and the neighbouring site of al-Hidjr (Hegrā') were occupied in ca.. dated ca.36. 25 B.37. one of several Arabian tribes mentioned in the Assyrian annals (Tamudi).38. were moving over great distances and living in tents.).D. Only the few Nabataean Arabic texts are written in Aramaic script. While the case endings of the nouns are still used correctly in the bilingual AramaicArabic of Oboda. Lihyānite should not be distinguished.. has become merely vestigial by the lst-3rd centuries A. as it seems. in a 5th-century Byzantine source. that had its own king in the 6th/5th century B. through the 1st century B. Pre-Islamic North and East Arabian dialects use a variant of the South Arabian monumental script.) and an-Namāra (328 A. 7.. the oasis was the capital of the kingdom of Lihyān. in North Arabian graf­ fiti from the Tayma' region.C. that testify to the evolution of the language.C. as appears from the inscriptions of Hegrā' (267 A. 7. the case differentiation between bnw and bny. ancient Dedān. Also in South Arabian. from the 4th century B.C. The so-called Thamūdic graffiti are named after Thamūd. which for nearly two centuries was home to a colony of Minaean tradesmen from South Arabia. by the Nabataean kingdom.. Then. a) Pre-Islamic North and East Arabian 7. Nabataean Arabic is represented by a few inscriptions in Ara­ maic script. These sources make it clear that the .D. 100 A. Lihyānite is represented by a series of graf­ fiti and of mainly monumental inscriptions engraved in a variety of the South Arabian script. 169 A. there was no longer a fully func­ tioning case system in the 3rd and 4th centuries A. both urban and Bedouin. where it can be detected. in a Greek inscription of a Nabataean temple in northeastern Hedjaz.D. in many passages of the Qur'ān. Lihyānite is the local dialect of the oasis of al-'Ulā.D.39. and in writings of Arab geographers. that had developed from the common Semitic alphabet.D.C. Different forms of speech have been distinguished.ARABIC 71 dromedary breeders and caravaneers. 7. dated ca.

Jordan. Hasaean is the name given to the language of the inscriptions written in a variety of the South Arabian script and found mainly in the great oasis of al-Hāsa'. D .. through the 3rd century A . dating from the 6th century B. often specify his job or the circumstances of his passage. scattered over an area including southeastern Syria. D . both urban and Bedouin. Many thousands of such texts. near modern Sulayyil.C. by the use of two differ­ ent articles. from al-Hāsa' down to 'Oman. through the 4th century A. Since the Safaitic graffiti have been found on the Nabataean territory and are con­ temporaneous with the Nabataean Aramaic inscriptions. For the period from the beginning of the 2nd century B. through the 3rd or 4th century A . some of them are likely to be written in Nabataean Arabic. 7. we actually possess the inscriptions from Qaryat al-Fāw. and North Arabia have so far been collected and in part published (Fig. which is very common in Safaitic inscriptions. 14-15). and belonging to different dialects. The oldest Thamūdic inscriptions. However. as shown e. South Arabian script was used also in southern Iraq ("Chaldaean" inscriptions) and on the East Arabian coast. The Safaitic inscriptions date from the 1st century B.40. the name ' T h a m ū d i c " was incorrectly applied to various types of graffiti found throughout Arabia. They are.D. Safaitic texts do not belong to a single dialect. These inscriptions can be dated from the 8th through the 1st century B.C. on the trade . Pre-Classical Arabic dialects.41. They are so called because they belong to a type of graffiti first discovered in 1857 in the basaltic desert of Safa. D . which is widely used in Nabataean Arabic proper names but appears exceptionally in names attested by the Safaitic graffiti.72 SEMITIC L A N G U A G E S Thamūdaeans were living between Mecca and Tayma'.g. In any case. b) Pre-Classical Arabic 7. namely h-.C. memorial inscriptions that mention the name of the person and of his ancestors. for the rendering of various local forms of East Arabian speech. to a large extent. have been found in the northern Tayma' area. in the east of Saudi Arabia.C. 7. probably from the 6th century B. southeast of Damascus. and 'al. and call on a deity to protect his memory and ensure peace to him.42. are described to a certain extent by early Arab philologists which have pre­ served some data on the forms of speech in the Arab peninsula around the 7th-8th centuries A .C.

2 route linking Nadjrān with the eastern Arabian coast. "(belonging) to him. 3° ls mt bn 'n'l. "(belonging) to Dahbānu the carrier. 2° Idhbn nql bn mnhl. They reveal the disappearance .ARABIC 73 Fig. son of Minhālu". Three Safaitic inscriptions on a boulder in Wadi Sirhān (courtesy of Abdu-Aziz al-Sudairi): 1° Ih Mm bn 'rm. to Blm. "(belonging) to Śāmitu. son of 'Ān'il". son of 'Amru". capable of expressing the pho­ netic features of Arabic unambiguously. They are written in fine monumental South Arabian script. 14.

of the nunation (e. except in the construct state. etc. (man) of 'Attara". d. Iwldhw. and the -t of the feminine end­ ing was preserved in some idioms.g. but attest the preservation of š (s ) and ś (s ). "(belonging) to A b ū s u . like in Aramaic. Thus the feminine ending -t is replaced by the mater lectionis -h. except in the construct state. However. is most likely a literary expression of the urban dialect of Mecca and Medina in Mohammed's time. ġ. t. dialects with and without case endings coexisted. mn 'zzm = Classical min 'azīzin mā. when indicated in script. There was no longer a fully func­ tioning case system in nouns and the case endings. have probably lost their functional yield. while it has dropped in others. written in a script developed from the Nabataean cursive. Safaitic inscription on a boulder in Wadi Sirhān (courtesy of Abdu-Aziz al-Sudairi): l'bs 'trw. 15. "from anyone strong") and of the case system (e. where ancient Qur'ān manuscripts preserve the spelling -t. of d. The consonantal text of the Qur'ān. The consonants not l 2 . "for his child").g.74 SEÌMITIC LANGUAGES Fig.

without altering the holy text. this lan­ guage was employed by poets whose vernacular may have differed strongly from the archaic Nadjdi dialects. etc. the language of the Qur'ān preserves certain fea­ tures deviating from ordinary Classical Arabic and proving thus that the consonantal text has not been tampered with.. thus testifying to the emer­ gence of an Arabic diglossia. Already before Islam. An important source for the investigation of early Neo-Arabic are South-Palestinian texts from the 8th-10th centuries A. is signified by "s" and z. just as / is indicated by " t " and d by " d " (Fig. emerged from the Pre-Classical Arabic dialects. which was in Old Arabic an emphatic lateral ś.. Neo-Arabic or Middle Arabic is the urban language of the Arab Empire from the 8th century A. Classical Arabic is the language of Pre-Islamic poetry.ARABIC 15 contained in the Aramaic alphabet are indicated by letters marking related sounds. However.D. on. The early Arab philologists of the 8th-9th centuries A. at the latest in the 6th century A. perhaps as early as ca. the system of the "pausal" forms. shaped further to satisfy the needs of poetical diction and of metre. d) Neo-Arabic or Middle Arabic 7. 500 A. with a number of diacritical symbols in order to fix its pronunciation and to adapt it to the rules of Classical Arabic. tā' marbūta (feminine ending). dating back to the 8th century A. It did not arise as a result of the great Arab conquests..D. and standardized in the Abbasid empire. with the Arabic version of Ps. although Mesopotamia and Syria-Palestine provided the Aramaic lin­ guistic substratum that stimulated the development initiated a few cen­ turies earlier and apparent already in inscriptions and in the consonantal text of the Qur'ān. is expressed by the corresponding dental "t". according to a system already established at Tayma' in the Persian period. Thus d. which was an emphatic interdental t.44.D. in Central Arabia. 78 written in Greek majuscules and thus exhibiting the vowel system. c) Classical Arabic 7. despite the various vocalic signs and the symbols for tanwīn (nunation).D. hamza.D.43. that had become sacred very quickly.D. probably based on an archaic form of the dialects of Nadjd. as well as a bilingual Graeco-Arabic fragment from Damascus. have provided the consonantal text of the Qur'ān. in the schools of al-Kūfa and Basra'.. 16). .

16.* •f h w. t 4 o-» J* t i J ii J r Û â J t I J a J r 4 -> — Â J4 ġ 3 5 r J f q k i m n Ji s: l . Arabic Script. ś k b t z.76 SEMITIC LANGUAGES Unbound Bound to the right I Bound on both sides — Bound to the left — Transcription \ ā Name of the letter 'alif bā' tā' tā' gīm hā' hā' dāl dāl rā' zāy sīn šīn sād dād tā' zā' 'ain ġain fa' qāf kāf lām mīm nūn hā' wāw yā' I J o *A* — x A A J A J b t t g c c ii J J h h d d r c a X . ū s J Fig. . A a•* 44 z s Š a- 7 s d.

3° East Arabian dialects of Kuwait. Palestine. With the spread of literacy. and its use with the substantive is limited. and Dosiri tribes are better known. the tenses are associated with the division of time. are no descendants of Classical Arabic but rather its con­ temporaries throughout history. 2° Southwest Arabian in Yemen and Zanzibar. The relative pronoun becomes invariable. Shammar. 7° dialects of southern Egypt. Except for Mal­ tese. Lebanon. and the 'Omānī dialects in 'Oman. Modern Literary Arabic. the adjective. Sudan. interdental spirants have shifted generally to the corresponding occlusives. a direct offshoot of Clas­ sical Arabic. 6° dialects of northern and central Egypt. and the pronoun. that imply different linguistic substrata. no spoken colloquial Arabic achieved official status as a written language. Jordan. and in some villages of Uzbekistan. Qatar. Bahrain. The dual disappears completely in the verb. in the Aleppo area and in oases of the Syrian desert. and they are closely related to Neo-Ara­ bic. Modern Arabic dialects. in southeastern Turkey. Rwāla. Among the Bedouin dialects. e) Modern Arabic 7. In almost all the Neo-Arabic dialects d has merged with z. etc. From the sociological point of view the Modern dialects fall into Bedouin and sedentary colloquials. 8° West Arabian dialects of the Maghrib with Malta and certain regions of western Egypt. Israel. becomes more and more widely known and it is used today for almost all written purposes and for certain formal kinds of speaking.ARABIC 77 7. the asyndetic sentences become more frequent. in Khuzistan (Iran). is instead the colloquial Arabic in its different forms of speech. In the dialects of the sedentary population.45. the following division emerges: 1° Hidjazi dialects in Saudi Arabia. those of the North and Central Arabian 'Anoze. no matter how well educated. but there is some popular literature in various dialects. and the United Arab Emirates. with a marked tendency to place the subject before the verb and to avoid the inserting of the object between verb and subject. According to geographical criteria. The Arabic which is used in ordinary conversation by all speakers of Arabic. 4° North Arabian dialects in Iraq. and Central Africa.46. spoken by some hundred and seventy million people. to which the Arabic idioms of Muslim Spain (al-Andalus) and of Sicily were closely related. . The disappearance of the case and mood endings led to a more rigid word order in the clause. 5° dialects of Syria.

South Arabian Alphabet. 17).78 SEMITIC LANGUAGES 8..J> y % 3 _> t s t(?) t z d Fig. S O U T H S E M I T I C 8. A total of at least 8000 such texts. and in Ethiopic. dating down to the 6th century A.2. 17. The present summary exposition divides South Semitic into South Arabian. In Yemen. with ancient Ethiopic or Ge'ez and various modern languages of Eritrea and Ethiopia. South Arabian 8. A.D. A t the end of the 8th century B.C. but also to shared linguis­ tic features. both epigraphic and modern. appear the oldest monumental rock and display inscriptions so far recorded. at the southern end of the Arabian peninsula. but only some of them have been fully deciphered and published (Fig. Monumental Cursive Transcription Y •y h 1 J* l Y 3 k © ) n X 3 h m q w s 2 _7 r b / t Monumental Cursive Transcription ri J> s 1 fi J k h V J h a o pi J> B 1 n s 3 f ?(d) g Monumental Cursive Transcription J) d Tl Á ġ m x H Jl ? . sometimes called "Ethio-Semitic" in order to distinguish them from the Cushitic languages of Ethiopia. whole or fragmentary.C. a seden­ tary agrarian civilization developed at least from the beginning of the second millennium B. .1. This subgrouping of Semitic languages corresponds not only to geographical criteria. hundreds of cursive texts incised with a stylus on sticks and palm-leaf stalks have been found in the Yemeni Djawf. Besides. have been so far discovered.

6. have been discerned besides the modern spoken South Arabian idioms: Sabaic. 2. Qatabanic. Mahra (Mehri) B .C. South Arabian Languages Epigraphic 1. and Hadramitic (Fig.D. 4. the realm of the ancient kingdom of Saba. Minaic. Sabaic is epigraphically attested from the 8th century B. Saba (Sabaic) Ma'in (Minaic) Qatabān (Qatabanic) Hadramawt (Hadramitic) Awsān (Awsānic) Himyar (Himyaritic) Modern A . • •V \ SAUDI ARABIA OMAN ® „ C O YEMEN (3) ® ^ * * -^ INDIAN OCEAN v . In the 4th to 6th centuries A.SOUTH ARABIAN 79 Four principal languages. its limits extended southward to include the region of Zafar. ) \ GULF OF ^-*^^*^OMALIA Fig.D. 5. the centre of the kingdom of Himyar. and eastward to cover the former Qatabanic and Hadramitic areas. through the 4th century A. Soqotra (Soqotri) a) Sabaic 8. 18. in north Yemen. attested by epigraphical documents. Djibbāl (Śheri) C . 18). where many "Yemenite" tribesmen have settled in the 8th century A. since these languages had by then ceased to be used for epigraphic .3. A number of ancient South Arabian linguistic features have been registered by early Arab grammarians and such occur also in the earliest materials of Andalusian Arabic in Spain. 3.D.

after the Sabaean conquest of Ma'in and of Qatabān. in 'Oman. Sabaic inscriptions dating mainly from the 5th-4th centuries B. at the southern marches of Qatabān. e) Modern South Arabian 8. The Modern South Arabian languages. when Hadramawt was conquered in its turn by Saba.. ancient Samhar. To judge from the name x\ AÛGIVÍTT| f|'icbv given to the East African coast in the "Periplus of the Erythraean Sea" (1st century A. The few inscriptions from the ephemeral kingdom of Awsān. the capital of Hadramawt. §8. in particular at the trading settlement of Khor Rori. are in fact written in Qatabanic.C. Their chronological spread is from roughly the 4th century B.D. and at several widely scat­ tered sites. b) Minaic 8. Qatabanic monumental texts have been found in the Wadi Bayhān. with a few texts from other sites in the east end of Yemeni Djawf. and from scattered places outside Arabia. ancient Yatil.D. to the end of the 3rd century A.C.80 SEMITIC LANGUAGES purposes. and on the plateau to the south of the two wadis. have been found also in Ethiopia. ancient Qarnāwu. Hadramitic inscriptions have been discovered so far in the royal residence Shabwa.5. d) Hadramitic 8. these texts date from the 4th to the 2nd centuries B. Besides.).D. the people of Awsān had led the way in the South Arabian trade along the eastern coast of Africa for which the island of Soqotra was undoubtedly an important sailing centre (cf. which are now confined to a relatively small area in and around Dofar and to the island of Soqotra. the capital of the kingdom of Ma'in.6. through the 2nd century A. they may be written in an Ethiopian language not classifiable properly as Sabaic.7). and at Qaryat al-Fāw. However. ancient Dedān. Chronologi­ cally. c) Qatabanic 8. They date from the 5th century B.7. Besides. Minaic inscriptions are attested at Khirbet Ma'in.C. resulting from Minaean trading activities.C.4. near modern Salālah. in the Wadi Harib. at Khirbet Barāqish. there are texts from the Minaean trading settlements at al-'Ulā. are .

Harari.ETHIOPIC 81 the last vestiges of a group of closely related South Semitic languages. Therefore. An answer cannot be provided easily since the majority of Ge'ez texts are translations and there is no certainty.53). The phonological division between North and South Ethiopic is shown by the Northern preservation of the pharyngals and laryngals. which were spoken in the whole of South Arabia. As for Soqotra. They share many distinctive fea­ tures with Ethiopic. that their syntax has not been influenced by the language of the original texts. 19). it was inhabited in the time of the "Periplus of the Erythraean Sea" by Arabs. Both are generally assumed to be derived from a common Proto-Ethiopic. Tigrinya. and Ge'ez has not yet been sufficiently investigated.8. and by a Greek colony the going possibly back to Hellenistic times. The modern languages exhibit certain features. which preserved its Greek name of Island of Dioscorides. also called Djibbāli. in particular. morphology. and Gurage (Fig. Certain features in phonology. The special attention paid to the Mahra tribe of this region by Arab historians and geographers was very likely due to its peculiar culture and unfamiliar language. and have their own language which none but they understand". the question whether Tigre and Tigrinya are direct descendants of Ge'ez or not should remain open. and it has been doubted whether they can be considered as directly related to the old literary dialects. The main modern languages. while South Ethiopic includes Amharic. The close relationship between Tigre. B. The main morphological differences appear in the secondary South Ethiopic gem­ ination of the second radical of the verbs in the perfect of the basic stem (§41. Tigre. and in the Southern sharp distinction in the conjugation of main verbs and subordinate verbs (§39. Hindus.5). Ethiopic 8. . although the speakers of South Ethiopic may descend from an earlier wave of Semitic immigrants (§8. and Soqotri. which are absent from Epigraphic South Arabian. in the widespread non-gemination of this radical in the imper­ fect (§38. Śheri.000 people. Gafat. however. are Mehri with the closely related Harsūsi and Bathari dialects. and Tigrinya.9). The North Ethiopian lan­ guages include Ge'ez. as it appears from the typical description by Ibn al-Mudġāwir (13th century): "They are tall and good-looking.7). and syntax justify the classification of the Semitic languages of Eritrea and Ethiopia into North Ethiopic and South Ethiopic. Its commercial importance was certainly great (§8. spoken by some 30.12). Argobba.

.Fig. Semitic and Cushitic languages of the Horn of Africa. 19.

8. It was the language of the Aksum Empire. and it is still taught in the Church schools. no defi­ nite conclusion concerning its ancient pronunciation can be drawn on this basis since present-day pronunciation of Ge'ez is influenced by the spoken language. A period of bilingualism fol­ lowed... Tigre is spoken in Eritrea by seminomadic tribal communities numbering some 300. which still endures. The South Arabian inscriptions found in Ethiopia. and particularly by Amharic.D. called also Ethiopic. When Semi­ tes from ancient Yemen settled in Ethiopia.C.D. which were influenced also by Oromo and by Somali (§2.9. as the language of worship and sacred literature.9-11). Harari. However. Tigre. The Bible was translated from Greek into Ge'ez between the 5th and the 7th centuries A. the Cushitic lan­ guages of Bedja. if not earlier (§3.11.000 people. Ge'ez remained a spoken language until the end of the 9th century A. but not with­ out having an impact on the structure and vocabulary of the South Ara­ bian idioms spoken by the conquering Semites.12. and Gurage. The Semitic languages of Eritrea and Ethiopia occupy a geo­ graphical area in which Cushitic was and still is employed. and Tigrinya. prove the existence of ancient rela­ tions between southwest Arabia and Ethiopia and might indicate that Semitic was brought to Eritrea and to Ethiopia from Yemen in the first millennium B.10. The influence of the Cushitic is stronger in the south than in the north.. while Eastern Sidamo or Highland East Cushitic covered the domain of Amharic. which was con­ verted to Christianity in the 4th century A. This influence of the Cushitic substratum on the Semitic languages of Eritrea and Ethiopia is a crucial problem of Ethiopic linguistics. especially at Aksum. in present-day Tigre province.. they imposed their South Arabian language on this Cushitic domain. although the oldest known manuscripts go back only to the 14th century. It survived as a literary language.D. 8. Argobba.ETHIOPIC 83 8.3). The Cushitic group lost ground. is attested by epigraphic texts from the 2nd century A. It is closely related to Ge'ez. and Saho-Afar appear as the linguistic substra­ tum of Ge'ez. and partially of Amharic and Gafat. a) North Ethiopic 8.D.C. Agaw. although it is not certain that it is the direct descendant of the language of the . Ge'ez. In the north. especially those of the 5th-4th centuries B.

mostly Christians. It is with Amharic that Argobba has the greatest number of essential features in common. 8. It was mainly influenced by two Cushitic languages: the Bedja and the Agaw. also on a few grammatical points. Amharic syntax and vocabulary are strongly influ­ enced by Cushitic. and Amharic lacks the archaic features discernible in other South Ethiopian languages. b) South Ethiopic 8. unless stated otherwise. the living Semitic language with the largest number of speakers. from the 19th century. The references to Amharic.13. dating to the 16th century. is only in its beginnings.04- SKM1TIC LANGUAGES Aksum Empire.D. 21). and on the vocabulary. 8. written in Ethiopic script (Fig. The absence of these features in Amharic is due to the fact that it represents an innovated type of South Ethiopic. The references to Tigre in the present Outline are based in particular on the dialect of the Mansa' tribe. it was influenced mainly by Agaw. Harari is spoken in the city of Harar in eastern Ethiopia.16. Some Harari texts. magazines. The oldest Amharic documents actually known are songs from the 14th century A.14). and books being produced. It was spoken also to the south of Harar. It is spoken in the central and southern highlands of the country by some fifteen million people. 8. the language is closely related to ancient Ge'ez. As in the case of Tigre. Amharic is the official language of Ethiopia. are preserved in Arabic script and more recent texts.D. Tigrinya is thus. after Arabic and Amharic (§8. but the lan­ guage disappeared in favour of Cushitic Oromo. in the Tigre province of northern Ethiopia — hence Tigrinya is called also Tigray — and in the central regions of Eritrea. Harari has several features in common with North .in the present Outline are based on the literary language. There are dialect variations in Amharic which bear on phonol­ ogy. with a marked difference between towns and the countryside. especially regarding palatalization.15. but it is developing steadily with papers. Tigrinya literature. Tigrinya is spoken by some five to six million people. Argobba was still recently spoken in a few villages to the north of Addis Ababa. The earliest known document written in Tigrinya is the code of customary law discovered at Sarda and dating from the 19th century A.14. have been written in Ethiopic script.

Leslau from four native speakers. which are alternatively considered as a sub-branch of West Gurage. This word appears as mossa in Amharic and as muda in Oromo..ETHIOPIC 85 Ethiopic and the opinion was expressed that Harar was a military colony from northern Ethiopia. From the three main groups of dialects. Ennemor.g. "little one". and Masqan. Its study is based mainly on a translation of the Song of Songs made from Amharic into Gafat in 1769-72 at the request of James Bruce and on the ample documentation collected in 1947 by W.17. Endegen.18. in western Ethiopia. related to ancient Egyptian ktt. w w . and mossa in Soddo. Wolane. There must have been a territorial continuity between the East Gurage and the Harari speakers. corresponding to Coptic mase. and Gyeto. Gafat was a Semitic language spoken in the region of the Blue Nile. the Eastern ones come closely to Harari and have several fea­ tures in common with North Ethiopic. e. later dis­ rupted by population movements. and the dialects spoken on the five islands of Lake Zway. 8. Gogot. an East Gurage group including Selti. "child".17). It also preserved the noun mossay. At present. Later. It is the only Semitic language preserving. The Gurage dialects are divided into three groups: a West Gurage group including Chaha. and a North(east) Gurage group represented by Soddo or Aymallal. "child". the movements of the Oromo tribes sepa­ rated them. "children". No extra-linguistic data help us yet in answering this question. Gurage is a cluster of rather divergent dialects spoken to the southwest of Addis Ababa by a population numbering about 600. the language disappeared com­ pletely in favour of Amharic. with a possible sub-group Muher. "to give birth". Eza. related to Egyptian mś. from the root mśì.000 persons or more according to other estimations. "calf": m asa in Chaha. m àssa in Muher. Gafat has some archaic characteristics and a number of features in common with the North Gurage dialect Aymallal. called also Soddo (§8. The Soddo and Gafat domains must have been once contiguous. the root is attested in Gurage with the meaning "calf". 8. the plural noun kitac (< *Jcitāti).

as emphatically expressed but unskilfully worked out in his book La langue hébraīque restituée et le veritable sens des mots hébreux rétabli et prouvé par leur analyse radicale (Paris 1815-16). formu­ lated in common types of script the rigid conservatism of which helps concealing local pronunciations. as well as the Amarna letters. often followed by . in particular for the knowledge of ancient lan­ guages. Cuneiform Script 9. the precursors of the phonemes as distinguished from their actual realization (§10. a student of linguistics must remember that writing is still only a secondary representation of language. This was already perceived by Antoine Fabre d'Olivet (1768-1825) who refused to identify the letters and the vocalization of ancient Hebrew writing with actual phonetic elements. A graph in the cuneiform writing system is a wedge or a cluster of wedges imprinted in clay. At a somewhat later stage. since writing systems may condition and even influence linguistic data. A treatment of Semitic scripts lies out­ side the scope of the present work. Most languages have existed and still exist as purely oral forms of communication. the following apercu deals with the essential facts of the Semitic writing systems.1. which uses alphabetic cuneiform signs. that it reflects a standard speech while true dialectal forms transpire but rarely. deprived of sound and gesture. Such a graph is called a "sign" and its referent in the language is called its "value". when Semitic texts first began to be written in it.7). His "signs" were. the graphs of which. The written records of North and East Semitic. However. being aware that these elements are "signs" of the real words. the elements of the cuneiform script consist of syllabic signs or syllabograms. Granted the importance of writing. L A N G U A G E AND S C R I P T 9. or imitations of such imprints in other materi­ als. make use of the cuneiform writing system. A. the texts were arranged in horizontal lines progressing from left to right. were arranged in vertical columns progressing from right to left. and that spoken language provides the final clue for understanding its written expression. in fact. With the exception of Ugaritic. Yet. graphic and largely inadequate representation of spoken language. Writing is no more than a secondary. written records also present indubitable advan­ tages and the debt of modern society to writing is enormous.86 SEMITIC LANGUAGES 9. of word signs or logograms. There is even a greater difference between a living language and a "dead" language.2.

and K I has the values ki. it is difficult. such as scribal conventions and later differentia­ tions of signs. and emphatic consonants belonging to the same "triad". "his head". ip. without being pronounced. is. neither in Sumerian nor in Semitic words. never reached a point where it could be said that every combination of phonemes found expression in the writing. in accordance with genuine East Semitic morpho-phonemic rules. laryngals. unvoiced. te.LANGUAGE AND SCRIPT 87 phonetic complements. use consonantal alphabetic scripts developed from an alphabet created in . the distinction of interdentals and dentals. de. and the changes occur­ ring between earlier and later texts cause problems for the correct analy­ sis of the Semitic phonology. B. ep. the sign D I stands for di.4. ez. eb. 9. correspond to an actual pronunciation qaqqassu. but the distinction of i and e does generally not find expression in the writing. but may also signify yib or yip at the beginning of a verbal form. Word dividers consisting in small vertical wedges occur irregularly in Old Assyrian texts and they are often used later in Ugaritic cuneiform alphabetic script. the cuneiform sign IB has the values ib. This applies in particular to the Ebla texts that cannot be under­ stood by taking the cuneiform signs at face value. The West and South Semitic languages. for instance. Thus. ke. the indication of the length of vowels and of the doubling of consonants never received a satisfactory and unam­ biguous solution. qé. therefore. the morpho-graphemic spellings like qa-qa-ad-šu. and semivowels. The sign GIŠ has the values iz. which are often described as reflecting the deep morphological structure of the language (qaqqad + šu). to reach phonetically satisfactory conclusions without using data drawn from comparative Semitic linguistics. ti. also the consonantic elements require an appropriate evaluation and an interpre­ tation. qí. es. as well as Ugaritic. Besides. In short. the notation of pharyngals. Alphabetic Script 9. besides giš. The writing system was not designed for Semitic and palliatives. and of determinatives that specify the class or category of the word which they determine. In other words. is. In particular. of voiced. the local variations in the use of signs. The Sumerian or Pre-Sumerian origin of the cuneiform writing system. es. The indication of vowels by syllabograms is of con­ siderable assistance to the linguistic analysis.3.

even long medial vowels: w was used to mark ū/o. The ambivalent use of w and y allows some­ times for the possibility that either the diphthong aw/ay or a long vowel is represented in a word. Greek o was not borrowed directly from Semitic but by application of the acrophonic principle to the Greek translation 6(p9aA. h was used initially to mark final -ē and then also final -ā. 9. 9. The main lines of the evolution of the Semitic alphabet are shown schematically in Fig. The Semitic alphabet was originally purely consonantal in charac­ ter. These three signs could be used also to mark the vowels a. However. written in monumental . The use of the matres lectionis w and y is also attested in the South Arabian type of alphabetic script. and later in Arabic. the Ugaritic script of the 14th century B.5. "the life".u. perhaps as early as the 8th century B. the West and South Semitic lan­ guages used its original linear form which developed into two distinct types of letters: the so-called Phoenician alphabet with twenty-two let­ ters and the South Arabian alphabet with twenty-nine letters. and based on Egyptian hieroglyphic signs.C. there is no notation at all for ā. 20. Besides. like in later Semitic texts. u.6.C. Three or four consonantal signs of the Phoenician alphabet received a supplementary function in order to indi­ cate long final vowels and. Christian Palestinian Aramaic hyy' l(h)ayyal. Only Mishnaic Hebrew and some Late Aramaic dialects show the practice of indicating consonantal w and y by a double spelling ww and yy\ e. y served to indicate lie.g. probably because its creation was inspired by the Egyptian hiero­ glyphic "alphabet".. This vocalic use of the letters under consideration was borrowed by the Greeks together with the Semitic alphabet and was extended to short vowels. Mishnaic Hebrew ywwny /Yawnē/ instead of Biblical Hebrew ybnh lYabnēl.88 SEMITIC LANGUAGES Canaan in the mid-second millennium B. While the Ugaritic script represents a cuneiform adaptation of this new writing system. to a limited extent. short or long. distinct from the original ' that received the value 'a. Instead. for which also ' served in Aramaic. "eye". of Semitic 'ayn. i/e. at least in Human texts written in alphabetic cuneiform script. not even in the Pre-Clas­ sical Arabic inscriptions from Qaryat al-Fāw. with the same vocalic values ū/d and ī/ē.C. already possesses two supplementary signs 7 and 'w.6c.C. a fully developed use of matres lectionis or vowel letters appears in Aramaic and in Moabite as early as the mid-9th century B.

Proto-Canaanite Beth-Shemesh tablet = "South Arabian" Phoenician Western Eastern Jewish Nabataean Palmyrene Syriac Mandaic Arabic Fig. 20. Evolution of the Semitic alphabet. .

however. 'dh /'idâ/. These deficiencies have been partly obviated in the 7th-9th centuries A. the Late Punic and the Neo-Punic inscriptions did employ w. "tight". 9. Besides. In some Late Phoenician inscriptions from Cyprus and in Punic. an abbreviation of yàlalla. "your brother". "that is loose".7. according to two differ­ ent systems (§21. The orthography. h can be used for a. Vowel notation by means of matres lectionis does not fix the meaning and the reading of texts in an unambiguous way. h for e. like in 'hwhm /'ahūhum/. there is a notable deficiency in the absence of any consistent marking of gem­ inated or long consonants. In addition to these 231 forms.14). Two additional symbols indicating gemination and non-gemi­ nation are often used in traditional grammars written in Amharic. The South Arabian script has been adapted in Ethiopic to denote seven vowels by a variety of changes in the shape of the consonantal symbols. while the non-gemination is marked by la.g. there are thirty-nine others which represent labialization and are usually listed as an appendix to the main list (see Fig. the pronunciation of Ge'ez preserved in the Ethiopic Church is influenced by Amharic. comparable to some extent with the cuneiform writing system.90 SEMITIC LANGUAGES South Arabian script (§7. and ' for e and o. Besides. mh /mā/. w and y are exceptionally used as vowel letters in foreign names or words. Contrary to the other West Semitic languages. h. "what".8. except in a few forms brought about by linguistic change. placed also above the letter. \h. Phoenician did not use any vowel letters. The best represented system uses ' for a. and ' as vowel letters. an abbreviation of pbq. they indicate an internal long vowel as well.' stands for a. and ' for o. However. and it uses the same set of symbols to mark the vowel d and the absence of any vowel. placed above the letter. y. In the second system. The gemination is marked by a small p. 9.42).D. by a complicated system of diacritical signs . "while". 9. Vowels have thus become an integral part of Ethiopic writing which assumed a syllabic character. Occasion­ ally. Besides. the Lihyānite inscriptions of the Hellenistic period follow the Aramaic scribal tradition and use h as a vowel letter for â. e.9. 21). The latter uses the traditional Ethiopic syllabary with additional signs: it has thirty-three characters. each of which occurs in a basic form and in six other forms known as orders. Besides. however. has two defects: it does not indicate the gemination or consonantal lengthening.

ft. <K Ax 0O <% •» IP V* 9 <h 1° *% I Ù n <* 6 ii it ** & ÌÍ y Q ft 75 •fl A A.T l <P <*. ? p. 7* ft. AM u tf A" A r h<h m s<š r s š q b t a <h ih. 21. h. *) »i tf 1 •* k h<k w z z y d è g t c p s S < Ś tb h. ft. . P «7 m m.L A N G U A G E AND SCRIPT 91 Consonant + w + Vowel 1 w 1 1 Name of the letter ! hoy lawe S hawt may šawt rees sat šat qāf bet tawe cawe harm nahas nahas 'alf kaf kaf wawe 'ain zay zay yam an dent ġent garni tait cait pait saday dappa af pesa Trans­ cription Consonant + Vowel 1 2 3 4 5 6 7 a/a u i a/ā e 9/0 o V V* a w 3 i w 4 a w 5 e 6 h 1 ii A- it /t y % A. * ÎÍ * * a- * p. h Ti. * ft- d T ft /» f p T X 7 r * Fig.ft0* ft. IT * 7 7« X m m* Ok fitfe ft 0 * ft. T *n C k Cr ft k ft ft 4. K ft. + + ì 7 h h Ii a* 0 II * i h<h n V ? n » T y. H & ll H Ii. Ti •» w ù /> II IT f A TP 7 T TC P ft. n. Amharic syllabary.

or by diachronic references to later spellings. etc. The real phonemic status of the languages using the Phoenician alphabet can only be established by synchronic comparisons with cuneiform. the three sounds /. placed one on top of the other. The use of diacritics is widespread and serves to distinguish various sounds expressed by the same consonantal symbol. and ś were all designated by the same symbol " š " . They go back either to a verti­ cal stroke used as word divider or to a pair of dots arranged like a colon (:).D. to much later diacritic signs. and Classical Arabic. That "punctuation" system was further developed by Arab scribes who called it naqt and used diacritical dots to distinguish consonantal phonemes represented by the same characters. the script of which derives from the Nabataean Aramaic cursive. Hebrew. 9. Thus. Using the same principle one gets h and c from " k " . The old­ est attestations of a diacritical dot distinguishing d and r are found in the Palmyrene inscriptions of the 3rd century A.10. š. and eventually to the pronunciation of some consonants in modern conserv­ ative idioms such as Modern South Arabian. and later for Classical Arabic. later reduced to one dot. The so-called Phoenician alphabet was used for Aramaic. The use of these diacritics is attested in the earliest Islamic papyri and inscriptions from the 7th century A. Hebrew. The three systems are used in the Aramaic Tell Fekherye . but no definite conclu­ sion concerning the older vocalizations can be drawn on its basis. the languages of Transjordan.92 SEMITIC LANC.. In Early Aramaic.g. and South Arabian spellings. in Arabic and in Neo-Aramaic. Jewish Aramaic. With a sim­ ple dot placed under " g " one obtains ġ. hieroglyphic. with a small upside-down v-like diacritic under the same letter. sometimes to three dots. by adding special diacritics to a number of the original twenty-two letters.DACES aiming at fixing the pronunciation of Syriac.11.D. The pronunciation thus fixed was a traditional one. and in Syriac. except in the Tell Fekherye inscription of the mid-9th century B. e. especially for the reading of the sacred texts. for example. Different punctuation signs have been used in the alphabetic script to divide each two words of a text. 9. one gets ġ. new sounds are represented.C. A similar system was adopted in modern times to write spoken Aramaic that contains an expanded sound system comprising some thirty-one consonantal phonemes. where / was indicated by the letter "s". The twenty-two symbols of that alphabet could not express the Semitic phonemes which did not exist any more in Late Canaanite and Phoeni­ cian languages.

Besides. § 38.. When the transcription differs from the simple transliteration of the signs. The pair of dots and the single dot are better attested. although occasional references to such tran­ scriptions occur in the part dealing with phonology. From the mid-first millennium B. However. The three dots occur on the Lachish ewer from the 13th or 12th century B. of mediaeval Arabic texts written either in Syriac script and named garšūnī. in accordance with the requirements of an introduction. there are Hebrew texts. for exam­ ple in Nabataean Arabic: fa-yafal lā fidā wa-lā 'atarā (pyp'l V pd' wl' 'tr'). instead.C. while the Moabite Mesha inscription uses small strokes to mark out sentences or contextual units. This is the case.12.C. No attempt is made in the present Outline to deal in a systematic way with the problem of transliterating foreign names and words into a Semitic writing system.13. follows the usual conventions and is based mainly on the standard form of the languages concerned. The vertical stroke keeps with the tra­ dition attested in Ugaritic by the small vertical wedge and anticipated in Old Assyrian texts (§9.C. which uses four square dots arranged in a square pattern (::) as a sentence divider. and in two lines of the Tell Fekherye inscription. Transcription and Transliteration 9. This practice was continued in West Semitic inscriptions of the 11th and 10th centuries B. the pair of dots (:) is used as verse divider. in ."who campaigned up to Thadj". In Masoretic Hebrew.. 9. in archaic Greek writ­ ing. A different but related problem concerns the use of one offshoot of the Semitic alpha­ betic script to write texts in another Semitic language.. there are West Semitic inscriptions and even Ethiopian newspapers where the words are run together.11). or in Hebrew characters and called "Judaeo-Ara­ bic". mainly biblical and liturgical. and in Epigraphic South Arabian. In particular. C.C.TRANSCRIPTION AND TRANSLITERATION 93 inscription of the 9th century B. the latter is also given. Allophones are indicated only in special circumstances. two square dots are employed as word dividers in the Ethiopian writing system. The transcription of Semitic words.2). and this practice began to be fol­ lowed also by printers of modern Ethiopic texts. "and he acted neither for reward nor for favour" (cf. space was used to separate words in West Semitic instead of dots. or dū 'asm li-Tāġ (dw 'sr' /fg). which is employed in this work. in particular.

94 SEMITIC LANGUAGES Arabic transcription. Such texts may have a great linguistic importance. . and of Hebrew words in Origen's Hexapla and in a few other works. Instead. but an Outline cannot enter into the discussion of questions they may raise and dialects they reveal. and there is a Berber translation of a Passover Haggadah in Hebrew characters. occasional reference will be made to the vocalized transcriptions of Punic words in the Poenulus of Plautus.

as well as the transcription of one language in the alphabet of another when this script is inherently unfitted to be the vehicle for an automatic transcription. Thus. I f the linguist and grammarian takes great interest in them. it cannot be neglected in the study of ancient languages and it will be used in the present work. Such material. as the Ashkenazic. There­ fore. therefore. proper names change pronunciation along with the rest of the language and. Although we are dependent on the orthography for discovering the phonology of ancient languages. it is a matter of great methodological importance to distinguish between orthography and phonology in considering written documents. However. the famous Sibawayh's treatise on Arabic gram­ mar. being part of speech. for example. As for the mod­ ern proununciation of Semitic languages. Now. The sounds of speech can be analyzed from various points of view (§10. written in the 8th century. Although this phonetic material is in general limited and subject to mishearing. we cannot base our phonological infer­ ences on the statistical predominance of a conservative spelling in the available sources. consists generally in proper names. apart from a few scattered glosses.1). their transcription in other languages may provide some help in following the evolution of speech sounds. although relatively static and isolated com­ munities.1. or Yemenite pronunciation of Hebrew. Particularly interesting and more revealing are the lapses. The twentyeight characters of the Arabic alphabet. Yet. enumerates forty-two consonantal speech sounds registered in Arabic by this doyen of Semitic linguistics.II PHONOLOGY 10.2). may preserve old South Arabian pronunciations and articulations. it does not reveal all the phonetic richness of the lan­ guage and does not follow its evolution in an adequate way. Sephardic. as those of the island Soqotra and of the montainous regions of 'Oman. he should bear in mind that the analysis of speech sounds of ancient languages is based mainly on their written notation which is imperfect and often conserva­ tive (§9. it is far from trustworthy in deter­ mining that of earlier periods. often concealed by the conser­ vatism of scribal practices. . it is because they are the phonetic manifestation of the morphemes which are the minimal units of any grammatical structure. are generally believed to correspond quite well to the consonantal speech sounds of Classical Arabic.

A synopsis of the two notation systems should make it clear. but word stress is shown in the present Outline by an accent placed above the vowel of the stressed syllable. airflow and intraoral pressure measure­ ments aim at explaining the aerodynamic conditions of speech produc­ tion. for practical reasons. and traditionally employed symbols and diacritics have been used to a great extent. [a]. as far as known and described precisely. e.g. Speech sounds can be classified first into conso­ nants. spectrography observes speech displayed in the form of acoustic energy. Arabic kátaba. can be expressed in a fairly adequate way when one uses the symbols of the international phonetic alphabet (in brackets). e. These experimental procedures go over into the field of phonetics as soon as they describe the bases for the classification of speech sounds as such. 'kataba. Linguistic Analysis 10. The study of the articulatory movements that produce speech sounds is prelinguistic.3. In the international phonetic alphabet. 10.g. It is customary to put phonetic symbols in brackets. the stress is indicated by an accent placed at the beginning of stressed syllables. "he wrote".2. e. However. Consonants = [?] = [Î] b = [b] ' (glottal stop) (voiced pharyngal) (voiced labial) . A l l the sounds of the spoken Semitic languages can be subjected. The various sounds of Semitic languages. B A S I C ASSUMPTIONS A. The linguistic analysis of the sound of language as a whole and of specific languages can be considered under three headings: 1° the study of the articulation of speech sounds. in accordance with the requirements of an introduction and with the widespread practice of teachers and stu­ dents of Semitic. to experi­ mental investigation: thus. and tones. while glottography and laryngography help in stating the function of the glottis.96 PHONOLOGY 1. 3° the functioning of speech sounds in the language structure (phonemics). [p]. in one way or another. the system of transliteration has been kept as simple as possible. vowels. being con­ cerned with physics and physiology.g. 2° the classification and description of speech sounds (phonetics).

spirantized) ~k = M (palatalized voiceless velar plosive) I = [1] (liquid lateral) I = [*] (velarized voiced lateral) m = [m] (labial nasal) n = [n] (dental nasal) h = [n] (palatalized nasal) h = [n] (post-palatal or velar nasal) (voiceless labial) P = [p] (do. [c] (voiceless palato-alveolar affricate) d = [d] (voiced dental plosive) (do. spirantized) (fricative palatal) S = [9] 6 = [tJ]. [t] (emphatic voiceless dental plosive. emphatic fricative lateral. velarized) b = [P1 y y y s 1 2 3 . q = M . uvular trill) r =M (velarized voiced trill) s = [s] (voiceless fricative dental) s = [?L [s'l. globalized or velar­ ized. spirantized) P = [q>] ]. [d] (emphatic voiced dental. w (emphatic voiceless interdental. [4]. globalized or velarized) t = [9] (voiceless interdental) t = [t]. [t'l. globalized or velarized) (palatalized and globalized voiceless velar plosive) q = ÍKì r = [r]. [k': [k] (emphatic voiceless velar plosive. affricate) ś = [4] (voiceless lateral fricative) ś = »] (emphatic lateral fricative) š = [/] (voiceless palato-alveolar fricative) š = [f'] (emphatic/globalized voiceless palato-alveolar fricative) s = [/] (voiceless palato-alveolar fricative) s = [4] (voiceless lateral fricative) s = [s] (voiceless fricative dental) t = [t] (voiceless dental plosive) t = [t]. [R] (liquid trill. spirantized or voiced interdental) d = [§] 4 = [d]. I ] > [ts] (emphatic voiceless fricative dental. spirantized) 8 = [y] (voiced palato-alveolar affricate) = [d3l Ì (voiced velar fricative) 8 = [y] (palatalized voiced velar plosive) 8 = [§] h = [h] (voiceless laryngal) h = [h] (voiceless pharyngal) h = [x] (voiceless velar fricative) h =w (palatalized voiceless velar fricative) k = [k] (voiceless velar plosive) k = [x] (do. velarized voiced dental) (emphatic voiced interdental) d = [d] (palatalized voiced dental plosive) = [41 (voiceless labiodental fricative) f = m (voiced velar plosive) 8 = [g] (do.BASIC ASSUMPTIONS 97 (do.

p. The spi­ rant form of b g d k p tin Hebrew and Aramaic is normally not marked (§11. Syriac. p .g. q . samā.g. lower-mid)) (non-syllabic) (nasalized) (non-phonemic) (mean-mid. t are used. ra'sun. Hebrew mappiq and Arabic hamza are simply indicated by transliterating h (e. are not transliterated unless the orthography needs to be pointed out. back) (non-syllabic) (central) (high. m = [a] = [*] = [e] = [e:] = [e] = b] = [i] = [i:] = m = [0] = [0:] =W = [6] = [e] = [u] = [u:l = [u] (low. o. g. A dot under the letter indicates its emphatic pronunciation. front) (long) (non-syllabic) (lower-high. e. "heaven"). velarized) (voiced palato-alveolar fricative) (voiced lateral fricative) w w w w w w Labialized consonants are transcribed b . p. back) (long) (lower-mid. /. "the king"). w. ī. "day".10). under­ lined symbols b. k. y of Hebrew. ydm. k . but the long vowels ā. Normally. back) (long) (non-syllabic) .g. Hebrew dageš forte and Arabic šadda are shown by geminating the consonant (e. g . r. front) (long) (non-syllabic) (mean-mid.g. Vowels a ā a â ā a e ē ê d i ī i 0 6 0 6 6 u ū ù = M = [a:] = [«]. ē. hammelek [ham:elek]. Aramaic.g. back) (long) (higher-low. Otherwise. h .98 PHONOLOGY í p w y z z z z = [ts] = ra = [w] = [j] = [z] = [?]. "head"). central) (high. Arabic. i f helpful for pointing out the etymology. d. The medial and final vowel letters h. 'arsāh. ū thus indicated are shown in transcription (e. [e] = [ft]. "her land") and ' (e. [a] = [3l = [fe] (voiceless dental affricate) (palatalized voiceless dental plosive) (voiced labial velar) (palatal) (voiced fricative dental) (emphatic voiced fricative dental.

BASIC ASSUMPTIONS 99 No graphemic distinction is made between long vowels resulting from a monophthongization or marked by a mater lectionis. "half-close". as in Rendille géèl. which mark the absence of any following vowel. as shown in the following figure with the letter 3 (b) as exam­ ple and with the names of the Tiberian vowel signs: 3 ba {patah) bi (hireq) 3 ba/b (hatef patah) a bā/b6 (qames) bo (holem) b6 3 be (segol) 3 bu (qibbūs) 3 bē (sere) 3 ba (tewa mobile) (hatef qames) bè (hatef segol) 0 The "furtive" patah. Consonantal sounds are described in terms of points of articula­ tion. these distinctions will not be followed for Hebrew and Biblical Aramaic. In the articulatory description of vowel height. and "open" to "high". when there are two level tones. some authors prefer the classificatory terms of "close". which is an artificial ultra-short vowel a inserted in Hebrew before a final guttural (§27. i. "mid". I. The šdwa quiescens of Hebrew. and other long vowels: they are all indicated in this Outline by a macron. Consonantal Sounds 10. ū. as Oromo beeka. the sukūn and the ġazma of Arabic.4.10). is transcribed like the hatef patah used in similar circumstances after a guttural. mainly mechanical transliteration of Tiberian vowel signs (§21. ē. for Libyco-Berber. low and high. as Tuareg. are not indicated. and "low" because of the belief that the former category provides clearer distinction. The same system is followed. although specialists in the field and a recognized orthogra­ phy of some languages. duplicate the vowel symbol. Instead. For instance. "to enter". [pi is called a labial because both lips are brought together to produce the sound. "he knows". of the various obstacles to the freely vibrating or moving air as it passes out from the throat passage. and Iraqw. a usual. as opposed to the [t].e. as in ā. the vowel symbol is doubled in this Outline. and Chadic. However. Considering the various traditional pronunciations of Hebrew vowels and their intricate historical development. B. as a rule. which is called a dental sound because the tip of the tongue is . "half-open". instead of bēka. Cushitic.19) will be adopted. Oromo.

C. Vowels 10. Vowels may be described as sounds produced in a resonance chamber such that there is a minimum of interference with the freely and regularly vibrating air as it passes out from the throat passage. [t]. which is called an interdental because it is produced by placing the tip of the tongue between the upper and lower front teeth. Instead. According to the first crite­ rion. The sound of palatal y ([j]) is produced by placing the front (not the tip) of the tongue near or against the hard palate. as [p]. otherwise it is voiceless. may be articulated with greater length (§23. and from [k]. Consonants are also described in terms of the activity of the air stream in the mouth and the activity of the vocal cords. Another articulatory contrast opposes a lax articulation to a tense one which is characterized by greater energy resulting either in con­ sonantal lengthening. since the tongue is bunched toward the back of the mouth. This differ­ ence has important consequences for the phonology since stops cannot be lengthened without changing quality. Both of these differ from / or [6]. either "king" (Phoenician) or "estate" (Arabic). called "continuants". When the air stream must pass through a narrow opening.1). 22. as in [f] or [s]. is the highest of the front í . and [u] can be described in the following way: the vowel [a] as in kalb. Vowels are classified by two criteria: 1° tongue height and tongue advancement or retraction. the three basic Semitic vowels [a]. is the lowest of the back vowels. [k]. Both differ from the pharyngals and the laryngals which are artic­ ulated respectively in the pharynx and in the larynx. the velum is dropped and part of the air stream passes through the nasal cavity. The various points of articulation are represented in Fig. 2° lip spreading or rounding. but it is not high toward its roof. It differs from the palato-alveolar š (\f]) formed with the front of the tongue touching the hard palate near the alveolar ridge. or in a sharper onset and/or wipe-off of the consonant.100 PHONOLOGY at or near the upper front teeth.5. e. When the vocal cords are vibrating the sound is said to be a voiced sound. the sound is called a fricative. which is articulated with the back part of the tongue somewhere in the region of the velum. as [p]. as [b]. In the case of the nasal consonants such as [m] and [n]. This may vary from a slight lengthening in time of the pronunciation to much more than double. [m:]. while other consonants. hence it is called a velar sound. the vowel [i] as in milk. [i]. "dog". The air stream is continuous in the frica­ tives but interrupted in the plosives or stops.g.

vowels. [u] . 23. The position of the tongue during the articulation of these three vowels is roughly indicated in Fig. the vowel [a] is open and has no significant rounding or unrounding. since the mouth opening is slight. 22. 23. The vowel [u] as in šulmu. while [i] and [u] are close vowels. [a] [i] Fig. Points of articulation. Articulation of vowels. is the highest back vowel. since the tongue is bunched forward in the mouth and is high toward its roof. but [i] is unrounded and [u] is described as a close rounded vowel. "well-being" (Akkadian). According to the sec­ ond criterion.BASIC ASSUMPTIONS JOJ Fig.

. bara.11) was a tone language distinguishing between high {a). intonation distinguishes one word from another. at present. for example. lexical dis­ tinctions may be based on tone.6. The intonation is the rise and fall in the pitch of the voice. low (à).g.. "They made peace". " D i d the kings make peace?". a rising tone. However. it is not practicable to mark this kind of into­ nation in Semitic orthography and there are no punctuation marks designed at signifying an interrogation or an exclamation. a falling tone. since this would explain the number of homophonous signs in Akkadian. "boy". High tone is indicated by an acute accent (á).102 PHONOLOGY D. Intonation 10. the two meanings of the interrogative šarrânu. islimú... while low tone is either left unmarked or indicated by a grave accent (a). and that sex gender of nouns designating human beings or animals was spec­ ified in "Proto-Sam" by the high-low tone for the masculine and the low-high tone for the feminine (e. such as Chinese and most Bantu idioms. for instance.2). "learner". In the languages called tone languages. Speakers may distinguish. with the pitch rising at the end of šarrānu. and fad­ ing at the end of the answer íslimū. Rendille ínàm. "girl"). among a high t6ne. as in gara. It is likely that the "ProtoSam" sub-group of East Cushitic (§2. tone plays an important part in some Cushitic languages. in general. "year". ìnám. The maximum num­ ber of tones systematically used in any one language to distinguish mor­ phemes seems to be about five. The rising into­ nation may be indicated in cuneiform script by an additional vowel sign (is-U-mu-u) but.3-4). In Semitic. a low tone. and high-falling tone (áà).. and gará "stom­ ach". islimū and of the declarative islimû are distinguished phonetically by intonation. In par­ ticular. Semitic languages are not tone languages and. In Oromo. This might have also been the case in the Sumerian or Pre-Sumerian language for which the Mesopotamian writing system was originally designed. in Old Babylonian. that may have been distinctive in Sumerian or Pre-Sumerian. and bará. Thus. and a tone that falls and then rises. and intonation can affect the meaning of whole sentences. "towards". the tone is not an integral part of any Semitic word. that may consist of a single word (§50. tone must have distinguished the preterite (*yíqtùl) from the jussive-(*y^íẃ/) (§38. intonation conveys shades of meaning which cannot conve­ niently be expressed by other means.

BASIC ASSUMPTIONS 103 E . The members of a phoneme are called its allophones. When two words differ by only one phoneme. then the two phonemes are minimally distinct phonemes. where the internal voiced labial plosive is geminated. "bit". the postvocalic p is spirantized and pronounced as a labial fricative [cp] or p. "he lived long". however. and 'amara. or 'ebed. etc. Current linguistics distinguishes sharply between speech and lan­ guage. In nearly every language the number of distinguishable sounds is often quite large and greater than the number of consonants and vowels indicated by a current writing sys­ tem (§10. "servant".e. i f there is a difference of only one distinction between the two phonemes in question. one is dental. or rabblm. Such a class of sounds is called a phoneme. By grouping the sounds in such a way. in Hebrew the word pat. where the postvocalic spirantized b is pronounced as a labial fricative [p] or b. "he perished". The phonemes /d/ and ' or / Î / in Arabic damara. the words are said to be a minimally distinct pair. Semitic languages have between 35 and 50 phonemes. These various p sounds are said to be members of a class of sounds which. are not minimally distinct. "bird". Phonemes 10. i. For instance. as a whole. one gets a limited number of phonemes in each language. for instance with the class b represented in such Hebrew words as bat. It is customary to represent phonemes by symbols enclosed in slant lines. "many". both are voiced sounds. short and long vowels. /p/ and /b/ are both plosives or stops. because one is plo­ sive. the internal consonant sound is a voiceless labial plosive geminated. which begins with a voiced labial plosive. both are labial and the only difference is that /p/ is unvoiced and /b/ is voiced. the phoneme /p/. the phoneme /b/. Of course. Instead. the other pharyngal.1). the number of significant differences is smaller and may correspond more or less to the number of consonants and vowels marked by a writing system. In a minimally distinct pair of words. begins with a voiceless labial plosive or stop. . phonemic distinctions differ from one language to another. the other fricative. and intonations.7. Thus in Hebrew pat and bat. "daughter". hence each language has its own set of phonemes or distinctive sounds. is in contrast with other such classes. that are conso­ nants. "north". such as Hebrew pat and bat. in sippor. in sāpon. between sounds and phonemes.

as well as the Egyptian transcriptions k-p-n and k-b-n of Gbl. b'l and p'l. The his­ torically attested spellings 'bd and 'bt.g. for instance. /d/ and /t/. Neo-Aramaic glabtā [ġlapta]. "to be heavy". "breath".5-6). is well known in the spelling of foreign names and it occurs in informal texts as well. 18. nbk and npk. "number". "Byblos". "this". "life". ndn and ntn. e. dmr > zmr/dmr and tmr > smr. e. "to protect". wst and wsd. dū and íẅ ("šu") > še. "to guard". the history of all languages that can be followed over a long period of time shows that voiceless occlusives become voiced.104 PHONOLOGY F . normal in speech. Early evidence points to a similar situation in ancient Semitic. or velar phoneme.10. and the frequent lack of differentiation between voiced and unvoiced sounds in Semitic cuneiform writing may in fact suggest that /b/ and /p/. interdental. Another development consists in spirantization or palatalization of occlusives in order to ease the enunciation (§11. . In this hypothesis. kbd and kbt. Tigre 'ādad ['ādat]. "to make". where the phonemic distinction between voiced and voiceless consonants is non­ existent. The description of the minimal differences has a certain impor­ tance in comparative Semitic phonology. "full moon". these allophones or phonetic variants would have received a phonemic status in the languages con­ cerned. Voiced and Unvoiced Sounds 10. "well". and spoken Semitic languages show that voiced consonants may become voiceless in contact with other consonants and in final position in the syllable. "to perish". the original Proto-Semitic consonan­ tal pattern could be compared with Sumerian and Chinese. the Syriac inscription 'lshāq bar Dāwīt or the Latin coin legend Turris Davit. dental. Still in prehistoric times. A similar devoicing of occlusives occurs also in some Indo-European languages. and perhaps some other similar pairs were origi­ nally allophones or free variations of the same labial. "who". but b/p is again treated as one phoneme in Neo-Assyrian. /g/ and /k/. In Mycenaean Greek. "victory".8. nbš and npš. "to give". etc.. "to be firm". since the distinction between voiced and unvoiced sounds. In any case. this distinction is expressed graphically only for the dentals.g. as in German before other occlusives. Such a devoicing. šbt and špt. /d/ and /t/. and it is missing in the Cypriot Greek syllabary. "sabbat". The phonetic realization is another question. might not be an original fea­ ture of Proto-Semitic.

a sombre «-like quality that tends to spread over the whole word. which probably originated from t and ś. ancient phonetic changes and transcriptions of the emphatics z and d. called tafhlm in Arabic. Their phonemic load. mty > mġy. with vocal cords tightly closed and pushed upward. "to be alert". "to arrive". and shq. Because of this spread of the suprasegmental velarization. and the surrounding vowels. "to laugh".BASIC ASSUMPTIONS 105 G. inassur instead of inassar. while h is a voice­ less velar fricative. Besides. k\ also transcribed p. support the primitive character of the pharyngalization which characterizes the Libyco-Berber emphatics as well. A different problem is raised by the Semitic emphatic sounds that are pronounced nowadays in the Ethiopian languages and in Mod­ ern South Arabian as ejectives. . accompanied by a velarization. This velarization gives them. the characteristic articulatory feature of all the emphatic phonemes is the contraction of the upper pharynx. "he guards". "mountain". such as dhk. Besides. instead. thus 'k. and followed by a glottal stop ': p'. velarization or pharyngalization in Arabic — should be considered as primary. as well as the appearance of the vowel u in the neighbourhood of emphatics in East Semitic. è. "to guard". s. t'. all the aforementioned changes and transcriptions point to a pharyngalization. how­ ever. However. is very limited. t. k/q. Emphatic Sounds 10. new emphatic consonants arose also in modern dialects. and qtl. In standard Ugaritic. It is uncertain which of these characteristics — glottalization in Ethiopic. there are pre-glottalized allophones pronounced with a closed and stationary glottis in the initial phase of the articulation. the latter can be seen by means of a radioscopy which shows how the emphatic phonemes are articulated with a raising of the back part of the tongue in the region of the velum. s\ c'. 't. ntr > nġr. i. tr > ġr. A comparable phenomenon is attested in Aramaic by the spellings " q " and " " ' used to mark d < ś and by the Neo-Assyrian transcriptions of this phoneme with hi or qi. like in Ra-hia-nu I Ra-qi-a-nu for Raśyān. The appearance of the velar fricative ġ signifies that the pharyngalization of the interdental fricative t had supplanted the basic character of this phoneme. the following changes are ascertained: tm' > ġm'. "to be thirsty". etc. may explain the variant forms of certain Semitic roots. śhq. Since " q " marks a velar plosive and " " ' was used to indicate also the voiced velar fricative ġ.9. or qtl. the spreading of the velarization over the whole word. as far as it is to be considered. and perhaps yqt > yqġ. In Arabic. ktl. "to k i l l " . 's. for instance in qurbum instead of qarbum.e. "near".

points to its being a secondary feature. the com­ mon Semitic or Proto-Semitic phonemic system can be reconstructed with a high degree of probability. on the other hand. supported in Ethiopia by the influence of the Cushitic languages. The geographical setting shows. orthographic peculiarities. and also Hebrew as pronounced by Georgian-speaking Jews.8). therefore. transcrip­ tions in other languages and scripts. Although there remain doubts and uncertainties. and it is out of them that the words and the grammatical forms are formed. The phonemes constitute the basic structure of the material of the language. that tends to be replaced by the glottal stop or by a glottalized velar plosive k' (among the Georgian-speaking Jews). that this development did not happen under the influence of Cushitic (§18.10. it must result from an articulation which is limited to the glottal contraction. Economy of effort seems thus to have brought about this development which does not indicate. such as tra­ ditional pronunciation. and in Hebrew as realized by some Jews of Algeria and Morocco. in Tigre before a consonant. Proto-Semitic Phonemes 10. As for the phenomenon q > ' in many Arabic urban dialects. Modern South Arabian. H. . description by mediaeval grammarians.11. The phonemes of spoken Semitic languages can be described and analyzed on the basis of observation of what happens when speech is produced. without the retraction of the tongue and a raising of its back toward the soft extremity of the velum.106 PHONOLOGY The fact that the glottalization of emphatics is not found in Semitic outside Ethiopic. Therefore the replacement of the pharyn­ galization or velarization of the emphatics by their glottalization may reflect the same phenomenon as the change ' > '. 10. The phonemes of ancient written Semitic lan­ guages are reconstructed on the basis of various indications. and comparative Semitic linguistics. that the dialects in question had a glottalized emphatic velar in an earlier period. One should rather note that all these forms of speech are also characterized by an almost complete non-occurrence of the pharyngal '. but it cannot be ascribed to the sole influence of Cushitic. while the laws of phonetic correspondences between the branches of Afro-Asiatic have not been sufficiently elucidated.

High/close front palatal unrounded / ([i]). some of which have certainly a phonemic status when used in concrete circumstances. [a] Intonations Besides the word-stress Semitic languages have various sentence stresses or pitches. The location of various vowels with regard to the front-back and high-low dimensions is indicated in Fig. but they acquired the phonemic status in several Semitic languages. Common Semitic or Proto-Semitic also possesses the three corre­ sponding long vowels: ā ([a:]). They are intermediate in height between the high vowels [i] and [u]. High/close back velar rounded u ([u]). Besides. ī ([i:]). The vowels e/ē and o/o do not belong to the common Semitic phonemes. 24.BASIC ASSUMPTIONS 107 Consonants Plosive Labial Dental Interdental (Pre)palatal Velar Pharyngal Laryngal pb t dt kg q ' Fricative s zs tdt (z) š hġ h' h Lateral ś ś (d) Liquid Ir Nasal m n Semivowel w y Vowels Low/open back velar a ([a]). Front high mean-mid lower-high lower-mid higher-low low [£] Central Back [u] m [e] [9] [o] [e] [9] m [a] Fig. . and the low vowel [a]. Location of vowels. vocalic functions of / and r are identifiable in some forms of speech and numerous vocalic variations are attested in Semitic since its most ancient historically attested phases. ū ([u:]). 24.

ś > d). Egyptian hk3.g. reconstructed changes cannot be treated in the same manner as historically attested develop­ ments which are revealed by orthography and its deviations.Semitic šlm. i. Only a careful consideration of historical changes enables us to draw any conclusions regarding pre­ historical developments.g. Prehistorical. The former are usually called "unconditioned" and they are examined. e. either short (e.g. of the velar fricatives in Assyro-Babylonian) and the rise of a new phoneme by bor­ rowing (e. "to be heavy".g. is compared with Semitic. The innumerable phonetic changes found in the history of Semitic languages represent three major types of phonemic development: 1° the phonemic shift consisting in the change of a phoneme of one sound-type into a phoneme of another sound-type (e. "to be healthy".g. since more radical divergencies are then revealed. anaptyxis. A l l phonemic changes may occur either in all positions or only in specific ones. The monophonemization consisting in the change from a cluster of two phonemes into a single phoneme is a phonemic merger. termed "conditioned".12.g. for instance. § 27.3-10). in the paragraphs dealing with the single phonemes (§11-22). of Turkish words in Neo-Aramaic) are two additional types of sound changes in a language. The latter changes. We may refer. dt > dd)\ 3° the phonemic split consisting in a bifurcation of two phonemes out of the allophones of one initial phoneme. m > mb/p) or dou­ ble/long (e.e. 2° the phonemic merger or total assimilation (cf. the coalescence of two phonemes resulting in the exclusive occurrence of either one of the two contrasting sound-units (e. The phonemic loss (e. possibly intermediate type (e. "to be right". and by evidence from contact between languages. while the diphonemization consisting in an opposite development is a phonemic split. metathesis. elision.. by comparative evidence. by the mod­ ern pronunciation of native speakers. nt > tt) or in the emergence of a new.e. i.g. dd > nd. "rule" = Semitic hqq. These changes are examined in the apposite para­ graphs dealing with gemination and various conditioned sound changes . prosthesis. dissimilation. independently from the conventional nature of Egyptological transcriptions. 10.13. corre­ spond synchronically to conditioned allophonic variations and manifest themselves through assimilation.108 PHONOLOGY 10. Egyptian wdn = Semitic wzn / 'zn. as a rule. etc. The equations may become quite interesting when Egyptian. Proto-Semitic phonemes underwent a great variety of phonetic changes in the course of time. ss > rs). to Egyptian snb . Prehistorical changes can be recon­ structed by comparison also with Afro-Asiatic languages other than Semitic.g.

Endegefi dāppārā. 2. word or name has p. from Persian parde. etc.g. Persian pirind. Common Semitic or Proto-Semitic has two labial plosives. Pliny's Latin carfiathum. w w w w w 11. a series of labialized consonants. as this is not made consciously. and Gyeto dàpàrā. "curtain". Ethiopic pos­ sesses.g. based on South Arabian hrf. and the Sabaic transcription bit of Greek 7taÀÂá5(ss) — designating the widespread Athenian tetradrachms with the head of Pallas Athena — indicates that the letter V could not be used to express the Greek n. from English "pipe". since each man's pronunciation is governed by the general conventions followed in his milieu. which nevertheless require qualifi­ cation. In other dialects. besides.D. and in Ethiopic.1. where a Persian. Chaha. Also Ethiopic possesses a p in addi­ tion to the/.2. it has further a labial nasal m and a labial semivowel w. "sword". testifies to this shift in South Semitic already in the 1st century A. k . parda. q (cf. h . The phoneme p occurs in Eastern Arabic dialects. e. Greek. however. e. in South-Arabian. may usefully be posited at this point. 27). all the speakers in a given speech community together. LABIALS 11. 4° sound change affects only certain sounds in a given language at a spe­ cific period of its history. Eza dàbbàrā.109 (§23. 3° the speakers of a given commu­ nity are unaware of sound change. e. "pipe". and in the lit­ erary language.g. Original /p/ is realized as the voiceless labiodental frica­ tive /f/ in Arabic. is pro­ nounced in Arabic firind or birind. Some general principles. §18.. These are four: 1° phonetic change is usually regular in that it affects all the occurrences of a phoneme in certain clearly definable positions in the utterance. Its presence in some Semitic roots could probably be explained as resulting from an original geminated bb as is the case in West Gurage where the sound p is an allophone of bb. 2° phonetic shift affects. the Arabs pronounce it as [f] or [b]. g . but its use is restricted to loanwords. It is relatively rare in Ge'ez and its symbol does not occur in the Aksum inscriptions.7). "he added". . as a rule. "autumnal". voiceless p and voiced b. "autumn". it probably origi­ nated through the spirantization of p into [q>]. in particular b . pēp. Ennemor.

*lbn > lūn. "liver"). "person. the sound p alternates with p. like p. In East Gurage (Selti). gabrā' > gora.g.5. 11. "cult". cer­ tain roots did not receive the same formalized expression in all the lan­ guages (§10.. 'bn > Gurage ūn. In NeoAssyrian b/p is again one phoneme. husband". with the result. and in Modern Ethiopian languages (e. "to be weak". Ethiopic possesses also a voiceless labial plosive p which is emphatic ([p']) and. When the allophones b and p reached a phonemic status.4.r|TOs.g.g. "to leap". in posta. "complaint". Syriac qwaz < *qbz. Caution is required in these matters because of the frequent lack of dif­ ferentiation between voiced an unvoiced sounds in cuneiform writing.g. become w.g. p'l. illustrate the transition b> b>w. e. e.110 PHONOLOGY 11. "table").o5co. However. "white"). many Ethiopians substitute b for p. but it is already attested by the Masoretic vocalization of ( . A non-geminated b in non-initial position can be spirantized into b (§11.g. Aramaic I'll-barak/. Pharyngalized labials are unknown in Classical Arabic and they play but a mar­ ginal role in modern Arabic dialects. in the minimal pair bāba. and both p and p are vari­ ants of b. 11. that the current Assyriological transcription pelludû or pilludû faces the Greek transcription (3tÀ. Nowadays. Nam-pi-gi for present-day Manbiġ. "father".g.10). and be reduced to the round vowel d/ū. which appears already in the Numidic transcription 'wdštr of Punic bd(')štrt. Interchanges between b and p are frequent in Semitic languages and some of them go probably back to the time when b/p was one phoneme. The further change is well represented in Neo-Aramaic (e. but b'l in Ugaritic and Amorite (i-ba-al-). parāqlitos = 7tapáKÀ. "to make". in Aramaic "spring site". Amharic saw as against Ge'ez sab'.8). in West Semitic. "man". tarappeza = xpáns^a. "its door". in Modern South Arabian (e.g. e. thus e. certainly corresponds to the usual qbr' with an additional change q > k. Nabataean kpr\ "the tomb". "post office". The shape of its symbol is imitated from s and it is usually employed to transcribe Greek loan­ words (e. Ar-pa-a-a and Ar-ba-a-a.3. and bâba. where one should not assign the unusual value bá to the sign PA. with variant spellings. "stone". "Arab". E. nabsā' > noša.À. Eastern Neo-Aramaic qbāltā > qwaltā. Different stages of this change are historically attested.g. "man.g. Ge'ez dabsa and dawasa. "Servant of Astarte". soul"). e. this glottalized labial may well be of Cushitic origin. Il-pa-rak-ka. pronounced and even written sometimes as bosta. "God has blessed". kbd > Harari kūd. of rare occurrence.

There are occurrences in which b alternates with m. "to escape". "four"). " I carried". "daughter-in-law". Tigre dabanā and damanā. "day". maġaha for bagaha. Sometimes the b disappears without leaving a trace in the labial vowel. . This phe­ nomenon is well attested in Palaeosyrian that exhibits spellings like Ha-lam for Halab.54). which shares many characteristics with South Arabian languages. "what is your name?". Syriac šarwaynā. correspond to Aramaic šurbīnā. "star". "frontlet between the eyes". darn and dāw. e. ymmt alternates in Ugaritic with ybmt. "hemp". "brother-in-law". The inter­ change of b and m is attested also in Andalusian Arabic.6. as well as to the verb br'.g. nšpt < mšpt.g.g. instead of represent­ ing unheard values of cuneiform signs.in Assyro-Babylon­ ian words containing a labial. where m and b can occasionally alternate (e. napharu. and inter­ vocalic m > b passing through b can become w (e. "cloud").g. Since there are archaic features in South Ethiopic.g. ki A A 11. and Arabic sarw. to Old Assyrian mer'um. "he rejoiced". Chaha and Eza ndm > Ennemor ndm > Gyeto ndb. Soddo dabdna and dàmmàna. "charm"). 11. since m reappears much later in the name of Halmān. to Middle and Neo-Assyrian mar'u. ġariiba for gamma. and Assyro-Babylonian šurmīnu. The nominal prefix m. "to shape. "cypress".g.7. Gurage amànāgà and awànàgà. The change was certainly not carried through consistently in any dialect. e. "total. habaltu for hamaltu. Sporadic examples occur in modern Maghrebine colloquials. bā smuk for mā smuk. u -bu for u -mu. explainable as Halam + Aramaizing -ān (§29. qinnam for qinnab. Besides. 'Iauvia stands for Old Hebrew Yabneh. Amorite yamamu{m) is the same word as yabamuim).g. "for whom?" Palaeosyrian sar-mi-na. ribs. ba for mā. "booty". Such spellings must echo real allophones.g. Ibdn for Imdn. Therefore Ara­ maic bar. e. "master").changes often into n. m can change into b (e. sum". might be related to Babylonian māru. "Aleppo". with a further change b > w or m > w.LABIALS 111 Hebrew kokāb < kawkab < kabkab. with a similar development of p in totāpot < *taptapat. "son". "water". "Aleppo".g. and various cases of substitution of b for m and vice versa are attributed to the Tayyi' and Bakr dialects in North and Northeast Arabia: e. especially in Amharic (e. "judgement". arat for Ge'ez 'arba't. "from". it is not surprising that South Arabian bn corresponds to West Semitic min. e. The same shift is sporadically attested in Aramaic. to create". dual of /šarmīnu/. "son". The same phenomenon occurs also in Ethiopian languages.

awád for Ge'ez Hamad. nqdš' < mqdš'.26). "the sanctuary".8. broken plural (§31. 'dn < 'dm. hkyn < hkym. name given to the god of Akko in the Babylonian Talmud ('Abddā zārā l i b ) . nwty stands for "Nabataean" in Talmudic Aramaic. when m is in intervocalic or postvocalic position. Also the third radical m may change into n. con­ firmed by Aramaic transcriptions (ss = sās < sa'as < šamaš). In Jewish Palestinian Aramaic and in Mishnaic Hebrew the change m > n in final position is very common.g.g. e. where one also finds a probable hypercorrection of w becoming m (Muher and Gogot tamuyà for *tawdyá. "God is (my) support". "orphan"). "skilled man". In addi­ tion. ndbkh < madbdhā.112 PHONOLOGY "bowl-like leg". The signs with m stand therefore for the phoneme w. as in Samaš pronounced Sawaš.g. and by Babylonian pasāmu / pasānu. "ashes"). related to Assyro-Babylonian kalmatu{m). may imply a previous spirantization of m.g. as shown by Palaeosyrian kà-ma-tum / kà-na-tù-um..g. e. "tongue. 11. "wise person". "to accuse". "south".9. and Libyco-Berber wsn. The same phenomenon is attested in Classical Hebrew with śtm / śtn. A-muk-a-nu or A-muk-ka-na. but the b is preserved in 'rby /'Arbay/. passing through the spirantized m. Amharic qâmbár from Ge'ez . transcribed 'wkn in Ara­ maic. "parasite". The occasional change m > n in medial position is found in the Aramaic name ll-su-un-ki = 'Ismk. language". which is written with m in cuneiform writing but transcribed šwš in Aramaic (Laos in Greek). Palaeosyrian zumūbaru. "louse". "to hide". "to be skilled". Gogot tambuyà next to tamuyà. which can­ not be interpreted as dissimilation of gemination. This change occurs frequently in LibycoBerber (§ 29. "is good"). "orphan". but the opposite change n > m is attested in Išm < Išn. In identical'conditions. The examples of the m / n alternation increase if the broader Afro-Asiatic area is taken into account with. "to pull away".28) of *zambāru < zamāru. "to be fit. dēq < de'iq < damiq. skilled". There are some examples of inserted b or p after m. the Middle Assyrian and Neo-Assyrian shift of intervocalic m to glottal stop and long vowel (e. attested already in Antiquity by the agent noun mwsn and per­ sisting nowadays in Tuareg a-mūssen. "Arab". drwn < drwm. "the altar". "man". 11. The labial m can become w. or baqāmu / baqānu. e. East Semitic wasāmu. The same change m > w may explain the shift from Babylonian I-lu-Me-er to Ara­ maic 7wr and it certainly occurs in Gurage dialects (e. This change is well attested in Neo-Babylonian. at Hatra. as in A-mu-ka-nu. "song".

Apppt transcribing Hebrew 'mry. both related to Harari and Arabic hubāb. The original nasal may disappear in front of the inserted plosive. and in France. Boni šimir. It is quite probable therefore that Eblaite si-pis and Ugaritic špš. These positional variants are attested in the Middle Ages in all Jewish Arabic-speaking communities.10. Mandaic 'mbr' from 'mr'. b g d kp t have survived only with one pronunciation except for p. but the plosive and the spirantized realizations have both attained phone­ mic status and are no more conditioned by their position. except in the loanwords of some dialects. with an inserted p after m. from popxos. also a secondary m may be inserted before b. Rendille cimbir. as in East Gurage dmbab. The spirantization of labials. originated from *śampšu or *śimpšu. Zap\|/ai marking the proper name Šamšay. p is never spirantized. "mortal". there are hesitations in the pronunciation of p / / . TT. including Spain. With the exception of a few communities. Arabic does not know spirantization of labials or velars. as in French chambre from Latin camera or in Greek au^poxos. Therefore we cannot be sure that Greek X. where the nongeminated consonants b g d k p t are spirantized in post-vocalic position. "sheep". both Eastern and Western. as in the secondary Greek form ppoxos. Although Aramaic p and k are never rendered in Demotic by / and h. "Omri". A similar phenomenon is attested in Cushitic (e. IappXaxos used for Yamlik.LABIALS 113 qamar. Among the Samaritans. "snake". and both attested also with the change b > w (Argobba hzwaw. t in the Bible: the Septuagint tran­ scribes these letters either by K. but these two series of transcriptions do not imply regular positional variants as in the tradi­ tional Jewish reading of Biblical Hebrew and Aramaic. "mor­ tal". "yoke". However. In Eastern Syriac. . only some of the six consonants are realized nowadays in Hebrew as plosives and as spirantized or labioden­ tal fricatives. "the Red (Castle)". but the labials 11.áiiJ35a reproduces a Semitic pronunciation of lamed. 9. which is pronounced b when geminated and mainly as/elsewhere. Spanish Alhambra from Andalusian Arabic 'alHamrā'.g. "sun". "immortal". p. traces of the double pronunciation of b g d k p t can be detected in Neo-Aramaic. however. and in the name of Humbaba. 0. or by %. and velars in the various Semitic languages is well known. Somali šimbir-ta. dentals. Old Babylonian Huwawa). the spirantization is probably indicated by the Greek transcriptions of k. "snake". "snake". "bird") and in Indo-European languages. T. But no evident case of such an insertion seems to occur before p. like in Ect|ii|/ai. In Tūroyo. Instead.

it may also result from a secondary diphthongization of a long vowel.g. the divine name Nusku = Nušhu). m. Ethiopic labialized consonants fol­ lowed by à may alternate with consonants plus -o (<-o). Amharic G àġġam or Goġġam. Similarly. Gurage wáz < b àz. u (e. "flexible [shoes]"). as in 'à-ba-al /wabāl/. g. The labial semivowel w has regular correspondences in all the Semitic languages.3-4. but there is either a compensatory lengthening of the vowel or a raising of a semi­ vowel. but there are some sporadic traces of spirantization of non-geminated velars k. spirantization of labials and of dentals cannot be detected.g.24).8). Assyro-Baby­ lonian pa-nu-ú-a = panūwa . "transporting". and from rounded phonemes b . h . "being read"). w w w w w w w 11. 27. indicated by signs with h. they are not necessarily connected with post-vocalic position (e. wa. neither b nor m may occur in intervocalic position.panū'a. "slave". The initial wa was sometimes expressed also by u+a or á+a. as in Old Assyrian Tan-bar-ta and Tan-mar-ta. The Semites were forced therefore to find ways of expressing w in their writing and they regularly used the sign PI in the function of wi. §18. except in the case of m (§11. 11. k . g .1. and by the sign É having the value 'à. the name of one Ethiopian province. cuneiform writing does not have any special signs to express it.23-24). cf. These replacements show that the scribes were aware of the phonetic .g.g. colloquial Arabic wakkil < 'akkil. facts that seem to imply a previous spirantization of the labials.g. and also as an on-glide in initial position before o.114 PHONOLOGY b. as in Ú-ar-ti-a /Wardiya/. In Śheri (Modern South Arabian).11. Ethiopic wof and of < 'op. Since the phoneme w did not exist in Sumerian or Pre-Sumerian.g. cf. In Assyro-Babylonian. Phonetic w occurs as a speech-sound throughout the life of East Semitic. The labial w can serve as a glide between vowels. which are allographs of Tan-wa(Pl)-ar-ta. The spirant consonants are not phonemic in Ethiopian lan­ guages and they can appear as free variants. sá-pu-wa-an. wu (e. but also ā (aw > ā) (§ 22. It may come from non-geminated b or m by spiranti­ zation. Replacement signs with b and m were also used. "bird"). and the velars k.7). e. q can be spirantized in Ethiopic when they are not geminated. but its graphic notation in cuneiform syllabic scripts is imper­ fect and doubts have been cast on its phonemic status from the Old Babylonian period on. Arabic maqrū'a > maqrūwa.g. q in Ethiopian lan­ guages (e. generally 6 or ū.12. especially after u (e. §19. In various diphthongs w may be reduced to a vowel. "he fed". and even before a (e.

"inheritance". A development w > y in initial position characterizes Amorite.g. in a few loanwords and in foreign proper names.C. but the hamza is there a purely orthographic feature.> iin initial position. its place in the system was taken by signs with m and b.> u. Besides.24). "lion".g. e. wa.fawha and fayha. The use of the cuneiform sign PI to mark yi. yú besides wi. wu (§11. In Middle Assyrian and in NeoAssyrian. in the name of the letter wāw.LABIALS 115 correlation of the labials b. meaning "hook". In Sabaic. These forms are written 'u. It is encountered in medial position after a consonant in the Hebrew and Aramaic word 'aryē. 'irtun. The sign PI having become restricted in later periods to the values pi. "goodwill". stands for wāzi'ahum. w and that the apparent changes w > m and w > b in East Semitic are to be considered as graphic replacements or allographs and not as real phonetic developments.and 7. §19. A South Ara­ bian Sabaic inscription shows a dialectal tendency to replace w by y at the beginning of words and the same phenomenon is attested once in an Arabic Hidjazi poem where yāzi'ahum. from warita. especially in Hudhail. Aramaic. probably following the occasional spirantization of these phonemes in the spoken language. fluctuation between the semivowels w and y is sometimes seen also in medial and final positions. The loss of w at the beginning of words can generally be assumed from the Old Babylonian period on. "to inherit" (cf. "gravity". yld. Ugaritic. 11. *ġāwiz > ġā'iz. "theif commander". Ugaritic. and Canaanite (Amarna correspondence) witness to this development which had a repercussion on scribal habits. *'iġrāy > 'iġrā'. "lawful". m.g.14. in kyn against normal kwn. "fragrant ema­ nation". "to be'\ in rdw and rdy. pe. s 11. instead of common Semitic *'arwiy-. as in Ni-nu-a. . "to bear". There is also the regular Arabic prac­ tice of substituting ' for w/y after ā.g. hawba instead of usual hayba. e. "enforcement". E. e. Initial w is preserved only in the conjunction wa-. The same fluctuation occurs also in Andalusian Arabic. some roots with first radical w have sporadically a variant with '. as well the "Canaanite" of the second and first millennia B. the other possibility consisting in not expressing w at all. also the vowel-sign u could be used to indicate wa.in the Arabic sources. as in M nu-u for Ni-nu-wa.g.13. a Hidjazi dialect. "and".and wi. "Nineveh". ya . e. is in opposition to wld in the other Semitic languages.13) in texts influenced by Amorite. In Arabic there is a possible development wu. except in archaiz­ ing script and in peripheral regions.

e. namely in West Gurage and in Argobba. Masyaf for Mediaeval Arabic Masyat. as the sign DI used with the values di/de and ti/te. In Modern South Arabian languages. 12. It was therefore of no phone­ mic significance whether the emphatic sound was pronounced with or without voice. who are unfamiliar with emphatic phonemes.os. and then by analogy lef. with Amharic tacc. Cuneiform spellings. classical nataqa is pronounced ndaq. and Hebrew. "inside". ġadat > ġadaf. "on". Common Semitic or Proto-Semitic has two dental plosives. Greek transcription <í>épe7i of the Syrian place name Tārib. Like the other emphatic consonants. Latin formus. Hebrew t is realized either as t or as d. as indicated by the traditional and colloquial pronunciation of Arabic. The Arabic colloquials of North Africa also show a tendency of voicing t. do not prove a voiced realization of t.116 PHONOLOGY 11. Ethiopic.g. This phenomenon would explain the Egyptian pronominal suffix -/ of the third person mas­ culine singular and the Argobba prepositions wdfc. / corresponded to a pair of non-emphatic ones. the post-glottalized t (t') has partially voiced and sometimes wholly voiced variants. The labiodental/may result in certain conditions from the inter­ dental t and from a lisping articulation of š I s so that the sound produced is [0] > [f] (cf. Greek Oepp.2. "grave. Arabic pharyngalized dentals. "under". DENTAL PLOSIVES 12.3.1. it has further an emphatic plosive t which was voiceless. The Jewish Sephardi communities of Italy pronounce Hebrew t as d and the voicing of final t into d also occurs in Ethiopia. The tendency of voicing a voiceless t or t is nevertheless attested in several Semitic forms of speech. as D-b-h for the toponym Tú-bi-hi. 3. This phenomenon is well-known to Arab grammarians and enters in their category of 'ibdāl luġawī or "lexical substitution". when com­ pared with Harari ustu. either plosive or . since the distinction between voiced and emphatic consonants of the same group is insufficient in both systems.15. one of which was voiced and the other voiceless.g. e. Lihyānite Rubaf for Rubat. which possibly goes back to a *tes alternating e. also in older Arabic. "hot"). and ancient Egyptian transcriptions of Semitic names. tef.g. "he spoke". In the Jewish Yemenite commu­ nity. voiceless t and voiced d. 12. at Cherchel (Alge­ ria). When pronounced by Bantus and Uzbeks. tomb".

the front orifice is contracted. [9] and [8].g. "daughter").6). INTERDENTALS 13. ti > c.transcribed 0pÚ7to~is in Greek). etc. i. w w 12. also known as Austronesian. i.e. . i. The precise phonological status of dental plosives in prehistoric Afro-Asiatic raises some questions because of the traceable alternations 11 k and dir. thus t > t . It is also attested for Arabic t in a few Tunesian dialects. at least in the consonantal cluster tb > kb (e. the lips are rounded. "daughter") and especially in South Ethiopic: di > ġ (§ 15. [ts]. A change t > k took place in the Indo-European Lycian of South Anatolia. s > s .g. "woman". and it occurs in languages of the Niger-Congo family (e.1. but an alternation 11 k appears in other circumstances as well (e. The first alternation occurs in Semitic pronominal ele­ ments (§36. often change into the corresponding labialized consonants. kbatra < *tbatra / rwatra. Instead of being original.g. The following table gives but a very partial idea of the development of the interdentals in the main Semitic languages and has to be explained below. berca < berta. Palatalization of den­ tals occurs in Western Neo-Aramaic (e. 4. Common Semitic or Proto-Semitic has two interdental fricatives.7). the upper pharynx. it has further an emphatic frica­ tive t. The alternation d I r has left traces in Cushitic and in South Ethiopic (§17. voiceless / and voiced d.6.g. Fulani debbo.5. t> c. Instead of the back orifice. An example of a phoneme realized as [t] or [k] is encountered nowadays in the Samoan language which is believed to represent the oldest form of Malayo-Polynesian. this opposition may result from a specialized function obtained by the allophones t and k of the same phoneme. which is often transliterated "z".4.INTERDENTALS 117 fricative.e. the PN Krupssi.e.11) and it led in Cushitic to an opposition of mascu­ line k vs. while a diacritical sign distinguishes it in Arabic from the emphatic dental fricative s. but the Maghrebine t is mainly characterized by its affricative articulation t. plur. 40.5). This consonant is represented by a graphic symbol of its own in Ugaritic and in Epigraphic South Ara­ bian. Further research is needed in both cases. widespread in the urban dialects of Morocco and of several Algerian cities. feminine t (§36. rewbe).

su^-a and Qatabanian s w.Ar. this interpretation is supported by the parallelism between su . "you will buy". "to divide". but traces of a distinctive sibilant seem to appear in the name of the Sun-deity *śpš. the regular occurrence of the su sign in the spelling of the independent and suffixed personal pro­ noun. the same city name as Ma-ša-du and Ma-sa-aa* . that are used for the three dental fricatives s. the voiced phoneme d is indicated by the same signs as its voiceless counterpart i . s (e. and tá-sa-am-ma (tś'm = ts 'm). as well as sa-am-si (śmšy = s ms y). si. As for the oppositions SÁ : SA and si : s i in Old Akkadian. šu expresses the syllables ta. used initially to express the prepalatal š.2. ši. zu. In fact. the overall picture corresponds to the Old Akkadian scribal practice. However. "he is well". while the signs SA. In the case of the Old Akkadian demonstrative. and in the expression mu-da-bil sí-kà-ri /mudabbil sikāril. as well as of the demonstrative of remoter deixis ("that". t dor d t Hebr.118 PHONOLOGY Pr. t d t Ass. root zht).S. Taš-má-Sí-piš II Taš-máUTU). instead. kl 1 .g. tu (e. su interchanges frequently with the SA. "story­ teller". "foundation".g.-Bab. "he made good"). S z s Aram. u-sa-ab luttabl. namely the voiceless t. š z s Ugar.g. No systematic distinction is made between ś and š. and between su -nu-ti and Qata­ banian oblique plural s mt. t d z E. thus obliterating the pho­ netic distinction between ŠA and SA. In Eblaite. zu'āzum. su stand for syllables containing the Semitic phonemes š or ś (e. Some irregularities occur in documents from Ebla where parallel texts quote e.2). In Old Akkadian. may signalize a phonetic distinction between the su sign. u-sa-lim /ušallim/.-Sem. where *dikr is spelt with zi {si) in agreement with Phoenician u 4 4 4 1 l 4 l A A l u 2 2 l 2 u d s*r(§14. zi. "he sits"). t d t Ge'ez s z s 13.A. in su-ru-uš (śrš = sVs ). and the su sign employed orig­ inally for the lateral ś.g. "people". "my sun". one could mention sá-lim (šlm . and ni-si (nś' = ns '). Palaeosyrian and Old Akkadian preserve traces of one interden­ tal at least.g. z. s i . "those"). A third set of signs SÁ. the group of syllabograms consisting of the signs ŠA.sHm). Proto-Semitic d and t (z) are expressed by the row of the signs ZA. written sí-piš (cf. si. ti. e. However. and both may be expressed also by the sign ŠÈ. between su -a-tum and Qatabanian oblique s wt. su group and seems to indicate that the phonemes origi­ nally differentiated by the two rows of signs were tending to coalesce in the period under consideration. t> t d> d t>t Cl.

ta ti tu da P. However. e. Sa-ap-si-A-du /Šapši-Haddu/. ši. T. The cuneiform spelling of Amorite personal names clearly pre­ serves the distinction between the interdental t and the sibilants š/ś.4. especially regarding the transcription of Amorite names (§13.. šu.2-4 and offers a com­ parison with the situation in Ugaritic and Epigraphic South Arabian. e. á à 13. ŠE. si.m or Šaap-si at Alalakh.g.g. e. and also in the case of Old Assyrian texts. SI. A chart of the principal signs involved in the discussion of early Semitic sibilants recapitulates the outline of §13. SU ZA/I/U 2 2 2 s 3 . However. These documents from Northern Mesopotamia still seem to reflect a distinction between / on the one side and š/ś on the other (e. ši. although the cuneiform signs of the different sets may occur in free interchange. while the signs SA. Sa-am-šu.3.Akk. SÁ.INTERDENTALS 119 13.-Sem. expressed in the cuneiform writing by the signs of the set ŠA. reservations have to be set forth in the case of the Mari and Qatara (Tell ar-Rimah) texts. Ia-šu-ub. su indicate that the development t > š must have begun in some Amorite dialects towards the end of the Old Babylonian period.g. sa-am-si < śmš). The picture that emerges from the Old Babylonian period on in Assyro-Babylonian is that /.4). t t t 'did did did š š š š Š Š s E. ŠA ŠI ŠU ZA ZI Amorite ŠA ŠI ŠU ZA ZI Ass. ú-sa-ás-ha-ar next to ú-ša-ás-hi-ir.g.Syr. but Ia-sa-rum /Yašarwn/. "he turned". The Proto-Semitic interdentals d and t (z) were coa­ lesced with the dental fricatives. etc. 13.S t (Si) ŠU ŠA t t d d d ~di du ša (SO ŠU SA/SÁ SI/ZI zu SÁ SI zu SA SI 4 s s s s s s l l 1 ši SU śa su/sù SA SI/ZI/SI U su su/zu SA SI SU ZA/I/U ši ŚU su ZA/I/U sa/i/u SA SI. š and ś have coalesced into one phoneme š. that were coalesced into š like later in Ugaritic.-Bab. were used to indicate the voiceless sibilant s.5. the rare interchanges of ŠA. ŠA O.Da-gan /Yatūb-Dagān/.. SU. ŠA ŠI ŠU ZA ZI ZU ŠA ŠI ŠU ŠA ŠI ŠU SA/I/U Ugar. šu with SA.

it can be replaced by a non-emphatic interdental in htm.C. In the 8th century B. The interdental d gener­ ally merges with d (e.5-6 points to the existence of / (tblt. dr'. The Greek tran­ scription Túpos of the Phoenician place-name Sr might indicate that original Tūr was still dialectally realized with t when it entered Greek under the form Túpos.g. ntr < *ntr. 4.24). '"Attar". but some words preserve the etymological d (e. the biblical shibbolet story in Judg. In Early Aramaic inscriptions the symbols " š " .g. / and t generally retain their independence. In addition. dbh. 'ahd. 12. 1. Ancient Egyptian D-r can simply transcribe Sr. "arrows".g. and Itpn.6. that appears about the 13th century B. a phenomenon which seems to prelude to the Aramaic shift / > t and to the merging of the two phonemes. "he seized"). which merged in Canaanite with dental and palato-alveolar fricatives at the time of the Proto-Canaanite inscriptions using the short alphabet of 22 letters. They should be distinguished from the assimilation tt > tt. "they seized him"). d > d (e.77. t respectively. "Tyre". "he guarded").120 PHONOLOGY 13. new spellings begin to appear.g. d. clearly differentiated from EtSuVv (Phoenician Sdri) and from the later attested name Eo(t))p.24). However.7. "gems". As for t.9). there are traces of an initiating process t > š appearing in the spelling 'ahrtp of the name 'ahršp and in the use of a new sign o to mark both š and t in three texts from Ugarit. 13. The phonetic process reflected in these changes . all of the few occurrences of etymological d are retained (e. but the dialect referred to may be Aramaic instead of being Hebrew. 4. the two words thrm. "s" stand also for /. 13. "stream. 'hdwhy < *'hdwhy. as in 'ttr > 'tr. flood") in Northern Transjordan. except in the Tell Fekherye inscription where t is transcribed "s". occasionally written hdm and tt. and in the theonym tt.710). since the " d " of Egyptologists usually cor­ responds to Semitic s. It is uncertain whether the Proto-Sinaitic inscriptions preserve the three interdentals.31. "kind". "may he bring back"). which are written from right to left in a shorter alphabet of 22 letters (KTU 1.C. "z".8. "arm"). í > t (e. besides the words in which an etymological t is velarized into ġ (§ 10. reflecting the shifts t > t (e. are spelt thrm and Itpn on one tablet (KTU 1.g. especially when the word contains a laryngal or r. However. although the alphabet borrowed from the Phoenicians makes such assessments difficult. In two texts ( K T U 1. yhtb < *yhtb. In Ugaritic.12.g. like later in Aramaic. Also Ammonite seems to have preserved /. "sacrifice").

D. e.g. tribe". A voicing of / is attested in the Hassānīya dialect of Mauritania. A limited evolution may be assumed in ancient South Arabian . by the early South-Palestinian Arabic spellings šahhāt for šahhād.44) uses 8 to transliterate Arabic d. like in md'w for mt'w. In fact. t. but the phoneme in question merged with s soon after the Aksum inscriptions (§ 16. which is attested in various Modern Arabic forms of speech. t. /d/ as [z]. "our hearts". t (conventionally transcribed "z") appears to have become a voiced interdental [d]. and dm'rn' for damā'irunā. as phd and phz for fahd. An intermediate develop­ ment phase seems to be attested by Sabaic cursive texts which indicate the etymological t by the sign for ś (d). "covering". e. " t " . North Arabian t is transcribed by t. d.9.g. "and he made a new sheepfold". as well as z it).4). 13. often realize /t/ as [s]. In most dialects of the sedentary population. or in the Algerian cities that have preserved the Andalusian dialect of Arabic (e. and x for t.g. for instance in the Hawrān. A particular feature of North Arabian inscriptions from the Tabūk (Saudi Arabia) and Ma'ān (Jordan) area consists in indicating the etymological / by the sign for ś (d).g. duhr. the frequency with which s and t appear as variant spellings in the same word (e. sm' îoītm'. 13. w-hdś śyt = w-hdt tyt.g.10. in Syria (e. In the North Arabian sphere. Pre-Classical and Classical Arabic maintain the three interdentals as independent phonemes and Classical Arabic uses the Aramaic symbols " d " . sll for til. "thirst") suggests that the phonemic distinctiveness of these two letters was. Besides. In cuneiform script. dofor. The opposite process of free spirantization of d > d and t > t cannot be assessed for Aramaic and Hebrew before the Hellenistic period. however. In fact. This scribal practice is somehow related to the phonetic shifts d > d or d and d > t which are attested around the 9th century A. The usual Greek transcriptions are 8 for d. and /d/ as [z]. readers of the Qur'ān who have no interdentals in their own language and try to pro­ nounce them. In Epigraphic South Arabian. d (f). in Ia-at-ri-hu lYatribl. " t " for d. to some extent at least. "they reached".INTERDENTALS 121 lasted probably for several centuries and the dental realization of the interdentals did certainly not happen at the same time in all the Aramaic dialects. The Dam­ ascus fragment from the 8th century A. this sign was borrowed by Ethiopic (dappa). "mid-day"). (§7. lost. Hesitations occur in Palmyrene Aramaic. "clan. x or 8 for t. but its acoustic value is not identical with that of d. "beggar".D. "nail"). interdental fricatives have shifted to the corresponding dental plosives.

It also possesses an emphatic dental fricative s. In some Modern South Arabian dialects there is loss of distinc­ tion between t and t. Although s and t remained distinct phonemes in Minaic. "to fence i n " . "he will reward". "Ptolemaios". Gurage atara. In Ethiopia. "Eleazos" ('EXeá^os). Argobba. vs. e. This would imply a local dialectal shift t > s in the 1st-3rd centuries A.11. and Gurage t. "to be thirsty".vs. Osarapis". Harari.12. t. it is / that is used for the rendering of a non-Semitic /s/. the sounds noted by t and s have fallen in Hadramitic together into a single phoneme noted indifferently by either letter.1. Amharic and Argobba tâmma. It is pharyngalized in Arabic. though here seems to be a preference for noting the sound as d. North Mehri dar.g. In Sabaic. Ge'ez mm'a. but post-glottalized ([s']) in Ethiopic and in Modern South Arabian. d. z. written qtwsf with s. Gurage tàma-. "upon"). In Semitic. Most South Ethiopian languages testify thus to a different development of / (z) than North Ethiopic. Harari hētàra. which is voiceless. South Mehri and Soqotri tar. Greek /s/ is represented by s .e. s. more appropriately noted by " t " or " š " than by "s " or "s". e. t. however. that became s< z. In Amharic. 'I'd. i.D. and 'trhf. and there is one case in Sabaic from Haram where Semitic t is spelt s . 13.g. against Sabaic 7'z. d. the hiss of s is very much stronger and more sibilant than the Greek s (§14. but different cases should be distin­ guished. Ge'ez hasara. Arabic zami'a. The Arabic z is in real­ ity the emphatic interdental d < t (§13.1). The same may have happened in the case of d and z. Tigre. Ge'ez.g. t became s.9). e. "Delos". 5. "Osiri-Apis. [s]: ys wbn. 3 3 3 3 3 3 13. D E N T A L F R I C A T I V E S 14. Since the Phoenician dialect of Lapethos in Cyprus uses š in the same period to transcribe Greek /s/ in ptlmyš. t and d (e. Arabic hazara. In addition. d and d. these transcriptions " t " and " Š " probably reflect a weakly enunciated Greek variety of the sound s which was likely to suggest a lisping effect. but [t] can alternate in Gurage with the glottal stop and become zero. Common Semitic or Proto-Semitic has two dental fricatives. voiceless s and voiced z. except in the place-name "Ctesiphon".g. .122 PHONOLOGY also for the interdentals t and d. tlmyt. in the divine name 'ttr or 'shrm. "Ptolemaios". Argobba hattàra. as in dlt. and Tigrinya lost the interdentals t.

14.DENTAL FRICATIVES 123 It is noteworthy that the Arabic plain /s/ is of higher pitch than most allo­ phones of English /s/. and in the Late Punic spelling s I st of the demonstrative pronoun z / *zt <d I dt. A similar palatalization of /s/ is attested later in Arabia and in Ethiopia (§14. Sibawayh makes it clear that. These variations most likely represent forms taken from different Phoenician dialects surviving later in Punic.C. in šh(y)mw (s hm).. as explained in §13. The cuneiform script does not distinguish the dental fricatives in an adequate way and a change in the writing practice occurred in the first centuries of the second millennium B. "to remem­ ber". "rewarding". "who owns the earth". e.D.3). the phonetic distinction of these consonants was blurred. Instead. In the 11th century B. which corre­ spond to South Arabian s and s . a Russian observer will be inclined to identify Arabic /s/ with his own /s/.g. "treasurer".D. where 'ks wt. z. both pronunciations zkr and skr are attested by the same name written zkrb 7 on three arrowheads and t-k-r-b-'-r in Egyptian transcription corre­ sponding to skrb'l. /s/ had a point of closure between the 1 2 3 1 x l 3 3 2 1 2 .2-4. comparable with the low pitch of Russian plain /s/. Therefore. "lucky". transcribed in Latin as syth.3.2. (e. use only the letters s (s) and s (f). Neo-Assyrian /s/ was palatalized into [š]. When Nabataean Aramaic script was adopted for writing North Arabian. and the Sabaic tendency to merge s with s in the 5th-6th centuries A. "clothing". The devoicing of z < d into s occurs in the Phoenician root skr < zkr < dkr. while the pharyngalized /s/ displays a noticeably lower pitch than the English /s/. as shown by Aramaic transcriptions. as in skr / škr (s kr). ms gd for the Aramaic loanword masgad.g. etymological Arabic š is transcribed "Š". Besides.. to understand an ancient transcription of Semitic /s/ by Hittite /s/ in ku-ni-ir-ša [korilrsa] rendering Semitic qdnl{'a)rsa. In Palmyrene inscriptions. by analogy. "arrow".C. "house of prayer") is already attested in an earlier period by a text from the Haram area. In the Arabic sphere. stands against standard Sabaic ks wy. no use was made of the letter samek ("s") expressing the [s] sound. as in š'd (s''<f). as a consequence. mšn /masennu/. the transcription of the etymological ś fluctuates between "s" and " Š " . In the 8th century A. 14. e. "garments". the North Arabian inscriptions. while a cuneiform sign za/sa was also available. in his time. for this sound was obviously palatalized into [š] and could be indicated by the letter shin ("š") serving to indicate both /š/ and /ś/. The three dental fricatives s.g. s are often interchanged in Western Late Aramaic manuscripts dating from a period when Aramaic dialects were no longer spoken in Palestine and. This might help. and the original s appears also as " š " .. even the earliest Lihyānic ones.

different from simple sibilants. Amharic. It introduces a clear distinction between sin and shin [s].g. and Gurage."he heard". a description which identifies it with [s] (§14.). and s are indicated by a (e. for luhūm.u%oup. "the Devil" (root śtn). š. which in Sibawayh's days had a point of closure "the same as for g and y. The Damascus fragment from the 8th century A.D. but there is no contact between the top of the tongue and the alve­ olar ridge.x P Y ° 0 for šabi'ū. for 'afsal. A. in which /š/ and /s/ had merged.e."it came up"). This transliteration shows at least that the modern standard shin was realized at that time as a com­ plex sound. "noble" (= Arabic šarīj).44) translit­ erates the etymological ś by % (e-g. probably similar to the fricative palatal [§]. "they were sated")."hs abhorred". the speak­ ers of which pronounce it with approximately the same tongue position as š. Harari. together with retroflexion of the tonguetip. "bread". as in qetan.g.4). "meat" plur. Some of them go back to earlier stages of Ethiopic and preserve the original phoneme. This sound derives from an original ś.4. OKp.124 PHONOLOGY tongue-tip and the hard palate. The merging of /s/ and / Š / occurred also in Ge'ez and in modern Ethiopian languages. (§7. 14. % P í for hubz. which serves for h and h as well (e. but it is not complete. The air is pushed out over the tongue and the lips are simul­ taneously rounded and pouted. Thus we may represent the development of the three sibilants indicated in Arabic as «š» in the fol­ lowing way: E l 0 l ) Proto-Semitic Sibawayh * • [?] Post-Sibawayh Arabic [5] [5] The fricative palatal [9] is still attested nowadays in Śheri. Although secondary palatalization of s into [š] is attested in Tigrinya. rjay[i8] for sa'id. between the centre of the tongue and the soft palate". or in girif. it is unlikely that all the attested cases of etymological š may have been occasioned by palatalization. must post-date Sibawayh's time. while etymological s. The modern standard pronunciation [s] of Arabic sin. erectly for samì'. i.rjEÀ. but .

or as [s]. when spn. s can change into š. "master". e. sġī' against classical šaġī'. Also in Modern Ethiopian languages. as in girś. Gurage. The palatalization of s into [s] led to the development of a new phoneme š for which an adaptation . "courageous". in zirāt. 14. e. In contact with q. e.g. Argobba.g. "lane". sirāt or sirāt. on the other hand. 14.g. "way" (< Latin strata). The principle at work here is the phonological tendency which makes the whole word either emphatic or non-emphatic. s can be realized as š.8. 14. in northern Jordan. and Harari under the influence of Cushitic.3). "to kick".7. In modern South Arabian languages. There are also unconditioned vari­ ants. z. In the colloquial of es-Salt. 14. as mentioned above (§ 14.9. "belly" (root krs). that shows an arbitrary interchange of the two symbols "s" and "Š". viz. This tendency is attested also in North Africa.DENTAL FRICATIVES 125 it can also be a palatalized variant of k. Both changes s > š and š > s are attested in modern Arabic dialects but they are generally conditioned: under the influence of a fol­ lowing ġ.5. by anxrip in Greek. and hsr. This [c] has phonemic status in Śheri. at least in Late Antiquity. but 5 can also change into s. and in the first case turns all or most consonants into emphatic ones (§10. e. in zaqf < saqf. although it seemingly corresponds to an old pronunciation of this dental fricative. e. In some Jewish Moroccan communi­ ties Hebrew /s/ and /š/ are both realized as a sound intermediary between [s] and [s]. in Mehri sdba'. "north". The realization of Hebrew s as a voiceless dental affricate [ts] by European Jews is probably to be considered as an innovation. In Ethiopian languages. "finger". as against 'dsba' in Śheri and in Soqotri. as in sqoq < zuqāq. It may reflect dissimi­ lation. the merging of /s/ and /š/ had already occurred in Ge'ez. the difference between s. as rafas and rafaš.9). as in the cluster st > st.g.g. "plant". The opposite shift s > s is attested as well in mediaeval and in modern Arabic dialects. z becomes s. in Amharic and in Gafat.6. s is often pronounced as an affricate [ts']. One should also mention the passage s > t in SouthAmharic. or simple develarization. s is suspended in the neighbourhood of emphatics. The substitution of z to s in the neighbourhood of an emphatic is attested in colloquials of central Syria. in soltān < sultān. was transcribed by tspn in Demotic. 14. "roof".

[j]. However.2). but is was replaced by the rough breathing in Greek (e.126 PHONOLOGY of the symbol for "s" is used (šat). e. in Armenian. and the prepalatal š was described in § 13. P R E P A L A T A L AND P A L A T A L 15.10. The sibilant s is almost universal. since the Neo-Assyrian cuneiform spelling remained unchanged. hik. i.1. "he has sent". and eastern Polynesian (Hawaii I Savaii).20).2-4. Instead. z. like Tigrinya Sam beside sdm. The changing practice of the cuneiform writing system with regard to the interdental t. the Middle Babylonian and Middle Assyrian orthographic change š > I before a dental is probably a scribal device that indicates a total reciprocal assimilation (§27. In addition.35) and the later NeoAssyrian spellings like issu for ištu. Therefore.11).4). Tigrinya and Amharic have an etymological š in some gen­ uine Semitic words and forms.10. In consequence.4). "since". i. Luwian maššana-.e. The change of an original sibilant (z. Arabic suq I siq. šoh'attà. 14.(§41. mahana. Per­ sian. West Semitic s was transcribed " š " in . e. [f]. for which no adequate cuneiform signs were available.g. there is a strong case for regarding š > s (> h) as the primitive phoneme of the Semitic personal pronoun of the third per­ son (§36. the Amharic causative preforma­ tive as. the dental fricative s. š) into h is attested also in the Tuareg dialects of Ahaggar and Air. "foundation". very likely [iśśu]. Common Semitic or Proto-Semitic has a voiceless fricative prepalatal or palato-alveolar š.vs. found mainly in borrowed words. "five". one may wonder whether the articulated š of Ethiopic results always from a secondary palatalization of s or preserves an old pronunciation. "make haste!". Modern South Arabian (§15.14).g. This would explain the spelling iu-se-bi-la {yuśśēbila}.g. "come here!". "seven". in an Amarna letter from Byblos (EA 88.g. 15. kušda > kulda [kuśśa]. vs. I f one takes this evidence into account.e. as shown by Aramaic and Hebrew transcriptions. and of the Semitic conditional particle (§61. Ghat dialect zik. "name". at least in a number of cases. Neo-Assyrian š was pronounced [s].beside as. 6. "seven") and by h in Lycian (e. nbwsmskn for Nabû-šum-iškun. and a palatal semivowel y. hammuštá. e. èitīá / septem.g. or for isdu. of the Afro-Asiatic prefix of the causative verbal stem (§41.5) producing a geminated dental lateral of the ś type (§ 16. "god").2. "urge o n ! " (root šūq).

Sa-me-ri-na = šmryn. "guardian") shows that the phonemic distinction between š and s was not lost. like šabe I sebe.: Root *šab' *šim *hamš *šma' *šqiy Śheri šo' šum hīš ìr šeqe Mehri hoba ham haymah hīma hdqu Soqotri yhobd' šem hamoš h emah hez3 y "seven" "name" "five" "he heard" "he watered 15.D. In Modern South Arabian and Ethiopian languages.g. despite some dialectal palatalizations of s into š (§ 14. "belly".5.PREPALATAL AND PALATAL 127 Neo-Assyrian texts.: "they put š . Gafat gàġġaš < gâġġaki. the use of the Greek symbol S in Latino-Punic inscriptions to indicate š (e.10. "seven". but also in Modern South Arabian. It is unlikely. e. "your (fem. as shown by a comparison of some nominal and verbal roots in Mehri and in Śheri.20). SVMAR = šmr.3. but mainly in foreign names from which one cannot infer that š and s were normally confused. noticed already among the Mahra by Masudī in the 10th century A. 15. Mil-ki-a-ša-pa = mlkysp. e.) house". In Neo-Punic there is an occasional interchange between $ and s in other cases as well. as stressed above (§ 14.3. Śheri šurś < krś. and finally developed into [s] in Arabic and in Ethiopic. "seventy".g. in the causative prefix of the causative verbal stem (§41. in NeoPunic that probably continues two different Punic dialects (§15.4.7).4). E. 15. however.g. while the situation in Soqotri is less clear and seems to reflect external influences. that palatalization may have brought about the change of s into š in the various cases where modern Ethiopian languages have an etymological š which is not con­ tiguous to a phoneme like / that may have occasioned the change s > Š. On the contrary. In other languages.2).9). e. an unusual variation š/s occurs in some words. "Samaria". in Assyrian and šb'm / sb'm.6) increased the number of words containing the prepalatal š. and in the conditional particle (§61.10).g. A similar phe­ nomenon occurs on a wider scale not only in Indo-European and other language families (§14.3-4. This is an old phonetic change. The prepalatal š merged with s in various Arabian and Ethiopian forms of speech. A change š > h is attested in the Semitic personal pronoun of the third person (§36.11).g. while West Semitic š was rendered by "s". the palatal­ ization of the velars (§18.

The Egyptian name ntr. In Libyco-Berber. The velar k is palatalized into c in various circumstances. to the phonetic repertory. Also t may be .g. Devoicing at the end of a word or of a clause may cause a fur­ ther change ġ > c.. It is either original. Palatalization plays an important role in Afro-Asiatic languages.g.6).g.g. e. Arabic yusr < 'usr. Similar phenomena occur in Arabic and Neo-Aramaic vernaculars (§18.g.6).(cf.8. As for the palatal ġ.g. e. or is derived from w (§11.6. Harari wūlâġi. in which it disappears in Assyro-Babylonian leaving behind the vowel which accompanied it (e. spirit". the voiced prepalatal z is found in loanwords and in gen­ uine Aramaic words in which š is voiced by assimilation. In Eastern Neo-Aramaic.g.g. "to greet").7). i-ig-muur. Śheri širet < qryt. " b i l l " . instead of a glottal stop. Gafat tzgàlġi < tdgàldi. the Old Egyptian pronom­ inal suffix -ki of the second person feminine singular became -c. Neo-Aramaic ydmma < 'emmā. yu. In cuneiform script it is not marked in initial position. "god". and also as an on-glide in initial position. for example. 'Omānī colloquial yāl < 'āl < 'ahl. but it should perhaps be linked with Agaw nkdra. "captivity". a-fġah < fallāh. he says. "soul. Amharic haya < *kil'ā. as well as the corresponding glottalized emphatic š or C. "you gird yourself" (fern.g. "mother".). or appears as a glide (§15. "sin"). and it is conventionally transcribed " t " by Egyptologists. The palatal y can serve as a glide between vowels. "heart") and // changes into ġ in Tarifit (e. Arabic hafi'a > hatīya.128 PHONOLOGY instead of k". yi. §22. especially after i (e. palatalization added â. z. e.13). Besides.> u-). it may be derived not only from a velar (§ 18. "town". is believed to be related to Semitic nkr. The Old Akkadian spellings of the type i-ik-mi. also with the change / > r. e. "twenty".> /-. thus. also in cases of usual spellings without initial i. uz < ul. 15. e. "strange(r)". probably indicate the preservation of initial y. "he wept". from wâlāda. The palatal semivowel y has regular correspondences in the var­ ious Semitic languages.7. like in Tarifit Mric < Melilla.g. Therefore. or results from a palatalization of /. but also from d. 15. / is palatalized into z in Tamazight (e. 15.13-14). in [xezbona] < hušbānā.5). e. especially when it is contiguous to the vowel i J u. East Gurage bāce < bakaya. "family". "peasant". "give birth!". "he captured". sdgdm < sallim. not only in Semitic (§15. "he conquered". any lexical study must take this widespread phenomenon into account.

the Semitic noun qāt-.3.'aśā. Otherwise. conventionally transcribed d. except in Old Akkadian where the sign su seems to have originally expressed the prepalatal š. "heaven") and śa (e.1. and -Xa. Greek $akaa\iov and its derivatives are borrowed from Semitic bśm. kis-. It might be a Phoeni­ cian loanword. "Kamosh . This phoneme cannot be distinguished from š in the North and East Semitic languages. ša (e. sa-te-e = śdh. "Soco").LATERALS 129 palatalized as.transcribing ś clearly shows the lateral character of this phoneme.2. voice­ less s and emphatic ś. is written in Phoenician 'sr. well-known in East Semitic. arm". ša-me-ma = šmm. ša-ah-ri = t'r. "field"). the merging of š and ś is apparently complete in all the languages attested in cuneiform script. while the sign su indicated the lateral ś (§ 13. The Neo-Assyrian transcription Ka-[ma-]as-hal-ta-a of the Moabite royal name *Kamoš. In the Amarna glosses. Nevertheless. "balsam-oil".g. and it seems therefore that ś has lost there its phonemic status.2). 2 4 16. "hand". for example. Common Semitic or Proto-Semitic had two dental laterals. namely s ( £ ) . with the tongue-tip in the /-position. The existence of both forms is best explained by an original 'śr.g. "gate").C. the sign ša can express ta (e. 16. Neither do the Amarna glosses and the Egyptian transcriptions indicate that an autonomous phoneme ś existed in the Canaanite languages of the I I millennium B. while the Egyptian transcriptions s and š are interchangeable (e. They correspond to [4] and [4]. but this is by no means certain: it can be directly bor­ rowed from another Semitic language. In the Phoenician alphabet ś and š are expressed by one symbol with the obvious consequence that the alphabetic script of no West Semitic language is capable of distinguishing the two sounds without using diacritical signs.g. ša-muma. "hand. LATERALS 16. 'śr. which corresponds to West Cushitic (Omotic) kite-. A systematic study of the Afro-Asiatic lexicon and of the palatalization rules in various languages is still a desideratum. and it is pronounced in Modern South Arabian by retracting the right corner of the mouth and forcing a stream of air between the teeth and the inside of the cheek at the right side of the mouth. 7.g. sk and šìk = śk'. the word "ten". but its "feminine" form is spelt 'šrt. Non emphatic ś has a graphic sign of its own in Epigraphic South Arabian.

perhaps close to [9] (§14. e. 5 5 5 2 16. the realization of ś is equal to that of s in all Jewish communities. In fact..4).5. the distinc­ tion made by the Masoretes is etymologically correct and it is confirmed by the incompatibility of contiguous ś and / in Hebrew roots. In Hebrew and in Arabic. The name "Chaldaean". sa-mì-iš. the differentiation of š and ś is expressed by diacritical signs. ti-mi-iš. Spoken Arabic leaves no doubt about the original character of this differentia­ tion. In Aramaic. "twelve". il-te-meš. śagīb. the Old Syriac inscriptions from the lst-3rd centuries A . "ten". "victorious". still preserve the spelling with " š " . However. 'šryn. etc. "five".g. D . kśdy in Aramaic and ks dy in Sabaic. is tran­ scribed sa-gi-bi or ta -gi-bi. since etymological š is written " š " in Jewish Aramaic texts. in standard Syriac. while š remains unchanged. Late Babylonian. Ethiopic.. indicates that ś was preserved in Moabite in the 7th century B. and in Egypto-Aramaic once also smš. and hammeš < hmš. "sun(-god)".130 PHONOLOGY has made". the graphic distinction introduced in Hebrew is absent from the Samaritan tradition and may be based on a comparison with Aramaic. Instead. 16. While š is regularly rendered by "s" in Neo-Assyrian and by " Š " in Babylonian. the phonemic distinction of š and ś is demonstrated by Neo-Assyrian. ta -meš.g. since Semitic languages generally avoid homorganic radicals in contiguous position. Neo-Babylonian. and Elamite tran­ scriptions.g. is written Kal-da/dà-a-a in Assyro-Babylonian. which in Sibawayh's time had a totally different articulation from the modern one. il-ta meš. and śmš. The Masoretes indicate the graphic dis­ tinction by placing a point either above the right side of the symbol (for s) or its left (for s). "moon(-god)". The phonemic distinc­ tion between ś and š results likewise from the later shift ś > s in Ara­ maic. which had the same development as Arabic in this case. but this sound 2 . in 'dsar < 'śr. il-ta-meš. and that it was a lateral phoneme. 16. śhr. Arab philologists distinguished sin (< s) from sin (< s) by placing three points above the right side of the letter (^à) serv­ ing to express šīn (< ś). borrowed the South Arabian letter s ( £ ) to indicate ś. te-er. il-te-{eh-)ri. as e. trt'šr'. appears as še-er. "twenty".6.C.4. where the spelling clearly shows the lateral character of ś. However. while etymological ś is rendered there by "s". is spelt šam-si. This also demonstrates the lateral character of original ś. etc. the various attempts to indicate the strange sound ś reveal its different phonemic status: e.

Andalusian Arabic 'al-qādì was still borrowed in Spanish as alcalde. 16. "to cut l . The dialectal treatment of ś in a single text from Ugarit (KTU 1. namely " d " . and in Hebrew. In Arabic ś is pronounced either as a voiced emphatic dental plosive [d] or as a voiced emphatic interdental [d]. i. The original lateral character of ś results not only from its pro­ nunciation in Modern South Arabian and by its articulation as / in Datura (§16. "mayor". at least in some dialects. as well as in East and North Semitic languages.2). con­ ventionally indicated by "z". though the sound itself has disappeared.g. The alternations between dād < ś and shīn < ś in Arabic indicate that the two phonemes constitute a pair: basaka and badaka. as e. "earth". It loses sometimes its emphasis and is then reduced to [d]. confirm the independent phonemic status of ś and its emphatic character.8.12). like z. but also from ancient transcriptions. as albayalde.e. "he was lost") and by its merging with 5 in Ethiopic. lā' for dā'. in 'rq > 'r'. without the glottalization which characterizes Modern South Arabian and Ethiopian emphatics. Early Aramaic practice of indicating ś by " q " and the later spelling e. "white". In Modern South Ara­ bian languages. The loss of the lateral glide in Arabic is therefore a quite recent phenomenon. expressed by the clear velarization of the sound symbolized by " " ' (§10. by Ru-ul-da-a-a-ú and the description of dād given by the Arab grammari­ ans leaves little doubt that d represented a lateral phoneme in early Islamic times. in the Hawrān dialect.7).7.g.. The name of the Arab god Rudā is transcribed in cuneiform script. 16. as arrabal{de). "suburb". also supports its original emphatic character and indi­ cates that ś had existed in North Semitic as an independent phoneme. "he finds". that was borrowed by Ethiopic to express the corresponding sound (0). "he laughs". The existence of ś at an early stage of Ge'ez is therefore attested in orthography. However. 'al-bayād. The emphatic lateral ś (d) has a graphic sign of its own in Epi­ graphic South and North Arabian. t'i.C. in the 7th century B. "white".g.9. ś and s are not con­ fused in the early inscriptions. "ceruse". However. 16. "go out!". soon after the early Aksum inscriptions. "you smite me".9). in todrobni. the original emphatic character of the sound is supported by its articulation as / in the Datīna dialect of South Yemen (e. abyal for 'abyad. where ś is expressed by t in ythq.LATERALS 131 merged with the one expressed by s (rS) to become [s]. and ymt'a. 'arrabda. in Phoenician. this phoneme is articulated like a voiced i (§ 16.

A. their distinctive phonemic status is nevertheless established in common Semitic as known in historical times. Cl.11. Aram. šafaza and dafaza. Hebr. The corresponding alternation between śzn and occurs in Hebrew śāhaq and sāhaq. *Pr. "to blind". especially before velars and palatals. šahaza and dahaza. while r was realized as a uvular non-rolled [R] in one of the traditional European pronunciations of Hebrew and sporadically in Gafat.of dangerous animals (§30.in West Cushitic (Omotic). The original phonemic distinction of these con­ sonants in Afro-Asiatic is in doubt.-Bab. "to explain". The following table displays the development of the ProtoSemitic laterals in the Semitic languages taken into account. mašaġa and madaġa. L I Q U I D S AND NASAL 17. exem­ plified by the noun "dog". The originally emphatic consonant ś corresponds to a single non-emphatic one.Ar.3-6).10).in Semitic. "jackal". 16. M. with the high probability that the latter reflects the loss of the emphasis. Common Semitic or Proto-Semitic has two dental liquids / and r. These alterna­ tions reveal an emphatic and a non-emphatic pronunciation of the same roots.1. haša'a and hada'a.S.10. waššaha and waddaha. This uvular articulation would explain the occasional non-gemination of r in Gafat . «āša and nāífa.A. considering the lack of a distinction / / r in ancient Egyptian and the frequent alternations (§17. and one dental nasal n.S. The dental basis of articulation of these phonemes is supported by their traditional and modern realiza­ tions. 'Maws and 'illawd. "to be submissive".in some Chadic languages. "to break". hašama and hadama. Ge'ez 5 Š Š Ś Ś > S ś s sit s q> ' f ś Š Ś(S ) 2 Ś Ś > S did ś z ś>s 8. 16. "to kick". "to knot". "to carry". n tends in some modern Arabic dialects toward a postpalatal n before most consonants.-Sem. Although 11 nl r still appear as allo­ phones of the same basic phoneme in Palaeosyrian (Ebla) and in Gurage. E. M. while the lateral glide of the phoneme was preserved ( | > ś). kan. with the gender determinant -b.132 PHONOLOGY off". and kal-b. Ugar. kar. However. Ass. It was therefore of no phonemic significance whether the emphatic sound was produced with or without voice. "to kindle the fire". haša'a and hada'a.Ar. "to laugh".

"army").LIQUIDS AND NASAL 133 and its systematic non-gemination in the Masoretic vocalization of the Hebrew Bible. "vine­ yard".10). "to go up".g. it may be palatalized into y (§15.3). The same phenomenon is l2 n .g. "cypress". "to go". Semitic lb'-. whb'tt < whb'ttr.g. several roots common to Semitic and Egyptian have l/r in Semitic but Hi in Egyptian. Egyptian kim. "he shall go". The variations in ancient and modern articulations of r have no phonemic value.g.. The weakness of the liquids is confirmed by ancient Egyptian transcriptions of Semitic / and r with an i in the Mid­ dle Kingdom period. ša-mi-nu Isar-mi-na (dual).g. in weg < wld. e. The non-gemination of r might also result from the articulatory shift r > ġ. "wife". "he went". Besides. e.2.g. hdġmi < harīm.g. "lioness". "single". "Ashkelon".g.6). rāh. Egyptian kib. 29. "heart". As for the liquid /. e. and hussaq. 'm' for 'mr. "to go". while Semitic languages generally avoid homorganic radicals in contiguous position. Zappa (Masoretic Śârā). Semitic karm-. Semitic libb-. The loss of final r is frequent in Sabaic personal names. Semitic qarb-. "boy". Xappáv (Masoretic Hārārì). The com­ mon use of the cuneiform sign NI to indicate Semitic ni. and slq.g. the weak­ ness of these phonemes may explain this apparent exception to the com­ mon trend. ('A)llāh. Egypt­ ian lb. while Gurage testifies to the occasional loss of medial / or r. e. e. "intestines". li. "to say". However. e. gund. "my god". However. "to eat". Semitic lahām. "God". the weakness of the liquids is reflected by the Aramaic verbs hlk. the Septuagint still shows gemination of Hebrew r. "gleanings".11). 'à-agú-um < hlk. Certain reservations have also been expressed concerning the dental nature of r and n because they are frequently contiguous to other dentals (e. and i. i-ś-k-í-n. The weakness of the liquids is amply exemplified at Ebla: e. as Touoppa (Masoretic 'Âtnorā). ġaġ»l < riġl. attested in some mediaeval Arabic dialects of Iraq and nowadays also in North Africa. A similar situation results from a comparison of Semitic and Cushitic roots.fard.and Rendille aham. e. i-i-h-b-w-m. and qema / qârma. even in the same word ì-lí. e. "he was brought up" (§43. 17. This phonetic phenomenon could be related to the appearance of the non-etymological cluster r' in the Ara­ maic 'r'm of the Palestinian Targum for 'ârīm. reflects the weakness and the interchangeability of l/n (§17. a city name attested as Irgt at Ugarit.g. Egyptian iby. final / and r are dropped occa­ sionally in Jewish Babylonian Aramaic. but the emphatic pronounciation of / and r in certain Arabian words deserves a mention. with forms like ydhāk. "he raised" (Gen. La-ru -ga-tù I A-ru -ga-tù. In the first millennium B. "Rehob". "foot".C.

e. Arabic layt. the aforementioned losses of liquids and of n should be distinguished from morphological phenomena like the surrender of nunation and mimation in Arabic and in AssyroBabylonian. while the phonetic change / > n is already announced by the possible intermediary sound in al-na-šuh.. "clarion".30) is to be considered as the result of an / > n shift. but its substitution by n or r is accomplished under well-defined conditions. laba > naba. the prefix n. As a matter of fact. The latter phenomenon can be dated in Eastern Aramaic to the 2nd-3rd century A.g. / becomes r. Spo­ radic occurrences of the interchange between initial / and n occur in Assyro-Babylonian (lamsatu I namsatu. "might". "waist".23) and in the jussive of some South Ethiopian languages (§40. and it is realized in nhwy'. e.3) — confirms the weakness of the liquids. However. "merciful".3. da'ānu < danānu. cf. in a dialect of central Syria (Nuhašše I N-g-ś > Luhuti I L's).instead of /. The change l> n occurs in initial position and in medial position when / was originally geminated. They are attested also in the Datīna dialect of South Yemen (e. "lion". In the field of grammatical morphemes. The disappear­ ance of the liquid is compensated by the lengthening of the preceding vowel.g. The shift / > n in initial position occurs sporadically in other Semitic languages as well. rēmē'ū < rēmēnū. e. in the 4th century B. otherwise. "disgrace").. in Nabataean with the proper name 'bd(')lg' transcribed once ApSaync. Thus.. which should be distinguished from the morphological change consisting in the use of the jussive prefix /. AeuKwaia > Nicosia).g.. 17. "may he be". qīqiltā < qilqiltā.instead of the imperfective y-. "rubbish dump".g. Aramaic laytā. the original liq­ uid / almost disappeared in West Gurage. and Greek Xiq are paralleled by Assyro-Babylonian nēšu. and in Aramaic. In some Gurage dialects the change / > n occurs even in initial position. in the 7th century B.. with hâsoprā < *hasarsarat. " f l y " ) . The interchange between / and n may be observed in various lan­ guages. Syriac qulqālā. Hebrew layiš. One should also mention the change of intervocalic n into ' in Middle Assyrian and Neo-Assyrian.in the Eastern Aramaic prefix-conjugation (§40.134 PHONOLOGY attested in Hebrew with qīqālon < *qalqalān.C.> n. unless a different explanation is offered for this form.4. lahna mā laqbil for nahna mā .D. "disgrace" (cf. "Nusku". The frequent assimilation of / and n to the following consonant — and even to the preceding one (§27. and in Cypro-Phoenician ( A á p v a ^ > Nrnk. 17.C.

etc. l-hāson for 'al-hāsilu. corresponds to Hebrew and Aramaic šlh. " I " . e.g.'à-aq-Da-mu. occur frequently in Libyco-Berber. that certain languages like ancient Egyptian and Mycenaean Greek (Linear B) do not distinguish graphi­ cally. as mentioned above (§17. in West Gurage r becomes n in initial position and in non-initial position when originally geminated.and 7. "for"). "stylus". The variation of n and / occurs in medial and final positions as well. in Nabataean Aramaic. "to send". ammàrd for Amharic ammana. attested in Demotic as Is and in Coptic as AAC. qartuppi < qantuppi. Lihyānite Himrāg is a phonetic variant of Himlāg. hdr. plural ngwaren. godārā. in Old Assyrian. In West Gurage dialects. "to believe". "statue". "to dispatch". fenġāl for finġān. e. mil for 'nk.> '1-.g. Hebrew gādal. . in Moroccan Arabic. "to become big".. the change / > r occurs in non-initial position when / was originally not geminated. "tongue". e.g. The variation l/n is a surviving feature of Afro-Asiatic. snm for slm. l2 17. in North Arabian.'à-aq-Da-mu / //. "we don't accept"). 'à-da-ru -um / 'à-da-lu-um.g. "camel". in EgyptoPhoenician. nd for la. the divine name ls-ha-ra I' Iš-ha-la. Ik. as exemplified by a comparison of Semitic Išn. e. Arabic sāraha. Interchanges between n and r are also attested. "sheep".e.g.g. "king". Harari gādāra or godâra. badenġāl for badinġān. "Damu caught up". "son". "gift". since the shift h > ' is widely attested.e. "cup".LIQUIDS AND NASAL 135 naqbil. "belly". e. great. Interchanges between / and r. and -mnkw for -mlkw. gamēra from gamdl. the non-geminated n becomes r in non-initial position. tall". "but". "seal!". ġlem or qlam for ġanam. ban for bal.g. "to be" or "to become big. "briefly". e. The same phenomenon is sporadically attested in NeoAssyrian.3). Logone ngun. in Modern Ara­ bic dialects. Tarifit r corre­ sponds regularly to Tachelhit / (e. i.in North Arabian also suggests a change hn.5. "aubergine". ìr. and Sabaic rzm. is etymologically related to Arabic Izm. In Assyro-Baby­ lonian the "stork" is called laqlaqqu or raqraqqu. or gudārā. and such interchanges are also fairly common in Semitic. Similar changes occur in Chadic languages. e. "the interior".g. "horn". In West Gurage dialects. corresponds to Gafat gāddàrā. bn. 17. "land-tax". kulkā for kunkā. "word").6. The l/r alternation is particularly frequent at Ebla. bl.g. i.g. where e. awal > awar. Instead. /Yilhaq-Da'mu/. mtn. The existence of the articles hn. qan(n) for qàrr < qârn. and in Tigrinya (e. with the etymologically and semantically corre­ sponding Egyptian term ns. as Aramaic br against bn. and East Gurage gāddrā. "son".

An insertion of r before another consonant is encountered in Mishnaic Hebrew hartom. The insertion of n should be rather explained in this case as the marking of the nasalization of the following consonant. related to Arabic hatm(un) and attested next to the usual hotām. The change / > d is reported in the Bantu languages (e.g. "to bring up". "anxieties".g.8). from the Aramaic root '11. "palm leaves". Therefore. frequent in South Ethiopian languages. and Gafat word qānd(à) < *qānr < qarn (Ge'ez). A nasal twang is quite audible with some Palestinian Arabs when they pronounce 'ain. 31.6-9). Tuareg a-mnukal. 17. e. "wheat" (root ś'r). in Greek). but not identical. ardent zeal". Proto-Bantu -fund. "statue of Baal". 17. Thus the alternation d 11 is encountered in Luwian and in Lycian (e. In the Central Mediterranean island of Gozo a peculiar shift / > d is attested in the Phoenician divine name sdmb'l < slmb'l. nowadays Ghawdex. Amharic anqafat for Ge'ez 'aqfât. can be observed in the Amharic variants sadsa (dissimilated from sassa < salsa) and salsa of the numeral "sixty".< -tunl-. However. "horn" (nd < rn).5) for *sa'appā. "obstacle". and in sar'appā (Ez. as the result of prefixing a . "passion. 94. Argobba. and in the name of the island itself TaoSos < gwl. explainable by the close articulation points of / and d.9. with the loss of the original pharyngal. "nose". and Oriental Jews use a strongly nasalized 'ain in Hebrew. Variant forms of this phenomenon.23) for śd'ippìm. Numidic mnkd. when a sequence of two consonants should appear in the beginning or at the end of a word. Plus-vocalic features of / and r are apparent also in Semitic. and in the change affecting the liquid r in the Amharic. 139. also with the original pharyngals h and '. probably in Proto-Berber (cf. "teach") and a similar phenomenon. a sequence of abutting consonants generally may not belong to one syllable so as to form a "consonant cluster" (§24.136 PHONOLOGY 17. related to Arabic šaġaf. attested next to he'ālā. The insertion of a non-etymological /. attested as sande in Highland East Cushitic and in Amharic. "to bring i n " . e.g.g. n or r is generally the result of the dissimilation of a geminated consonant (§23. An insertion of r before 'ain occurs in Hebrew śar'appīm (Ps. related to Arabic sa'af. In classical Semitic languages.19. or in the Oromo word sinra. e. this explanation is hardly correct in cases like han'ālā. in Aramaic hansāqā < hassāqā < *haslāqā.7. and as sandâ in Gafat. the Lycian PN Dapara transcribed Aanapác.8. "king"). occur in a number of languages.g.

ġlem. 18. 9. brát. especially in the case of plus-vocalic sonorants or liquids (/.g. and even to some ancient Semitic languages. e.V E L A R PLOSIVES 137 morpheme or dropping the case endings. gran. For the initial plosives.8). Nevertheless. there was also a way of pronounc­ ing q in Arabic that led to its occasional representation by " k " and q can alternate with k in modern Ethiopian languages. this rule does not apply to colloquial forms of Arabic. e. or be pre­ ceded by another consonant in the beginning of a word. kr-.g. q. It also possesses an emphatic velar plosive q. "soul". tr-)\ e. Common Semitic or Proto-Semitic has two velar plosives. IG is used for ig. voice­ less k and voiced g. e.g. but "fortis". "left-handed". between GA. -nz. ak or aq. gl-. "river". "theft" in Eastern Neo-Aramaic. since the one emphatic velar plosive corresponded to a pair of non-emphatic ones: voiceless k and voiced g. "iron. a Gurage dialect. to Amharic and other modem Ethiopian forms of speech. and even QA in certain regions like Mari and Eshnunna.1. gener­ ally articulated as the emphatic consonant corresponding to k and there­ fore also transliterated k. which does not mean "voiced" (g). "sheep" in colloquial Arabic. etc. qr-. to Eastern Neo-Aramaic. màngdst. r) that may be followed by another consonant at the end of a word (e. there is a wide tendency to use prosthetic or anaptyctic vowels. "rainy season". certain modern Arabic dialects either support a voiced pronunciation or reflect the shift q > g (cf. br-. slābā. §18.). The only initial clusters which do occur in Amharic are those involving / and r as second mem­ ber (bl-.3. 18. -st. Sibawayh defines Arabic q as maġhūra. as usual.g. there was of no phonemic significance whether q was produced with or without voice.2. as generally assumed. nàfs. V E L A R PLOSIVES 18. Throughout the whole course of cuneiform writing no attempt was ever made to indicate the exact char­ acter of a final plosive: AG serves as ag. qart. wànz. The syllabic cuneiform writing system is. a certain distinction is introduced from the Old Babylonian period on. rifle" in Chaha. KA. g. However. gr-. ik or iq. kalb. However. "government". but the emphatic velar plosive q . "city" in Phoenician). inadequate to indicate the distinction between k. In reality. kràmt. etc. More possibilities occur in final position (-fs.g. "dog" in Arabic.

kábiru. e. and of g > ġ. e. "to seize".5). "to bow". "to equalize in value". cf. kim and qim. and in Neo-Aramaic (§15. and a similar phonetic situation may occur under certain circumstances in modern South Semitic languages. and may be written then as h. and GIM for gim. but GAB is used for gab and qab. GI/KI. In Neo-Aramaic.6. from the Late Babylonian spelling tamāku of tamāhu. "he is rich". The voiced pronunciation is attested in intervocalic posi­ tion by Aramaic and Hebrew transcriptions.138 PHONOLOGY is generally indicated by signs with the "voiced" or "voiceless" conso­ nant. hulātt. 18.g.g. but yahahbir. In Tigrinya. which simply signi­ fies that the Aramaic verb tmk is used in this occurrence.2). in Ethiopian languages. but it is not phonetically conditioned in the same manner in the various forms of speech. Tigre kal'ot. e.g.resulting from the change k > k > h attained a phonemic status already in Antiquity. The Hadramitic preposition h. e.. is borrowed in Syriac under the form mkr. "grave". Such cases should carefully be distinguished. 18. §15. hanāšu instead of kanāšu. "two". . the spirantization of velars is widely attested. In Classical Arabic. as against tkltšr for Tuklat-lštar (> Iśśar. daġāġa. why the Babylonian loanword mahāru. KI serves for ki and qi.g. 18. In Modern South Arabian and Ethiopian languages.9). but cannot prove the existence of two phonemes /k/ and /g/ in Assyro-Babylonian. e. g > i. is attested in Ara­ maic as kwk (§63. The affricative pronunciation of k > c or k > š. mngsr for Mannu-kišarri. In Neo-Assyrian [g] and [k] seem to be positional variants of the same phoneme. and why Babylonian kimahhu [kiwah]. e. KU for ku and qú. the spirantized velar plosives attained phonemic status and their "hard" or "soft" pronunciation does not depend on their position.g. "he will be rich". vs. GU/KU may reflect a dialectal voiced articula­ tion of q in some areas.10) and explains. The occasional orthographic interchanges GA/QA.5.g. e. the pronunciation ġ is considered as the correct one (e. of q > c.g. especially in Amharic.4. The spirantization or fricativization of non-geminated and nonemphatic velar plosives is attested in various Semitic languages. "hen").. the non-geminated k is frequently spirantized into k or h in post-vocalic position. Hebrew and Aramaic spirantization of k/g follows the same rules as the spirantization of labials (§11. or g > z is attested in Arabic. in Mod­ ern South Arabian.g. The occasional cuneiform spelling with signs of the series " h " instead of signs with g/k reflects this change.

g. q .g. In Semitic the labiovelars phonetically conditioned occur in Arabic colloquials of Tripolitania and Morocco.g. ndk . indicated e. Mainly in the neighbourhood of a palatal vowel. "to cry".11). the direct change k > š is attested in some Arabic colloquials (e.i ) . Instead. and in Ethiopian languages (e.g. A series of labiovelars g . bâššâ < bky. but is rather unstable except for ġ. the passage g > z is attested in Soqotri and Śheri (e. while North Arabian inscriptions from the Tabūk (Saudi Arabic) and Ma'ān (Jordan) area seem to testify to an ancient conditioned g > 6 change. e) is common in Central Arabian bedouin colloquials (e. and also in Mehri. These sounds have a phonemic status and exist along­ side the ordinary velars (e.VELAR PLOSIVES 139 while the analogous tendency k > é is viewed as a dialectal deviation (e. developed from the signs for the velars. k . Therefore. kīfìš < képiki.g. Northeast Gurage). sīġān < sīqān.g. Amharic idb and gab. y 18.g. and some Gurage dialects (§11. The sounds ġ and z interchange in South Ethiopic under Cushitic influ­ ence (e. In Cairene Ara­ bic. "he was mourning". and occasionally if. "slave-girl"). in a verb that was probably belong­ ing to the fa HI class (*wagim > *watim). but not in Mehri (garēt. This fricative variant [z] is encountered also in Algeria and in southern Iraq. "liver"). "your stone"). In Modern South Arabian. although their phonemic status is due to the impact of the Cushitic substratum. zirit. "hyena"). "legs"). The palatalization occurs extensively in both Soqotri and Sheri. in Soddo. Palatalization of q into d or ġ before or after front vowels (j. the feminine suffixed pronoun -ki becomes . in Modern South Arabian (e. both colloquial and literary. The lack of traces of the Ge'ez labiovelars in an unvocalized text is probably due to the fact that the new symbols for the labiovelars.g. A non-conditioned q > é change occurred in the 'Azd dialect of northern Yemen. šubdet < kbdt. were invented at the same time as the vocalic signs. by the spelling wtm instead of wgm.g. the original pronunciation [g] is either preserved or revived. w w w w . die < dlk.7.g. the devoicing ġ > c occurs in the dialect of Palmyra. where the influence of Arabic is a factor of importance. The same is witnessed on the southern coast of Arabia and in early Andalusian Arabic. in the Neo-Aramaic dialect of Ma Tula (e. while [g] as well as [z] are attested in Moroccan Arabic. " I am"). Harari. "slave-girl"). occurs in all the Ethiopian languages. except in Tigre. "cock"). their develop­ ment in Ethiopian languages cannot be ascribed solely to Cushitic influ­ ence.

"mosque". On the other hand. wēh is monophthongized from *wayh < waġh. the change q > g is actually attested in Hadramawt and in Dorar. Id 'tal for hqtal).14). "stone". e. there are many cases of written g for q in Mandaic (e.g. for Syr­ iac qayta). espe­ cially in East Arabian. q becomes ' in some Arabic dialects. for baqara. This development seems to be different from an earlier change that occurred in Aramaic at least in two distinct phases. as in rll that in Kuwait means as well "man" (rīl < *ruyil < raġul) as "foot" (rīl < *riyl < riġl). "mosque".140 PHONOLOGY 18. in the realization of Hebrew q among certain Jewish communi­ ties of Algeria and Morocco. as in bgara. The change q > ' hap­ pened probably as the result of an economy of effort (§ 10. In some Gurage dialects.g. thus in a region that had contacts with Ethiopia.9-19). However. . The actual evidence is provided by the change q > ' in the Aramaic spelling of ś (d) (§ 16. Historical implications are obviously involved.9. and in East Arabian where it may result from a partial assimilation by voicing. In Bahrain. "few"). both from *masyid < tnasgid. Arabic q is almost invariably transliterated by Ge'ez " g " . In some cases. "stay!". g'yt'.7).8. after a change of the syllabic structure. "face". the syllabic structure has been influenced by this change.g. 18.g. hayar < haġar. or in ibg. The second stage ' > ' is attested in Late Aramaic and in Neo-Aramaic (§ 19. e. but the lack of glottalization differentiated Arabic q from Ethiopic q and may explain this transliteration. the velars k and q can become zero in medial position (e.g. and in Tigre at the end of a syllable (e. and Hebrew q is realized in some Jewish Yemenite commu­ nities like /g/ or /ġ/. tit < tsqit. "summer". from bqy. it represents the widespread reduction of the voiced pharyngal ' to a glottal stop (§ 19. It occurs also in the Mesopotamian g3/3i-dialects (gaht for qultu. It is noteworthy that a similar change is attested in Algerian Ara­ bic by the word msīd.10). in some of the Arabic dialects spoken in the ChadSudan area. The change ġ > y is attested in some Arabic colloquials. " I said"). and by the Maltese place name Msida.. the first one consisting in the shift q > ' which is supported by the spectrographic analysis showing the q between d and '. Besides. "cow".

they share some common features and. There are two frica­ tive pharyngals: voiceless h ([h]) and voiced ' ([Î]). because of their historic developments and of the way they are indicated in the various writing systems. while Arabic uses diacritics in order to distinguish the various phonemes. or a-la-ga-am Ihalākaml. 19. voiceless h ([x]) and voiced ġ ([y]). they oppose each other as spiritus lenis and spiritus asper in Greek. Palaeosyrian and Old Akkadian writing allows distinguishing the laryngals. as well as the two pharyngals h C ) and ' (' ). h by ' . pahhur. Common Semitic or Proto-Semitic has two laryngals: one glottal plosive ' ([?]) and one voiceless laryngal fricative h. "my father". a name which has been accepted in several circles even though it does not accurately describe all of them from the point of view of their articulation. while the twenty-two letters of the Phoenician alpha­ bet are insufficient to express all the phonemes of the languages which have adopted it. Nevertheless. are indicated in two ways: 1° by zero. h is essentially a pharyngalized laryngal. 19. a fortis.LARYNGALS. Greek m3p). These phonemes are often classified under the heading of "gut­ turals".3. ' by ' and ġ by ' is fol­ lowed in the present section. LARYNGALS. the pharyngals. The syllabic cuneiform writing system disposes only of signs indicating ' and h.g. It is convenient to examine Semitic laryngals. among them. il-ga 3 4 2 . The two laryngals ' (") and h (' ). the pharyngals. and the velar fricatives. This phenomenon is paralleled in other AfroAsiatic languages and in Indo-European. However. a tendency to be phonetically weakened and even reduced to zero.20). while the air consumed by the voicing of ' leaves it as a lenis.1. as shown by Hittite that has supplied the clinching evidence for the existence of laryngals and pharyngals in Proto-Indo-European (e. as in Amharic (§19. 2 3 4 5 19.2. PHARYNGAL AND V E L A R F R I C A T I V E S 141 19. as in a-bi /'abī/. PHARYNGAL AND VELAR FRICATIVES 10. "fire". 19. Of the two pharyngals. "to go" (accusative). and the velar fricatives. The Assyriological practice of indi­ cating etymological ' by '\ h by ' .4. Only the cuneiform alphabetic script of Ugarit and the South Arabian alphabet have adequate symbols for the laryngals. as well two velar fricatives. an articulated ġ is generally transliterated in syllabic cuneiform writing by h and not by '.5. pharyngal frica­ tives and velar fricatives in the same paragraph.

'À-da-ša IHadatal. "to empty". "upper" (feminine accusative).g. or harā'u{m). Ha-zi-ir /ùazzīr/. "Hero". also ' may be indicated by 'à. "fierce". however. means that the voiced phoneme ġ in sġr and ġrw had definite characteristics which influenced the vowels and separated it from the voiceless h of 'hd and mhr. and arābum.' the change of contiguous a to e. or Eb-du. a-zum I'azzuml.6). herū(m). but * does not influence. the pharyngals h and ' influence. "(The god) made". the phoneme h C ) when followed by the vowel a is expressed quite often by the sign É = 'à. This a > e change proves conclusively that h and ġ are to be distinguished. Besides. §21. 2° by special signs.. "fierce". thus differentiating the laryngal fricative h from the glottal plo­ sive '.11). The conventional transliteration of É as 'à does not indicate that the word or the name contains a true aleph. ' is exceptionally indicated by 'à.g. such as MÁ. "Servant of Resheph". da-la-'à-mu /talahhamu/. there is little doubt that a phonemic distinction tnust have existed between.142 PHONOLOGY lyilqahl. became sehēru(m). but expresses any of the to ' consonants. "he took". "he heard". or 'k-da IHaddal for the divine name. The fact that Old Akka­ dian mahāru(m). " M y god is favourable". as a rule. This spelling is at least a leftover from a period in which the phoneme h was independent from '. . as in iš-má lyišma'l. and É ('à). This change does not affect the laryngals ' and h.6. while this spelling occurs frequently with h. but it may have disappeared in course of time (§19. arābum. īb-'à-lu /Yib'alu/ or /Yip'alu/. ahāzu(m) in Old Babylonian. 19. In any case. "Youthful" or the like. e. Besides. "to enter" ('rb). Ha-la-ll IHala-'ll]. "to equalize in value". The two velar fricatives h and ġ are both indicated in Palaeosyr­ ian and in Old Akkadian by signs with h. a-li-dam /'alītam/. the vowel in the Palaeosyrian texts from Mari (but cf. remained mahāru(m). Occasionally.g. under certain conditions. "to seize". "to be small". "you will taste" (subjunctive). as in En-na-ì-lí IHenna-'Ilīl. or ahāzuim). In fact. u 4 d 3 The change 'a > "e occurs regularly at Ebla. There is also a convincing way of distinguishing ' from h in Palaeosyrian and Old Akkadian writing system. giving rise to homophones.Ra-sa-ap I'Ebdu-RaSapl. "to combat" (hrb). " E l is a maternal uncle". which are therefore to be distinguished from the pharyngals h and ' in Palaeosyrian as well as in Old Akkadian. as in Old Akkadian 'à-zum I'azzuml. while Old Akkadian sahāru(m). e. as in gu-la-'à-tum next to gu-la-a-tum and gu-la-tum (meaning unknown). e.

but ' and h are clearly distin­ guished in Amorite names found in Egyptian execration texts and in later alphabetic texts from Ugarit. contrary to h and ġ which are always expressed by signs with h.g.a w . 'bdnt < 'bd'nt.r a ' . This is easily explainable since the narrow orifice of the labial articulation scarcely affords a contrast to the narrowing of the pharynx. pharyngal and velar fricatives are indicated by a distinct symbol. "she called". not only in Human texts.7-8). and A-bi-hi-il /'Abī-ġēl/. dtnrd < dmrhd). An actual reduction of ' may occur when ' is contiguous to a labial. Ya-ás-ma-ah. all the laryngals. or Ha-ab-du. "Servant of Yarah".7. e. h > ġ ('bdyrġ < 'bdyrh).Ha-na-at /'Abdu-'Anat/. The pharyngals are often indicated by signs with h.LARYNGALS. like in i-ba-al lyib'all or in Da-mu /Da'mu/ (cf.g. These changes already announce the later development of the phonemes under consideration. e.g. and An-naDINGIR /Hanna-'Il/. 'u. 'abġl. yšm'. . In any case. 'i.8. "El is favourable".D i N G i R /Hanni-'Il/. These signs were employed also as vowels. there is no reason whatsoever to suppose that the articulation of these phonemes in Amorite was different from the one known from Ugaritic alphabetic texts. h> h (e. "The word of El has made". "Servant of 'Anat". Whether this reduced ' was then lengthening the adjacent vowel or was simply absorbed by the labial depends on the interpretation of the second / in the name Ib'ûfl [Yibāl-pī] or [Yibal-pl] of a prince of Mktry /Magdalay/ in an Egyptian execration text (E 5). 'ahrtp < 'ahršp). e. However. as well as the reduc­ tion of ' and of h to zero (e. all the pharyngal and velar fricatives are clearly dis­ tinguished in parallel names attested in texts from Ugarit: e. "El is favourable". personal names show occa­ sional changes ' > ' (e. PHARYNGAL AND VELAR FRICATIVES 143 19. à á 19.g. 'bdyrh. the laryngals are reduced graphically to zero. qr'at.g. Three cuneiform signs are used for the glottal stop ' according to its vocalization 'a. hn'il.8).g. even at the end of a word. h> h (e. e. Ab-di-a-ra-ah /'Abd-Yarah/. However. "Dagan has heard". but even in Semitic when the etymological ' was not pronounced in postvocalic or intervocalic position. In Amorite proper names written in syllabic cuneiform script.g.Da-gan /Yašma'-Dagan/.g. In Ugaritic.g. 'bd'nt. Therefore. 'abdhr < 'bdhr). e. Ammu-ra-pí-i /'Ammu-rāpi'ī/. / / a . " M y father is snatched away (?)". which provides a shortened form of the well-known Amorite name 1-ba-al-pi-El. h and ' may be reduced graphically to zero. §45. "The Ancestor is my healer".g. hnn < hnri). corresponds to qarāt and not to *qara'at (§45.

In reality. rahābum Ira'ābuml. "debt". "ask!".g. "to be terrified". Dialectal vari­ ations could influence the standard practice of the scribes. the conclusion that AssyroBabylonian words could begin with a vowel. hakāmu. contrary to the classical Semitic rule. rēšu (r's). sa-'-a-le and ša-a-le.g. haslsu (hss). On the synchronic level. irregular and optional in medial position.g. A partial identification of the etymological consonants which have coalesced in Assyro-Babylonian ' is at times possible. "to dig". "dust". e. In later periods. is usually absent at the beginning of words. "head". "to be". "to under- . In Assyro-Babylonian the laryngals and the pharyngal fricatives have been reduced to the glottal stop '. sehēru (sġr). 19. In particular. is unwarranted: the absence of a symbol does not neces­ sarily coincide with phonetic reality. "this". it is generally assumed that gemination of aleph.9. by the Ara­ bic 'alif 'al-wasl which is not pronounced in the classical language. e. However.g. So does the presence of the symbol not always mean that a glottal stop has to be articulated. as shown e. "wisdom". 19. hapārum /hapārum/. However ' and h did occasionally influence the same change a > e. retained in earlier periods.g. Amorite influence can probably be detected in spellings like Old Babylonian e-hi-il-tum for e-'-il-tum. and Aramaic influence in the NeoAssyrian form ha-an-ni-e for the demonstrative anniu. Besides. for h and ' had influenced the change of contiguous a into e. "to be small".10. e. esēdu I esādu (hsd).g. Yet the pharyngals h and ' are often indicated in Old Assyrian by signs with h showing that they were still preserving their phonemic status. by the glottal stop in English an aim contrasted with a name and by the very Neo-Assyrian variant hanniu of anniu.g.11. e. Assyro-Babylonian h corresponds in general to h or ġ. ša-ale stands for /ša 'all and that a is an allograph of ' or '-a. was lost in later dialects. the glottal stop is omissible and could therefore be considered as an allophone of the zero phoneme. the older praxis of indicating /'a/ by "a" may as well lead to the conclusion that e. however.144 PHONOLOGY 19. "to grow rich". The graphic notation of '. "to reap". e. it may indicate an etymological h in cases in which a change h > h had occurred in a period in which Proto-AssyroBabylonian still had an independent phoneme h.g. e. the use of a particular form of the sign AH to indicate ' from the Middle Babylonian and Middle Assyrian periods onwards leaves little doubt about the phonemic status of the glottal stop. epru ('pr). hanāmu (ġnm). ewûm (hwy). as shown e.g. However.

LARYNGALS, PHARYNGAL AND VELAR FRICATIVES

145

stand" (< Mm). A general shift h > h occurred in Eastern Syriac and in Neo-Aramaic (§19.14). Instead, the exceptional Neo-Assyrian spelling hanâšu for kanāšu, "to bend", is the result of a spirantization of k (§18.5). 19.12. The Canaanite dialects of the second millennium B.C. pos­ sessed not only the two laryngals and the two pharyngal fricatives, expressed by distinct letters in the "Phoenician" alphabet, but also the two velar fricatives h and ġ. While h is indicated in Egyptian by h, e.g. r-h-h, "Rehob", h is transliterated by h, e.g. d-b-h = Tú-bi-hi in an Amarna letter (EA 179). Semitic ' appears also in Egyptian as ', e.g. '-k-3, "Akko", while ġ is transliterated either by q (k) or by g, e.g. q-dt or g-d-t, "Gaza". These distinctions do not appear in the West Semitic alphabetic scripts of the first millennium B.C., when h is represented by "h" and ġ by However, e.g., the name of Gaza (ġzt), in Hebrew 'azzā, is consistently spelt TáQa in Greek and ġzt in Minaic inscriptions; the place-name ġufrā, "covert", in Hebrew 'Oprâ, is called Tocpepa in Greek, while Akko ('ky), in Hebrew 'Akko, appears in Greek as "AKT). Besides, e.g., Bethlehem (byt Ihm), in Hebrew Bēt-Lehem, is transcribed Br|0A,eeu. in Greek, but Jericho (yrhw), in Hebrew Ydrlhd, appears in Greek as 'Epi%(o or Tepi%G), and the proper name 'bhyl, in Hebrew 'Ablhayil, is transcribed 'Afh%<xiÀ,. These examples indicate that a pho­ netic distinction between etymological h and h, as well as between ġ and ', persisted in spoken Hebrew until the Hellenistic period. This phonetic distinction had a phonemic status allowing the Greek translators to dis­ cern, e.g., the 'zry (< ġzr) hmlhmh of I Chron. 12,1, who are "men valiant in battle", from the 'zry (< 'dr) rhb, "the helpers of Rahab", in Job 9,13. In Phoenician, instead, nothing suggests the survival of a dis­ tinction between the velar and the pharyngal fricatives. Any real trace of this distinction vanished also in the pronunciation of Hebrew in Roman times, and St Jerome (348-420 A.D.) never represents ' by g, the same being true of the Punic passages transliterated in the Poenulus of Plautus. 19.13. In Aramaic, the situation is also quite clear despite the use of the "Phoenician" alphabet. The velar fricatives h and ġ are always tran­ scribed by h in cuneiform script, e.g. Ba-hi-a-nu /Baġyān/, "the desired one". Instead, under different conditions, the pharyngal fricatives may either be transcribed by h (e.g. Ha-ab-di-ia = 'bdy) or by ' (e.g. Ba- '-lu = b'l), or correspond to an orthographic zero (e.g. Ab-di-ia = 'bdy;

146

PHONOLOGY

Ba-al = b'ī). These different spelling conventions mean that the velar fricatives h and ġ had definite characteristics which separated them from the pharyngal fricatives h and ' despite the fact that the alphabetic script used the same letter " h " for both h and h, and the same letter " " ' for both ' and ġ. These phonemes probably remained independent until the Hellenistic period, while the increasing cuneiform use of signs with h to transliterate the laryngal fricative h (e.g. Na-ga-ha-v.v /Nagah-Hadad/, "Hadad has shined") demonstrates the strength and the stability of this phoneme in the period under consideration. However, the h of the divine name hdd I hd is often reduced to ' or assimilated in personal names, perhaps under influence of Assyro-Babylonian Adad (e.g. 'dntn < hdntn, "Hadda gave"; mt'dd < mt'hdd, "Protected by Hadad"). 19.14. Middle Aramaic generally retains the independent articulation of the laryngals and of the pharyngal or velar fricatives, but original ' and h are liable to disappear in certain situations. The velar fricative ' may change into ', as in the Old Syriac proper names 'bdnhy < 'bdnhy, "Servant of Nuhay", and 'bd't' < 'bd'î', "Servant of 'Attā", while ġ changes into ', that is finally pronounced in Neo-Aramaic as an ' which is always retained. The h tends to be articulated /h/ in the West, but it is pronounced as /h/ in Eastern Syriac and in Eastern Neo-Aramaic. The consonants in question are frequently interchanged in Samaritan Ara­ maic, but the etymological spelling is generally retained in the other dialects. In Neo-Aramaic, the final syllable of a word was often written phonetically, but the actual tendency is to write it etymologically, e.g. -leh fief instead "of earlier -/<?. In loanwords the phonemes ', h, ', ġ are also found, e.g. Arabic hākim, "ruler", is written hakim, but is articu­ lated [hakim], while Syriac hakīmā', "wise man", is pronounced [hakktma]; Arabic ġalaba, "victory", is written glabtā, but articulated [ġlapta]. 19.15. In Middle Hebrew, the process h > h and ġ > ' is complete, but the Jewish European tradition realizes h as [h], while it generally reduces ' to ' or to zero. Instead, no velar fricative articulation is attested among the Jewish Arabic-speaking communities, which all retain the pronunciation of the pharyngal fricatives. In the Samaritan pronunciation of Hebrew, the laryngals and the pharyngals are reduced to zero. It is possible that the Masoretes have aimed at preventing a similar develop­ ment by means of their peculiar system of vocalizing the pharyngals in the biblical text (§27.10).

LARYNGALS, PHARYNGAL AND VELAR FRICATIVES

147

19.16. The laryngal ' was weak in Phoenician, as appears from the number of changes and elisions which it suffered. In Late Phoenician ' and ' seem to be losing their distinct consonantal values, as suggested by bd'štrt where ' occurs for '. In Punic, the gradual weakening and the final reduction of ', h, h, ' to zero are seen from spellings like Idn for I'dn, "for the Lord", o r ' d for 'hd, "one", from the frequent interchange of these letters in the orthography (e.g. b'l'mn, bhlhmn, b'l'mn for b 'lhmrì) and from their use as vowel letters in Late Punic and Neo-Punic (§21.14). Latin transcriptions of proper names, like Hasdrubal ('zrb'l), Himilco (hmlkt), etc., cannot be considered as proofs of an actual articu­ lation of Punic pharyngals, for the initial h- just reflects a fashionable Latin pronunciation. 19.17. In Arabic, the laryngals and the pharyngal and velar fricatives are generally retained. The laryngal h was pronounced distinctly in ancient North Arabian, since the divine name Nhy /Nuhay/ is transcribed Nu-ha-a-a in Neo-Assyrian and Nhy in Syriac (§19.14). However, there was a shift h > ' recognizable later in the prefix of the verbal form 'af'ala < haf'ala or in the particle 'in < hin, " i f " .The Arabic sounds h and ġ are usually represented in Greek by % and y. Also ' is transliter­ ated by y in the Damascus fragment (§7.44), e.g. XzyaX for la'all(a), "perhaps", but it is unlikely that something can be inferred from this fact. In vernaculars, dialectal changes ġ > ', h > h, ' > h, ' > ' are attested, and ' may disappear in certain situations, or be replaced by w, y, or be compensated by the lengthening of the contiguous vowel, e.g. in early South-Palestinian tarawwas far classical tar a" as, "he became chief", rayyis far classical ra'ls, "head"; in Maghrebine wekkdl for clas­ sical 'akkala, "he made (him) eat", mya for classical mi'a, "hundred", ūden for classical 'udn, "ear". These cases should be carefully distin­ guished from spellings like Safaitic my, "water", or s my, "heaven", where y is etymological, while it is replaced by hamza in Classical Ara­ bic mā' and samā'. The etymological y is preserved also in the Neo-Ara­ bic broken plural 'amyāh instead of classical 'amwāh.
d x

19.18. In Sabaic, one of the three letters ', ', h is occasionally omitted in a place where it would normally have occurred. These omissions reflect a phonetic trend towards the reduction of these consonants to zero in certain circumstances, e.g. yz for normal yz', "he shall do again"; imperfect ts r, "she will be aware", against perfect s 'rt immediately before; the divine epithet twn instead of the usual thwn. This trend
2 2

148

PHONOLOGY

appears also in cases where 'b, "(my) father", and 7, "(my) god", are reduced to b and / in compound personal names. In Hadramitic, the 'd is the equivalent of Sabaic'd, "to", which points to a change ' > ', widely attested also in Modern South Arabian and Ethiopian languages. 19.19. In Modern South Arabian, there is a shift ġ > ' and h > h in Soqotri, as well as a tendency for both pharyngals to become glottals. This trend is attested for ' also in Śheri and Mehri, although this conso­ nant occurs explicitly as a radical. E.g. 'h, "brother", is articulated in Soqotri asahi or dhi, while b'l, "possessor", is pronounced in Śheri b'ctl, ba'l or bāl, with the vowel lengthened and realized with the pharyngal constriction required for the pronunciation of '. 19.20. Ge'ez had all the phonemes in question, except ġ that has become ' in all the Semitic languages of Ethiopia. However, some speakers of Tigrinya articulate the voiced velar fricative ġ, which exists also in the Agaw dialects of the Qemant-Qwara group. In Tigre and Tigrinya, etymological h and h have coalesced into h, although the nongeminated k is frequently fricativized into k or h in Tigrinya (§ 18.5). A l l the laryngal and pharyngal fricatives tend to become zero in South Ethiopic. The h is still fairly common in Amharic, but it may be dropped as well; e.g. the word for "fifty" may be pronounced hamsa or amsa. However, in Harari, in Argobba, and in some Gurage dialects ', h, h are preserved in certain conditions, e.g. in Harari harāsa, "to plough" (root hrt); hadâra, "to pass the night" (root hdr); sàma'a, "to hear" (root šm'). In modern North Ethiopic, on the contrary, ' may disappear alto­ gether in word-final position (e.g. Tigre mulu', " f u l l " , pronounced [mulu]), while ' and ' may be in free variation with each other (e.g. Tigre ['adddha] or ['addaha], "noon"). The Amharic pronunciation, which reduces the phonemes in question to zero, has affected the spelling of Ge'ez texts, so that inconsistencies and interchanges blurred the orthog­ raphy of many manuscripts. 19.21. According to the Masoretic tradition, the laryngals and the pharyngals, as well as r which shares several of their characteristics, cannot be geminated in Hebrew and in Biblical Aramaic. In the Ethiopian idioms in which gemination is a regular feature, all the conso­ nants can be geminated except ' and h. In Neo-Aramaic, instead, the doubling of the consonants has largely been eliminated and replaced by the length of the preceding vowel. In Arabic, all consonants may be

L A R Y N G A L S , PHARYNGAL AND V E L A R FRICATIVES

149

subject to gemination (e.g.fa"ala, "he caused to make"), and this might have been the original situation also in the other Semitic languages. In any case, the Samaritan tradition geminates Hebrew r and this, doubling of r is confirmed by the Greek transcriptions Xappav, "Harran", Eappa, "Sara", etc., while Late Babylonian transliterations of Jewish names, like Mi-na-ah-he-mu, "Menahem", attest the gemination of pharyngals as well. 19.22. In conclusion, the correspondences of the laryngals, pharyngal and velar fricatives in the principal Semitic languages may be presented as follows:
Pr.-Sem. P.Syr. O.Akk. Amor. Ass.-Bab. Ugar. Hebr. Aram. Arab. E.S.A. Ge'ez

h h

h(' )

2

h h

h h b>b ġ>'

h h h> h

h h h 8

V)
h gib) h gib) h śib)

19.23. In the broader area of Afro-Asiatic, an alternation ' / g, inde­ pendent from the Greek transcription of ' < ġ by y (§19.12), can be observed when comparing Semitic and Cushitic roots; e.g., Hebrew 'ereb I Somali galab I Rendille geléb, "evening"; Hebrew 'es I Sabaic 'ś I Somali geid, "tree, wood"; Aramaic and Syriac 'all I Oromo and Somali gal / Rendille géèl, "to enter". A similar hlk alternation occurs e.g. between Semitic hrt and Libyco-Berber krz, "to plough"; Semitic hšb, "to assume", and Libyco-Berber kašaf, "to guess". Further research is needed in these comparative fields. 19.24. An initial ' may alternate with w (or y; cf. §15.7) without being the result of a change of wa- into 'a, or in the contrary. This alternation rather implies the existence of variant on-glides, as in Arabic 'ahad and wāhid, "one", from *had (§35.3); 'alifa and walifa, "to be familiar"; 'asmā' and wasmā', "the beatiful one", from *šmay/w; classical 'anātun, "languid woman", from Arabic wanā but Hebrew 'ānā, "to languish"; in Lihyānite 'āfaqū for usual wāfaqū, "they agreed"; Old Babylonian 'ahārum and wahārum, "to be behind". These analogical formations are particularly widespread in the Arabic verb, and the main methodological danger would consist either in considering colloquial wforms as newly formed from classical verbs with initial glottal stop or in

150

PHONOLOGY

assuming a passage of verbs with first radical w from Stem I to Stem IV because they appear with initial ' in Andalusian Arabic or in modern dialects (§41.11). Besides, this ' may simply introduce a prosthetic vowel. Further research in this matter is needed throughout the whole Afro-Asiatic field, because the alternation ' / w appears also when com­ paring e.g. Semitic waqru and Egyptian Ikr, "excellent".

11.

SYNOPSIS O F T H E CONSONANTAL S Y S T E M

20. To summarize the evolution of the Semitic consonantal system in the principal languages of the group, the following table may be of some use: .-Sem. P.Syr. O.Akk. Amor. X' ) b d 4 8 sib) h(' ) hO h k I m n P a r s s š s š t t t s w y z
2 4

Ass.-Bab.
>

Ugar.

Hebr.

Aram.

Arab.
<

E.S.A. Ge'ez
<

b d d 8 8 h h h k I m n P a r s s ś ś S t t t t w y z

r ) b d z g 8(h) h(' ) bn h k I m n P < 7 r s s š s Š t t t s w y z
2

4

b d 8 8(b) h h h k I m n P a r s s š s Š t t t s w y z

«(-4) b d z 8 b h k I m n P q r s s š s Š t t Š s w y z

b b b . b d d d d z d>d did 4 8 g Š 8 Ś ġ 8> ' ġ> ' h h h h h h h b h>h h>h h h k k k " k I I I I m m m m n n n n P P P f a <? < 7 •° r r r r s s s s s s s s š Ś ś > s g> š s sit q> ' did Š š Š s t t t t t t t t t Š t>t t t s z i >t w w w w y y y y z z z z

b d 4 8 8 h h h k I m n f a r s(s ) s ś(s ) Ś Ẁ) t t t t w y 1
3 2

b d z 8 h h h ìc I m n f < 7 r s s Ś> s Ś1 s s t t s s w y z

SYNOPSIS

151

^Haccforma ^pyrationc cametsfub^lcgibusrcncturpuru â i camcts primo loco manétis, ideoq; regimine vcrtitur in fchc ua ^ l ^ p r i m a nimirum per chi fie, quod alioquifchcua'gcmitnim exiftcrer inirio didiohisj dequo p3gin.5&\ Ai reliquaefof roar cameras vocalesanrefcemi* ninuiii murare non confueuc runt ctiarhalîoqiii mutabiíẃ;
B 5

~ Qtúnquarb. Vd<cCtenardihocdixerirrt,r$ quod lint fctminittd, qua* uocalem ante mum iant,ut n s y l p abominatiojn regimine mutat " prim* fyllabd* fub $ in ~\ (nam inputtfy lid* btsfcemimnis non conhūmeranda eft primafyl īabàí quantumfyeftatad ntotionum mutatio* tiem>mft UU prr canonrs mafculmorum fint mutábdcs) C7 dicitur TO%}r* Prouerb. 8. ey
4

alias Difiioporro

quam ClenardusM*

tulit,non in ufu repentur in liatu regiminis p* fitdidut compoftta cum dffixis. lohan. *faac. Etfidiflio T^ZnonriiJìftâ* tuabfolutoftng num. inBiblijs ufurpeturitá» 'ẃen nomma etusforms (cuius rKgẂÌà hie X Clendrtto popta eft) quam plurimd re)>eriutufz

yumtttd* non mutant. EXcipîunturì^ty
quod in regimine habet rtaft : in affixis

^Vaiuw'í.j.Dfí(ffro«,2i.n^nS in regit
mint n ^ n a . D « t f r W k $ . n i Ẁ in regimine Tti&iinpluráiuyỳmore mafcul.

Fig. 25. Fol. 62/3 of the Tabulae in grammaticam Hehraeam by Nicolaus Clenardus (Cleynaerts) (1493/4-1542), professor of Hebrew at Louvain in 1521-31, with comments by Johan. Quinquarboreus (Cinqarbres) and Johan. Isaac Levita, in the Cologne 1561 edition.

152

PHONOLOGY

12.

VOWELS

21.1. Common Semitic or Proto-Semitic has three short vowels (§10.5): low/open back velar a, high/close front palatal /, and high/close back velar u with strongly rounded lips. It also possesses the three cor­ responding long vowels ā, ī, a. Although additional vocalic phonemes have arisen in various Semitic languages, there are no sufficient grounds to suppose that other vowels beíbng to the original core of the Semitic phonemic system. The three vocalic -a-, - i - , -u- classes in the basic stem of the Semitic verb (§37.1; 38.3) and in the basic patterns of the Semitic noun (§28.5-12), the three Ugaritic 'a, 7, 'u signs for the glottal stop (§19.8), and the three vocalic phonemes of Classical Arabic show that these are the sole vowels constituting the vocalic core of the system. The situation is identical in Libyco-Berber. Besides, if one takes the evidence of primitive languages, such as those of America or as Australian Arunta, and considers the Bantu languages of Africa, there is a strong case for regarding a, i , u as primitive vowels, of which e and o are acci­ dental variants, unless they result from diphthongs. However, the real­ ization of the Semitic vowels a, /, u in actual speech can produce other vocalic sounds, mainly in the case of short vowels (cf. §10.11). There is a widespread tendency in Semitic to pronounce high and low vowels, especially when they are unstressed, as mid vowels [e], [a], [o]. Thus short [i] and [u] tend to become [a], as in Ethiopic (§21.30), and the same can happen with [a] (§21.6-8,10,13). Besides, [i] can easily become [e] by lowering the tongue, [u] becomes then [o]. The lack of appropriate vocalic signs, especially for [a] and [o], does often not allow determining the presence of these vowels in an accurate way, and "e" will then stand for [a] and " u " for [o] (§21.3). On the other side, a stressed short vowel tends to become long, and its articulation may at the same time be lowered (e.g. i>l> e) or raised (e.g. a > ā > 6). Some of these new vowels may acquire a phonemic status in a determined lan­ guage. 21.2. Despite their smaller number, the vowels are not second to the consonants with regard to their phonemic importance, as shown e.g. by Hebrew 'ab, "father", 'ēb, "bud", and 'ob, "bag". These words differ by only one phoneme, which is a vowel. Statistical examination of the relationship between consonant and vowel shows that an average Arabic text contains ca. 52% of consonants versus 48% of vowels. Statistical samplings of an average Ethiopic text give similar results: ca. 55% of

VOWELS

153

consonants versus 45% of vowels. Such statistical calculations offer a salutary corrective to the impressionistic approach to Semitic phonol­ ogy, in which the vowel is considered just as a secondary modifier of a consonantal root. 21.3. Besides a, ā, i, F, u, ū, North and East Semitic languages possess the vocalic phonemes e, ē. The existence of the vowels o, d cannot be proved directly, because the cuneiform writing system does not use spe­ cial signs with o, while variants like qurbu and qarbu, next to qirbu, "near", do not point to [qorbu], but indicate the existence of dialectal variations. Even Ugaritic 'u < 'aw cannot be considered as a conclusive proof of 'o, for the monophthongization aw > ū is as plausible as aw > o. However, the vowels o, o appear in Greek transcriptions of Late Babylonian words in the Seleucid period, e.g. o^ov for uznu, "ear", coei, for ūmī [ūwī], "days". 21.4. The vocalic quantity in North and East Semitic can often be determined only by comparison with other Semitic languages and by application of phonetic principles. In fact, the rule that long vowels can be expressed in cuneiform writing by an additional vowel, as in the type ka-a-nu for kānu, "to be stable", or sa-qu-u for šaqū, "to drink", does not apply to the older phases of North and East Semitic, when this scribal convention was still unknown. Even in later periods, the writing itself rarely indicates length by inserting a vowel sign after the sign for open syllable (e.g. ka-a, qu-u). 21.5. The alternating cuneiform notation of a long vowel in one case and of a "doubled" consonant in the other, e.g. sarru-u-ti and sarru-utti, "kingship" (genitive), should presumably be considered as a spelling convention and not as a phonetic phenomenon which is usually described as doubling of the consonant to compensate for the shortening of the preceding vowel. This variation must represent two different scribal devices used to indicate a long vowel by writing an additional sign which expresses either the sole vowel (e.g. sarru-u-ti) or the vowel plus the consonant of the following syllable (e.g. sarru-ut-ti). The latter practice is just a particular case of the so-called "continuous" spelling in which the final consonant of one sign announces the initial consonant of the following sign without aiming at indicating its gemination, e.g. liil-lik-kà instead of li-il-li-ka /lillika/, "may he come". Signs expressing consonant plus vowel plus consonant {C vC : UK) are not used to
x 2

154

PHONOLOGY

indicate geminated or long consonants, while pairs of syllabograms graphically doubling a consonant (VCJ-CJV) may either indicate a gemi­ nation (e.g. du-ub-ba Idubbāl, "speak!", root dbb), or express the lengthening of the preceding vowel (e.g. sarru-ut-ti Išarrūtil), or be devoid of any phonetic significance (e.g. Im-li-ik-ku-um, variant Im-likum llmlikumf). The alleged dialectal variation of Assyrian -uttu versus Babylonian -ūtu is hardly sustainable, for the spellings in -ut-tu /-ūtu/, etc., characterize the scribal practice at Ugarit, notably in the "General's Letter", at Boghazkoy, in the Amarna letters from Amurru, etc., where they cannot be regarded as Assyrian dialect forms. 21.6. The Palaeosyrian vowel e is secondarily derived from a under the influence of the consonantal phonemes h and as in En-na-ì-lí /Henna-'III/, " M y god is favourable", Eb-du- Ra-sa-ap /'Ebdu-Rašap/, "Servant of Resheph", both at Ebla, or Eš -tár-ra-at /'Ettarat/, "Astarte", at Mari. The long vowel ē is derived at Ebla from a diph­ thong -ay, as in Ti-iš-te-Da-mu /Tištē-Da'mu/, "Damu has drunk", with a prefix ti- of the third person feminine because of the sex of the name bearer, regardless of the syntax of the name. A short e (rather than i) may result from the reduction of a in unstressed syllables; e.g. ba-tá-qì i-dim /batāq yidim/ next to bí-tá-qì i-dim [betaq yidim], "cutting by hand"; ne-sa-qù(-um) [nešaqu(m)], "to kiss"; a-a-u mi /(l)a'ayu miyl, "getting of water", with loss of initial / (§17.2), next to lé-a-ù ma-a [le'ayu maya], "to get water", with initial sign NI. The long ā does not change into d.
d d 4

21.7. In Old Akkadian, e is secondarily derived either from a, follow­ ing the same conventions as in Palaeosyrian, or from i, as in E-li- /'Elī/ from 'ill, "my god". Long ē is derived from i followed by a "weak" consonant, as in ip-te /iptē/ from *yiptih, "he opened", from a plus a "weak" consonant, as in be-lí/bēlī/ from ba'll, "my lord", from a diph­ thong ay, as in Me-sar /Mēšar/, from *Mayšar, the deified "right", or from an original /, as in šÁM-me /ši'mē/ from ši'mī, "prices" in the oblique case of plural. Vowel i changes into u before š (e.g. ištu > uštu, "from"; cf. §48.18) and before an emphatic consonant (e.g. usārum < *hiśārum, "court"), confirming the velarized nature of the emphatics (§10.9). 21.8. In Amorite, the phonemic status of the short vowel e is uncertain, for e seems to be a positional variant or allophone of i in front oíh,h, I,

VOWELS

155

r, e.g. El J'El/ or // I'll/. Instead, the long phoneme ē appears in Amorite mēqtil names (e.g. Me-es -li-mu-um /Mēšlimum/, "Well-doer") which formally are causative participles *muhaqtil > *muyaqtil > *miyeqtil > mēqtil, attested in Semitic as late as Neo-Punic (e.g. mysql, "honour­ ing"; myqm, "raiser"). Some interchanges of i and u (e.g. Bi-ni- / Bu-ni-), of a and u (e.g. Sa-mu-1 Su-mu-), and of a and / (e.g. Ba-sa-ar / Bi-siir), in words like bn, "son", šm, "name", or in the name of the mountain Bishir, may possibly suggest the existence of secondary vowels of the i, ii, o, 6 types, but Arabic parallel cases of dialectal i against u (e.g. mishaf I rnushaf, "codex"), of u against a (e.g. summ I samm, "poison"), and of a against / (e.g. lahya / lihya, "beard") rather indicate that binum is used in one dialect, bunum in another, etc., and that these variations were originated under the influence of certain consonantal combinations.
s

21.9. From the three Ugaritic symbols 'a, 'i, 'u we may probably infer that the Ugaritic vowel system corresponds substantially to that of ProtoSemitic (§21.1). The existence of the phonemes ē and o cannot be deduced from the monophthongizations 7 < 'ay and 'u < 'aw, for ay, aw can also evolve into í (e.g. i-nu < yyn, "wine" in Canaanite; BiGia < byt'n, "House of the Spring" in Punic; BuxoÀAiov < *byt'l, "House of E l " in Phoenician) and ū (e.g. Mou6 < mwt, "Death" in Phoenician), like in Assyro-Babylonian (§21.10). In any case, syllabic transcriptions of Ugaritic names indicate a shift ay > I, as in Mi-ša-ra-nu = mšrn /Mīšarānu/, from *mayšarānu (root ysr, "right); I-nu-ia /'īnuya/, from *'Aynuya (root 'yn,"eye"); I-ia-um-mi /'īya-'ummī/, from *'Ayya'ummī,"Where is my mother?". Like in ancient Hidjazi poetry, also the diphthong iya can become I, as in Bi-di-'-lu /Bīdi-'Ilu/, from *Biyadi'Ili,"By the hand of god". An occasional shift ā > o occurs in personal names, e.g. A-du-ni- \J /'Addnī-Ba'al/,"Baal is my lord". It is attested also at Emar where the same persons are called Da-a-du or Du-u-du, Abba-nu or Ab-bu-nu.
d

21.10. Assyro-Babylonian presents a vowel system identical with Proto-Semitic, but with the addition of the phonemes e and ē, which were secondarily developed at various periods from a, ā, i, I. For e < a and ē < ā one can refer, e.g., to sebēru < *tabārum, "to break"; qebēru < qabārum, "to bury"; qerēbu < qarābum, "to approach", where r brings about the change ā > ē, like the velar fricatives (§ 19.11). Vowel i followed by any of the so-called "weak" consonants can change into e, e.g. i-ru-ub or e-ru-ub, "he entered", from *(y)i'rub. The change i > e

156

PHONOLOGY

can also occur before r, h, m, e.g. kal-be, "dog" in genitive, from kalbim, while the shift i > u is attested mainly before emphatics and labi­ als, like in ussu, "arrow", from *hittu, or ummu, "mother", from *'immu. The vowel a characterizes many Old Assyrian words which have i in other dialects; e.g. gamrum, "expenditure", as against Middle Assyrian gimru. There is also a frequent aju alternation; e.g. azni and uzni, "my ear". In Ássyro-Babylonian, the original diphthong ay changes either to I or to ē, e.g. i-nu or e-nu, "eye", from *'aynu. In the writing system, the signs can often be read with i or with e, e.g. gilge, ri/re, šìr/šèr, zikjzek, ib/eb, ir/er, etc. I n the late periods, the loss of final vowels occurs not only in the case of short vowels, as in aẁìl, "man", for the older awllu (nominative), awīli (genitive), awlla (accusative), but also in the case of originally long vowels, as i n rab, "chief", nāš, "holder", for the older rabī, nasi, which were shortened in the interme­ diate period to rabi, nasi. 21.11. Late Babylonian reveals a certain tendency towards alphabeti­ zation of the syllabary with use of odd vowels. This tendency appears not only in the transliteration of Greek words like 7TpoaTcn;r|s, "chief", spelt pu-ru-su-tat-te-su, but already in the spelling of genuine Akkadian forms like ú-zu-na-a-šu for uznāšu, "his ears", li-qi-bu-ni for liqbūni, "may they speak", or i-rak-ka-si for irakkas, "he ties". These odd vow­ els are devoid of any phonetic value and should be explained on a purely graphic basis. 21.12. Old Canaanite, known by texts written in cuneiforms, displays the same vowel system as Assyro-Babylonian. However, the change ā > o, expressed by cuneiform signs in u, is already attested at Hazor in the Old Babylonian period by the theophorous element Ha-nu-ta /'Anot/ < 'Anāt, while North Semitic preserves the long ā: Ha-na-at (Mari), A-natu/ti/te (Ugarit). In the 14th century B.C., the shift ā > o is shown at Pella, in northern Jordan, by the Amarna gloss sú-ki-ni /sokini/ (EA 256,9), from sākinu, "prefect", against Ugaritic sà-ki-ni /sākini/, and it is confirmed in Jerusalem by the pronoun a-nu-ki /'anoki/ (EA 287,66.69), " I " , against Ugaritic a-na-ku l'anākul. The long vowel ē resulting from the monophthongization ay > ē is then found, e.g., at Byblos with the probable qè-e-sí/qēsi/ (EA 131,15) from *qaysu, "sum­ mer", at Tyre with mé-e-ma Internal (EA 148,31), plural of may, "water", and in Palestine with ša-me-ma Išamēmal (EA 264,16), plural of *šamay, "heaven", "sky".

VOWELS

157

21.13. The Phoenician vowel system can be partially reconstructed with the help of Assyro-Babylonian, Greek, and Latin transcriptions of Phoenician words and names. The many dialectal variations result from the geographic and chronological dispersion of the sources, that witness a number of varying pronunciations. The impact of the Old Canaanite change ā > 6 (e.g. macom /maqdm/ < *maqām, "place") becomes stronger in Phoenician after the accent shift to the last syllable and the lengthening of the stressed vowel, which created a new group of long ā vowels. While the original vocalization of the verbal form is preserved e.g. in Ia-ta-na-e-li [Yatan-'Ell, "El has given", in the 7th century B.C., the change yátan > yatan > yaton is attested in the same period by Sa-mu-nu-ia-tu-ni /Samun-yaton/, "Eshmun has given". The new long a vowel, which resulted from the lengthening of a after the loss of a "weak" consonant, also changed into 6 and later into ū. Thus, the origi­ nal vowels are still preserved in Ba-'-li-ra-'-si /Ba'li-ra'ši/, "Baal of the Cape" (9th century B.C.), but a' > ā is finally reduced in Punic to a, e.g. in Rhysaddir /Rūš 'addīr/, "Mighty Cape", and a' > ā appears finally like d in Anniboni from Hanni-Ba 7, with the loss of final / or a change I > n (§17.4). In closed unaccented and in doubly closed syllables (e.g. A woo for Hanno) the vowel a was short and remained unchanged, although it could be pronounced colloquially as e (e.g. felu < *pa'lū, "they made"). Short / was rather lax and open, so that by the side of the usual MiÀx-, Gi(r)-, there occur the variants MeÀ,K-, rep-, for mlk, "king", and gr, "devotee". The diphthongs ay and aw could develop to ē/ī (e.g. aaur|u- /šamēm/, "heavens"; BiGia /Bīt-'ī(n)f, from byt'n, "House of the Spring") and o/ū (e.g. ICOLUAKOU from Yihaw-milk, "May the king give life!"; Moo0 /Mūt/, "Death"). The use of the matres lec­ tionis in Late Punic and Neo-Punic (§21.14) seems to reflect the lack of a phonemic distinction between u and o. It is noteworthy that the letter y is used in the Latin transcription of the Punic passages in Plautus' Poenulus where we would expect u or i. One should keep in mind that the letter y was not yet used in Latin in Plautus' time, so that it must have been inserted into the Mss later. This happened probably in Accius' day, some fifty years after Plautus' death, when the latter's work seems to have been edited first. At that time, however, Punic was still a very alive language and the representation of a Punic vowel by y would normally signify that it was pro­ nounced [ii] and corresponded to Ionic-Attic u. 21.14. The more widely followed system of vowel letters in Late Punic and Neo-Punic can be schematized as follows:

158

PHONOLOGY

"y" "w" ""' "'"

for i (e.g. tyt\ "Titus"), sometimes for e (e.g. syntr, "senator"); for u (e.g. Iwqy, "Lucius"), sometimes for o (e.g. rwm'n', "Romanus"); for e (e.g. p'lyks, "Felix") and o (e.g. 'nt'ny', "Antonia"); for a (e.g. grm'nyqs, "Germanicus").

A second, less successful but partly older system, uses " y " for i and " w " for w, like the first system. Besides, it uses ' " " for a (e.g. rm\ "Roma"), " h " for e (e.g. šhqnd', "Secunda"), and " " ' for o (e.g. 'd'n for 'adon, "lord"). In addition h and h could be used for a (e.g. hdn for 'adon; bhrk' for barako, "he has blessed him") and o/u (e.g. qlh for qolo, "his stem"; yhly', "Julia"; šmh for šamo', "he heard"). 21.15. The vowel notation in Hebrew was limited before the 7th/8th century A.D. to the matres lectionis used at least from the 9th century B.C. (§9.9). Therefore, any investigation of ancient Hebrew vocalism is as difficult as that of Early Aramaic. The use of Masoretic punctuation indicating the vowels of the consonantal text began only five hundred years after Mishnaic Hebrew had ceased to be a vernacular. It was inspired by the Syrian practice of vocalizing the Bible by means of dia­ critic marks and it is nearly contemporaneous with the similar vocaliza­ tion of the Qur'ān by early Arab philologists. The names and descrip­ tions of vowel sounds in Arabic, Hebrew, and Syriac show that much of the phonological theory then current was common to the students of all three languages; e.g. ptāhā, "the open one" in Syriac, is the vowel a, in Hebrew patah and in Arabic fatha. From the fact that Syriac manuscripts with diacritical marks go back to the 5th century A.D. and that similar vowel signs, with similar values, were used for the sacred texts of the three languages we can safely deduce that the system of Eastern Syriac served as model for the Hebrew and Arabic vocalizations. Besides, Syr­ iac influence is visible in writings on other facets of Hebrew and Arabic grammar, so that impact on the development of the vowel marks is not an isolated phenomenon. 21.16. Within the Masoretic system itself three different traditions can be distinguished: the Babylonian one, the older stage of which is very close to the Eastern Syriac system, the Palestinian tradition, continued by the Samaritans, and the Tiberian one, which is not attested before the 9th century A.D. The first and the second of these vocalization systems indicate the vowels by means of supralinear signs, while the third one uses sublinear symbols (with one exception). There are a few notable differences in the qualitative distinction of vowels between these

VOWELS

159

systems: the Babylonian and the Palestinian vocalizations do not distin­ guish a (patah) and e (segol). The lack of this phonetic distinction in an ancient pronunciation of Hebrew is confirmed by the Greek transcrip­ tions of the name of Esther as 'Acmip or 'Aa0f|p, which correspond to modern Jewish Yemenite pronunciation. Besides, the Palestinian system did originally not distinguish either a and e ox o and u. The combination of vowel signs with the matres lectionis suggests a certain quantitative evaluation of the vowels. None of these systems allows for a distinction of earlier Hebrew dialects. 21.17. According to the older stage of the Babylonian system, consist­ ing entirely of dots located above the letter and to the left of it, the vow­ els may be represented as follows: ā a e i o u

This system has probably led to some confusion because of the multi­ ple use of the set of two dots, and it was replaced by a system in which ā is symbolized by a small 'ain, a by the shape of aleph with one leg missing, and u by a small waw: j. •< i ā a e i o u

The transcriptions ā and a are approximate and simply correspond to the system followed in the transliteration of Hebrew (§10.3). However, they may be correct since vowel quantity is phonemically relevant in modern Tigre precisely and exclusively in the case of ā/a. A similar sit­ uation cannot be excluded for Hebrew as pronounced in the Babylonian tradition. 21.18. The Palestinian system is not a crystallized one. A few mss. do not distinguish between o and u, while the use of two different symbols for a and e in some classes of mss. imply the existence of allophones. Besides, symbols for a interchange to some extent with e and / in many mss. Such variations appear to reflect different Palestinian pronuncia­ tions of Hebrew over a period ranging from the seventh to the tenth or eleventh centuries A.D. Like in the Babylonian system, the diacritics are located above the letter and a little to the left of it: — or— a — or — e i o — u

In the Jewish Sephardic tradition. 21. e. ā.19. the three dots are replaced by a dot added to the waw. back [o]. The Tiberian system was developed by the Karaite families of Ben Asher and Ben Naphtali. the qames was considered as a representation either of a long ā or of a short 6. Besides these eight signs. both vocalized by members of the Ben Asher family (§7. the qames stands for ā in qām. Their values. and it stands for 6 in an original kull. e to express furtive vowels of the same quality. an additional one ~ (šdwā) is used to indicated a zero vowel or a furtive vowel a.20. "he stood up".160 PHONOLOGY This vocalization allows for a clear distinction of the basic vocalic phonemes of Common Semitic with the sole addition of the vowel e. . the qames was probably realized as a lower-mid. It prevailed in Hebrew manuscripts and later on in printed books. simply correspond to the current transliteration of Hebrew: A a ā (patah) (qames) ē (sere) e (segol) i (hireq) o u (holem) (qibbūs) When a wow is adjacent to the letter followed by the vowel u. This tradition was followed by the Samaritans with a few changes in the shape or the value of the signs: — J.y_ <. best represented by the St. " a l l " . and Latin texts.) contain a transliteration of the Hebrew Bible in Greek letters and the Dead Sea scrolls are characterized by an intensive use of . Greek. Petersburg Codex B 19 and the Aleppo Codex. The vowel system of earlier stages of the Hebrew language can be learned to a certain extent from the transcription of Hebrew words and names in cuneiform. the fragments of the second column of Origen's Hexapla (3rd century A.. often symbolized by . The corresponding translitera­ tion is used also in the present work. This tradition is phonologically justified for. Besides. the Hebrew vocalic system consists of seven full vowels. as well as from epigra­ phy. In the original' Tiberian pronunciation.g. This sign can be added to the vocalic symbols for a.5).D. e 21. In this system. This pronunciation was officialized in the State of Israel and it is gener­ ally followed in the teaching of Hebrew. instead. A a b e i ujo The transcription 6 is approximate. as indicated below.

i. with the use of matres lectionis '. where we have even long medial vowels without matres lectionis. These sources con­ firm the Canaanite shift ā > o. initially indicated only in final position by h or '. used by the Monophysites or Jaco­ bites.5). spelling tends to be plene. 21. generally when they are long (§9. is based on Greek vowel symbols and probably does not antedate the 7th/8th century A. The exceptions are rare.D. goes back to the 5th century A. In Late Aramaic. long or short. Aramaic uses the matres lectionis w. Syriac is the only Late Aramaic dialect to have standardized vocalization systems of its own.21. in Mandaic each and every vowel. h. w. rby. while Biblical Aramaic and the Targums or Aramaic translations of the Bible use the same systems as Hebrew.VOWELS 161 matres lectionis indicating medial and final vowels. but point to a dialectally differentiated sit­ uation of the diphthongs (§ 22.g. and ' to indicate the final and medial vowels o/u. "my teacher". y. long final vowels ē. The Eastern Syriac vowel signs are located above or under the letter. y.D.. 6 that disappeared in speech are in certain cases preserved in writing. pronounced rab. To indicate the vowels old. and a. even in the case of internal short vowels. '. "houses"). However. There are two different methods of vowel notation in Syriac: the Eastern system of dots. b'tyn.e. used by the Nestorians. ā. is spelled plene. while the Western one. The use of matres lectionis increases with the time and the vowel a. In Syriac.g. except in Samar­ itan Aramaic. contrary to the practice in other Aramaic dialects.6). h. begins in the Hellenistic period to be written plene with ' (in Mandaic also with ') even in medial position (e. 21. i/e. e. while the dot indicating a long ī is employed with the letter y: a iptāhā) ā/o izqāpa) or r e/i karyā) (rbāsālzlāmā e rbāsā/zlāmā qašyā o o/o ('esāsā rwīhā) o u/ū ('esāsā 'allīsā) (hbāsā) . u/ū and I. they are combined with matres lectionis which are widely employed in Syriac: the dot marking old or u/ū is used with the letter w.22.

Spellings like MoXmxos do not prove the contrary since Greek at in those times was pronounced like e (cf. as results from Greek transliterations like Avrjak'kas for 'wš'lh. a sometimes changes into o.g.24. "house". ā does in general not change into d. long ā at the end of words was expressed by ' as in other Aramaic dialects. " I " . Pre-Classical Arabic.g. neither the long vowels nor the diphthongs are indicated in Thamūdic and in Safaitic inscriptions. rīšo. mt. but pāroqā in Eastern Syriac.g. K E E I X O S < Kahll). Mo}ie%r|). "lord". In Aramaic as in Arabic. "head"). 21. e. Greek translitera­ tions show also occasional shifts u> o. at least. except in Western Syriac and in Western Neo-Aramaic. pdrūqd. i > e. but rabbúni. and a followed by a laryngal changes into e (e. is best represented by the con­ sonantal text of the Qur'ān. e. rábbon. It does not indicate the short vowels. from which derives Middle or NeoArabic. probably due to the following emphatic consonant (e. and a > e. "death". "gift". etc. In fact. ē and ī merge in Samaritan Aramaic: their quality and quantity is conditioned by stress (on the penultimate) and syllable (open or closed). "saviour". that system differed appreciably from that of Classi­ cal Arabic. "servant". The diphthongs were certainly monophthongized aw > o and ay>ē like in many modern Arabic colloquials. V < 'awš. bt. for the phonemes 6 and ē certainly existed in Thamūdic and in Safaitic.g. PoaaouaGos = Rdwt /Raśawatf). O^euos < 'Ulaym. 'n for 'anā. a > o.162 PHONOLOGY The Western Syriac vowel signs use the Greek vowel letters mostly irrespective of their quantity: A — a/ā (ptoho) o/o izqofo) JL e/ē irboso) A -L u/ū ('esoso) i/ī (hboso) 21. Thus in Nabataean Arabic names. In some forms of speech. Pre-Islamic North Arabian inscriptions do not furnish sufficient indications for a full reconstruction of the vowel system.g. but in Thamūdic and in Safaitic no long ā is indicated in writing.g. e.Instead. but . 21.25. e. Western Syriac presents cases of a shift 6 > ū and ē > J (e. while the old phonemes o and ū. as well as the modern colloquials. tm < taym.g. in Western Syriac. "my lord". Oae5os < 'ws^d. In Nabataean. at Ma'lūla. as a rule. Besides. the diphthongs are partially preserved in Nabataean names.23. the demonstrative dā is written d. This is shown by Greek transliterations as Noiepos < Ntyrw.

VOWELS

163

contemporaneous Greek transliterations of Arabic names show that it possessed the vowels a, d, e, i, o, u, e.g. OPaiSaAAa psv ApiXaaq xwv (3avi Ieaxcop ['Obedallah bdn 'Ab'ilaha' min banl Yds hof\, in the 7th century A.D. However, the phonemic structure of the short vowels was characterized by the opposition a : i/u, with allophones. The long vowels are indicated by matres lectionis: w and y express respectively ū/â and ī/ē, while ' indicates ā, except at the end of feminine nouns in the "absolute" state where h is used as mater lectionis for ā, like in Ara­ maic, e.g. klmh /kalimāl, "speech". However, these letters may also be etymological, e.g. in sw'l /sū'l/, "demand" (root š'l), later shortened to su'l in a closed syllable, or hdy Ihadiyl, "guidance" (root hdy).
x

21.26. Classical Arabic, formalized in the 8th-9th centuries A.D. by Arab grammarians, presents a vowel system which corresponds phonemically to the Proto-Semitic one with the three short or long vowels a/ā, i/ī, u/ū. The existing system of matres lectionis was complemented in the late 7th cen­ tury A.D. by a system of diacritics inspired by the Eastern Syriac system: a dot above the letter for a, a dot under the letter for i, and a dot in the midst of the letter or to its left for w. Duplicated dots, placed in the same posi­ tions, indicated the nunation -un, -in, -an (tanwīn). This old system, attested in the 8th century A.D. by Kufic manuscripts of the Qur'ān, was expanded towards the end of the same century by additional diacritical signs, and the dots were replaced by other diacritics, used henceforth in Qur'ān and Hadīth manuscripts, i.e. from the 9th-10th centuries on: ia (fatha) i (kasra) 1. u (damma)

For the notation of long vowels, these diacritics are consistently com­ bined with the matres lectionis ', y, w and added to the preceding conso­ nant. A special symbol called sukūn (^-) denotes the absence of a vowel; it is also called ġazma when placed on the final consonant of a word. In order to distinguish the glottal stop (hamz) from the mater lec­ tionis ā, a sign called hamza ( —) is placed above the letter '. To signify the reading 'ā with the glottal stop followed by a long ā, another symbol is placed above the letter, namely madda (-^). In the Qur'ān, however madda is used to indicate ā', V, ū'. The prosthetic ' (§ 27.15), albeit tra­ ditionally written, is not supposed to be pronounced in the Classical lan­ guage and it is therefore distinguished by a symbol called wasla or sila {—), e.g. in 'sm, "name"; 'bn, "son".

164

PHONOLOGY

21.27. Early Arab grammarians and descriptive studies of modern col­ loquials reveal the extensive variations in the timbre of Arabic vowels. The main tendencies are already described in the traditional Arabic grammars which single out two principal phonetic phenomena: 1° imāla, a non-conditioned palatalization ā > ē, e.g. sēra instead of sāra, "he arrived"; 2° tafhīm, a velarization ā > o, sometimes conditioned by the neighbourhood of emphatic consonants, but heard also, e.g., in saldmun 'alaikum. In modern colloquials, 6 is often a free variant of ū (e.g. bākor, " f i g " , in Algeria), but it can also result from the contraction, e.g., of the pronominal suffix -ahu > -o. In several dialects, the short a is preserved, but i and u change into a, unless they occur in a final closed syllable where they are pronounced e and o, e.g. in Damascus. In some colloquials of North Africa, all the short vowels are elided or reduced to 9. Thus, the very short vowel a can derive from original a (e.g. sahra, "rock"), i (e.g. ddll, "shade") or u (ba'd, "distance"). In general, the variations can be very important as shown, e.g., by the different pronun­ ciations of the word "name" in urban colloquials of Algeria: Isdm at Algiers, sdm at Djidjelli, āsam at Cherchel. These developments are partly depending on the Libyco-Berber substratum and adstratum. 21.28. In Epigraphic South Arabian, no vowels are indicated, except for the use of w and y ambivalently for either consonant or vowel nota­ tion. The spelling of the pronominal suffix -hmw indicates a pronuncia­ tion ending in -u and variant spellings like ywm and ym, "day", in the same inscription obviously express the same pronunciation yam or the like. Similary, y might represent I / ē. Instead, there is no notation at all for long ā, except in two or three aberrant cases. 21.29. In Modern South Arabian languages, vowel quality and quan­ tity are closely related to stress. In Sheri and Soqotri, there is a large range of vowels a, e, ā, a, i, o, d, u, which are generally long in stressed syllables. In Mehri, there are five long vowels ā, ē, F, o, ū, and two short vowels d and a, which occurs only in stressed closed syllables. E.g. kdtob, "to write", corresponds to kt'db in Śheri and Soqotri. 21.30. Old Ethiopic or Ge'ez was at first written in a purely consonan­ tal way, like Epigraphic South Arabian, but in the 4th century A.D. the consonantal symbols were provided with regular vowel markings by adding short strokes or circles and other alterations in the shape of let­ ters. The vocalism which is manifested by this notation distinguishes

VOWELS

165

seven "orders" or syllabograms for each consonant, as can be seen in Fig. 21 (§9.7). These distinctions are essentially qualitative, with the probable exception of Ge'ez ā and a. Also in Tigre, vowel quantity is relevant phonemically only in the case of a/ā. Beside this case, the Proto-Ethiopic origin of the single modern Ethiopic vowels can be reconstructed as follows: ā < a; u < ū; i < F; a < ā; e < ay; d < i/u; o < aw. In classical Ge'ez, the vowel a is reduced to d before laryngals and pharyngals. The Ethiopic change i > d and u > d is paralleled to some extent in Tuareg (e.g. ergative d-mnokal < *u-mnokal, "king"; ta-barart < *tu-barrart, " g i r l " ; plural d-ndsldm-dn < *u-ndstem-9n, "Moslems"; non-active case d-lkas-dn < *i-lkas-dn, td-lkas-en < *ti-lkas-en, "gourds") and a similar tendency is widely attested in Arabic colloqui­ als (§21.27). 21.31. The modem Ethiopian languages have several additional vowels, short and long. The two central vowels à and d are the most frequent ones, but they show variation in pronunciation depending on the surrounding sounds. A l l the vowels can be nasalized and the long vowels are generally phonemic. They may result from various phonetic developments, as dis­ appearance of intervocalic, prevocalic or postvocalic consonants, or con­ traction of contiguous short vowels. Short vowels are liable to variation and allophones occur frequently. In particular, [o] seems to function some­ times as an allophone of a zero phoneme. For further details, studies of the various languages and dialects concerned have to be consulted. 21.32. There is a widespread tendency in spoken Semitic languages to weaken short -a- to -d- or -e-, and to -i-. The resulting vowel has an indistinct timbre, especially in an unaccented syllable preceding a stressed one. Very often such vowels disappear altogether. There is little doubt that this unconditioned weakening of -a- took place also in ancient Semitic languages, but it is concealed by the conventional and system­ atized spelling of the scribes. At Ebla, however, where there was obvi­ ously no longstanding tradition of writing the local idiom, the variant spelling of proper names reveal the same tendency. Thus, the same place name may be written a-ga-lu or i-ga-lu \ the same personal name may be spelt a-da-ad-mu or i-da-ad-mu, 'à-gi or i-gi. At Ugarit, instead, the occasional change a > i results from a vowel assimilation and is thus phonetically conditioned; e.g. 'ihqm /'lhlqam/ < 'ahqm. There are also examples of a change a > o in close syllable, e.g. in the name Sabbat(ay) transcribed Io|3|3aeo(<;) in the 3rd century B.C. (§21.24).
kl k

166

PHONOLOGY 13. DIPHTHONGS

22.1. Diphthongs are continuous monosyllabic speech sounds made by gliding from the articulatory position for one vowel towards that for a semivowel ("falling" diphthong), or the opposite ("rising" diphthong). They usually undergo a different development from that of their compo­ nents. The combined sounds are subject to a number of conditioned changes which will be dealt widi in the appropriate paragraphs. Some changes, however, are not conditioned: they affect, in particular, the falling diphthongs aw, ay, whose development presents several varieties. When the semivowel w or y is not long or geminated, as in ayyābu, "enemy", or qawwām "manager", the diphthong is often monophthon­ gized. Thus aw is reduced either to ā or to o/u, and ay changes either to ā or to ē/ī. 22.2. Besides, diphthongs can arise when two vowels meet or they can originate from long vowels, the diphthongization of which leads to the creation of new nominal and verbal patterns. E.g. the colloquial Arabic verb ġawraba, "he put on socks", derives formally from Stem I I I ġāraba of the root grb (§22.17), with lengthened first vowel, while Stem I I of the same verb means "to test, to try". 22.3. The phonetic shifts aw > ā and ay > ā are found in several Semitic languages. In Palaeosyrian, at Ebla, the variants a-mu 'à-mu-tum and a-aw-mu 'à-mu-tum f(y)a(w)mū hammūtum/, "hot days", both with the elision of initial y, suggest the dialectal coexistence of the diphthong and of the contracted form ā < aw, while the change ay > ā appears fre­ quently, e.g. in ba-tu < baytu, "house", or in Ba-na-a-hu /Banā-'ahu/, "The brother is nice". The name of Ebla itself, spelt lb/Eb-la , later E-eb-la-a- or I-ib-la-a in Hurrian texts, testifies to this monophthongization, since it is still written Yb3y = Yiblay or Yeblay in an Egyptian execration text which mentions its Amorite king Šmšwìpìrìm = Šamšu'app-'ilim, "The sun is the face of God" (E 43). Also in Amorite names, we find the variants Ia-aw-si-DiNGiR and Ia-sí-DiNGiR /Ya(w)śi-'El/, "El went out", with the change aw > ā, while the shift of final ay > ā appears e.g. in Ra-sa- Da-gan /Raśā-Dagān/, "Dagan is pleased". The shift 'ayn > 'ān, "eye", is implied at Ugarit by the spelling of lGl-at l'Anāt/, and the widespread change ay > ā is confirmed there by syllabic transcriptions like Ma-ag-da-la-a for Mgdly, Sá-am-ra-a for Tmry, etc., which indicate the monophthongization ay > ā at the end of a word. A n
kl á md

DIPHTHONGS

167

earlier pronunciation is attested for the first place name by Mktry /Magdalayl in an Egyptian execration text (E 5 ) . 22.4. The change ay > ā is widely attested in Arabic. Thus, e.g. Safaitic my, "water", and s my, "heaven", become in Classical Arabic mā' and samā', while early Arab grammarians mention 'alāhā, "upon her", for 'alayhā; salām 'alākum, "peace upon you", for 'alaykum; i/āka, "to you", for ilayka; ladāka, "with you", for ladayka, etc. In modern vernaculars, the change ay > ā occurs in closed syllables of the Syrian dialect of the Nusayris (e.g. bāt, "house", but bayti, "my house"), and before accented syllables in a few dialects of Southern Tunisia (e.g. zātun, "olives"; bada, "white"). In verbal roots with "weak" w/y as third radical, the singular of the base-stem shows the monophthongization -aw > -ā and -ay > -ā when the semivowel is not retained; e.g. ġafā < *ġafaw, "he treated harshly"; bakā < *bakay, "he wept". In fact, the termination -a of the perfect in Classical Arabic is not attested either in modern colloquials or in Safaitic, judging from Greek transliterations like MaaaxnA-Os, Lautx%T|}ios, etc. And a pro­ nunciation 'atā, bakā, etc., as in Classical Arabic, is unlikely in Safaitic because of the spelling 'ty, "he came", bky, "he wept", etc. The monophthongization of -ay > -ā seems to have taken place in Arabic quite late. E.g. the name of the Arab goddess al-'Uzzà is written 'zy not only in the Nabataean name 'bd-'l-'zy, "Servant of al-'Uzzā", but the Syrian writer Isaac of Antioch, in a poem describing events of ca. 4 5 7 A.D., still renders her name as 'wzy. Also the frequent Arabic spelling of final long ā with a -y goes back to the Pre-Classical language in which the monophthongization had not yet taken place. But the final -ay becomes -ā in the interior of Safaitic composite names, e.g. wh'l /W ahā-'It!/,"God has revealed". This development is attested in most Semitic languages.
! r

22.5. In Old Akkadian, the original diphthong ay changes to ē or f, as in Me-sar < *Mayšar, "right"; e-ni-a < *'aynīa, "my eye"; bi-tum < *baytum, "house". The original diphthong aw is reduced to ū, as in u-mi- < *yawmi-, "day"; u-su-zi < *ušawśi', "he led on". The reduction does not take place when the semivowel of ay is long or geminated, like in a it-ti-in /ayyiddin/, "may he not give". The same changes occur reg­ ularly in Assyro-Babylonian (§21.10), with the same exception of ayy, e.g. ayyābu, "enemy". In North Semitic, changes ay > ā (§ 22.2) and ay > ē/ī are attested at the same time, suggesting the existence of such

168

PHONOLOGY

unknown factors as the dialectal distribution, e.g. in texts from Ugarit (§ 21.9; 22.3). 22.6. The shifts ay > ī/ē and aw > ū/o occur in Old Canaanite (§21.12), in Phoenician and Punic (§ 21.9,13), and in ancient Hebrew, where spellings like yn in the Samaria ostraca instead of yyn, "wine", 'b instead of 'wyb, "enemy", and 'nm instead of 'wnm, "their sin", in a Qumrān fragment (4QPs 89), indicate the reductions ay > Hē and aw > ū/6. In Masoretic Hebrew, the diphthongs aw and ay remain generally unreduced when the semivowel was originally long or geminated (e.g. hay < *hayy, "living"), when the diphthong constitutes the final syllable of a word which is not in the construct state (e.g. layl, "night"), or when it precedes the enclitic particle -h expressing the direction (e.g. šāmaymāh, "towards heaven"). However, there is a tendency not only to preserve or to restore the diphthong, but even to split it into two sylla­ bles, e.g. bayit < *bayt, "house"; māwet < *mawt, "death". 22.7. This tendency to expand and split the diphthongs radically dif­ fers from the general trend observed in Pre-Islamic North Arabian (§21.24) and in Arabic, except in the classical language which preserves the original diphthongs in medial position, but often reduces -ay to -ā at the end of words (§ 22.4). The reduction -ay > -ē is indirectly attested at al-Kūfa where Qur'ān codices frequently write -' instead of the final -y and where long ā was subject to imāla (§21.27). In the Syro-Palestinian dialects, the diphthongs ay and aw, followed by a consonant, are gener­ ally reduced to e (e.g. lēl < layl, "night") and 6 (e.g. tor <tawr, "bull"). However, there are also cases of a monophthongization ay > I (e.g. bltār < Syriac paytārā, "farrier") and aw > ū (e.g. ġū'ān < ġaw'ān, "hun­ gry"). Instead, these are the best attested reductions of the diphthongs ay and aw in Arabic colloquials of North Africa; e.g. bhlra < buhayra, "pond"; bin < bayn, "between", mūġa < mawg«,"wave"; yūm < y#wm,"day". Cases of aw becoming ū are rare outside the Maghrib, but the change of initial aw > ū, as in ūlad < awlād,"children", is ascribed to the ancient Tamīm dialect of northeast Arabia. 22.8. The contraction of diphthongs in Early Aramaic is attested in certain dialects and in certain circumstances. In the Tell Fekherye inscription from the mid-9th century B.C., the closed syllable of the nonsuffixed construct state of byt in bt hdd, "temple of Hadad", shows the reduction, while the open syllable of the suffixed form byth is spelt with

DIPHTHONGS

169

yod. The monophthongization of bayt might have been bāt, as suggested by Bat-ti-il-, Ba-ti-il-, Bathillo in Latin transcription, i.e. /Bāt-'Il/, an Aramaic divine name meaning "God's house", but the reductions ay > I and ay > ē are also attested, e.g. by E-ni-il designating the same person as A-i-ni-U, "Eye of God", or by Sa-mir-i-na besides Sá-ma-ra-'-in, "Samaria". It is clear that forms preserving the diphthong ay and forms reducing it to ā, ē, or ī could coexist in the same area and in the same period. Also in later times, the diphthong ay could be either preserved or monophthongized. In Syriac, it is preserved unless occurring in a doubly closed syllable, e.g. 'aynā, "eye", but 'ēn in the construct state. In Eastern Neo-Aramaic the reduction ay > ē is general, e.g. bēsa < baytā, and ē shows a tendency to change into i. A similar reduction occurs in Tūroyo with the monophthongization ay > i; e.g. milef < maylaf, " i t is learned". Instead, the diphthong ay is generally preserved in Western Neo-Aramaic. 22.9. The parallel use of 7 yrwh and 7 yrwy, "let it not be sated", at Tell Fekherye, suggests a reduction to -ē indicated by -h, while -y prob­ ably represents either a historical spelling, inherited from an older stage of the language, or a secondary diphthongization -ē > ~ey. Since the diphthong -ay at the end of perfect forms of the basic stem is invariably reduced to -ā (e.g. band < *banay, "he built"), the reduction to -ē at the end of a word should be rather explained here by the contraction -iy > -I > -ē (*yarwìy > yarwī > yirwē), well attested in Aramaic by variants as Zab-di-ia /Zabdìy/ and Zab-de-e /Zabdê/, Ba-ni-ìá /Bāniy/ and Ba-né-e /Bānē/, designating the same person. 22.10. The diphthong aw can be preserved in Aramaic or be reduced to o I ū. The spelling with w, even in Early Aramaic inscriptions, does not allow deciding whether w indicates the semivowel, or is used as a mater lectionis for 6 / ū, or is simply a historical spelling. E.g. the name mwdd, from the root wdd, "to love", appears in cuneiform transcription as Mu-da-da, Mu-da-di, and in Greek as Mco8a5. In Syriac, the noun "death" is spelt mawtā, but its construct state appears as mūt, because the preservation of the diphthong would then result in a doubly closed syllable. In Eastern Neo-Aramaic, the diphthong is always reduced to 6 (e.g. motā, "death"), unless the a is long or the w geminated, as in qawwama, spelt qāwemā, "to get up, to go". The same reduction is applied to diphthongs originating from the change b > b > w, as in gabrā > gabrā > gawrā > gorā, "man". The o shows a further tendency to

170

PHONOLOGY

change into ii. A similar monophthongization dw > u occurs in Turoyo (e.g. ktúli < *ktáwli, " I wrote"), while aw is generally preserved in Western Neo-Aramaic. 22.11. In Epigraphic South Arabian, variant spellings like bt and byt, "house", or ym and ywm, "day", even in one and the same text, indicate that both can be facultative variant orthographies for a single pronuncia­ tion, probably bēt and yom, according to the modern Hadramawt collo­ quial which contracts ay into ē (e.g. ēdā < 'aydan, "also"; 'ēn < 'ayn, "eye"), and aw into 6 (e.g. yom < yawm, "day"); cases of aw becoming ū are rare outside the Maghrib. A n alternative interpretation, which would introduce a distinction either between open and closed syllables, or between stressed and unstressed syllables, does not explain all the variant spellings byt I bt and ywm / ym, since they occur in the same forms, as ywmtn and ymtn, "the days". 22.12. In ancient Ethiopic or Ge'ez the diphthongs appear in reduced form, e.g. yom < yom, "today", lelit < lēlīt, "night", but there are many variations. In the modern Ethiopian languages, the number of divergent realizations of diphthongs is even greater. E.g., while the "threshing floor" is called awd in some Gurage dialects, it is od in others. Simi­ larly, "sheep" is said tay (< tali) in some Gurage idioms, why it is te in other dialects. The reduction pattern appears to be aw > o and ay > e. 22.13. The diphthongs iw, uw, uy are reduced to ū in Assyro-Babylon­ ian, e.g. ūbil < *iwbil, "he brought"; Šūbil < *šuwbil, "send!", šūšur < *šuyšur, "is kept in order". Original iy is instead monophthongized to I, e.g. ide < *iyda', "he knows". In Arabic, instead, iw changes to iy > t at the end of a syllable; e.g. īqā' < *'iwqā', "rhythm", from the root wq'. Also uy develops to iy > ī, sometimes to uw > ū; e.g. bid < *buyd, "lay­ ings (of eggs)", from the root byd; mūqin < *muyqin, "certain", from the root yqn. When the Phoenician orthography was fixed, the suffix -iy of the first person singular was still pronounced -iya after nouns in the oblique cases, e.g. 'by, " o f my father". With the loss of final short vowels it was reduced to -iy, which in time was simplified to -J, but the writing with y was preserved and even extended by analogy to nomina­ tive nouns (e.g. 'my, "my mother"), despite the fact that Phoenician does generally not use any matres lectionis. The same development is attested in Palaeosyrian by synchronic variants; e.g. i-a-la-nu ['iyalānu] and i-la-nu-um or ì-la-núm ['īlānum], "a large tree". In Hebrew, uw

DIPHTHONGS

171

becomes ū and iy changes in i ; e.g. hūšab < *huwtab, "he set"; yīraš < *yiyrat, "he will inherit". Corresponding reductions are also common in Ethiopian languages. 22.14. The rising diphthongs yi- and yu- of the prefixed verbal forms of the third person are not indicated in Palaeosyrian, but they are proba­ bly signified in Old Akkadian by the spellings i-ik-mi- lyikmī-l, "he cap­ tured" (root kmy), i-is-e- /yiš'ē-/, "he searched" (root š'y), u-ub-lam /yūblam/, "he brought" (root wbl), u-ur-da-ni /yūrdanni/, "it went down on me" (root wrd). Similar spellings occur in Old Assyrian; e.g. ú-ub-lu lyublūl, "they brought"; i-ìš-qú-ul /yitqul/, "he weighed out". In Amor­ ite, yi- is only expressed by i-, but yu- is attested by the name Iu-um-raa5-DiNGiR /Yumraś-'El/, "El grieved". In Assyro-Babylonian, yi- is reduced to i- and yu- to u-. Also ya- is monophthongized to e.g. idu < *yadu, "hand", but the alternative spellings with a- and /- in Palaeosyr­ ian indicate the change ya- > yi-, without monophthongization; e.g. a-me-tum /yamittum/ and i-me-tum lyimettuml, "right side". This change explains the form yi- > i- of the prefixed personal in most Semitic lan­ guages (§40.31) and, occasionally, of the first syllable ya- in the basic stem of verbs like yāda', "he recognized"; e.g. in the Edomite proper name Qwsyd' transcribed KOOT8T| in a bilingual ostracon from the 3rd century B.C. 22.15. Secondary diphthongizations are to be found in Semitic lan­ guages when two vowels meet. In such a case, either the two vowels coalesce and there is crasis (e.g. Arabic ī+ū > ū; ū+ī > i; Tigrinya d+a > a; Amharic a+a > a), or a "hiatus-filling" semivowel y or w is pro­ duced. The so-called "weak" verbs, the root of which is monosyllabic and contains a long vowel ā, l, ū, give frequently rise to such secondary diphthongs, e.g. qūm, "to get up", śīm, "to place", kūn, "to be". Thus, the active participle of qūm is in Aramaic either qym /qāyim/, or q'ym with a medial mater lectionis, or q'm with ' substituting the y after ā. While the form qym goes back to the 6th century B.C., the glottal stop replaces the glide y only in the 2nd century B.C. Also in modern South Ethiopic, the glottal stop may replace w or y, as in e'àdā, "to tell", and we'a, "to go down", in one of the Gurage dialects, against ewādā and wayā in other dialects. The situation in Arabic is similar to that of Ara­ maic. In Pre-Classical Arabic, as apparent in the consonantal text of the Qur'ān, the active participle of the same verb is q'ym /qāyim/, "stand­ ing", which was reinterpreted in Classical Arabic as qā'imun. Such

172

PHONOLOGY

changes are well-known to Arab grammarians who call them 'ibdāl nahwī or sarfì, "grammatical substitution", but consider usually that hamza is replaced by wāw or ya, although the etymological form is, e.g., miyar, "provisions", while mi*or is a secondary form historically. Safaitic inscriptions use sometimes ' as in k'n /kā'in/, "being", but in some cases y is written even instead of an etymological ', chiefly in the neighbourhood of the vowel i, as in hnyt /hāniyat/, "maid" (root hn'). In Epigraphic South Arabian, there is no trace of the practice of substitut­ ing ' for w/y after ā, and the modern Arab colloquials are identical in this respect with the Pre-Classical language. E.g., in Syro-Palestinian dialects, the participle "seeing" of šūf is šāyef and, in Maghrebine dialects, the participle "lodging" of bāt is bāyit. Because cuneiform script lacks specific signs with semivowels, spellings like sa-i-im, sa-imu, "fixing" (root śīm), are ambiguous. Assyriologists explain them usually as šā'imu, but occasional Standard Babylonian forms as da-a-aik /dāyik/, "killing", seem to indicate that one should always read šāyim, etc. In Hebrew, the forms qām, "standing", met, "dead", imply the monophthongization of the secondary diphthongs: *qāyim > qām, *māyit > met. In modern Ethiopian languages, w can be used in medial position as a transitional consonant between two vowels, e.g. duwa from Arabic du'ā', "Muslim prayer". 22.16. There is a series of nominal patterns extended by a diphthong, like fay 'al, faw 'al, fay 'āl, fay 'ūl, fu 'ayI, fu' 'ayI, fi' 'awl, known not only from Arabic but also from other Semitic languages. In particular, the patterns faw 'al "and fu 'ayl — the latter used for diminutives — are attested also in Aramaic (e.g. 'lym, "lad") (§29.10). The monosyllabic patterns fayl má fawl alternate sometimes with noun types ClC and CūC (§29.9), and a possible example of a fay I noun paralleled by a CāC type is provided by the word bayt, "house", apparently related to Cushitic bati, "roof" (Oromo), borrowed in Gafat with the same meaning. 22.17. Verbal Stem I I I with lengthened first vowel — attested in Ara­ bic, Ethiopic, and Syriac (§41.5) — kan give rise to a secondary diph­ thong developed from the long vowel. This is perhaps non evident when comparing, e.g., Classical Arabic ġawġa'a or ġawġā, "to cluck" (of hens), with Syro-Palestinian colloquial gāga, because the verb derives from an onomatopoeia, but colloquial horab, "to strike up a war song", is best explained as *hawrab < hārab, "to wage war", since aw > 6 is the normal reduction in Syro-Palestinian colloquials. The existence of a

GEMINATED OR LONG CONSONANTS

173

fā 'ala > faw 'ala I fay 'ala stem in Ethiopic is implied by forms of the types qotala and qetala. As for Syriac, e.g., the form gawzel, "he set fire on", is best interpreted as a Stem I I I ā > aw of the root gzl, "to plun­ der". These developments are important for a right understanding of the secondary stems of the monosyllabic verbs with a long vowel (§44.5-9). 22.18. Another type of secondary diphthongs can explain the forma­ tion of some secondary verbal roots with "weak" first radicals, e.g. wld I yld, "to bear" (§43.6-7). Thus in Ethiopian Gurage dialects, e.g., the prepalatal y can be an on-glide before the vowels à, e, 9, i, e.g. ydrbat and àrbat, "evening meal". In the same way, labial w can serve in South Ethiopic as an on-glide in initial position before o, u, e.g. wof and of, "bird". This phenomenon cannot be equated with the appearance of a diphthong at the beginning of a Berber noun to which the ergative w-prefix is added, e.g. wagmar < u+agmar, "horse"; yildf < u-ibf-, "wild boar".

14. GEMINATED OR L O N G CONSONANTS

23.1. Gemination or consonantal length can be justified etymologically or grammatically, but it occurs also when a long vowel plus a single consonant is replaced by a short vowel plus a doubled consonant, as in Hebrew gdmallim, "camels", "dromedaries", plural of gāmāl (§24.7). Some Semitic languages and dialects are non-geminating in part or in general (§23.5). A compensatory lengthening of the contiguous vowel may then correspond to the gemination, as in Neo-Aramaic dābāšā, "bee", instead of dabbāšā. Gemination is phonemic in the Semitic lan­ guages in which gemination or lengthening of consonants is a regular feature, as it appears, e.g., from Arabic kabara, "to become great", and kabbara, "to make great", or North Ethiopic (Tigrinya) qátànà, "to be small", and qàttānà, "to liquefy", 'abay, "wild", and 'abbay, "Blue Nile", and South Ethiopic (Gurage) abar, "dry season", and abbar, "young man", where gemination and non-gemination of b and t consti­ tute the sole phonemic difference between the two words. It has been suggested that there may have been a phonetic difference in Semitic between long consonants and double or geminated consonants. In fact, there is a category of "continuant" consonants that can be held continuously, with variable tension but without changing quality, and a second category of socalled "kinetic" or "interrupted" sounds that cannot be so held. The first group

174

PHONOLOGY

comprises the nasal, lateral, fricative, and rolled phonemes, while the second one includes the plosives and the affricates (e.g. [ts]). The gemination of the phonemes of the second group does not imply length, properly speaking, but increased tension which is perceivable in the case of a voiceless plosive, while a voiced one is reckoned less tense since a considerable part of the air it uses is consumed by voicing alone. Therefore, really geminated voiced plosives have to be pronounced either by doubly stopping the chamber of the mouth and sucking in the breath, or by changing the quality, as /bb/ > [mb] or [bb], [dd] > [nd] or [dd], /gg/ > [ng] or [gg]. The first articulation is encountered, e.g., among native Tūroyo speakers and among speakers of Western Neo-Aramaic who even insert an anaptyctic vowel between the geminated consonants; e.g. amehl < amell, "he said to them" (Ġubb 'Adīn). Concrete examples of the second pronounciation in ancient Semitic languages are probably provided by such transcriptions as Eercipcbpa for Sìpporā, 'AK^W for 'Akko, MaxGaGias for Mattityā, which aptly illustrate the changing quality of geminated plosives. In other circum­ stances or forms of speech, and especially in the articulation of "continuants", the so-called "doubling" of a consonant does not consist phonetically in its double articulation, but either in its lengthening or in its amplification. This may vary from a slight "tightening" or lengthening in time to much more than dou­ ble. We keep nevertheless using the traditional terminology and the current notation of consonantal length or tension by transcribing the long or tense con­ sonant twice, e.g. bb. This notation is interchangeable with the symbol /b:/ employed in the international phonetic system and with the capital letter B adopted by some authors. 23.2. Gemination is sometimes hardly audible, particularly at the end of a word (§24.5), where it is not recorded either in Amharic or in Hebrew, e.g. 'am, "people", instead of 'amm. However, it becomes evi­ dent when the final consonant is followed by a vowel, e.g. Hebrew 'amml, "my people". Gemination is at times missing also in the middle of a word, as shown by the Masoretic notation mdbaqdšīm (Ex. 4,19; 10,11), "seeking", instead of the expected *mdbaqqdšlm. Besides, there is no regular marking of long consonants in cuneiform script and there is no such notation at all in Semitic alphabetic scripts, except in some rare cases (§23.3), until the introduction of special diacritics in Hebrew and in Arabic (§23.4). 23.3. Early essays aiming at indicating a geminated or long consonant are found, e.g., in the Hebrew Bible, where the variant spelling hrry of hry must express the plural construct state *harre, "mountains". In Lit­ erary and Official Aramaic, spellings like 'mm' for 'ammā', "the peo­ ple", or dššn and dššy' for the plural of the noun dašš(ā), "door", should be explained in the same way. Besides, it is very likely that at least the liquids and the nasal n, when geminated, were sometimes written in

GEMINATED OR LONG CONSONANTS

175

Safaitic with a double / and a double n, e.g. kllhm = klhm, "all of them", tnn'l = tn'l, Tavvr|À,os, "God has considered" or the like. 23.4. In the Hebrew vocalization systems, the symbol called dageš — generally a dot placed in the letter — is used to mark the gemination of a consonant, but it is in reality an ambiguous sign, since it can also indicate the lack of gemination and the plosive pronunciation of the con­ sonants b g dkp t. This was probably the original function of the dageš used with the plosives, since these phonemes cannot be lengthened, properly speaking, but only amplified by other means, as a pronuncia­ tion with greater pressure. Only Arabic sadda (*) indicates in an unam­ biguous way that the consonant is long or geminated, e.g. 'ammu, "paternal uncle". 23.5. In principle, all the consonants can be geminated, but ' and h are not geminated in Ethiopian languages and the Masoretic punctuation of Hebrew and of Biblical Aramaic in principle excludes the gemination of the pharyngals (h,'), of the laryngals (', h), and of r. In Neo-Aramaic, the doubling of consonants has largely been eliminated and replaced by the lengthening of the preceding vowel, e.g. yāma < yammā, "sea", but a secondary gemination can oppose a word to its counterpart characterized by a long vowel followed by the single consonant, e.g. /mīta/, "dead" (masculine), and /mitta/, "dead" (feminine). There are also non-gemi­ nating dialects in the modern Ethiopian languages, although gemination through assimilation occurs in these dialects as well, e.g. wàsse < wàsfe, "awl", in a non-geminating Gurage dialect. 23.6. Assimilation and resulting gemination occur in all the Semitic languages and will be examined in the appropriate paragraphs dealing with conditioned sound changes (§27.3-7). Instead, dissimilation of gem­ ination is a common Semitic phenomenon which is not conditioned by any particular phonetic environment. It amounts to a phonemic split or diphonemization, if the resulting sounds become significant (cf. §10.7,12), as in Neo-Arabic where the dissimilation may serve as means to distinguish verbal Stems I and I I (e.g. ġarmaš < ġammaš, "to scratch", vs. ġamas, in Lebanon). A geminated consonant can be dissimilated through n and m, through the liquids / and r, sometimes through ' and y. It should be stressed that the dissimilation of geminated plosives, especially when they are voiced, proceeds from the nature of these phonemes that cannot be lengthened, properly speaking, without

176

PHONOLOGY

changing quality. Thus there arise equations like lb: I = [mb] or [lb]; /dd/ = [nd], [md], or [rd]; /tt/ = [nt], etc. The dissimilation through ' (§23.10) belongs to the same phenomenon, since the p t k series is "geminated" in some languages by spirantization or glottalization. 23.7. Dissimilation through n occurs in Palaeosyrian, e.g. si-na-ba-ti (gen.) < *sibbātu, "sunbeams", from a variant root sbb of Hebrew sby, "splendour", with a plural siba'ot. It is attested in Old Akkadian, e.g. by Ha-an-za-ab-tum as compared with Ha-za-ab-tum /Hassabtum/, a per­ sonal name derived from the root hsb, "to break off". In Amorite, one can mention the names An-du-ma-lik - Ad-du-ma-lik, "Haddu is king", and Samaš-ha-an-zi-ir - Samaš-ha-zi-ir /Samaš-ġazzīr/, "Shamash is a hero". Besides the frequent Babylonian form inandin < inaddin, "he gives", one can refer to ta-na-an-zi-iq = ta-na-az-zi-iq /tanazziq/, "you are angry". The geminated consonants of both verbs can be dissimilated also through m: tanaddina > ta-nam-di-na, "you give me", anazziq > anam-ziq, " I am angry". Dissimilation through m is attested also in Assyro-Babylonian sumbu < subbu, "wagon" (cf. Hebrew sabblm\ Aramaic sabbā). The name of the Palestinian city Eqron /'Aqqaron/ is dissimilated in Neo-Assyrian texts in An-qar-u-na and Am-qar-u-na, and the name of the Aramaic tribe Gabbūl, "kneader", appears as Ga-ambu-lu, etc. The Aramaic personal name hdy /Haddîy/, "rejoicing", is transcribed Ha-an-di-i in Neo-Assyrian and Ha-an-di-ia in Neo-Babylonian texts. The noun kkr < krkr (§27.3), "talent" (weight), may be dis­ similated in Aramaic in knkr, with parrallel Coptic forms ġinġor, ġinġor, and Greek KÍvxotpss. The geminated t of the Ammonite name htš /Hattaš/, also attested in Safaitic and in Nabataean with a Greek translit­ eration XoiTTeaos, is dissimilated in Neo-Assyrian texts in Ha-an-ta-si and in Neo-Babylonian texts in Ha-an-ta-šú. In Ethiopian Gurage dialects, e.g. goġġo, "hut", is dissimilated into gonzo.
á à

23.8. Dissimilation through / is attested in Hebrew by pltyš in the Isaiah scroll l Q I s for pattīš, "forge hammer". One can also quote galmūd < *gammūd, "sterile", from the root gmd, "to be hard"; zal'āpot < *za"āpdt, "deadliness", from the root z'p, "to kill instantly"; hlmš = Assyro-Babylonian elmešu < *hammīš, a precious stone, from the root hmš, "to be steadfast". The Assyro-Babylonian name of the reptile hulmittu corresponds to Hebrew hornet and is obvi­ ously dissimilated from *hummittu. In Arabic, there are several verbs without and with dissimilatory / in stem I I , e.g.fattaha and faltaha, "to
a

GEMINATED OR LONG CONSONANTS

177

make broad"; habbasa and halbasa, "to mix"; kahhaba and kalhaba, "to strike"; dammasa and dalmasa, "to hide"; etc. The North Arabian name *Faddās is dissimilated in Nabataean in pndšw, but it appears in the Hebrew Bible as pldš. In Amharic and in Gurage dialects, one can mention salsa, "sixty", dissimilated from sassa < sadsa.
1

23.9. Dissimilation through r is attested in Old Babylonian by la-marsú-[u]m < lamassum, "guardian she-angel" (ARM VI,31,19), in Ara­ maic by kursi' < kussi'u, "throne", by the name of Damascus: Dammeśeq > Darmeśeq, and by šarbīt < Babylonian šabbitu, "staff, sceptre", borrowed further by Hebrew (Esth. 4,11; 5,2; 8,4; Sir. 37,17). In Biblical Hebrew, the verbal form yakarsamennāh (Ps. 80,14) derives through dissimilation ss > rs from yakassamennāh, "it gnaws i t " , and the participle makurbāl (I Chr. 15,27), "wrapped", is dissimilated through bb > rb from makubbāl, attested in Mishnaic Hebrew. In Ethiopian Gurage dialects, the dissimilation through r is frequent, e.g.: gard < gadd, "misery"; korda < koddà, "water bottle"; irda < iddà, "carding bow", etc. Rare examples of progressive dissimilation through r are attested in Hebrew with dibrē < dibbē (Assyro-Babylonian), "words", followed by the denominative verb dibber, yadabber, "to speak", and other derivatives, and in Arabic with batta, "to cut off", dissimilated in batara or in bat{t)ala, with / (§17.5). 23.10. The glottal stop serves for the disjunction of a geminated con­ sonant in some West Gurage dialects, e.g. gum'a vs. gumma, "club"; gun'àr vs. gunnàrjn, "head". This feature can be interpreted as dissimi­ lation of gemination. A similar phenomenon may perhaps be observed in Middle Assyrian and in Neo-Assyrian, in words like bi-'-ti < bittu < bintu, "daughter"; pe-'-ta < pettu < pēntu, "charcoal"; sa-'-te < *sattu < san/mtu, "morning dawn", although the change n > ' is well attested in these dialects when n appears in intervocalic position (§17.2). How­ ever, this is not the case in the present examples, since the spellings bittu (bi-it-ta-, bi-it-tá) and pettu (pé-et-tum) are likewise attested in cuneiform texts. 23.11. The semivowel y serves, e.g., for the disjunction of a geminated liquid in the Arabic verbal form tayla' < talla', "he brought up", attested in Lebanon.

Assuming that every syllable begins .g. probably "asylum". Such elisions are not exceptional in the Old Babylonian of Mari (e. CCvC). 24. SYLLABLE 24. thus mas-. which is followed in its turn by a consonant (CvC. 2° a closed syllable consisting of a conso­ nant or a consonant cluster followed by a vowel. Arabic 'aġāra.g. as masa-. is contrary to the old scribal practice. short or long.g. e. "enter" a sanctuary). by a "weak" / (§17. may represent one closed syllable. with an assimilation Im > mm.(§29. The orthographical ambiguity of syllabic cuneiform has a bearing on syllabic structure. and constituting a word or part of a word. CCv) or long (Cv.2). e.1. at least in conse­ quence of the phonetic and morpho-phonemic evolution of the lan­ guages. CvC. or *kam-ma-tum.with a consonant. ilkamma for illikamma. the first member of which is often a liquid (CvCC). Such cases occur instead in Late Babylonian (§21. well-known from Old Akkadian. which is followed either by a long or geminated consonant or by a two-conso­ nant cluster. Sumerian KU ("to enter") = ma-sa-gàr-tù-um or mas-gàr-tum Jmašagārtum/. especially in Palaeosyrian and in Old Akkadian. 4 . Instead. On the other hand. Thus. "grant asylum") of the verb gūr (cf. However. a nominal derivative in ma. "cause to enter" a sanctuary.11). a word can also begin with a vowel or with a two-member clus­ ter of consonants (§17.2.21) from the causative *šagār (cf. "louse". one orthographically closed syllable may stand for two open syllables and represent a value of the type CvCv.*or by a geminated consonant. Sabaic gr. 3° a dou­ bly closed syllable consisting of a consonant followed by a vowel. the often assumed hypothesis that two orthographically open syllables of the Ebla texts. unless one assumes that a short unstressed vowel was elided. One might even invert the reasoning and conclude from spoken languages that the above-mentioned phonological principle of classical literary languages results from standardizing ten­ dencies which aim at committing speech to writing.178 PHONOLOGY 15.9). one can distinguish three types of syllables in Semitic: 1° an open syllable con­ sisting of a consonant or a consonant cluster followed by a vowel. short (Cv. CCv). CCvC. and it may lead to an incorrect interpretation of variant spellings. Authors generally assume that every Semitic syllable originally began with one consonant and one only. "par­ asite".5) and they occur also at Ebla. Palaeosyrian Da-mu stands for the divine name *Da'-mu and kà-ma-tum represents either kal-ma-tum. The syllable is a sound or combination of sounds uttered together or at a single impulse of the voice. §25. an orthographically open syllable may represent a syllable actually closed by a guttural. "he came here".

dā/llun). In Arabic. This phenomenon leads to the elimination or restriction of syllables of the type CvC. 24. e. qam < qām. CvCC: e. long or ultra-long: 1° a syllable is short when it ends in a short vowel (Cv: e. yátan > yatan > yaton. "straying.g.g. "he gave".g. e. and in verbs with a sec­ ond long or geminated radical (e. the accent shift to the final syllable occasioned its lengthening with the eventual change ā > o. "dog"). zaliltu).g. zaltu or ziltu < *zall-tu. Short vowels tend to become long in open and in stressed sylla­ bles. It is difficult to perceive this phenomenon in a correct way in lan­ guages written in cuneiform syllabograms (§21. "descendants"). nāzilūn < nāzilūna. Hebrew qāra' > qārā. Also long or geminated consonants show a tendency to become short.. in fa-qat < *fa-qatt. qām. especially at the end of a syllable (§23. a syllable may be short. qurdr < qurr. "in"). lā.5.2). The vowels are always short in a closed unstressed syllable and long vowels show a tendency to become short when their syllable closes. there are excep­ tions in pausal forms of ^Classical Arabic. 3° a syllable is ultra-long.6. 2° a syllable is long when it ends either in a long vowel or in a consonant fol­ lowing a short vowel (Cv: e. or in a two-consonant clus­ ter (CvC: e. "only". 24.g.g. when it ends either in a consonant following a long vowel. or in a geminated or long consonant. bi-.g.g. the shortening of long vowels in closed syllables became a general rule. In Arabic.g. " I became"). but opens the way to syllables of the type CCvC.3.g. "people".SYLLABLE 179 24.g. CvC: e. e. while modem Ethiopian dialects can avoid it by splitting the long or geminated consonant by means of an anaptyctic vowel (e. "paternal uncle". "not". Quantitatively. 'am < 'amm. but this is the case in certain forms of West Semitic verbs with last radical ' when the latter loses its consonantal value. this shorten­ ing appears. 24.g. with a plural 'ammīm). when the final vowel is dropped in pronunciation (e. In Phoenician.5).g. e. However. 'amm. dallūn > dā/lun. min. erroneous") or be pronounced long in the beginning of the following syllable (e.4. "he stood up". Ara­ bic nabbā < nabba'(a). kalb. "basket" in Gurage).g. "he called". "he stood up". . "from").g. unless the long consonant is split by an anaptyctic vowel (e. "he announced".g. and in the case of long consonants that can either be shortened (e. This shortening is a gen­ eral feature in Hebrew at the end of a word (e.

.9.17). Instead. kâlbi. to Masoretic Hebrew štayirn and to Syriac štā. with or without an ultra­ short vowel d. 'dzni. There are also some cases of consonant doubling after a short open syllable (§23. This tendency is absent from most modern colloquials and its partial absence may be traced back at least to the Late Antiquity and early Middle Ages. or in Aramaic 'attānā < 'atānā. In fact. nabdr.g. "foot". A vowel can also be added at the end of a word. "was made". or the noun kràmt. one can point. e. "camels". "he opened" (§27.g. or apa.1). e.. to the transliteration KO5CF of Hebrew qds. In both cases. besides the two-consonant clusters formed with a liquid. "touch! *\firri orfirra. There is a wide tendency in classical Semitic languages to elim­ inate two-consonant clusters at the beginning or at the end of a word by adding a supplementary vowel either between the two consonants or at the beginning. while it is inserted in Hebrew between the prefix n.g. " I wrote" in urban Maghrebine Arabic. Neo-Aramaic ebra.g. In all these cases.g. the fluctuating pronunciations of Hebrew sdwa (§21. "she-ass". the addition of the vowel results in a new syllable 'in/fa'ala or nif/'al. "leop­ ard". has an anap­ tyctic vowel u splitting the geminated consonant. e. The same phonetic device is used in Tigrinya with nouns ending in a cluster of consonants. e. "ear". the imperativeptuh. some modern Arabic dialects and Western Ara­ maic show a clear tendency to introduce prosthetic vowels in such cases. 24. "palm trees". .g. "speak!". e. at least announce the modern colloquials with. Although the palato-alveolar š may conceal some phonetic affinities with liquids in these particular cases. This results in a change of the nature of the syllable in question which becomes closed and long.19) and of Syriac words. in Origen's Hexapla. the perfect ktdbt. one can refer to the Hebrew verbal form nif'al.g.5).7. like nahl. "dog". The Assyro-Babylonian imperative dubub. iftah.8. However.and the first radical of the verb. differing from the corresponding Arabic form 'infa'ala by the place of the supplementary vowel i which is added in Arabic at the beginning of the word. "run away!". agar. in the Hebrew plural gdmalllm < *g9mallm. "earth" (§17. the addition of a vowel results in the appearance of a new syllable. e. 24. Amharic often breaks final clus­ ters by inserting an anaptyctic vowel d\ e.180 PHONOLOGY 24. "two". "open!" in Eastern Neo-Aramaic. "rainy season" in Amharic. "son". to the imperative of the verbs with second long or geminated radical: massi or massa.9). Beside the anaptyctic vowels of qurdr and zaliltu (§24. both in Arabic. "holiness". respectively at the end of the word.

with the reduction of the perfect to one syllable.10. "servant of the king".WORD ACCENT 181 24. "ten". an expiratory or stress accent exists in Hebrew. "vow". after the loss of the short unstressed syllable. in certain Arabic colloquials such as the Syrian. The opposite. A similar evolution can be observed in verbal forms. a semivowel or the glottal stop is inserted in order to avoid the hiatus. above all i f there is a stress accent with a phonemic status (§25.[ba'al] < *ba'l. báraqa > bráq. y. WORD ACCENT 25. Early Arab grammarians did not deal with the subject and little can be said about "written" languages. "tomb". Orthography rarely indicates tone and stress. Now. this phase was preceded by a dissyllabic realization of the nouns under consideration. A parallel phenomenon is attested in the Neo-Punic of NorthAfrica thanks to the matres lectionis marking the stressed syllable in words like nd'r [ndár] < *nidr. "lord". 16. these elements are of the highest importance. unless there is elision or contraction (§22.15). but this does not lead to a radical elimination of original short vowels. sk'r [skdr] < *sikr. the loss of ancient short vowels in open syllables and the shift of the accent from the stressed syllable of Classi­ cal Arabic to the following one completely changed the syllabic struc­ ture of the language. it later became qamáh and finally qmdh. Yet. although the traditional pronunciation of Ge'ez seems to show some principles for the stress of isolated forms and words. Af3e5peXe%e (vocative) ['Abed-melek] < *'Abd-milk.8). Maap.g. Aramaic. as shown by Late Phoenician |3aaX.11. w introduce a new syllable.g. l 2 3 x 2 y 24.[Mahar-] < Mahr-. "memory". In Arabic colloquials. Also the Ethiopian languages make a fairly extensive use of prosthetic and anaptyctic vowels. Like in Arabic. The pattern observable in nouns and verbs can be schematized as C vC C > C C vC E. tonal marks were added only in Alexandrian times and Hebrew accentuation was only introduced by the Masoretes in early Middle Ages. " i t lightened". and in certain Gurage dialects. nonexpiratory type of accent can be heard in Amharic and in Ethiopic in general. although some variant spellings may reveal the . "wheat". e. classical qamh(uri). 's'r ['sár] < *'asr. "courier". The inserted '. was pronounced qamah in Pre-Classical and Middle Arabic. qb'r [qbar] < *qabr. In Greek. When vowels meet.1. in Semitic.

e.g.182 PHONOLOGY impact of the stressed syllable or a shift in the position of the accent. " I was wounded". Assyro-Babylonian šanu < *šaniyu. "second". "per­ fume".3) is considered as belonging to the following word. i f there is no such. Besides. 25. ktebna. the stress falls then on the penult syllable which can be long or short. the position of the accent on the penult syllable is limited to the cases when the latter is long. "you killed". raqabatani. 25. e. katábū. in Maghrebine Arabic: ríha. "quest of knowledge". even i f it is long. Generally.g. sàbára.g. makātíb. in Arabic of the Hawrān: katabuha. However.2. "letters". e. in Maghrebine Arabic. in Syriac: kétbat. "he killed". Common Semitic or Proto-Semitic principles in this matter can only be highly hypothetical. "they wrote i t " . Concerning the 3rd person of the perfect in Ge'ez. "he is". tálaba/tu l-'īlmi. but qátala. it comes to rest either on the first syllable of the word or on the antepenult. Considering modern colloquials from various regions. "he said". e. "they wrote". a 25. the word accent falls on the final syllable when the latter is ultra-long. The penult syllable is systematically stressed in the Samaritan pronunciation of Hebrew and Aramaic. thus. in the Arabic of the Hawrān.3. swdm /Sādom/ in lQIs 1. It was assumed that the position of the word accent in East Semitic followed analogous principles.4. " i n the beginning God created heaven and earth" (Gen. contrary to the Jewish traditional pronunciation of Hebrew and of Aramaic. The monosyllabic proclitic particles bear no stress. e.1). against Masoretic Sddom but in agreement with Greek Eo5opa. This accentuation seems to be confirmed by the matres lectionis of the Dead Sea scrolls. Otherwise stress recedes until it meets a long syllable and. "slave". the word accent does not fall on the ultima. néhwē. thus qatálta. thus nagára. barášit bára 'àlúwem 'it eššámem wit áres.g. "we wrote". the final syllable of a noun in the construct state (§33.g. e. but such a reconstruction is .10. "he broke". In other cases. in the Egyptian tradition of tfre Qur'ān reading. 25.5. nêġráht. The word accent may perhaps fall on the final syllable also when the latter is the result of contraction. "you observed me". but raqábatun or ráqabatun. there is gen­ eral agreement that the penult syllable is stressed.g. the same situation is found in the South Ethiopian Harari. 1. e.g. "she wrote".

but on the penult syllable of the nouns. though it falls usually on the final syllable of a word. še-e-mi-šu /šémišu/. It is also difficult to determine where stress falls in Tigrinya. which was tone-lengthened with the consequent change ā > 6 (e. "he washed himself". e. In Tigre. . that appears to have had a strong stress accent. Thus. Also Phoenician.7. may be judged from the apparent reduction of short unaccented vowels in the penult syllable (e. "king". usually accentuates the final syllable of the word. the accentuation of the final syllable is dominant in the Jewish traditional pronunciation of Hebrew. the stress falls also in Mod­ em South Arabian languages on the final syllable wherever there is no penult long syllable in the form under consideration.g. However. but the accent falls on the penultimate in some particular cases. But no systematic attempt was made until now to establish whether Assyro-Babylonian had an expiratory accent. However. giîlát.g. the accent falls on the last syllable of the verbs. Although it is difficult to generalize. and not that of Old Babylonian. are best explained by assuming stress on the first syllable. 25. il-ku for illiku. e. as can be learned from Hebrew grammars. "beloved"). This tendency is even stronger in the Arabic dialects of the area.g. the reduction and complete elision of some vowels. instead.WORD ACCENT 183 based mainly on secondary deductions. Ie8ou8 [Yddud] < yadud. "he gave"). "beam". qá-a-ab-la-at ta-am-ti Iqáblat tāmtil. "they went". In some West Gurage dialects. The situation in some languages is the result of complicated developments.. especially in "segolate" forms like melek. "who".6. yaton < yatan < yátan.g. for which no clear phonological rules can be determi­ nated to-day.g. plene writing occurs in accented syllables and does not necessarily mark phonemic length. te-se-em-ma for tešémmema. however. "she killed". and its strong stress character. stress is regularly on the penultimate and all the vowels are phonologically short. but it is conditioned by the general sentence stress or pitch. damiqtu / damqatu. Babylonian by-forms like litmudu / litamdu. "good" (feminine). which would have been a force in the word capable of bringing about. Such elisions frequently occur in Mari letters (e. e. tā'àbá. The place of the accent. In Soqotri. 25. Middle Babylonian ma-a-ni Imánnil. "you will hear"). stress is non-distinctive and shifts eas­ ily from one syllable to the other. but waka. "he understands". "the middle of the sea". as in Dofar habdl.g. but they very likely reveal the stress accent of the Amorite linguistic substratum of the writers. "his hearing". "preg­ nancy".

Inves­ tigations into Semitic phonology and syntax must include intonation and pitch. "they captured". and interrogative sentences generally differ merely in intonation.1. which falls on the last syllable of the verb. "may he k i l l ! " (§38. but on the penultimate syllable of the noun. one can assume that the word stress may have had a distinctive or phonemic status in ProtoSemitic. e. "he made".g. in other words: whether there are tonemes in Semitic. Instead. and qūmí. in Arabic and in AssyroBabylonian. dekkd. the stress is a distinguishing feature between certain pairs of words. It gives expression to sentences and conveys shades of meaning which cannot conveniently be expressed by other means. e. In Masoretic Hebrew accentuation has a phonemic value in some grammatical constructs rather than in lexical items. and dékkâ. also lexical items can be contrastive in stress in West Gurage dialects: some verbs can be distin­ guished from certain nouns only through the place of the stress. "beam". ēnza. ba'ā. nášele.g. "he seized". qumī.g. "my getting up" (infinitive). wàká. "he killed". "she came". e. though much important material can be adduced from spoken and even . though sometimes — particularly in written language — a question may be introduced by a word indicating that the sentence is a question or the order of words in the sentence may differ from the ordinary one. eppa. Considering the existence of archaic features in South Ethiopic. "get up!" (feminine). and bānu. and énzâ. A similar situation occurs in spoken Neo-Aramaic. S E N T E N C E S T R E S S OR P I T C H 26.184 PHONOLOGY 25. and wâka. "they returned". "long".g. "he pierced".2). "she is coming". and našéle. Thus. but so far no serious work has been done on the subject. 17. declarative.8. and of the jussive yaqtúl I yiqtúl. Sentence stress or pitch affects the meaning of whole sentences. in the case of the preterite yáqtul / yíqtul. "they built". A n important question is whether word stress or tonic accent has or has not a phonemic status. In Semitic languages. e. "among us". "he is their man". but it is not an integral part of any word. and šābu. "he is long". "thatch" (grass to cover the houses). even i f authors often assume that word stress is not phonemic in several Semitic languages. But we are moving here on ground that has not been satis­ factorily investigated. and bā'a. and éppá. "he is a man". since the word accent seems to be the only distinguishing feature between this pair of verbal forms.g. banū. šabū. exclamatory. e. "toga".

vocalic changes (e. "young [camell"). case endings. "magistrate". byúktob for byúktub. or pausal consonants may be pre-glottalized and devoiced (e. etc. Thus. . the different speaker's commitment is discernible mainly from a higher pausal pitch. In an interrogative sentence.g. and feminine noun endings are dropped (e.g. In Syro-Palestinian and Egyptian dialects. "he will write").g. This may be expressed in writing by the additional final sign -ú in is-li-mu-ú. In Yemen. 'al-bakrun > 'al-bakur. "did they make peace?". This common and spontaneous phenomenon has a repercussion on the traditional recitation of the Qur'ān and of the Hebrew Bible.g. final short vowels. with a con­ sequent anaptyxis in monosyllabic roots (e. 26.g. the last syllable is generally lower in pitch than the penultimate. in an Old Babylonian letter from Mari suggesting a pitch islimu instead of the normal íslimū (§ 10. on the other hand. as shown by the "pausal" forms used in Classical Arabic for the vocative (e. vowel lengthenings (e. "oh! friend!"). 26. especially from letters.3.g.g. "bitch"). 26.4. In other words. stress retrocedes to the penult syllable (e.. in Arabic. In a declarative sentence.g. kalbi for kalba. "he wrote". yā 'ammá. The exclamatory intonation implies a high pitch and a conse­ quent shortening of the words.g. yimšow for yimšū. malikun > malik. and gives rise to the so-called "pausal" forms. "they will go"). e. qādīn > qādī. may occur. "chil­ dren"). diphthongizations (e. This is confirmed by the apocopated forms of the jussive in various Semitic languages (§39. but no accent shift has been registered. "he went"). malikatun > malikah.6). "earth"). "king". [awlā't] for awlād. "oh! uncle!") and even by the occasional dropping of the final radical of the word (e. particularly toward the end when the latter corresponds to the point of greatest sig­ nificance in speech.g. We must therefore limit ourselves to some general observations on the falling and rising pitch patterns.SENTENCE STRESS OR PITCH 185 "written" languages. In Masoretic Hebrew. máyim > mayim.. either a nasalization may affect vowels standing in the pausal position (e. yā sāhi < sāhibu. in any case. 'ānokí > 'ānokī) and a short accented vowel is lengthened or changes its quality (e.14-18). "queen"). 'eres > 'ares. kataba > katab. [wallf] for walli. Pausal phenomena occur also in modern Arabic dialects.2. the last syllable must be heard as being on a higher pitch than the penultimate. "water".g. a question must be said with a rising tone.g.

g. between vowels. anaptyxis. though the interval between accents. etc.186 PHONOLOGY 26. which appears to be conditioned by proclitics. haplology. is essentially governed by semantic considera­ tions. while elision. The natural tendency of the speaker is to limit effort in his speech and to avoid sharp shifts in the use of speech organs.. contiguous and distant. Only when this assimilation is particularly sharp is the change felt. to a phone­ mic merger or shift. a sequence of similar sounds demands a greater effort from the speaker and he tends to dissimilate them. subjoining words. aim at facilitating or simplifying the emission of speech sounds in various ways.2. Assimilation 27.... Arabic kállimī rasúla llahi yukállim..5.7. enclitics. but only one of them is the main sentence stress. "tell that they should forgive". §10. logical hypotaxis seems to be indicated by the stress falling not on the verb of the main clause. and reciprocal assimilation. This leads to a chain of assimilations of one sound to another. Assimilation may take place between consonants. and between a consonant and a vowel. "speak to God's envoy he should speak. In particular.. whether primary or secondary. At other times. CONDITIONED SOUND C H A N G E S 27. . 12). sentence stress has a syntactical function. Assimilation is the main type of conditioned sound changes. prosthesis. A. There is partial and total assimilation." In other words. It amounts to a monophonemization. Hypercorrection is instead an intentional but erroneous correction of the spelling or of the pronunciation of a word. like the order of words in the sentence. Arabic qui yáġfirū. e. especially when there is formal asyndetic parataxis ( § 5 5 ) . The stress of conjoined clauses appears to be influenced in the main by the meaning of the sentence as a whole and. A sentence of any length has several accents. e. Another type of change is metathesis. is scarcely more than two or three sylla­ bles.1. progres­ sive and more often regressive.g. The place of the secondary accents is based princi­ pally on the word accent. i f the resulting sound becomes distinctive and sig­ nificant (cf. 18. but on the beginning of the logically dependent clause.

in Ge'ez inscriptions 'dnza > 'dzza.g. "he gave". in Amoriteyansib > yassib. Andalusian Arabic 'anta > att. In Thamūdic. "since". we find *s tnsr > s tsr. In Sabaic. in Gafat samdt / samdttâ vs.g. 't /'atta/ < 'anta. " l i p " . "week".g. in Phoenician and NorthIsraelite Hebrew *šnt > št /šatt/. North Arabian mundū > muddū. "while".g. "you would see".. "you". kīn tarā > kittarā. mg't < manga'+t. "year". the common type being here total regressive assimilation. in Gurage kànfàr > kàfâr.ASSIMILATION 187 a) Assimilation between Consonants 27. 'tt /'attat/ < 'antat. The main types of Semitic consonantal assimilation are the following: bk > kk bt > pt dn > nn dš > ss dt > tt dt > dd ~kr > kk Id > dd Ik > kk In > nn Iq > qq lr>ll Is > ŠŠ It > ss mb > bb nb > bb/mb nd > dd nd > dd nf>ff ng>gg nh > nn nk > kk nl > 11 nm > mm np > pp ns > ss nt > tt nt > tt nz > zz qt > qt rd > dd rk > kk rl>ll rn > nn rs > ss rz > zz sf> ss ?f> ?? st > ss/st td > dd th > tt tk > kk tn > nn tš > ŠŠ tt > tt tz > zz tt > tt zt > st a) Thus. but in agreement with cer­ tain colloquials where forms like bitt < bint occur occasionally. e. in Lihyānite. vowelless n assimilates regularly to a following consonant: e. but also in l l . "woman") contrary to Classical Arabic. Assimilation between consonants takes most often place between a liquid /. with the auxiliary verb kin (classical kāna) used to express an eventuality. b) Vowelless / assimilates to various consonants. 'al-sams > [aššams]). r or the nasal n and another consonant. "feed-place". "he raised". The occasional or dialectal loss of a vowel may also lead to the regressive assimilation of n.3. perhaps better termed "anticipatory".(e. in Old Akkadian and in Assyro-Babylonian *indin > iddin. "you". and in Safaitic. most prominently in the case of the Arabic article 'al. the n is sometimes assimilated to a following consonant (e. since the vocal organs "anticipate" the position of the next sound. Amharic samant. as in i(3cop0 transcribing Late Babylonian ina būrti. "he summoned to his support". "through the cistern".

"he imprisoned".g. rests upon the ground "like a cake of bread" (E. e) Transcriptions of proper names in other scripts allow us to distin­ guish assimilations from elisions. " a w l " . "Handmaid of Samaš". Total progressive assimilation frequently occurs in verbal forms with infix t (§41. "talent". the divine name 'Attar > 'Attar. e. 27. "he sought". in colloquial Arabic (e. It is attested in Hebrew with the pronominal suf­ fix -hū. in Assyro-Babylonian *(w)ālidtu > (w)ālittu. e. "morning dew").g. Egyptian Demotic krkr.g. gurda > gudda. "speaking to himself". "the one who bore". Instead. e. halla 'al-ka > [hallakka].g. 27.W. " o f the two horns (of the moon)". "round loaf". in Arabic *ittalaba > 'ittalaba. in Aramaic. There are also other cases. in Classical Arabic *idtakara .25).g. "half". "Servant of Nerigal". "she weaned him". "he shall capture".g. the assimilation of a sibilant to / is exceptional. "he didn't kill us").5. man la-bet > [manna bet]. *yilkadenhū > yilkadennū. "because of"). results from the assimilation tš > šš. as in Palaeosyrian su-lu-la-a. Lane). bârzaz > bâzāz. in the Palaeosyrian noun kak-kab < kabkab. " i t is to you". in Aramaic (7 dbr > 'dbr.g. Total reciprocal assimilation implies the change of both conso­ nantal sounds in an intermediary one. c) The assimilation of r to the following consonant is well attested in various West Gurage dialects. e. "solemn oath". from *sul-sul (cf. etc. Old Syriac 'mšmš < 'mt-Smš. "shoots"). ma-qtel-nā-š > ma-qten-nā-š. as shown by Greek Apotacaparis. Hebrew salsillot. in Assyro-Babylonian *ustabbit > ussabbit. "star". in colloquial Arabic nisf> nass. to compare with Sabaic krkr. when the animals lie down. "from the sea". "he takes"). gamālathū > gamālattū. in Hebrew *mitdabber > middabber.g. e.188 phonology Assyro-Babylonian (e. "round disk". e. d) Also labials and dentals may be assimilated. in the Phoenician name of the god mlqrt < *milk-qart. "one" (feminine). as well as other con­ sonants. in Hebrew and in Sabaic *'hdt > 'ht. " I sent". e. as indicated by the Latin form Abinneric(h)us.4.g. nalšu > naššu. in Ge'ez inscriptions 'ambaharu > 'abbaharu. "dream". in Gurage wâsfi > wàsse. "from the house"). shows the assimilation dn > nn. in Hebrew (*yilqah > yiqqah. It occurs in North and West Semitic. Palmyrene 'bnrgl < 'bd-Nrgl. also with Arabic kirkira(tun) used metaphorically to designate the callous protu­ berance on the breast of dromedaries which.g. e. "talent" (weight). kakkaru or kikkār.g. *attarad > attarad. in Tigre (e. "the king of the city".

and Arabic 'urf. in Arabic.g. "base of neck". "lords". Partial progressive assimilation occurs frequently with verbal infix t which changes into emphatic t when it is contiguous to another emphatic consonant. 27. "base of neck". and to ra'a < ra'a. *istabaġa > istabaġa. where the contact of the voiced inter­ dental d with the voiceless dental t gives rise to the geminated voiced dental dd. correspond to Hebrew 'orep.g. occa­ sions the change nk > ng. e..g. frequent in Neo-Babylonian and in Late Babylonian. Besides the possible occurrences which can be best inter­ preted in a different way (§10. pronounced /iśśeqe/. The trill consonant r occa­ sions changes ' > ' and ' > ġ in Arabic where.g. in Hebrew *nistaddāq > nistaddāq. In Maghrebine colloquial šrapt < šrabt. rostrum". Another partial progressive assimilation. the devoicing b > p partially assimilates b to the voiceless t. e. Śheri ġarb and Mehri ġoreb. while the latter's feminine counterpart BaOvavaia / BiGvavaia. Partial regressive or "anticipatory" assimilation occurs e. 27. "you yourself". "he remembered". one can mention the partial regressive assimilation of voice in Gurage *timbárâkā > dimbârākâ. ġarb. There are also cases of non-contiguous assimilation between consonants. " I approached". "to be narrow".4).2).. in Neo-Assyrian *aqtirib > aqtirib. attention was paid to barra' < barra' < barran.7. "behold!" . the contact of the liquid / with the dental t gives rise to the geminated dental lateral ś (§15. Regressive nasalization explains the West Semitic names Minyamēn < Binyamèn and Mivvavaíos < BinNanay. in Ge'ez 'agā'azt > 'agā'ast. The assimilation is total when the contiguous consonant is t (§27. "outside". e. Arabic 'anbar > 'ambar. "ambergris".8). " i n order that you live". remains unchanged. "pulpit. " I drink". in Aramaic *yistabba' > yistabba'.g. "Daughter of Nanay". and its derivatives are etymologically related to Ugaritic and Epigraphic South Arabian 'rb. "we said".8. kangu < kanku. 27. e.ASSIMILATION 189 > iddakara. "he took". Inversely. "to kneel down". In Hadramawt colloquial. minbar > mimbar. 'anbe > 'ambe. " i t is immersed". In Neo-Assyrian ilteqe > isseqe. " i t was dyed". "sealed". Tigre 'agal tdnbar > tdmbar. "mane (of a horse)". e.6. ramānga < ramānka. The change nb > mb occurs frequently. "West". and the partial progressive assimilation of voice in Gurage tàpàbà > 'epâpà. where the devoicing z > s partially assimilates z to the voiceless r.g. "we shall prove our innocence".

"hammer". or total. at Aleppo. "mouse". "chief" (regres­ sive). "he struck" (regressive). "dog"). 18. qaqqada (accusative). or velar (§11. as in Palaeosyrian 'ahírtum > ì-hir-tum ['ihírtum]. In Arabic words borrowed by Libyco-Berber. and 'akbar. e.9. *sāriqīhu > sāriqīhi. e.in nouns like 'aqrab. or in Gurage dbbut > dbbdt. "wicked".g.10. as in Assyro-Babylonian hiblātu > hiblētu. to study". . " o f his thieves") and in Hebrew "segolate" nouns (e.5). keleb < kāleb < *kalb. q may change into ġ in the proximity of r. 'imri'in (genitive). "scorpion". r may also cause the change of a nonemphatic consonant in an emphatic one in modern Arabic dialects. sâlot > solot. Also a preceding dental may be palatalized. attested next to ġr.g. kdfu' > kufu'. especially in Assyrian. "prince". b) Assimilation between Vowels 27. -warġ-. ra's > rās. Vowel harmony occurs also after -i/-l in the pronominal suffix -hu/-hū of Classical Arabic (e. especially in Ethiopian languages.g'.10.g.g. "rest". Besides.190 PHONOLOGY The influence of r also explains the prosthetic 'a. "man" (nomina­ tive). from wariq. The regressive total assimilation of vowels may take place in Tigrinya. "issue".g.g. e. or uhappi > uheppi.g. "to fall". -ġra-. "damage" (progressive). " o f his thief". "road. qaqqidi (genitive). at Essaouira (Morocco). c) Assimilation between a Consonant and a Vowel 27. Assimilation between a consonant and a vowel can consist in the influence of the vowel either spirantizing the following labial. *sārìqihū > sāriqihī. 'imra'an (accusative). "to call". in Classical Arabic 'imru'un. Tigre gazaz(z)e. Vowel harmony is widely attested in Semitic. or palatalizing the following velar (§18. "yellow". Assimilation between vowels is always at distance. since the structure of Semitic syllables does not admit contiguous vowels (cf. in Ugaritic *'allūpu > 'ulp /'ullūpu/. den­ tal. e. from qarā. §22. "head".15). Typical cases of regressive total assimilation occur in Assyrian and in Classical Arabic when the vowel of the noun is assimilated to the vowel of the case end­ ing. "king". darb > darb. in colloquial Arabic dlk > die. in Assyrian qaqqudu.6). e. "to read. "prayer". e.g. ndgus > nugus. There are also examples of partial assimilation of nonemphatic to distant emphatic consonant. "head" (nominative). e. in Gurage wâdâqà > wàtàqâ. mādosa > modosa. "cock". It can be partial. "pile" (progressive).

e. laryngals. §24. killim. "he taught". e. vs. ū. and sometimes h. *rūh > rū h. "he spoke"). ġ. "stone". This phonetic notation of the Masoretes aimed probably at insuring a distinct pronunciation of the gutturals and may not reflect any really spoken language. One should also mention the phenomenon attested in the Western NeoAramaic dialect of Gubb 'Adīn and in the Arabic dialect of surrounding villages where the long vowels o. i.9) stands for Masoretic šdlah. yafruġu > yafraġu. "friends". where a appears also in contiguity with emphatics. "Jacob".5. However. Besides the emphatics. which were normally vowelless.g. '. contiguous or . šlwh /šaloh/ (lQIs 58. h. Neo-Aramaic kē fa. yikammil. "evening" (but cf. a a a x 2 x 2 2 a a B. 21.g. e. e.g.g.11. Ya'qob (Icuccop) > Ya' qob. from the transcription Koai5r| of the Edomite -name Qwsyd' in a bilingual ostracon from the 3rd century B.C. Instead.g. 21. hamar for 'ahmar. "glass". However. it is comparable to some extent with East Arabian dialects which are characterized by the change C aC > C C a when C is h. vs. "he achieves". but also in West Semitic. Dissimilation is the reverse of assimilation. or in the influence of pharyn­ gals. *yišloh > yišlah. "my friends". in ancient Arabic dialects (e.6-7). as it appears e. Dissimilation can be progressive or regressive. Dissimilation 27. "he takes out for a walk". ē. mġarib for maġrib. as against masānit. 'allim. "he sends"). where e. it is a diphonemization or a differentiation of two or more identical sounds in a word by substituting for one of them another sound of similar type or position. and in modern Egyptian colloquials of Cairo and of the Delta (e. yifassah. as it appears from Arabic and from the Hebrew of the Dead Sea scrolls. the gutturals bring about a change of other vowels into a in Masoretic Hebrew (e.9. and velar fricatives which frequently occasion a vocalic shift a > e in North and East Semitic (§19.g.g.ASSIMILATION 191 "my glass". this change presents some dialectal variations or is optional. h.e. "he is at rest". The Masoretic vocalization is paralleled by the appear­ ance of secondary a-timbre vowels next to the consonants h. as against gazāz. luhd > lahd.27).10). masānicce. also labial consonants may cause other vowels to change into u.g.g. Assimilation can further consist either in the velarizing effect of an emphatic consonant which brings about a change of other vowels into o / u (§10. "grave-niche"). and I are prounounced with an offglide before any consonant. "breath". *libb > lubb. "heart" in Arabic (regressive assimilation). "red".

and Gurage tarraqqa > dàrraqqa. "to save". Metathesis or transposition of sounds in a word occurs in all the Semitic languages. dissimilation is most often non-contiguous. in Arabic layl.g. Phoenician his /halds/. from the East Cushitic verb q)al / d'el ("Sam" languages.12. e. Metathesis can be contiguous. that is. A dis­ similation of vowels occurs also in Syriac.g. "meeting place of two rivers". Early Aramaic hsl lhasall. with the first or the second emphatic develarized (but see §10. and in Masoretic Hebrew. or Sabaic 'wld and 'Iwd. E. "he bit". There is also dissimilation of homorganic sounds in Arabic *wawāqī > 'awāqī. It results in the develarization or deglottalization of one of a pair of emphatic consonants. hehākām for *hāhākām. "night".g. "language". Neo-Aramaic [sāhid] < [hazid] (root hsd). "to harvest". "coat". the consonants are separated by a vowel. shows a deglottalization of t which changes into a voiced d. *Madīnīy (< Madīnà) > Madānīy > Madanīy. that is. "to spend the night". Metathesis 27.g. "lamb". e.g. Ge'ez nakasa and nasaka. Hebrew 'āsam and Arabic ġamada. e. and a dissimilation of homorganic sounds appears e. Tigre mawarri and marawi.7-10).g. "children". A dissimilation of voice is attested e. "faces". "star". > lūn. Hebrew Simla and śalmā. . It can also be non-contiguous. as it appears e. Harari sinān < lisān. since q is voiceless. "to shut the eyes".g. wuġūh > 'uġūh. "ounces". e. w w C. "to lie". where the dissimilation of the long vowels i is qualitative (J > a) and quantitative (ā > a). or in the dissimilation of two homorganic sounds.9). "the wise man". the consonants that undergo metathesis are in contact without any vowel between them. Abstracting from the differentiation of long consonants or the so-called disjunction of gemination (§23. "water-course".13-14). Slēmon for Hebrew Sdlomd and Greek SaA-copcov. and Tigre lahasa. It is related to the phenomenon aptly expressed by the phrase "his tongue tripped". "moon". Aramaic ktl and Arabic qtl may point to an original qtl. in the voicing or devoic­ ing of one of the consonants.192 PHONOLOGY non-contiguous. Metathetic relations appear also in the larger Afro-Asiatic realm. "to k i l l " . although a spirantization of the first b may have helped the process in this particu­ lar case. Hebrew td'ālā and Arabic taVa(tun). in Gurage d3m d > d3m t. Hebrew kebeś and keśeb. changing into kwkb in Hebrew and kawkab in Arabic. with a concomitant change / > n (§17.g. "sticks". "Medinan". Gurage kdbàzà < kàzàbâ. and in Amorite kabkabbu.

as in 'estamek. tataqātalūna. Haplology is the omission of one of two contiguous and almost identical syllables which occurs occasionally in various languages and can also be expressed in writing. with an additional change / > n (§17.13. etc.e. i. "lion". "he is on his guard"). i.g. there is little doubt that one of the consonants involved in many cases is either / or r. Haplology 27. may be reduced to taqātalūna.15. "ladder". "eminent". "scor­ pion". the prefix is placed after the first radical (e. can be reduced to bēt. In Assyro-Babylonian. *sitbutu > tisbutu. corresponds Tigre 'arqab. In Ara­ maic. Both 'arġal and 'aġral mean in Arabic "sluggish". E. "sieve". 27. 27.HAPLOLOGY 193 Oromo. "to pass". or Oromo dabra or darba. In Ara­ bic. the infix can be prefixed to the verbal form. mdzdssē. "he leaned". is borrowed in 'Omānī Arabic as humra. there are not enough examples of metathesis in the same language to warrant a definite statement on the phonetic conditions in which metathesis occurs. "to give birth". at least. "date". further research is needed to see whether there is metathe­ sis of a tf'l stem or simply an example of a ft'I stem. . AssyroBabylonian simmiltu. z. These explanations are not really convincing and another point of view will be presented below (§41.g. Beside the examples quoted in §27. and Per­ sian hurma. *yatsaddaq > yassadaq. e.g. "gate".e. becomes 'aryā. "door". instead of being totally assimilated to it. "to pass".14.g.g. to the name of the Cassite goddess known in Babylonia as Sumaliya but called Šnm at Ugarit. *yitšammer > yištammer. corresponds to Hebrew sullām. Hebrew babēt. *zitqāru > tizqāru. one of the two "liquids". s.3-4).12. to be compared with Semitic (w/y)ld. However. In Hebrew. related to Arabic dabara. becomes màssē in another one. as in Pre-Classical Arabic (e. Saho). "in the house". when the first radical is a sibilant s. "to grasp". in one Gurage dialect.24-25). Another allegedly paradigmatic example is provided by the ver­ bal infix and prefix t of the verbal stems (§41. while Aramaic tar'ā < trġ parallels "Canaanite" tġr > š'r. In general. To Ge'ez 'aqrab. š. one may refer e. 1 D.20-32) which is generally believed to be subject to metathesis with the first radical of the verb in precise conditions. Syriac *'aryiyā. "you fight". "he will prove his righteousness").

8). 'etkdteb. as Aramaic and Arabic. "table". and it is introduced by ' or h in idioms which require the presence of an initial consonant. from rák(k)âsâ. either pronounced or simply written. "son". "prayer" (cf. "tears". "he was impure". "mule".17. Gurage drkus I ârkus. §29. etc. Ugaritic 'usb'. the Phoeni­ cian divine name 'šmn = Eapouvos < šmn.16). In order to disjoin an initial two-consonant cluster by producing a new syllable. Hebrew 'ptlmys and 'btlmys for IlxoA-spatos. "night-watch".g. "broken" plural of balad. Arabic 'ibn < bn. The prosthetic vowel is not required when the initial cluster contains a liquid or the palato-alveolar s. inhās from nuhās. while the other languages. one ought to mention the use of a prosthetic vowel in nouns like Palaeosyr­ ian ís-ba-um /'isba'um/ < *sba\ "finger". sàbri. where an auxiliary vowel -/ is added at the end of a monosyllabic root (§24. Ge'ez 'dstifanos for Zi8(pavos. Punic 'klyn for KAecov.g. 27. e. Hebrew 'ašmoret < *šmrt. use e.17). "cemetery". 'udm't.9) and in some modern Arabic dialects after elision of an original short vowel. Assyro-Babylonian ikribu. e. and Arabic 'Aflātūnu for Plato. "dog". Beside the cases of composite verbal forms with initial cluster. Harari qâbri. 29.194 PHONOLOGY E . Amharic dsport. "country". the Humanized name of the god Ea. h(y)kl. or with loan­ words like Neo-Aramaic 'ustol for Russian CTOJI. "impure". There are also animal names with a prosthetic 'a. both in Egyptian Arabic.8. The same devel­ opment occurs in Western Neo-Aramaic (§24. as Tigrinya hbbi. The prosthetic vowel is employed also with foreign names. bàġli.g. " o i l " . is not a prosthetic letter. probably by analogy with the stem hif'il. The initial h of some North and West Semitic words borrowed from Sumer­ ian. à. The final -i can be dropped if these nouns are followed by another word. The use of written h is limited to the Hebrew verbal stem hitpa'el. This -i is attested also with roots ending in a gemi­ nated consonant. "heart". u.19). "finger". "endurance". Instead of a prosthetic vowel. Prosthesis 27. Tigrinya kálbi. This prosthetic vowel can be i.that form a special category (§27. hyn. "sport". an anaptyctic vowel can be used (§27. a prosthetic vowel is generally prefixed to the first con­ sonant.16. "palace". but a reflex of Old Sumerian h. e. as Late Babylonian Ik-se-nu-nu for Sevcov. A third mean of disjoining a two-consonant cluster is attested in Tigrinya and in Harari. as hbrk. "steward". iblād from bilād. already explained when dealing with syllabic structure (§24.8). "cop­ per". d < i. 'ifta'ala. like in the sound st .

s.9). Neo-Aramaic grībā. nif'al = N-stem) or at its end. e.g. A so-called "prosthetic" vowel à or d occurs in modem Ethiopian languages. But the use of the anaptyctic vowel is attested in other cases for the same nouns: ri-gim < rigmu. Anaptyxis is the insertion of a supplementary vowel in a word in order to disjoin a two-consonant cluster by producing a new syllable.g. Also the con­ nexion of a noun in the construct state with its complement can make the insertion of a vowel superfluous. the Djebel Sindjar identified with the Moongod and known later as Šangar or Šaggar.ANAPTYXIS 195 (§17. without serving to dis­ join an initial cluster. "king". anaptyxis did not spread automatically to all doubly closed syllables. māwet. "holiness". "rainy season" (pronounced also as kdràmt). 24. ša-ki-in I šá-kin I šá-kan < šaknu. Amharic krâmt. perhaps originally Šanar. màngdst. etc. "two". in kalbu > kalb > kāleb or kalib.g.9). A similar d l2 . zi-kir < zikru..18. ša-ak-ne-Ellil [šakn-Ellìl]. "death") is a unique feature which is not con­ firmed for a somewhat earlier period by Origen's Hexapla mentioning. 27. Modern Semitic languages permit similar consonant clusters in initial and final position. Anaptyxis 27.19. "earth". "city". especially when a two-consonant cluster was created by the loss of case endings (§24.g. "evening". in Gurage. an Old Persian word already borrowed into Imperial Aramaic (grb).9. as ktub. e. The vowel u is found in Harari urūs. A widespread use of anaptyctic vowels gives a peculiar flavour to the Ebla texts where they cannot be reduced simply to a particular way of using the syllabograms. e. e. However.g. r. YLobafor qodeš. "head" (root r's).g.9. "the voice of Adad". where the presence of a liquid can dispense from using the anaptyctic vowel (§17. app for 'ereb. 24. Sa-nu-ga-ru . "dog". e. although the Tiberian Masoretes were pronouncing it 'ištē. ktabt. F. "the trustee of Ellil". "books".g. "the mention of your god". melek. "thorn". especially before /.g. e. as shown by Phoeni­ cian qart. "king". Hebrew štē.8). zi-ik-re-el-ka [zikr-elka]. "you wrote". àsok. The systematic use of the anaptyctic vowel e in Masoretic Hebrew (e.g. KopP for qereb. e. "bushel". "government" (§17. in Assyro-Babylonian ri-ig-maAdad [rigm-Adad]. or milk. š. Even plosive clusters occur regularly in Maghrebine Arabic. "middle". The anaptyctic vowel can be used at the beginning of a word (e. apafor 'eres. It is also called "epenthesis".9).

"the king of the City".. ends in a vowel or in a consonant producing the spiranti­ zation. especially in the traditional recitation of the Qur'ān.g. Accord­ ing to Sibawayh. speakers pronounced it without assimilation. "tell me!".4). Tamīm speakers of Central and Eastern Arabia were saying [harra'ayta].. thus e. to produce the phrase hal + ra'ayta. designates the assimilative changes occurring in a word under the influence of neighbouring words uttered in consecutive speech.g. y. "did you see?". w. with the assimilation l-r >rr\ in Hedjaz. e. uttered together in consec­ utive speech. the spi­ rantization of the consonants b g d k p t becomes operative also at the beginning of a word when the preceding one. The sandhi-writing 'zlh for 'zl Ih. " i t is to you".20. in Sūra 24. etc. is pro­ nounced [hallakka]. Sandhi is widely attested in Arabic. G. In Tigre.44: halaqa kulla dābbatin. e.g. A similar development gave rise to the divine name Mlqrt < *milk qart. . instead of the traditional 'adbār. "mountains".g. is pronounced [halakkulla dābbatin] with the dropping of the final -a of halaqa and the assimilation q-k > kk. H. for hy yhwh. In Hebrew epigraphy. A sandhi-spelling.196 PHONOLOGY situation occurs with the Ethiopic syllabograms of the 6th order that are tradi­ tionally pronounced in Ge'ez either vowelless or with the vowel d < i/u. Sandhi. hs trb'l byrm for *bn yrm. 'addbār. is attested in some Old Phoenician inscriptions. i. Sandhi 27. "he went away" (cf. h.21. n) is amply attested in Semitic languages. which involves the assimilation of n to the following consonant. is common in Galilean Aramaic and it became a rule in the Neo-Aramaic of Ma'lūla.e. Similarly. e. "king of Kition". Elision 27. the spelling qwly. halla 'al-ka. son of Yarīm". I. §65. It stands to reason that these syllabograms were originally articulated with a very short vowel in all circumstances. "he created all the animals". Elision of vowels and "weak" consonants ('. "Yahwe is alive". the sandhi phenomenon leads to spellings like hyhwh. on the other hand. is used regularly in early South-Palestinian Arabic for qui ll. as already reported when dealing with these sounds. Thus. "arrow of Tūra-Baal. and to mlkty < *milk Kitti. meaning in Sanskrit "a placing together". "you have".

The remaining vowel is originally long and the process is practically identical with the contraction of the diphthongs ay I iy I aw > ā (§ 22.g.3-5). e. kdhla. w can be elided when they are not long or geminated. Simi­ larly. 1. thus reducing e. § 45. 2° the quality of the long or stressed vowel tends to prevail. The elision of short vowels occurs also in modern Arabic dialects and in other languages under influence of a strong word accent (§25. The h of the presentative *han (§49. it is often kept as his­ torical and etymological spelling without being articulated.38.4.g. while fa 'il mdfa'ul were reduced to fa 7 and the verbal forms fa'ila to fa 'la. ydkl. there is elision of intervocalic h in pronominal suffixes: klm < klhm.g. V. " I carried".28).15).16. one of the so-called "Allah's daughters". "they have built". at Tell Fekherye. fu'l. y. In Early Aramaic.31).6) may be elided in Ugaritic in intervocalic position: wn < *wa-hanna/u (KTU 1. When two vowels meet. corresponding to the frequent Hebrew wdhinnē.24. North Arabian mnwt(w) = Latin Manavat > mnt /Manātl. following rules are generally applied: 1° the meeting of two like vowels results in the same long vowel. e.8).22.23. "heavens". y. collo­ quial Arabic dale < dalā < dalaw. either one is elided or a glide w. 27.g. Postvocalic and intervocalic '. ' is produced (§ 22. the ' was still pronounced in some forms of the words where it was later elided.25. . Pre-Classical Arabic baqat < baqāt < *baqiyat. "he drew".Išamēml < šamayīm.IV. Punic nasot Inašotl < *našāti < našá'ti. syllabic ú-ma-lu-ú. like a and /. the dissyllabic nominal patterns fi'i]. at the time the orthography was fixed. Phoeni­ cian o~aur|u. "to be able". Thus. e. the imperfect being yabl. the verb in rht[h] yml'u /*yumalli'ū/. e. It is omitted in the imperfect of the frequently used Ge'ez verbs bdhla. 1. When there is elision.ELISION 197 27. Assyro-Babylonian ibniū > ibnū. e. was very likely pronounced [yimallu] or [yumallu] (cf. "to say".50. 27. the h of the pronominal suffixes -hu I -ha is elided in Ethiopic (> -u I -a) and the masculine suffix -hu can be elided also in Hebrew (> -6). may produce a vowel with an intermediate point of articulation. "all of them". According to Sibawayh.V.6. "and their children".3. "she remained". 3° two phonetically distant vowels. Nabataean Aramaic wldhm for wyldhm. In West Semitic.5).g.g. "they filled his hands" (KTU 1.24. in ancient Arabic dialects unstressed i and u were elided. kin < klhn.fu'ul to fi'l.V. 27. This was probably the case also at Ugarit.

to become" root kūrì). "Baal of the Skies". also by -'. 'aha). "Stone of the Mighty one". in the New Testament (Aá^apos). provided by parallel passages (KTU 1. In North Arabian. in var­ ious Middle Aramaic dialects. composed of the negative element 'a and of the verb kona. " i f he is not". "family".IV. y l 27.29). "She is a sister" (rather "brother".. which is often written 7 in Safaitic.g. and in Arabic by the vowel letter -h.g. A similar assimi­ lation or elision must have occurred in the Libyco-Berber verb llukk < *hlukk.g.26.28. "plough-land". by the name L'zr instead of biblical 'El'āzār. there is no doubt that the ending -at gave way to -a > -ā at some point in the history of several Semitic languages. Also / and n are elided in certain conditions (§17. "to be. Hīrom < * 'Ahīrom). for balhonà. from Tell Beydar (cf. mhrtth). d-'l Ms k. and Ikt. "you (feminine singu­ lar) break". It is uncertain whether -h functions already as mater lectionis in the Ugaritic equation mhrtt = mhrth.2). "go!".g. 27. of the very same verbal root. e. "one".3). Aramaic had. as shown e. or the / of B'l-šmyn >B'sm(y)n. e. like in Ge'ez 'ako. An elision in the middle of a divine name is attested. I f one hesitates to recognize this general trend in the Palaeosyr­ ian personal name Si-a-ha /Šī-'ahat/. "to go". "people".29). but it is certainly used as a vowel letter from . The nasal n is often assimilated (§27.4). the elision of h is attested in the word ahl. The same phenomenon occurs in Mishnaic Hebrew. The elision of the feminine ending -t at the end of a word is a widespread phenomenon attested in Semitic and in Late Egyptian (§30. as the auxiliary -al in Harari before suffixes. The elision of an initial unstressed syllable 'a occurs in Phoeni­ cian personal names (e. preserves the original monosyllabic char­ acter of the word (§35.198 PHONOLOGY Elision of h occurs further in North Gurage hono. when h is preceded by the / of the negative al-. "to tread on". "May my brother be exalted!".6. The residual final -a was then indicated in Hebrew.g. thus balonà. and in the West Semitic forms Ik.27. 3 = 14 cor. in Ara­ maic.3) or elided at the end of a word. "from the people of Mašaku". §27. 27. This form is found in contemporary Palestinian inscriptions. and also in some Aramaic names as hlrm /Hīlarīmf for 'Ahīlarīm. and in manuscripts not affected by hypercorrection (§27. " i t is not". tisàbraš < tisàbralš. e. by Neo-Punic Abaddir < *'abn 'addīr. The liquid / can also be assimilated or elided. Instead. "to be".

In ancient South Arabian anthroponomy. the elision of final consonants is attested in various places. "just". "praying to him is sweet".. or tm'. Safaitic bkrt. fallaciously. The same process took place in Phoenician. "year".29. that was already used as mater lectionis in Lihyānite (e. "young she-camel". Moabite Mhdbh. kl k I . the theophorous element 'Attar occurs frequently without r at the end of a name. illustrate this phenomenon. "bales (of straw)". "perfect".t.C. e. maqarrāt(e). confirmed by Latin transcriptions such as Anna for Hnt. Aramaic Pt'sy for P3~dl3ś. "first fruits".g. "hundred". the place name Mug-ri-i occurs next to Mug-ri-du \ the personal name A-mi-i occurs next to A-mi-du and prob­ ably designates the same person.. A similar process occurred in Late Egyptian. Lihyānite s nt.. on.g. rèsāti. since "whom" sounds more formal and hence seems. Hypercorrection occurs when a speaker or a writer over-com­ pensates for an error which he fears he might incur. final -t was dropped in Assyro-Babylonian pro­ nunciation after a I ā and after ē. "folk"). Hebrew m'h.g.g. 'dh /'idā/. Thamūdic nqt /nāqat/. show the feminine ending -ā > -o. "while"). At Emar. but the spelling of feminine nouns in -h throughout the Qur'ān shows that final -t was elided in the Pre-Classical period and that the residual vowel -a was indicated by -h. e. mh /mā/. a hypocoristic suffix -iy is added to the name. Such mis­ placed changes occur frequently in Mss.30. In both examples. "she-camel". "Medeba". first in the verb where this occurred before the fixing of the orthography (hence p'l. The feminine ending -t was still preserved in Pre-Islamic North Arabian (e. as when an English speaker uses "whom" where "who" is required. but especially when he tries to harmonize the idiom used by a writer with a "classical" form of . Hittite pseudo-hieroglyphic transcrip­ tions.HYPERCORRECTION 199 the 9th century B. Hypercorrection 27. "what". transcribed rsh. In proper names. At Ebla. "Whom Isis has given".g.g. as shown by number of Semitic and Greek transcriptions of Egyptian proper names. x 27. e. Palmyrene 'mt. transcribed mqrh. In the first millennium B. tà-ka for Dagān. more correct. Neo-Assyrian Ekallāte transcribed 'glh in Aramaic. "she did"). later also in the noun where NeoPunic spellings like sdyq'.C. not only when a copyist endeavours correcting a normal scribal error. e. Aramaic tbh in tslwth tbh. like ma-li for malik.

"your word". zdmā for damā.g. but lack the necessary knowledge. Thus. E.g. but its vocalization was "cor­ rected" into -kā.g. "blood". The pronominal suffix of the 2nd pers. masc. tried to harmonize Mishnaic Hebrew with Biblical Hebrew because they considered departures from the latter one as mis­ takes. . ddbārāk. Fortunately. and finally grammars and dictionaries. e. it can be shown. Hypercorrection occurs also in Mandaic manuscripts. e. but the word was systematically "corrected" in more recent Mss.200 PHONOLOGY the language. and later the printers. etc. that Mediaeval copyists. all these words with etymological d and d are pronounced exclusively with d in colloquial Mandaic.g.. This phenomenon is not at all unusual where speakers and copyists try seriously to follow the rules of a "cor­ rect" language. the Mishnaic Hebrew word for "man" was 'ādān instead of biblical 'ādām.g. sing. thus writing zeqlā for deqlā. was -àk. e.. This "hypercorrecting" tendency let to a complete distortion of the linguistic structure of Mishnaic Hebrew in many manuscripts. when scribes extend the archaizing spelling with z for d to words in which d is etymological. "palm-tree". printed texts. and it disap­ peared in printed texts. e. ddbārdkā.

"to write").D. kitāb-. Their number even increases significantly i f one accepts that only two of the three radicals of the triconsonantal roots are the main bearers of the meaning and that the third one had at one stage the task of a determinant or modifier in very much the same way as occurs with vowels in the fully developed triconsonantal system. kataba. kātib-. T H E R O O T M O R P H E M E 28. but it may undergo a certain degree of change through internal diachronic development or by contact with other languages. This is the reason why a rather synchronic assess­ ment of the characteristics of the Semitic root morpheme may differ somehow from a diachronic appreciation. This is illustrated by the well-known example of the Hebrew verbs prd. and of one or more vowels. 28. by Hayyudj of Fez whose ideas are generally followed up to now. morphology is relatively resistent to radical linguistic mutation. I f Semitic languages can be considered as genetically related.1. Arabic ktb. yaktubu. maktabat-.g. the three con­ sonants. "he writes"). "library"). "writer". Now.4 ff. form the smallest lexical unit of the language and constitute the root morpheme (e. According to this traditional grammatical analysis. Such a conception was strongly advocated in the 10th cen­ tury A.Ill MORPHOLOGY 1.2. prm. it is because they exhibit a systematic correspondence of words in their morphology. . or determine the grammatical category and act as grammatical morphemes (e. even apart from the roots that became biconsonantal in consequence of the dropping out of one of the radicals.g. besides the triconsonantal ones.g. "book". cannot be denied. especially in the inflectional patterns. This basic semantic element is assumed to be further qualified by a number of vowels or vowels plus consonants which either specify the meaning of the root and serve as lexical morphemes (e. called radicals. Most of the words of the historically attested Semitic languages are usu­ ally analyzed as being a combination of three consonants. "he wrote". The material of a language is generally taken to be its words. The existence of biconsonantal roots in Semitic languages. as the one exposed in §28.

"seize!") that can be extended by affixes. 28. However. g u b .. Semitic roots are con­ tinuous morphemes which are instrumental in derivation but subject to vocalic and consonantal change in this process which is based on continu­ ous or discontinuous "pattern morphemes".g. In Sumerian. "great". eight monosyllabic types of Semitic root morphemes can readily be distinguished in historical times. like those specifying the grammatical gender (§30. prs. Such a situation does not occur in other language "fami­ lies" where the roots also include vowels and can be pronounced by any speaker of the tongue under consideration. e. The same happens in languages as different as Chinese or Sumerian. the Semitic triconsonantal or biconsonantal root. "he seizes"). were already agglutinated to the root. that have the radical pr in common and express the basic notion of "dividing". "boy". which should therefore be regarded as forming part of the root. In fact. "fatherhood") or grammatical morphemes (e.g. "to stand". the basic stock of the Semitic vocabulary appears to consist of monosyllabic root morphemes (e. is only the abstract basis of a family of words used in the language and did never exist as a living reality in a spoken idiom.1. In English.g. e. the root is a stable reality. for practical reasons and to keep in tune with the common usage of the Semitists. the roots include vowels and they constitute pronounceable realities. "father".g.g.202 MORPHOLOGY prs. In other words.g. The derivational process can occasion phonetic modifications (e.4. The root morphemes in question consist . abbūtu.10-11). g a l . espe­ cially in the verbs (e. con­ ceived as the smallest lexical unit. and further changes are due to the standardization of the monosyllabic root in accordance with a dissyllabic stem pattern. This practice should be considered as a simple short­ hand. the morphological analysis of basic Semitic words and forms — especially the three -a-. ya-hud-u. However. and -u. without any morphologic bearing on the Semitic word structure.classes of East Semitic and Arabic verbs (§37. we shall often refer to the roots by indicating their sole consonants. Contrary to the traditional opinion. d u. 'ab-. both lexical and grammatical. 'ab-ūt-. etc.g. 28. "fatherhood" in AssyroBabylonian). an evolution which took place already in the Proto-Semitic period. Arabic ya'hud-u). when complementary mor­ phemes. hud. prq.. "love". "to go". Despite these changes that Semitic has undergone. 15) — reveal a relative stability of radical vowels.3. which are either lexical morphemes (e. prr. "to cut". 38. -i-.3. "god". for example. prš. as in Indo-European languages in general. as d i n g i r .

u. gaš. pá-šu. one can then distinguish twenty-four sub-groups of monosyllabic roots. The inflection of the nouns qā. or with -i attested also in the nominative and in the accusative. "truly". Since the vowel length behaves like a consonant and since the initial or final two-consonant cluster is just a variant of a long or geminated consonant. pá-i. l 2 v x 2 y { 2 2 2 3 28. "come near!".5). lū. 28. da'. hah.6. or ultra-long syllables.6). or long. Ci: hi-. pu-šu. "know!". In general. "my mouth". "ewe". all in the accusative. pi-šu.THE ROOT MORPHEME 203 either of short syllables of the type Cv. "that (one)" (cf.> u-. "not". lā. §28. pū can bring about a change of the vowel. gū. "father". "that (one)" (cf. yad-. Ci: ki. du/tu. "what?".ī. "voice". or of long syllables as Cv. while the Arabic construct state fū favours a Cū root morpheme in the case of "mouth". "in". "and". in Old Akkadian pu-i. "give!". e. "gauge". ha. . "for". and C. ham-. The three sub-groups of the Cv class are distinguishable on the basis of the long vowels ā. 28. had. "one". In this approach. as CvC. ka-. qā. Taking the three fundamental vowels a.vC C (§24. may-. dū/tū. Cū: gū. "father-in-law". 'ah-. u into account.g. "oh!". "because". so that". pū. distinguishable on the basis of the vowels a.3).5. or with the vowel -a in the accusative. "let it be.7. For clarity's sake. The Assyro-Babylonian nominative form qa-a confirms the Cā pattern for the noun "gauge". The CvC class is well represented by monosyllabic nominal roots and verbal basic forms: CaC: 'ab-. the pro­ nounceable two-consonant clusters are acceptable as well as in modern colloquials. "water". U-. C C vC C vC C . "his mouth". mā. one could also say that the monosyllabic roots are either short. la-. "as. There are three sub-groups of the Cv class.> a-. CvC and C C vC or of ultra-long syllables. "brother". and their construct state may be used either with the -u of the nominative. CM: lu-. "hand". "take!". "mouth". it is better to divide them into eight groups. i. like". §28. "and so. interrogative.ū: Cā: yā. "truly". pa-lfa-. these morphemes are proclitics or enclitics: Ca: wa. however. ta'-. i. -ma. pi-i. qah. be it!".

e. E. although Hausa mutum and East Semitic mutu(m) exhibit a non-geminated t. "to get up". Now. may originate from *hlik. "to place". ktub.9. 1 2 3 C C aC : rba'. since the initial tense or long 11 probably derives from *hl. or qūm. tib. "go!". "seize!".g. "bear!". sba'-. The CvC class with an internal long vowel is well attested among nominal and verbal roots. CuC: mut-. "sit down!". "son". They base themselves mainly on the Classical Arabic "weak" verbs. there tend to be strict constraints on the formation of clusters in most languages.g. lidil). which is often identical with that of the radical. The Hebrew plural mdt-īm. "god". r (§17. "take!". particularly in the imperative of the simple stem: e. "learn!". "people". although there is a widespread opinion among scholars that Semitic and even Afro-Asiatic "weak" verbs have triconsonantal origins.g. din/tin. would indicate. "olive-oil". "man". sabat. 'inzil.19).9). "write!". CjC /C : mri'. "man". but 'iftah. "night-watch". the best attested combinations involve one of the con­ sonants /. in Neo-Aramaic. the Hebrew imperative lēk < *lik. "go!". rid. tin-. "open!". §1. "four". "belonging". 'arba'. The C C vC class with a consonant cluster in initial position occurs frequently in modern Semitic languages. the Phoenician divine name "Eshmun" derived from *šmun. ptah. and seems to imply a development nt > tt > dd. "go down!". How­ ever. in Arabic. "write!". is apparently related to the Libyco-Berber plural midd-dn or mddd-dn. but it was largely represented also in Pre-Classical languages. the socalled "weak" verbs of the type śīm. lik.2). "sheep". as a comparison with Libyco-Berber seems to suggest in a few specific cases. su'-. hud. "men". šim-. 'uktub. "two". mil.16-17) or an anap­ tyctic vowel (§27. The clusters involved by these patterns are generally resolved in Clas­ sical languages by the addition of a prosthetic (§27.8. "give!". but limad.g. This class comprehends. in particular. bin-. gb) may play a role as well.and Eapouv-. in Assyro-Babylonian. 28. "seize!". Alternative forms with prosthetic or anaptyctic vowels can coexist. "go down!". šmun-. m.204 MORPHOLOGY CiC: '//-. kusud. ] 2 3 2 3 C C uC : { 2 3 dltmur-. "men". e. Samun. ru'-. n. "name". "companion". "four". Some specimens of this group may go back to a Proto-Afro-Asiatic pattern CCvC or CvCC. but labiovelar consonants (kp.g. by the imperative of triconsonantal verbs in the simple stem. as shown e. sgul-. "tread on!". 28. in Hebrew and Arabic. "man". where the tense dd suggests a link with Bantu mu-ntu. since they are frequent in African languages (cf. "go down!". Arabic . "finger". "open!". as Libyco-Berber llukk.

"rock". "three". kapp-. mas(s) > mass(i/a).> kdlš. tmāl-.THE ROOT MORPHEME 205 morphology in general. 28. "shade". firr(i/a). "be at rest!". "ancestor. gud(d). The situation is here comparable with the opposition of voiced and unvoiced consonants (§10. This pattern characterizes many nominal roots and it constitutes the stem of the Assyro-Babylonian stative (pars-). bāš. tmān-. "wound". "cage" (Greek K^COPOS). tlāt-. "be ashamed!". "feature(s)". This extensive analogical process makes Arabic less fitting for an analysis of the monosyllabic roots of the CvC class than most other Semitic languages. nūn-. qirr(i/a).12. "forehead". "gnat". x 2 3 C. where the tense consonant qq could go back to a labiovelar q (§11. "foetus". 2 3 x 2 x 2 3 28. Instead of the final long or geminated consonant. C J M C C : dubb-. *kdīś. "begin!". but its existence is firmly attested. jade" (Classical Arabic 'ikdīš or kadis).11. zimm-. "cart horse. "clay". "feast. pilgrimage".11. "speak!". "cave". "kill!". x 2 2 C aC C . "heart". sis-. x 2 2 šinn-. and that of the verb in particular. "oven". "flee!". "pitcher" (Classical Arabic 'ibrīq). "street". sim. 2 2 28. as shown by words like klr. "get up!". or blr and būr. "fish". hurr-. "palm of the hand". šušš-. mūś-. with the exception of 2 3 . 18. 'amm-. "oven". "arm". libb-. sir > sir. thus adapting monosyllabic roots to the triconsonantal system. kīs-. till-. kūn. "tooth".8). CūC: būl-. in other Semitic languages. "place!". "roll!". More insight might be found in Libyco-Berber verbs like qqim. "night". "well".C āC : drā'-. founder of a family".in Assyro-Babylonian and kūr. "eight". it has a two-consonant cluster at the end. The C vC C class is characterized by a long or geminated final consonant. C C ūC : klūb.10. CīC: zīm-. Such monosyllabic roots would sim­ ply require the transfer of some samples to the classes CvC or C C vC w w x 2 y CāC: kāp-. "mother".7) and imply an original form *q im in Proto-Afro-Asiatic. Ml > sal. "be firm!". "touch!". The CjVC C class is a variant of the preceding one. kīr-. pan-. in particular by some basic numerals. nag. "stay!". C C īCỳ briq-. x 2 2 gal{l). dubb(u). "yesterday". put-. "flower". tit-. qūm. "fly". "livestock". sūq-. "chief. "sixty". kūb-. hagg-. C iC C : 'imm-. "ask!". śarr-. king". reflects exten­ sive late analogical formation. "bag". "face". baqq-. The C C vC class with a consonant cluster in initial position and a long radical vowel does not occur frequently. One should note that the opposition l:ū is not absolute.

"vow". qudš-. "hand". Thus.206 MORPHOLOGY the 3rd pers. ġunm-. "dis­ missal" (root wś'. The following list includes only examples of nominal roots: C aC Cỳ 'amr-. derives from such a secondary form damā with a w-glide separating the two vowels. 'igl-. damā for dam. yadā for yad. In Palaeosyrian. "ten". Similar enlargements of biconsonantal roots are found in Palaeosyrian and in Libyco-Berber. instead of damly. These forms go back to the following elements: 'ah-ā + atum. ba'l-.-"she-bear". "lord". "order. the Semitic enclitic -ā often occurs in cases where " m y " is implied and it is probably related to a pronominal suffix of the first person singular (cf. "you". plural of Us. sing. šab'-.7). "booty". ilsawdn. "valley". 'idr-. C uC Cỳ 'udn-.9) *ilsā and *ismā. 'aśr-. sa-zu-wa-tum /ša(w)śu(')wātum/. instead. e. "joy". and šmh'. A comparable phe­ nomenon occurs in dialectal Aramaic with dbhh. The original tf-morpheme is preserved. besides the normal form an. Arabic damawly. According to the com­ mon opinion. In fact. like in the name 'Abraham < 'Abrām and in the plural form of some nouns (§31.26). that are unlikely to have been borrowed from the Aramaic emphatic state (§33.13. "document. "king". "the memorial" (lit. In reality. 'ahā for 'ah. "bloody". "wall". The enlargement of certain biconsonantal roots with -ā should be compared with the colloquial extension of Arabic pronouns to anāya. "help". in Libyco-Berber. some ancient Arabic dialects used 'abā instead of 'ab. "five". tūb-ā + atum. and with the widespread suffixing of an enclitic -a in various Agaw dialects. ravine". Now. the second vowel of the stem is to be considered as consistently suppressed. "blood". hamš-. These forms might be compared with Mishnaic Hebrew 'abbā. "father". "(my) father". especially to the pronoun of the first person singular an-a. x 2 x 2 x 2 28. masc. . C iC Cỳ biśr-. "(my) mother". nidr-. gadr-. "gorge. kalb-. tù-bù-a-tum /tūbuwātum/. "ear". book". This enclitic produced derivatives that are sometimes considered as proofs of the original triconsonantal or dissyllabic character of the roots under consid­ eration. "seven". we encounter at Ebla ù-hu-wa-tum /'uhuwātum/. "dog". both are "sound" plurals of older internal plurals (§31. plural of ism. §36. "to go out"). "tongue". "holiness". Thus. 'umq-. ša{w)sā(') + atum. where a /z-glide separates the two vowels ā. (paris. " I " . sipr-.19). śi'b-. "the name"). "darkness".g. "fraternity". "goodness". "brother". hušk-. however. and 'immā. "is separated"). "calf". dntāya. word". instead. "name". the a-vowel changed into u under the influence of the w-glide. with the Libyco-Berber use of an expressive parti­ cle ay. malk-. and ismawan.

"double!". kalb-.with respect to the number of animals being referred to. When -im is added in Hebrew. For example. can func­ tion only as a noun. because they cannot stand alone as independent words. 43. "name". and "empty". The "full" and "free" morphemes are not fully defined by their semantic and phonological properties. and sipr.g. The task of lexical individualization and grammatical catego­ rization is assumed. the effect is to further specify kalb. 28. but the "full" and "free" verbal morpheme Ictub-. 28.11). a "dog". E.and kalblm are nouns..14.16. Similarly. "two". while the suffixed verbal morpheme -ū usually indicates the plural. when the morpheme -y is suffixed to the root. affixing the imperfect and indicative morphemes to the Arabic verbal root *ktub. They are "full" morphemes because they have a more or less independent meaning. kalb. and never. by lexical and grammatical morphemes (§28. although they are not all empty of semantic content. Affixes and infixes have varying effects when they are added to roots.gives another verbal form .to form kalb-im. 28. "write!". because they do not have an independent meaning. adding the plural morphem -lm does not change the grammatical class of the word in question. In fact. some nominal morphemes may become verbal when an appropriate lexical morpheme is added to it. They are "free" morphemes because they can stand alone as indepen­ dent words. Conversely. gives rise to tniiy).li-malk. indeed. and 'abnā' of 'ibn. say. Such cases cannot advocate the alleged tri­ consonantal origin of the roots in question. These morphemes are called "bound". so that one or a series of full morphemes in isolation can be fairly meaningful. e.13. The roots are sometimes called by linguists "full" and "free" morphemes.1) which are affixed to the root or infixed in it according to a series of well determined patterns. this morpheme seems to have a causal function (§41. They also have syntactic properties which determine how they function with respect to the gram­ matical processes of the language. "dog". the prefixed morpheme ma.21).means "a book (belonging) to a king". Both kalb. tin-. "dogs".. e.g.1). e.15. as a verb.g.suggests a determined ani­ mal. "son".THE ROOT MORPHEME 207 corresponding to the Arabic internal plurals 'asmā' of 'ism.g. is subject to a shift in grammatical class or part of speech when determinate lexical morphemes are added to it (§28.forming nouns often implies a notion of place (§29. to kalb.

Uninflected morphemes are subdivided in adverbs (§47). most adverbs and prepositions are derived from nouns.vs. "because" (§58.16). and the noun sum.208 MORPHOLOGY ya-ktub-u. for instance. "library". The addition of affixes or infixes to roots is one way of con­ structing complex lexical items from simple ones and of indicating their grammatical function.is a deverbal noun.. As for the grammatical function of differ­ ent units in a clause. with no affix in a Semitic language lacking a functioning case system in nouns. the pattern of which. belong to different grammatical classes. However. As a rule. These distinctions are based on the actual use of the parts of speech in Semitic languages. and malkat-. we obtain the East Semitic subordinate conjunction aššu(m).17. "he writes". Some words. yaktubu and maktabat-. 28. yaktubu is a verb. "name". as maktab-at. libb malkat. 28. " i n " . are formed by combining two or more roots. The derivational suffix -at-. and even vocalic change. but they can also be used as relative pronouns. bēt. *ktub. linguists often distinguish between "inflectional" and "derivational" affixes. as prefix. On the other hand. the conjugation of which is examined in the corresponding chapter. On the basis of the categories for which certain classes of words inflect. and verbs (§37-46). meaning "house" and "place". 28. combining the preposition ana. However. Thus.18. yet both are nouns. Affixation and infixation involve adding an "empty" mor­ pheme to a " f u l l " morpheme or to a larger unit containing a " f u l l " mor­ pheme. Nor is a shift in grammatical class always signalled by an overt marker. "king". infix. There are various other ways of forming complex lexical units and expressing the relative function of various items in a sentence. "grammatical" and "lexical". suffix. however. where the nomen regens is followed immediately by the nomen rectum. but maktabat.are basically nouns.. to use older terms. with no affix (§57.g. connective and deictic particles (§49). prepositions (§48). is dealt with in the chapter on nouns.19.and ašar. pronouns (§36). etc. "queen".5). e. it can be expressed by their simple juxtaposition in a determined word order. by way of contrast. or. and the distinction . relates malk-. a shift in grammatical class is indicated by overt markers. Accordingly. three inflectional classes or parts of speech may be distin­ guished: nouns (§29-35). derivational affixes do not always effect a change in grammatical class as in the example of maktabat-.g. E. "heart of the/a queen".

They are derived i f their pattern repre­ sents an extended or modified verbal or nominal root morpheme. in Afro-Asiatic. but Somali qād < *qāt is a verb meaning "to take". The same paradigms will be employed in the sections dealing with the inflection of the noun. or the usual paradigms will be used: fl mainly for Arabic. gender (§30). e.1. called "stem". They are primary i f they correspond to a root morpheme. 2. In East Semitic. 29. and often also through their morphological type. number (§31).3.NOUN PATTERNS 209 between nouns and verbs is not clearly cut in Semitic and. as well as in the further chapters of this book. and habār "to curse" or "curse". In the following. and case. Semitic nouns are either primary or derived. the main noun patterns will be presented with their principal semantic fields.g. or "pattern". number. qufa' means either "to cough" or "cough". state (§33). in general. Arabic 'ubūwa. prs for AssyroBabylonian. e. "noun-form". as well as numerals. Nominal patterns are said to be "simple" when they correspond to a root morpheme or appear as its allomorphs. e. qtl for most North and West Semitic languages. In Somali.. qātuim) designates the "hand".g. Noun Stems or Patterns 29. "fatherhood". case (§32). For the identification of the patterns. because there are a number of Afro-Asiatic roots which are used both as nouns and verbs without distinctive affixes. "father". The participle and the infinitive are verbal nouns. These nominal subclasses are usually distinguishable through their various degree of subjection to the inflectional categories of gender. adjectives. T H E NOUN 29. A noun is a member of a class of words which has a descriptive function and comprises substantives. either the symbols CvC. They are "extended" . viz. etc. inasmuch as they can be established. "to help" or "help". There are many deverbal nouns and denominative verbs. which is associated in several instances with a specific meaning or function.g. while the stative or suffix-conjugation stands on the threshold between nominal and verbal predication. Comparative Afro-Asiatic linguistics shows even that the distinction between verbal and nominal roots is not always clearly cut. and participles. gargar..2. A. 'ab-.

x 2 2 29. The "simple" patterns are distinguishable from each other by vocalization. "lay down". Ugaritic or South Arabian. and basic numerals (had-. afformatives or infixes are added. and phonetic develop­ ments. The "extended" patterns are often discontinuous and may be superimposed on a root.12) are likewise subject to phonological developments. Their semantic field includes kinship (e. "blood".19).6-7. 'ab-. "brother". "sit down". "hair". In Arabic. "hand".g. "dog". These stems mostly denote concrete nouns. "bed".g. unless such a spelling of stems belonging to the type C vC C is to be attributed to the inadequate character of the writing system.g. "river". Arabic maġlis-. and by lengthen­ ing or gemination of consonants.5. or a long vowel (CvC). a) Simple Patterns 29A.g. "one". when a diphthongization occurs. kalb-. Considering the uncertainties resulting from lacking vocalizations. "hunt". "conference room". but Aramaic miškab.9. The monoconsonantal (Cv) noun stems are fairly rare. tard and tarad. E.11). 'ah-. sa-ma-nu. 'imm-. dam-. ša'r and sa'ar.g. "father".g. e. or a long second consonant (C vC C ). The same phe­ nomenon seems to be attested in Palaeosyrian (e. but are subject to phonological developments according to the principles exposed in the section on phonology (§10-27). "palm of hand").to the verbal root *glis (/-class). E. The triconsonantal nouns with one short vowel of the type CjvC C (§28. nahr and nahar. " o i l " ) . the patterns fa 'al and fa 'il may occur as phonetic variants of fa'I. yad-. is formed on the pattern miCCaC which is superimposed on the verbal root *škub (w-class). diachronic factors. conceals a wide variety of morphological "simple" noun stems. by lengthening of vowels. e.g. pū-. tin-. Hence it is evident that a purely con­ sonantal script. we shall not enter in this Outline into a discussion of the vocalic components of discontinuous patterns. "mouth". but they may convey abstract meanings 2 3 { 2 3 . kapp-. parts of human body (e. simply implies the prefixing of ma. "two"). They corre­ spond all to the root morphemes of the same type (§28. The bicon­ sonantal "simple" patterns can have either a short vowel (CvC). especially to anaptyxis (§27.210 MORPHOLOGY either when preformatives. can become kalab in the construct state of Assyro-Babylonian. or when the whole root morpheme or one of its radicals are reduplicated. "mother"). it develops to kāleb > keleb in Hebrew and to * kalab > kdlab in Aramaic.

6. in particular. 29. "holiness". these nouns are vocalized a-a. 29.10). "going out". It is worth noting that this segolization is still unknown to Origen (3rd century A. probably originate from *tlāt. Assyro-Babylonian sulmu. becoming šopēt and later šūfēt in Punic. "three". occasioned mainly by anaptyxis. "great". 29. ālikuim).g.g. e. kabid. It is subject to important vocalic changes in Hebrew and in Phoenician. talāt-. §33. 74. malik. halba < halab-ā. often substan­ tivized as agent nouns. The second short vowel is lost in Syriac and in Neo-Aramaic. "eight". Numerous vocalic changes occur in modem Arabic dialects. as Arabic tiqal. Dissyllables with long vowel in the second syllable (CvCvC) par­ tially derive from monosyllabic morpheme roots of the type C C vC (§28. "misleading".g. "bitter". Palaeosyrian ba-ša-nu-(-um) and Arabic batan(un) vs.g.and tmān-. hence "quitter".8. Hebrew nouns belonging to this category are called "segolates" since they are vocalized e-e. "great".g. while their Tiberian pausal vocalization is ā-e.). Others are adjectives. "peace.NOUN PATTERNS 211 as well. "greatness". Babylonian bašmu(m). in the Tiberian Masoretic tradition. "king". The patterns CvClC and CvCūC are predominantly adjectival or participial. to be compared with the broken plural tiqāl of the adjec­ tive taqll. "milk" (cf.7. e.). Arabic kablr. kātib. In the Babylonian tradition. wa-zi-um /wāśi'um/.D. "serpent". Ge'ez marir. "judge". E. *tāpit-. which reappears frequently in modern colloquials. "strong". participle "writing" in Aramaic and substantive "scribe" in Arabic. to be related to the broken plural kibār of the adjective kablr. Dissyllables with long vowel in the first syllable (CvCvC) are either active participles of triconsonantal verbs (CāCiC). well-being".7).g. kà-šè-bù(-um) /kādibum/. e. "liver". with two segols. Hebrew 'āsūm.g. hence "liar". from malk-. barely attested in the Latin transcriptions of Jerome (348-420 A.g. or patterns very rare outside Arabic. e. Dissyllables with short vowels of the type CvCvC may be vari­ ants of the preceding group.from kabd-. participle "going" and substantive "envoy" in Assyro-Baby­ lonian. E. where kerem appears as charm and zemer as zambr. and appar­ ently unknown to the author of the Latin transcriptions of Hebrew in the 10thcentury Ripoll Ms. e. CvClC. or kibar. Arabic quds. while the present colloquial forms of Damascus are tlāt. is used in Aramaic as a passive participle { 2 3 . This pattern is attested also in Palaeosyrian. "heaviness". e. partially substantivized. and tamān-.and *tmān-. "heavy".D.

g. CuCūC.6). while kawkab. attested already in Sabaic (kwkb). "the small ones". e.2).g. "olibanum". becomes kiblr in Cairene Arabic and gblr in a "Mesopotamian" dialect.g. *Kutar. "belt". Assyro-Babylon­ ian qināzu(m). and in Sabaic hwlm vs. "killed". "lime-plaster" (cf.g. Hebrew 'Is). e. also in kawtar. 'awfar. "killed" in Arabic. while the same function is assumed in Hebrew by the type CvCūC. Dissyllables with diphthong in the first syllable (Caw/yCaC) may have different origins. "rice". "smallness". the short vowels may disappear or be affected by a qualitative change. "man" (cf. "garlic". kdtlb. A similar use of both stems occurs occasionally also in other languages. e. "small". The patterns CiCāC. "the great ones". "whip". siġār. Aramaic himār. Since early Andalusian is related somehow to ancient South Arabian (§8. Arabic hln). "generous". In modern dialects. are inter­ nal plurals (§31. a divine name. cf. hyn. karūbu. tyb. while 'awnuq or 'aynuq. Hebrew 'ēzor < *'izār.g. "dream".g. bihār. 'izām. West Semitic and South Arabian hykl or hykl goes back to Old Sumerian (§27. tawm. swr. the short vowel i is lost or changes into eld. E. b) Patterns with Diphthongs 29. Arabic ġlr). Monosyllables with diphthong appear in Semitic (e. 'ys . and CaCIC are used in Arabic to form broken plurals. from wafara.9. "donkey". is based ultimately on the biconsonantal reduplicated root kabkab-. šayham. siġar. kablr. Ge'ez qdnāt. "house").212 MORPHOLOGY (e.5). In some languages. They characterize the Andalusian dialect. "written"). fay sal.(§34. qatll. The diphthong aw appears instead of u in Andalusian Arabic lawbān for lubān. "day". particularly in Arabic: e.26-28) of nāqa in ancient Arabic dialects. known from Amorite onomastics. "arbiter". "seas"). especially the first one (e.g. bayt. "to abound". e. Arabic hulm. "she-camels". The same pattern CiCāC is employed also for tools and instruments. Arabic nitāq. "star". is used for 'ā'(d)dug. cf. "lake". "belt". see also §29. yawm. "donkey".g. "wool". 'izam.g. sawf for suf. "more abounding". Other Arabic examples are haydar.g. "men". Several nouns of this group are simply Arabic elatives introduced by 'a. and ā may become o. riġāl.g. gyr. Arabic sūra).g. the Sabaic patterns fyl mdjwl should probably be interpreted as/ay/ and fawl: e. e. "great". "time" (cf. "image" (cf. "greatness". "blessed" in AssyroBabylonian poetry. 1 . hawr. "hedgehog". qātūl.g. "belt". "ass". as suggested by rawz for ruzz. e.16). vs. which appears as an extension of the pattern CiCaC (e. Arabic tlb). Ethiopic 'aydug. Neo-Aramaic hmāra. "scent" (cf.

Tuareg a-fdrrad. e. šappīr. "sailor" in Aramaic. e.g. dabbūhā. qaddūs or quddūs. "(are) outstanding". "statuette" in Sabaic. qattanu. "young boy" in Aramaic.NOUN PATTERNS 213 29.g. "bolt"). "butcher". e.g. The vowel ā should normally have changed into 6 in Hebrew. c) Patterns extended by Gemination 29. Dissyllables with the diphthong -ay. Stems of the types CjVC C vC and CjVC C vC are also employed throughout the Semitic languages to indi­ cate adjectives with intensive meaning. Clear traces of this pattern subsist in LibycoBerber. "territories"). rakkāb. "grave-digger". but belongs then to different semantic fields. pillar". Aramaic ptwr. e. "horse­ man" in Hebrew.10.g. "interpreter (of dreams)". Most of these stems occur also in Assyro-Babylonian. "sacrificer". "column. naġġār. but it is also encountered in proper names. "workman" in Ge'ez. "drunk­ ard". "harness collar". but this did not happen for some unknown reason. "very small" in Assyro-Babylonian. e. "piglet". da-nu-nu Idannūrìul.in the second syllable (CvCayC) and a dissimilated vowel u<a'm the first one are largely used as diminutives. "small dog" in Arabic. while Assyro-Babylonian uses patterns with vowels ī / ū for the same purpose (e. and /tubbūhu/ in Palaeosyrian. It might be identical there with the Arabic fi"awl pattern which occurs in some diminutives. "barber" in Assyro-Babylonian. gazzūzā. hasslnu. "contusion".g. a-nabbal. a kind of "bolt").g. The pattern C aC C ūC is used instead of C aC C āC in Assyro-Babylonian šakkūru.11. "carpenter" in Arabic. especially in Arabic. The pattern appears also in Hebrew 'ammūd. 'iġġawl. gazzūrā. gabbār. i f Tù-bù-hu. habbūrā. The pattern/V/ occurs in Epigraphic South Arabian mainly as a broken plural stem (e. kaddān. "beautiful" in Aramaic. The pattern C^aC C āC is largely used in Semitic for names of professions. "shearer". perhaps slym. sikkūru. zammām. should belong here too. mdwr.'À-da means "Very slaughterous is Hadda". e.g. "small calf". The same noun stem is employed for tools or instruments in colloquial Arabic (e. "sweeper". or with one of the diphthongs -aw-1-ay-. "axe". gallābu{m). This pattern occurs in Libyco2 2 3 l 2 2 3 ] 1 2 z 2 2 3 2 2 3 á . and in a number of Syriac nouns. 'addīr.g. "powerful" in Hebrew and in Phoenician. "most holy" in Ara­ bic. either with short or long second vowel. ni-bù-hu /nibbūġu/.. 'ulaym. "(are) very strong". Twelve different noun stems with geminated second radical consonant are attested in Arabic. kulayb. while their number is somewhat reduced in other languages.g. mallāh. hinnaws.

or ahuzzatuim). "old". The 2 2 3 2 3 3 2 3 2 3 . gursidakku(m). Other nouns are Semitic.g. The pattern with complete reduplication of base is easily recognizable also in Libyco-Berber.g. "white". CjVC C vC ). e. "skull". e. "marriage". Noun stems with reduplication either of the second or third con­ sonant of the root morpheme (CjVC vC vC . Hebrew baqbūq. "chain" in Ge'ez. and kdlakkā. or of both of them (CjVC vC C vC ). qaqqadu(m) < *qadqadu(m). "splinter". are two Sumerian loanwords in Akkadian.g. "oak". as kunukku(m). "wheel. in Phoenician. and sirsur. "flask". "butterfly" in Neo-Aramaic. with vocalic dissimilation. siyyūd. is unknown. a-wdssar. changed into kakkabu(m) or 'kawkab > kokab in other languages. hdrrād.g. In Arabic there are a few nouns and adjectives of these pat­ terns. e.g. E. "game". nesakkuini). Aramaic gulgultā (Greek ToÀyoGá). "broker".g. e. "wharf. a suffix -akku{m) or -ikku{m) was added to several Sumerian loanwords. and Arabic ġalġala(tun). "galbanum". "star" in Amorite. "raft". Ara­ maic nouns like pdrakkā. "flour basket". Dissyllables with geminated third consonant occur chiefly in Old Akkadian. Hebrew gulgolet. The gemination is lost in Neo-Aramaic (e. gizzūr. globe". e. The vowel of the reduplicated base may change by dissimilation. with different vow­ els. Mishnaic Hebrew pilpēl. galgal. and Aramaic. Hebrew. are attested in various Semitic languages. "head" in Assyro-Babylonian. sdttār. "altar". in Assyro-Babylonian. d) Patterns extended by Reduplication 29. and in Arabic. pirpira. "piece". and haluppu(m). palāha < pallāhā. "skull". while the etymology of Assyro-Baby­ lonian arammu. l 2 2 ?> l 2 2 3 29. like in sirsur. "vast (ocean)". ta-kdlkabba. e.13. baluhhu(m).14. "new"). ramp". sansal < *šalšal. In Old Akkadian. e. and tâ-kdrkort.214 MORPHOLOGY Berber adjectives as well. Patterns with reduplicated root morphemes are attested in most Semitic languages.g. "seal". "(non-precious) metal". a-mdllal. The pattern C dC C āC is used in Tigre to form names of products or results of actions. 29. "occiput".g. are borrowed from Assyro-Babylonian. "city ruler".12. e.g. ġitamm. išši'akku(m).g. kabkab-. "what is slaughtered". The same meaning characterizes the C iC C ūC pattern in Samaritan Ara­ maic.g. "worker") and in some modern Ethiopic dialects (e. filizz. a priest. especially in Tuareg. both related to Assyro-Babylonian gulgull(at)u(m). "pepper". haġis < *hâġgis.

"babbling"). the executor and the heir of my father In Pre-Islamic North Arabian.g. e. nāhiru in Assyro-Babylonian). 'af'āl. "movement"). "withered ". šdharhar. "crooked paths"). cf. 'sdq w-yrt. 'arakrak.g. and 'anhb. Besides. "green"). especially with 'sdq. 'by. this pattern may lead to the splitting of the long second radical.g. 'âqalqallot. šdlamhmā. . colour names (e.g. "scor­ pion") and in modern Ethiopian tongues.g.. "brightness. in Hebrew (e. 'araġ. morning light"). in Syriac (e. "great".. Both denom­ inative and deverbal derivatives are represented in this large group of patterns. ydraqraq. m-. 'a'raġ.g. 'af'ilat. Arabic 'asdaq. 'af'ilā(')w are 2 7 7 .. "ant"). annabu). Patterns with reduplicated second and third consonants of the root morpheme occur sporadically in Aramaic (e. "hare" in several Semitic languages (but cf. Noun stems are extended by various prefixes. in Arabic (e. "green"). both attested in Ugaritic. and bodily qualifications (e. and Š-. "cruel"). "reddish". 'ahmar. infixes. "nobler. "shellfish " (cf. in Aramaic (e. in Tigre (e. 'anhr.g.g. The noun stems 'af'ul.. comparatives or superlatives (e.g. in Tigrinya (e. e. wa-. 'af'ūl.5). "lameness"). Patterns with a prosthetic vowel introduced by '. "narwhal" (cf. šdparpārā. in Amharic talallaq. 'akzār. n-. "noble"). The second one is encountered in Assyro-Babylonian (e. Preformatives '-/-' 29. in Arabic (e. t-. ra nān. " I .NOUN PATTERNS 215 first pattern occurs rarely in Assyro-Babylonian (e. hamalmāl. "blackish". hatamtam. The most frequent of these is 'af'al which forms elatives (§34. 'akram. sdwunwun. very noble".g. in Hebrew (e. kulbābu.16.g.g. greenish". "executor" (cf. nibu in Assyro-Babylonian).e. karlm. "complete"). "lint").are well rep­ resented in Arabic. The pattern is attested also in Lihyānite and in Nabataean Aramaic. The main prefixes consist in a prosthetic vowel introduced by ' or sometimes '. "yellowish. this pattern is used in Hebrew for diminutives of colour names: 'âdamdām.15.g... "red"). A few Hebrew adjectives are related to these cate­ gories (e. "lame". šumlūl.g. "the most reli­ able"). "small amount"). martūt.g. i. 'arnab. 's ll /'aślal/ against Arabic 'ašall.g. in the morphemes ya-. in Ge'ez (e. <a e) Patterns with Preformatives and Infixes 29. e. cf. V . e.g. The same pattern is used also for some ani­ mal names. zuqaqlpu. "thick"). and suf­ fixes.g.g.g.

so m. "slough" of a snake in Arabic). E. However. yahmūr. Hebrew. in South Arabian. "prayer". "hammer". nouns of instrument are formed by the prefix wā. from dānāgām. "scorpion" in Hebrew. and yabrūh. from nàtàràm. Yarmūk.g. "sparrow. Ethiopic.g. just like the instrumental and locative functions of the ergative case are closely related. Preformatives w-lm-ln- 29. "filter".8).20.17.g. time.that is used in the various Ethiopian languages to form verbal nouns. 'aqrab. the morpheme m. Ygrš. There are few nominal patterns having a particular meaning in South Ethiopic. The patterns with preformatives ma-/mi-/mu.and the suffix -ya.216 MORPHOLOGY employed in Arabic.26-28). small bird" in Arabic (cf.8). "Expeller". e.g. ring".are confined to names of animals and plants.derives from mâ. išdihu(m). "mandrake". including nouns of place. 'qšr. In the Chaha dialect of Gurage.18. 29. original name of Medina. "profit". e.19. agent. §31. 29. Patterns with preformative ya. name given to a staff in Ugaritic. The shift could simply be explained by the tendency of Gurage to change m to w (§ 11. "birds".belong therefore to the categories discussed in §29. The nouns with prefix wa. a kind of snake in Ugaritic (cf. however. verbal nouns. The . insabtu(m) or ansabtu{m). Yatrib. qišr. and in North Ethiopic to form "broken" or internal plurals (cf.has an instrumental function and expresses the instrument or the means by which one performs an action. and participles. wâdràgya. Preformative ya- 29. 'usfūr. in Hebrew). "locust" in Sabaic. ipteru(m).have the widest possible range of meanings. and Arabic. wāntiya. instrument.may also express the place where or the time when the action occurs.20 ff. "deer". both in Aramaic. Patterns with prosthetic iare attested in Assyro-Babylonian. since the antelope or gazelle is the holy animal of this deity. 'rgl. sippor. " i t melted". and to proper names. ikribuim). "ransom". Aramaic. "he h i t " . e. Basically. Arabic. "ear-ring. Some animal names containing the consonant r have a pros­ thetic vowel introduced by ' instead of ' (§27. This morpheme wâ. Yarmouk river. Per­ haps the divine name 'Attar is to be explained in the same way.

má-kás-híi. from wrh. "dwelling". "instrument of killing". "market place" (cf. viz.g. miftāh. midbār. and mar-a-tum /mar'aytum/. Punic *ma'gal > Latin magalia. Various nouns of agent are formed with m. má-kás-áa. a a 29. e. mdbyāt. Arabic manāh. etc. "west" in Hebrew.g. "fat".g.is likewise used to form nouns of place in Hebrew (e.g. the prefix mi. mdsraq. mišqāl. mdšrāq. grave"). several nouns in má. "halting place. in Punic (e. "dining-room"). "place" in Phoenician. mizān. "to k i l l ' .g. mahdár. mdkwāl. In Hausa. "weight"). as apparent in place names like MápKx(3a (Mryb) or Maicpa (Myf't). Hebrew mānd h. but other languages furnish valuable information con­ cerning the use of different patterns. from wnh. "east". "tower" in Ge'ez. maqom. "dam structures". maškanum. maġlis. miškab. "pas­ ture-land" in Palaeosyrian. i. also in Ugaritic. the prefix mi. "temple").> n. má-kà-híi. and in Amharic (e. "weapon". It is heavily exploited in scientific and other modern coinages. "home. "to anoint". but rarely in Arabic where a noun like mihrāb. is borrowed from ancient South Arabian mhrb. mismār. e. . "kingdom" in Amharic. and nouns of place. "nail". mrht.g.. "height. The same use is attested in Hebrew (e. e. mnht. "to clothe one's self". ma-a-al-tum (cf. "dwelling place". "enclosure"). mizrāh. "key".g. "east"). "east"). "killer". Egyptian mnw. from hni. as shown by the syllabic spellings ma-ah-ha[-du] (m'ahd).g.g. "city".26). often in Ge'ez (e. mana. "bed. "settlement" in Old Akkadian. particularly with nouns of instrument. ma' rāb. "site of killing". mhnt. step". thus in ancient Egyptian. "desert". "dwelling" in Assyro-Babylonian.g.is also attested in the other branches of AfroAsiatic. resting place"). in Aramaic (e. "hiding place"). "place of sacrifices. mdrfaq. e. "to row".e. altar" in Aramaic. mūšabu(m) < *mawšabu(m). f a 29. and in ancient South Arabian. "skate. madbah.22. "ferry-boat". Instead. e.g. ma-sa-batum /mada'bātum/. "audience room". although no general rules can be established. "conference room.derive from the verbal root kas-. The prefix ma. mizlaġ. "sanctuary". court" in Arabic. and magālā. mahfad.in Libyco-Berber (§22. house" (cf. The prefix m. "clothing".forms Semitic nouns of place. maqdàs.21. However. Hebrew ma'gāl "encampment". "prayer niche". ski". in Tigre (e.. occur in Cushitic.is largely employed in Arabic for nouns of instrument. ma' leh). in Qemant-Qwara. myqdš.g. "scale". e. and màngdst.g.NOUN PATTERNS 217 unvocalized Ugaritic and South Arabian texts prevent us from further specifications. both in Oromo.

218 MORPHOLOGY which also employs the prefix ma.g. "chariot". in Amharic. One should also notice that the local and instrumental acceptations are sometimes difficult to distinguish.g. "writing implement".g. a-maddaz.g. in Palaeosyrian. "burning" in Assyro-Babylonian.when the root morpheme contains a labial. "book.g. muslālu. is one of the defence means and a place (cf. "anchor". miġzal. can be regarded as a tool and as the place where the driver sits or stands. Both cases occur mainly in East Semitic. and verbal substantives or abstracts. masaddaqi. e. "fork".26. mušpalu. especially Aramaic infinitives of the basic stem. This semantic use of the pattern occurs also in Libyco-Berber (e. maġzal.for the same purpose (e. ma-qar-tum /maqqartum/. "sieve". "dress").g. mazlēg. "chisel". "battle-axe" in Old Akkadian.(e. from the Assyro-Babylonian verb nadāru. "closeness" in Arabic. "spoon"). maktabi.g. "set of javelins"). Ge'ez (e. "midday") and of extension (e. "sum" in Assyro-Babylonian. 29.can either derive by dissimilation from stems with prefix m.g. maptē h. mahzīt. also infinitives (§42.g.> n. maqraba-. "spindle". munhul I munhal.g. 9).also serve to form various participles (§42. 29. can be consid­ ered as an implement and as the surface on which something can be seen.23. malhdq. "judgement" in Hebrew. "mirror"..14-16) and derived verbal stems in Neo-Aramaic. maqlūm. the suffix -/ is attached to the stem (e. mišpāt. Some substantives are attested in Arabic with different vocal­ izations of the prefix m-. "fearful".25. e.g. in Hebrew e. malbas. merkābā.g. "key"). muġzal. or be deverbal nouns formed from the verbal stem with preformative n(§41.24. . naplaqtum < *maplaqtum. "to be wild". and migdol. "action". nanduru.). as does Palaeosyrian (e. "mallet"). whereas it appears in Arabic as a variant of mi. napharu(m) < *mapharu(m). mdgbar.is used in Assyro-Babylonian for nouns of time (e. mashaf. Phoenician *magddl). mushaf. mdhràt. ma-za-rí-gú /mazārīqu/. e. 29. "tower".g. mishaf. codex". mânka. and Amharic (e. "mercy".4. with alternative forms as ma-pá-hu(-um) /mappahum/ and na-pá-hu-um /nappahum/. The change m. 7. The prefix mu. "altar"). "depth"). borrowed as nphr into Aramaic. In Tigre. Patterns with prefix m. "bellows". Patterns with prefix n.15-19).of the prefix appears exceptionally in other Semitic languages. a 29.

e. §29. A professional or social meaning appears already in Palaeosyr­ ian and in Old Akkadian. e.C. mostly before velars (e. They characterize professional or social situations with reciprocal connotations. the change m > n occurs frequently in Berber dialects when the root morpheme contains a labial or the labiodental/. ma. in conformity with one of the basic functions of the affix r-.produce. from *tml. dnqolo.g. "cut").> wa-. "a great quantity of butter". "new-born crop" in Palaeosyrian. "roper". -ndġmar < -*md'mal. e. "hunter". "woodcutter". -ndzdam < -*mdsdam. 1 ì 29. by the nouns nbb-n < *i-mbabīn .29. from tdlfii).NOUN PATTERNS 219 and with the rare examples of nbl'at.g. also at . This phonetic development is attested as early as the 2nd century B. "to roast"). "chair". "roasted grain". and nšpt.27. as against Tigrinya q dr'a. from qollá.can appear in the Chaha dialect of Gurage before a collective noun in order to express a plural or a great quantity. takbaru.g. "cut­ ters" (root bbdy. "flames" in Ugaritic. This element na. "soldier" (cf.g. -ndhšam < -*mdhkam. they frequently form verbal nouns sig­ nifying an action. -nazum < -*masūm. "blessed.19). dntdlfit. "judgement" in Old Aramaic (cf.g. kind of "hawk". wáttaddàr < *māthaddàr. e. viz. tá-da-bí-lu /tadābìlu/. nouns derived from verbal stems. Although less common than the patterns with prefix m-.7). may occur in South Ethiopic. but also before other consonants (e. in Amharic dnq drarit. w w Preformative t- 29. and sometimes occur with nouns of place. lucky". tá-aš-tá-me-lum.are also widespread in Semitic. na-qdb. "divide").g. Another dissimilation. e.corresponds to Amharic dllâ with the n derived from an original *//.g. those with t. "frog". and nbt-n < *i-mbdttūri . wânfit < *manfit. a preformative nà. for the most part. Besides.g. also with animal qualifications. -nbarš < -mbārdk < -mubārak. tù-la-dì /tūladu/. e. "splitters" (root bdu / bdttu. In the Ethiopic dialects of the Gurage group. 29. -ndfsad < -*mdfsad. The patterns with preformatives ta-lti-ltu.28. "interpreter". "sieve". However. there are several nouns with a non-etymological prefix dn-. This prefix occurs sometimes in other modem Ethiopian languages as well. Amharic wâmbar < *mânbâr. "the man of the mourn­ ers" at Ebla. §11. in a Punico-Numidic inscription from Dougga. "fastener". "judge". "fattened sheep" in Assyro-Babylonian.

§41. in Ugaritic (e.g. tnhyt. in Arabic (e.220 MORPHOLOGY Ebla (cf. Hebrew for tdšāb.g. from w'm\ etc. tafsām. tá-er-iš-tù-um Itaheríttuml. e.31.are found in Assyro-Babylon­ ian (e.g. "supplier.20-32). Nouns of place with the prefix t. "courtyard").g. like Tadmer / Tadmor. "ornament" in Hebrew. in Ge'ez (e.g. "battle").g. semantic evolution into concrete nouns must be allowed for. "confession of sin"). metic". tradesman" from makāru(m). "south"). "resident. "store-keeper". tapšahu.g. LibycoBerber takarza). "layerage" or "soakage". "going"). tdmhdrt. "summing up"). "interpreter". from Imd. "help"). "fold". from dgr. "covering. Ugaritic does it for tdgr. "partition").(cf. in Ugaritic (e. Infix -Í- 29. in Syriac (e. "praise").g. tdhillā. tarbū(m). 29. tamhārum. "chosen". "work". tágbar. in AssyroBabylonian (e. "to give away".g.g. ta'dīrā. is implied already by Ugaritic targumyānu and Assyro-Babylonian targumānu.32.g. in Phoenician (e. in Old Akkadian (e.). "awful". talmīdu(m). Gafat and Gurage dàbbâlā. "perfect". takrik. from rgm. "disciple". in Hebrew (e. "offering"). tū'amu{m). in Amharic (e. tamkārum. tarmiktum.) and Batahra (fern.g. trbs. to speak". tamšīlum. "counsellor". in Neo-Aramaic (e. tafrlq. in South Arabian (e.g. "completing"). "twin". targlgu(m). garment"). tip'eret.signifying an action are well attested in Palaeosyrian (e. in Hebrew and in Aramaic "(e. tklt. tešmeštā. Despite its late appearance in texts. "pupil". talmld. Their existence in North Semitic is ascertained by the Amorite names Batahrum (masc. However.g. "dis­ ciple". to'ām. nominal patterns with infixed -toccur in East Semitic for adjectives with intensive meaning. pitluhu(m). e.30. trmmt. Assyro-Babylonian uses patterns with prefix ta. Verbal nouns in r. "ploughing". tarbāsu. "heel" in Amharic. formed by the addition of the suffix -yānu or -ānu (§29. in Aramaic (e. "to heap". "teaching"). "to utter. "service").g.g. gitmālu{m). that are formed with this prefix. 29. "twin". Apart from the participles and substantivized infinitives of the verbal stems with -t. There are also ancient Semitic place names. . tēmān < toyman. the noun targūm.for tamlāku(m). cf. "resting-place"). "evil-doer". "Palmyra". from wtb. tallaktu. "image" in Assyro-Babylonian.g. "to repeat").39).g. tàràkâz.

29. This pattern has similar functions in Libyco-Berber. šalhēbītā. "strong"). "lazy").NOUN PATTERNS 221 Preformative Š- 29. Tachelhit a-skdrz. Afformative -ān 29.g.g. "eastern"). Hebrew (e./ šu.g. as well as for some elatives (cf. Numerous proper names belonging to this group are found in Assyro-Babylonian (e. -ya. e. like šulhān.g. occurs especially with adjectives and proper names formed as adjectives. In Hebrew and Phoenician -ān usu­ ally changes into -on. built like the verbal stem with causative preformative. "completion". šagapūruim).35. "ploughshare". EXiouv. and qorbān.33. "to kindle".g. Phoenician (e. Sinnānu. 'lyn = EAioov. Adjectives with the suffix -ān occur in Arabic (e. "black". rahmān. a-siban.g. "luminous"). in the Ugaritic proper name Š'tqt.g. f) Patterns with Afformatives 29. šuklālā. there are some Hebrew nouns.g. "to complete". šūbultu(m). kaslān. and it develops later to -ūn (e. Aramaic (e. "present".g. "dismissal". Instead. -at. "bald". which is sometimes attached to other nom­ inal patterns. "flame". from šaklēl.and št. "Most High"). Neo-Aramaic (e. and broken plurals. Ugaritic (e. e.10). However. There are several noun patterns with afformatives or "nominalizers": -ān > -on > -ūn. with verbal substantives. from -kdrz-.g. "golden").are employed in East Semitic for verbal nouns of the stems with preformatives š. -it. E. -aym/n. -ā'u. diminutives. -akku. -o.(§41. and in Aramaic nouns derived from verbal safel stems (§41. where this change does not take place. -ut. The morpheme -ān. "She-who-removes(-evil)". "table". from šalhēb.34. -awl. -iy > I and -ay > ā. a-bdrkan. Apart from Assyro-Baby­ lonian there are a few examples of these patterns in Palaeosyrian sa-zuwa-tum lša(w)śu(')watuml. and šutābultu. "plough". qadmon. both from wabālu.g. "interpretation" of omens.g.g. -àhhā. "toothed"). Aramaic (e. . haylānā. borrowed into late Biblical Hebrew as šalhebet. is attested also in Libyco-Berber. Nūrānu. Neo-Aramaic uses a variant morpheme -ūn to denote diminutives (§29.29). from gbr I gpr. "Most High"). "to bring". "to plough".5). -nàr. "merciful"). Hebrew (e. Patterns with prefix ša. nouns of agent.36.8-10. e. This pattern.g. "offering". "very strong". *Dahbān. §34.38).

"son"). some nouns have a concrete meaning.g. were formed with a suffix -ūn (e. "greeting. "small dog". "growing old").g. or Arabic qur'ān from qara'a.g. bribe").g.222 MORPHOLOGY Šimšon. goes back either to the adjectival or to the diminutive function of the afformative. "sunny" or "small sun"). "to recite". hafaqān.41). "stomach-ache"). "heartbeat. "sickness").g. saltan. in Neo-Aramaic tūranāyā. zbln. 'aqrabān.g. "remembrance"). which is attached also to verbal forms (e. 'ahūnā. Amharic (e. kalbon. śah ron. in Hebrew (e.30. targumyānu. našūnā. šulmānu. fluttering"). ras'ān. especially in names from Ugarit (e. "interpreter". in Hebrew qadmonl. e. Phoenician 'In . and later still in the West. yalūnā. q antan. the gentilitial endings -iy > -I or -ay can also be added to the suffix -ān > -on. Some Neo-Aramaic words have nowadays lost their original diminutive meaning (e. Ugaritic (e. 29. However. in Modern Hebrew where diminutives end in -on with a feminine -onet (e. ġišmānī.39.g. qorbān. "(man) of Astarte". The number of adjectives in -āríi without any gentilitial connotation increases in Post-Classical Arabic.g. šoltān. amulet called "littlemoon").g. "(man) of A l w " .g. "authority". "power". Ibn Halfūri). "little fel­ low"). mērānu. "mountaineer". ruhānī.g. "table". cf. brūnā. "oriental". Aramaic (e. in Arabic (e. like Palaeosyrian i-a-lanu /'iyalānu/.g. A great many Arabic names of the first Islamic centuries. e.g. related to Hebrew 'yl. Diminutives in -ān > -on occur in Assyro-Babylonian (e. "youngster".g. sipron. Ugaritic tlhn and Hebrew šulhān. but also in nouns of agent (e. puqdānā. "offering". "large tree".g. "sunny" or "small sun"). "spiritual". "whelp"). in Phoenician 'štrny. Verbal nouns or abstracts in -ān > -on are attested in AssyroBabylonian (e. Ibn Haldūn.g.g. w 29. zikkāron.37. Arabic (e.g.g. The nouns of agent formed mainly from active participles by addition of the suffix -ān are well represented in Assyro-Babylonian and . but they testify to the existence of a suffix -ūn. present. Šapšìyānu. 'Abdūrì) and in the Neo-Aramaic diminutives (e. Ge'ez (e. Yamlikān in Amorite). Conversely.39). which should be compared with the suffix -ūn appearing in Syriac names (e. "small scorpion"). The widely used hypocoristic suffix -ān(um) of personal names.g. "brother". The suffix -ān can be added to the afformative -iy (§29. "order"). "booklet"). variant of -ān. §29. a 29. "corpu­ lent".'alon.38. Hebrew (e. in South Arabian 'Iwny. "god".

"Horus of Nhn". The difference between -iy. hdrfān. Su-mi-ia) from the feminine -aya (e. The Amorite anthroponomy appears to distinguish the masculine hypocoristic suffix -iya (e. "writer". nādinānuirrì). roshān. the so-called nisba. HrNhny. katbānā.g.seems to have been originally dialectal. In subsequent times. in Assyro-Babylonian. subyān. this distinction lacks clarity.g.g. used also to form the genitive marker -/ (§32. and probably in Epigraphic South Arabian. Ma-ar-sa-ia. rhty. "seller". They most commonly signify an individual member of a social group and they are widely used as gentilitial and hypocoristic endings with Semitic and even non-Semitic proper names. šarrāqānu. are attested also in ancient Egyptian (e. "steers­ man". in the Palaeosyrian texts from Ebla one finds the suffix -ay. qunwān. etc. in Neo-Aramaic. 29. "legs". This use of the afformative -ān should be dis­ tinguished from the external plural -ān attested in several Semitic lan­ guages (§31. At Ugarit. "the Ombite". may have originated from a postposition. from Nbt. from Sumerian u m m i a.does not exist in Old Akkadian and in stan­ dard Assyro-Babylonian before the Middle Assyrian period. "to steer". which is also used later at Mari (e. A parallel difference characterizes the hypocoristic ending which seems to be -ay. Ia{ f pu-ha-ia). "boys"). ġizlān. gozlān. Ekallātayum. Afformatives -iy I -ay I -āwī I -ya I -iyya 29.42) or to the first singular pronominal suffix. also in Mishnaic Hebrew.in Old Akkadian (e. plur. Nbty.g. E.and -ay. hmy. Ar-ša-ti-a). The genitilitial ending. "bunch of dates". [orāna] < 'abrānā. Iš-ra-ià). Instead.g. sīgān. The relation between these different functions of the suffix is not evident. Zi-im-ri-ia. "lambs"). ìr-su-ti-a. "Ombos"). contrary to the opinion of some authors. They do not imply any particular individualization of the person acting.g. whereas it is -iy. En-na-ià. from rht. Ma-ti-ya). "to wash". from hm. Sukkalliya. "robber". also in mod­ ern colloquials (e. In fact. and later in Old Assyrian and in Old Babylonian (e. "passer-by".(kà-na-na-um/im /Kana'nayum/. In this hypothesis.7). A-si-ya. Ha-ra-ià.g. "murderer". but -aya is generally used for feminine names.41.12). "master".> -ā. the hypocoristic ending should have a different origin and be related either to a form of diminutive (§29. Elahutayum). extended to professional qualifications. Broken plurals in -ān are found in Arabic (e. The gentilitial or adjectival suffixes -iy. "thief". ġazāl. ummiānu(m).40. "gazelle".NOUN PATTERNS 223 in Neo-Aramaic. "Canaanaean").g. Ir-bí-ia.> -I and -ay. "washerman".g. there is a clearly cut division between . the gentilitial formation -ay.at Ebla (e.g.

hsw. "military".44. "magic". sanbatāwi. Sīdonī. "town". e. daggetāy. Makkāwī besides Makkī. How­ ever. Misrī. "calf". from ba'at. "[Born] on a holiday"). e. 29. e.g. an ancient use of this afformative to develop diminutives may be attested. The suffix -I can be added to the afformative -ān > -on (§29. "craftsman". while the feminine suffix -at or -it is added to masculine nouns. "little well" (cf. as l'bs 'trw /'Attarāwī/. "goat". "large cave".g. "official". "Meccan". world". Kaśday. e. Haggay.g. "craft". Under the influence of Aramaic. hk3w. "little house". ba'āy.g. "maritime". augmentatives can be derived only from nouns ending in -at by substituting the suffix -āy for this morpheme. "cave". 'anglizawi.g. "(belonging) to Abūsu.g. rawwasi. 29. Tigre 'dgelāy. Safāwī. e. in Ge'ez (e. e.42. "to sing". the suffix -iy > -I is used in Arabic (e. "magician". to form pejoratives. This suffix may have existed also in ancient Egyptian as suggested. E.42) are also used to form pejoratives and augmentatives.). "state". "earthly. (man) from 'Attara" (Latin Otthara). Examples occur already in Safaitic. Afformatives employed in Tigre to produce diminutives (§29. e. by Arabic ġady and Hebrew gadī. "runner"). against 'anesāy. from mangast. 'agelat. "ruined house". on the other. from dagge. "Egyptian"). where -ay is often written -a-a. "singer".224 MORPHOLOGY Aramaic. the hypocoristic ending -ay appears frequently in Hebrew names and in Assyro-Babylonian onomastics. "little calf" (fern. in Hebrew (e. betāy. "soldier". A related suffix -awl appears in Arabic with place names end­ ing in -al-ā (e. "ploughman". the masculine suffix -āy is added to feminine nouns. "worth­ less man".g. worldly". wâttaddàrawi. from hk3. when compared with Libyco-Berber -*ġatt-. on the one side. "little man". by hmww. The corresponding Tigre feminine endings -at and -it go probably back to -*ayt > -āt or -It. "Englishman") and in Amharic. bāhrāwi. from dunyā. however.. 'dnesat. The hypocoristic function of -ay may have a bearing on its use in forming modern Ethiopic diminutives.43. from 'agāl. Instead.36). It is also productive in Tigrinya (e.g. Haggī. mangastawi. 29. "young goat". "earth. "[Born] on a holiday"). from hśi. "Chaldaean"..g. from wâttaddàr.43).g. krastiyanāwi.g.g. harrasi. 'dlatit. against betatit. and the bulk of West and South Semitic languages. from hmt. §29. dunyāwī. "Sabbatical". . "man from Safā") and with some other nouns ending in -a/-ā.g. However. e. "little calf". While the Aramaic gentilitial and hypocoristic ending is -ay (e. "small town".g. "Christ­ ian". "Sidonian". This suffix -āwi is widely used in Ge'ez.g.

e. alonuth. "mother")..15) gave rise in many Phoenician dialects to a new -āt that became -dt and later -ūt (§21. i3w. "queen". Some of the abstract nouns thus formed were subsequently used as masculine concrete nouns.in some modern Gurage dialects of South Ethiopic to form nouns of instrument. with a probably original long vowel. "goddesses"). most nouns with a suffix in -t came to be regarded as feminine. e. "elders" (cf.NOUN PATTERNS 225 29. "balance").15).g. ummata. Oromo abbotī.16). e. produces abstracts like. misdāqiyya. and the Hebrew singular hokmot. 'usūliyya. when attached to masculine root morphemes. The feminine plural ending was -āt-. from sarru. e. may witness to the same phenomenon.g. §31. in Assyro-Babylonian). šarratu. 'amatu > 'amāt> 'amot. "holy things") have shifted to the category of abstract or collective substantives (cf. "hen". "-ism". Old Babylonian ummatum). "cock". Modern Arabic uses the suffix -iyya to coin the ever increasing number of abstract nouns corresponding to English "-ity". Phoenician mmlkt.1-3). it is quite possible that this ending is to be considered as an originally plural mor­ pheme.t. In Ge'ez. "king". indicating that some concrete plurals (e.g. "youth". Late Babylonian pāhātu. "old age") and in concrete nouns (e.45. "people" (cf.g. Arabic hallfat-. e. mh3. nominal patterns with -t are attested in ancient Egyptian. "king". "Wisdom".46.g. Outside Semitic. "fundamentalism".g. "governor". na'asāt. Cushitic nouns belonging to an old stock seem to confirm the originally collective meaning of the pattern.47. but also in abstracts (e.g. "holiness".19). 29. the feminine singular is formed by prefixing and suffixing t. they produce feminine nouns (§30. "king". Some Palestinian place names. By analogy. Numidic gldt. but it became -ot in Hebrew and in Phoenician.g. 'Aštārdt. "-ness". "maid". Afformatives in -/ 29.g. not only in truly femi­ nine nouns (e. .t.in the singular (e. §29. successor". Noun stems with suffixes -at f-ut f-it have often an abstract or collective meaning and. "credibility".t. m3w. from e-kahi. Tuareg te-kahi-t. wàdrāgya.g. like 'Anātot. Amorite abbūtu). Rāmdt. and even -ūt in Punic (e.g. In LibycoBerber. The usual feminine ending is -at. e. qddsāt. "hammer" (cf.13).g. e. Since this derivational suffix -āt is homophonous with the "feminine" plural mark (§31.g. but the tonelengthening of a after the dropping of case endings (§32. "deputy. the ending -āt. The suffix -ya is used with the prefix wà.

as mu'assasāt. North Ethiopic has an -o suffix used with concrete and abstract nouns. §31. as e.g. Ge'ez fatlo. sdrqot. "covering"). malakūt. "institutions". e. and in Amharic (e.g. mmlkt. "a grain of wheat". "medi­ cine"). .g. in Hebrew and in Phoenician (e. "jar". "beginning"). in East Semitic. although a formation with addition of -at > -ā(h) is also attested (e.g. šarrūtum. zàndo. rarely in Ge'ez (e. Also the exceptional Arabic adjectives in -ūt are borrowed from Aramaic substantives. dannūtu(m) from dannuim). ù-mu-tum /'ūmūtum/. "a grape".g. qadāmit. most likely in Phoenician (e. hirut. in Ge'ez (e. Mo'âbiyyā). Tigre mdhro. "teaching". "kingship"). Abstract substantives in -ūt occur in Palaeosyrian (e. "longing".g. "pregnant". powerful" (cf. manage­ able".g. "future"). "little glass". e. tarabūt.16. malkūt.g. e. "kingship"). but it served sub­ sequently as -ūt to form abstracts derived from other roots and.g. Modern written Arabic widely uses the -āt suffix to designate concrete or abstract entities. Modern Hebrew adds -it to form diminutives. "spell-binding"). "training".226 MORPHOLOGY The same remote origin might be ascribed to Amharic abstracts like nafqot. in Old Akkadian and in Assyro-Babylonian (e. "royalty" > "king"). 29. in Hebrew and in Aramaic (e.48. rē'šīt. kasit.g. Hebrew kasūt. "little pitcher". 29. Other Afformatives 29. "python". "trained. "beginning"). The original suffix -ut was probably added initially to root mor­ phemes ending in -ū (e. 34.4).g. "kingship"). àmhkot. "Moabite"). e. "strong. that were perhaps borrowed from North Ethiopic. from tarbūt.g. the suffix -It forms singulatives (§31. In Arabic. or taqallubāt. in Aramaic (e. "theft".g. "goodness") and in Arabic (e. "domination". Md'ābīt. "spinning". kadit. these nouns are rightly regarded as loanwords from Aramaic.49. and they have no relation whatsoever to the East Semitic masculine plural of the adjectives.50. "crying") and to gentilitials in -iy > ī (§29. Palaeosyrian 'à-rí-tum /harītum/. màdhanit. Tigrinya 'atro. 'ahârīt. "fluctuations". it produced the masculine plural forms of adjectives. "coercion". In Neo-Aramaic. It is attested also in Amharic.g. It served subsequently as -ūt to form abstracts in Palaeosyrian (e. "fat­ tening"). Hebrew bdklt. and they are masculine.g.40).g. a-za-me-tu /lazamītu/. The suffix -it was most likely added originally to root mor­ phemes ending in -I (e.41) in order to form their feminine (e. hitītā. e.g.g.g. 'anbītā.g.g.

"little donkey". gazetâMa. from Arabic kataba. it is paralleled by the Arabic -ā'u class (e. This formation preserves the final vowel of the Sumerian word.NOUN PATTERNS 227 kâbāro. Abstract nouns are formed by adding the suffix -{a)nnat to concrete nouns and to adjectives. there is an additional suffix -nàr serving to form abstract nouns.51. "correspondent".g. nàz-nàr. "president". e. from muziqa.g.g.g. "parentage". but follows the Old Akkadian nominal pattern purussā' (e. The suffix -cik has a Pol­ ish origin. asportânna. e. derived from the Sumerian genitive morpheme -a k (e. In modern Amharic. agaranha.g. "comrade". e. the suffix -iko is Judaeo-Spanish or Ladino. "maturity". followed by -I'um > -urn (e. Thus. Sumerian loanwords are characterized in Old Akkadian and in Assyro-Babylonian. "little lad". "music". basahanna and basahannat.g. "little father". the suffix is added also to Ethiopic words. 'dbale. "destruction"). e. "tambourine". issV akkum from ensi. "to be heavy". hirbā'u. 'imale. bahúrcik. 'attalātā'u. habériko. e.g. "musician". "sanctity". The two suffixes are interchangeable. among others. e. "teacher".52).g. the Yiddish suffix -le is used in nursery words and in proper names.g. katabānnà. "Tuesday". e. muziqànna. kussl'um from g u z i . "throne") or -ā'um > -ūm (e. prezidentdnnāt. "sport".g. where a suffix -anna or -dnnat is used to form abstract nouns. Nouns borrowed from another language are often extended by particular suffixes. huluqqā'um. from asport. "partner"). e.g. nouns of agent are formed by adding the suffix -(a)nna to loanwords denoting objects and occupations. waladannat.g.7). §29. Thus. tappā'um from taba. qaddsdnna.g. Adammā'um). "newspaper". or by a geminated last radical. Parallel formations occur in Tigre. "pedestrian". However.52. "to write" (cf. "sportsman". but is attested in Palaeosyrian (cf.g. e. 29. Besides. from naza-m. Suffixes of non-Semitic origin are added in Modem Hebrew and in Neo-Aramaic to Semitic nouns to form diminutives. which is infrequent in later East Semitic.g. "presidency". In Gurage dialects of South Ethiopic. "city ruler"). Instead. Its origin is not clear. "little mama". . Several Arabic loanwords are used in South Ethiopic idioms of the Gurage group with the palatalized suffix -ānnā (and variants). e. hamorCik. d 29.12). by the suffix -akkum (§29. from gazeta. which is employed also for loanwords borrowed from Aramaic with the ending -ā' of the emphatic state (§ 33. "chameleon"). from prezident.g. "heaviness".53. 29. e.

known also with an early change ay > ā (§22. from the Egyptian execration texts of the early second millenium B. but toponyms formed with an archaic locative morpheme which probably consists of the gentilitial suffix -ay /-iy with the genitive ending -«-. e. Arabic place names of Syria-Palestine ending in -in may have the same origin. or 'štr'sy. "vineyard". "Egyptians". The process of word formation in Semitic includes coinage of nominal compounds comparable to English words like "motel" from "motor + hotel". and Horonayim (Hebrew) = Hwrnn (Moabite). " A k k o " (E 49). 'ky /'Akkay/. "shepherd". and some Phoenician divine names as Mlqrt. The dis­ tinction of the cases is still preserved in the place name Ta-al-ha-yu-um (nom. from 'štrt ("Astarte") and 'sy ("Isis"). Afformative -ayimln of Place Names 29. The later reduction -iyi. "Ebla" (E 43). from *g ord ("neighbour") w w w w w . N-h-r-n in Egyptian texts of the New Kingdom. The latter usage was not general. from milk ("king") + qart ("city"). known especially from the Mari documents. Ta-al-ha-ya-am (ace).g. "the day before yesterday". from ištēn ("one") + ešeret ("ten". followed by the mimation or the nunation. from ab a ("father") + àlam ā ("cow"). and Nahrīn in Aramaic.54. "the gods of Egypt". "Melqart". e.allows at Ugarit for the orthographical distinction between the petrified country name Msrm {/Musrlmf). "neighbour".C. "Egypt" as in 'il Msrm. Similar formations occur in Gafat. "small vineyard". Such blended nouns are no strictly modern innovations in Semitic. from šālos ("three") + yom ("day"). 'Ib(il)lin from Hebrew 'Iblayim. Ta-al-ha-yi-im or Ta-al-hi-yi-im (gen. A reduction ay > ā appears in Dotayin = Dotān. from karmā.g. "North Mesopotamia". fern. The same reduction explains the forms Na(h)rima/i in the Amarna correspondence. "Magdala". since numerals as Assyro-Babylonian istenseret. i.55.> -i. "eleven".and by the feminine gender indicated by the afformative -tā [-ta]. e. and the always productive gentilitial formation Msrym (/Musriyyūmaf). karm-ik-tā.e.).3).g. Mktry /Magdalay/ (E 5). "father of cows".. 'Énayim = 'Ēnām.g. g drâbetà.). adverbs like Biblical Hebrew šilšdm. as it appears.). belong to this category.228 MORPHOLOGY Neo-Aramaic diminutives are characterized by the affix -ik. abâlam à. while Hebrew has Nah rayim. The place names ending in -ayim l-īm /-ām or -ayin l-īn /-ān are no duals. e. with place names like Yb3y /Yìblay/. a g) Nominal Compounds < 29.

e. lit. dmmáġàtit. "car-lead­ ers". e. i. "lighthouse". for instance. but the plural marker (§31.g. "car drivers".1. masculine and feminine. "husband and wife" + plur. the Semitic noun has two genders. bal-dnna mist-odd. except for the cardinal numerals (§35.e. although Semitic nouns have preserved some traces of a completely different grammatical gen­ der institution (§30. Yet. Greek. while Semites speak under the influence of a two-gender language system. as 'ab. i. i. are condi­ tioned by three-gender language patterns. "ass" vs. "owners". migdalor. A particular feature of Libyco-Berber con- . "master of a house" + plur.e. from pārahtā ("winged animal") + lēle ("night"). marker. from somer ("guard") + tap ("little children"). Primarily gender has nothing to do with sex: human beings and animals with sexually distinct social or economic function have simply different names. Instead.GENDER 229 + bet ("house").vs. qeštīmāran. "babysitter". "well-known". People speaking Latin. from qešteh ("his bow") + māran ("our Lord"). "married couple". "father" vs. Sanskrit. gender correlates with sex only in those nouns where sex in expressed semantically. Blending of words to form new nominal compounds plays an increasing role in Modern Hebrew. smartaf. but Semitic pronouns and verbs carry gender characteristics in addition to the nouns. Gender 30. the femi­ nine is marked by the ending like in ancient Egyptian and in the other Afro-Asiatic languages. in historical times. "she-ass". bald bet-oéé.17) is added to the second element. from dm à ("mother") + ġátit ("mistress"). from migdal ("tower") + 'or ("light"). §64. màkina nàġi-w-oâd. maršema. "rainbow". parhalēle. "mother". 'imm-. Amharic has a considerable variety of compound nouns. "the bow of our Lord". marker. but their components are written as separate words and may even be joined by -anna.2-4) with a genitival qualifier (cf. "and". Only later was the attempt made to relate them according to their kinds and different patterns. "the winged animal of the night". e. lit. "famous". w B. "the owner of a name".5). lit. "bat". like the Romance languages. The most common formation consists of a noun in the construct state (§33. The masculine possesses no spe­ cial endings. but other compositions occur as well. and the Slavic languages. e.g. himār.g. "lady". However.4).vs.10-11). from mārā ("owner") + šemā ("name"). Comparable compounds are widely used also in Neo-Aramaic. 'atari-.

Tigre walat (< *waldat).in the plural of nouns of masculine singular does not demonstrate that the gender of the nouns in question is or became feminine. 30. 'atān. the Libyco-Berber i-prefix belongs to the case markers ta-.3. the appearance of so-called "feminine" endings with -t. e.g. — e. hams-u. hallfat-. Hence the absence of the -t. "fathers" in Hebrew. 'imm. t-ihf-t.to indicate female beings led by analogy to the attribution of the feminine gender to most nouns ending in -t-. e.g. the r-morpheme was used in Semitic to form collective and abstract nouns as well. Hebrew tarblt. "son". sarrutuim). . "elderly woman". in consequence. e. e. "queen". Such a proof can be provided instead by the gender of the adjectives. and even to the addition of -t. from wad (< *wald).230 MORPHOLOGY sists in both prefixing and suffixing the morpheme t in the singular. "mother". "woman".g. except for adjectives and participles that are not always recognizable by their form as such. "city". person".morpheme does not necessarily indicate masculine gender (cf. to the Semitic abstract nouns with both prefixed and suffixed V. "large profit". Originally it designated a female being only in nouns derived from a root morpheme signifying a male.> 'umm-. §30. common Semitic 'imm-. and imittu(m) as compared with common Semitic 'rś. "kingship" in Assyro-Babylonian.(§32. and verbal forms referring to the noun under consideration. "soul. "mother". Assyro-Babylonian šarratu{m). t-ussdn-t. 1 w 30. "man". napistu(m). as in ta-mġar-t.to some feminine nouns originally deprived of this mark. — or it was connected with the root by a vowel. "lady". Conversely. e. "daughter". "king­ ship".1).g. Other feminine nouns were not characterized by this ending -t-. pronouns. "deputation" > "deputy" in Arabic. The three formations in -at-. Old Assyrian and Old Babylonian takšītum. Hebrew malkut. and the resulting forms cannot be assimilated. -it. in Assyro-Babylonian bēltu{m). and ymn. tu-. either collective or abstract. ti. and in Phoeni­ cian qart. warq-u. Arabic malikat-. The morpheme -t. "right hand". 'ābdt.of the feminine or to their allophone td. or Palaeosyrian ì-ma-tum /'immatum/.g.2.was either attached directly to the root mor­ pheme. from malik-. e. "wild sow". -ut. A third pattern may occur with nominal bases ending in a conso­ nant cluster.> 'āton. participles. "queen". and Soddo (Gurage) dmmit vs.g. In fact.6). Gafat dtn it.are attested. "she-jackal". "she-ass". "yellow-green". Assyro-Babylonian ersetu{m). npš. like gabr-u. However.g. "earth". "king". "increase". Ge'ez bd'dsit. The widespread use of the morpheme -t.

Assyro-Babylonian unqu. and Arabic by the mater lectionis he. "sun". but it has masculine concord in Neo-Assyrian.5. "mother's sister". "fox". "kingship". A number of nouns may be either masculine or feminine. musahhinu. Hebrew 'iššah. "palmgrove".g. The quality of the vowel occurring between the last two conso­ nants of the base is not predictable but. the phonological solution may then be -CC+t. as a rule. "wide".6. 30.C. by Aramaic mšqy. as against malkut in the construct state. i like in hamistu. e. Aramaic. which was indicated in some lan­ guages as Hebrew. rarest is a. "watering-place". On the opposite. 30. The loss of the t in the endings -it and -ut is reflected.19-20). Aramaic millah. testify also to the elision of the feminine ending -t in ancient Egyptian. 30. in a restricted number of cases u. like in rapaštu.28). ta'lab.GENDER 231 "five". it is the same as that of the vowel which may occur in the same position in the stative (§38. Masculine concord may refer to species. Besides the sex of human and animal beings. "vine­ yard". e. Tigre bet. are attested in both genders. faras. Neo-Aramaic šimšā. and the formal con­ stitution of the noun with the ending -t-. The ending of the absolute state was thus reduced to -a. E.g. Arabic hālah. Assyro-Babylon­ ian nam. "mare". "stove". "other". is generally feminine. although it was consistently retained in the construct state (§33. can designate a "cow" in Hebrew. and lb' can be a "lion" or a "lioness" in Sabaic. hâmēšet < *hamistu. The vowel is. šor. "lady". Hebrew kerem. Instead of adding the morpheme -at-1-it-1-ut-. This pattern is clearly attested in East Semitic and it subsists in Hebrew under the form of the feminine segolata. from the base hamš-\ 'aheret < *'ahirtu.C. e. "river". "bull". on (§27. like in waruqtu. -CiCt-. or -CuCt-.3). "five". it was preserved in the cuneiform writing. Besides. like gdberet < *gabirtu.was eventually lost in many Semitic idioms from the first millennium B. rapš-u.g.g. "woman". in most cases. Animal names can be feminine when they designate a female. without any marking of the gender.4. can also mean "she-fox" in Arabic.. e. The elision of -t in late Assyro-Babylonian is attested by Aramaic transcriptions as rsh for Neo-Assyrian rēšāti.g. abstracting from sex. "house". some categories of nouns deter- . The gender variation may also be dialectal. "signet ring". The t of the most frequent ending -at. can also designate a "horse" in Arabic. Sabaic nhl. "word".> -CaCt-. and malkū. "first fruits". from the base gabr-. Transcriptions of the first millennium B. from the base 'ahr-.

"salt").g. In other words. "foot". tuppu(m). especially paired. as well as of nouns the gender of which changes when they are used in the plural. qarn-. "log of wood". "hand".1. or samnuirn). e. "stone". instances occur. has-. relatively small. "house". "face". girru(m). but for some semantic fields there is a def­ inite preference. Instead. "ear". In an Old Babylonian letter from Mari. "bowl").55) (cf. 17. "hand". although they are used in a proper sense. but they tend to be masculine in Assyro-Babylonian because of the usually masculine gender of the word ālu(m). names of cities and countries are gen­ erally feminine. napš-. "breath". e. Besides. the plural qar-na-at na-li. "tooth". However. šìnn-. and weapons tend to be treated in Ethiopic as mas­ culine. meteorological phenom­ ena.5). and riġl. but yacc mdshaf is feminine because it might refer to a favourite booklet.15). rūh-. since byt is a masculine noun in Hebrew. "horn". "foot". or toward which the speaker feels affection. just as Hebrew place names beginning with byt. then "throat" (§63. Even the feminine pronoun anā. "heart". rivers. "horns of roebuck". yah máshof. In Assyro-Babylonian e. Thus. "breath". where such nouns are treated as masculine. march". and gandāy. can be either masculine or feminine in the singular.g. Thus..157) and rgl (TAD 111. names of various tools are feminine (e. lyb.g. harb-.. almost any noun can be treated as feminine i f it refers to something female. In Amharic. " o i l " . Semitic languages show instances of nouns which can be either masculine or feminine (§30. may be applied in Addis Ababa to a male friend to show affection. "road. 30. "city". milh-. in Neo-Arabic kaff. are masculine.170). §31. In Ethiopic. is referred to by a masculine pronominal suffix ( A R M XIII.1. but can be masculine or feminine in the plural. etc. as well as names of different stuffs (e.16). "heart". 'ayn-.12) and yāmīn.1. 27. However. perhaps yad. are generally feminine in Semitic: 'udn-.g. Names of parts of the body. 'abn-. dialectal differences and even personal pref­ erences must obviously be taken into account. Nearly all other nouns occur in either gender. in Christian Palestinian Aramaic 'dr\ "arm". are masculine. "eye". rigl-. "sword". "this book".g.7). yad-. "hammer".1.g. is masculine in the singular. Tigre madoša.232 MORPHOLOGY mine their feminine gender. "palm of the hand". is masculine in the singular. 'pyn. In these questions. e. names of months. stars.7. in Hebrew yad (Ex. while its feminine plural . metals. gender usage is predictable only for nouns denoting human beings. E.g. "you". even in literary texts. "tablet". no generalizations are possible. is masculine. and libbu(m). in Aramaic 'yn (TAD 111. "right hand" (Prov.

"women of the same clan"). sing. "girl". set lag. It is interesting to recall here that no gender distinction based on sex existed in Sumerian that has a gender classification made on another basis entirely — that of animate and inanimate categories. In the Lowland East Cushitic languages of the "Sam" sub-group. 30. Other Ethiopian languages have gender specifiers for the male and female sex of animals and of human beings. One can observe in Tigre. "plants". e. "this one". arc. "so-and-so" from the masculine ebdrya.g. waned) ba. as shown by the many basic feminine nouns without any special morpheme. including "human beings".g. of fern. "animals". and there are traces of another class with the opposite gender-number relation. the -t morpheme characterizes some feminine proper names and it occurs in a couple of substantives (e.) wanaccit (plur. For the sex dif­ ferentiation of animals.8.. arast kutara. tdbat tagga. 30. "she-calf". substantives are divided into two to six classes or genders. gàrād. anger". "boy". "places". e.g. "he-calf". The distinction in gender is indi­ cated by the verb or the pronoun referring to the noun. "boy".9. ardst. Amharic wand lag. This opinion is apparently confirmed by the South Ethiopic idioms of the Gurage group which have no feminine mark.g. the formal distinction between masculine and femi­ nine is not an original feature of Semitic languages. etc. e. ast or am the female (war kutara. e. there is a large noun class which is feminine in the singular and formally masculine in the plural.g. The same gender classification exists. The Bantu. an additional word may be used: tābat or war may serve to designate the male.GENDER 233 libbātu(m) is used in the special acceptation of "heartstrings. ansatà bušā. In any case. greġāt. Gafat tàbat bušā. although the -t morpheme is used to a certain extent as a feminine marker as well. In the east Caucasian languages.. hata gdrrum (masc. "cock". that inanimate feminine nouns may have a masculine singular concord when they are used in plural. which they designate by differ­ ent prefixes. as Chechen. or Darghi. e. . "girl". "hen"). Avaro-Andi. Other living beings have different root morphemes for the male and the female.g. although this situation may also reflect that of Highland East Cushitic. divide things into eight to fourteen categories. in the pronominal element ebàryàt. e. in the Algonquian family of North American Indians.. However. while the Bantu languages of Africa and the Cau­ casian languages have many grammatical genders. anast taġġa. and probably in zak-it. "she has beau­ tiful bracelets".g. "boy". "girl".

"lion". which can hardly be separated from Sumerian a n š e. Semitic languages preserve traces of a similar gender classifica­ tion with a distinction. "dog". "calf". 'ag-al-1 'ig-I-. "ewe". Instead. di'-b. kuta-ra.C. e. "ram. "ox" (cf. "elephant". "frog" (Tigrinya). kal-b-. jackal". e. hilam-ār. dab-b. "swine. raven". dmb-ab / hub-āb.g. dromedary". between wild animals and domestic animals. " k i d " (Afar). baq-ar-. ġurā-b I *ġāri-b. "elephant". a-ġy-ul. dawb-al. gam-al-. kir-r-. Thus. "cattle". nam-l-. Egyptian 3-bw. "camel. "hyena" (Harari. "donkey". "deer". "hare". bak-dl. "dog" (dangerous and originally a wild one). iy-r. "bear". cf. kura. but dáb-el. 'arn-ab-. Tuareg a-g ān-ba.C. e. ġura). the determi­ nants -b and -r / -/ became constituent elements of the concerned nomi­ nal roots in the historically known languages. "goat" (South Ethiopic). lah-r. "lizard". nuh-b. hab-b-. although these formations were still operative in the late third or the early second millennium B. §65. "ant" (!).10. "fox. Gurage).g.in Cushitic and Chadic.5).. Contrary to -t. e. "wolf".234 MORPHOLOGY 30. lamb". "crocodile". s-rw. but it must be posterior to the domestication of animals. "poultry" (Gurage).. hulizzlr-. 'aqr-ab-.g. izimm-dr. "pigeon". dub-b-. taw-r-. however. "goat. lamb"./ rah-l-. 'ay-r-. lacking the final b in q dr'a (Tigrinya) and in other Semitic languages. dub-b > dub-āb. "spider". A significative example is provided by Gafat ansd"donkey" (Argobba hansia). and therefore seems to indicate that *hanše was a West Asiatic culture word used around 3000 B. "lamb". "deer". related to Somali warā-be and perhaps to Egyptian whr. gam-al-. mainly in the seventh millennium B. "snake" (South Ethiopic. "ass". "ass-foal". domestiw w y . the postpositive determinant -/ / -r qualifies the grammatical gender of domestic or tamed animals. fdyy-al / fjġġ-àlà.9). "ram. zab-ba. but still lacking the -b. "flies". "goat" (Saho). 'ayy-al-. Some of these suffixed animal names and similar formations appear in other Afro-Asi­ atic languages as well. Libyco-Berber a-tbi-r. as shown by the Semitic name of the "dromedary".C. §11. without the final morphem in Somali and in Rendille (East Cushitic). just as the postpositive morpheme -t categorizing the collective or abstracts nouns. p i g " . roebuck". This gender classifi­ cation should thus go back to a common Afro-Asiatic background. q dr'-ab. "scor­ pion". "donkey". "young ass" (Arabic). ta'l-ab-. still lacking the -b in South Ethiopic (qurà. "bee". 'ank-ab-. "hip­ popotamus". gawz-al "young pigeon" (Arabic). wârā-ba or urā-ba. etc. 'imm-ar-. "crow. the postpositive determinant -b qualifies the grammatical gender of wild and dangerous animals. Egyptian d-b and Highland East Cushitic lo-ba.) and in Cushitic (qura. still lacking ~ba. nayyal-. "sheep".g. attested as lah.

Cushitic). Number 31. sometimes. qar-n-. a-z-ġdm in Tamazight. The same morpheme is prefixed in Libyco-Berber to *gam I *kam. without the morpheme -(d)l. as Ugaritic td. and it lacks the determinant /. "bosom". "tooth". bat-n-. "cow". Omotic qaro). with metathesis). while "ear" and "horn" lack it in Cushitic (*waz. "horn". with the typical change / > z of that dialect. Semitic languages possess three numbers: singular. gah-ān. Oromo funnān. It is the same word as Gafat ālam à. pa'~n-. "mouth". w y C. in Proto-Semitic. "stomach". Besides. afu-na.11. There are also several exceptions in Semitic. Gafat. wadne in Somali. corresponding to Ge'ez 'af and Harari af. cf. ša-du-un (Amorite). "small finger". Syriac tddā. tad-anlun. whereas "ear" is called kàzàr(à) in the same Agaw dialects. thus a-lg m in Tachelhit. Inter­ estingly dián designates the "heart" in some Agaw dialects of the Qemant-Qwara group. Within the general domain of number expression. Its Tuareg name alam has a different origin."belly". with other names in various Cushitic languages. qut-n-. "rib" (Ethiopic. udder". "thigh" (Amharic. The noun tad-an (Arabic). etc. "side"). Thus. pa-n-. šin-n-. yam-an-. "teat. perhaps from the same root as £d-n. there are collective . w w 30. dual. the postpos­ itive determinant -n qualifies this gender as shown by 'ud-n-. also Oromo af-ān. q dn-n. "nose" (Gurage. a development attested in Boni. "ear". gēl-a (collective) in Somali. in Libyco-Berber (i-ls). Hebrew šad. "lap". his-n. trial. g ad-n. with the regular Rifan allophone r II I The East Cushitic name of the dromedary. "teat. and a-r-ġdm in Tarifit. gāl in Rendille and Boni. A number of languages distinguish a grammatical gender com­ prising names of parts of the body. but these are mostly either loanwords or derivatives. The word "tongue" is attested without the gender morpheme in Egyptian (ns). and in Chadic (*/i-). 'ay-n-. udder". from Semitic 'anf + na. "right hand". td-n (Ugaritic). Oromo cinā. corresponds to a form without final -n.NUMBER 235 cated in that period in Arabia and still called kām (plur. "eye". Argobba. is best explained by the loss of intervocalic m. and plural. wazana in Saho. "face". Gurage.. "foot".1. "tongue". languages dif­ fer on the basis of whether they limit themselves to singular and plural. or include also dual and. intensified as usual to *ġam. karri) in Cushitic (Bedja). "buttocks" (Gurage). laš-lliš-ān-.

ba-ntu.g. e. The plural distinguished by additional mor­ phemes from the singulars and from the collective singulars is then the result of a further development. while the plural was expressed by the collective singular. or a group of beings or things regarded collectively. "little man". "by virtue of the two horns of the moon". in ba-ta-a. in Class 1 (human beings) mu-ntu.12) by nunation or mimation. Old Akkadian bēlān. e. a) Dual 31. at Ebla. Ugaritic. Besides these general observations. as "man". Its regular occurrence in Palaeosyrian. the singular was originally the only number. The absence of nunation or mimation does not necessarily indicate its dropping in a later period. the eyes —. viz. in Class 7 (diminutives) ka-ntu. Many language groups present plural formations radically different. "two ostriches". Thamūdic h-bkrtn.23-34). Arabic. "the two masters". In the Bantu languages.g. corresponding to the so-called "bro­ ken plural" (§31.35-37). al su-lu-la-a 1 ITI.2.236 MORPHOLOGY nouns (§31. they denote either a single being or thing. e. Safaitic n'mtn. the singulars do not require any particular presen­ tation. "the two young she-camels". needs further investigation. Old Akkadian. nunation is generally missing. bēltīn.g.38-39) and paucatives (§31.40-42). except for the singulatives derived from collectives (§31. The dual is formed by special endings attached to the singular. bēlīn. Comparative analysis indicates that the dual is diptotic in all the Semitic languages. but also to express simple duality. In all probability. In Palaeosyrian. and in Modern South Arabian idioms indicates that its restricted use in other lan­ guages results from the widespread substitution of the plural for the dual. for example. attested at least in Arabic and in Ethiopic. E.g. "little men". as "mankind". the plural is distinguished from the singular by other sets of class-prefixes. "the two mistresses". 31.3. It serves to denote not only pairs of objects — mostly parts of the body occurring in pairs. "men". and they are fol­ lowed in the absolute state of the noun (§33. The latter category. "man". It is formed by adding -ā in the nom­ inative and -ay > -ē / -I I -ā in the oblique case. characterized in Semitic by a different pattern of the basic root. which can be missing. As for singulars. bēltān. it distinguishes only the nominative and a single oblique case (genitive-accusative). These dual morphemes are attached to the stem or to its "feminine" ending -f. tu-ntu. " i n the two houses". 6 .

but the plural of the same feminine noun is yudlyun or 'ayādin.g. §22. paws".4). while "two eyes" is 'ēntēn in Dam­ ascus and in Baghdad. "two brothers". The attested Pre-Classical forms are -āni or -āna. §22. . this assumption is by no means certain. The mimation is used also in Ugaritic. the original dual ending is used as a "pseudo-dual". Hebrew raglayim. because the whole extension of the same single form for all cases of the dual in ancient Arabic dialects calls for caution.NUMBER 237 31. A similar situation occurs in Modem South Arabian languages where original duals can function as plurals. and in Modem Ara­ bic colloquials. Instead. 'ēnēn. nasfi.g. "feet. appears also the feminine plural in 'šr ydt.g. 'ēnēn. among modem colloquials. and are so considered by native speakers (e. that of Hadramawt has -an for the dual in all cases. with the dual ending added to the feminine mor­ pheme . "two horns". in particular a part of the body. occurs also the feminine plural qrnt when more than two horns are meant. Phoeni­ cian 'hym.13). However. It can be contracted to -ē (Arabic colloqui­ als) or to -ī > -i (Modem South Arabian). e. but it may also appear in the plural when more than two are meant or when the word is used in a derived or different meaning. "eyes". in most modem Arabic colloquials. in the construct state -ā.g. A noun designating an object that comes naturally in pairs. qarnayin. In classical and literary Arabic e. "ten handles" or some similar tools. However. in Ugaritic e. "eyes". is generally used in the dual. we have mimation instead of nunation. from sing. and besides the dual ydm. thus -tayn > -ten. with various meanings. in the Arabic dialects of Damascus and of Baghdad (§31. 31. e. but the ā might result from a monophthon­ gization of ay (cf.. it is fol­ lowed by the nunation in Aramaic. in Neo-Arabic.3-4).g.5.12-18). "halves".6).. In Hebrew and in Phoenician.g. Thus. "two eyes". the dual "two hands" is yadāni in the nominative..4. The ā of the oblique case might result from a contraction ay > ā (cf.i . "hands". In the Semitic languages which have lost the case endings (§32. nasf). In the absolute state. but the unvocalized texts and the contraction of the diphthongs reveal no formal distinction between the nominative and the oblique case. 31. Authors generally believe that such a distinction has to be admitted since Ugaritic possesses the three basic cases (§32. the dual morpheme -ay of the oblique case is commonly used also in the nominative. besides the dual qrnm. and 'ēntēn. e.6. "horns" in Aramaic.

liqān and liqānāt. E.7. When the same nouns occur with the regular plural ending.238 MORPHOLOGY 31. but the regular feminine plural p 'mt is used when the word means "occurrences. which could be related to the broken plurals. times". "saying". However. but they are then used even i f the number of these objects is more than two. it may consist either in the lengthening of the characteristic vowel of the singular (§31. in Ge'ez where the external plural ending -āt may be added to the broken plural.g. that refers to the paws of sacrificed ani­ mals. the contrast between originally dual endings and plural endings is henceforth on the lexical level and it does not express any more the distinction between the numbers dual and plural. Besides.8.21-22). dual endings are replaced by plural end­ ings from the Old Babylonian period on. "sayings". as in the case of liq and of masāle. the dual ending occurs in p'mm. they have a different mean­ ing.33) and both forms may coexist (§31. 'amsāl and 'amsālāt.10-11). "six wings" are called in Hebrew šēš kdnāpayim. E. e. Such a double pluralization is frequent in Berber dialects (§28. i. there are examples of double plurals.10. Therefore.12-20). In Punic.g. There is also a plural pattern in -h (§31. E. while qātāti designates two or more "shares".33). Comparative analysis indicates that the external plural of ProtoSemitic or Common Semitic is diptotic and that it is formed by the . Broken plurals... they continue to occur with nouns that denote a pair of objects. are sometimes further pluralized by an external plural (§31. The same evolution is attested in Hebrew and in Punic.25.19). The plural may be formed in Semitic by the attachment of special endings to the singular or by the use of a noun pattern different from the one employed for the singular. 'aqwāl. 31. with the dual ending.e. External Plural 31. in particular.23-34). "proverb".g. plur. "elder". two or more "hands" are qātā(rì).g.28. 'aqāwīl.13). "chief". and a plural by reduplication of the root morpheme (§31. consisting in pluralizations of forms already plural. or in suffixing an ending -ān(-) or -āt{-) (§31.9. The first type of plural is called "external" or "sound". The second type of plural is referred to as "internal" or "broken" plural (§31. In Assyro-Babylonian. Ara­ bic qawl. while the regular feminine plural kanāpot means "extremities". plur. b) Plural 31. Ge'ez liq. "ensembles of sayings".

12. "houses". as suggested by the construct state bnw and bny. "kings". "gods") appears in AssyroBabylonian from the Old Babylonian period on. the plural of the feminine nominative kalbatuim). The later Assyro-Babylonian plurals kalbē. can best be explained by the common ten­ dency of the oblique case to stand for all cases. It is as yet uncertain whether the Lihyānite construct plural bnw /banū/.g. 31.g. yrhw 'sp. mlkw. Hani. and modem Arabic colloquials (e. plur.g. implies an absolute masculine plural banūna. e. Aramaic (e. Phoenician (r|À. sūsīm. šišiltā.g.g. i.g. the mimation of the singular is omit­ ted. a plural in form). torīlē. the plural of the nominative kalbu(m). is kalbū and the plural of the oblique case kalbiim) is kalbl. "sons". Ammonite. In the masculine plural of Old Akkadian and Assyro-Babylonian. e. haddādīn. In Neo-Aramaic.e. in Classical Arabic (e.g. This apparent innovation might thus be explained as an expanded use of . šarrānu. used also for the nominative. but the distinction of the cases is preserved in the Samalian dialect of Aramaic (e. "months of ingathering"). "thieves". in the Canaanite dialect of the Gezer calendar (e. "bulls").g. which regularly occur in Old Assyrian and occasionally in Old Babylonian. while it changes to nunation in Aramaic (with the exception of Samalian). but sāriqīna in the oblique case). e. bīrānyāt with a double plural ter­ mination -ān + yāt (§31. at least in Ugaritic. in Arabic. bi-ti-e. "chain". and most likely in Lihyānite and in the earlier stages of South Arabian. Late Babylonian blranātu). "dog".NUMBER 239 lengthening of the characteristic vowel of the singular. "kings") and -āni/ī (oblique case. plur. In the "plene" spellings a-wili-e. also in Hebrew and in Phoenician. like in Classical Arabic. from the time when the distinction between the numbers dual and plural was abandoned. This formation is already attested in Late Aramaic. Neo-Aramaic (e. 31. cf. the -e functions as the phonetic indicator of this long vowel. "words"). The allophone -ē of the ending -I in the oblique case of the plural is already attested in Old Akkadian. Thus. Similarly.g. "smiths").eip = 'lm. "stronghold". is kalbātu(m) and the plural of the corresponding oblique case kalbati(m) is kaìbāti{m). "deity". "horses"). Moabite. šišilyāti. e.11. "sons". but mlky in the oblique case). "bitch".g. the ending -yāti is the plural morpheme of the nouns ending in -tā. This trend is well known from Hebrew (e. sāriqūna. millln.12. A masculine plural in -ānu/ū (nominative.g. but it is preserved in the absolute state of plural North Semitic nouns. bīrtā. "men".

g. masihān. "cut­ ters".g. However. "girls"). corresponds to the plurals ri'ūna and rVātiun). "gazelle"). "garden". Significantly. as well as the plural patterns fi'lān and fu'lān in Arabic (e. and ri'atiun).7). a-la-ni-i-ka. kahdnāt. "cities". malkān. Some nouns with the overt -t. ġagotan. has the plurals sìnūna and sanawāt(un). and nbt-n "splitters". has the plurals ginnln and ginnayyā or ginnātā in the emphatic state (§33. The apparently double indications of the plural in East Semitic -ān-ū or -ān-ī (e. "belly". ngwaren). the ending -ān appears in Aramaic as the feminine plural morpheme of the absolute state.240 MORPHOLOGY the originally dual ending -ān (§31. "year". Tigrinya qaddusan.C.15). although the "feminine" suffix -āt is used for masculine sub­ stantives as well (e. "year". The Aramaic noun ginnā.g. l . The ending -āni < -ānī/ē is also used to form masculine plurals in Neo-Aramaic and it is rightly seen there as combining the element -ān. The plural s nn lšinūn(a)l is attested also in Safaitic. as early as the ninth century B. the external masculine plural ending can be also -ān (e. "Mes­ siah").g. the same suffix forms the plural in Chadic (e. It is used mainly with participles and verbal adjectives. has the plural šānīm in the absolute state. plural of masih. ġizlān.and the usual plural morpheme -lie. E. Besides. "queens"). "eyebrows"). in the nouns nbb-n. which kept only few traces of the dual.g. plur.g. §31. šānā.C. Arabic sanat(un). sitt riġlēnāt.g. "lung". "priests".g. Logone ngun.3). thus inde­ pendently from any influence of Arabic. e. "six pairs of paws").in the colloquial Arabic of Baghdad (e. In Hebrew. -tan or an allophone (e. while the most common fem­ inine ending is -dtdn. a-la-nu-u. 31. followed by the case markers -u/ū and -i/l. Tigre qdddusām kdtubām. Modern North Ethiopic still preserves the ending of the external plural -an in Tigrinya and -ām in Tigre. The external masculine plural in Modern South Arabian languages is -In or its allophone (e.13. "Holy Scriptures".g. There seems to be no contrast between the two sets of forms that can be determined at present. "your cities") could perhaps be compared with the addition of the plural end­ ing -āt to the dual morpheme -ēn. one should bear in mind that the plural in -n is attested in Libyco-Berber as early as the 2nd century B.g. while 'âšērā. "sanctuary".. In Ge'ez.g. gdfnln. "saints".g. but sanot in the construct state.mark of the feminine in the sin­ gular have a masculine and a feminine forms in the plural. of a bilingual Punico-Numidic inscription from Dougga. with the change n > m\ e. and it is being substituted for the feminine ending -t of the singular (e. has both plurals 'âšērīm and 'âšērot. "broken" plural of ġazāl..

"chair". "plough". some of the nouns in question are really feminine also in the singular. appears with the plurals r'ašm. plur. "peasant". but are treated as feminine in the plural. with idū and idātu(m). has the plurals grnm and grnt. marl. "hand". are masculine in both numbers. plur. as e. plur. A similar phenomenon occurs in Assyro-Babylonian.g. e. This alternative plural regularly adds a special connotation and agrees with masculine forms of adjectives and of verbal pronominal affixes. yom. "known owners" (masc). r'iš. as e. We know at least that ks'u is also . "day". In Assyro-Babylonian. "ship". is attested with the plurals biblāni and biblāti. "gate". "gift". These nouns cannot be considered automatically as feminine.in the singular. "head".g. mārwātā. yomāt. "dwelling". yamot) and in Phoenician iymm. plur. biblu. ymt).NUMBER 241 31. qaqqaduim). in Ugaritic. In Imperial Aramaic. Thus. "threshing floor". eleppētu(m). grn. Other nouns however. take the "feminine" ending in the plural. or eqlu(m). without the -t. in Assyro-Babylonian. has the plurals bābāni and bābāti. epinnētuim). Both forms of plural occur also in Hebrew iyāmīm. as shown by the masculine plural termination of the modifying adjective rbm. plur. "owners" (masc). abulluim). indicates that the morpheme -t does not change the gender of the noun.g. which was the ending of the plural oblique case: the vowel i caused palatalization of t and was absorbed in the palatal. Some nouns. in the construct state.14. has the attested plurals yomln and. Also the external plural in -oca / -ac. as shown by the plurals nasīkāni and nasīkāti of the name nasīku.g. "city gate". epinnu(m). originates from an ancient *-āti. ikkarātu(m). idu(m). both masculine and feminine. "sheikh" or "prince" of the Aramaeans. which have the plurals ks'at and mtbt. In Ugaritic. 31. bābu. e. "head". This phenomenon assumes larger proportions in Assyro-Babylonian and especially in North Ethiopic where the "feminine" plural ending -āt is widely used for masculine nouns. plur. The situation is similar in Ugaritic with nouns like ks'u. qaqqadātuim). Some Neo-Aramaic masculine nouns have a "feminine" plural formation besides the regular masculine plural ending. or eleppu(m).mark of the feminine in the singular. A third group consists of nouns which are masculine in the singular. eqlētu(m). and in Aramaic with nouns which do not have the feminine morpheme -t. or mtb. "many days".15. while the word "wheat" (root hnt) is attested under the forms htm and htt.17) and in some Tigre nouns. mārā. "owner" (masc). r'ašt and r'išt. used for both genders in modem South Ethiopic (§31. abullātu(m). or ikkaru. while the Ammonite phrase ywmt rbm. "field".

"house". plur. rabiūtu(m). . is dnġdccà in East Gurage. therefore. This plural morpheme is pro­ nounced -occ in Amharic. -deed.22). "baby". 31. as in hbbi. or a substantive with collective meaning. e. the plural of dng.g.g. "water".15). culo. plur.g. it is sometimes difficult to decide whether the noun ending in -ūtu is an adjective masculine plural. also in Highland East Cushitic. hadisān). e. This morpheme is homophonous with the deriva­ tional afformative -ūtu that forms abstract nouns (e. East Gurage dialects add a vowel to the element -c: -cá.16. plur. E. hbbdtat (cf. "great".g.is the most common Cushitic marker of the plural. plur. gâza.g. "languages".17.g. §31. In North Ethiopic the ending -āt is used instead of the masculine plural morpheme -ān without influencing the gender of the nouns (e. I n Tigrinya.20). "king­ ship") and. and the Soddo dialect of Gurage. whereas it is -àc in Harari and -ac in Argobba. e. sâbat).g. "house". "mankind". plur. šarrūtu. Tigrinya sáb. "new". amēlūtuirn). plur. Ge'ez may. Gafat. rabiūtiim). "hand".12) is employed for adjectives and participles (e. Ge'ez kadis. māyāt. the plural of bet. East Gurage dialects also repeat the last radical to express the plural (§31. The Amharic broken plurals are borrowed from Ge'ez (§31. plur. Besides the plural ending -occ (§31. For the adjective masculine plural. as šibūtuim). However.242 MORPHOLOGY feminine in the singular. Contrary to Ge'ez and to the modern North Ethiopic.g. witnesses".17). a side influence of the Cushitic adstratum on Ethiopic should not be excluded. however. is betoâé in Amharic. Amharic continues using the Old Ethiopic ending -āt.1). In the context. South Ethiopic uses the ending -occ / -ac deriving from an ancient *-āti as plural marker without making a distinction between the masculine and the feminine (§31. "per­ son". the plural ending is -tat after vowels (e.34) and the preserved patterns are no longer productive. "apostles". Assyro-Babylonian uses the morpheme -ūtu. §31.. mainly with masculine nouns or with nouns unspecified as to gender. South Ethiopic has an external plural only. "heart". whereas it is bedac in Argobba. and for a small number of substantives. gádamat. hsanat.g. "elders. while the morpheme -ān (cf. in the oblique case -ūti. hawaryat. since -t. 31. gâzatat).16). "convents". The wide use of the ending -āt can best be explained by the original function of the morpheme -tforming collective nouns (§30. e. rabiu(m). even when the final vowel has only an auxiliary function (§27. the question can be decided on the basis of grammatical concord.

and in some modem Arabic colloquials. 'ummahāt. "handmaids"). fàràz means both "horse" and "horses". 'mh. "ag(h)a". "elder brother". pehā. "the mothers". a a 31. handmaids" (sing. waġi. qešet.g. "companion". plur. Ugaritic 'aht. qrht. with 'amht. plur. ddlātot. The same usage is attested in Arabic. e. Assyro-Babylonian išātu(m). Sabaic plural patterns in -h are found with biconsonantal nomi­ nal roots in 'bh. The number of the noun is indirectly reflected in the verb and in the pronouns. in Ugaritic. 'ilht. "mothers"). "fathers". 'aġā.g. šdmāhāt. "fathers". in the Hawrān ('abbahāt. plur. gamela. "sister". and with the addition of an external feminine plural morpheme in 'mht. "doctor". The Tigre plurals 'abayt. "bow". Tigre sadāyat. thus 'āsē. thus. 'ās( )wātā in the emphatic state. korsāwān. qašātot. "names" in the construct state). Similar plurals occur in North Ethiopic. plur. 'hint. construct state).g. plur. construct state). "fathers". "mouths". 'aġāwātiun). "seat". "women". and 'afayt. 'umht.15). with gu-la. 'immdhātā. "master". "cups". has a plural gdzawdtti besides the usual gdzatat (§31. .g. 31. 'aryē. "house". bāšawāt(un). gamelalo.20. barakatāt. The plural in -h is attested also in Arabic (e. plur. 'ahtt. e. pah( )wātā. Instead. The same ending -āt > -ot of the plural is sometimes superim­ posed upon the singular ending -t. "mothers". Hebrew delet. E. West Gurage has no plural marker. "doors"). likewise in Minaic (e. "blessing". in Hebrew ('âmāhdt. sadāyatāt. plur. plur. kinā. "clans").g. 'âbāhātok. "camel". "moth­ ers". in Tigrinya where gdza.g. Aramaic uses the feminine plural morphemes to form the plural of some masculine nouns — mainly loanwords — ending in a vowel. "door". "towns". "your fathers". while the formation of the plural in -ht occurs in Palaeosyrian. "fathers".19. The same situation is found also in some Cushitic dialects of the region. plur. plur. 31. "lips") and in Qatabanic ('hh-. the ending -t of the plural seems to be superimposed upon an earlier ending in -h. plur. besides 'aht. waġiġo. suhartātu(m) besides suharātu(m).'à-tum I gullahātum I. plur. kdnāwān and kdtiāwātā. "pasha". bāšā. "female clients. korsē. suhartu(m). e. in Aramaic (e. have probably the same origin.g. in Hadramitic Cbhty. plur. "girl". išātātu. "help". plur. 'ntht. plur. 'mi). bet is "house" and "houses". šifāh. 'ustādāt(un) or 'asātida(turì). Ge'ez barakat. "handmaids". 'ustād. plur. "lion". "brothers". "governor". "fire".18. In all these instances. "goddesses". in Phoenician (dlht. 'aryāwān and 'aryāwātā.NUMBER 243 éulalo. plur.

gázatat (§31. A plural formation by reduplicating the root morpheme is attested for some biconsonantal nouns.g. plur. dambedambe). This suffix -otāt < -atāî is common in Tigre with nouns whose singular ends in -at. Assyro-Babylonian birbirrū. "brother". "hidden things" (cf.g. àmar.g. plur. to a lesser extent in Cushitic (e. "neighbour". with addition of a -t. plur. wâyzazdr). sdrqotāt. Palaeosyrian ha-ba-ha-bi (gen. plur. -t) should be considered as a secondary feature. plur.AN may indicate the plural "gods". Arabic hab').g. This partial reduplication occurs also in Chadic (e.244 MORPHOLOGY 'āmat. "lady".g. Hebrew mēmē. culo. while the double writing AN. -In I -ān. "struggle". "day". "donkey". also with the addition of the usual plural ending.g. sanasana). Hausa dambe. "waters" (con­ struct state). plur.15). plur.AN signifies heaven". The repetition of the root mor­ pheme is probably one of the oldest methods to express the plural. This type of plural marking is unlikely to have arisen from a distributive context (§35. plur. gâza. "brother". wànddmamocc). wànddmm. g abbabit. sdraq. wolla. Bedja san. hadhddānē. Also Amharic uses the repetition of a radical as a device to express the plural in a lim­ ited number of nouns. w w . "gods". Aramaic rab. âmararà.g. g àbbe. plur. Syriac daqddqē. rumah-rumah.17). plur. e. "little ones". wàyzàro. and in the Malay language (e. "year".g. rahrdbln (masc. 'āmotāt. k u r . with the addition of the -t morpheme.).22.g. while ancient Egyptian initially redupli­ cated the hieroglyph to mark the dual and wrote it thrice to signify the plural. "great".g. wollalu. ft 31. The addition of the usual plural morphemes (-Ū. but it is used also with other nouns. especially in Gurage (§31. "glare". plur. "houses"). for reduplication can have various functions. plur. "mountains"). e. md'altat. culalo. "certain ones". "house". "brother".22). A n analogous plural formation is attested in some South Ethiopian languages which express the plural by the repetition of the last radical.AN. "cutting edges". with no singular but with the addition of the plural morpheme -ū. cf. in Sumerian (e. "theft". but this consonant is not necessarily the last one (e. and the plural morpheme -occ can still be added to the extended root (e. Hebrewpe.g. The Tigrinya plural end­ ing -tat appearing after vowels is probably related to these cases. "baby". Sabaic 'Vlt. A similar practice is attested at Ebla where the triple writing AN. plur. It is used in Chadic (e.) and rabrdbān (fern. e.k u r . plur. Plural by Reduplication 31. e. pīpiyydt.21.). mà'alti.

attested besides the usual k3w.NUMBER 245 Hausa magana. "word". and it is more pervasive in ancient South Arabian than in any other Semitic language. Therefore. where the Coptic internal plurals can hardly be considered in mass as a secondary development. sanan). "free men"). plur. viz. bākik-6. plur. Traces of broken plurals are also preserved in dialects belonging to Semitic languages in which these plurals are not regularly attested. the broken plurals may be regarded as Proto-Semitic. Kafa bāk-6.g. plur. 31. plur. in Cushitic. "cock". These nouns form their plural on the pattern C vC C a(C ) to which the ending -ū / -ī or -ūtu / -ātu of the external plural is added: ] 2 2 3 . "goodness". "soul". and in Egypto-Coptic.g. in Libyco-Berber. a rare plural of k3.23. "good". The use of broken plurals is widely attested in Arabic and in Ethiopic. "small".25. "door". the pattern of which corresponds to Classical Arabic fu'lun. kofa. The same noun pattern is used in Old Akkadian and in Assyro-Babylonian for socalled abstracts like dumqu(m). The patterns so used are rightly regarded as original collectives and their function as plurals can only be established in the light of grammatical concord. Thus. as suggested by the old Egyptian plurals ipw and ipn of pw and pn. while the Assyrian individual singular is sahru(m). "nose". Internal Plural 31. Semitic internal or "broken" plurals are formed by the use of noun patterns different from those of the singular. 31. Old Assyrian suhrum. especially in Bedja and in Afar-Saho. and there are clear parallels in other branches of AfroAsiatic. A restricted number of Assyro-Babylonian monosyllabic nouns preserve traces of a broken plural — sometimes called "infixed plural" — that parallels the Arabic pattern fu "al (e. at least in the sense that the collective function of some of their patterns is common to several Semitic languages in different areas. buhhal. Some of them may be quite archaic. maganganu. which is in reality a collec­ tive noun "good things" derived from damquim). or ik3. Somali san.24. "this". has the collective meaning "the small ones". kofofi) and in Cushitic (e.

both fu'al and fi'al are well represented in North Ethiopic by the pattern CdCaC (§31. Besides. "young camel" damq-u. "herb" qešet < qašt-. The stem with vowel ā most likely occurs in ahlāmu. "man" Plur. "mountains") and it was used in North and West Semitic. The stem vowel is either a ('af'āl > 'af'al) or u ('af'ūl > 'af'ul). "mountain". in Ge'ez. subbdk-o 'innab-ē 'assdr-ot 'iqqab-ē I -ot 'iśśdb-ot qasšat-oî miqqadāš 31. "grape" 'âseret < * 'asr-. and to Ethiopic dabr. They generally have a superimposed external plural ter­ mination which causes the shortening a > d in the pattern C vC C aCỳ l 2 2 Sing. sdbak < *sbak-. "brother" alk-at-. du-ba-lu /dubbaru/. This kind of ambiguity does not occur in Hebrew where similar survivals of broken plurals — traditionally explained by a dageš dirimens — are preserved by the Masoretic vocalization of some Hebrew nouns. a noun related to Aramaic dabr-. "heel" 'ēśeb < 'iśb-. the probable Old Babylonian designation of the Proto-Aramaeans and their congeners which must transcribe a .31). as well. in Epigraphic South Arabian. the Palaeosyrian orthography does not allow distinguishing patterns corresponding to fu 'āl. fu' 'al. "fine" ebr-u. abba-ū > abbū *ahha-ū > ahhū alkak-ātu arrak-ūtu bakkar-ī dammaq-ūtu daqqaq-ūtu ebbar-ūtu hannab-ātu sahhar-ū / sehher-ūtu zikkar-ū This pattern might be attested also in Palaeosyrian by e.g. amsal. "good" daqq-u. adbar. to which du-ba-lu might be related as well. "voluptuous" sahr-u / sehr-u. However.246 MORPHOLOGY Sing. "bow" miqdāš. and in Tigrinya is 'fl with the preformative 'a-. "pasture". "small" zikr-u. "sanctuary" Plur. "friend" hanb-u.26. "long" bakr-u. By far the most frequently used broken plural pattern in ancient Arabic. "parables". "way" ark-u.g. "father" ah-u. This stem is preserved in a few Amharic forms bor­ rowed from Ge'ez (e. "pastures". ab-u. "crowd" 'āqēb < * 'aqb-. "thicket" 'ēnāb < *'inb-. fu"āl and fu'al.

some of them with geminated second radical (e. Vocalic variations as Ugaritic 'utkl and Aramaic 'etkālā might suggest that the pattern was broken. They may occur also in Palaeosyrian. "city". The trace of such a plural. "daughters".26) have almost disappeared and coalesced with other stems. in a Punico-Numidic inscription from Dougga. with the addition of the usual plural morpheme -m. used with nouns designating persons (e. others extended either by the prefix 'a. 'ahqul. in consideration of mt kin.(e. 'amrād) or by the suffix -ān (e. a-sa-lu /'āšālu < 'a'šālu/. the patterns with the prefix 'a. "lands". both 'af'āl and fi'āl became . where they are attested already in the 2nd century B.NUMBER 247 native 'aġlām. with the meaning "surety". It is used also in modem North Ethiopic. -zlufa. from sing. ġazāl. "rushes". except in Yemen (§31. There is also a pattern fu'alā'.g.g. plur. from sing.28. with diphthongs. ġizlān). though it is often to be translated as a singular. This stem with the vowel u < ū is very com­ mon in Ge'ez. "grapes".g. "all the lands". plur.29). "scribe". "rush". "springs". marad. corresponding to Hebrew 'eškol or 'aškol < *'atkāl.C. in the Aramaic Tell Fekherye inscription. 31. 31.C. hagar.. kuttāb).g. -gbalu). "illness".g. and zl'. Tigre kaldb. 'ahgur. as well as some other possible instances of broken plurals (§31. áš-kà-lum /'atkālum/. "warranty".25). and bhnt. probably occurs in Ugaritic bhtm. The stem with vowel ū ('af'ūl) is found in 'dqwr. e. plur. "despicable (men)". plur. "boys. "jug". "sons".g. an Aramaic noun used in the Tell Fekherye inscription (9th century B. with short or long vowels. "field". "iron".C. haql.) and apparently employed as the broken plural 'adqūr of dqr. in which there are clear examples of syncope of internal -h(§27. 'aklub next to 'aklāb. since they seem to appear in Palaeosyrian.27. by gú-a-tum Iquhatumj. -ġbula. e. "dog". The existence of such a plural may also be assumed for *mht. "gazelle". šā'ir.(§31. plur. 26). probably a plural of qātum. lads". plur.g.are found in bhn. -zlaf'. Classical Arabic has thirty different patterns of broken plurals (Fig. kātìb. e. šu'arā'). deprived of their short vowel. "houses". as suggested. as shown by šqr'.g. and by sa-a-dum /šahadum/. plur.. "hand". In modem colloquials. a barley species. Some of these pat­ terns go probably back to the third millennium B. seem­ ingly a plural of šēdu(m). It is probably found also in Palaeosyrian ar-ša-lu /'ardālu/. "wood". Minaic plural stems with infix -h. but similar broken plurals have a wider range of application in Libyco-Berber (e. "poet".g. e.26).

The pattern fa 'īl. e. Thus. widely used pattern: harāf "lamb". . ġabal. 26.g. plur.9). sing. "slaves".> plur. "barley". "songs".g. "donkeys". "star". while the Hebrew plural of "segolate" nouns derived from the type CjVC C (§29. "feet").g. which is a new.> rēgāl+en. "story-teller". The same situation occurs in Samaritan Aramaic (e. sa-i-lum or sa-ì-lu-um /śa'īrum/. 'hdr. ġbāl instead of 'aġbāl or ġibāl. plur. represented in Arabic by words like 'abīd(un). Four-consonant singular patterns have plurals formed on the stems fa'alii and fa'ālīl. e.5) is an external plural superimposed on a broken plural of the same pattern//'āl >f'āl. zu-mu-ba-ru /zumūbaru/. from *zambāru < zamāru (§11.g. and hamlr(un). *milāk+īm > m(à)lākīm. For further details the grammars of classical and colloquial Arabic have to be consulted. e. It should be stressed here that some of the last-mentioned patterns are very old or are used also in other languages. "devil". e.29. lit. is likewise attested in Palaeosyrian. plur. sing.g. "Hadramites". 2 3 l2 Patterns of "broken" plurals in Classical Arabic fa 'al(un) f'ial{un) fu 'al(un) fu 'ul{un) fa'īl(un) fi 'āl(un) fu'āl(un) fu 'ūl(un) fa 'ala(tun) fi'ala(tun) fu 'ala(tun) fu"al(un) 'af'ul(un) 'afilaitun) 'af'āliun) fu' 'āl(urì) fi 'lān(un) fu 'lān(un) fa'ālā fa 'ā 'il(u) 'af'ilā 'u fi'āla(tun) fu 'āla(tun) fi 'la(tun) fa 'ālil(un) fa 'ālīl(urì) fu'alā'u fa 'ālilaitun) fawā'il{u) fawā 'īliu) Fig. "teller of stories". rigl.248 MORPHOLOGY f'āl. kawkab. as well as the pattern for four-consonant singular stems. "mountain". malk. kawākib. hrāf instead of hirāf. the broken plural of thefì'āl pattern occurs in Palaeosyrian mu-dabil sí-kà-ri (mudabbìl sikāri/.g. As said above. 31. the use of broken plurals is more extensive in ancient South Arabian than in any other Semitic language. inasmuch as it occurs even with nisba formations in -ly. šayātīn. plur. šaytān. e. but with a different set of vowels.

'abaġalti). šafattit). 'abqal. CaCaCt (e.and/or the afforma­ tive -t. Tigre ġana. plur. "foal". "friend".f'ln. person". Tigre walat. 'azmad). šanaggal). "horse". sometimes accompanied by the preformative 'a.NUMBER 249 The most used broken plural pattern is 'f'l (§31. e. plur.26. one can assume that epigraphic f ' l (e. 'aCCāC (e. plur. 'aCāCaCit) (e. 'aCCdC (e. plur. Ge'ez bahr. plur. plur. with broken plurals showing the vocalic change of the singular basis. C^aCflCyCjit) (e. 'dgdr. the patterns of which show more similarity to those of Ethiopic than to those of Arabic. plur. dabr. nasr. "mast". 'afras. We give examples of the main patterns CiCaC I CuCaC > CdCaC (e. 'f'lw. Tigre safta. and for four-consonant singular stems the patterns CaCdCaC (e. 'aCCūC (e.g. karas. "field". hagar.g. plur. "shin". 'abatarti. "eagle". gdwf. plur.g. plur. "sea". plur. C aC aCyC aC (e. "uncle". 'n/s ) was vocalized simi­ larly. "rebel". 'amm. "child". 'dzni. Tigre mansaf. plur. plur. plur. 'ahqul. "lip". "ear". "foot". nanyhon. Tigrinya fàrās. zâmàd. CaCāC (e. "tail".g. plur. f'lt.26).31. 'albās.g. plur. Ge'ez q ays. but its vocalization cannot be established directly. "staff". gawaft. "brother".g. "relative". plur. "ear". q ayās. "mule". 'anaggal).g. plur. plur. "soul. plur. "mountain". "chal­ ice". kànafar) and CaCûCaCt (e. w w l 2 2 3 { 2 2 x 2 3 A .f'yl. 'a'mūm). manāsaf. plur.f'ylt. "younger brother". 'adbār. plur. Ge'ez 'din.g. sāhl. Tigre kaldb. 'adbar).g. 'aklāb. Ge'ez baql. plur. "carpet". fit.g. "dog". "wall".g. Tigre daqal. Tigre karšat. malā'akt). Tigre dabar. 'abhart. fdlhi. Since present-day Yemeni colloquials use the stem 'af'ūl (e. Tigrinya bâtri.f'wlt. plur.f'wl. Additional or variant patterns occur in modem North Ethiopic: 'aC aC C aC (e. plur. danāgal. 'aCCaCt (e.g. plur. Ge'ez haql. Ge'ez zanab. 1 1 The Modern South Arabian languages have broken plurals. In North Ethiopic the situation is similar to that of ancient South Arabian. 'ahgur. plur. It consists in the change of the last vowel i/a of the singular into o/w. Tigre naggal. manadaq. 31. plur. ġabāb).30. "writer". nanhan. sahaft). 'aC aC C it (e. labs. plur. plur. Ge'ez darigal. "mountain". nfs . Another kind of broken plural has developed within these languages. plur. Other attested broken plural stems of ancient South Arabian are f'l. Tigre masni. 'azan. 'ashdl. "kid"." girl". 'ansart). plur. aġayw. best attested in Ethiopic (§31. 'aġannit). plur. "dress". plur. plur. bàġli. "messenger".g. 'aznāb. "town". 'awālad. Tigre ġdbbat. CaCāCi(t) (e. 31.g. Ge'ez sahāfi. "belly".f'lw. kanfdr. "mule".g. Tigrinya mândàq. 'agar). plur. instead of usual Arabic 'af'āl. Tigre šangul. 'adquī).g. plur. plur. "adult". "jacket". e. masānit).g.g.g. "chest". aġā. Ge'ez mal'ak. 'a'zan.31). plur. plur. "daughter". fdlho.

plur. or 'alwahāt. where the -at ending probably goes back to the plural morpheme -āt. This plural form occurs also with some other noun types like katam. tree". Thus.g.26). 'alwāh. E. e. " b u l l " .32. kdtāb. from mánbàr. Tigre bara'. "to be strong". 'awharat < *'awahrat. luh. mànabart. 'ab'arat < *'aba'rat. Some nouns of this group are used only with the additional external morpheme -āt. The external suffix -at > -at is added also to broken plurals of the type 'aC aC C aC when C is a guttural. 'akalla'at or 'akalla'at.. and with a metathesis when the second radical consonant is a guttural. plur. 'adgām or 'adgāmāt. ddgdm. plur.33.34. plur. "king". 'asagdat. one should mention the CàCaCaCt pattern which is used for four-consonant singular stems and is the best represented. "book". is a form superimposed on the formally broken plural amlak. plur. wahar. Amharic has borrowed several broken plurals from Ge'ez. The plural amalakt. there are many examples of external plural superimposed on broken plurals. In modern North Ethiopic.g. The same occurs with a Tigre noun like kokab < *kawkab < *kabkab. a broken plural 'atahalti. plur. "plant.g. the broken plural of Tigre bet < *bayt is 'abyāt. stamp". and a combination of both in 'atahaltat or 'atahaltat. with a regular external plural tàhlitat (§31. "neck". e.g. is hasāyar. "seats". The same pat­ tern is used for aga'azt. A characteristic of broken plurals in modern North Ethiopic is the preservation of original diphthongs. "clay pot". 'abra'āt or 'abra'at with vocalic changes next to the pharyngal '. Besides the aCCaC type (§31. is zalāyam. e.g. plur. "seal. that of hilat. "sovereigns". from màshaf. "rain". Tigrinya tāhlì. Tigre bd'ray. 'amdār or 'amdārāt. "tale". "star". as a kind of plurale maiestatis (§50. plural of agzi'a.250 morphology 31. "ox". "land". is hayal. "God". l 2 2 3 3 31.g. plur. the broken plural of which is kawākab. Tigre kala'. "book". "strength. the plural suffix -āt can be added optionally to the broken plural of the type 'aCCvC. "books". the broken plural of Tigre zdlām. coined from malik. plur. and that of hasur. "gods". Some nouns have alternative plurals. "sovereign". e.g. Besides. "pen". power". e. Other diphthongs reveal the use of a broken plural pattern which corresponds to ancient South Arabian f'yL Thus.g. from the verb hela < *hayla. 'akatbat. "enclo­ sure". or 'alwahat with vocalic changes next to the pharyngal h. 'akatmat. "seat".15). e. Tigre madar.24). 31. e. e. sagad.g. másahaft. "board". several Tigre nouns of the pattern CdCāC have a plural 'aCaCCat. plur. and .

"cow" "small monkey" "little girl" "houses" "hyenas" "a few cows" "a few monkeys' waletāt. "people. E. "tears". E. As subjects of a sentence. Collective nouns express a plurality of individual objects or persons.35. walatit.g. e. 'ab'ār 'āliha.23). plants.37. "relative" "some rivers" "some wells' "some gods" "some relatives". Tigre qataf. Old Akkadian sabūm. the plural of ndgus. "God" Plural nuhur.g.38. It is attested at least in Arabic and in Ethiopic.: Singular nahr. "a few houses" 'akarritām. Paucative "rivers" "wells" 'anhur. species of animals. Instead. they may be considered as masculine or feminine (§50.. is nàgást. and the plural of liq. to the basis of diminutives. 'anhās. Collectives are often . 31. but may have existed in other Semitic languages as well. The suffixes are added to the basis of countable singulars. "a few hyenas" wa'āt. Paucative is a grammatical category expressing smallness of number or quantity. "king". I 'ābār.NUMBER 251 for anabdst. habbeyām. and 'af'ilā'u corresponding to fu'alā'u. etc. Further research is needed. 'aqribā'. d) Collective Nouns 31. Arabic dam'(un). habbeyāy.36. 'akarrit. "well" 'ilāh. bi'ār. 31. qarīb. workmen". and to broken plurals.with the following patterns: 'af'ul(un) corresponding to the plural fu 'ul(un). under the singular form. "lion".: wa'at. the collective animate nouns may take their verbs in either the singular or the plural. Paucative is indicated in Tigre by the plural-type suffixes -ām for the masculine gender and -āt for the feminine.g. "river" bi'r. is liqawant. In Arabic. "lions" plural of anbàsa. 'af'āl(urì) corresponding to fi 'āl(un). the paucative derivation is characterized by the pre­ fix 'a. 'af'ila(tun) correspond­ ing to fa'alaitun) or fi'alaitun). "learned man". "leaves". "some girls" 'anhesām. c) Paucative 31.

"soldier". in Tigre. -ay > ā. "donkey". gabil. "people". in Old Akkadian. from 'aéay. ummānum. "iron". "pigeons".41-42) can be considered as singulatives. hamāma(tun). Tigre qadrāy. or -ām (§29. and baqar.g. "a piece of iron". "coffee". širā. bunat. The presence of a collective noun does not imply the absence of either the countable singular or the plural. 31. "a gnat". derived either from the singulative. there is a correspondence between determinate categories of beings and the linguistic classes: e.40-43). . "eagle". in Hebrew. hadid(un). e. "a piece of wood". hadlda(tun). "bricks". no principle of classification differen­ tiating individual from collective nouns is apparent from a mere inspec­ tion of the members of each class. such a countable singular is called singulative (§31.39. "wood". in Hebrew. from tdkān. "crowds". "single hair". "army". fleet". Besides being fre­ quently equivalent to plurals. "eagle"). and in Ethiopic. "ship". This afformative may also express the notion "a piece of". e. "lead". "single brick". Yet. "a coffee grain". Arabic hamām(un). "pinion". 31. "a piece of lead". animals living individually ("donkey".41. "army". "hair". 'oniyyā. Hebrew 'onī. post-classical Arabic 'askariyyun. collectives are often difficult to distinguish from abstracts. and tayr. men". "songs". labina{tun). A plural form may exist beside the col­ lective noun. E. Tigre 'dcyāy. The suf­ fix -ay may also be used without any gentilitial connotation to form a singulative.g. "small cattle". e. "people.g. Nouns formed by the addition of the gentilitial suffixes -iy > -I. labin(un). Tigre rdšāš. "single song". tdkenāy. The more common afformative of the singulatives is -at{un) > ā/a in Arabic. śē'ār. "birds") vs.40. e) Singulative 31. or nasr. 'ēber.252 MORPHOLOGY feminine in gender when they are considered as a plurality. "bugs". "cattle". Tigre 'addām. and 'addāmātāt. "ships. "a bug". so'n. should belong to different classes. bun. from 'askarun.g. A singular may be derived from the collective noun by means of an afformative. "birds". animals living in groups ("cat­ tle".g. śa'ârā. On the other hand.g. There is no apparent reason why enti­ ties of such similar nature as Arabic himār. and riisum. or directly from the col­ lective in the absence of a countable singular belonging to the same root. 'ebrā. rasasat. "single pigeon". "gnats". from qadar. sir. e. "people". "pinions".

was orig­ inally replaced by one between "ergative" and "predicative". In Arabic. D. tā. wam-an and yam-an. hama. Instead. instrumental). The "ergative" is marked by u. A similar situation is found in Libyco-Berber which shows close links with Semitic. The feminine t. the former being used when the noun is acting (cf. the familiar con­ trast of nominative and accusative. while the "predicative" or non-active case is marked by a in the singular and by / in the plural.) for the object case. and 6. "drinkers' company". with a plural in -ot-a. "water(s)" — a plural noun attested in all Berber dialects — can be interpreted respectively as *ū-am-an and *ī-am-an: . "bad".1.(plur. i. The Cushitic case system appears quite clearly in Oromo which possesses an "ergative" in -n(i) (e. This contrast is close to the distinction of the "agent case" (casus agens) and the "patient case" (casus patiens) in the so-called "ergative" languages. partly postpositions. "the man". It has two cases expressed by the vocalic alternation which affects the first syllable. Cushitic and Libyco-Berber have two basic cases. Greek epyarns) as instrument or subject. plur. šārib(urì).g. "a drinker". We can assume that the plural case marks were originally pronounced ū and I so that. tū-1 to-. Bedja has a prefixed case marking with ū.g.CASE INFLECTION 253 31.. and the syntactic relations of nouns were indicated either by the word order or by the use of prepositions and the like.) for the subject case. Ancient Egyptian.. Instead. shows no trace of case inflection. harkan. the active participle may function as a singulative of collective nouns of the pattern fa7 / fa'al which designate human beings.e. object case.(sing. e./ tē-). namni. Thus. "the man". but they do not correspond exactly to classical usage. but the name casus patiens suits the Afro-Asiatic "predicative" only in part.) or ē(plur.) or ā. and Southern Agaw (Awngi). "by hand". e. or subjective and objective.(sing. There are other nominal suffixes in Cushitic.morpheme precedes the case marking (sing. subject case. and a "predica­ tive" in -a (nama. as the singular of šarb(urì).g.g. has developed a rich operative case system. Case Inflection 32. while the latter at once defines the predicate and the object. for instance. e.42. predicate). To what extent the Semitic languages originally possessed case distinctions is a debatable question. generally reduced to a after the feminine tprefix and in Tuareg. with a plural in -on(n)i < -ot-ni. the non-active component of the sentence.

the "ergative" or "agent" case in -u. several facts suggest that Semitic nouns were ini­ tially diptotic and that two cases were distinguished like in Cushitic and in Libyco-Berber: the subject case or "ergative" in -u and the non-sub­ ject case or "predicative" in -a. "cow". "may one draw water from the well" (both examples in Tarifit). Several peculiarities of this diptotic sys­ tem indicate that it is closely related to an ergative language structure: 1° the coincidence of the "nominative" case with the "instrumental" or "locative".5) and partly corresponds also to the two oblique cases of the "classical" languages. 5° the use of pronominal affixes of the verb referring not only to the "agent" but also to the non-active compo­ nent of the sentence (§36. the active principle of a process is not viewed as the subject of the verb expressing the action. a-funas i-funas-dn u-funas u-funas-dn fern. a) Diptotic "Ergative" Declension 32. and the non-active or "predicative" case in -a. " b u l l " . The well-founded assumption that Semitic originally had two cases (§32. which has a predicative function (§32.11. leaves us with the same morpho-syntactical opposi­ tion u : a as in Libyco-Berber. non-active ta-funas-t ti-funas-in ergative tu-funas-t tu-funas-in Although authors generally believe that Semitic substantives and adjectives originally inflected for three basic cases. viz. plur. The questions related directly to the case system will be examined briefly in the following paragraphs. 32. In ergative languages. "he filled the jar with water". 4° the existence of an absolute form of the noun which originally corresponded to the case in -a. and accusative.3. the genitive and the accusative.16). the vowels of the plural ending being those of Tuareg dialects: w masc. 33.254 MORPHOLOGY ydššur a-ġarraf s waman. nominative. 2° the function of the non-active case in -a in intransitive verbal or nominal clauses. 6° the lack of a common Semitic passive voice (§41. gen­ itive. viz. traditionally called "nominative". with the examples a-funas. and ta-funast. ad ydksi yaman z g anu. called "accusative".8-12). but as the instrument of .2. 3° its use to denote the construct state of the noun.43). The system may be presented schematically as follows. non-active ergative sing.

marker. "for Enlil". although historical develop­ ments often introduce a formal distinction between these two functions. d 32. called "ergative". which uses the same basic forms in both functions.CASE INFLECTION 255 its realization. is an "ergative" plural which is introduced by the preposition s. and I-bi-i-lum or ì-lum-i-bí. while the concord with the non-active component of the sen­ tence is established in certain conditions by means of pronominal suffixes. This particular example does not raise the question of the concord between verb and subject. as a possessive. with the difference that the Semitic "ergative" finished by losing its instrumental function and became a nominative subject case. Both names pro­ vide incomplete sentences. e. "with the hand" (instrumental). with /'Hum/ indicating the agent (§32. while the a-case may appear with the construct state. as a rule. In intransitive utterances. u-rgaz. ba ti-'à-ma-tù /mā tihāmatul. However. Thus. and the instrumental case coincide. As for a-ġarraf. "God has named". from kūjguo. For example. e. . Ves­ tiges of the same system are preserved in Old Akkadian onomastics where the same "ergative" case in -um is still attested in names like En-num-ì-lí. gender. This is the reason why the agent case. it is the entity affected by the "filling". the Tarifit clause yaššur u-rgaz a-ġarraf s waman. si-in I-li-lu. called "locative" in Assyriology. É I-li-lu. the predicate is represented by the nonactive case. but the archaic form ktijga still functions as ergative and as instrumental.. i-na-'à-áš na-'à-su I-li-lu /yinahhaš nahāšu lllilul.g. as indicated by the a. "the house of Enlil". "the water of the sea". where /hennum/ is clearly an instrumental case.4. where the child is obviously understood. "the man filled the jar with water". "the hand" (ergative). and as instrumental in a verbal clause. the personals of the verb in the historically attested Afro-Asiatic languages agree with the agent in person. by contrast.g. like in Libyco-Berber where it is indicated by the a. which indicates the object affected by the action. the Caucasian Chechen language distinguishes kujgaca. since the verb is missing in the first one and the subject in the second one. contains two "ergative" cases and one non-active case.18). while the "predicative" became.prefix. This distinction was not introduced in Afro-Asiatic. A similar situation occurs in Caucasian languages. Palaeosyrian ritual texts seem to preserve some archaic phrases with the "ergative" w-case used like in Libyco-Berber after a preposition. "water". In Libyco-Berber. an accusative object case. "By the grace of my god". "jar". and number. is an "ergative" singular and waman < *ū-am-an. "he will certainly recover thanks to Enlil". "man". because the god and the child are both masculine and both in the singular.

"his mother is the sister of P N " .3). 32.g.3) has the same origin.5. "the child is buried". by the Oromo clause niti-n hamtu-da. "why does your tablet not make it clear?". Old Akka­ dian.g. "church"). e.5. but it should rather be consid­ ered as a vestige of an old syntactic feature. In fact. Old Akkadian tērtakunu lillikam. Tarifit Muhnd d a-mqqran. and in East Gurage (§32.g. used independently from the accusative. Arabic gā'a rākiban.2). "the king of the city") and in Amharic (e.8). "you are coming (in) crowds". Babylonian alaktašu šaniat. the logical subject was originally con­ ceived as the instrument by means of which the signified condition was actually realized. ummašu ahāt PN. "Servant of Dusares". "the wool of the sheep". as shown e. however. occur in Nabataean Arabic as reflected by the Greek transcription Ap5a5oi)aapos. the "ergative" case characterizes the log­ ical subject of these intransitive sentences. while the non-active component of the sentence is expressed by the so-called "predicative accusative" (§52. 54.256 MORPHOLOGY e. while the nomen rectum is marked by the w-prefix. viz. The personals of the verb agree always with the logical subject. "his gift". The original system was later reinterpreted along the lines of the contrast of nominative and accusative. a-ham u-rgaz.e. still operative in Ethiopic. as a non-active component of the phrase. batd krasriyan. kl tuppaka pānam ul šuršu. in Classical Arabic call phrases . in Classical Arabic. " E l is high". Traces of a construct state in -a. The same mor­ pheme a is suffixed to the nominal predicate in Palaeosyrian. by the Amorite name E-lura-ma. high.. "his way is dif­ ferent". Old Babylonian mārašu. "the tent of the man". i. Old Akkadian qīštašu. "house of Christian". This -a is usually explained as a paragogic vowel. being bad. ndguša hagar. in Ge'ez. or by the Tachelhit verbal clause imdl u-fruh. buried. tātūna 'ajwāġan.g. but also in the nominative. before pronominal suffixes.g. The construct state in -a appears occasionally also in East Semitic. and Amorite names. since the perfect of the suffix-conjugation goes back to a stative which is basically a nomi­ nal predicate (§38. "the woman bad-is". This construction exactly parallels the syntax of the Libyco-Berber noun phrase with the nomen regens hav­ ing the (3-prefix.g.g. "he arrived (as) a rider". The ending -a of the perfect in Classical Arabic and in Ethiopic (§40. "his son"). etc.g. ta-dut-t w-ulli. not only in the accusative (e. "may your instruction reach me". The -a ending characterizes the construct state in Ge'ez (e. "Muhend is great".11. e. The noun determined by another noun can be regarded as a kind of recipient and be considered. 33. Instead. e. therefore.

Also the Classical Arabic vocative in -ā I -an may go back to this "absolute" form of the noun. "our Lord!".g. and why several Semitic divine names also end in -a.g. etc. To what extent this ending cor­ responds to an ancient usage is a debatable question. One might posit an initial distinction between a genitive denoting an "agent" and a genitive denoting a "patient". and nib'al hab-bayit. In Afro-Asiatic.CASE INFLECTION 257 and exclamations (cf.12). "the tent of the man". and in Syro-Palestinian colloquials before pronominal suffixes.g. "he owns the house".r a . the likelihood that. "oh! uncle!". "bull" in Tuareg. "country". "battle". "the house of the chief") from the object case in -a (e. the construct state in -a occurs only when the nomen regens is an accusative. A similar sit­ uation can be assumed in the ancient Semitic languages before the devel­ opment of a new case alignment. e. The existence of an "absolute" form or citation form of the noun is a characteristic of ergative languages. In Afro-Asiatic. may go back by nominalization to both bā'al hab-bayit. Cushitic Oromo distin­ guishes the subject case in -n(i) (e.. d d 32. "the owner of the house". "meat". rabbanā.6).g..e.. ba'al hab-bayit.g. dust") and in some West Gurage dialects. A-dam-ma. the first clause would require the a-"predicative" *bayta and the second one would use the «-"ergative" *baytu. The genitive of the diptotic Semitic declension ends in -a (§32. no case endings are used in Modern Arabic and. like in LibycoBerber. e. Ishara. mana motti). like yā bna 'ammī. Such a distinction inevitably touches upon the question of the ambiguous status of the so-called genetivus obiectivus (§51. i-rgaz-dn imdqqr-an. in Classical Arabic. "oh! heroic rider!". 32. "our mother". "oh! son of my uncle!". muk-a. §32. like Abba.. Palaeosyrian 'À-da /Hadda/. "earth. The same origin may be attributed to the ending -â of the absolute state of the nouns in Gafat (e.?" The answer is.g.d a. "tree" in Oromo. This would explain why a large number of words passed from Old Akkadian to Sumerian in a form end­ ing in -a. i. 'ummane.h a . a-ham u-rgaz. Otherwise. "big men" in Tarifit. such a s d a m . e.g. since it is the "ergative" that often denotes the genitive relation.12). the mark of the "predicative". Labba. 'a-rākiban kamlyan. e. m a . afàrà.g. manni motti. as indi- . in the Hawràn dialect. "our road". bàsàrà. yā 'ammā.7. This is the form of the Libyco-Berber or Cushitic noun given in answer to a question like: "what is the word for. e. "the house is owned (by him)".6. a-funas. darbane. this form coin­ cides with the non-active <z-case.. Ela.

maf'al. and among the Semitic loanwords in Sumerian. and with theophorous elements.9.. Ma-lik) have been regarded by some scholars as vocatives in form. at least in part with names ending in -ān. "from a man".10). Old Akkadian or Amorite languages. there is a postposition -i which includes the idea "out from" or "away from". e. several nominal patterns of Classical Arabic ('afal. as well as many categories of proper names. the use of which is still inconsistent in Palaeosyrian and in the Old Akkadian onomastics in general. Contrary to the case endings of the actually used Palaeosyrian. "the child is buried". Dagan. the stems ending in -ā 'u).258 MORPHOLOGY cated by Tachelhit imdl a-fruh. this -a ending never appears as -am(a). Also in the singular. As for the "new" genitive marker -/. the plural stems of the type fa'ālilu. etc. this explanation cannot be accepted.g.10. it should be made on a compara­ tive Afro-Asiatic basis. ilma nama-ti. such as personal.6).15-17).8. The divine names with no case endings (e. geographic. 32.g.3. e.g. and imdl u-fruh. In Highland East Cushitic.. "the son of a man".g.> -i (§ 29. 32. with some place names. Plausible as it may appear for theonyms. This situation can best be explained as still reflecting Semitic languages or dialects having no well-established declension.fu'āl. which may derive from a postposition. e. b) Use in Proper Names 32. probably because it reflects idioms spoken before the introduction of the mimation (§33. because forms without case endings appear also among d d d . manni kun (subject) kan (object) abbāko-ti. There are only two cases in the plural and the dual of the Semitic languages (§31.g. fu'al. "from Mecca") and it is a postpo­ sition -ti that expresses the genitive relation in Oromo (Lowland East Cushitic) . or using the non-active case with the ending -a as citation form (§32.fa'lān. "this house (is) the one of my father". Makkī. The Semitic gentilitial suffix -iyexpresses a similar idea (e. divine. and month names. it is likely to have the same origin as the gentilitial suffix -iy. Many nouns without any ending or with the ending -a appear in Palaeosyrian and Old Akkadian proper names.13. and syllabic spellings indicate that the situation was similar in Amorite and in Ugaritic.41). Hadiyya mene-i. "he buried the child". the diptotic declension char­ acterizes Ge'ez. Further research work is needed in this field. Ra-sa-ap.

sa . ìr-ra-na-da. "The root is firm". they possess three basic cases for the singular of most nouns. always charac­ terized by conservatism (e. uru d c) "Classical" Triptotic Declension 32. Su-ru-uš-ki-in. They rather reflect the stage of the language prior to the introduction of case marking. e. "hunger". names of months (e. sum < turn-.g.an < karān-.g. As for the ending -a. tradesman".g. in Ge'ez: konki bd'dsita. an explanation which is generally proposed for a few basic Libyco-Berber common nouns which are in the same situation.g.11. "The heir is firm".tu < śadw-. "he was a brother to me". " I I is father". in East Gurage: giddirân ydhanál. in combination with the verb "to be. thus distinguishing one subject case or "nominative" ending in -w. bdrhāndka sdlmata konaka. "His ancestor is a rock". ga . 'atr b'l = Ki. g 9 kì 32. cf. present a somewhat different picture. Sú-ra-Ha-am-mu-ú. kra. Ha-ab-du-A-su-ra. and it is used as such in a productive way with diptotic Ugaritic names (e. The ending -a is likewise the morpheme of the non-subject case of diptotic nouns in all the Semitic idioms which have preserved the case inflection. and common words borrowed by Sumerian (e.g. Ba-ah-la-DiNGTR. "vineyard". A-ba-Il. in Arabic: kāna 'ahā ll. "mountain".g à r < tamkār-.g. With the exception of Ge'ez. A-šùr^). The same -a morpheme still character­ izes the predicate in Classical Arabic. "Settlement of Baal"). §33. and Amorite proper names. 32. "Servant of Baal". "lira is exalted". contrast subject cases Nu-ri-ia-nu and Pu-lu-zi-nu with genitives Nu-ri-ia-na and Pu-lu-zi-na).13).ra . and in some Gurage dialects of South Ethiopic. "The progeny is firm"). "Servant of Ashur".g. Ha-ni-it). Ba-hi-ir. dam .CASE INFLECTION 259 Semitic toponyms (e. U -ga-ra~at . . Palaeosyrian and Old Akka­ dian included. Such words could not possibly be explained as vocatives. The morpheme -a can also be the morpheme of the genitive (§32.7) in Amorite and Ugaritic proper names (e. Li-da-at-G\.13. as fad. Old Akkadian. The "classical" Semitic languages. Ha-ab-du-Ba-ah-la. "El is lord". "thirst". it is the morpheme of the predicative in Palaeosyrian. in Ge'ez. laz. to become". "supplier. "you became a women". A-ru -ga-a^\ Ba-sa-ar^. "your light has become for you darkness". "(some)thing". and two non-subject cases. usually called "genitive" and "accusative".12. "he became big". A-pil-ki-in.u-/a I'Atru-Ba'lal.g. "garlic") or used in anthroponomy (e.

Old Akkadian. acquiring thus an adverbial meaning (e. It is even used with words functioning as prepositions (e.g.14. mu-lu-iš. "Rely upon the king!"). Its ending is -i (cf. but no functional relation connects these forms.7) with a pos­ sible allophone -e. also at Ebla. and Old Babylonian have also a dative-adverbial or terminative-adverbial case in -iš or -eš (e. sa-da-bí-iš. with infinitives (e. for". Old Assyrian.5). inherited from the -a of the unique non-sub­ ject case which has lost its predicative function with the exception of the formations still attested in Arabic and in Ethiopic (§32.16. . "on behalf of"). but the <2-case develops for sure from a nomi­ nal predicate through the participial predicate (e.2). such as an adjective. an adverb. a-li-iš. "Damu is standing") into the verbal perfect in -a (fa'ala.3).g. 40.2-4).5.18-20).g. to which they are related genetically.260 MORPHOLOGY 32. called "construct state" (§33. Its ending is -a.g. The following picture of the case inflection emerges thus for the "classical" Semitic languages: Singular Nominative Genitive Accusative -u -i -a Plural -u -ī/-ē Dual -a -ay d) "Adverbial" Cases 32. The singular endings -u and -i are quantitatively distinct from the corresponding morphemes -ū and -I of the plural. Beside the examples with -iš.g.g. 32. §47. "into the Tigris") which expresses the idea "with. Idiglat-es. with adjectives and participles. This mor­ pheme occurs frequently with nouns forming elements of personal names (e.17. The genitive is a subjoined case (nomen rectum). "strongly". a determinative pro­ noun (§51. as it seems (e. etc. Qāma-Da'mu. na-da-ni-iš. mah-ri-is. The accusative is used for the object of a transitive verb and for the term of reference of an intransitive verb.7). " i n front of"). determined by its antecedent which can be a preposition (§48. It is uncer­ tain whether this predicative function reappears in the Aramaic "emphatic" state (§33. 32. §33.11). "instead of". There is also quantitative opposition between the sin­ gular ending -a and the dual morpheme -ā of the subject case. §32. or another noun (nomen regens) in the bound form.15. to. Sar-ri-iš-da-gal.g. "to give"). da-ni-ii. " i n addition to".

suprānuššu < supr-ān-um-šu. within") and could be com­ pared with Arabic adverbs ending in -u (e. a kid with both hands".g. Proto-Semitic had a so-called "locative" in -u(m). 4 32.1). libbu. "By the grace of my god". "to the enemy") and from Ebla (e. which may have the same origin. "for/with its/his messenger". and rittū'a (< *rittū-ya). as already mentioned (§32. e. "for a pair of shoes".g. as shown by ma-za-lum-su. e.3). kì 4 . and En-num-ì-lí. "By the hand of my lord". In reality.18. Forms without the final -m are attested already in Old Babylonian (e. The func­ tion of the instrumental / locative suffix -uml-u appears in Standard Lit­ erary Babylonian fixed phrases like šēpā'a (< *sēpū-ya). "the day before yesterday" (§ 29. Its traces survive in several Semitic languages. and there is little doubt that it coincides with the subject case. éš NÌ. Most of the nouns to which the ending -um is attached form adverbs or preposi­ tions (§47.14. as well as with Syriac and Ge'ez adverbs with final long -ū. ba'du. which may derive from an original -um > -dm (§47. the instrumental suffix -(u)m used in one parallel member is balanced by a preposition in the other. and of the East Semitic and South Ethiopic enclitic -ma. "with his claws").g. "on a sudden". In Ugaritic. "later". enough".3).55-57). "for the journey to Harzu". "and". but employed also as preposition in Palaeosyrian. which should more conveniently be called "instrumental".g.KAŠ -S Û. The pronominal suffix follows -um also in Palaeosyrian.3). like I-dum-be-li. §48.g. "above".CASE INFLECTION 261 there are several East Semitic forms with -uš. "and".g.111. "for his journey". since the alternation w : m is well attested (§49. Ge'ez lā'lu < lā'lū. .55). and in a few Old Akkadian names. Syriac kaddū. in texts from Mari (e. Iqh 'imr dbh bydh IIIVa kl'atnm (KTU 1. This double use of a particle as preposition and postposition is not exceptional in Afro-Asi­ atic and it may be compared with the parallel existence of the common Semitic conjunction wa-. Old Akkadian and Assyro-Babylonian balum. but not šilšom.g.g. "in the heart of. "sufficiently. és NÌ.10). the morpheme iš is no case ending but a particle used mainly in ancient phases of Semitic as postposition. with which it formed the "ergative". iš mas-a-né-en. "he takes a lamb of sacrifice with his hand. fawqu. e. "above". iš na-ak-ri-im. "below"). "at my feet". Hebrew pit'om. The suffix -{u)m is preceded here by a morph (a)n which appears in this position also in Assyro-Babylonian (e. which might have devel­ oped secondarily from -iš. " i n my fingers". "with­ out".KAŠ Har-zú . tahtu. Besides the "classical" cases already considered. at Ebla.

allows of the explanation of this -a/-a morpheme as the ending of the non-sub­ ject case. as shown e. "each of them entered into his house"). In East Semitic. while others end either in a consonant or in a vowel.).2). e) Historical Survey of Case Inflection 32. ūmisam.19.g. have the same ending -am without the final -m and with the consequent lengthening of the vowel (§47. e. "he lifts his hands heavenward". 4 A 32.IV. e. Šmmh. and per­ haps in Ge'ez (e. This ending -am is employed in Hebrew without the postposition -iš to form the adverbs yomām.g. bārā.21.11. "emptyhanded". hadaddni gar gardni hid. Subsequently. Many names are not declined at all. but the concomitant use of verbs of movement. Bābelāh. employed also with the accusative (§52. The weakening of this postvocalic -h (*-ah) and the consequent reduction of the postposition to the vowel -a are already noticeable in Ugaritic. daily" in Old Akkadian and Assyro-Babylonian.g. we may surmise that the Aramaic adverbs such as 'ar'ā. "to the orchard they went down (dual fern.14. there was no longer a fully functioning case system in proper names and in nouns.3).g. sarqan wa-ġarban. The Hebrew ending -ā(h) denoting a place relation (e. in Arabic. "heavenward".262 MORPHOLOGY 32.14.22-23) compared with ns'a ydh šmmh (KTU 1.g. "he entered into the city") and in mod­ em East Gurage dialects of South Ethiopic (e.20. In view of the analogy with the suffix -um/-u (§32. This reduced directive postposition -a is preserved not only in Hebrew but also. From the Middle Babylonian period on. as it seems. The postposition -is can be combined with the ending -um and alternate with a prepositional phrase. (§47. 'arsh. by š'a ydk smm (KTU 1. like in an Old Akkadian love incantation from Kish: ki-rí-šum tu-ur -da tu-ur -da-ma a-na kirīm. case endings on proper names were either dropped or drastically shortened. However.g. "on the ground.18).2). "by day". bo'a hagara. which is mostly i. "to Babylon") was regarded by some scholars as a survival of the accusative ending -a. "outside".C.g. where it is regarded as an adverbial accusative (e.4-5). "earthward". case inflection is in full use down to the middle of the second millennium B. . "day by day. below". Ugaritic has shown that the -h in question was originally a consonantal postposition expressing motion towards a place. they went down to the orchard". The postpositive -is can be used also with another adverbial ending -am. etc. "eastward and westward"). rēqām.

ks'i (genitive). case inflection is in full use down to the twelfth century B. and -a are encountered without reference to their syntactical function. and the -ā of the dual (e. "my eyes": EA 144. -He.. ks'a (accusative). ks'u (nominative). ša-me-ma: EA 264. without any case distinction. The relation expressed by the former genitive is sufficiently indi­ cated as such by its antecedent. the ending -w often occurs instead of the -a of the accusative. "small cattle": EA 263. 32. case inflection is fully used in Old Canaanite. 'i. hi-na-ia = 'ny. At Emar. the use of the Latin case names "nominative".17). 'u. "throne".24.7. as may be seen in Ugaritic nouns whose final con­ sonant is \ vocalized 'a.g. "shades (of the dead)". with the following scheme: object (accusative) precedes the verb. The syntactical function of the case inflection is taken over by word order.17-19) and by the optional use of a particle intro­ ducing the object of a transitive verb. 32. subject (nominative) precedes the object.22.16). In West Semitic.9). rp'im (plural non-subject case).g. the case system was not in full use any more and the irregularities are numerous. e. in the early 12th century B. sú-ú-nu = s'n. had been dropped in the spo­ ken language or had become merely vestigial.g.12).17). and with them the entire case system. ma-at-ni-a = mtn'. the -i of the genitive (e. "governor": EA 256.g. while the plural is marked by -ī/ē or -āni. the -a of the accusative (e. all three endings -u.g. sú-ki-ni = skn.13. "supply": EA 337. In North Semitic.21). and that there was no longer an operative case system. the former genitive. there is no longer a distinction between subject and nonsubject plural forms. In the NeoBabylonian and in the Neo-Assyrian dialects. kept in writing because of the syllabic nature of the script. while the subjoined function of a noun. 32. ša-mu-ma = smm. This indicates that the vocalic endings of the singular. "accusative" in reference to these languages is more convenient than strictly scientific. rp'um (plural subject case).g.. as may be seen in the Amarna glosses showing the -u of the nominative (e. the former accusative (§52. "heaven": EA 211. results from .23. Therefore.C. "genitive".g.C.9. and the vocalic case endings of the singular disap­ pear.CASE INFLECTION 263 As for nouns. the -ūma of the plural subject case (e.C. and in Aramaic. In the "Canaanite" languages of the first millennium B.10-11). It simply suggests the syntactical function of the nouns which is indicated by word order (§50. the -Ima of the plural non-subject case (e.

g. especially at the junction of a noun in the construct state with its complement. have no longer a syntactical function. §7. "after the priest has taken water"). Because of the lack of vocalization in the script of ancient South Arabia. which means that the case differentiation had become merely vestigial by the lst-3rd centuries A. apparently with the -i. The modern South Arabian languages do not possess any case distinction. 32.of the former genitive. 32. the so-called "paragogic" vowels of the Hebrew grammar. In fact.43). Ge'ez distinguishes the subject case and the non-subject or oblique case. and in the modern colloquials which developed from Pre-Classical Arabic. and in combination with the verb "to be. In Neo-Arabic or Middle Arabic. one encounters the nunation in the -in form (e. standardized for diction in the schools of the Abassid empire (§7.D. 42). band Ba'or. baytu. However. this distinction is already blurred by the mid period of the epigraphical evidence. while -u < -un is preserved in some Yemenite colloquials when the noun is indeterminate (e. However. "a house". the case inflection does not exist. we are limited to the contrast between the two forms bnw and bny. e. The faint traces of case endings. "son of Be'or". "Face of God". but al-bayt. the construct external masculine plural is the one gram­ matical feature in which a case distinction would be apparent.g. saba nas'ā kāhdn may a. darbin. where another explanation is however possible (§32. They may play an euphonic or rhythmic role.38. 32. 'Abdī. "a road").'Ēl. Panū-'Èl.20).g.11). The archaizing use of the case system in Classical Arabic derives from conservative dialects. "Servant of God".25.g. This morpheme is used in the expression of the direct object (e. Pre-Classical North Arabian had no longer a fully functioning case system in the 3rd and 4th centuries A.27. "the house"). where the morpheme -a expresses the syntactic predicative situation (§32. often in poetry and in proper names. The same use of the suffix -á is encountered in East . to become".26. which testify to the existence of a subject case and of a non-subject case in the plural. In some dialects of Northern and Central Saudi Arabia.264 MORPHOLOGY its place and from the bound form of its antecedent. a few faint traces of case endings subsist in Bedouin vernaculars of the Ara­ bian peninsula.D. marked by -a. in the expression of place relation.. although the Nabataean and the South Arabian scripts continued to indicate final vowels indis­ criminately (cf.

2. §33. and the absolute or indeterminate state. The "States" of the Noun 33. betu. However. testifying indirectly to the earlier existence of two distinct cases in plural nouns.g. Dāwit walada Salomonhā. E .14). the determinate state. called in the tradi­ tional terminology the "states" of the noun: the construct state.14).THE "STATES" OF THE NOUN 265 Gurage dialects of South Ethiopic.g. "the house").g. a feature paralleled in other Ethiopian languages.15. either noun (e. "David begot Solomon". bidā'a-w musāfìr-u yàthidādral. Ge'ez occasionally introduces the direct object by a prefixed la-. "he administers the merchandise (and) the merchant(s)". A few Greek transcriptions apparently show that Early Ge'ez had the -u (nominative) and the -i (genitive) case endings. Traces of a case distinction are preserved also in South Ethiopic where various developments can be observed. in Argobba. "he has sold the sheep") and in some other phrase types. Ge'ez has a par­ ticular case ending -hā for the accusative. especially in the expression of the direct object. The definite article -u is occasionally paralleled in Amharic by -itu for the feminine (usually -wa < -u+a) (cf.17). and in Gafat when the definite noun is used as direct object (e. "the house") and in Argobba (e. Clas­ sical Arabic suyūfu l-'a'dā'i. The South Ethiopic plural ending -ac I -ode originates from the plural oblique morpheme -āti (§31. This marker is identical with the postpositive definite article -u in Amharic (e. A noun followed by a genitival qualifier. "the house"). For proper names. 32. It is a unlikely that these postpositive articles -u I -i have originated from the case vowel preceding the nunation.1.g. and there is a corresponding -/ morpheme in various Gurage dialects (e.g.g.28. Amharic bāgun šāttà. 17).g. "the swords of the enemies"). or attached . gari. There is a direct object marker in Harari which is -u after a consonant and -w after a vowel. Their origin seems to be different (§33. these were first reduced to -d. bedu. e. the pred­ icate state. The Semitic noun can appear in four different syntactic situations. a) Construct State 33. I f so. although nunation appears in Amharic. which may imply four formally differentiated forms. and then disappeared. e. as in ancient South Arabian languages (§33.

"the time he died". Ilū-da-nu.5) and it cannot be regarded as a simple paragogic vowel. especially before pronominal suffixes (e. "Ishtar is great".g. Eštar-ra-bí-at. This ending derives from the case-form of the non-active component of the sentence (§32.g. e. 33. "the king of the city". -o remain unchanged in the construct. a different picture seems to emerge from Ge'ez and from Amharic that indicate the construct state by an additional ending -a. "your lord"). The contraction of the vowels explains why Ethiopic nouns ending in the long vowels -ā. the dual or plural morpheme and the case ending are usually added to the stem (e. a-za-me-tù du-hu-rí si-ne-mu /lazamītu duhūli šinnīmu/.6-10) corresponds in East and North Semitic to that of the absolute state (§33. but they are capable of deletion. or an asyndetic relative clause (e. It repre­ sents the case-form of the predicate in a nominal clause of an ergative . Sabaic rglhw.15) or of the stative (§40. cf. or Amorite (§32. and they may be not represented in the consonantal writing systems when they are reduced to a simple vowel (e.11). Arabic rabbuka. "the gods are powerful".g. Ge'ez ndgusa hagar. Besides. while those ending in -i have their construct in -e. §57. l 33.7-14).2ff. the article may be prefixed to the construct state in well defined circumstances. Phoenician bt 'Inm.g.266 MORPHOLOGY pronoun (e.).11). e. e. "his two legs"). "temples of gods".3. a noun in the construct state has the stem form with no prepositive or postpositive article (§33.5. in Old Akkadian: Ea-rabi. As a general rule.g. Aramaic 'aynē 'ânāšā. Old Akkadian.4. "spell­ binding of the rear of the teeth". -e. However. is said to be in the "construct" state or in a "bound" form.g. "the eyes of a man".5). as opposed to the "free" form. b) Predicate State 33. in parallelism with "spell-binding of the tongue".g. "the wealth of the tradesman"). The standard form of the predicate state in a nominal clause (§50. "Ea is great". This rule is followed already in Palaeosyr­ ian. and with no addi­ tional mimation or nunation. Assyro-Babylonian māssu < māt-šu.g. However. Arabic hīna māta. a form ending in -a is found in personal names regarded as Palaeosyrian. Classical Arabic mālu tāġirin. as in Modern South Arabian (§33. Safaitic s nt b'yt hwlt m's. "the year when Hawlat has over­ come Ma'as". At first sight.9) and in Tigre (§33. "his country").

The mimation. The predicative in -a of the Palaeosyrian. Amorite. "cattle and flock" (cf. It often marks class and species determination. as when affixed to a substantive already mentioned or being in natural connection with a given situation. e.8. šm-tì. It may also express the idea of "the particular one". "she is gone". the predicate of the nominal clause is in the nominative of the absolute or indetermi­ nate state (§33. "the boy is small". "God is the (particular) father".g. "who is the (i.7). for that matter. to become" is used (§32. "he owns". c) Determinate State 33. as in Safaitic klmh h's d.5) appears to have such a secondary function when one compares names like Su -be-li. A noun which is neither a proper name nor in the syntactic situ­ ation of a construct or a predicate can be made "determinate" by an additional prefixed or suffixed morpheme which may mark individual determination. Enclitics are here much less frequent than proclitics.g. instead. Arabic malaka. which presents a close resemblance with the Semitic stative.g. the -a of the predicate state subsists in Classical Arabic and in Ethiopic when the verb "to be. "this") man?" (cf. of the Indo-European languages as well.14).10).6. and probably in North Semitic (§33.3). Ge'ez qatala. seems to lack any connec­ tion with the determinate or indeterminate state of the noun. e.3). " I am/was saying". "Muhend is great".e.15). hr-ti. exactly as in Libyco-Berber. both definite and indefinite. instead. and replace a pronominal suffix. § 33. dd-kì. though not indispensable. 33.THE "STATES" OF THE NOUN 267 language (§32. in South Semitic (§33. and Old Akkadian onomastics (§32. ] 4 4 4 . e. "he killed".11. Tarifit Muhnd d amqqran. In Classical Arabic. contributed to the growth of identification. e. As for the article proper.4. How­ ever.g. "He is my lord". "My god is my father". and A-ba-Il. § 33. as in Arabic man-i rraġulu. it is a very late acquisition of the Semitic languages and. e.10). This mor­ pheme -a must go back to Afro-Asiatic since it occurs also with forms of the Egyptian old perfective or pseudo-participle. as in Hebrew habbāqār wd-hasso'n.g. and Su -be-la. and it also explains the verbal perfect ending -a in both languages (§40.13).11. 'al-waladu saġīrun. "you are content". at least in the North and East Semitic languages (§33. This grammatical device.4). or A-bi -ì-lí. "a (particular) lion roared at him". but they are attested in Aramaic (§33. "the man" as distinguished from "man" in general. "He is the (particular) lord".16). 54.

36).) qora.g.C. thus ['ū-malko]. onwards and by h already in the 8th century B. Its earliest attested form is ha-. [malka] or [malko]. In this particular case. In the "predicative" hypothesis. Mehri a-saar. in Arabic.8. Lihyānite.13-14). e. sing. "he has opened (it) the grave". and in the Modern South Arabian languages where the definite article a-1 áis prefixed to definite nouns the initial element of which is a voiced or glottalized consonant (e. in Pre-Islamic North Arabian. In the first hypothesis. Moabite.). therefore. e. since it is unaccented in speech. "mother". and the usual transcription -ā' is more convenient than strictly scientific. used in Hebrew. Phoeni­ cian. the Aramaic postpositive article is a suffixed pronominal element.C. either the original form of the postpositive article is preserved dialectally. In Soqotri. 33. 'a'am < *hā-'umm. "the kings" (plur.). "the king".g. "king".. "the king"). sing. yuspul-la hdudca. Edomite. Safaitic. this "emphatic" state became the common form of all the nouns in the Eastern dialects of Middle Aramaic and in Neo-Aramaic.7. Yam-malke] < 'an-malkē. This mor­ pheme -ā is indicated in writing by ' from the 9th century B. originally iden­ tical with the determinate state (e. like in South Semitic and probably in North Semitic (§33. which character­ izes the so-called "emphatic" state of the Aramaic noun. Other means ful­ fil the former function of the "emphatic" state: the Neo-Aramaic of the Tūr 'Abdīn area (Tūroyo) expresses definiteness by prefixing shortened allomorphs of the demonstrative (§36.g. Thamūdic. it has apparently lost its spe­ cific function and has become a constitutive part of a number of nouns. "the daughter" (fern. however. This implies that the original morpheme *-ah was weakened very early to -a' and finally reduced to a vowel -ā marked either by ' or by the mater lectionis h. ['ī-barto]. In other Neo-Aramaic dialects. e.g. at Hamath.268 MORPHOLOGY 33. Similar examples are attested in . The predicative morpheme -a might reappear in Aramaic with its determinative value as the postpositive article -ā. Śheri à-ġarb. Ammonite. ptehlē-la (fern. definiteness is expressed by proleptic pronominal suf­ fixes added to the verb form. "the king" (masc. in the word mlkh. also in the Western ones. and in Tigre by a prefixed definite article. The determinate state of the noun is marked in the "Canaanite" languages of the first millennium B. Anyhow.C.). "they will take (her) the bride along" (Ma'lūla). in Modern South Arabian languages.g. malkā'. or -h is already a simple vowel letter. "the gazelle"). Its ending -ā is pronounced nowadays as a short vowel. the original consonantal value of ' as a glottal stop is in doubt. "the large well-bucket".

e. the article is sometimes preserved in Late Biblical Hebrew as well. melek. it has long been assumed that its original form was hn-.g. found in Lihyānite before the consonants ' and ' (e.g.10). which does not geminate the gutturals. e. "the grave"). "Romans". This hypothesis is weakened by the fact that the Mehri numeral "one" is tāt (§35. However. hā'āres. "this day. e. Mehri bayt. today". "this camel") and by some Hebrew expressions using the article: hay-ydm. contrary to Thamūdic and to Safaitic where it is preserved. a-bdtk.THE "STATES" OF THE NOUN 269 Mehri. l-hmr't.g. /-. as suggested by its demonstrative meaning in Pre-Islamic North Arabian (e.g. as gdr. "house". This usage was later extended to some foreign place names. but this happened under Greek influence (e. hn-'sl.g. "king". serving as an indefinite article. In Hebrew. b-hā-'ām. "father". hap-pa'am. hn-'zy. as shown.7). "the smelter"). A peculiarity of the Mehri and Śheri definite article con­ sists in being prefixed to nouns determined by personal suffixes. e.g. "with the camels". for instance. However. initially a deictic particle "this here. by Aramaic 'ârdmā'ê. hn-'lt.g. "this night". "in the camp".C. was originally limited to names originated from common nouns.e. "this time". f| T6ur|). as Neo-Punic h-Rm\ "Rome". which should be related to the Arabic word han. k-.10. "for the lady". in languages that have no article h-l'-.is already attested around 400 B. e. hnqbr. with assimilation of d to the following consonant.25). "wall. com­ pound" (§67.3d).g. hal-layld. and to .18). "your house". "the socle". hayd. that here" (§49. 'nsk.6). it was also sug­ gested that the Zia-prefix in Mehri may go back to *had. ham-melek.g. In Masoretic Hebrew. b-h-dr. "the king"). haš-šānā. "Gades". the article is elided after the prepositions b-. The use of the definite arti­ cle with place names. the gemination is compensated in simi­ lar cases by the lengthening of the vowel a. The weakening of the h led to the Punic spelling ' of the article (e. "from" (§48..g. as Phoenician h-Gdr.g. and seem indeed to indicate that the original articulation of the article was han-. "this year". hayb. Some particular points concerning the use of the article have to be mentioned. "Rome". 10. These allomorphs parallel the Hebrew forms of the particle min. 33. or Meroitic Arome. Thamūdic hgml. Since the prefixed article is followed in Hebrew by the gemi­ nated initial consonant of the noun (e. "the land". b-h-'bl. "the goddess" (TSSI 11.g. — sometimes before other consonants as well (e. 33. "for the people" (II Chr.9. "one". i. "something". e.12). like and in Amharic (§33. DN han-'Uzzay). "hand". This form hn.

"the old man". "this here".in ancient Yemenite colloquials (e. should normally resume the preceding direct object and the postpositive 2 . although it is generally kept in writing. la-bd'ds-a. la-wdlād la-ddgge. "this" (§36.is used as definite article and its absence may signify indefiniteness. 'am-raġul "the man"). the / of the definite article 'al. and to the dual (e. the boys of the village").. "(some) boys of the vil­ lage". "the sun"). 33. sàbo-h tālsālam. "the boys of the village".38). dl-. "the four hundred").is assimilated in Arabic to most initial consonants of the noun (e.8).12.4). The -n could alternatively mark the accusative in Gafat (§32.270 MORPHOLOGY its local forms 'an.g. 'an-hulm. e. 33. to the broken plural (e. " I don't eat millet" ("frumentum non edo"). 'aš-šams. "her husband".g. s 'bynhn. The (')a is commonly dropped after and before a vowel (e.g.g. with the change n > I (§17.g. two Gafat phrases mentioned by H. Contrary to other Semitic languages.g. this opinion is confirmed indi­ rectly by the Mishnaic Hebrew demonstrative hallāz < *hal-'az. la-wdlād la-ddgge. viz. "the dream". 'rb'tn m'nhn. but a pronominal suffix. contrary to the situation reflected by the modern South Arabian languages and by the ancient Yemenite col­ loquials (§33. "a friend of the girl's parents". The determinate state of the noun was marked in Epigraphic South Arabian by the morpheme -n. il-. 'nhln. In modern colloquials one encounters the pronunciations al-.has also been regarded as a variant form of han.and probably younger 'am.g.. corresponding to Biblical Hebrew hazze < *han-ze and to NeoPunic h'z. Although there is no article hi. Ludolf in 1681 might belong to a dialect still using the same morpheme -n to express determination.g.g. §33.g. walād la-ddgge. with the word order: article + qualifier + qualified noun. "on the earth".in Lihyānite (hlhm. dl-bēt l-dkbīr.g. fatāy wāldat la-walat. "the statue). Like the n of hn-. is the beginning of the "South Arabian" and Ethiopic alphabet).11. "the palmgroves"). the particle la. e.28). attached to the verb.> 'al-.9) or by another noun (e. bdle-h tàlbàlam. to the external plural (e. This ending -n — or -nhn with duals and external mas­ culine plurals — is attached in epigraphical texts to the singular (e. 'alā l-'ardi. the Tigre particle la. Since the Semitic article and the determinate state serve also to mark class determination. vs. cf. " I don't molest a man" ("hominem non laedo"). "the two tribes").may be prefixed to a noun qualified by a pronominal suffix (e. it may imply the indefiniteness of the qualified element. "the big house" in the Damascus colloquial). la-gandāb 'anas. slmn. The Nabataean and later Arabic definite article 'al. I f it is prefixed only to the qualifying noun. In Tigre.

) / -wa (fern. and in Argobba (e. betu.g.g. dabru.'It. "to give birth". and *(bēt)u-ha > (bēt)-wa "her (house)". "the woman"). hn. "his". "her". "his (house)".) of the Ethiopian languages evolved by elision of intervocalic h from suf­ fixed third person pronouns like in *(bēt)u-hu > (bēt)-ū. wrh-h is encountered frequently in Sabaic date formulas. and with some Palaeosyrian divine names. AMA-ra-sù. and -wa (or -itu) for the femi­ nine singular (e. Similar cases occur in Harari where the suffix -In expresses a strong determination. instead of AMA.13). "do you eat (truly) Amharic meat?". 1 1 l á The use of the logogram TU. vs. "(T)his/The Adamma".36). ahmara bāsàr-īn tolak ?. Latin dominu-s. 33. despite the fact that the Gafat operative suffixes are of the -h type: -{d)ho. as in Tigrinya (§36. in Amharic (e.13.g. I-la-kab-ka-bu-ú /'īla-kabkabuhu/.17). Thus.THE "STATES" OF THE NOUN 271 article -dš should be affixed to the nouns (§ 33.g. "lord"). by North Arabian hn. gàġġā.g. e. This particular way of expressing determination occurs in Epigraphic South Arabian with the noun wrh. That structure may explain the Ara­ maic emphatic state as well (§33. dndscawa.'zy. -{d)ha. followed by the pronominal suf­ fix of the third person and by the name of the month. "the mountain"). or by a vocalic postposition which goes back to a suffixed pronominal element -ih)u functioning as demonstrative. "the house") for the masculine and the plural. One North Gurage dialect (Muher) uses the definite article -we regardless of the gender and of the f . which has been explained by the demonstrative so (Greek 6). "(T)his/The AMA-ra". Another related Semitic construction is attested in Amorite personal names ending in -Cu-ú (e.36). "month". etc. "house") can best be explained as an unchangeable petrified pronominal suffix of the -s type.g.g.g. "mother". The article is thus -u in Ge'ez (e. "the house").g. "the house". 33. "(T)his/ The star is the god".7) and it is paralleled in Indo-European by the suffixed -s (e.C.14. This must be a fossilized structure going back to a period when the South Arabian languages had not yet devel­ oped the system of determination and indetermination based respectively on nunation and mimation (§33. wrh-s d-tmn'). as Adam-ma-sù. The determination may be expressed also by a demonstrative. §36. In the light of wrh-h/s\ also the postpositive article -dš(a) in Gafat (e. used as demonstratives (cf. while wrh-s is used in Qatabanic (e. The related postpositive definite article -u (masc. which are paralleled in the first millennium B. bedu. in a parallel passage of the Ebla texts and the absence of the suffix -sù in similar sec­ tions are no sufficient reasons to postulate a new value for the sign AMA. gâġġ-aš(ā).

like 'il-m and '//-?.6-14. seem to confirm this explanation of the -m/n morpheme. tdnms betu. which preserve some archaic features. tdnndsu bet. "the man". However. 'zzm. The indeterminate or "absolute" state is that of a noun which is neither construct. Since nunation (-n ending) in Classical Arabic (e. "the children"). Most likely it was originally a masculine marker. nor determinate in the sense described in §33. "from beginning to end" in Assyro-Babylonian.g.15. the appellation "absolute state" is likewise used in Semitics to designate the citation form of the noun.6).16. etc. b'l-m and b'l-t. mdštawe. with the article and the North Arabian nunation. g àbbabìtwe.. not to the head (e. "his small house").g. and by divine and per­ sonal names. "a statue") denote the undefined state of the noun.272 MORPHOLOGY number of the noun (e. slmm.g. as well as the state of the undefined or indeterminate noun as opposed to the other states. átiti. This is confirmed by early Lihyānite forms like h-slmn. However. Some preserved pairs of divine names. mlk-m and mlk-t. like Mlkm. and marked in consequence. However. "the brothers"). w d) Indeterminate State 33. 'ttr-m and 'ttr-t. mzsszwe. -y after a vowel (e. which may be identical with the predicative (§32. there is no reason whatsoever to believe in the original deter­ minate or indeterminate values of the mimation and nunation (§33. bayocdi.g. "the sister". the postpositive article in a noun phrase is suffixed to the modifier. "the house". sāriqun. nor predicate. only later used with feminines which already had their own marker -t. it corresponds to the bare stem form of the noun. "the camel"). 33.g. gari. The two endings -m and -n are allophones of the same original morpheme which initially characterized the non-construct state of the noun without denoting determination or indetermination. "the small house"). with no additional affixes. while the possessive suffix is attached to the modified substantive (e. some scholars have assumed that mimation characterizes the indeterminate state of the noun also in East Semitic. Its earliest . In principle. "a thief") and mimation (-m ending) in ancient South Arabian (e. "this statue". with the mimation.g.g. as known to us best from various kinds of proper names. the postpositive definite article is -i after a conso­ nant (e. "the woman". but also from some frozen expressions like ul-tu re-eš a-di ki-id. gamelay. In other Gurage dialects.16). The -u and -wa articles are identical with the Amharic third person possessive suffixes "his" and "her" so that sometimes there is ambiguity.

a "petri­ fied" mimation and nunation seem to appear in adverbs ending in -am or -an (§47. Am. the -m morpheme is missing in the masculine plural.k u = nisqum. an enclitic particle -ma may exceptionally occur with nouns in the construct state. Otherwise. the language introduces a functional distinction between the mimation and the nunation. appears regularly in some modern Semitic languages which have intro­ duced a formal marking of the indeterminate state of the noun by means of a word functioning as indefinite article.2).g. "Servant of Dagan".Da-gan next to Ha-ab-du. In Amorite and in Ugaritic. še .17. which undoubtedly preserve archaic features.18. in the Old Akkadian language and in Early Assyro-Babylonian dialects which reg­ ularly use the case endings -um. However. Henceforth. An innovation characterizes the ancient South Semitic languages represented by epigraphical South Arabian documents. 33. and Arabic employ the -n ending for the same purposes. n i . Classical Arabic is adding -n also to the singular and to the feminine plural of the undefined state. Instead.g. the use of the -m ending is restricted in Ugaritic. e. d d 33. slmm.Da-gan. "a statue"). while the -n ending was used for the dual. slmn. a kind of servant. It is obtained by a semantic weakening of the number "one". "a . just as in Palaeosyrian and in Old Akkadian.g. Whereas mimation continues to be extensively used in proper names. instead.g. Its use is standardized. — contrary to East Semitic. a quality of figs). but it becomes later a free variant of the case endings. and Phoenician to the absolute state of the dual and of the masculine plural. and in Semitic loanwords borrowed by Sumerian with the case ending (e. and Amorite proper names. Old Canaanite. -am with determinate and indeter­ minate nouns. already encountered in Biblical Hebrew. Hebrew. In consequence.is . Another innovation. "the statue"). Old Akkadian. and its use is inconsistent in Palaeosyrian. Amorite Hab-du-ma. Ugaritic bnm 'il next to bn 'il. as in the Hebrew phrase 'īš 'ehād. while the nunation denotes the absolute or non-construct state of the determinate noun (e. Besides. the mimation charac­ terizes the absolute or non-construct state of the indeterminate noun (e. "son of El". Moabite. but it is no mimation. — while Ara­ maic.THE "STATES" OF THE NOUN 273 attestations in Palaeosyrian and in Old Akkadian already display the -m ending attached to the case vowels of the masculine singular and of the feminine.gu = šerkum.er . the -n ending became the mark of the determinate state and the -m ending the mark of the undefined or indeterminate state of the common noun. it can be omitted in Palaeo­ syrian.

hadā. Fem. šarrū šarrī/ē sarrātum šarrātim Dual Masc. In Egyptian and Syrian Arabic. at ârà.28) gâràd. is used for that purpose. Nom.g. Construct State Singular Masc. Acc. e) Paradigms 33. In Tigre.8). In Iraqi Arabic. šarrā šarrī šarratā šarratī šar šarri v sar šarrat sarrati sarrat šarrāt šarrāti ( šarrī/ē Determinate I indeterminate State Singular Masc. fern.(e. in Amharic. It results from this investigation that the states of the noun influence the case inflection in the languages which have preserved the distinction of cases. in Tigrinya. woro(t).1). the noun fard > fadd.g.g.274 MORPHOLOGY (certain) man" (I Sam. where the indefinite article can be expressed by at (< had). Fem. Acc.l-. hatte. "a boy". Fem. "a woman"). in Gurage dialects. and Old Assyrian nouns. fad ādami.19. It is combined with the article to wāhid dl. A similar practice can be observed in modern Ethiopian languages. hantit. Although the endings have various phonotactic vari­ ants.g. Fern. "a girl". Fern.g. and. "a man"). attd gàrdd or quna (§36. the following paradigm can be proposed for the triptotic inflection of Old Akkadian. instead. while the Algerian indefinite article ha-l. fern. e. hd-rragdl. andit. "hand" (§33. šarrū 1 Dual Masc. "a man") goes probably back to 'ahad.in Moroccan Ara­ bic (e. The same usage has been assumed for Mehri forms like hayd. masc. "some­ one". and it is reduced to fad > fa in the Arabic of Uzbekistan (e. "single". šarrān šarrīn šarratān šarratīn . is used alone for this purpose (e. masc. "one". šarrum šarrim šarram sarratum šarratim sarratam Plural Masc. masc. Old Babylonian. the numeral wāhid. "a man"). fern. is used in the same way. "one". wahda sitt. Gen. Nom. Fem. wāhd arrāġal. 1. A Plural Masc. Gen. the numeral "one".

"who is the one who stands?". milkūma milklma milkātu milkati Dual Masc. sariqi ^ sāriqātu . . allows a pared with the alphabetic spelling of nouns ending in reconstruction of a Ugaritic paradigm. Gen. K Plural Masc. Gen. 'a. Fern. sariqata . and the indeterminate state. Nom. milkā milki milkatā milkati Determinate 1 Indeterminate State Singular Masc. "thief". sāriqū 1 > J . Fern. Fern. Construct State Singular Masc. Nom. Fern. . Here. Gen. Fern. . The syllabic writing of Ugaritic proper names and nouns.20. Acc. . Classical Arabic distinguishes the construct state. Acc. sanqati ^ Dual Masc. we present the paradigm sāriq. the difference between the determi­ nate and the indeterminate states consists in adding the -n ending to the indeterminate singular and feminine plural. milku milki milka milkatu milkati milkata 1 \ Plural Masc. milkām milklm milkatām milkatīm 33. milkū milkī milkāt milkati Dual Masc. Fern. e.THE "STATES" OF THE NOUN 275 33. probably identical to a great extent with an Old Canaanite paradigm that could be based on the glosses of the Amarna correspondence: Construct State Singular Masc. the determi­ nate state. sāriqā sanqay i J sāriqatā sariqatay v J sāriqu sāriqi . sariqa sāriqatu sāriqati _ . Fern. milku milki milka milkat milkati milkat 1 J Plural Masc.21. Besides the use of the definite article 'al in the determinate state. On the other hand. Fern.g. Acc. . mani ('a)m-qā'imun. . com­ 7. the nunation is known in the determinate state of ancient Yemenite and Eastern dialects. . This distinction is not attested in Pre-Islamic and Pre-Classical Arabic. Fern. Nom.

as-sāriqāni 'as-sāriqayni Fern. The most apparent differences are encountered in the Old Aramaic dialects that have the "emphatic" or determinate state: Singular Masc. Fern. sāriqānì sāriqayni sārìqatāni sānqataym 33. Gen. as-sāríqūna 'as-sārìqlna Dual Fern. Hebrew. Masc. Fern. as-sārìqātu 'as-sāriqāti Masc. Plural Fern. as-sāriqatāni 'as-sāriqatayni Indeterminate State Singular Masc. Acc.22. 'as-sāriqu 'as-sāriqi 'as-sāriqa 'as-sāriqatu 'as-sārìqati \ 'as-sāriqata) Plural Masc. Masc. . Fern.276 MORPHOLOGY Determinate State Singular Masc. as Aramaic. Phoenician. sāriqun sāriqin . Acc. Neo-Arabic. sanqan sāriqatun sāriqatin 1 _ . sāriqūna sānqlna sāriqātun sāriqātin Dual Masc. Fern. Construct State malk > melek malkat mal(d)ké malkawāt mal(d)kē Emphatic State malkā(') malkdtā(') mal{à)kayyā(') mal{d)kātā(') mal(9)kayyā(') Absolute State malk > melek malkā(h) mal{3)kīn mal{p)kān mal(3Ì)kayin malkdtayìn . Fern. Nom. Nom. The different states of the noun are clearly apparent in the Semitic languages without any case distinction. Gen. . > sanqatan J Plural Masc. Dual Fern.

although it is not used in the dual. The construct state of the feminine sin­ gular has the ending -it more often than -at. "teacher". at least in the writing. Construct malk > melek malkat mal(a)kē State mal(d)kdt malkē malkatē Determinate ham-melek hammalkāih) ham-mdlākìm State ham-mdlākdt hammalkayim hammalkātayim Absolute melek malkāQí) mdlākīm State mslākot malkayim malkātayim 33. mi'allimlnak. . and that its ending -n is not dropped in the construct state of the modem colloquials. not even before a pronominal suffix. Singular Masc. The Hebrew paradigm can be considered as valid also for Phoenician with the exception of the feminine determinate and absolute states where the Phoenician noun preserves its original -t ending. Plural Fem. Dual Masc.THE "STATES" OF THE NOUN 277 33. We give here a paradigm based on the Cairene pronunciation of the noun mi'allim. "your teachers". Fem.23. In Neo-Arabic and in modem colloquials there are many dialec­ tal and phonotactic variants. Masc. e. One should add that the Maghrebine dialects employ the external masculine plural suffix -In only with adjectives and partici­ ples.g. Fem.24.

they indicate the quality of another noun in a specific and concrete situation.44). 34.1. participles. the adjective is more often in the singular (e. "pious men". which are grammatically singulars. Classical Arabic riġālun sālihūna. "the good". From the morphological point of view.g.2. Some nominal patterns are used more often to form adjectives (cf. And because they are referring to another noun. Construct State mi'allim mi'allimit mi'allimīn mi'allimāt Determinate State il-mi 'allim il-mi 'allima il-mì 'allimīn il-mi 'allimāt Absolute State mi'allim mi'allima mi'allimīn mi'allimāt F . Some Semitic nouns are plural in form though not plural in meaning. Adjectives 34. the adjectives.g. Singular Fern. "the people who walk".g. Hebrew 'êlohīm . but no strict rules can be established. Their concord is in general plural ad sensum. When adjectives are not used as substantives (e.41. As a rule. Fem. Hebrew hā'ām hahotekīm. thus e. §29.g. that can be either masculine or feminine. even with collective nouns and with Arabic broken or internal plurals. and case. In such cases. and verbal adjectives belong to the category of nouns. number.35-36. the adjectives agree with the substantive they modify in gender. the adjectives have a proper basic characteristic which is gender inflection expressed by formal grammatical means.278 MORPHOLOGY Masc. e. "God": Hebrew 'êlohīm. Phoenician 'ēlīm. Plural Masc. The main difference between adjectives and substantives is rather of a semantic kind. "the true").

tannas bet-un. the -n marker of the definite direct object (§32. "a small house".ADJECTIVES 279 hay. Aramaic šinnayin rabrdbān. 'akbar. These forms are no longer used in the colloquial . in Aramaic. "his small house". 34. fi'la. "good-looking boys". so we have. "the receivers of money". In several languages. tannas bet-u. Amharic adjectives only occasionally form plu­ rals. e.14) in an Amharic noun phrase is attached to the modifying adjective.16). Tigrinya does not distinguish gender in plural adjectives.28) follows the same rule: tannas-un bet. tannas bet. tannas-u bet. the mas­ culine plural of adjectives and participles ends in -ūtu{m).fa'la. in contrast to the -ūt of Assyro-Babylonian.g. Interestingly. -ūti(m) in Old Akkadian and in Assyro-Babylonian (§31. the case inflection of the adjectives is in gen­ eral the same as the declension of the substantives. plural adjectives agree usually with dual substantives. e. Besides. 34. although the homophonous possessive suffix -u (§36. 'ētān < *'aytān. On the opposite. In Tigrinya. 'èlohīm hayylm). not to the substantive. Classi­ cal Arabic has also feminine patterns for emphatic qualification. māhirū kaspim.4. bintun mulāhun.g.20) is attached to the substantive. 'uzma. "good-looking girls". e. 34. Traces of this formation are found in a few Hebrew adjectives. "how nice!".g. sdbbuqat 'awâddat. "big teeth". In other languages. and distinctive feminine forms are used optionally only in adjec­ tives derived from Ge'ez. except that inanimate plurals are treated as singulars. e. "greater". but its construct state ends in -ū in Old Akkadian. "a very pretty girl". and sdbbuqat 'awahd. the suffix can be added also to the adjective: tannaš-un bet-un.5. there are adjectives belonging to some nominal patterns of Arabic that are not capable of concord. "his small house". "deceitful". e. "racing she-camels". is formed in Arabic on the pattern 'af'al. "the small house" (direct object). "great". "living God") than in the plural (e. The comparative degree of adjectives. mdfu'la. called "elative". "how beautiful!". e. ni'ma.g. "the small house". in the latter case. However. viz. thus kablr.3. in Hebrew. like 'akzār. In consequence. "how mighty!". unlike Ge'ez. hasna. "his small house" (direct object). the definite article -u (§33. thus in Assyro-Babylonian. in modem Arabic colloquials.g. nūqun hiġānun. The concord of adjectives and substantives in modem Ethiopic frequently deviates from the common rule.g. "lasting". 'akzāb. however.g. "cruel". adjectives agree in gender and number with the substantives they modify.

"very large" (cf. 'awrab. "celebrated. in particular Bedja. 8 = 5+3. The comparative may be followed by min or. "greater". 34. loanwords from Arabic may be deceitful. but its exact meaning and function are unknown. Most numerals have a Proto-Semitic origin. A n East Semitic elative is represented by a small number of adjectives with a prefix ša-1 su\ e.2. The pattern 'f'l appears in many South Arabian proper names. "the greatest". šūturuim). §29. and distributives. The former. This limits our knowledge of the numerals. it has been borrowed into the West­ ern Neo-Aramaic of Ma'lūla. šanūdu. which serve to express "than". G. "widower". and they even exhibit close connections with Libyco-Berber and Egyptian. where it was operative. especially in North and East Semitic languages using syllabic and logographic script. Cushitic and Chadic stand apart. as well as with Egyptian and selected Libyco-Berber numerals. The standard cardinal forms of the numerals in the principal Semitic languages are given below with reconstructed Proto-Semitic forms. For instance. a) Cardinals 35. Numerals 35. e. "the most excellent man". 'al-'akbar. fractions. "great". Here. e.g. "bad". a noun probably derived from lemnu. lit. 9 = 5+4.1. multiplicatives or iteratives.6. "very violent". "taller than a palm-tree". Besides. famous". a remnant of a larger use of this pattern in Semitic languages may be preserved by Hebrew 'almān and Assyro-Babylonian almānu. Instead. 'afdalu raġulin. and meaning etymologically "worse". "highest of mountains". 'atwalu min nahlatin. 'a'lā l-ġibāli.280 MORPHOLOGY speech. by 'an. the . The same pattern 'af'al is used for the superlative and may then take the article or be defined by a genitive.g. 7 = 5+2. from rab. šalbabu / šalbubu.33). The numerals belong morphologically to the category of nouns. in the post-classi­ cal language. writing them in cipher instead of spelling them out. Beside the cardinals. Semitic languages possess derived forms or differ­ ent stems to express ordinals. "from". has cardinals based on the quinary Nilotic system and forms the numerals "six" to "nine" by composition: 6 = 5+1. The scribes have often represented the numerals logographically.g. e.g.

Various phonetic developments occur in modern Aramaic. while Amharic and (fern. and in South Ethiopic where Gurage had. someone". in Tigrinya (hadá). fern. and tād. "about. "one.3. "eleven". followed by the palatalisation add > aġġ. yahad. "one". a) The best known numeral "one" is had. but South Moroccan Tachelhit ya-n. dġġát). ištān). -ēn (ištērì). without pretending to go back to Proto-Berber numerals. some". § 33. which is used next to 'hd. but the same numeral is attested by Libyco-Berber iġ. b) The second root is 'išt-.NUMERALS 281 Ghadamsi numeral "one". wahada. woro-. ad.16). The forms with initial '.in Hebrew which results from a change i > a occasioned by ' in a closed stressed syllable. and Ethiopic colloquials. The feminine ištiāt and the Hebrew construct state 'aštē 'āśār. "by the unit (of measure)". attested in Arabic ('ahad). The numeral "one" is represented in Semitic by four different root morphemes. Arabic. as shown by other derivatives from had. "eleven". in Arabic and in other Semitic languages. Gafat aggā (fern. "one". "one". goes back to *(h)and. use a secondary root morpheme brought into line with the triconsonantal system. andit) and Tigrinya feminine hantit show an inserted n before the dental. The proposed etymologies of 'št are highly conjectural. Hebrew ('ehād I 'ahad). or -ān (ištiān-. way id. "one". and at(t) must derive from had. "alone" in AssyroBabylonian. follow the declension of the plural. and of Minaic 's t. "one"). "to be alone". wēdum. derives from the same form *wa'(-n) as ancient Egyptian w'(y-w) and must be considered as an authentic Libyco-Berber word. Tuareg forms are in general the most archaic ones. in Tigre (had. They cannot be reported here. Ugaritic ('ahd). and Ge'ez ('ahadu). hatte. but it shows an assimilation nd > dd. "to be united" and "gathering" in Hebrew. viz. had. 35. with an allophone 'ašt. but a formal singular is implied by the Aramaic expression b'št'. It is the only numeral "one" attested in East Semitic where it is used with a suffix -In (ištīrì). Epigraphic South Ara­ bian ('hd). l . the original form of which is preserved in Aramaic (had). also in Ugaritic (yhd). viz. South Arabian. is borrowed from Arabic. in some early Arabic vernaculars (had). wāhid. We do not know the vocalization of Ugaritic 'št 'šr(h). which is diffi­ cult to explain unless one considers -n as the masculine singular ending of the Afro-Asiatic pattern of agreement (cf. present also in the Libyco-Berber number "one": yiw-an /ya-n. feminine išt. 'išt-. A selective approach is thus necessary.

5 m. f. mraw mi 'tli'mribb- . kil'\ ślat\ rba '> hams\ sidtVšab'>tmān>tiš'> 'aśr- 0. f. f. šart kkuz kkuzt sdmmus. 6 m. f. \had-. iġ yiwdt. 8 m.-Sem. 'ahd 'aht tn(m) tt tlt(t) tlt 'rbXO 'rb' hmš(t) hmš tt(f) tt šb'(t) šb' tmn(t) tmn tš'(t) tš' 'šr(t) 'šr 'šrm tltm 'arb 'm hmšm ttm m'it 'alp rbt . ištēn ištiāt šinā šittā šalāšat šalāš erbet(t) erba hamšat hamiš šiššet šiš(š) sebet(t) sebe samānāt samānē tišīt tiše ešeret ešer ešrā šalāšā erbā hanšā < *hamšā šūš me 'at līm ešeret līm Ugar. 9 m. Bab. f. išt sin. sard kratt. f. 'išt\tin-. 3 m. afus sdmmust sdis sdist sa sat tarn tamt tza tzat mraw mrawt td-marwin *t9-ni9rwin d mraw *sin-id td-msrwin *sin-id td-mdrwin d *krad-id td-mdrwin td-mede a-zimlgim *Pr. f. 2 m.282 MORPHOLOGY Cardinal Numerals Egyptian 1 m. 20 30 40 50 60 100 1000 10. f. f.000 w'(y-w) w't śn-wiy) śn-tiy) hmt-w fd-w dl-w dl-t śrś-wfśîś-w śfh-w hmn-w pśd-w md-w dwt m'b3 hm dlyw ÌŚŚš(n)t h3 db' Libyco-Berber yiwdnfya-n. san snat. sdnt krad. f. 4 m. 7 m. 10 m.

s b't s b' x x l s tt st l l ydtēt hēt ydbayt hoba tmt tdmdnyīt tdmoni sāt sā 'āśdrēt 'ośdr \sds sddddst šib 'ā šdba' tdmanyā îdmānē tiš'ā tdša' 'aśrā sab 'atu sab'u. 'cśrlm šMošīm 'arbā 'īm hàmišším siššīm mc'ā 'elep ribbo 'aśar tmny. šcba' šjmonā šab'â sost 'arbā 'tu 'arbā'. sdddstu sdssu. 'ašartu 'ašru. šdls Tigre woro(t) hatte kdl'ot kdl'e f salas Amharic jand | hulātt šāldš 'arbā 'ā 'arba' hâmiššā húmēš sisšā šcs šib'ā."ninety" is replaced by Coll. tds 'atu tds'u. s dt.sjlošā 'ahad Aramaic Arabic had hâdā trên tartēn tdlātā tdlāt 'arbd 'ā 'arba' hamšā hâmēš šit šiš wāhid wāhida 'itnāni 'itnatāni talāta talāt 'arba 'a 'arba' hamsa hams sitta sitt sab 'a sab' tamāniya tamānin tis 'a tis' 'ašara 'ašr 'išrūn talātūn 'arba 'ūn hamsūn sittūn mi'at 'alf 'ašara 'ālāf Sabaic 'hd 'ht tny tty sHtt. Arabic numerals. . màto ši(h) Blf sdlsa 0) 'Blf Ẃ (') The series "twenty" . samāntu samāni. sdds hdms rdb' > 'arba' aratt > hamds ammdst s dtt. 'ahat šnayim stayim . 'dšrā šalāsā 'arbd 'ā hamsā sdssā md'dt 'ašartū md'dt 'dšr tds' samn sdb' >sabu' sàbatt tmn(y)t. hamdstu hams. tlt 'rb't 'rb' hms t hms l 1 x 2 Mehri tāt tayt tdrd tdrayt śātayt śhdlēt rdbot 'arba hdmmoh haymdh Geez 'ahadu 'ahatti kdl'e(tu) kdl'eti šalastu šalās.NUMERALS 283 Cardinal Numerals Hebrew 'chad. tlty 'rb'y hms y s ty m't 'lf 'śrt "If l l 2 2 2 2 1 l y samān sdmmdnî >Sd' zātànn 'âśar 'eśrīn tdlātln 'arbā 'īn hamšīn šittīn md 'ā 'âlap ribbo > 'asdr 'dsra salāsa 'arbd'a hdmsa sdssa md'dt šdh assdr haya sàlasa arba (h)amsa sddsa. tmn ts 't ts ' 's rt 's r 's ry s lty. šeimdnē tiš'ā tcša' 'âśārā 'cśer. tltt s lt.

The same numeral is attested in Egyptian (śn-w). etc. "second". "some".284 MORPHOLOGY c) Tigre woro.). the substantive śn means "brother" in Egyptian (śnt. tāt in Mehri. tnēn. tād in Harsūsi. An early change n > r. Śheri. "to cut off". The second . "to do (something) for the second time". this is just a guess. while there may be a link with the root *dad. clus­ ter". attested also in td 's r.28): asra qdmcâttá. 2 e) Another root is used by Egyptian w' and Libyco-Berber yiw-dn / ya-n. The numeral "two" is represented by two different root mor­ phemes.13) or through the ordinal *tāniy > tānī. d) The numeral "one" of Modern South Arabian. tin. 35. while the South Arabian numeral tdro shows an ending -6 which is related to the Qatabanic ending -w of tnw. "sister"). as well as in Cushitic (e.and kil'-. "something". itnēn). "ten-one".4.is apparently related to Tuareg mraw. with the dual form śnwy. But. a) The first one is employed in Old Akkadian and in Assyro-Baby­ lonian with the dual ending -ā (šinā.g. and with the nunation in modern Arabic colloquials (tnayn. as of dates.of tin. To express the unit in the numeral "eleven" Gafat makes use of a derivative of *qmt-. but one finds an ending -w in Qatabanic (tnw).6). goes back to Arabic say'.in Aramaic and in Modern South Arabian. This form is used in Aramaic with the dual ending of the oblique case followed by the nunation (tryn > trērì). to which the mimation is added in Ugaritic (tnm) and the nunation in Classical Arabic ('itnāni).either directly (§41. kilaltān > kilattān. "single" (cf. hence "ten". and Soqotri. so far. lit. "eleven". "ten". has an initial glottalized t and goes back to Qatabanic td. and in Libyco-Berber (sin. 'šnm).g. and originally designate a " b i t " or a "bunch. Its relation to had is improbable. "to repeat". while Berber indefinite ša. derives from tin. which parallels the situation in Gurage where a nongeminated n becomes r in non-initial position (§17. Bedja san) and in Chadic (e. used in the Eastern "Sam" languages belonging to the Lowland East Cushitic group. explains the byform tir. in the feminine. b) The other numeral "two" is attested in Assyro-Babylonian under the form kilallān I ūn and. The Semitic verb *tanāyu(m). The word might belong to the same root as Assyro-Babylonian wurrū or murrū. sena). viz. §36. The dual ending -ay appears with the mimation in Hebrew (šnayim) and in Phoenician (šnm. sdri). The dual oblique case -ay is used in Sabaic and Minaic (tny). East Chadic sin. "someone".

next to a feminine kl'ty. §17. is used as numeral "two". with the meaning "two" or "both".when followed by a pronominal suffix. The Classical Arabic kilā. "two villages"). go back to kdl'et-. sometimes called "gender polarity". but it has an oblique case kilay.5. Ara­ bic 'ahaduhum.g.6. "on one of the mountains". "not one among them has escaped". The numerals "three" to "ten" are abstract or collective substan­ tives marked by the suffix -t in historical times. "both". is replaced in the mid-period by kly. "pair". creates the false impression that cardinal numerals are used in the gen­ der opposite to that of the noun which usually follows in the genitive . e. 35. "one of them") or adjectives which agree with the noun they determine in gender (e. The numerals "one" and "two" are either substantives (e.g. case (e. kdl'e < *kil'ay is the normal numeral "two" for both genders. Aramaic.2). and Hebrew.I kiltay. the noun zawġ. where the labialized velar k < k is spiran­ tized into h and the / reduced to the vowel e.g. is unchangeable when it is followed by a noun. or kdVetu may stand for the masculine and kdl'eti for the feminine. Its basis is probably related to Cushitic *kal-. "a set of five" in Middle Egyptian. Arabic qaryatāni tnatāni. Also Amharic hulàtt and Gafat (h)dhttâ > dhé(êā). pronounced zūz. "one.g.NUMERALS 285 liquid / corresponds to the glottal stop of the other Semitic languages (cf.g. "identical words"). Hebrew 'al 'ahad he-hārīm. diw. "two". Duality is also represented by kpl. However. "double". "both". the bare root morpheme continued to be used with feminine nouns. Hebrew ddbārìm 'âhādīm. "of two kinds". w w w c) In Maghrebine Arabic. Arabic kulluhum li-'ummin wāhidatin. In Ge'ez. This development. 35. or zuz. "they are all from one mother"). the noun has the dual ending with the nunation and it means "both".t. in Hebrew and perhaps in Moabite (kVy: Mesha 23). in Ugaritic. and corresponding to ancient Egyptian collectives in -t. Assyro-Babylonian ištēn ina libbisunu ul ūsi. but it may be expressed likewise without the numeral by using the sole dual formation of the noun. alone" in the "Sam" languages. Only the feminine form kl'at is attested so far in Ugaritic and the masculine form kil'ayitn. It is in reality a dual *kil'ā which has lost its not pronounced ' like in ancient South Ara­ bian where the archaic form kl'y. This noun "two" is used likewise in the modem Ethiopian languages where the phonetic development can lead to forms like h et in Gurage. zūz. with the feminine kiltā. and the numeral "one" even in number (e.

as in the numeral sidt-.g. the "gender polarity" tends to disappear. A similar evolution is encountered in Neo-Aramaic. The Proto-Semitic root morphemes of the numerals "three" to "ten" can be established as follows: ślat-. Tigre uses the numerals "three" to "ten" without -t. šab'-. tit sswm. and tldt rġāl. genitive) with the sign SA used to express the syllable salśa. also in the ordinals and in the fractions (§35. and in Neo-Punic.2). "three horses" in Ugaritic. "three girls". This is confirmed by the Old Akkadian spelling sa-liiš-tim of the ordinal (fern.g. rba'-. The use of anaptyctic short vowels explains the vocal­ ization attested in the "classical" languages with the exception of the numeral "four" where a prosthetic vowel was added to the root mor­ pheme rba'. sidt-. instead of ŠA which would indicate ta (§13. This grammatical princi­ ple is rarely or irregularly observed in Late Babylonian. e. Classical Arabic talātatu riġālin. finally eliminating the distinction between masculine and feminine. in Damascus: talāte. only one form is employed. regardless of their gender. "six". In most modern languages. how­ ever. while the opposite tendency can be observed in the other Ethiopian languages where the numerals with the ending -t are more frequently used. e. this apparent inversion of gender seems to operate inconsistently in North Semitic.30).7.g. while most modern Arabic colloquials use the forms with the -t ending for independent numerals and those without -t in connection with plural nouns. while ancient Egyptian and Libyco-Berber add the mark -t to numerals qualifying feminine nouns. Nevertheless. one might also surmise that certain nouns. still used without prothesis in the suffixed form rdbot of the Mehri and Harsūsi numeral. er-bu u -mi. in the Mehri and Harsūsi numeral rība for counting "four" days. 'aśr. hamš-. *talātat > *talāta). Thus. "three". literally "a set of three men".286 MORPHOLOGY plural. are feminine instead of being masculine and that the lack of the ending / in Ugaritic results from its dropping in the spoken idiom (e. Chadic languages do not seem to distinguish the gender in numerals. contrary to the com­ mon practice in "classical" Semitic languages. A 35. e. but tldt banāt. tiš'-.8. The later form tit results from a regressive assimilation. especially in Ugaritic where two regressive assimilations took . for instance "day". In later Semitic languages. "during four days". "three men". However. in Syriac. The archaic and the modern South Arabian forms of the numeral "three" indicate that its original root morpheme had an initial lateral fricative ś (ślat-).g. 35. in the Babylonian dialect of Mari influenced by Amorite. tmān-.

This shift š > s is paralleled by 7 d w . 35. The Old Akkadian mentions of the dei­ fied "Seven" (planets ) imply the initial š since the sign si of Si-bi stands for ši I še (§13. and to Libyco-Berber sdmmus.had an allophone *śrat. e. Bedja fádig. and then a second regressive assimilation št > tt. asa-rama. "seven". with the exception of spellings influenced by North Semitic. which reflects a change h > s. muha vs. and Hausa fu'du. In the quinary system employed occasionally also in Libyco-Berber. originally "fist". is certainly related to Egyptian śrś I śiš-w. The numeral šab'-.10. "to take with the fingertips".11.9. In fact.2). "five". dative musē / muse. especially in Ugaritic where the form tt first implies the regressive assimilation dt > tt. "seven" (5+2). it must be related etymologically to Sabaic hms\ "main army force". etc. related directly to Egyptian hps. has an origin which is inde­ pendent from the corresponding numeral in the other Afro-Asiatic lan­ guage families. "three". The change in question resulted from a dissimilation caused by the originally lateral character of ś (§16. These forms testify to phonetic changes.12. as se-eb-i šanāti. as such. to Libyco-Berber sdis. 35. may be used instead of sdmmus.NUMERALS 287 place (§35. which can easily be related to *śrat.5). The numeral sidt-. "six". "fly". and to the Egyptian noun hps. "handful". but later Assyro-Babylonian texts have sebe or seba. "six" (5+1). "seven years". The numeral hams-. The Libyco-Berber numeral kkuz is related to Hebrew and Aramaic qums-. The / of the numeral "three" is labialized in Gafat s osta and elided in Amharic sost. is certainly related to Egyptian śfh-w and to Libyco-Berber sa. attested also in Semitic languages. signifies a hand (five fingers) and.11). used in asa-gwir.. The Semitic numeral rba'-. at Alalakh. and to Arabic qamaza. Its existence is confirmed by the Libyco-Berber numeral krad / sard. 35. "hand". The numeral *ślat. testifying to the change / > r in non initial position (§17. also to the Bedja numeral asa < *assa < *hassa < *hamsa. "four". The word originally signified the bending of the four fingers over the hollow of the hand. attested also in other Semitic languages. it is the basic unit of a quinary system.g. "strength" (m > p). "five". and the š appears also in the Old Assyrian orthography sa-be. A third Afro-Asiatic word meaning "four" is represented by Egyptian fdw. 35. the word afus. and to Hausa sidda. paralleled before front vowels in Slavic languages.1).

The Semitic numeral 'aśr. the passage / > s is not usual. and hoba. The numerals from "eleven" to "nineteen" are normally formed by the juxtaposition of the unit-numbers or digits and of the numeral "ten". "ten". "nine". South Ethiopic *ziht > *ziht. zīzātu). samānēšer. i. and from Arabic zahzaha. The original vocalization *tmān-t. zi'â). "clan". šdnē 'āśār. attested also in Gafat (zātâhnâ). and Harari (zdhtārì). but the phonetic differences are still in need of a consistent explanation of / : t : h. The numeral tiš'-. w 35. in Aramaic (e. Also in East Semitic. the numeral "eight" is written ša-ma-né in Old Assyrian.g. § 15. cannot be related to forms attested in other Afrasian language families. The numeral tmān-. 2 35. "share".g. "fifteen").21). with a noun zittu (plur. "eighteen"). They appear in four variant forms: a) Digits in the construct state. the same root as Old Akkadian zu'āzum and Assyro-Babylonian zāZu(m). are found in Assyro-Babylonian (e. "to divide". is related to Libyco-Berber tza. where the emphasis of z is secondary.10). is related to Egyptian hmn-w and to Libyco-Berber tarn. Attempts to relate this numeral to Egyptian pśd-w should be abandoned. 35. goes probably back to zht (zhz). 35. "to rip off". is the portion or the amount that remains after one part has been removed. "eight". 36. "portion". Argobba (zâh tâhh). but the later form is samānē. . "six". To judge also from the related Aramaic verb zūh. "to cut off". "nomad group" (cf. hamšat 'âśar.15. a spelling which probably implies the pronunciation /tamānē/.3). and in Hebrew (e.g.13. the forms 'sr and 'šrt are attested for the numeral "ten". in Mehri. "to dis­ place".seems to be reflected by old Amharic sant. "to go away" or "to remove". they reflect different dialectal shifts of the original ś phoneme (§ 16. In Phoenician. and Epigraphic South Ara­ bian 's rt. "seven". followed by "ten". yha't and yhob?' in Soqotri (cf.e. Its original meaning might be preserved by Arabic 'ašīra{turì).4.288 MORPHOLOGY another shift in Mehri and Soqotri where the numerals "six" and "seven" appear with an initial h or yh instead of the expected palatoalveolar š: net.14. §35. The final h of the Ethiopic numeral is a suffix which is missing in most Gurage dialects (zâtâ. num­ bering one digit below a full ten or two hands. Gurage (zâtāh). "twelve"). The Amharic numeral zdtdnn.16. In fact. "nine".

g. "eighteen"). 'arbā'ā 'āśār. asra s ostá. 'asartá šamante. "thirteen". and exception­ ally in Hebrew ('âśrā wa-hâmiššā. in Hebrew. āsrâhulâtt.g. Besides. in Tigrinya (e. "twelve"). 'sr w-hmš. "eleven"). 'asartà kdldttd. "eighteen".g.g. in Ugaritic. and in ancient South Ara­ bian.g. Syriac 'arba'sdrē. "twelve"). but only forms without -t are employed in Neo-Aramaic (e. 'šrh w-sb'h.g. arba'ta'sar. "fourteen") and only forms with -t occur in modern Arabic colloquials (e. in Gafat (e. "thirteen". Babylonian hamisserit. which is usual in Tigre (e. w d) The component "ten" preceding digits and joined to them by wa or -m. the ending -it or -ih > -ē is added to the numeral "ten" when the teens are used with a feminine noun. "fifteen". "thirteen") and in ancient dialects (e. although the lack of vocalization excludes any certitude and allows of an interpretation of the first type (a). 'ašru wa-salās. "seven­ teen"). âsrand.g. 'šr w-tlt. followed by "ten" with the same ending. in Aramaic. Mehri 'āterēt wd-śātayt.g. and can perhaps be assumed in ancient South Arabian (e. "thirteen". in Ge'ez (e. "thirteen"). "fourteen". 'šr 'arb'). e. Gurage asrd-m h et. regardless of the gender. "twelve". 'ohr wd-śhdlēt.g. in AssyroBabylonian.g. 'arb' 'šr. . also with inverted order (e. and in Amharic (e.NUMERALS 289 b) Digits with the fixed ending -a.g. "fourteen") and in Hebrew (e. appears in Phoenician and Punic (e. "seventeen". "thirteen"). Ge'ez 'ašartu wa-šalastu.g. but the practice of the Ugaritic scribes is not consistent. "fourteen"). and in Ge'ez. w 35.g. Ugaritic sb' 'srh. Ugaritic tmnt 'šrt. "sixteen"). "sixteen". tldtta'sar. Tigre 'asdr wa-sds. "fifteen"). in Nabataean (e. and in Hebrew. in Modern South Arabian. in Modern South Arabian (e. "fourteen"). In Aramaic. talātata 'ašara.g.g.g. "fourteen"). "fif­ teen"). s dtt 's r. in modern Ethiopic (e. "and". the digits of the numerals from "thirteen" to "nineteen" have the -t ending when used with masculine nouns. 'asdr 'arba'.g.g. "fourteen").17.g. are attested in Classical Arabic (e. asra httá. it is reflected in modern colloquials (e.g. arbāsar. "thirteen"). "eleven". this form is developed into a compound in Neo-Arabic (e. l 2 c) The mere asyndetic juxtaposition of digits and of "ten" occurs in Ugaritic (e.g. 'ihdā 'ašrata / 'aširata I 'ašarata. both components of the numeral have the -t ending when used with a masculine noun (e. arba'tāšar. Hebrew šdloš 'eśrē. In Ugaritic. "thirteen"). "thirteen"). "twelve". in Arabic.

hammdst assdr. sddddst assar. erbē . from fifty upwards. etc. "three thousand". in South Arabian ('s ry). thus in Hebrew. Aramaic. and some Gurage dialects (Soddo and Gogot). in modern Ethiopian languages. One may ask therefore whether Cushitic has not rather borrowed this use from the Semitic languages of Ethiopia or assume that this manner of expressing the tens is common to both language families. In Argobba. where it is generally replaced by Arabic 'ašrin. "thirty". so much the more that a similar system is attested in Libyco-Berber with the numeral "twenty" as basis. except in Late Babylonian where gender concord is attested with masculine and femi­ nine plural endings (e. thus sost assdr. Otherwise. also sdmmus-id 'ašrin. Neo-Ara­ maic tlāy. while šūš. However. French "quatre-vingt"). However. "twenty" is expressed by the plural of "ten" and the following tens are formed analogically by adding the plural ending to the numerals from "three" to "nine". The odd tens are formed by addition: sin-id 'ašrin d 1 2 1 . formed as duals of the numerals from "three" to "nine". krad-id 'ašrin. Hebrew šdlošim. hanšā < *hamšā. Tachelhit mrawin). "fifty". In other Semitic languages. Soqotri śìle 'eśarhin. another system probably existed in Semitic languages. "twenty"). literally "five tens".g. Phoenician. e. "fifty" in Old Akkadian. "three hundred". "three times twenty" = 60. "five times twenty" = 100. amsa in Amharic.18. thus e. tlātīn in Damascene colloquial. This formation parallels the situation in ancient Egyptian where the tens. "thirty". ešrāt. "sixty" is lētonnēte. "twice twenty" = 40. §35. Modern South Arabian has pre­ served some numerals of a similar old series (e. in Classical Arabic). Decade numerals have no gender differentiation. "forty". talātu mi'atin.290 MORPHOLOGY 35. "four times twenty" = 80 (cf. proba­ bly in Libyco-Berber (Tuareg td-mdrwin. "sixty". "six tens".g. and the same system is used in other Semitic languages for hundreds and thousands (e. "sixty".g. and Arabic. kkuz-id 'ašrin. talātatu 'alāfìn. The dual afformative -ā spread analogically from "twenty" to the following tens. remains unchanged. Classical Arabic talātūna. as suggested by the situation in South Ethiopic. arbdt assdr. literally "three tens"). and perhaps in Ugaritic ('šrm).g. This manner of expressing the tens is generally believed to have been taken from Cushitic. Harari. the numeral "twenty" is expressed by the dual of "ten" in Assyro-Babylonian (ešrā).g.31). "forty". are plurals of the units. since "fifty" is ontētonnēte in Sidamo. literally "ten hundreds" ). Aramaic tdlātīn. Ge'ez forms the numeral "thousand" in this way ('ašartu md'dt. the tens can be formed by compounding a unit with the numeral "ten". in Ge'ez ('dšrā). "thirty". unfortunately in its borrowed Arabic form 'ašrîn: sin-id 'ašrin (cf.

is employed when the digits follow (e. šlš hmšm. where it is combined with the decimal system. but an additive / precedes the "ten" when the digit stands first (e. In South Ethiopic. w l 35. and tza (§35.g. md'dt in Ge'ez. Old Akkadian (mi-ai). The Semitic dual form for "20" seems to point to a former vigesimal system which is thus pre­ served in Libyco-Berber. but to Semitic m't. with an emphatic d replacing the voiceless t.g. māto in Amharic. and in modern Arabic colloquials (e. "twenty-eighth nights"). 'esrī trē.g. "fifty-two"). literally "three tens"). 'šrm 'rb'. sdis (§35.20. but the conjunction w. . It is written m't in Phoenician and in ancient South Arabian. 1+20. "twenty-one"). the various systems can be presented as follows: 20+1. indirectly in Amorite with the orthography me-et at Mari.or -m (e. like in modern Ethiopic using the conjunc­ tions wa. The Assyro-Babylonian use is unknown. and mā in Syriac and in Neo-Aramaic.g. ma-a-at). "twenty-one" in Gurage [Chaha]).g. 1-/-20.8).g.g.joins the numerals also when the digits precede (e. AssyroBabylonian (me-at. "thirty". me-at). literally "four above twenty"). "twenty four"). mi 'at in Classical Arabic and mlye in the colloquial of Damascus. "twenty-two"). hms w-s ty.g. It has a plural timad. there is another term for "hundred". in Classical Ara­ bic (e. while the plural is m'at. 'rb' l-'šrm. It is vocalized mē(t) in Late Babylonian. Several systems are followed to join tens with the digits. which explains the Ugaritic singular m'it (*me'et or *mēt). "sixty-five"). mē'ā in Hebrew. "hundred". "twice twenty with ten" = 50. 35. "twenty-seven"). sab'ā w-'dšrīn. md'ā in Aramaic.NUMERALS 291 mraw. the situation is opposite: no conjunction is needed to join tens with the digits that precede (e. 20-Wm-l. but the conjunction wa. Schematically.g. besides. a change par­ alleled in krad (§ 35.14). "fiftythree"). In Phoenician and Punic. "fifty-four" in Tigre. The numeral "hundred" is derived from a common origin. The latter scheme is the normal way of expressing the compound numerals in ancient South Arabian (e. The latter appears also in Indo-European (e.g. This is also the most common construction in Hebrew. hdmsa wa-'arba'.g. but two different systems are attested in Ugaritic: no conjunction is used to join tens with the digits that follow (e. exactly as in Neo-Aramaic (e. hu ya-mat. is related not to Egyptian md-w.11). l-w-20.19. tamānin wa-'išrūna laylatan. 'ehadwd-'eśrīm. "ten". hmšm w~šnm. It is attested in Palaeosyrian (mi-at. The numeral is attested also in Tuareg where td-mede. etc. Latin trī-gintā.

and in Tigre. k dm and other forms in Gurage. literally "ten hundreds" (cf. In Ugaritic. Arabic mi'āt and mi'ūn. in Old Akkadian. while the use of 'alf with the meaning "thousand" in Tigre and in Harari is due to Arabic influence. while "one thousand" is expressed by 'ašartu md'dt. In Palaeosyrian. used also in Cushitic (ših). Tigre 'am'āt). ? u b) Ordinals 35. Aramaic mā'tayirì) and in the plural (e. Tachelhit alim > Tamazight azim. as just mentioned. The numeral "ten thousand" has a special name in Palaeosyrian (n-ba ). They generally derive from the cardinals by adding a suffix to the root morpheme or by adopting the CāCiC pattern. The numeral may be related to the Ethiopic name of the "mule". It may be related to Tuareg a-gim. with variant forms in other Cushitic languages. bâqdr or bàqdl in Gurage (l/r).22.18).292 MORPHOLOGY viz.23. §35. Mod­ ern Ethiopic sdh / ši(h). one also finds a noun "thousand" borrowed from Cushitic: hum in Harari.g. There are five different ways of expressing the numeral "thou­ sand". "people. in Ugaritic (rbt). with an internal plural 'a'lāf I 'ālāf. . but the South Ethiopic baqdr may suggest "ox". since Libyco-Berber z often derives from / (e. in Ge'ez baql (cf. in Amharic. i-zīm-ān). bāqlā in Harari and in Gafat. It suggests the idea of "magnitude" and can be used in the dual and the plural. and Hebrew (ribbo).21). in the West Semitic languages. clan". 35. 'dlf is used for "ten thousand". In South Ethiopic. number. except in East and North Semitic where they precede it. Aramaic (rebbo). and in South Arabian. which is related to Ugaritic Vim and Hebrew Id'dm.21.g. The numeral "hundred" is used also in the dual (e. "straw"). "thousand". might belong here too. and case. In Ge'ez. as well as to the divine name Li'im. Tuareg a-zim. suggests the notion of high number (root šyh).which is related to the noun meaning either "clan" or "ox". In Ge'ez. and baqol in Somali. instead. "thousand" (plur. w 35. The ordinals are adjectives and they follow the rules of the gen­ der. 'dlf has the meaning "ten thousand". and in Assyro-Babylonian one finds li-im. §35. one encounters the noun 'alp. The numeral "thousand" can be used in the dual and the plural. Since the ordinals are adjectives. they normally follow the substantive. The analogy with li-im is in favour of "clan". but there is a dialectal variant a-gim that may suggest a link with Cushitic and South Ethiopic hum.g.

etc. "eighth"). both meaning literally "the one in front" or "former". in Tigre (e. The ancient Aramaic ordinal tinyān. šādištum. sālištum. the only known terms for the ordinal "first" derive from root morphemes different from the ones used for the cardinal "one". 35.g.25. "first".26-27). to South Arabian qdm. result from the loss of the short i and from the subsequent shortening of ā to a > e in a close syllable (e. This pattern can be assumed also in Ugaritic.). kāh'. "eighth"). In Assyro-Babylonian.). With the exception of the rare Old Babylonian ištiyūm and of the South Ethiopic atānā. probably by allusion to Ex. "to begin".13. Ge'ez (sādas). with Mehri sodas. fern. and probably in ancient South Arabian judging from Arabic and Ge'ez. where the suf­ fix -y is likewise missing. followed by Modern South Arabian (e.g. as well as to Arabic 'awwal or 'awwil. *mahrlu(m) > mahru(m) and pānīuim) > pānū(m) are attested.g. "eighth").. does not reflect the change n > r of the cardi­ nal. and South Arabian.g. "third" fern. l . Hebrew rVšon derives instead from the word "head". 13. and kā'db. Arabic (sādis). The same basic meaning is attached to Aramaic qadmāy. *šādišu > *šaššu > šeššu).g. the more so if the later Assyro-Babylonian forms saisu. and from the identity of the ordinal s dt. while Tigrinya fálāmay etymologically means "redeemed".24. The form of the ordinal "first" in Old Akkadian. and Ugaritic cannot as yet be ascertained. "sixth" fern. while the ordinal tdt parallels the ordinal "sixth" in Old Assyrian (šādiš). Palaeosyrian.NUMERALS 293 35. tāmin. "third". also in Ge'ez (e. from the absence of the -y suf­ fix. sāhs. The ordinal "second" is formed in general according to the same patterns as the ordinals 3-10 (§35.26. The CāCiC pattern is used for the ordinals up to "tenth" in Old Akkadian (e. "second". 35. in Arabic (e. toman. viz. andànna. and to Ge'ez qadāmi. dāgam. The pattern CāCiC can therefore be regarded as ProtoSemitic. šadtēt. in Old Assyrian (e.g.g. Gafat màzámmàryà. "sixth". originates from the Ethiopic root ġmr I zmr. "third"). The alternative explanation for these Assyro-Babylonian forms would be a different pattern CaCC or CiCC. which has been borrowed as hāwīl in Modern South Arabian and is suffixed into awwalānl in Syrian and Egyptian colloquials. but the Neo-Aramaic form is treyāna. Ge'ez uses for "second" three different root morphemes. Amorite. while Modern South Arabian employs mašēġar as ordinal "second". "first". without any suffix. sāman.

'dšrāwi. md'dtāy. and Aramaic: the cardinal numerals are used beyond "tenth".g. In Arabic.g. In Old Babylonian. ešrū. hmšy.30. "fifth". "third". "fifth". etc. "a fourth".31). sālsāy. "one hundred and first".294 MORPHOLOGY 35. md'dtāy wa-qadāmāy.27. "twentieth"). "thirtieth". e.g. sā/sēbi'atum. "second".. in Hebrew (e.g.g. šalāsāwi. from tre. "second". "four­ teenth". c) Fractionals 35.g. hamsāwi. 'asdr warāb'āy.52) prefixed to the corre­ sponding cardinal numeral. "twentieth". sālsāyt.28. we encounter šālistum. Above "tenth".g. "fifth"). e. from tla. ddtla. Tigre masc. "third". in Phoenician and Punic (e. 'dsrāy. "ten thousandth". e. "eleventh". šalāšiyu.g. sostânâ. "two". tišā'iyu. 'asdr wa-qadāmāy. "twentieth". §35. "eleventh".26). From "twentieth" onwards.29. we have hādiya 'ašara. the ordinal is followed by 'dd. tdlītāy. hâmīšī. -I in the genitive (e. The best proof that fractionals were initially derived from fem­ inine ordinals can be deduced from the fact that CāCiC is the normal formation for fractionals as well as for ordinals in the early period of East Semitic and in Ge'ez (§35. the cardinal numeral "ten" is used with the ordinal of the digits. hamsāy. hamāšiyu. but endings of the ordinals may be applied in Ethiopic. "ninth"). "fourth". masc.g. the ending -āy may be applied. e.g. "fourth"). "a third". "a quarter". In Tigre. and -and in South Ethiopic (e. etc. no ordinals are attested in Ugaritic and in ancient South Arabian. rdbī'āy. and no special forms exist in Hebrew. -āy is used in Tigre (e. 'asdr Šdhāy. "third"). In Ethiopic. and fur­ ther the ordinal followed by 'ašara. "hun­ dredth".g. tdšī'ī. In Ge'ez. literally "fourth part" (cf. "fourteenth". In Neo-Aramaic. In Ge'ez. erbēsērī. 'rb'y. in Aramaic (e. "twelfth". "fifth". "third". fem. e. likewise. Patterns with a suffix are attested in Middle Assyrian (e.relative particle dd (§36. "three". and in Ethiopic: there are parallel series with the endings -āwi and -āy in Ge'ez (e. either the cardinals or forms with the suffix -āwi can be used. rābītum. h etànâ. w 35. etc. sādsāy. ordinal numerals can also be formed by means of the determinative .g.g. "third"). 'asar wa-kāl'āy. Phoenician. ddtre. sādsāwì. A related stem is . "hand". rābd'dt 'dd. "ninth").g. The ordinals above "tenth" occur in Assyro-Babylonian with a suffix -ū in the nominative. e. "sixth"). "a seventh". 35. the cardinal numerals are used.

"third". "one eighth". "four times". in Arabic (e. There are also forms which can­ not readily be attached to common patterns. in Sabaic. "fifths". or sar in Tigre. other patterns are attested as well and the stem CuCC is widely used in West Semitic languages. meaning "half".g. mrb't.g.g. roba'. hst in Ugaritic and in Hebrew. "a third"). An innovative Libyco-Berber use of the morpheme id with­ out preceding numeral aims at marking the plural of not-Berberized loanwords borrowed in the singular. etc. id hali (< Arabic hall. "for the third time". "second". while the Phoenician and the South Arabian vocaliza­ tions are unknown. Ugaritic šb"id or šb'd. adi I ana hamšišu. The meaning of multiplicatives or iteratives is not only "once". This suffix -išu goes back to -* 'itu and corresponds to the South Arabian -'d and to the Ugaritic -0)d. "a third"). From the Middle Assyrian and the Middle Babylonian periods on.g.31.NUMERALS 295 found also in Hebrew (e. "third part". and idiomatic ways of expressing frac­ tions. "a fourth"). kkuz-id. viz. often with the preposition adi or ana. Sabaic s dt'd.g. e. in Hebrew (e. as shown by ištiššu. in Tigre.g. etc. "thrice". as well as to the plural fraction mhms t. "for the second time" or "in . but also "for the second time".e. or šinepiātum > šinepātu(m) in Old Babylonian and snpt in Ugaritic. labá-ad.g. tult. These numerals are formed in East Semitic by adding -išu to the stem of the cardinal numerals. "twice". special words like mislu(m) in Assyro-Babylonian. A morpheme -ad appears in Somali where it forms the ordinal numerals. e. "a fifth"). "a fourth". krad-id. mtltt.g.is dropped after ištīn. "three times". "my maternal uncle"). hums-. at least among the Iġšan (Tachelhit): sin-id. "for the sixth time". e. However. "seven times". The vowel -/. "once". used to express the iterative. "multiple maternal uncle(s)".and suffixed -t. etc. 'sb'm bntmny 'sb' in Sabaic.g. saddehá-ad. 1 35. e. another method is attested for signifying that something happens "for the first time" or "in the first place". In Ugaritic. The same formation is preserved in some Libyco-Berber dialects. also in Late Babylonian (hunt. "five times" or "for the fifth time". and to masallas. e. etc. "a fifth"). "a third".32. kobá-ad. the attested fractions have prefixed m. 1 d) Multiplicatives 35. "first". e.. This pattern is related to the feminine ordinal "second" in Mehri and Harsūsi (mzšdġdrēf). "one finger from eight fingers". i.g. "two-thirds". in Aramaic (e. "twice". salīšīt.

there are also two patterns used to express the distributives. the ordinal with the suffix -ānu. "three by three". however. "time". talāta talāta. tulāt.33. fourth time". "four by four". "he adjured you". In Arabic. "two". Hebrew hišbī'ekā. or the numeral had. "two by two". "firstly.g. "make x-fold". matlat. etc. secondly. "two by two".). rābi'an.g. rabiānu. derives from tin-. salsiānu. like other derivatives of the same type. 41. the expressed or understood noun pa'am. viz. etc.34. but this pattern has not been identified as yet in other Semitic languages which usually express the distributive numerals by a repetition of the cardinals. This formation is attested also in Arabic with the old ordinals of the pattern CāCiC: 'awwalan. pa'am hâmīšīt. šinā'. had sib'ā. Because of its use in spells and conjurations. third. used in Aramaic with a following cardinal (e.13). e. e) Distributives 35. thirdly. "three by three"). f) Verbal Derivatives 35. Palaeosyrian si-ba jsibba'l K I . means "to sit cross-legged". e.g. tunā'. e. "do for the x time". tālitan. fu'āl (e. "td repeat". gave rise to differ­ ent denominative formations. "three by three") and maf'al (e. The root morphemes of numerals are used in Semitic languages as base of verbal derivatives that sometimes have a particular meaning going beyond the basic acceptations "divide into x parts". "seven times"). Aramaic mšby' 'ny 'lykm.g.g.4. "adjure the lands!". like the Hebrew and the Jewish Aramaic causative stem. There are also forms specific to one language or idiomatic ways of expressing iteratives. in Hebrew and in Phoenician with the cardinal (e. "one by one". "to quadruple". the verb *tanāyu. . K I . Although Arabic Stem I I of sb' simply means "to make sevenfold".g. E.296 MORPHOLOGY the second place". derived from Stem I I rabba'a. šaniānu. "the fifth time"). rubu'ā'. "three by three". e. The distributive numerals have the characteristic formation CuCuCā' in East Semitic (ištinā'. "seven". the corresponding Palaeosyrian D-stem expresses the idea of adjuring. "two by two". Arabic Stem V tarabba'a.g. viz.g. and the phrase tarabba'a 'alā l-'arsi has the specific meaning "he mounted the throne". "once") or the ordinal (e. the numeral sab'-. "one". to which a morpheme -y was added (§35..g. Instead. matnā. " I am adjuring you". fourthly".g. sulusā'. "for the second. pa'am 'ahat. tāniyan.

g. The Chadic branch. Such constructions do not occur with Semitic per­ sonal pronouns. "they have hunted" (perfective). and suffix-pronouns are found only in Egyptian and in South Ethiopic. inni. where the first set is used as copula (§49. "his").g. The personal pronouns are subdivided into independent or separate pronouns and suffixed pronouns which can be used with nouns. su-na halbi. plur. so that it is the pronoun which seems to be inflected. as predicate. isā.g. su-ka halbi. §11. and the indefinite and interrogative pronouns. although full sets of independent pronouns. "they will hunt" (future).g.g. "he swore". has the greatest number of peculiarities. both morphological and morpho-syntactical.7) of the cor­ responding pronoun in the other Afro-Asiatic branches. This leaves us in Oromo with three sets of personal pronouns that correspond to an active or subject case (e. . and a genitive or pos­ sessive (e. e. but the fact is that it is identical with the West African pronoun of the Mandingospeaking Vai tribe (e. One set is used as isolated citation form. "him"). In Cushitic languages like Oromo. and prepositions. the Hausa pronoun mu of the 1st pers. mu-ro. Semitic languages have five types of pronouns besides the per­ sonal preformatives and afformatives of the verb inflection which will be discussed in connection with the verb (§40).g.1. For example. "they were hunting" (imperfective). 3. The system of the pronouns of the personal group can be said to be genetically identical in all five branches of Afro-Asiatic. the independent possessive pronouns. the determinative-relative pronouns. The per­ sonals are expressed in Hausa conjugation by separate pronouns that precede the verb and are fused with morphemes indicating aspect and tense. e. while the second set serves as possesssive pronoun fol­ lowing the noun. These five types are the personal pronouns. dependent pronouns. s(u)-ā halbi. mu-fa. pronominal suffixes are replaced by two sets of separate pronouns. kan or isā. verbs. "our father". "they hunted" (preterite). and mainly as direct object pre­ ceding the verb. a non-active or non-subject case (e.PRONOUNS 297 the reflexive N-stem is used in Hebrew with the meaning "to swear". nišba'. "we say"). may be considered as the assimilated element nu > mu (cf. and may form reflexive pro­ nouns. as usual.19). the demonstra­ tive pronouns. PRONOUNS 36. su halbi. "he").

and for Tuareg. Dual 1 1 šuwa šiya nt-a 'an-kā 'an-tjk{an)ā š{u-rì)ā *attunā *šunā 2 3 Plur. Independent Personal Pronouns 36. *'antīn. 1 m. 'an 'ami. In-k nt-k nt-t nt-f nt-s n-sk kay kdm Rendille *Pr.2. The following paradigm of the independent personal pronoun. for Rendille. 'attā 'an hū '(a) hī'(a) 'ānokī 'ânā 'anta. f. 'attemimā) 'attēn(ā) hèm(mā) hēn(nā) 'ānu 'ânahna(n) 'antūn/m. P. a Lowland East Cushitic Language spoken in Kenya. f. although the existing pronouns Egyptian Tuareg Sing. 3m.-Sem. a paradigm is added for Egyptian. Ugaritic Hebrew Aramaic an(i) 'an-a 'an-talka^ 'an-ti/ki šu-wa ši-ya a) 'anna 'anta ? anāku atta atti ŠŪ ŠĪ 'an. 'ank 'at 'at hw hy 'ânī. Palaeosyrian is based on the Ebla texts. 'inriin . 1 2 m. in-n n-zkkà-ni n-akkâ-wti kâw-ni kàmâ-ti dntâ-ni inno nih-nu 'an-ta-nu 'an-ti-na šu-nu ši-na ? 'antanu ? sunu ? nlnu attunu attina šunu šìna hm hn ('ā)nahnū.Bab. hū(') hī(') 'an j am us(u) ice 3 m. nt-tn atin nt-sn 3ntâ-n3ti ièo himmd(n). is limited to the principal Semitic languages and to *Proto-Semitic. 'attūn 'attln 'innūn 2 m. hinnīn. For compari­ son. O. subject case. which generally preserves archaic Libyco-Berber forms.Syr. free form.298 MORPHOLOGY A.

intu 'antunna. fem. attested with the 2nd pers. like in Egyptian. nassatna n3ssdk(atk)um nzss3k{atk)dn n3ss(at)om nsss(at)an j dhha . intu hum(ma) hunna. hin 'ntmw snha 'dtēm 'dtēn ndhna 'antdmmu 'antsn wa 'dtomu. 'dmānîu hdna 'dntum 'antsn hdtom hdtan ndhna. dnnàrsu . inti huwa. However. It should be reminded that Old Egyptian /k/ is palatalized into [c] = " t " before the front vowel / i / (nt-k. used for the 1st and 3rd pers. hū hiya. nt-t < *nt-ki. hi 'n '(n)t 't h(w)' hy' ho{h) hēt hēt ha(h) sē(h) 'ana 'anta 'anti wd'dtu yz'dti 'ana 'dnta 'dnti hdtu hdta 'ana ndssdka ndssdki ndssu mssa {'anta) sne anta anii 3ssu.). masc. drsu 3ss a. 'dmuntu ys'dton.PERSONAL PRONOUNS 299 seem to belong to a mixed paradigm. inta 'anti. 3rs a w w dkdy 'antumā humā dîdy hsy nafynu. the element (d)nt. snnanta hmw hn hēm sēn ' snnāssu. nt-tn < *nt-kin.is likely to represent an expanded form of and both elements form the basis of South Ethiopic copulae (§49. the other without that ele­ ment. resulting from the contamination of two older paradigms: one with the demonstrative element (d)nt-.dllantā. Arabic Sabaic Mehri Ge'ez Tigre Tigrinya Amharic 'una.. (ni)hna 'antum. plur. ana 'anta..19-20).

Sabaic u Qata.11). from the Old Akkadian period down­ wards.. " I " . 1 2 m. Palaeosyrian. f. while Phoenician (hmt). while Gafat has even a first per­ son singular pronoun anāt(ti). Assyrian Old T> i i • Babylonian T T . 3 m. "me". " I " . The independent personal pronoun of Proto-Semitic most likely possessed at least one non-subject case. tana (nonsubject). In fact.300 MORPHOLOGY 36. in Ugaritic. banic kuwāti šuwāti yāti ku(w)āti šu(w)āti ' šiāti yāti kāti/a šuāti/u. PalaeoSyrian Sing. Old . Oromo ani (subject).1). its trace can be found in the personal pronouns of the third person singular and plural with an element t (§36. "me". " I " . 1 2 3 m. Dual 2 3 Plur.3. Old Akkadian. and some Arabic vernaculars (humā denoting the plural and not the dual) probably preserve a trace of the oblique case in the third person plural. šā/ēti hwt hyt hwt hyt s wt s yt l l *kunīti šunīti *kuntti šunīti hmyt s myt l ni{y)āti niāti kunūti kināti šunūti šināti nīāti. Ugaritic „ . . Walamo tani (subject). The fol­ lowing paradigm gives the genitive/accusative form. but also in Palaeosyrian. A n oblique case of the independent personal pronoun is attested also in Cushitic languages (§36. an oblique case is attested not only in East Semitic. and in South Arabian. šāti/u šuāti.. ana (non-subject). Hebrew (hemmā). and Old Babylonian distin­ guish two oblique cases: the genitive/accusative and the dative. In the Ethiopian languages. nēti kunūti šunūti šināti hmt hmt hnt s mt l šināt(i) . to which Soddo àdi. is probably related (< *ati).g. e. Fairly complete paradigms can be established only for East Semitic. f. . f.

a. 'ânā do raise the question whether 'ânā wasn't once an oblique case. There can be no doubt about the original nature of n since it is present not only in . Dative forms are attested in Palaeosyrian. In Semitic languages. "seal!") and in North Semitic (e.or dn. anân. 1 2 3 Plur.g.in all persons of the Egyptian pronoun. and Old Babylonian. i-. "he". at Ebla (an-tdnu).is followed by morphemes indicating the first person (-a). The first and second persons singular and the second person plural of the subject case have a common element 'an-. When compared with Cushitic (§36. In most West Semitic and South Semitic languages. Sadun-laba > Sadum-laba. but the original vowel did not disappear com­ pletely: it is still present in the Neo-Assyrian plural form attanū-ni. and the plural masculine (-nu) and feminine (-no).or n.3). "we".in the first and third persons of the Tuareg pronoun. The only distinct feminine form so far encountered is the third person singular šiāšim which occurs in Old Babylonian poetry. as n. as suggested also by the South Ethiopic copula nwhich must go back to the same pronominal element (§49. d-. " I " . 'an. An alternative explanation is suggested by the colloquial use of 'ana for the masculine and of 'ani for the feminine in the Djebel ed-Drūz (Syria) and in Yemen. 1 2 3 Old Akkadian Old Babylonian yāši(m) kāši(m) šuāšìm.g. the contrasting West Semitic forms 'ânī vs. The second vowel a was subsequently dissimilated into u or u > i in all Semitic languages. Middle Assyrian kunkā > kumkā.5. e. dnt. Qwara (Agaw) an. The initial 'a-. and in several persons of the Cushitic pronouns. the n of the masculine pronoun changed into m. Palaeosyrian Sing. šāši(m) niāšim kunūši(m) šunūši(m) kuwāši šuwāši ni(y)āsi kanūšì šanūši kuāšim šuāšim 36.PERSONAL PRONOUNS 301 36. possibly under the influence of the preceding vowel u. ni. an Amorite personal name). etc. The original form of the sec­ ond person masculine plural is attested in Palaeosyrian. "you".20). the use of which was generalized. Old Akkadian.4.g. which appears as in. the second person masculine {-ta) and feminine (-ti). a phenomenon attested sporadically also in East Semitic (e.seem to originate from a prosthetic vowel.

in the Oromo demonstratives (kuni I tuni.6. 36. this ending is added to the plural stems.for both genders (but cf. 36. one might assume that once upon a time there had been an opposition of masculine *an-ka vs.). 22. "this".g. oka (masc. as exemplified e. The alternation -ka I -ki vs. these remote South Arabian idioms may represent the original situation of the dual.5). and in Aramaic. 24). -tny < *-kny. There is a probable relation between this suffix and the pronominal suffix of the second person singular (-ka / -ki I -ku) and plural (-kun / -kin) (§36.17). in East Semitic.19.).is the basis of the second person pronoun not only in Tuareg and in ancient Egyptian (§36. akāk in Harari. Old Canaanite (a-nu-ki). The additional suffix -ku I -ki of the first person singular is attested in Old Akkadian (a-na-ku^ a-na-ku-ú).4). exactly as in Old Egyptian where the suffixed dual pronouns are -ny.or -t. §12. feminine t. etc.2). These forms seem to imply a Proto-Semitic variant *'an-ka I * 'an-ki of 'an-ta I 'an-ti for the second person singular and may suggest that the addition of -ku I -ki to the independent pronoun of the first person singular arose by analogy with the variant suffix of the second person. Instead. In Arabic. Ugaritic ('ank.7. k. plur. . Since more and more Proto-Semitic features are being discovered in Semitic languages and dialects which are still spoken. Therefore.) with variants in West and North Gurage dialects. In fact. A morphological difference characterizes the Tuareg pronouns of the second person which are formed on the basis k. Samalian ('nk). but also in Egyptian (nttn < *ntkn. "you". -ta I -ti raises the problem of the alleged AfroAsiatic opposition of masculine k vs. "my". a-na-ku). Assyro-Babylonian. §36. Tuareg (nak). Hebrew. " o f you two".) and possessives (kiyya / tiyya. " o f us two". but also in some South Ethiopian languages: the singular pronoun "you" is ank in Argobba. " o f them two". we assume a suffix -kā with the -ā of the subject case. For the ProtoSemitic dual. like in the Arabic dual. feminine *an-ti (cf. and ancient Egyptian (ink). The ending of the first person pronoun dkdy of the Mehri dual corresponds to the dual -ay morpheme of the oblique case. the Modern South Arabian forms suggest that the dual mor­ pheme was added to the singular stems of the pronoun. as well as with the first person dual in Mehri (dkdy) and Soqotri (ki).302 MORPHOLOGY Palaeosyrian.of the Semitic suffixed pronouns of the second person (§36. but that the Semitic languages have later used the forms with -k. -śny = *-šny. Moabite ('nk). Phoenician ('nk). etc.

is probably due to the influence of the following pharyngal (§27. instead. 'antan). is replaced by secondary formations in most modem Ethiopian languages. Argobba prefixes the same element anna to ankum. colloquial Arabic.). plural of the singular ank. 36. annantā for both genders. The West Semitic vocalization nah.g. the -a of Aramaic. where the dissimilation did not take place. and fem. in Cushitic.10. 36. although Egyptian inn. thus nassakum for the masculine and nassakan for the feminine. collo­ quial Arabic (nihna). thus àkàkac. etc Instead. Modem South Arabian. while Cushitic nu. thus annantá or dnnantum. 45) to the singular pronoun antā.PERSONAL PRONOUNS 303 36. 41. while the form nih.8. as well as dual. while the Gurage dialects. "you". nuna.was used also for the first person plural. and in Egyptian. while Gafat prefixes it to the sin­ gular anta or to the plural antum. use forms basically identical with the suffixed personal pronoun. e. which changed into h.in several Semitic languages (§15.u> d) This explanation is con­ firmed indirectly by the Assyro-Babylonian change nīnu > nīni. and in some East Gurage dialects (atum.33. "you" (§36. For the third person singular and plural. and Ge'ez (ndhna). As for the variants in the final vowel. exactly as in the suffixed personal pronoun . in Semitic.10). intena (non-subject). thus allantā. inno seem to indicate that nu is the only common element of Afro-Asiatic. It is still found in Tigre ('antum.which is used for the first person plural also in the suffixed personal pronouns. nuni. thus dnnankum. we may posit a Proto-Semitic element Š-.4). akma (fem.9. Amharic prefixes the element dllá I dnnà of the plural demonstrative (§36. some Cushitic languages of Ethiopia preserve the original plural pronoun "you". in Chaha: aku (masc). Harari adds the nominal plural ending -ac to the singular pronoun akak. vocalized anon or anan in Coptic.). masc.6). and all the Ethiopian languages is likely to be occasioned by a dissimilation of the final -u of nihnu from since the opposition /: u is weak in Ara­ bic and in South Semitic languages (i > d. like Walamo inte (subject). The Semitic element nih is followed by this morpheme -n. The Proto-Semitic personal pronoun of the second person plural. indicates that the element 'an. "person". in LibycoBerber. other than East Gurage. 'antarì). in Tigrinya only in the vocative "O you!" ('antum. attested in Ge'ez. Tigrinya uses the suffixed noun nāfs > mss. For the first person plural of the subject case we may posit the Proto-Semitic form nih-nu. the weakly opposed vowels i and u were harmonized.is implied by East Semitic (nīnu).

"he". and they distinguish the masculine -u ending from the feminine -/. thus with the addition of the plural morphemes -mu and -n(a). It is likely that this vowel was originally short. and šīt. e. their". "she". Oromo isān. his". like for the second person plural. "these" (fern. ti(n).< šu. The plural was marked by the addition of the morphemes -nu and -na. with a vowel correspond­ ing qualitatively to the semivowels w and y. ./ hi./ ši. §61. in Middle Assyrian.17). As for the feminine sē of Modern South Arabian languages.5). ta. "she". and in the conditional particle (s-m/n > h-m/n. which resulted in a masculine pronoun šu-wa and a feminine pronoun si-ya. the initial element hu. e. hut. anterior to the change ša > ha of the mascu­ line pronoun. yi(n). and hdta. "she". thus 'dmuntu and 'dmāntu. or huda. "they. "this" (fern. The distinction between masculine and feminine was indicated by the suffixed morphemes -wa and -ya. si Isil.> hd.g.11. that must be related to the Libyco-Berber determinatives wa. and in the Western dialects. In an early Ge'ez inscrip­ tion. 9 36. "these" (masc). and derive from the oblique case (§36.2).> '-m/n. however. it may have resulted from an early shift šī > sī. plur. These forms are paralleled in Cushitic. in the causative verbal stem (š. "he". "she". hida.and ya-. "he". The resulting forms are wd'dtu and ya'dti for the singular. "he". "her".).> hd. The h also survives or is rein­ forced to h in the Gurage dialects.> '-. masc sing.is omitted and the ending of the former oblique case of the independent personal pronoun is added to the elements wa. which likewise preserve the / or voice it to d. Among the modern Ethiopian languages.11).g. These dialectal forms are obviously related to Tigre which has preserved the hu. and hmnt (*hdmuntu) for the 3rd pers. "him. these morphemes are placed before the ending of the former oblique case. Oromo isā. especially šūt. "he". with the t of the oblique case. §41. with parallel changes (§36.3). also in Old Assyr­ ian and in later Assyro-Babylonian dialects.). Variant forms appear in Palaeosyrian (su-u /šū/. "this" (masc). "she".20).> h. wd'dtomu and yd'dton for the plural.304 MORPHOLOGY of the third person (§36. In the variant form of the plural pronoun. as in Ge'ez. "she"). cf.g. isī. "he".of the masculine and the hi. In Ge'ez. isāni. the endings -u and -a being those of the suffixed pronouns of the third per­ son masculine and feminine (§36. which are used in Old Assyrian. and ydt. masc. as shown by hdtu.of the feminine. The same morpheme characterizes the plural in Cushitic. e. cf. hit. the independent personal pronouns are h't (*hd'tu) for the 3rd pers. "them". Gafat comes closest to Ge'ez with the pronouns wdt.

The form of the second and third persons sin­ gular can be traced back in the Egyptian language of the Pyramid texts and of the Old Kingdom (§2.is very likely related to the demonstrative han.e. Gafat.32). lit. for the feminine azze. *cuwāti < *kuwāti). ydt. ndssa. They obviously go back to the determinative-relative element z (§36. Argobba. The vocalization of the South Arabian pro­ nouns hwt and hyt of the third person singular is reflected in Andalusian forms transmitted by Pedro de Alcalá as huet and hiet. -ze. for the plural azziyac. "he".122) and they are closely related to the old use of the suffixed form of the same noun to express the reflexive pronoun (§36. drsu comes from *rd'su. viz.g. when the corresponding masculine inde­ pendent pronouns were twt (i. "my person will not know the way": TAD I I I . and kdssa.17) and a second element -(w)āti / -ūtì. differ from Ge'ez.9). from nafsu. from kársa. . "her person" > "she". and nafsa. "me". "she". "he". Classical Arabic qāla li-nafsihl. "he said to himself". "he".2). "his head" > "he". and dss a. kut. C 1. Other Ethiopian languages. probably from kársu. Amharic. "us". npšy V td' 'rh'. The morpheme -t(i) characterizes the object-case of the personal pronoun also in Cushitic languages.> -zz. Hebrew 'al tašši'ū napšotēkem. and Harari. Zway aya). The element yaof the first person singular apparently parallels the independent personal pronoun " I " in Argobba (ay) and in some Gurage dialects (Chaha and Ennemor dya\ Masqan dyya. Amharic. the endings -zo. and śwt (*šuwāti). drs a.1. Amharic dssu. but y results there from the palatalization n > h > y. "you". e. Tigrinya. 36. and Argobba express these pronouns by a noun with a suffixed pronoun of the third person.PERSONAL PRONOUNS 305 36.41). anat. w w 36. "her head" > "she". The oblique case of the independent personal pronoun has a first element corresponding to the suffixed personal pronouns (§36.14. the final vowel of which is occa­ sionally replaced by a or u. Tigrinya. from *rd'sa. Tigrinya has ndssu. "she". in the Agaw dialects of the Qemant-Qwara group.like in Hebrew hazze.g. While the element az.12. "you". "do not deceive yourselves".13. "her belly" > "she". Tigre. Argobba has kdssu. and Gurage in the formation of the personal pronoun of the third person. Harari has for the masculine azzo.28). These formations of the personal pro­ noun parallel an Aramaic use of the suffixed noun tips (e. with an assimilation -nz. e. "his person" > "he". like for the sec­ ond person plural (§36. "his belly" > "he". "they".(§36. -ziyac are the respective suffixed personal pronouns of Harari.g.

" 3 m.e. [-a] -ka. -iyi -k -m -s f -t -s — -ka -ki -s -s -na -ka -ki -sa -ta -iy -ni -ka -ki -ŠU -ī. 1 (noun) ] " -/ (vei-b) j 2 m. Egyptian Sing. -ni -kunu ? -ni[ātifāši] -kun(u)[/ūti/ūši\ -kin(a)[lātilāši] -šun(u)[lūtilūši] -šin(a)[/āti/āši] - -śn -san 1 -tan 1 -sna -sant 1 -tant J -su{m) -šunu -šini . Old Akkadian. -na -warn -wamt 1 J -n -kna -mu -ku(m) -na -kun -kin -šun -šin -na. [-sum] -Sa. or by way of assimila­ tion or contraction. -nu. [-ku] -ki -šu -ša. f. Allomorphs may occur after a verb. [Si] -Jfc -t -f -Ś 1 -t -ša -ny -tny -śny -nay{a) -k{un)ay{a) -š{un)ay(á) -naya. -š -I. -ya -{an)ni. [-kum] -ki. -k -Su. to prepositions to express various relations. -šumā -šunī[ti/ši] -n Ì ' J 1 ' J -í n -na.16. 3 m. " i t is ascribed to him". The bound form of personal pronouns can be attached to nouns as possessive pronouns. as accusative and as dative. -kumān -kunī[ti/ši] -iśumaya. Dual 1 2 3 Plur. -ya -ni -ka.g. pending on the consonantal or vocalic ending of the verbal form. B. i. e. For details. ana šu(w)āšim šateršum. 1 2 m. f f Tuareg Bedja Hausa *Proto-Semitic Palaeosyrian Old Babylonian -i.306 MORPHOLOGY 36. and to verbs both as direct and as indirect object. -š. and Old Baby­ lonian independent personal pronoun. In the dative of the Palaeosyrian.15.17). Suffixed Personal Pronouns 36. -niya -kumaya. f. the second element is -(w)āši I -ūši instead of -{w)āti / -ūti. The attachment of these suffixes to the noun may be effected either by means of case endings or glide vowels. There is no evident connection between this element and the postpo­ sition -iš I -es of the so-called dative-adverbial case of the noun (§32. grammars of the various languages and dialects should be consulted.

•-k -hi. -hū/i -hā *-n •k *-k -hw -h. ki-lāmda. The particular morphemes of these forms are placed between square brackets without the mimation which is often added to the dative ending -ši(m). "my soul". nay-ki. -ā(h) -ki -Š -U. but not in Old Assyrian. -hinna -hn . -hw -i. like the Agaw dialects of the Qemant-Qwara group. e. -ha -ki -km -kumā -humā. -u. -) -t •w.g. -ih -hā. a paradigm of Egyptian. -āw. attested in Old Akkadian and in Old Babylonian. "his place". -a. ni-sdbra. -o. -nn] -hū. Bedja. anadàra or anâ adâra. "all of them". -hu -u. Tuareg. -wo -a. and Hausa suffix-pronouns is added for comparison. -ya -rii -ka -ki -huli. -n. -him(ū) -hmw -hunna. ys-nkdra.SUFFIXED PRONOUNS 307 In some Cushitic languages. I fi. -ān(ā) -hēn -hum(ū). -6 -i -nī -k -kl. ka -š. -hà -hu -s.i -h. in the accusative and the dative for Palaeosyrian and East Semitic. The following paradigm of the suffixed personal pronoun in the principal Semitic languages includes the East Semitic and Ugaritic suf­ fixes of the verb. Ugaritic Hebrew Aramaic Classical Arabic Sabaic Mehri Ge'ez Tigre Tigrinya Amharic -V -// -i -ni -kā -k.14). etc. -wa -wa. -sa -hā -nh. the pronouns are not suffixed but prefixed or placed before the noun in the possessive form and before the verb in the form of the object-case (§36. -nn] -hā. -o. ām{6) -hon/m -hen(ā). 36. -yà -ay -k. "our God". -himā -ki -hmy -hi -nā. "your shadow". Besides the paradigm for the main Semitic languages and for the reconstructed Proto-Semitic forms. -šá -ya -ni -ka -ki -ye -ni -ka -ki -k -k -āy -ni -ka < )e -nn -h y V/. -ah -I. -an -km -kn -hm -hn -n -kem{ā) -ken(ā) -kon/m -kēn -{aíía)n -kammu -kum -kan -kan -{h)omu -(h)om -{h)on -(h)an -kum -kan -{w)om -(w)ān -aéèahu -aíéàw -kum(ū) -kunna -kmw -kam -km -ham -san -hem{ā). -ki -n.17. '-nh.

colloquial Yemenite. is paralleled at sight by Mishnaic Hebrew yes-no. like in other South Ethiopian languages. 36.308 MORPHOLOGY 36. "he is". "like me". the suffix of the verb is -ni. The third person singular reflects the same changes š > h as the independent pronoun (§36. The suffix of the first person singular. Therefore. 36.19.). Both forms -y and -h occur as verbal suffixes also in Gafat (South Ethiopic). results from the adding of -I to the ending of the masculine dual or plural construct state. The Amharic masculine object suffix -(a)w and the Hebrew suffix -āw / -6 result from contractions of the type -ahu > -au > -aw > -o. Old Phoenician 'b. This nominal -n. *sūsay-ī > sūsāy. and in 'ddennī <*'odē-nī (cf. Gafat -c / -š < -ki. "my father". while Amharic -t of the masculine is an allomorph of -u I -w after the vowel -u. is usually absorbed in the palatal. . regardless of the termination of the governing word. is -iy > -1 after a consonant or a short vowel. which causes palatalization.) and śy (fern.10). and spread later to other persons. occurring first with the suffix of the first person after nominal forms terminating in long vowels to prevent hiatus. The vowel i. Instead. but it is still -dni in the Harsūsi dialect.18.suffix is a probable transference from verb to noun or preposition. correspond to Egyptian -k (masc. with the exception of Mehri -dy. "she is". " o f my mistress". "my time". yeš-nāh.20. "they are".8). lit. Palaeosyrian exhibits a shortened form -š of both the masculine (l-suf) and the femi­ nine il-saf) passessive suffixes (genitive). The feminine forms in -s of Modern South Arabian should be related to the sē of the independent pronoun (§36. -ki. e.1-3). "my father's"). The latter derives from a palatalized -ki > -c ("t") (§15. added to a noun or to a preposition. a better parallel is provided by the Phoenician and Punic suffix -nm of the 3rd pers. but yeš is a frozen form of a verb (§49.30).10) and corresponds exactly to the Egyptian "dependent" pronouns św (masc. The femi­ nine suffix -at is the corresponding nominal ending (§30. and -ku for the masculine dative. The Hebrew nominal suffix -ni in kāmd-riī. but 'by. and -ya after a long vowel and after the originally short vowel i of the genitive (e. attested in Hebrew and in Aramaic. -the Amharic masculine suffix -h and the Gafat mascu­ line suffix -hà have in reality a [x] derived from a spirantized -k(d). "my horses".) (§36. It corresponds to the Egyptian suf­ fixed pronoun The form -ay. Assyro-Babylonian be-el-ti-i-a.) and -t (fern. The second person singular suffixes -ka.g. and Amharic -s < -k. Perhaps by analogy with the suffix of the first person plural.23). plur. yeš-nām. 'ddē-nūlhū). exactly as Modern South Arabian.g.

A-hu-na.SUFFIXED PRONOUNS 309 36. Gafat (-na).22.g. "Our father". -nn of the third person sin­ gular can be explained by the use of the energic endings -anna or -an of the prefix conjugation (§39. AssyroBabylonian. -su-ma-a l-šumayaj. without final vowel: Amharic fâl- . -śny = -šny. and Hebrew is most likely the result of analogy with the final vowel -ū of the inde­ pendent pronoun (§36. the Gurage dialects (-na. Since -nh alternates with -n in other­ wise identical contexts.g. "was given up"). lš-hi-lu-na /Yit'(u)-'iluna/. should then be explained as *-an+hū+un > -annūn. yqbr. in-ne-du-ú for normal Babylonian innadū. In several languages. "of both". 'aqrbrn. "he buried him".g. " I shall bury him"). "Our mountain"). The latter group of languages either did not preserve the plural morpheme -n/m.of -u. As for the second n of the -nn suffix. Sa-dú-na. " I shall bury him"). while the suffix -ni of Old Akkadian. -Una. The Ugaritic object suffixes -nh. 'aqbrnh.2). Arabic. The dual forms of Proto-Semitic ended most likely in -ay(a) which is also the ending of dual nouns in the oblique case and which appears in Ugaritic -ny. The Proto-Semitic suffix of the first person plural was most likely -na. Harari (-zina).23.g. 36. -tny < *-kny.33). A-bu-na. 36. Ge'ez. e. The Ugaritic suffix nn of e. it should be compared with the enclitic -dn which can be added in Gurage dialects to the object suffix of the main perfect or imperfect without an apparent change in meaning.g.in the position u-u. while -n stands for -an + hū > -annū (e. "you make us die": EA 238. §36. but also by the archaic or dialectal suf­ fix -na in Old Akkadian (e. Aramaic. corroborated both by Amorite names (e. -šuni. -kuni. a spelling which seems to imply a colloquial reduction of an original Amorite -na. the suffix of the first person plural is attested also under the form -(a)n. with the allomorph -t. and it is implied by the Modern South Arabian -ki < -kay and -hi < -hay. and Tuareg.nn. "Our brother".8-11). "he killed him".g.or reflects the Proto-Semitic situa­ tion (cf. and apparently Palaeosyrian (-ne I-nil) results possibly from the generalized use of the old ending of the oblique case (?) -ni. as suggested not only by Palaeosyrian. "The Saviour is our god") and by the frequent form -ne of the suffix in the Mari documents.21. Tigre. often written as a separate word. gú-ma-a l-kumayal. The vocalization -nu of Palaeosyrian. Old Canaanite (ti-mi-tu-na-nu. gāddàlā-nnd-t-dn < *gáddālâ-nnu-u-3n. like in comparable cases at Mari (e. Tigrinya (-na). in Old Egyptian -ny.g. -nh represents the ending -anna + hū (e. -Ma).7). -n. in Palaeosyrian -na-a l-nayal or -ne-a l-niyal.

As for the final vowels. The vowels marked by h in the suffixed pronouns of the Hebrew Qumrān scrolls (-kmh. the situation is the same as for the second person (§36. 36.5. The observations made on the consonantal elements of the inde­ pendent pronoun are relevant also for the suffixes of the third person plural (§36. -šum. are attested also in Palaeosyrian. In Modern South Arabian. -knh. The ending -āt(i) / -ūti is used for the plural suffixes of the Babylonian accusative and of the Assyrian dative. with a great variety in their Old Akka­ dian use (-šunu/i/a) and a generally attested feminine -šin in that idiom.9). The suffixed pronouns sometimes have dative force in other Semitic languages as well. -hmh. only traces of the oblique case of the pronominal suffix are attested outside Palaeo­ syrian and East Semitic. the personal suffixes are affixed to definite nouns (§33. For the second person plural. -humū. and there is an additional suf­ fix -a(m) of the first person. Mehri and Harsūsi -abydtidn.26) or represent a late development by analogy with singular suffixes. "our house".5) have to be taken into account and Proto-Semitic forms -kun and -kin posited. and -sum. "to him". The dative suffixes -kum. Both types of endings correspond to those of the independent pronouns (§36. and it is not certain that this is due to the oblique case.24).26. followed by the Old Babylonian poetry. 36. With the exception of the accusative/dative pronoun -ni(m) of the first person singular. the observations on the indepen­ dent pronoun (§36. Neo-Aramaic bētan. 10).g.25. In Old Babylonian. In Old Akkadian. "our houses". attested in all Semitic languages. while the Arabic forms -kumū. but there is no evidence to show that there was a formal distinction between accusative and dative suffixes. e.24. The Old Akkadian and Assyro-Babylonian dative of the singular suffixes is characterized by the frequent use of the mimation (-nim. -kim..3-4). while the ending -īši(m) / -ūši(m) I -āši(m) is employed for the plural dative suffixes in Old Akkadian and Babylonian. with -himi. there is either no final vowel in poetry (-kun) or the vowel is -u (-kunu). -šim). the attested vowel is -/ (in qá-tì-ku-ni. without the final vowels which are unstable and which are missing or can be omitted in most Semitic languages. -kum. since the same text has in na-ap-ha-ri-su-nu. -himū are poetical and can alternate.310 MORPHOLOGY làgán. " i n their total". -hnh) may either be a trace of the ProtoSemitic oblique case in -āt (§36. 36. "to you". . "he wanted us". " i n your hand").

g. "himself". quna.R E F L E X I V E PRONOUN 311 36.g. It is likely that qnūm-. "person.16). or napš-.g. qīqrii. "head" (cf. Christian Palestinian Aramaic has a word qlqn. Sabaic grmk. A similar use is attested in other Semitic languages. l-qn'm. Syriac ba-qnūmeh. Two suffixes can be added to a verb without intermediate preposition in Old Akkadian.g. "for himself"). "about herself". "he said to himself").functioning with a suffix as a kind of reflexive pronoun. especially with nafs "soul". "belly" (cf. "head" (e. lit. "by himself".g. "he sold himself").12). e. Punic. the other as indirect object. and Arabic. Neo-Punic p'l mqr. Hebrew 'āmar bd-libbo. §36. "about her bone". "he gave it to me".. one act­ ing as direct object. Syriac. However. "self". Reflexive Pronoun 36. besides the generally recognized ramanu (e. is an allophone of qdn. This word probably derives from the reduplicated root *qdnqdn. one finds sometimes another noun. Classical Arabic 'a'tā-nī-hi. "bone" (e. Hebrew wayy'aś Id 'Ēhūd hereb (Judg. .g. being". "yourself").. This construction is attested frequently in Ethiopic.12). with the required pronominal suffix. instead.g. Arabic ba'ata 'dā Marwāna fada'āhu 'ilayhi. The first person suffix precedes the sec­ ond and third persons. "Ehud made a sword for himself". There is no distinct reflexive pronoun in the Semitic languages which can use. "he swore by himself"). "heart" (e. the simple form of which is used in Gurage dialects with the sense of "single". > n 9 s s . while its derivative indicates the unit in the Gafat numeral "eleven": asra qdmcatta {<qdmt < *qdn-t + atta). and *qdn have the same origin and that Amharic qdl. qdl-u. "oneself" (e. also §36. pagaršu ina šīmim iddin. Assyro-Babylonian. In East Semitic. ana ramanisu. "he sent for Marwān and summoned him to himself". the second precedes the third. Semitic languages prefer to employ the noun raman-. with the change n > I. but also with nouns meaning "head". qdl-eh. Old Babylonian atrudakkušsu < *atrud-am-kum-šu. regardless of their syntactical function. "I sent it to you".27. and ra'as.g. "you yourself"). Phoenician. 3. e. "body" (e. "person". "alone" (quna > qura-).g.28.g. e. " I myself". Syriac 'al garmah. C . qaqqassa ana šīmim iddin. not only with napš(e. or pagru. the usual suffixed pronouns that are then refer­ ring to the subject of the sentence. Hebrew nišba' bd-napšo. "she sold herself"). "Maqer made it for himself". as qaqqadu. e.g. and Samaritan Aramaic use also the noun qnūm-.

Beside the suffixed personal pronouns of the noun which act as possessive pronouns (§36.312 MORPHOLOGY D. f. Independent Possessive Pronouns 36. has such a pronoun which is formed by the comple­ ment of appurtenance yâ. 3 m. 36. ya '-um yatturn kūm < < ya'-t-um 2 m. while the process of its transfer into the adjectival category is already accomplished in Old Assyrian.17). In East Semitic. in the way Latin does. śt <*kn kuwa-um kan(?)-t-um šuwa-um < šan{l)-t-um kattum < šūm < šattum nūm < m-um *kunūm < kuni-um šunūm < šuni-um . Plur. tw < *kw tn św śy. the independent possessive pronoun originally formed a separate inflectional class which is indicated by some Old Babylonian forms. There can be little doubt about the Proto-Semitic or even Afro-Asiatic origin of the posses- Independent Possessive Pronouns Egyptian Sing. wì Old Babylonian sing.30. Semitic languages do not distinguish. f. with the exception of Harari.combined with the personal pronoun. and South Ethiopic. East Semitic has two types of indepen­ dent possessive pronouns which are formed on the same basic mor­ phemes as the suffixed personal pronoun.29. between suus and eius.

the Old Babylonian forms attested with "pronomi­ nal" inflection. and "plur. since it is paralleled by the Old Egypt­ ian "dependent" pronoun. This inflection probably reveals a Proto-Semitic or even Afro-Asiatic origin of this independent possessive pronoun. etc.48) in a pattern comparable with the Arabic collective ending -atun (e. This ending is added in the masculine forms to the abstract-collective morpheme -vtt (§29. mei. it is added to the fem­ inine morpheme -t. "Moslems").INDEPENDENT POSSESSIVE PRONOUNS 313 sive with pronominal inflection." refer respectively to the singleness and the plurality of the items possessed. like Latin meus. muslim-un. Old Assyrian plur. "Moslem". yūtun < ya '-ūt-un ya'-t-un kuwa-ūt-un kan(l)-t-un šuwa-ūt-un šan(l)-t-un ya'um yātum ku(yv)a 'um ku(w)atum šu(w)a'um šu{w)atum ya 'ūtum yattun < kūtun < yātum ku(w)a ku(w)ātum *šu(w)a'ūtum *šu(w)ātum > šā'ūtum 'ūtum kattun < šūtun šattun < < nūtun < ni-ūt-un ni(y)a'um ni(y)atum kunu'um kunūtum *šunu šunūtum 'um ni(y)a ni(y)ātum *kunu 'ūtum *kunūtun < kuni-ūt-un 'ūtum kunu(w)ātum 'šunu 'ūtum *šunūtun < šuni-ūt-un šunu(w)ātum . In the feminine forms. meae. Independent Possessive Pronouns Old Babylonian plur. Old Assyrian sing. The two sets "sing". The "pronominal" inflection of the Old Babylonian pronoun is char­ acterized by the plural ending -un which is considered as common to the Afro-Asiatic languages. muslim-atun. and the adjectival inflection of the mainly Assyrian forms of the possessive pronoun in the nominative case. The following table shows the Egyptian dependent pronoun.g. mea.

combined with the suffixed or independent personal pronouns. attukunu f. and the Amharic para­ digms in parallelism.g. 'ēt of Hebrew. A similar formation is attested in Cushitic and in Egyptian. and yt of Aramaic. attuki m. 'iyyā-ka I -ki.31.parallels that of the Ethiopic complements of appurtenance nāy. attušina 3 nāyna nāykum nāykdn nāyom nāyan yàrína < yà-dhna "ours" } \ yànnantā < yā-annantā "yours" J ] >• yànnàssu < yā-dnnāssu "theirs" J . wt. and tā. either noun or pronominal suffix (§52.10). "his" "hers" Plur. when the object possessed is of the feminine gender (e. t of Punic and Mishnaic Hebrew (Bar Kokhba letters). "his"). when the object possessed is of the masculine gender (e. attušunu f. We give here the Middle Babylonian. 1 2 Amharic 3 attu 'a m. attukina m. the Tigre. "you".< * 'aytu-. and Moabite. which is to be identified with the Arabic particle 'iyyā < 'iyyatintroducing the suffixed pronominal object ('iyyā-ya. attušu f. tā-sa. Middle Babylonian Tigre Sing. "me". etc.g. and it occurs also in Hausa (Chadic) with the complements of appurtenance nā-. In the non literary language of the Middle Babylonian period appears a new type of independent possessive pronoun formed by addi­ tion of possessive suffixes to the complement of appurtenance attu.(< ntā-). "yours" fem. Edomite. "his").and yá-. attuša nāye nāyka nāyki nāyu nāya yàna < yā-dtie yantà < yà-antà yanéi < yā-ancì yâssu < yà-zssu yâss a < yà-dss a w w "mine" "yours" masc. 'yt.) and with the "accusative" particle 'yt of Phoenician. nā-sa. 1 2 attuni m.314 MORPHOLOGY 36. which is an optional mark of the definite direct object. The Middle Babylonian use of attu. attuka f.

and with its later syncopated form hart. Arabic hādā (§36. "he". "this". to Harari azzo. za.> hā.38).32. which formally corresponds to the later Hebrew hazze.13). while the North Ethiopic Tigre demonstrative is 'dlli < *'ulli. Gurage zd.appears in Old Akkadian and in Assyro-Babylonian under the form anni-u(m) > annūirrì). and anna (plur. Two series of demonstratives can be distinguished in the Semitic languages: demonstratives of remoter deixis or "far" demon­ stratives ("that. "these".). with a Babylonian variant ullūm.DEMONSTRATIVE PRONOUNS 315 E. It is used also in Libyco-Berber (§36.). and demonstratives of nearer deixis or "near" demonstratives ("this. vs. and to Amharic dnnàzzih or dllāzzih. "that". ya. However. Gafat dhhd < *hinni. while dhhd occurs in various Amharic compounds. azze.with its variants *halli.from the "far" demonstrative in -a I a. dnna (fern.33-34).). sometimes under a variant form or with affixes. The initial h was also preserved in the Aramaic dialect attested indirectly by the Neo- . the opposition d : a distinguishes the "near" demonstrative in -d / d. those").45). Demonstrative Pronouns 36. The demon­ strative position of "previously mentioned" can be assumed by the def­ inite article or its equivalent (§33.g. §36. annāz. besides the Amharic base -āzzih preceded by various prepositions. can be used in both acceptations. "these" (§36. This distinction can be estab­ lished for each language only on a contextual basis. "this".6 ff. "this ". are attested in Gafat. Amharic yd(h).in nearly all the South Ethiopian languages.). "this".and 'ulli-. In South Ethiopic. anne or dnnā. "this". with the same final vowel -ā as in Arabic plural 'ulā. vs. and an-ne /hanni/. "she" (§36. viz. are found in East Gurage.of the West Semitic definite article. ahhd < *hanni. and the determinative-relative pro­ noun have the value of a demonstrative in several Semitic languages. thus hn-d (*hanni-dū or *hinna-dū). "those". the independent personal pronoun of the third person. a/dhhd (masc). vs. "that". The same demonstrative is composed in Ugaritic with the determinative-rel­ ative pronoun d > d. to Gafat dnnāz(dh). *hanni. Besides. these") (cf. a/dnnà (plur. There is one Proto-Semitic root morpheme that functions essentially as demonstrative. It does not yet appear clearly whether the same opposition exists in Palaeosyrian between i-ne /hinnil. in Assyrian also ammiu(m) and allū. (?) (?) 36.34). e. "that" (§36.33. "that ". placed before or after a substantive. both literary and dialectal.45). because the same root morpheme. The demonstrative * hanni. The equiva­ lent of annium in Mishnaic Hebrew is hallā and in Syriac hand. "these".

Amharic and Argobba alia. 'It in ancient South Arabian. 'illēk. 7 in Phoenician. with a variant 9n(n)ahuš.of the definite article (§33. The Ethiopic verb of presence hallo < *hallaw. end.316 MORPHOLOGY Assyrian hanniu. where the fossilized pronominal suffix -š used as definite article (§33. in the Punic dialectal demonstrative hnkt (§36. 'ēllū in Hebrew. f. while other Ethiopian languages adapt its final to the usual ending of the perfect. wanndt and wállāt. Plur. "he is (present)". 'ēlle. "forked digging pick"). matter". m. This shift should be explained by the ancient tendency of the Semitic languages.> 'al. that go back to the allophone *hanna of the demonstrative.35). > *hallau > *hallaw. preserved in Tigre ha(tu) and in the Gafat -ho suffix: *halla hu. Related Demonstrative Pronouns Assyro-Babylonian Sing. "here he is".34. 'In. "here he is". best attested in the Gurage dialects (e. and in the Sabaic indefinite pronoun hn-mw or hl-mw. The change h > ' is otherwise widespread in the West and South Semitic lan­ guages which kept using the morpheme *hanni to express the plural demonstrative: 'ēlle. 'Men in Aramaic.10). anniu(m) annitu(m) anniān annitān anniūtuim) anniātuim) Ugaritic hnd = *hanni/a-dū hndit) = *hanni/a-dā(t) hndn = *hanni/a-dān *hndtn = *hanni/a-dātān . and Harari hoi.> -11. 'dllū in Ge'ez. in the 18th century. thus "thing. Gafat and other Gurage dialects have forms based on and. The demonstrative annitān at Mari is interpreted here as a frozen feminine dual originally meaning "this and that". "there he is".3-4). some Gurage dialects alā.reflected by all these forms paral­ lels the shift han.are given below in the non contracted form (-ium > -ūm).g. and it is used in Tigrinya under the form 'alio. The latter is still attested in Gafat. goes prob­ ably back to the frozen demonstrative halla followed by the personal pronoun hu. f. Dual m.'ulā. 36. which is etymo­ logically and functionally related to the demonstrative. The change -nn. The principal forms of the Assyro-Babylonian demonstrative *(h)anni. in the very compound dn(n)aho < *hanna-hu. f.13) is added to the demonstrative. thus Tigre halla.in Classical Arabic. m. a phrase comparable with ancient Hidjazi huwa dā. . "whatever". Ge'ez compound hallo is further inflected like a perfect notwith­ standing its present meaning. to alternate the liquids / and n (§17.

Mishnaic Hebrew. 36. agannūtu (masc. The plural was usually formed by adding the demonstrative anniūtu > annūtu (masc) or anniātu > annētu (fern.). šunu (plur.). a-ġyul-inn.). — followed by the complete regressive assimilation ng > gg. — frequent in Late Babylonian (§27. 'ln/t in Sabaic. The Gafat "near" demonstrative is dhhd (masc).). while Mandaic hānāt. Hebrew and Punic ko < ka.) to the element ag(g) < *ang < *hank.(§36. plur. e.). Also the Punic demonstrative hnkt combines hnwith the deictic element -ko (cf. while the "far" demonstrative has an initial a. agātu (fern. It is invariable and is suffixed to the noun.32): ahhd (masc). a-ham-dnna.35. This demonstrative is attested also in Libyco-Berber where it is used as a "far" demonstrative. ammiu(m) and allū in Assyrian.).). The Neo-BabyIonian and Late Babylonian "near" demonstra­ tive agā (masc). "that jackass".). possibly derives from *han-kā with a partial progres­ sive assimilation nk > ng. A by-form ending in -a adds the nuance of "mentioned before". Mandaic. are employed as "far" demonstratives. although forms from other idioms could also be referred to. and Tigre. Syriac. The pronoun anniu(m) is used in Assyro-Babylonian as demonstrative of nearer deixis. to the element agā (agāšū I -šī I -šunu). "here".7). "the tent (in ques­ tion)". A parallel "far" demonstrative was formed by adding the independent personal pronoun šū (masc). hand in Syriac. and ullūm in Babylonian. while its variant forms. anna (plur. dnna (fern. agannētu or agātu (fern. The demonstratives hn-d in Ugaritic. anna (plur.g. plur. and 'dlli in Tigre are also used for the nearer deixis.DEMONSTRATIVE PRONOUNS 317 demonstratives are selected only from Ugaritic. Demonstrative Pronouns Syriac hānā hādē Mandaic Mishnaic Hebrew hallā Sabaic (hn/l-mw) Tigre 'dlli 'dlla In hallēn h 'n 'twn h 'n 'tyn 'In 'It 'dllom 'dllan . e. šī (fern.and Mishnaic Hebrew hallā appear to be "far" deictic pronouns. Sabaic.g.

šī (ha)hV h'. hy hy Ipn win- . The independent personal pronoun of the third person is used as a demonstrative in East Semitic. m.9) followed by the ending -t which is suffixed in Phoenician-Punic also to other demonstratives (hmt. 'st). Hebrew. st. .. . f. sunatunu r (ha)hem(ma) hmt innun iptn tin- v . "this is her stele". hnkt n'bn'. . hnkt qybr tht 'bn zt. a form *hafinnilakdt has to be assumed.36. nominative oblique case Plur. The distinction between the personal pro­ noun and the demonstrative is here not formal but functional. Since there is no assimilation of n. sunuti. "the memorial of 'dyt (PN) is this stele". [innokot] i f the word appears in the "Poenulus" of Plautus. _ . m. . nominative oblique case Plur. Phoenician. Aramaic. f. "this one (a man) is buried under this stele". satina ' v hmt . although the demonstrative employed as adjective has a case inflection in East "Far" Demonstrative Pronouns Egyptian Sing. . . This demonstrative is used for both genders and appears to function as an adjective and as a pronoun of the nearer deixis. 36. this results from the following examples: n'pš š 'dyt hnkt 'bnt. . y (hā)hennā sinati(na). West Gurage. nominative oblique case Tuareg East Semitic Hebrew Phoenician Aramaic šū pw wu-1 wa- šuā(ti/u) (ha)hū' h' hw(') tw tu-1 ta- si šiāti. nominative oblique case Dual m. ancient South Arabian.-f. nominative oblique case Sing.318 MORPHOLOGY §49.

"Far" Syriac Neo-Aramaic Sabaic Demonstrative Tigre Pronouns Tigrinya West Gurage (Chaha) Qatabanic hāw (h)o h'. but have no direct correspondent in Semitic.2. a shorter form of which is used with the def­ inite article la-. Their p. its "far" demonstrative goes back to an older form of the independent personal pronoun. and the main variants of the Tuareg pronominal bases of demon­ stratives are added in the first and second columns of the paradigm.< *nàfs. hw' hwt sw s wt x l lahay 'atu huta hāy (h)ē h'. but used as definite article as well (§33. a nominative and one oblique case. at least in South Semitic and probably in North Semitic (§33.elements are probably related to the demonstrative and pronominal 6-prefix of Bedja and of West Cushitic (Omotic).and w. hy' hyt hmy hmyt *s y s yt ] laha 'ata hita *s my s myt i i hānon (')ān(i) hmw hmt sm s mt l l lahom 'atom hano hānēn (')ān(i) hn hnt *s'nt lahan 'atān hanāma . the Old Egyptian demonstratives. Also the suffixed personal pronoun was used as a kind of demonstrative and as definite article. As for Tigrinya. manifestly corre­ lated. viz.13-14).13). "soul / self" (§36.DEMONSTRATIVE PRONOUNS 319 Semitic and in Epigraphic South Arabian. replaced in the latter function by the suffixed noun ndss. For comparison. The Tigre "far" demonstrative is also related to the inde­ pendent personal pronoun.12).

hādā iha)dā (hā)dihī. Ara­ maic h' thwmwhy. m. The personal pronoun can also function as isolable demonstrative pronoun.33. Chaha arc huta. 'ulā .8-10) but which is functionally the demonstrative parti­ cle employed also in Aramaic and in Arabic (§36.38). e. "those kings". hallāz(e) (haz)zot. (hā)dī. the pronoun used adjectivally is preceded by the deictic element hā. In West and South Semitic languages. zn z('). Qatabanic bs wt mhrmn. dayni (hā)tāni.g. st ddnā. "that is what I said". Most languages distinguish demonstratives of nearer and of "Near" Demonstrative Pronouns Hebrew Sing. Hebrew hū' 'âšer dibbartl. The determinative-relative is employed for the singular demonstratives. 'ēlilū) 'I 'ēllē. e. 'illēn. which shows the mark of determinate status (§33. t(ih)ī. tā Dual m. hā 'ellayin (hā)dāni. " i n that sanctuary". 'z. as a rule.37. Plur. In South Arabian.which is formally identical with the defi­ nite article (§33. "on that day". (h)'z. often with the addition of a variety of deictic affixes. m. tayni (hā)'ulā'i. (')st. "that boy". f. In other languages. hāden dā(t). Phoenician Aramaic Arabic (haz)ze. (hā)'ēlle. ] 4 36. also the determinativerelative pronoun du (§36.< han. it precedes the substantive. hallēzū. zo z('). It is generally used as demonstrative of remoter deixis.g. Aramaic malkayyā 'innūn.g. e.320 MORPHOLOGY The personal pronoun accompanying a substantive functions as demonstrative adjective.12). it is generally placed after the substantive. Palaeosyrian in u su-wa-ti. while the plural is generally formed by the common Semitic demonstrative *hanni-/'ulli or by its derivatives.46) is used as demonstrative. "that are its borders". In Hebrew.

in some Phoenician dialects. "Near" Demonstrative Pronouns Sabaic Qatabanic Sheri Ge'ez Tigrinya dn dt dn dt dánu dinu zd(ntu) zā(tti) dZU dza dyn ? 'In dtn izanu 'dllu. 'dllojāntu 'dzom 'It 'dlla. Besides. These demonstratives do not show case differentiation. dn or dt). a-rgaz-ad.) and tayni (fern. in Middle Aramaic dialects.9).g. fern. in Arabic. with the exception of the Arabic dual attested in the oblique case also as dayni (masc.38.< han.< *halinnila-.g. "these". For the "near" demonstratives formed with the determinativerelative pronoun. A comparable usage is attested in LibycoBerber where an invariable near demonstrative -ia)d may be suffixed to nouns (e.DEMONSTRATIVE PRONOUNS 321 remoter deixis by means of different affixes. "this one"). The prefix is used in Hebrew. viz. The element -k seems to be related to the Semitic deictic particle -k(a) (§36. but vocalic variations may play a role as well (§36.33). there is in Tamazight a suffixed form dak > tak which qualifies the person or thing present or visible (e. 36. "this man") and to pronominal bases (e. 'dllo/āntā 'dzàn . mainly two affixes are employed in West and South Semitic. the prefix hā.g. Thamūdic (masc. dn. "this here man").41-44.32). Ethiopic. in Tigrinya. Both the prefix and the suffix are used in the Gafat "near" demonstrative anndzdh < *hinnazin. while the suffix appears in Aramaic.33). like earlier in Ugaritic (§36. and the suffix -n. wa-d.). 49. and in the Phoenician dialect of Byblos (zn). a-rgaz-adddk. South Arabian. and in South Ethiopic (§36.

hayye). The situation is similar in Gafat.40). 36. "these". — used also independently as dā and dl in Harsūsi. In South Ethiopic. ddndmdh (fern. "this" (masc). where hā.(e.). The elements in the Mehri and Harsūsi "near" demon­ stratives are the base form. "these" (masc) (§36. as zi or zd. hd(y)n. didha (fern. at Hassānīya (Mauritania). — beside the basic da (masc. also with new devel­ opments in the form of the feminine hādī. presents the forms hādawlā < *hā-dā-'ulā and dawlā'i < *da-'ulā'i. The determinative-relative develops its own plural form dū. a change which reflects the shift d > d (§13. New forms are encountered likewise in the Modern South Ara­ bian languages. while the Damascene form is haddle.is added to Classical Arabic (hā)'ulā('i) or develops its own plural form and replaces the element 'ulā('i).40. The shift d > d occurred also in many modern Arabic colloquials.g.). "this" (§36.). this plural is generally extended in the Maghrebine colloquials to hādūma/na with a feminine hādāna. "this which (is) here" (masc). with the exception of the demonstrative az-dhha (§36. A l l North Gurage dialects and the West Gurage Chaha preserve z as demonstrative also in free position.322 MORPHOLOGY 36. " i n this") and in the plural dnnàzzih or dllazzih. the situation differs from one language to the other.34). Instead. How­ ever. the demonstrative element z is used in Tigrinya: 'dzu. Beside Ge'ez. — are par­ ticular in the sense that the deictic element hā. The Aramaic demonstrative is written znh. 36.41. d'/h.8). The Yemenite colloquial of San'a. — and a -m?h suffix which occurs as a deictic in other com­ pounds.38). haydi. and especially in the plural demonstrative where d.) and dds (fern. 36. dandmdh (masc).).39. "these". 'dzom. The z of izànu in Śheri is the palatalized / of 'In.) in Mehri. §49.. which is reduced to dol(a) in the Meccan and Cairene collo­ quials. The Soqotri forms.) in Harsūsi. e.) and h'z' (fern. A similar "far" demonstrative occurs in Gafat: az-dhha. and dlyomdh (plur.g.is placed after the base form: dddha < *dd-d-ha.42. the element z is pre­ served only after preposition (e. Mandaic preserves the archaic spelling h'zyn (masc. and hn/fomzh (plur. "that". Tigre has the demonstrative 'alii. dlha (plur. ba-(z)zih. In Amharic.). "this". z'/h.9).). The final -h derives from the spirantized deictic -k (cf. The principal forms of the "far" demonstrative based on the determinative-relative d are compounded with the deictic element -k . "these". hd'/h. and later dnh.g. dlmdh (fern.changed locally into hay. thus domdh (masc).

hā'ulāk / hawlāk.45. cf. and in South Semitic languages. Amharic has developed three degrees of the demonstrative pro­ noun. The demonstrative referring to an object near the speaker (first person) is ydh (masc). f.12): ?rsu I BSSU (masc). the determinative-relative z forms "far" demon­ stratives in North Gurage dialects with the suffix -k. hādīk. m. tilka. dnndssu (plur.). or with other extended suffixes. The lateral z of izok in Sheri is the palatalized / of 'Ik (§16. or with the prefix ha.34).and the suffix -k (Arabic: hādāk.DEMONSTRATIVE PRONOUNS 323 (cf. 36. w w .). ydec (fem. which occurred also in many modem Arabic colloquials. dss a I ?rs a (fem. yacc (fem. They are attested in Aramaic.7).). 'antākti Plur. 36.). denāk dāk. Beside Ge'ez.^ 1 n Arabic Mehri Śheri Ge'ez dāka tāka. dēk. "this". 'ulālikà).). dnnazzih / dllazzih (plur. Other "far" demonstratives are formed with suffixes -lika (Classical Arabic: dalika.44. Arabic. Classical Arabic preserved a dual with a subject case and an oblique case. §36. "that". as za. ddkdmdh. which can change occasionally to -m. hlyk). The "near" demonstrative is expressed by za. 'ellayin 'ulāka. and they are also used with an additional suffix -n. thus reflecting the shift d > d (§13. ille.is used as a suffix in the Arabic colloquial of Egypt (dukha. The opposition d : a distinguishes the "near" demonstrative from the "far" demonstrative like in Gafat (§36. tīka dēk dayk dáku dikun zdk(t)u 'antaku. 36. "that". -kdmdh (Mehri. dikha. The demonstrative referring to an object near the addressee (second person) is the Amharic independent personal pro­ noun of the third person (§36. Aramaic Sing. as zak. 'ulā'ika dlyēk iíok 'dlhku The Aramaic demonstrative is written with i in the earlier periods. didbuk.8). §49. 'abmdh). dlbuk). or without. dnnàzziya I dllàzziya (plur.32. The demonstrative referring to an object near a third person is ya (masc). The element hā.9). comparable with Latin hie. iste. "f ^illēkī } '^n. Harsūsi: dākdmdh.40). dēkī j J ^ . -buk (Soqotri: dddbuk.43.). Christian Palestinian Aramaic: hdk. later with d. dukham I dukhumma.

Zu-ú4-la /DU-'ilaf) and in Ugaritic (du-ú). 9-10). Judg.8). 36.g. or in a relative sentence (e. ca. This means that its original North and East Semitic form was generally tu. The following paradigm contains the fully or partly inflected forms of the pronoun. lit. number.6). it acts as a pronominal or adjectival antecedent of a relative clause (§57. or Zu-Aš-tarti. accred­ ited the erroneous opinion that the determinative-relative pronoun of East Semitic is formally connected with the independent personal pro­ noun šū (§36. 1200 B. already noticeable in some variants at Ebla and in the post-Ur I I I texts from Mari.47.g. 36.g.g. as shown by names like Zu-ha-ad-ni /Du-ġadni/. 5. "the two horned". The determinative-relative pronoun was originally fully inflec­ tive. ŠÈ or ši.48. at Emar. with the well-known opposition of voiced and unvoiced consonants (§10.C. who has car­ ried out its repair": K T U 4. at Mari. The subsequent change tu > šu.145. but were lengthened in the course of time. In the first case. dū l-qarnayn. "the chief of the craftsmen. lit. The vowels of the singular were ini­ tially short. "The pleasant one". Further dif­ ferentiations are hindered in Epigraphic South Arabian languages because of the lack of vocalization. and partly in Amorite. in the second. dialectal Hebrew (e. . Determinative-Relative Pronouns 36.2) and the related demonstrative (§36.7). in Old Akkadian. The determinative-relative pronoun is written usually with the signs šu. but it became indeclinable without gender. and case differ­ entiation in practically all Semitic languages. The unvoiced form tu existed also in West Semitic and it is attested by the Phoenician. "the [man] of two horns". an epithet given in Arabic to Alexander the Great). while the voiced form du appears also in Amorite dialects and at Emar.18).46. The determinative-relative pronoun tu I du introduces a deter­ mination which can consist either in a noun or proper name (e. ŠA in Palaeosyrian at Ebla and at Tell Beydar. and Mishnaic Hebrew relative pronoun še-. "The (man) of Astarte". already in Amor­ ite (e. Ugaritic rb hršm d šs'a hwyh..324 MORPHOLOGY F. "The (man) of pleasure".36). paralleled in West and South Semitic by du. it functions in a genitival structure (§51.

nominative obi. dātay. case Dual fem.6). . Only in rare cases have šūt and šāt survived in the first centuries of the second millennium B. nominative genitive accusative Dual masc.DETERMINATIVE-RELATIVE PRONOUNS 325 Old Akkadian Classical & Palaeosyrian Arabic Sing. 'wlw tat(u) tāt(i) dawātu. as well as the feminine forms ši /ti/ and si-i /šī < til. 'ulātu 'ulāti dtw(l) 36. dn tā dātā. 'ulū 'ulī 'iy 'hl. dawāti. since its variant spelling is sa-ti [šati]. masc. masc. fem. case Plur. The Palaeosyrian determinative-relative pro­ noun is attested by the singular feminine oblique case sa-ti jtatil rather than Idatil. nominative obi. nominative obi. The Assyro-Babylonian determinative-relative pronoun appears from the end of the Old Akkadian period on under the indeclinable form ša of the accusative. dawātā dawātay dty dtyn tūt(u) tūt(i) dawū. case Sabaic Minaic Qatabanic tu ti ta dū dī dâ d. nominative obi. dl dtw.C. case Plur. . which is in reality the old citation form (§32. dawī. fem. the masculine forms Zu(-ú) /dū/ and šu /tu/ occur. In Amorite onomastics. nominative genitive accusative Sing. dw tat(u) tati *tata dātu dāti dāta dt dt dt tā dawā daway dy dy dw.49.

3839. "the (God) of Sinai". 'allatī (sing. while simple di continues to be used in Yemen and in some Maghrebine colloquials.). Ugaritic d appears sometimes in place of the expected dt. 'alladāni (dual masc).53. It is written z. 'allāti I 'allawāti (fern. Its old feminine z't (*dāt) is attested as demonstrative at Tell Fekherye (9th century B. and the modern colloquials either reduce it to ildī. while the masculine genitive *di > ze and the feminine *dat > *zat > zot > zo are employed as demonstratives (§36.). gender. the archaic theonym zū-Sīnay. and d with a very short vowel (da).38). The form dt stands for the feminine singular (*dāt-) and for the plural (*dūt-). while the function of determinative-relative was usually taken by še-lša< *ti. In Aramaic.6). since the determinative-relative dū was already indeclinable in the pre-classical poetry. exactly as in Phoenician. The pronoun zū is attested as indeclinable relative in poetry. which have used it also as an indeclinable alladi. Only the vocalization of d as du-ú (du) is provided. However.56). There are many variants of this pronoun in the ancient dialects. and case. as relative pronoun (§36.).38). Arabic devel­ oped an extended relative pronoun combining the deictic 'alia < *hanna with the determinative-relative: 'alladī (sing.). In Hebrew.C. preserves the nominative of the pronoun du used as determinative. 'allatāni (dual fern. "place". 36. often with a prefixed vowel ('š). In Ugaritic. iddl. illl. only d < d and dt < dt are attested. zy. Besides.51. the determinative-relative di in the genitive case is used in its original function and as element of demonstratives (§36. while those with š < t are employed as determinative-relative pronouns.326 MORPHOLOGY 36. dy. fern. plur. 'alladlna (masc. In Phoenician. The Classical fully declined Arabic pronoun reflects a system­ atic archaizing intervention. However. the forms with z < d are used as demonstratives (§36. directly from a base * 'allay.). 36. or derive allī. but they may be inflected according to number.50. 36. 42). masc). plur. either because the pronoun was reduced to a single form or because the final -t was dropped as in possi­ bly similar cases (§35. the attestations of se in Classi­ cal Hebrew are rather scarse owing to the widespread use of the noun 'âšer. .52. but one can assume the existence of a gen­ itive *dī and of an accusative *dā.

.for the relative pronoun. except Harari. e. (place) which you have given". Babylonian eqelšu. "(the house) wherever the king will order. prefixed to the verb of the relative clause. the word started soon to be employed with any qualified element. Then it was used in apposi­ tion to another noun designating a place. "he who broke"). 'alia (plur. šupru bēt sūtūni.25) and then to introduce a relative clause (§57. The Hasaean (§7. this noun was simply followed by a relative asyndetic clause: e. (place) which I have chosen"..55. use the element yd. Hebrew habbayit 'âšer bāriītl. zi-sdbdra. Semitic languages also use some nouns as relatival antecedents. 'anta (fem. bēt šarru iqbūni lillikū. But since the meaning of 'âšer was forgotten in Hebrew. zd is used in Tigrinya and zi in Harari (e. regardless of their original meaning. Ya' qob 'âšer bdhartīkā. as in other Semitic languages.g. which was used first to express a genitival relation (§51. the Tigre relatival antecedent is la-. "Jacob. 36. Instead.g. "place" (< 'atr-). dd like the Arabic Yemenite col­ loquials.41) feminine determinative-relative d't shows the use of alif as in Classical Arabic and in Old Aramaic z't (§36. they shall go". The best known is the construct state ašar of the Assyro-Babylonian noun asm. "the house (place) which I have build".which corresponds to the -tl of Arabic 'alla-tl. the Hebrew and Moabite 'âšer. or Sabaic hnmw and hl-mw. The Ge'ez relatival antecedent is za (masc). has to be explained as a palatalized la-.. Epigraphic South Arabian had most likely a fully inflected determinative-relative pronoun. an inter­ rogative and indefinite pronoun can also be used as a relatival antecedent (§36. "whatever". also the noun bēt. where the older forms of Arabic 'allaappear.. also with some rarer forms as Sabaic feminine t. The South Arabian determinative-relative antecedent is also employed in the sources as an indeclinable pronoun. e. 36.54. "write (the house) where he is". and the South Ethiopian languages.g.62). Mehri uses di.52).). Initially. e.56.9).DETERMINATIVE-RELATIVE PRONOUNS 327 36. "he forgot (the place) where he was bom". "house". "his field. In Modem South Arabian. ašar tattadnu.g. Assyro-Babylonian imtasi asar iwwaldu.. is used in this way.). In several Semitic languages. In Neo-Assyrian. This particle.g. a .

There is also an inflected Semitic interrogative 'ayyu. The interrogative pronouns go back to a common Afro-Asiatic element transmitted in ancient Egyptian as m.6). attested in Old Akkadian. "what?". and mi.) and manna (fern. mdnt (masc. Hebrew ('ayyē). the n of man changes into r (§17. as *mahna > manna. 36. Tuareg ma and mi correspond to the situation in North Semitic and in "Canaanite" lan­ guages.) and mdnta (fern. It is used as adjective in Assyro-Babylonian (e. Ara­ maic. ma-an-na. "who?".). The Old Canaanite form mi-ia. appears in several Amarna letters. and South Semitic.). followed by the gen- . and in the "Canaanite" languages.57. "who?". like Latin guis? and quid? The pronoun referring to animate subjects is character­ ized by two different morphemes: -an in East Semitic. certainly related to Ugaritic mn. "what?". §36. and ma-a. -ah(a) > -ā in the other Semitic languages. 36. "what?". in Hausa as mèe. min has generally replaced man and its vowel is often lengthened. as well as an extended form mi-ia-ti with the affix -i (cf. "Who is like I I ? " . "what?". Semitic languages provide here the only examples in which animate subjects are distinguished from inanimate subjects. "until which day?") and as pronoun in Classical Arabic. derived from the interrogative particle 'ay. "who?". Old Canaanite (/'ayyāmif). show neither case endings nor mimation. -iy(a) in Palaeosyrian. "what?". Interrogative and Indefinite Pronouns 36.5). "Who is his opponent?". are both attested in proper names. The pronoun referring to inanimate subjects is likewise marked by two distinct morphemes: -in in East Semitic and Ethiopic.58. there is an interrogative ma-an-na (EA 286. in Tuareg as ma. In colloquial Arabic. The archaic forms man and min. "what?".328 MORPHOLOGY G. "which?". Besides. Amorite. but they have a mascu­ line and a feminine form in Ge'ez: mannu (masc. l'ayyāmaf). In Arabic. In Amorite.59. The Palaeosyr­ ian pronoun ml appears at Ebla in the proper name Mi-kà-il. while mā is still used only in Yemen. "who?".g. but these pronouns had already become mannum and mlnum in the Old Akkadian period. Arabic. "where?".35). man and mā are uninfected. adi ayyim ūmim. Ugaritic (7y). Amorite (/'ayyaf. and at Mari in the pre-Sargonic Mí-ma-hir-sú. "what?". which can be explained in the light of Minaic mhn. "what?". "who?". Old Babylonian. Ugaritic. In some Gurage dialects. "which?". which is attested in Old Akkadian.

"this". "which thing?" Various reductions are attested. e.g. the feminine 'aydā.and s. or short forms as ds. in question". etáta 'ay . ay-ad.. The indefinite use of ay is attested also in Libyco-Berber where this pronominal base combines with demonstrative suffixes: e. wāš and wūš with substitution of w for original '. mā mā mh(n) mdnt mi 'dntay man man/r mu 'aynā ērii 'ayyiun) ay'y 'ayyiflt) 'ayi 'ayyānay yātu yitta. "who is.INTERROGATIVE AND INDEFINITE PRONOUNS 329 itive ('ayyu raġulin. and the plural 'aylēn. ēs and ās with monophthongization. "that". ay-inn. "which man?").. 'y is encountered up to now only as an indefinite pronoun ("any"). while it is employed in Syriac. man min mn mannu man man man m{ )a{ri) w What?" m ma min mlnu(m) Which?" ay ayyu ? mā mh.?" In Ugaritic and in Minaic. "this here" (§36. as ayš. ay-inna. Its use is widespread in Arabic colloquials with an ending -š which goes back to classical 'ayyu say'in. ay-ad-ddk. šenhu. The forms assumed by the interrogative pronouns in the various languages are as follows: "Who?" Old Egyptian Tuareg Old Akkadian Assyro-Babylonian Palaeosyrian Amorite Ugaritic Old Canaanite Phoenician Hebrew Aramaic Syriac Neo-Aramaic Classical Arabic Colloquial Arabic Minaic Ge'ez Tigre Tigrinya Amharic Gurage m mi man mannu{m) ml manna my miya my ml man man man. 36.g.37). "that. with the affix -nā. both as interrogative and as indefinite. mn manna m mā mā mā mānl mā.60.which often preserve the -n of say'in when they are used with the agglutinated pronoun -hu employed as copula.

"anything"). e. "did any foe rise against Baal?". . "whatsoever"). mdndm. min-ma > mimma. in Amharic. The indefinite pronoun can also be formed with the deictic particle -k added to the interrogative. "a certain man". "whatever"). More often a suffix is added to the inter­ rogative. in Arabic ('ayyumā. "nobody". is preserved in Western Neo-Aramaic monmi l-lt < monmi d-lt or mūnma l-lt. Arabic raġulun mā. as indefinite in the clause qnmy 't.61. "a certain". y . do not exist in Semitic. "whatever"). strictly speaking. mainly -ma used in Old Akkadian. e. 36. in Poenician and Punic (mnm.330 MORPHOLOGY F i g . The pronouns mn in Ugaritic and mā in Arabic may be placed in apposition to nouns with the meaning "any". in Ethiopian languages (e. "any­ one". in Assyro-Babylonian (man-ma > mamma. Oxford).g. "whoever". Amharic manndm. Phoenician uses the reflexive pronoun qnm-y. Gafat mandm. mdndm or mdnâ. Indefinite pronouns. "whoever". 27.g. māmā. "whoever he is". Aramaic ostracon from Elephantine. or by reduplication of the interrogative.g. Aramaic *manmi or *manma. "what­ ever". Tigre manma. "whatever". as mhk and mnk in Ugaritic. in Ugaritic (*mannama. as mamman < *man-man in Assyro-Babylonian and mdndmzn.28). The forms used as a kind of indefinite pronouns are based on the inter­ rogative pronoun. Ugaritic mn 'ib yp' Ib'l. in the very same phrase: mnm ib yp' Ib'l). C . reverse with lines 10-17 of a letter (Bodleian Library. "whoever you are". "whoever". lit. "he himself" (§36. "whoever". 5th century B .

and "passive". but both categories can be distinguished also in Libyco-Berber where they are represented. or "modal" elsewhere..VERBS 331 36. "cohortative" in a third one. South Arabian. for aspect. the traditional explanation of Semitic verbal forms is based on the conception of a triconsonantal discontinuous morph or root. This artificial approach cannot lead to an understanding of the Semitic verbal system which was originally characterized by trimorphous <2-class. CaCiC.g. and fashionable resorting to modern linguistic analyses of Indo-European tenses may lead to a misinterpretation of the basic characteristics of the Semitic verbal system. and w-class roots. §28. or C aC C (cf. V E R B S A . e. 2 4. independently from the formal distinction between tenses and aspects (§38. "subjunctive" in one language.1. C vC C or C C vC while the aspectual conjugation originates from a "nominal" base of the types CaC. Authors call corresponding verbal forms.. CvC. CaCuC.g. "whoever removes my name from the objects.1-14). derive from a "verbal" base of the types CvC. CāC. The interrogative pronouns can be used also as relatival antecedents in several Semitic languages. like the imperative and the jussive refer­ ring to futurity. for stem. the latter being further expanded to CaCaC. and the preterite expressing a genuine past. as the distinction between the categories of tran­ sitive and intransitive.7-12). "nobody knew what was his malady". which is unpronounceable and did never exist in a spoken language. The problems raised by the verb are among the most difficult in Semitic linguistics and the varying terminology used in grammatical studies bearing on the single languages does not help in solving them. and between transi­ tive and the intransitive conjugations (§38. Sabaic 7 mn s 'r k-mhn h' hlthw. broadly speaking. and for voice. for mood. as Aramaic. for actor. and the Semitic aspects of action. of active (event) and stative (state). Aramaic mn yld šmy mn m'ny'. Arabic. Preliminaries 37. Also occa­ sional confusions between "stative". /-class.. The primitive tenses. e.15-17).". Lihyānite. "jussive" in another one. The verb is the grammatical category which inflects for tense. Important shifts from one group to the other occurred in the course of time. considered either in a synchronic or in a diachronic perspective. Furthermore. like in ergative lan­ guages.62. "intransitive". by { 2 2 { 2 y x 2 3 .

3. The types CvC. are. . Forms of the most ancient Semitic languages. As for the basic patterns of Semitic aspectual forms. the pattern CjC vC will be followed in the general presentation.and suffix-conjugation. it serves to express formally distinguished aspects which cannot be equated with telic and atelic situ­ ations. This bipartite distinction is the normal one in a wide range of languages. while the patterns CvC. a telic situation involves a process that leads up to a welldefined terminal point (e. the "telic/atelic" distinction is neutralized in the Semitic imperfective aspect. and "future" is not a universal characteristic of temporal systems. while the type CjC vC corresponds to those which are triconsonantal. In other words. on the one hand. the distinction between a future action and a past action. "he killed". e. while an atelic situation lacks such a determinate goal (e. The essential function of the "nominal" base of verbal forms is to indicate a condition or a situation with respect to circumstances. 37. and CjVC C occur with bicon­ sonantal roots. as a rule.332 MORPHOLOGY the proper verbal conjugation. "to separate". it serves to form a kind of tenses which tell us something about the relative order of events. "he made". In other words. 2 2 2 3 2 3 2 2 37. we shall occasionally refer to Semitic roots or bases by designating them by the sole consonantal signs. CvC. Since the last group is more dominant in historically attested verbs than in any other part of speech. and CjvC C will be examined in a complementary section (§44). while the verbfa'ala. viz. and by the so-called "qualitative" which derives from a nominal base. while the semantic feature of "accomplishment" replaces the "well-defined terminal point" in the perfective aspect.g. Both types of conjugation occur in other Afro-Asiatic languages as well. In fact. "John is making a chair"). and unaccomplished or not completely performed. attested in cuneiform script.2. the types CaC. for the other languages. The essential function of the "verbal" base is to express. "John is singing"). is generally used for Arabic. since the familiar tripartite division of time in "past". on the other. designated by the paradigmatic verb parāsu. whether some event took place before the real or fictitious time of speaking or had not yet taken place at that moment. and qatal. for example in the "Sam" sub-group of Lowland East Cushitic (§2. CvC. It means that the "telic/atelic" distinction is of no use in the analysis of Semitic aspects. in grammatical categories. accomplished or perfected.g. Now.11) with its prefix. permanent or static.g. — To avoid tedious repetitions or unwarranted hypotheses. "present".

37. 3° a West Semitic subjunctive marked by -a. jussive. to the category of nouns.2).4.14). In fact. The actor affixes or personals specify person. and corre­ spond to well-known patterns of verbal adjectives. are thus purely grammatical-syntactic categories of coor­ dination and subordination. and C vC C are used for the biconsonantal bases. and called "cohortative" in grammars of Hebrew (§39. and qatal.7) and which is expressed by the indicative with a suffixed "direc­ tive" morpheme. the imperative. The imperative. Despite some difficulties. and is called "apocopate(d)" in grammars of Arabic (§39. while the other types will be examined in §44. notifies a fact or what is alleged to be a fact.5-6). Both are probably related to the East Semitic ventive or allative mood which signifies a general movement of the action towards the speaker (§39. having a final/consecutive function. there are three basic moods in Semitic: 1° an unmarked jussive which derives from the imperative (§38. l 2 2 2 3 37. and the participle are not considered as "modal" forms.3).8-11). subjunctive. Besides the indicative.6. broadly speaking. ventive or allative. gender. unmarked or marked by affixes (§39). while the infinitive and the participle belong. fa 'ala. corre­ sponds to the Indo-European injunctive. They appear as both prefixes and suffixes.5. it serves as the base of all the inflectional forms connected with a specific meaning. The moods of Semitic verbs. energic or energetic. 2° an East Semitic subjunctive marked by -u and used in all kinds of subordinate clauses (§39.VERBS 333 CāC. denotes the manner in which the action or state is expressed. a clear connection can be traced between the personals of the verb and the com­ ponents of the personal pronoun. while its suffixed variant in -n(na) is termed "energetic" or "energic" (§39. The indicative. etc. which is used essentially for statements expressed in main clauses. Their distribution depends on tense and aspect (§40). The stem is a verbal pattern deriving from a root. the infinitive. The general presenta­ tion of the Outline will be referring to the triconsonantal pattern supplied by parāsu.. The category of mood. strictly speaking. Each verbal root has a simple or basic stem — not always used in his­ torically attested languages — and a varying number of derived stems. . which denotes a command. as its name implies. 37. called indicative. is simply a base-form. while the type CjaC C and its derivatives serve for the triconsonantal ones. and num­ ber.

East Semitic B/G-stem D-stem (L-stem) Š-stem ŠD-stem Hebrew pa'al/qal pi'el Aramaic pe'al pa"el Arabic Stem I Stem II Stem III Stem IV Ethiopic Stem I.g.l/A Stem II.g. While some of these stems have an Afro-Asiatic origin. §38.2/B IV.18). 1/A IV. in many cases — also in the present Outline — the only available means of translating Semitic aspects in an intelligible manner (cf. 1.3/C • StemII.3/C — — hif'il — N-stem B/Gt-stem Dt-stem (tL-stem) Št-stem ŠDt-stem — — — haf'el nif'al hitpa'el — — — — — — itpe'el itpa"al ittaf'al — — — — — — — Stem VII Stem VIII Stem V Stem VI Stem X — Stem Stem Stem Stem Stem Stem III.3/C IV. Arabic. the so-called present-future (iparras) of the grammars of Akkadian. e.7. the Semitic verbal system underwent sev­ eral important changes. Hebrew.. is far from being common to any type of Semitic verbal system. The symbols reported for Ethiopic. Thus.334 MORPHOLOGY In some languages. Only a diachronic and comparative method can insert these changes in the wider context of the whole Semitic system. a morphosyntactic approach is needed here. The terminology and the symbols used in the grammars to denote the various stems are not identical for all the Semitic languages.1 and IA. By "voice" we intend these passive and active forms. Tense. take two different usages into account. Aramaic. which refer both to the same forms. in the proper sense of the word. 37. showing vocalic differentiation. 1/A III. noticeable either in a whole group of languages or in a particular idiom.2/B Stem I. In order to illustrate the changes occurring in the verbal system.l/A Stem I. l / A " .2/B Stem II. the stem can exhibit an active and a passive theme. .8.2/B III. and Ethiopic. either with an affix or with lengthening. viz. e.3/C — — 37. An exlusively morphological presen­ tation is insufficient to explain the changes involved. also in the derived stems which are formed from the root or from another stem. The symbols will be explained in §41. In the course of time. Therefore. it is useful to present here a synoptic table of the main customary terms and symbols employed for East Semitic. "Stem I . other seem to result from an internal evolution of a particular group of Semitic languages. while European tense forms are.

The Semitic perfect may. B. The imperative stands outside any possible system of aspect. the preterite. That does not preclude the fact that most grammatical categories have a variety of meanings and thus may be ambiguous as long as a con­ crete context or life situation do not remove the ambiguity. be rendered by a past tense of verbs which express action or by the present of verbs which express a state. and certain modern innovations. Paired with the impera­ tive. Semitic languages have no tenses properly speaking. Therefore. a present. indicating basically that a process has not reached completion at a certain moment of time. in the Arabic colloquial of Damascus.. it refers by def­ inition to futurity. as yet not performed action.1. e. the misuse of the translations in attributing. This formation can be translated in European languages by a present. to a future. is clear in itself. the aspectual and the temporal. a future. viz. i. It is of the highest importance to distin­ guish these approximations from the real functions of Semitic verbal cat­ egories.TENSES AND ASPECTS 335 really is an imperfective aspect. an imperfect. exhibiting the same personal suffixes. outside the system of aspect. a future. in complex sentences. ktob. A grammarian is not called upon to explain a function i f it does not exist in a given language as a distinct form. It may also indicate that an activity is in progress or a state is being entered upon under the influence of another activity or state. Often however. a future perfect. but referring to the past and signifying that an action has or has not been performed.g. but exact translation from the one to the other is very dif­ ficult. Each of the systems. there is another purely verbal form. i f not impossible. Tenses and Aspects a) Fully Developed System 38. but only aspects. for practical reasons. e. In the basic triconsonantal model C C vC the simplest form morphologically is placed in the most unmarked category which is the imperative across the whole spread of Afro-Asiatic languages. "write!". It is characterized morphologically by the same basic model C. Except for the imperative. but having besides personal prefixes. also by European subjunctives and other categories.C vC as the imperative. the Assyro-Babylonian preterite.g. x 2 v 2 3 . or a future perfect will be needed in translating. the functions of a future perfect or of a continuous pre­ sent to the Assyro-Babylonian iparras results in a complete misunder­ standing of the Semitic verbal system.e.

which was phonemic (§25. so that he ran away". Harsusi ydlbed. yíprus. "don't play!" (plur. 38.g.). ayiprus.8). expressing a genuine past. The use of the jussive. and as subordinate past tense in ydsla ydrwdl (80-70 msec). and the vetitive or prohibitive. "may he separate". in ydrwdl used as jussive in ydsla ydrwdl (60-90 msec). Subsequent changes intro­ duced special prefixes aimed at characterizing the cohortative. "you shall not k i l l " . and it is not used in negative clauses. These cohortative (1st pers.g. e.). which continued to be used by the West Semitic languages in a narrative context until the mid-first millennium B. Soqotri liqbar.g. Maghrebine ma tal'âbuš. "don't cry!" (fem. "he heard that he had run away". Rendille y-igis. or not at all. "may he hit".g. by the Assyro-Babylonian verbal adjective pars(um). This form is attested also in Libyco-Berber. "may he write".2. optative or precative (3rd pers. the opta­ tive or precative. "may he bury".10-11). represented e. called also "subjunctive". Aramaic Ihwy.g. e. well-known in Old Akkadian and in Assyro-Babylonian: CaCiC { 2 3 . "may he be".g. lā taqtul. "he killed".3.g. yáqtul). was extended to various subordinate verbal clauses. liprus. the simplest form of which would have been identi­ cal with the preterite i f there were no differences in the stress. Mehri ydkteb. "he / she separated". e. Arabic li-yaktub. despite the changes that the verbal system had then undergone (§38. "he heard.). yaqtúî).). This is confirmed by the stress of the prohibitive in modern Arabic colloquials and of the so-called subjunctive in Modern South Arabian. Therefore. "he knotted". (y)iprus / taprus. Tuareg yd-krds. 38. as shown by the quantitative vowel gradation. and beyond.336 MORPHOLOGY e. These proclitics express the expectation on the part of the speaker or active subject that the process w i l l indeed take place. "may he not separate". especially in Ethiopic. and vetitive or prohibitive functions are assumed by the so-called "jussive". "he should write". The very limited evidence provided by the Kabyle dialect of Libyco-Berber seems to go in the same direction.g. Mesopotamian lā tdbkáy. "may I separate". e.g. one has to surmise that the stress rested in the preterite on the prefix (e. Assyro-Babylonian luprus.C. The aspectual category of the verbal system is based on the adjectival C aC C pattern. Semitic imperative has no first and third persons. It is a narration oriented form. e. yiprús. while it rested in the jus­ sive on the basic syllable of the imperative (e. It occurs in Cushitic as well. and developed to the stative / permansive forms.

"you are content"). "he is wide").e. CaCuC are disrupted in many Arabic colloquials following the elision of short vowels in open unstressed syl­ lables or their qualitative change occasioned by vowel-harmony. the basis is either CaCaC or CdCdC. while it is CiCaC or . which express the accom­ plished and the unaccomplished aspect of the action signified by the verb. e. CaCaC. the quality of the inserted vowel is not predictable (e. the completed (per­ fect) and the incomplete (present-future) aspect of the action. This leaves us with one basically unmarked category. "he is deaf".g. in the East and especially in the West.g. These marked categories. i. The Afro-Asi­ atic origin of this morphological category C aC (v)C is demonstrated by the Egyptian old perfective. the so-called "qualitative".g. "he is walking" denotes a continuous process. Thus in Mesopotamian ^p/íw-dialects (qdltu. Assyro-Baby­ lonian ilmad. which is a stative or permansive form. Maghrebine katbat. "he is learned"). while "he is striking" connotes a series of actions. which may function either as transitive or as intransitive forms. Assyro-Babylonian rapaš. "she is alike". but lamid. hrt/d. many Arab vernaculars. and with two marked categories.g. and the latter is not identical with the thematic vowel of the imperative-preterite (e. also called "pseudo-participle" (e.g. masc. "she wrote"..19). "he is wide"). but its use was extended in several standard languages. e. in the 3rd pers.g. influence of contiguous consonants. vowel-opposi­ tion. Classical Arabic malakat. damiq. "he is good"). have forms without anaptyxis. Ge'ez masalat. sing. The originally static aspect of this morphological category is opposed to the perfective and to the imperfective which express. "she is the proprietress". Despite relative distinctions made between an active pattern exhibiting a and stative patterns exhibiting / > e or u > o. Neither does it mat­ ter whether the action tends to a determinate goal ("teli'c) or is simply considered as durative ("atelic"). are conveniently termed "perfective" and "imperfective".TENSES AND ASPECTS 337 (e. qarub. "he learned". rapaš. They express a fundamental contrast between an event which is in the process of transpiring ("he is walking") and one that has already taken place and exists only as a resultant state or condition ("he is in a condition subsequent to walking"). It does not matter whether the process is continuous or repetitive.g. Tuareg mâzâg or rndzdg. The anaptyctic vowel a I i I u should appear only when there is a two-consonant cluster (§27. in grammatical categories.g. l 2 3 The morphological categories CaCiC. Instead. " I said"). CaCuC (e.g. Damascene katbet. for this distinction results from the meaning of the verb.g. CaCaC (e. e. and by the Libyco-Berber suffix conjugation of stative verbs which express a quality. "he is near"). e.

However. whereas 'imths ksp with a direct object means " I have seized / laid hands on silver".g.. and in Ugaritic. perfective originally conveys involvement of the acting subject.4. has given to P N " . This aspect-deriva­ tional gemination signifies actuation in reference to the action. whether it be caused by another or by himself / itself. Vistbm tnn. e.111. there is no formal difference in East and North Semitic between the preterite of the Gt-stem (§41. and obviously uses the per­ fective of the basic stem.111.g. Similarly.40). In other words. Thus. The perfective is formed with the r-infix. but when the same verb has a reciprocal connotation. "they fought". The same analysis can be made in Ugaritic.g. the basic function of which can be characterized as "effective" in the sense that a state is pro­ duced in someone or in something.g.5. " I have seized silver. " I myself have gone" ( K T U 1. "he has chosen us"). Further synchronic and diachronic studies of verbal forms with infixed are needed. like in Old Assyrian qātka itntahas. the form is an Assyro-Babylonian perfect. This definition implies a functional congruence between the aspect-derivational -t. but the perfective in P N kūm P N PNj ana P N ittadin implies that "PNj has taken upon himself to give P N instead of P N to P N " . notwithstanding their distinct structural planes. ia-ab-ta-haar-na /yabtaharna/. (y)iptaqid. 'an 'itlk. while preterite marks the simple past.20) and the perfective of the basic stem. the distinction can be made on a contextual and syntagmatic basis. the form can only be a preterite of the Gt-stem. as in the Old Akkadian dual imtahsā. contains a preterite of the Gt-stem. E. and CiCaC in East Arabian. "he has heard").3.'mq. e. The base form shows the types CaCaC.and the stem-derivational -t(§41.338 MORPHOLOGY CuCaC in the Mesopotamian #a/ar-dialects (gdfot.11. "he has separated". while the .20 ff.g. CiCiC. " I said"). showing a shift from original pattern C vC C to C C vC 1 2 3 1 2 y 38. when the verb mahāsu governs a direct object. "didn't I have muzzled the Dragon?" ( K T U 1.). viz. in Amorite (e. "he has struck your hand". The perfective is represented by the Assyro-Babylonian perfect (y)iptaras. 46-47).15). the Neo-Babylonian preterite in PNj ana P N iddin plainly means that "PN. iš-tá-má /yištama'/.6. by gemi­ nating the second consonant of a triconsonantal root. acquired gold" ( K T U 1.3. "he has rushed". CCiC. where 'nt tmths b. "'Anat fought in the valley". or CCdC. It is also attested in Palaeosyrian (e. and not simply "has have it sent". išpur. (y)irtapud. 38. while the Maghre­ bine basis is CCaC. The imperfective is formed by a lengthened root. 'imths ksp 'itrt hrs. 2 f f 2 3 4 2 f f 3 4 2 As a rule. "he has com­ manded". mār sarri šipirti iltapra umma means that "the king's son has issued the following writ".

but in modern Ethiopian languages as well.. "he learns"). More­ over. The imperfective is represented by the Old Akkadian and Assyro-Babylonian "present-future" iparras. with the m-prefix of the first person singular like in Maghrebine Arabic (§40.44.3). which influences the spelling of syllabic Ugaritian i-le-qa-aš-šu-nu-ti. Besides. "he stands upright". its existence has to be assumed in Epigraphic South Arabian. from imperfectives like y'ahd (*[yahhud\) (KTU 4. a form which occurs not only in Ge'ez. "you will bind". The source of such anomalous imperfective forms is to be looked for in South Arabian dialects brought to Spain in the 8th cen­ tury A.6. you desire". ydlāmmdd. geminated imperfectives of the basic stem appear in early Andalusian Arabic. "he/she took". in Amorite. as reported by Pedro de Alcalá: nihammí [nihammī] instead of 'ahmī. ilāmmdd.3: 2°).28). Harsūsi ydlobdd < *yalabbad. "El will protect". and in Ugaritic. since the imperfect of the Modern South Arabian goes appar­ ently back to such a pattern (cf. "he shoots". "they will come near". or irāmmdd with the change I > r. The avail­ able texts distinguish preterite forms like y'ihd (*[yīhud]) and t'ihd (*[tīhud\). while the differences noticeable in Ugaritic verbs with first radical ' point to the use of a yaqattul (iparras) next to the yaqtul (iprus) and to the yuqattil (uparris) (§41. §38.TENSES AND ASPECTS 339 stem-derivational gemination expresses actuation in reference to the actor (§41. akantib < *akattib. This imperfective is attested most likely also in Palaeosyrian. 38. has a form i-taha-à /yitahhawū/. "he takes". A Palaeosyrian incantation text from Ebla. This imperfective must have an Afro-Asiatic origin since its close cognate appears in Libyco-Berber (e. e. by "Yemenite" tribesmen. " I am writing"). Mehri yarokdz < *yarakkaz. The frequent reference to the Ugaritic form yqh ("may he take"). The imperfective under consideration is documented in North Semitic. ipaqqid. and Amorite proper names like Ia-ma-at-ti-El /Yamatti'-'El/.g. "he/she held fast". e. "he will take them". "he will k i l l " . irappud. or t'asrn (*[tassirūn]). nixehéd [nìšehhed] instead of 'ašhad(u). can hardly be explained without admit­ ting the use of a yaqattil form.8).25). or y'arš (*[yarris\) and farš (*[tarris]). "El will name".g. instead of normal Babylonian ilaqqē-šunūti. and by the Ethiopic ydqattdl. One should refer also to fairly contemporaneous imperfectives from Emar which are clearly influenced . or Ia-na-ab-bi-El /Yanabbi'-El/. fails distinguishing between a jussive yiqqah and an imperfective *yilaqqah.g. "he desires. allegedly incompatible with an imperfective corresponding to East Semitic iparras.g. "he kills". " I cer­ tify". while the singular forms of the Bedja present seem to contain a dissimilated double consonant (e. or y'uhd (*[yuhhud]) and t'uhd (*[tuhhud]). " I protect".D.

the stative / permansive differs from the perfective and from the imperfective by the distribution of the actor affixes: they are all suffixed to the stative. e-ez-zi-ib-ka /'e'ezzibka/. The purely "verbal" and aspectual categories were somewhat reduced in several Semitic languages.. while the non-gemination is compensated in Mehri and Harsūsi by the lengthening of the preceding vowel (§38. one should say that the evidence points to an original imperfect ydqattdl / yaqattal in South Arabian. Besides the i-infix and the lengthening of the root. closer to Proto-Ethiopic than the South Ethiopic group. vs. tu-uš-ša-ab. yàdakdm. Therefore. and occasional feminine forms like te-er-ru-ub.6). Tigre. Gafat yddakkam. ta-lak. on the whole. However. East Gurage dialects occasionally show the gemination of the second radical. "she goes". "she enters". " I shall dismiss you". "when . like East Gurage and Harari.. In summing up the situa­ tion. "he takes prisoner". ta-al-la-ak. "he speaks"..g. while they appear as both prefixes and suffixes with the perfective and the imperfective. while changes in function and meaning also occurred. the South Ethiopic imperfect of Stems 1/A has very likely lost its gemination. this consonant is generally not geminated in South Ethiopic which pre­ sents a pattern ydqàtdl. e. §41. so much the more so that North Ethiopic is.. instead of Middle Babylonian ezzibka. "she will stay". "may he speak". the South Ethiopic imperfect of Stem I. and Tigrinya — certainly uses the ydqattdl form with gemination of the second radical consonant. The perfective with f-infix is attested in East and . despite the fact that some of these idioms. However. 38.4. b) Simplified Systems 38. contrary to the jussive where the gemination was not required by the system. The Modern South Arabian languages do not have the second radical consonant geminated in the imperfect.8. "may he take prisoner". ydmark.340 MORPHOLOGY by North Semitic.g. vs.7. Amharic ydmarrdk.10. 38.C/3 generally preserves the gemination of the second radical. Sheri and Soqotri are again languages of the non-geminating type (cf. While North Ethiopic — Ge'ez.5). Moreover.9. or ki-i-mee . as well as in Ethiopic. she will go". 38. are precisely languages of the nongeminating type. although the close structural analogy between gemination and vowel lengthening allows of an allophone ydqātdl / yaqātal. e.

as it seems. Although the old preterite was the regular narrative form in myths and epics. "and I made this high place"). e. no certain examples of the stative used as a real perfect before the later half of the second millennium B. 40. e. And the difficulty of distinguishing past action and completion led to the dis­ appearance of one of these categories. but it was supplanted in West Semitic by the preterite yiqtul I yaqtul. with both band and danna in the predicate state (§33. The perfec­ tive with i-infix has obviously appeared as redundant. There are.g. "messenger".g. which can be explained in a different way. in the 15th-13th centuries B.5. and even Ma-laak-ì-lí can be interpreted either as a stative malak. Arabic malaka. "(if) PNj (the pawned slave) will have fallen sick (or) will have escaped.g. The situation in Amorite is similar. or as a noun mal'āk. This last evolution was certainly on its way at the time of the Amarna correspon­ dence and of the Ugaritic texts. "to be master"). All Palaeosyrian cases cited refer to alleged perfect tenses of verbs ultimae or mediae infirmae. ma-hi-ir is often used in the sense "has received".BAR mandattašu PN ušallam. in Phoenician (e.C. since it referred to an action already accomplished and thus belonging to the past. "and Baalshamayn answered me"). PN shall pay four litre of barley a day as his clearing". "he seized". "and". at least sporadically (e. 2 2 38.g. 'atwt. w-y'nny b'lšmyn. only some Late Babylonian forms influenced by Aramaic may be considered as perfects. Ba-na-a-hu means "the brother is nice" rather than "the (divine) brother has created". at least sporadically.11. w-yp'l b-hlb [šl]m. The old preterite preceded by wa-. in Hebrew (e. expressed by the preterite. "and he made [pea]ce in . and Da-na-LUGAL means "the king is powerful" rather than "the king has judged". kept on acting as a narrative past tense. in Moabite (e. as some authors believe. PNj ma-ra-su hal-liq ūmu 4 SÌLA ŠE. "she came").TENSES AND ASPECTS 341 North Semitic.C. way-yo'mer. The question is whether this change goes back to Amorite or even to Palaeosyrian. which in turn was superseded by the stative that acquired the meaning of a perfect without losing its original function. "and he said"). 'ahd.g. As for East Semitic. In fact.3). This led to a further simplification of the verbal system. The second change was prompted by the parallelism between a situation existing at a determined moment and the situation resulting from an event anterior to the moment in ques­ tion.g. in Aramaic (e. there is some parallelism between the "aspectual" notion of perfective and the "temporal" notion of past.g. "my god is king" (cf. In Late Babylonian contracts. w-"š h-bmt z't. the stative appears already as a real perfect.

and said"). which dis­ tinguish an imperfect for the main affirmative clause and an imperfect for the subordinate clause.g. use a form of the ydqat{td)lu-iyx)Q in the main clause . ydsàbru.g. the North Gurage dialects. but tisàbdr < *tdydsabdr. be-lí i-ma-ar-ru-šu. "he struck her . 38. w-yhtb mwy dhbhw. It certainly occurs with the preterite (e. "and he restored the water-supply of his alluvial land").g. developed from the jussive yaqtúl (§38. although an unconvincing attempt was made to discover it in Hebrew on the basis of the orthography of the Qumrān scrolls. in clauses where formal parataxis expressed log­ ical hypotaxis (§55. The development must have occurred first in spoken dialects.. This aspectual form was replaced by a new indicative yaqtulu. ú-ul aš-ku-un-nu. wa-yaqūlu. and biblical Dead Sea scrolls occasionally substitute a suffixed form belonging to the contemporaneous spoken idiom.2-3). in South Arabian (e.2) by adding an -u. "why do you keep so silent that you do not say to the king that he should send the bowmen?" (EA 71. as suggested by the Oboda inscription: fa-yafal lā fidā wa-lā 'atarā (p-yp'l V pd' w-V 'tr'). and in Arabic (e.. A R M I I . which cannot be completely independent from the -u of the Palaeosyrian and East Semitic subjunctive (§39. we have evidence of the imperfective . including the Arabic vernaculars. and was already affixed at Mari to preterite and imperfective forms without apparent change in their mean­ ing and their function (§38.13). darabahā . "and he acted neither for reward nor for favour".g. This narrative past tense with the so-called "converted" imperfect was not used any more in Mishnaic Hebrew. The imperfective iparras I ydqattdl is not attested until now in West Semitic languages.23-26) but it most likely reflects occasional lapses of the scribes into their native Amorite idiom in which the -u suffix must have taken root (cf. This final vowel -u is preserved in Classical Arabic. " I did not assign") and with the present-future (e..14). e. "my lord will see i t " ) .1-8). However. while its shedding is quite general in the other idioms.10-14).g. the old preterite is probably used also after fa-. "he breaks" or "he shall break". thus justifying the hypercorrect use of a sub­ junctive marker.13. 136. 38..g. "when he breaks".12. §38. as examplified by the following sentence of a Byblos letter: a-na mi-nim qa-la-ta (stative) ù la-a ti-iq-bu (jussive + u) a-na šàr-ri ù yu-wa-ši-ru-na (jussive + u + energic n) sāba pí-tá-ti. In Nabataean Arabic. I f so. The final -u added sometimes at Mari to verbs of main clauses may result from an hypercorrect use of the subjunctive (e.342 MORPHOLOGY Aleppo").

"her heart was filled with joy"). Besides. forms in -u are perhaps unveiled by Ugaritic literary texts where the final radical 'u might characterize the singular preterite yaqtulu (e. Besides. The morpheme -u of the indicative may therefore be considered as a simple generalization of the -u suffix of subordinate clauses (§38. ki 38. The bulk of the material is provided by Arabic that uses the indicative imper­ fect yaqtulu also in all kinds of subordinate clauses. and it seems therefore to have been an optional literary feature of the Ugaritic idiom. ma-ti -ia u-ki-in-nu.TENSES AND ASPECTS 343 yaqattalu and of the preterite yaqtulu in North Semitic.7th centuries A. I did guard the king's city") and of the presentfuture (anumma inassaru âla sa šarri.C.4-6). with Hidjazis using the indica­ tive in -u where others employ the apocopate (cf.3). except those that are final/consecutive or conditional. The examples from Mari and from Alalakh. stands simply for *yimlū I *yamlū (cf. it has to be examined in the paragraphs dealing with moods (§39. testify nevertheless to the use of -u forms in main clauses.25-26).14. However. As for the origin of the subjunctive in -u and of the Arabic subjunctive in -a. and by the Gurage dialects. "now.D.g. A question related to aspects is the existence of certain mor­ phological distinctions in the conjugation of active and stative. the situation in ancient Arabic dialects of the 6th . unless yml'u.g. this ending -u appears neither in proper names (e.15.g. I shall guard the city of the king") in the same letter from Megiddo (EA 220. corroborated by the Amarna correspondence. 39.g. "now. a distinctive -u sub­ junctive had become superfluous. e. Further evidence is provided by Idrimi's autobiographical inscription from Alalakh (e. suggests that the use of the marked indicative was dialectal or optional before the systematization introduced by Arab grammarians. §45..8). since its role was assumed by subor­ dinate conjunctions that have developed in the course of time. of transi­ tive and intransitive verbs.12. yml'u Ibh bšmht. This explanation is supported by the structural changes which had occurred in Semitic already in the third millennium B.. especially in their basic stem since the mean­ ing of the derived stems generally obliterates these fundamental .15-16. By that time. Ia-qub-Ba'al /Ya'qub-Ba'al/. c) Transitivity — Intransitivity 38.14-18). §39. " I gave strength to my country") and by the Amarna correspondence with parallel examples of the preterite (anumma issuru āì šarri. "Baal has protected") nor in syllabic texts from Ugarit. by Classical Arabic.

In East Semitic and in West Semitic. "to drink" ydlbas. yanqud. e.g. The proper meaning of a transitive verb is "to per­ form an action" directly affecting another person or thing (e. In other cases. "to be cut off". iqrab next to iqrib.g. the "Sam" imperfect is added for comparison.g. "to come near" yiqrib. In conclusion. "to mn" yiktub.g. while the 0-class consists of mainly intransitive verbs. there are Ethiopic verbs that are inflected according to two patterns. "to write" yanqud. "to separate". "to become great". "to value" ydngsr. The Semitic jussive is the ver­ bal form which exhibits these differences at best. "to save" yamut. "to save".16. a change of class implies a modification in the meaning of the verb. yafsul. the stem . "to learn" yilbaš. Arabic yadbaġ. as Migāma. passive yabtar. the /-class and w-class verbs came together in the mainly transitive 2-class. This class distinction goes probably back to an Afro-Asiatic scheme. "to speak" yagis. "to tan". or expresses the state of being in a certain condition (e. "to go away". "to be saved". but this distinc­ tion is not absolute and dialectal variants exhibit. viz. the /-class. since it is attested also in Cushitic. Arabic yafsil. always inflect like û-class verbs. yadbuġ. "to k i l l " ) . "to dress" yašrab. especially in Rendille. islam next to islim in Assyro-Babylonian. as in Arabic yabtur. "to eat" /-Class iqrib. e. "to depart". These class alternations must be distinguished from vocalic modifications resulting from the use of the passive voice. "to cut". However. however. while intransitive verbs like Arabic yadhab. Some verbs may be inflected according to two or even three classes without any semantic differences. and the w-class. most stative and intransi­ tive verbs belong either to the /-class or to the w-class.. "to kill" M-Class irpud. passive yanqad. whereas the intransitive verb either signifies an action which is complete in itself and affects the subject (e. "to come near" yahsib. The dis­ tinction of transitive and intransitive verbs is apparently somewhat clearer in Ethiopic. where the statistical data are inverted.g. in the prefix-conjugation of the "Sam" sub-group of Lowland East Cushitic (§2. "to die" 38. Authors often assume that this kind of distinction is indicated by the quality of the stem vowel which divides the Semitic verbs into three classes: the aclass. "to lie"). Aramaic Arabic Ge'ez "Sam" ilmad.11). a-Class Ass. e. or Hebrew yigdal. e. "to be pleased"). The same situation occurs in some Chadic languages. "to dress" yaham.-Bab.g. yadbiġ.g.344 MORPHOLOGY semantic differences.

and imdl u-fruh. Tachelhit imdl u-rgaz a-fruh. In consequence.10) resulted in the course of the first millennium B. Therefore. e. especially in Libyco-Berber. the mor­ phological distinction between transitive and intransitive verbs as such is no more essential in Semitic. There was no return to the basic "ergative" opposi­ tion of transitive to intransitive.8). and Semitic belonged originally to this linguistic type. . as it seems: it is radicated in the C C vC model of the root morphemes (§28.1-4). the originally ergative character of Semitic was reduced mainly to the opposition of the active and non-active nominal components of the sentence (§32. Similar cases occur also in other Afro-Asiatic languages. l 2 3 38. used with a subject in the non-active a-case. Nevertheless. there is no doubt that the categories of transitive and intransitive are extremely important in any ergative language. Its origin is not functional.16). "the man came near". 38. "the man buried the child". Besides. but to any moment fixed in the utterance. although it introduces a relative morphological distinction between stative and active verbs. this dis­ tinction was based mainly on the intransitive function of the basic stem (B/G). These categories imply a reference not to the absolute moment of speaking. It is the accomplished or unaccomplished aspect signified by the verbal form that is relevant. the basic stem of the historically attested languages contain exclusively transitive verbs as well. The major distinction of category between the new perfect and the imperfect can be seen simply in terms of the aspectual contrast "accomplished" (perfect forms) / "unaccom­ plished" (imperfect forms). However.g.17. Arabic qariba r-ragula. and qariba r-raġulu.TENSES AND ASPECTS 345 vowel is neither predictable nor does it allow distinguishing transitive and intransitive verbs. used with a sub­ ject in the ergative «-case (§40.g. e.18. since the predicate of a nominal clause — thus lacking any verbal form — may be inflected in -a under certain conditions like the object of a transitive verb. "the child was buried". and the transitive function of the causative-factitive stems (D and/or S). But this distinction has no direct bearing on the transitive or intransitive nature of active verbs.C. a semantic development took place in an early phase of Proto-Semitic or even in Afro-Asiatic with the result that the basic stem of numerous Semitic verbs can be used both transitively and intransitively. "he approached the man". in a new scheme with two main morphological categories of the indicative: per­ fect and imperfect. In Semitic. Further changes in the West and South Semitic verbal system (§38.

or lay sa 'ahadun 'afqara min ġaniyyin 'amina l-faqra. being able to do. fa-tālati š-šakwa wa-huwa yabkī 'aharra bukā'in. "he had already seen through you just as we have seen".3). as Arabic 'alimtu. fa-lammā qadima l-Hazraġīyūna. "and the complaint took a long time.g. "and the Moslems began working". 38. having power.20. The imperfect is used naturally in clauses express­ ing finality and after verbs that denote setting in. Besides. " I know". e.. etc. I f we now turn to the tense formations which have been developed in some modern Semitic languages to express time relations in imitation of the western Indo-European tense scheme. "as long as he will have been alive".346 MORPHOLOGY This is the main reason why these categories do not correspond to any particular tense of the temporal scheme which has been evolved in the Indo-European languages. acting must often be translated by a present. perfect forms of verbs denoting feeling.19. As a matter of fact. while he was crying bitterly". While the "classical" verbal system of the Semitic languages is based on aspect. Instead. e. present. This tense is related to Classical Arabic kāna qad or qad kāna followed by the perfect of another verb. we can see that these compound tenses are partly based an old formations which were used in the past to express particular aspects or situations and not time relations.g. the perfect pre­ served its original stative function. "when the Hazradjites had come". e. e.. e. "nobody is poorer than a rich man (if) he feels safe from poverty". The pluperfect "he had written". As for the imperfect.g.. wishing to. qad kāna ra'ā minka mitla mā ra'aynā. can be expressed in mod­ ern Arabic by using the perfect kān. as in Arabic 'azza wa-ġalla.g.g. "as long as". like those introduced by mā. discontinuing. . e. In temporal clauses referring to the future. modern speech tends to found the verb inflection on the notion of time and to express it by means of "tenses". and future. kān katab. "he was". per­ fect forms may correspond to an English pluperfect. etc. mā dāma hayyan. halaftu. speaking. it expresses unaccomplished actions in the past. " I swear". "he had written". the perfect should be translated by a future perfect.. in a temporal clause introduced by lammā which in standard Arabic usage implies anteriority of the subordinate clause to the main sentence (§58. Thus. wa-ġa'ala l-muslimūna ya'malūna.g. thinking. "he is mighty and great". with the perfect of another verb.g. In the past. d) Modern Languages 38. e.

with the imperfect. These constructions are not operative in Neo-Aramaic (§42. The future sense can be expressed also by the participle rāyih. "he will write". By using the imperfect ydkūn with the perfect of another verb. or the continuous present denoting an action actually performed.g. de'mrēt (hâ)wēt fokon. e. Additional morphs. the imperfect with b(i)indicates the future. various particles are prefixed or suf­ fixed to verbal forms in order to express either the general present. This construction is used in Classi­ cal Arabic to signify a situation resulting from an action which will be accomplished in the future: e. 38. modem Arabic can express the future "he will write".5). nehwē kāteb.g. "then we shall already be in the situation of having taken an equivalent". kāna n-nabīyu ya'ūdu l-marida.g. e. "going". e. e. Only some examples can be given in the frame of this Outline.24). Syriac can express the future. In reality. "the prophet used to visit sick people".is employed to express the continuous present. other than verbs but acting as verb modi­ fiers. or the future as opposed to the present. "which I had said to you". bi-yruh.g. In Eastern Arabian. with the pre­ ceding perfect of another verb. instead.22. §58. and he entered Egypt". usually with a volitive connotation. ydkūn katab. 38. the auxiliary does not alter the time reference of the verb in such constructions. In several Arabic dialects of the Arabian Peninsula and in NeoAramaic. kān ydktub (cf. The duration in the past (past contin­ uous) can be expressed also by the perfect of kān with the active partici­ ple. e. In particular.g. "he was". ana rāyih asma'. e. are used in several modem Semitic colloquials to express time relations and aspects. " I am going to hear".TENSES AND ASPECTS 347 kāna is a stative expressing a situation existing at the moment when "we saw" it and it does not shift the tense of the clause automatically to the pluperfect. a preverb b(i).21.g. Authors generally assume that Syriac has created a pluperfect of the same type as Arabic by combining the auxiliary verb (hâ)wā.g.18-22). This compound tense goes back to Classical kāna yaf'alu which denotes a stable situation con­ sisting in doing something. but its use is extended to the general present in Syro-Palestinian and Cairene colloquials. kān kātib. "and he boarded it (a ship) and sailed with them on the sea. fa-nakūnu qad 'ahadnā 'iwadan. wīteb bāh wardā 'amhon bd-yammā wd-'al (hâ)wā ld-Mesrēn. A similar analysis explains the modem use of the perfect kān with the imperfect of another verb to express the European imperfect or past con­ tinuous "he was writing". e. however. thus: "he was already seeing through you just as we saw". By using the imperfect nehwē with the participle of another verb. "he was writing" (§42.g. .

in turn. In Eastern Neo-Aramaic. This construc­ tion parallels the Neo-Aramaic tense formed with the preverb bit(§42. kā. "let him prosecute". but is used at San'a in phrases like bayn-aktub. the particle is prefixed to the infinitive followed by the preposition / with a pronominal suffix. this formation must be linked to the earlier Qatabanic use which exactly parallels most Syro-Palestinian colloquials: the b.cannot be discarded i f one takes some nominalizations into account. kdm 7 s knw w'l bys knwn. and ki. " I am just writing". "we write".21).g. In Arabic colloquials.is prefixed to the indicative imperfect (§38. an innovation of the imperfect.D. The particle k.. Hawrāni vernacular la teftahis). Damascene bydktob.g.g.23). The same use is encountered in the Western Neo-Aramaic dialect of Ġubb 'Adīn. "she is opening".marking a present tense. but mndktob.19) and the widespread Indo-European use of an auxiliary verb expressing desire to form a future tense (e.g. In Neo-Aramaic. An explanation based on the preposi­ tion hi.g. e. but not to the jussive or in some other way not-indicative. 42. . bi-ptāhā-lā. etc. In Arabic colloquials. bydktbū. In any case. e. e.as the shortened form of yibġi "he w i l l " . the preverb b(i). "he is opening". "he will do"). "he ordered his killing".23.g. "he will sleep". " I am writing". e. it is used with the imperfect and serves to express the continuous present.is prefixed to the imper­ fect. "she writes".. "he wants to go" (Kuwait). Although the earliest Neo-Arabic instances of the ^-imperfect date from the 9th century A. bi-ptāhā-lē. both attempts to explain this b(i).seem to be undermined by the regular use of a preverb b. Qatabanic wl ylsq. t l l 38. and is vocalized kū.(< kūn) in Anatolia. "they write". the Neo-Aramaic preverb bi. "don't open".24.(< kin) in Neo-Aramaic. "to be". which is often used in Bedouin dialects and in Libya to signify an action that will be performed immediately.348 MORPHOLOGY "he will go". 38. in accordance with the Neo-Aramaic verbal system (§42. etc. Besides. e.(< kārì) in the Maghrib. ka-niktib. "he wants to sleep". "because they did not decree and will not decree".g. According to one opinion.originated from the conjunction baynā which means "while" in Classical Arabic.is used in other dialects with the same function. e.23.21). with the partial assimilation bn > mn. bi-yudmuk. Another explanation considers b(i). "he writes".is the common Semitic preposition b-. but it cannot serve as an explanation of the preverb b(i). like Classical Arabic 'amara bi-qatlihī. since it governs an original infinitive (§38. instead. It derives from the verb kāna I kūn. the preverb b(i). that may have prompted. btdktob.in the Qatabanic indicative imperfect.

ki-pātih. the future tense is formed by the particle kdprefixed to the imperfect and by the copulative old pronoun 'dyyu. kaldb 'db gabay hs'e 'ala. to establish tenses expressing continuous actions either in the present ( " I am writing") or in the past ( " I was writing"). The gerund (§42. while the future is expressed by the imperfect with various affixes. 38. in Tigrinya and in West Gurage dialects. e. the imperfect + halla expresses the present continuous.g.26. "a dog was running on the road". and the participle. "he is breaking". In most Ethiopian languages. The modem North Ethiopian languages have developed several compound tenses. "they are opening". to".g. "he breaks"). kisābbdr (< kd-ydsàbbdf) 'dyyu. in Tigre. or the jussive/subw . Instead. especially in order to distinguish the present from the future. Instead. in Harari.g. followed by the jussive and by the copulative pronoun tu. The place of the Arabic negative mā in Mesopotamian <p/iw-dialects deserves a special mention. 38.25. but kū-mišrab at 'Aqra. hdna hddāy ndtfarrar hallena. the imperfect + nàyru or nàbārā expresses the past continuous. There is also a noticeable tendency. "she is opening".g.g. "he is open­ ing". the simple present or past from the continuous present or past. e. ki-pāthā. ydsābbdr. A similar development can be observed nowadays in the Mansa' dialect of Tigre. the present is signified by the simple imperfect. 38. In a past context.g. "he is not drinking". "for. faġdr Basd' 'dgdl nigis tu. "he (is)". the imperfect + 'ala expresses the past continuous. It may be placed either before or after the verb modifier k-.g. the imperfect + alio expresses the present continuous. there are two ways of expressing the future: either the imperfect is followed by -te / -k e. In a past context. Other compound tenses are used as well with the imperfect. e. e.g. "he will break". ydsābbdr-allo.g.19). In West Gurage dialects. ma-kišrab in the Irbil dialect. in Argobba. e.TENSES AND ASPECTS 349 instead. "he was breaking". in Amharic. e. thus in Ge'ez. In the Mansa' Tigre of today the future tense tends to be expressed more and more by the preposition 'dgdl. the imperfect expresses the pre­ sent and the future. Instead. in East and North Gurage.27. the perfect. especially in Tigrinya and in Amharic. ki-pāthī. "tomorrow we shall go to Massawa". Tigrinya exhibits a paral­ lel development: whereas the old imperfect expresses the general present (e. ydsàbbdr nàyru I nābārā. e. e. "he (is)" (§49. in Gafat. it is prefixed to the active participle.12) enters in the composition of other compound tenses. "(only) we are going out to the wedding".

he had spoken". "he was". 5° Past perfect. The simple imperfect expresses the present and the future in the main negative clause and in subordinate clauses. he w i l l speak". whereas "he will find" is signified either by the suffixed imper­ fect ydrakdbte or by the suffixed jussive ydnkàbšà. In main volitive and negative sentences. The simple imperfect followed by a frozen or a conjugated form of nàbbār(â) expresses a continuous.350 MORPHOLOGY junctive is followed by -šà I -se. the future. Thus.28. "he has spoken". durative or habit­ ual action in the past: ydnàgdr nâbbàr(â). he was used to speak". formed by the combina­ tion of the gerund (§42. both affir­ mative and negative. The compound gerund. especially when the action occurs at the moment of speaking: nāggâra. "he spoke. he has spoken. Chaha ydrakdb means "he finds". w Amharic uses the auxiliary verb alia. 4° Present perfect. expresses a past action the outcome of which continues into the present: nāgfall. With certain verbs. 3° Past continuous. with the simple imperfect and with the gerund. yanagral. "he is". and the future perfect in the main affirmative clause: ydnāgr or yanàgdr. and it combines it with the gerund to form the present perfect. "he speaks. The combination of the gerund with nàbbàr{a) expresses the pluperfect or past perfect: nàgro nàbbár(à). e.. he is speaking. This leaves Amharic with five tenses used in main positive clauses: 1° Imperfect. to form the imperfect of the main clause. the jussive yangar is used instead of form 1. the perfect may express the present. Besides. Amharic developed a past continuous and a past perfect or pluperfect by using the verb nàbbāria). .g. The compound imperfect with the auxiliary verb alia > all expresses the present. "he was speaking. and form 2 serves as the negative for tenses 2 and 4. while the imperfect with -te I -k e implies doubt or simple intention. 38. "he had spo­ ken". 2° Perfect. It would appear that the jussive with -šà / -se expresses certainty.12) with the auxiliary verb alia > all. The perfect normally expresses the past and may also express the pluperfect.

to some extent. The subjunctive or "relative" is by definition the mood or form of verbs in clauses which are subordinate to another clause and intro­ duced by a conjunction or a relatival antecedent. two types of the so-called subjunc­ tive. to the ven­ tive. Moods 39. the jussive with the energetic.12). The enclitic -na occurring in dialectal Old Babylonian may have the same grammatical function. 39.13). at least with some prefix-conjugations (§38. in Old Akkadian. The use of the same marking with the verb must imply that an analogy was perceived between the ergative and the subordinate . There are good reasons to believe that this marked indicative originated from a generalization of the -u ending of the subjunctive which denoted subordinate clauses (§38. and to the pronominal suffixes of a verb.3) and is functionally identical. it can be added to the subjunctive. with the jus­ sive. It is also possible to consider as moods the paradigm that comprises optative or precative forms. to which the so-called apocopate or apocopated imperfect of Classical Arabic and of Hebrew is closely related.1.4. and in Assyro-Babylonian. The same suffix is attested in Classical Arabic and in North Gurage dialects (§38. 39.14. in the Amarna correspondence. 39. One can distinguish five moods in Semitic languages: the indica­ tive. 39. Except for the jussive.2. The so-called "subjunctive" ending -ni occurring in the Assyrian dialect is not a mood ending. for a verb with a ventive suffix does not take the subjunctive suffix. may simply be the ventive suffix with­ out mimation (§39. are prefixed and hence structurally differ from the moods marked by a suffix.MOODS 351 C. but a suffix -u seems to appear in Amorite.7). The subjunctive suffix is generally -u in Palaeosyrian. however. which occurs in Old Akkadian and in Palaeosyrian subordinate clauses. and in Assyro-Baby­ lonian. while the ending -a. We can assume that the -u suffix of the subjunctive derives from the -u ending of the ergative-instrumental case in nominal constructions (§32. but an enclitic indicating the end of a dependent clause. these moods are all characterized by suffixes. which was initially unmarked. the ventive or allative. The indicative is unmarked in Palaeosyrian. and the one which comprises the vetitive or prohibitive forms.Iff.3. and perhaps in Ugaritic.) which historically preceded the appearance of formally subor­ dinate clauses. in Old Akkadian. These.

U-. "so that".hā-'ām hazze bd-yādī wa-'āslrā 'et-'Âblmelek. so that I may enter in the paradise!" The subjunctive is also used after lan. e. of (sawfa) yaf'alu. of the stative. "so that". does not expect that "he will do" or " w i l l pay a visit". the -a subjunctive. 'aw. cannot be equated with the Arabic subjunctive in -a which is a marked form of the jussive expressing wish. which is the sub­ ordinate form of the preterite.352 MORPHOLOGY clause.g. should not be understood as statements and negations. the subjunctive occurs sporadically also after wa-'. expectation. whereas the apocopated imperfect. but it is called "indirect cohortative" in the grammars of Hebrew. The Palaeosyrian and East Semitic subjunctive. "would God we were taken back (from hell). may govern the subjunctive in certain circumstances.. It is remarkable. in any case. or lan yazūraka 'abadan. of the perfect. or consequence after well determined conjunctions. In the pre-classical language. Also other conjunctions. "may he (the Pharaoh) keep his servant in life so that I may guard his faithful city". They signify that one does not foresee.to express finality or consequence. It is used in the classical language after fa-. 39.g.g. ('i)ġfir līyā Rabbi fa-'adhuia l-gannata. e. and time (§57-58). This subjunctive in -a alternates in the Amarna correspondence with the East Semitic ventive.29). "who will give this people in my hand that I might get rid of Abimelech?". ml yittēn 'et. e. but phrases like lan yaf'ala. (Judg. " I pre­ sume. wazannīyā bn-a l-'Arwā 'an ta'ūda. and the energetic appear mainly in other subordinate clauses. O! my Lord. "he will never visit you".g. Modern colloquials no longer distinguish the subjunctive. 39. kay-. "he will do". O! son of Arwa.5.g. and of the present-future. like 'an {'allā < 'an-lā) and hattā. yuballit ardašu u anassara āl kittišu (EA 74. space. yā laytanā nuraddu wa-lā nukaddiba bi-'āyāti Rabbinā.6. e. 9. "forgive me. This is understandable i f -u was suffixed at first to verbs of clauses which were situating the main action in operational circum­ stances of cause.55-56). that Classical Arabic uses the -u imperfect precisely with these cate­ gories of subordinate clauses. so that we might not contest the signs of our Lord". instead. and its particular use with expressions of wish or expectation. especially in the post-classical language. this subjunctive in -a is used in Old Canaanite and in Hebrew after the conjunction wa. especially in the first person . that you will come back". finality. "unless". Like in pre-Classical Arabic. e. "he will not do".

g.g. "to be sitting for someone. "to be teaching for". was used "to indicate that the action of a verb was destined for. "to comb". although it belongs to a subordinate relative clause. Boni has retained this suffix as a productive mor­ pheme. the Ur-III name Tu-ra. ublūnim. in particular of the "Sam" sub-group. (e. šu a-na PN a-ti-na.7. from Mari) and there is no need. thus enhancing the proposed interpretation of the East Semitic ventive/allative. In other words. therefore.g. "Return. in fil. and it characterizes actions which profit to the logical subject or affect it in one or the other way. "he came here".g. "(flour) that I gave away to PN". This positional allomorph -im strengthens. it "would thus have been used with motion verbs to indicate simple action in the direction of the speaker" (P. e. . which indicates that action takes place in favour of someone or is directed toward a specific end. This mood originally indicated motion toward the speaker or the focus of action.Da-gan /Tūr-a-Dagān/. e. which is suffixed to the verbal stem. Newman). It is characterized by the afformative -a which is believed to have originated from the pronominal dative suffix of the first person singular.g. Since subjunctive and ventive suffixes are mutually exclusive. leaves little doubt that the West Semitic subjunctive in -a is but a ventive or allative. illik. that probably gave rise to the dative suffix "to me".g. e. šūbilīm. the ventive or allative is not unique to East Semitic. the East Semitic suffix appears as -im after the -I ending of the 2nd pers. d 39. káàd. fem. even in Old Akkadian. but very soon also signified a motion coming from the speaker or the focus of action. sing. "to buy for one's self". in its turn. However. However. done for the benefit of. and kád-o. Sotho ho-lúl-al-á.affix. this verb attina with the ventive suffix -a does not take the subjunctive suffix. with some distinctive syntactical features. since it appears as -da after verbs in plural. "he came". The mimation can be missing. "to comb one's self". "to buy". "send in!") and as -nim after the -ū termination of the plural (e.g. to have recourse to an East Semitic borrowing to explain the -a suffix (§39. to be waiting". This morpheme. Kwena hu-rút-al-á. -am. "they brought in"). or otherwise affected or pertained to some­ one". I f this interpretation is correct. the similarity between the afformative of the ventive/allative and the "destinative" verbal extension -*in reconstructed for Proto-Chadic. The directive a-morpheme is attested in Palaeosyrian (e. the historical process may have been in the contrary direction: it is the Afro-Asiatic directive morpheme. This suffix probably derives from -*a.MOODS 353 singular. Dagan!". illikam. a verb of motion with the "destinative" -*in suffix would constitute a kind of pre-dative verb form.7). which is paralleled also in the Bantu languages by the verbal forms with the direc­ tive -al. the function of which appears e. and fil-o. There is a conspicuous analogy between the ventive/allative suffix -a and the "benefactive" suffix -o I -oy of Lowland East Cushitic.

and -ā in pause: yafa'lanna / yafalan / yaf'alā. which might indi­ cate either that the use of -n was optional. 7 tlqhn.g. but there is a large number of controversial cases (e.9): 'immā tarayanna mina l-basari ahadan faqūli " i f you happen to see a human being. "may they perish". and it is used in this way also in the protasis of conditional sentences and in interrogative sentences. wasin. yqsn.g. where the energetic signifies a desirable possibility and is thus comparable with forms in -an of Modern South Arabian (§39. It is character­ ized in Classical Arabic by the endings -anna. and I shall well come to know These semantic connotations of the form appear also in con­ ditional sentences introduced by 'immā.g. "they shall draw") and in Aramaic (e. including relative clauses (e. sing. "don't abuse". .g. "so do not try to imagine God as a breaker of his promise concerning his envoys" (Qur'ān 14. The energic in -in or -ina is the usual form of the jussive in the Agaw dialects of the Qemant-Qwara group. la-yaqūlunna. "and may he grant them"). "let him hear!". say . "may you sustain us!" 39. It corresponds exactly to the so-called "direct cohortative" of Biblical Hebrew. .9.g. These endings are all added to the "short" jussive and they are generally introduced by the optative pro­ clitic la. The energetic denotes a strong wish rather than an emphatic asseveration or prohibition. Mehri ydslēmdn. "he would be safe"). "and all what he might acquire"). by the ending -un(n)a\ e. 'āsuranna. 'ēldkā. The suffix -n of the energetic is attested also in Phoenician (e.: T A D I I I . and -dn is the comparable suffix in Mehri and Sheri. layta ši'rī wa-'aš'uranna "would I have known. 1. anāt alšina. like fa-lā tahsabanna llāha muhlifa wahdihi rusulahu. This mood. la-tarawunna. No proclitic is used in a sentence introduced either by the negative lā or by wa-. " I should go").g.167). sing.354 MORPHOLOGY 39. ù yu-wa-si-ru-na. halafa la-yaqtulanna. Sabaic wyhmrhmw. yš'n. which uses the suffixes -anna (e.g. . Sabaic 7 Vyrn. as a rule.. written 'srh-n'. " I should go across") or -ā (e. "you may not take". " .g. C I . " . "he swore that he will try to k i l l " . "may they speak at last". e. "that he should send" (EA 71. . -an. where it is characterized.13).g.8. "you will well see". or that -n could be assimilated . wa-'immā tahāfanna min qawmin hiyānatan " i f you should fear treason from people . in the 2nd pers. occurs already in Old Canaanite. with its various functions.48/47).13) in affirmative sentences. Minaic wkl dyqnyn.(§39. The -n imperfect of ancient South Arabian occurs in jussive and subordinate clauses of all kinds (e.).g. used only in sen­ tences involving unreal conditions (e.

g. hīs ydrkēz. Ethiopic and Modem South Arabian distinguish two moods: the indicative of the enunciative clause and the jussive or subjunctive of the volitive clause.12-13). "when he stands upright").1) and by the phrase yqr'a mt b-npšh ystrn ydd b-gngnh (KTU 1.23). or a desirable possibility (e. 39. A particular problem is raised by the double -nn of the Sabaic dual and plural -n imperfect (e.g. But the form in -ā is probably employed as well. " i f you happen to see"). "may I invoke the gracious gods". the Mari forms iškunanna and imhuranna in subordinate clauses may reflect the native Amorite idiom of the scribes using the energetic ending -anna.10. may the Beloved hide in his inwards". These convergent data seem to indicate that the energetic goes formally back to the ventive/allative. and therefore had no graphic expression in these cases.17.47-49).MOODS 355 to the consonant of the following pronominal suffix. Although the energetic does not appear in Amorite proper names. yqnynn. where the syntactic status of the two verbs yqr'a and ystrn is absolutely the same. 39. "they will make").g. Hebrew nēhkā. although the interpretation of partic­ ular examples is sometimes open to question. or even a predictable fatality (e. It denotes especially either a strong desire of the speaker (e.4. or 'iqr'an 'ilm n'mm (KTU 1.VII.23. as suggested by the parallel passage 'iqr'a 'ilm n'mm (KTU 1. or that -*an was sometimes reduced to -ā like the Arabic pausal form and the Hebrew direct cohortative in -ā. to the alternate forms -n and -a in Ugaritic (§39.12. and . Hebrew 'āmūtā. one can assume that -n was also the original ending of the Sabaic plural imperfect and that this -n was pre­ served before the energetic suffix -*an(na). "let us go!"). "may I too sit down and be at ease". thus -*ūnan(na).g. but semantically has optative or prospective connotations.11. 39. The evidence of Mehri and Harsūsi indicates that the subjunctive of Modem South Arabian formally corresponds to the ancient yaqtúl jussive (e. "may Mot cry out from his throat. Since the Qatabanic simple imperfect ends in -wn (yf'lwn.g.11. " I shall surely come to know").10). "they may acquire"). Its use in Ugaritic has long been recognized. The energetic is attested also in North Semitic. " I shall have to die"). as well to the frequent use of this mood with verbs of motion (e.23. 'immā tarayanna. Arabic wa-'aš'uranna.g. The origin of the energetic mood is linked to the element n of the suffix. The suffix -an or -anna is certainly used in "cohortative" cases like 'atbn 'ank w-'anhn (KTU 1.

is his duty". e. hqtal. the phrase 'dgdl + jussive + tu is used in Mansa' Tigre as an expression of futurity without any modal connotation (§38. appears in subordinate clauses. mi 'ide. which is used as a volitive mood. the jussive preceded by the con­ junction 'dgdl. "you should take". Nowadays. ul cannot be considered as proclitics (§47.16). barhat tdgba'. "may they grant prosperity". The negative adverbs lā. or implying a doubt. The so-called "apocopate" or "apocopated" imperfect in Clas­ sical Arabic and in Hebrew is a shortened form of the prefix-conjugation corresponding to East Semitic iprus. 'dgdl tdššayam. "to go is his duty". Tarifit at-tdksid. is used for the main clause./ la-. "so that he should go. "let there be light".g. 40. the volitive or injunctive forms of the verb are composed by prefixing a proclitic to the basic verbal pattern. ydqātl or yaqàtdl in Amharic) is used for the subordinate clause (and also for the negative clause). 39. the simple imperfect (e.18. The resulting paradigm can be considered as a kind of mood. In several languages. 39. lit. e. The vetitive or prohibitive is formed by prefixing ay or one of its derivatives (§38. composed of the simple imper­ fect and of a variable "auxiliary". In most South Ethiopian languages. especially with the third person./ li.13.356 MORPHOLOGY the same usage is attested in Ge'ez which has two jussive patterns: ydqtdl for the mainly transitive verbs and ydqtal for the intransitive ones (cf. viz. e.to a verb occasionally entails graphic deletion of imperfect y-. or requiring an answer in the imperative. Sabaic Ihslhnn < l+yhslhnn. ad-ydksi.g. i. Prefixing of the proclitic /.g.2.26). even when they are attached to the verbal form. 'dgdl ligis waġġdbbo. It is found also in interrogative clauses.30). The historical appreciation of these peculiarities must . ydruh ad-ydġar. e. Only one jussive pattern occurs in Tigre. when the question is either rhetori­ cal.23.e.g.59). Besides.14. while the compound imperfect. the so-called verbs ultimae infirmae.8). and by the shortening or the loss of the final long vowel in verbs with a third weak radical. A similar construction of the volitive is attested in Libyco-Berber with the particle ad-1 at-1 ad. There is no doubt that this prohibitive particle is originally iden­ tical with the interrogative 'ay (§36.g. to express the optative or precative (§38. "he left so that he might study".and the jussive. Widespread is the use of the proclitic lu. "what should I do?". It is characterized by the absence of the indicative -u suffix in Arabic. 'al. §38. "so that you will be appointed".g.2). e. "so that". Amharic does not use the jussive in subordinate clauses. "one should draw".

§58. The variations in the spelling. and lammā ya'ti. "he didn't build". 39.11). "he should not do!". Besides.g. ia-aq-bi. sw and swh. lā ya'ti.. The analogy suggests that Arabic and Hebrew apocopate reflects an earlier stage of West Semitic. §43. as i f the stem itself was shortened. "not yet". — ysw and yswh. — ydsaw and yasawwe. " i f you do not tell. and real conditions after the particle 'in. "not". "speak to God's envoy (in order that) he would speak". li-yaf'al. "he should do!". saw and sawwē. it preserves the old function of the jussive in asyndetic final / consecutive clauses following an imperative. its function exactly parallels the one of East Semitic preterite.g. — gave rise to different Masoretic vocalizations. "he built" (root bny). — although the differences are purely graphic or dialectal. preterite ibni instead of ibnl. 39. e. lam ya'ti. "he should not come!". " i f he does not depart. "have pity (in order that) you will be pitied". and it is to be consid­ ered likewise as reflecting an earlier stage of the language. you will die". This interpretation is confirmed in Hebrew by the fact that apocopated forms are used as jussive and in wayyiqtol (§38. documented already by the Amarna correspondence (e. li-ya'ti.g. 'irham turham. "he didn't yet come" (cf.g. "he didn't come". or dropping it altogether.15. and lammā. parallel East Semitic lā ibni. As for the apocopate expressing negative statements after lam. "building" (par­ ticiple). " i f " .g.e. "he will build".4). their inflection precisely exhibits the phonotactic feature that length is dropped in absolute final position and that vocalic ending may drop altogether. ban instead of bāni < bānī. It preserves the old volitive function of the yaqtúl after //.g. "may he speak") and traces of which were later systematized by the early Arab grammarians and by the Masoretes (cf. They reflect a spelling and a pronunciation either expressing the final short -e < -i (< i) and indicating it by the vowel letter -h.(inclusive wal-. parallels Assyrian šumma atta Id taqbi tamuat. kalliml rasula llāhi yukallim. The situation is similar in Classical Arabic where the jussive is operative only in determinate kinds of syntagms. when the original function of yiqtul is preserved. I shall not be satisfied". 'sw and 'swh. E. i. wa-li. present-future ibanni instead of ibanni. Now. lā yaf'al. "he should come!". "to order". 'âsaw and 'âsawwe. e.12). fa-li) and after the negative lā. fal-.MOODS 357 reckon with the inflection of the verbs ultimae infirmae in Assyro-Baby­ lonian. e. e. . 'in lam yabrah lam 'arda. of the verb swy.16.

instead of the indicative in -u.19.358 MORPHOLOGY 39. the apocopate was used also after 'an.'ayn 'annak. The following branching diagram of tenses and aspects summa­ rizes the presentation of the common Semitic development of basic ver­ bal forms in the "classical" languages: ya/i + CCvC + an(na) (energetic) ya/i + CCvC (jussive) CCvC (imperative) yá/i + CCvC (preterite) ya/i + CtaCaC (perfective) ya/i + CCvC + a (subjunctive-cohortative) ya/i + CCvC + u (indicative imperfect) (imperfective) CaCC (adjective) ya/i +CtaCiC (perfective) ya/i + CaCCiC (imperfective) CaCuC ya/i + CtaCuC (perfective) ya/i + CaCCuC (imperfective) . just like the wayyiqtol in Hebrew. "if. We can assume that those dialects didn't have the subjunctive and the indicative in -u. in some categories of negative clauses.17. Allah yardi 'alayk. when".g. Allah yakūn ma'ak wa-yahmīk. In some ancient Arabic dialects. In Hidjazi dialects. In modern Arabic colloquials. On the other hand. The gram­ matical analysis made by Arab philologists on a synchronic level should in fact be reinterpreted in both a diachronic and synchronic perspective. and lan (< lā-'an) instead of the classical subjunctive. 39. the apocopated jussive is widely used without the particle /*'-. instead. 39. the indicative in -u was operational in cases where others used the apocopate. e. "may he put the (evil) eye to shame before you". and after law. "be Allah with you and may he protect you!". Thus in subordinate clauses they employed the jussive. "Allah befriend you!".18. "that". the old lam-yaqtul and lammā-yaqtul continued to be used to express the preterite. yuhzi I.

The following paragraphs (§40. two types of paradigmatic sets determine the two types of conjugations: the suffix-conjugations and the prefix-suffix-conjugations. a) Suffix-Conjugation 40. . but later agglu­ tinated to the verbal base. but also by Hausa which expresses the personals by separate pronouns that precede the verb (§36. they preserve clear traces of case inflection (§40.2. both as prefixes and suffixes with the derived forms where person is designated by prefix morphs. The Ge'ez paradigm can serve also for Tigrinya. In the paradigmatic set of the suffix-conjugation of the stative. like in Hausa.1.1). as in the standard form of the third person masculine singular.ACTOR AFFIXES 359 D.16). usually called prefix-conjugations.3-12) will offer some observations on this table. Remembering always the gaps in our knowl­ edge and the dangers inherent in any argumentation from analogy. the hyphen (—) indicates that there is no marking. the pre­ fixed personals are survivals of pronouns once separate. In consequence. gender. and num­ ber. Whereas prefixed Semitic personals are not fused with morphemes indicating aspect and tense. Actor Affixes 40. 28. in 1845/6. While the actor affixes of the suffix-conjugations go back basically to a form of pronominal suffixes of the noun. The actor affixes or personals specify person. we can use the evidence supplied not only by important languages of the Niger-Congo family. gender and number by suffix morphs. They appear as suffixes with the stative and the imperative. The square of fountains at Ghadames. Fig.

ša-pár.24).5). e. This suffix is generally believed to belong to the normal inflection of the Ugaritic stative/perfect because of forms like nš'a which are vocalized by some authors *naša'a. Qd-ma-Da-mu.g. f. "he lifted up". 45. "he killed".g. in any case. as damqa. while the bulk of the material shows no -a. the general trend in Semitic elides the final ' and lengthens the vowel (*našā. The stative. f.8). the syllabic spelling of Ugaritic proper names shows no -a ending in the stative/perfect (e.g. " E l is good"). This may be the correct explanation of the few forms with final -a in the Amarna correspondence. 2 m. §27. 1 East Semitic Ugaritic Hebrew Aramaic — -at -ka 1 -ta -ki 1 -ti -ku — -at -āt(a/i) -āti -āk(u) — -t -t -t -t — -ā -tā -t -tī — -at -t -tī -et -ā -atā -kā (?) / -tanā (?) -kāya / -nāya -ā -tā — -t -tm -ny -ū -ā -kan(u) 1 tanu -kin(a) l-tina -na -ū -ā -ātun(u) -ātin(a) -ān(u/i) — — -tm -tn ? -ū -ū -tem -ten -nū -ū -ū -tūn -tēn -nā 40.360 MORPHOLOGY *Proto-Semitic Sing. However. f. "is good". f. has a third person masculine singular in -a in Classical Arabic and in Ethiopic. which became a mark of the perfect through the participial predicate as in Qāma-Da'mu. Old Akkadian (e.3). either substantive or adjective. 3 m. "Damu is standing"). this -a is the mark of the predicate state of the noun (§33. which is indicated by 'a (§19. and Amorite (e. f. 3 m.g. 2 m.3. which became also a perfect (§38. du-ak. "the god is beauti­ ful").g. Since the stative essen­ tially represents the conjugation of a noun (§38. cf. "Baal is king") and. 1 Dual 3 m.10). Ta-ba-El.8. ìl-ba-na. The assump­ tion that the vowel a/ā linking the pronominal suffix to the stative/per- . "he sent". 2 1 Plur. also in some personal names occurring in Palaeosyrian (e. Ba'al-ma-lak.

as this can be seen in parallel names.g. -k -Š -hu.g. another value attested in the same period. "Baal has blessed". like Si-'-pa-rak-ka /Si'-barak/.g. "The Moon-god has blessed". Amharic nābbār. Sàwa rāggād. Arabic -a -at -ta -ti -tu Coll. d . isbatanni. e. "Shoa trembled". Judging from the Greek transcription of Pre-Islamic North Arabian names the third person masculine singular of the Old Arabic perfect did not end in -a. "he killed me") is a residue of the ancient -a ending is also questionable. qdtālanī. Arabic Sabaic Mehri Ge'ez -a Amharic -ā -add -h.ACTOR AFFIXES 361 CI. not sal. Harari hal.g. "Hadad has blessed". do not exhibit a termination -a. or lM-ba-rak-ka /Hadad-barak/. Se-erba-rak-ki /Sehr-barak/. since the same vowel appears in AssyroBabylonian with the preterite (e. "he was". "he seized me").-ku -at -t -ti -t 1 et 1 -it -t -k -dt -k -Š -k -at -ka -ki -ku 1 -it ? ? -ā -atā -tumā -y -ty ? -o -to -ki -ki ? -ū -na -turn -tunna -nā -u -in -tu -tin -na -w -aw-y -u 1 1 -u -nl — -kdm -kdn -dn - á } J -u ? 1 -tu ? ? -kdmmu 1 -kdn •na -accdhu -(d)n feet in Hebrew (e. Gogot). "the Moon-god has blessed". and the Classical form is therefore based on dialects having connec­ tions with South Semitic. Only auxiliary verbs and proper names may lack this -a or its equivalent -o (Soddo. e. "there is". at least in Ethiopic. where the -a predicative had become a firm element of the stative/perfect. Aramaic names in cuneiform script. Bé-il-ba-rak-ki /Be'dlbarak/. The final signs -ka or -ki are phonetic complements indicating that the penultimate sign is to be read rak.

masc. just as the plural sufl . while the Amharic -àcé results from the palatalization of -ati. 40. "you wrote").g. The survival of these variant forms.g. Chaha). "you heard"). "she can be confident"). Sabaic 'wdk. in Epi­ graphic South Arabian (e. §33. It is not possible to exclude the Proto-Semitic origin of the second one.g.) appear in dialectal Neo-Assyrian (e.5). kunk. What is not generally known is that a similar formation is encountered in dialectal Neo-Babylonian (e. perhaps under the influence of the masculine -a termination (§40. "you are spell-bound).1-2). both ancient (e. transcribed -t by Egyptologists.5: -ta.6. The feminine ending in -t corresponds to the feminine morpheme of the noun (§30. "why are you silent?". "she heard". Gafat is the only language having an ending -àttà in -ā. and because the Old Egyptian suffix-pronouns of the second person singular are -k = -ka for the masculine and -t < -*ki for the fem­ inine. which later philologists explained saying simply that "some Arabs occasionally substitute k for r" (Lisān XX. "you reach"). because the Tuareg independent pronouns of the second person are formed on the basis k-. possibly influenced by the second person feminine singular. "you were disloyal") and modern (e. a-ta-a qala-a-ka. the endings -āka (masc. fem. lu-ú ha-ma-ti.occurs also in some Gurage dialects (e. The Old Egyptian palatalized -ki. They are suffixed as nominatives to the simple tenses of the verb (e.g. -ti) or suffixed (§36. śdm-t. while the Modern South Arabian and Ethiopic feminine suffix -š (Argobba -â) originates from a palatalized -ki.) and -āki (fem. exactly in the same way as the Semitic personals of the stative/perfect. the Egyptian "pseudo-participle" sdm-ti. as an euphonic vowel.5. Zway (-ha). While the palatalization -ati > ac(c).3). a form with a final -/ which appears in some Gurage dialects {-átîi in main clauses of Soddo and Gogot) and. Masqan. so much the less because it is used with the stative in Palaeosyrian (e. Besides. śdm-k.4. aza-me-kà or a-zi-mi-kà /lazimika/. In any case. and Amharic masculine suffix in -h results from a spirantized -ka. the original ending -t is preserved before pronominal suffixes.g. also indicates the presence of a final vowel (cf. -ki). In Hebrew and in Phoenician.g. in Harari. either independent (§36. ka-aš-da-ki. was written also -t in later periods. 'asayka. s trk.362 MORPHOLOGY 40.g. "you brought back". The second person masculine and feminine is characterized by the same morphemes as the personal pronoun of the second person. and in Yemenite dialects of Arabic.330). is the strongest evidence for their use outside the proper realm of South Semitic. The Gafat (-áhâ).g.19: -ka. "you were").

" I wrote"). and the additional ending -ī of Old Canaanite (e. marsā.7.). The suffix -t of North Semitic and West Semitic languages is almost certainly to be explained by analogy with the second person singular. The first person singular suffix -ku.7).g.ACTOR AFFIXES 363 fix -tn < -*kin was later indicated also by -tn. Old Babylonian -āt) which usually finishes by being absorbed in the palatal. these endings can therefore be considered as Proto-Semitic. bahalku. " I became king") derives from the possessive suffix . 40. like the -h of the second person (§40. can surely be considered as Proto-Semitic.g. while the endings -y (masc.5). " I built": EA 292. k'tbty.g.4). 'amartī. " I was".6) and it is attested also in Yemenite dialects of Arabic.29).g. used both in East Semitic and in South Semitic.C. ba-ni-ti.g. " I bore". " I spoke"). and Harari. and in Modern South Arabian with -6 < *-ā and -to < *-tā. It is identical with the morpheme -ku of the independent personal pronoun (§36. katubk.6. the second person forms in t could have originated from Proto-Semitic or Afro-Asiatic forms in k (§12. The masculine ending -ā and the feminine ending -tā are attested for the third person dual in East Semitic (used until the mid-second mil­ lennium B. §32.18).) and -ty (fern. there is certainly a relation to the Modern South Arabian forms -o and -to. which seems logical (e. kunk. The intermediate spirantized form -ku(m) occurs in Chaha. w 40. This might imply a pho­ netic change -ki > -c> -t through the depalatalization of à by the loss of final -i (cf. -uh) — has a wide­ spread allophone -ku and it results from the spirantization of -ku. "sick-we-two"). They correspond to the dual morpheme -ā of nouns in the subject case. Although one must allow for the possibility of a pronunciation -aw and -taw. The Qatabanic masculine dual has the ending -w and there is an alternative feminine dual ending -tw in Sabaic. mlkty. both ancient (e. Phoenician and Punic (e. the Libyco-Berber suffix -ġ of the first person singular represents a pharyngalized velar followed by a vowel. thus *-ku. The Amharic suffix -hu — attested also in Gafat and in some Gurage dialects (-h .f (§36. " I spoke") and modern (e. I f one assumes a similar evolution in Semitic. Finally. in Classical Arabic.) of Sabaic are due probably to anal­ ogy with the oblique case (cf. and Moabite (e.g. Muher. Hebrew (e. waladku.g. . " I wrote").

For the third person plural of Proto-Semitic we may posit the end­ ings -ū (masc. The resulting form -ki is then identical with the suffix of the second person dual. which so far do not appear as such in any known language.(-*kay > -ki or -Si after palatalization. the pronominal stem of the singular (-k-) is followed by the same elements -ā-ya. and with the metathetical Mehri pattern kdtawb. 40. pá-na-ù /pānayū/. The latter should be compared with the colloquial Arabic -aw ending of the Persian Gulf region (e. The dual end­ ing -ay of the oblique case would have been monophthongized in Ugaritic to -ē or -I without being marked in writing. The element -ya is the only one introducing a clear distinction either from the plural suffix of the first person (-na I -na) or from the proposed dual suf­ fix of the second person (-ka).364 MORPHOLOGY 40. "they wrote".22). and Amharic use the ending -ūl-u for both genders. are attested in . It stands to reason therefore that the Proto-Semitic suffix was either -nāya or -kāya.7. The Yemenite dialectal ending -ayn may be related to the Sabaic alternative feminine plural in -y which parallels the masculine -w. the Modern South Arabian languages add the oblique ending -ay to the sin­ gular pronominal stem -k.) which appear as such in Palaeosyrian (e.12) is added to the feminine suffix with various vocalizations (-ūn. the feminine plural ending -n (§31. masculine -w and feminine -y. ni-bu-ha /nibbūġā/. as expected.as deriving from -tan. Aramaic.22). one may pro­ pose the alternative Proto-Semitic endings -*tanā and -*kā. "they wrote"). but Hebrew. resulting in -*nāya. lit.is apparently added to the pronominal stem of the plural (-n-) and followed by the possessive suffix of the first person. reduced to i > i. -ši < -*ki). -ēn. "are outstanding"). ktibaw. In South Arabian. but it must have existed also in Palaeosyrian (§36.9. I f one considers the -ā ending of the subject case as ProtoSemitic and the form -turn. The first person dual is attested only in Ugaritic (-ny) and in Modern South Arabian (-ki. in several modern Arabic colloquials. in East Semitic. and in Late Aramaic dialects. and in Ge'ez. 40. The second person dual raises the same questions as the personal pronouns (§36.(§36.10. "have a face of".8. While the dual ending -ā is added in Classical Ara­ bic and probably in Ugaritic to the plural pronominal stem -turn-. In Ugaritic. in Classical Arabic.g. -an). -ana. Both Sabaic endings.5). some Arabic colloquials.) and -ā (fem. The same purely vocalic endings can perhaps be assumed for Ugaritic.g. "are clothed i n " . in Śherí). the dual morpheme -ā. In Epigraphic South Arabian. -ayn.

Instead.g..g. The Proto-Semitic origin of -kan(u). cf. in a main clause.5: -a-tanu. fern.g. e. As for Soddo. kâmâ-ti. The final -m is added to all the persons singular and plural in the positive main perfect of several Gurage dialects: it is an enclitic reinforcing the meaning of the word to which it is attached. Both the masculine and the feminine plural have a suffix in -m in Gafat and in Soddo (North Gurage). As for the problem concerning the consonants t and k. nâqâr-o-m. "they went out". The passage from -a-tanu to -a-tunu and -turn has to be explained in the same way as in the case of the independent personal pronoun (§36.g. nàqār-āma-m. e. There is a relation between this element -m and the plural morphemes -mu and -na of the personal pronoun (§36. at-tu-nu qa-la-ku-nu. w w 40. since the same suffix-pronoun -tn < -*kn is attested in Old Egyptian for the second per­ son plural. it distinguishes. katabkan). -kin(a) has to be taken seriously into account.appear not only in Modern South Arabian and in Ethiopic. either independent (§36.5). -a-kiri).g. sàffàr-ma. also kunkū. East Gurage and Argobba pre- . In fact. masc. Instead. used for both genders. while the Tuareg independent pronouns of the second person plural are. Chaha masc. "they pulled out". A masculine plural in -m is attested in the Western Ġiblah dialect of the Arabian Peninsula (e. kdtdbdm). e. used for the third and second persons. sàffàr-ma-n. The second person plural of Proto-Semitic should be character­ ized by the same morphemes as the corresponding personal pronoun. masc. "they measured". Similar forms in -k.17). "you wrote". sáffàr-mu-n. fagr-aw. This particular form is due to the fact that in some respects Amharic represents an innovated language type in the South Ethiopic group. fagr-aya.11.ACTOR AFFIXES 365 Tigre: masc. which adds the morpheme -kum > -kdm > ku to the nominal plural ending -āti (§31.g. fern. fern. "you were") and -kan for the feminine (e. -a-tina) or suffixed (§36.kan be assumed for Epigraphic South Arabian. and it has no connection with the actor suffix as such.5). "you keep silent"). in a subordinate clause. fern. A common plural termination -i m < -*um occurs in Gafat.g.g. The Amharic ending -accdhu and its variants. and in dialectal Yemenite Arabic with -kum for the mas­ culine (e. indicate that the suffix originates from *-ātikum. it is the same as in the case of the singular (§40. probably by analogy with the second person ending -hu m < -*kumu. katabum.24: -a-kun. sàffár-dm. kaw-ni and fern. and masc. the suffixes in -k.11). katabkum. a feminine plural in -m is used in the West Gurage dialects which make a distinction of gender in the third person plural. "they wrote") and in South Arabian Harsūsi (e. but also in dialectal Neo-Assyrian with -ākunu (e.

obvi­ ously by analogy with the ending -nan of the independent personal pro­ noun (§36. 40. and in Amharic. Arabic. masc. Dual Plur. sàffár-kdman. in the subordinate clause). which is spirantized into -kum. Chaha masc. the hyphen (—) indicates that there is no marking.14-15) offer some observations on this table. "you pulled out"). except in the northern Amharic dialects that preserve -ná.13.g. the distinction of gender in the second person plural is preserved not only in North Ethiopic. probably by analogy with the element -nu of the personal pronoun. -hum. fem. -hu m. It appears as -o in Gogot and in Soddo. while the Assyrian allomorph is -āni (§36. and in West Gurage (e. Soddo sàffár-nā. •u -u -na East Sem. or -hu in other South Ethiopian languages. Ugaritic Hebrew Aramaic -i -a -i -i -u -a •a . and most South Ethiopian languages (e. -ku. "you went out"). a subsidiary -n is added to -na (> nan). fem. Sing. sàffár-kdmun. The conjugation of the imperative is likewise limited to the use of actor suffixes. Actor Affixes of the Imperative Pr. Besides. but also in some South Ethiopian languages. The final vowel is shedded in Modern South Ara­ bian.12. "you measured"). f. fagar-kdn. viz.g. In Late Aramaic dialects. b) Imperative 40. 2 m. For the first person plural we may posit the Proto-Semitic actor affix -na which appears as such in Aramaic. 23). in Argobba. masc. fagar-kum.g. náqâr-ku-m. and it is replaced by -u in Babylonian and in Hebrew. very likely as the result of a change à > d > o. as in Tigre (e. "we measured".8. 2 m. nàqàr-kdma-m. Ge'ez.366 MORPHOLOGY w serve -kum.g. fem. f.-Sem. In the paradigmatic set of the personals of the impera­ tive. in Soddo. The following paragraphs will (§40. a North Gurage dialect (e.2).

"bite!". Instead. Actor Affixes of the Imperative CI. in the Chaha (Gurage) fern. Arabic Coll. "get up!": Ps.g. sdbar-a. "bring back!". please!"). 3 l 40.19). added to the imperative already in Amorite (§40. "be ready!". the masculine ending -ū is used also for the feminine and this usage is implied likewise by the Aramaic suffix -ūn < -ū + n. in Neo-Aramaic. No par­ ticular ending can be proposed for the feminine plural in Proto-Semitic. -n is added to a plural ending -ū (> -ūn / -In) by analogy with the inflexion of the prefix conjugation. s tr-n. ridkdš < ndkdsi.g. In other Arabic colloquials. The regular Sabaic ending -n of the mas­ culine singular (e. since the -na suffix of Hebrew and Classical Arabic is most likely related to the precative particle -na. there is probably no connection with the -a which can be added in Old Canaanite (ku-na. "break!").g. which is used also with the imperative. In some Late Aramaic dialects.8). s hl-n. please!"). 'âlē-nā.ACTOR AFFIXES 367 40.15. 'zl n'.g. "go. in Hebrew (e. "climb.14). 82. in Hebrew (e. The bare stem of the imperative is used for the second person masculine singular.14. please!"). "turn back. and the feminine is formed with the suffix -i which characterizes the second person feminine singular of the stative/perfect (§40. and in Aramaic (e. EA 147. 'wd-n. This -a suffix may be related to the -a ending of the subjunctive (§39. This -i may cause the palatalization of the preceding consonant.g. "come!".5) and of the personal pronoun (§36. The dual ending -ā is employed as plural ending in East Semitic for both genders. Arabic Sabaic Mehri Ge'ez Amharic -n -i -a 9 -i -o -i -u -na u u I -in -w ? 9 9 •u •a .g. e. and might be related to the Amharic interjection na. qūmā.36). "write!") goes back to the precative particle -na which is used with the imperative in Amorite (e. "take care!".5-6) and to the Assyro-Babylonian ventive/allative. and for the feminine in Aramaic and in Ge'ez.g. and in related Arabic colloquials.2. šu-ub-na-. and in Amharic to the imperative for emphasis (e.

in Modern South Arabian. the vowel of the prefix is indepen­ dent from the thematic vowel in East Semitic. in Hebrew. There are also cases of vowel harmony. "bite!". i in the plural (§32. but u occurs when the basis of Stem I contains this vowel. Soddo masc. ta-qrib. Instead. inclusive the basic stem which must be considered as orig­ inally intransitive. hence a subject in the ergative case. e. galbdm. galbdma. c) Prefix-Conjugation 40. "to bring near": 2 m. fern.g. This question brings us to the problem of the origin of personals and to the ergative foundations of Afro-Asiatic. Chaha masc.24). The Geez paradigm can serve also for Tigrinya.368 MORPHOLOGY Also Amharic employs -u for both genders. In Common Semitic. while a/i marks the non-active case: a in the singular. i f we disregard the derived i-stems for the moment. "we brought near" Barth's law stating that the prefix of the first set was vocalized with i when the thematic vowel was a (yiqtal). fern. qurrubu. which by definition have a transitive meaning. and with a when the thematic vowel was either i (yaqtil) or u iyaqtul). hence having a subject in the non-active case. yuktub besides yiktub.1). sing. "gallop!". in Aramaic. with final vowel -a < -ā characterizing the feminine plural like in Ge'ez. e. "we came near" tu-qarrib. This comes out very clearly in such examples as the following Assyro-Babylon­ ian verb qerēbu. does apply only to a later stage of some Semitic languages. The distinction of two sets of prefixes is lost in Neo-Arabic which mostly uses the i-vowel with all the stems. the use of the w-set of personals characterizes the causative or fac­ titive D-stem and S-stem (§41. 1 plur. The personals of the prefix-conjugations were represented origi­ nally by two paradigmatic sets characterized by the prefix vowels either a/i or u. "he writes". "you brought near" nu-qarrib.7 ff.16. but some South Ethiopian languages of the Gurage group distinguish the two genders. although the pair fa 'ila / yifalu is productive in Sibawayh's time (§40. The a/i-set is employed for the other stems.). in Amor­ ite. in Palaeosyrian.g. nakso. in Ethiopic. These vowels are no "root-augments" but case endings of personals once separate. "you came near" ni-qrib. naksama. and generally in Arabic. but later agglutinated to the base of the verbal stems (§40. .3.1-6). where u characterizes the ergative case. "to come near".

Text of Deut. 1. Fol. It will be commented in §40.x\^f/^2 /VKr^ZJtS'V^'rtr tkO/sC. 29.1-11.18-31.ACTOR AFFIXES 369 •2f rfr 2 \ - ^A-ev^ l£3X7*Kf<*>^ m Wf*2L .\ Fig. The first set of the prefix and suffix elements is shown in the following table. 237 of the Samaritan Pentateuch from the collection of Pietro della Valle (1586-1652) in the Vatican Library.17. . r Set I 40.

Tàš-má. prefixes of a Western type occur as well. " D N D N will go out").-ā ni- ylt- ... . yatatata-..-l 'a- (y)itatata-.-ī 'e-l'a- yitititi-.g... f. Mari tablets use the form tiqtulū for the third person masculine plural (e. "The Sun-goddess heard").Akk.-ā *ti-lta-.AMA tá-sa-a /tassa'ā/... f.. m..-ā ni- (y)i-. because of the sex of the name bearer) and texts (e..-ā ta-. m.g..-l 'a- yttt-.. ne-'à-la-a Inihallall..-ūn ti-... fern.. Ugaritic HeWew Aramaic Sing...... 2 ya-.. . f. " ) . "they have hammered . O..Bab.g.. i-ta-ha-u fyitahhawū/.-ā ... A prefix ti. .370 MORPHOLOGY *Pr.g.g. Ebla texts provide examples of affixes which are identical with those of Old Akkadian: third person singular masculine (e...-Ū yi-.. timhasū . "they will come near").. tištayū .. Besides... .-ā ti-..-ā ni- Ì-.(-«) . ..mu /Tašma'Sepešl.-n 'a/'i- yi-lyati-ftati-ltati-/ta-.18. . na-tì-lu ti-na-ta-ú Inātilū tinattalū/. an-na áš-táma /'anna 'aštama'/... .-u ti-. tikkulū ... rubbed themselves . t-.... c f. .-a 1 Plur.ā ta-.g.. tiltaptū.. 3 2 1 m..-n n- 40..g. Palaeosyrian and Amorite are not included in this table because a full paradigm cannot be established as yet on the basis of the available evidence and because this evidence points to important dialectal varia­ tions.-īn 'i- (>- Dual 3 m.. 3 2 1 m. drunk .'ū (y)i-. i-.ā \ r (y)i-.-ū ti-lta-... "the wailers strike up"). Ti-iš-te-Damu. like in Ugaritic and in Old Canaanite texts from the later d d d 2 D 4 .g.-ān ni- f. O.-ān ti-..g.a ta-. ..-ūn yi-..g...-Sem..--a J *ta-. ..-nā ni-lna- yi-..-« yi-lya-..a ? *ta-.. .-l iitata-.-ū ti-lta-.-ū *ti-/ta-. yi-. lš-má-ll /YiŠma'-'Il/. . and this form occurs at Ebla as well (e.....-Ū i-.of the third person feminine singular appears in some names (e.. "Damu has drunk". first person singular (e. "the Sun-goddess will dry bricks"). . However. . third person plural masculine (e.. " I I heard") and feminine (e. there is a distinct dual feminine (e. ... Ba-li-haa siG. "we purify").-ā ta-. eaten . Thus...-nā ti-lta-. . U T U ti-a-ba-an /tilabban/ SIG GAR... " I heard myself") and plural (e..

.-ā •-y -y ydtd-. . . ni-lna-lnu-ln- .. in Amurru (e.21). . a-pákà-ru /lapakkarū/. and in Canaan (li-ba-lu-ut-ni. Arabic Coll. .15)....2.. n d d d .23.-i 'a- yi-lya-Hti-lta-ltu-ltdti-lta-ltu-ltdti-lta-ltu-ltd-..-o td-.-ā td-. As for Amorite. "may he protect": EA 169. ti-lta-ltu-lt-.-u ya-. ? ? half of the second millennium B. as na-nasa-ab Inanassabl.-o ...-i 'd- ya­ tdtdtd-.. and Iu-um-ra-as-El. there are Ebla texts with verbal forms having a-prefixes of the first person plural. . .. yi-lya-lyu-lh.-n t- yd-.-6 ya-..-dm td-.-Ū ta-. li-na-sa-sár. .. "he should take"...-u(mln) .. 39.20)..-dn td-. but most cases do decidedly not conform to this principle.-ul-aynl-ēnl-an -u(mln) -ul-aynl-ēnl-an . and Iš-ma.. "they should join". liqattal is encountered later in the precative-optative function at Alalakh (e.-a ta-. indicated by a..ACTOR AFFIXES CI. or Ia-ás-ma-ah.-i d- yatdtdta-./laltaqqah/.-dm td-.-ā ta-. (§40.-na na- yi-lya-lyu-ll. Finally. "we are staying". a-a-tá-qá.m...\M..-dn nd- ya-.. li-nasa-ru-su.16). a-na-pá-ap /lanappap/. important dialectal variations are shown by third person forms like Ia-am-ru-us-Èl.2).13. d- ya-.. "Haddu did hear".{-il-iri) a-lni-lnd- ytt? ? yatdtdta-. -u td-./yi-/.-Ū ya-..-na ta-. just like the _yi-prefix may be expressed by i..-w y-. Ia-am-ra-as-El.-ā nd- ydtd-. Arabic Sabaic Mehri Ge'ez 371 Amharic yarntata-. .g..g. -ti-ba-al /tib'al/.. e./Yib'al/... "may they protect him").. .. "El did care".. and of the third person. "may he give me life": EA 198.-U y-. "they should join" (dual).. ti-lta-ltu-lt-. As a matter of fact./ya-/ or rather precative-optative /la-/ (§38..M. like all the names in I-ba-al. Some of these examples seem to favour Barth's law (§40.... Ia-ás-mi-ih.C.g. a-pá-kà-ra /lapakkarā/. 40... These variations and the probable precative-optative use of the imperfective — a construction which would be unusual in East Semitic — obviously reflect the intrusion of local forms into a text writ­ ten originally in another language (§4. "he should besprinkle". ..30).

tilqūna . Besides.22. "may they both take pos­ session".21. Dual forms are rarely encountered in Old Assyrian and in Old Babylonian. ilānu tiddinū bāštaka. one must reckon with exceptions.may occur in Assyrian and in archaizing or poetical Baby­ lonian. " I shall ask").and feminine ta. as well as in Late Babylonian which happens to reflect the spoken Aramaic language. 40.in the Sargonic period.18) and in Old Canaanite as reflected in the Amarna correspon­ dence. The Ugaritic feminine dual is reconstructed in the paradigm (§40.".. it occurs occasionally in the 13th12th centuries B. u tidūkūna. but they are attested at Emar (§40. and for the -t. and they will k i l l " (EA 104. The form tiqtulū{na) occurs along with the usual yiqtulū{nà) also in Palaeosyrian documents (§40.infixed stems which secondarily derive from the two above-mentioned stems. "he conquered". in texts from Emar where it reveals an influence of the local North Semitic idiom.17) according to two nearly contemporaneous North Semitic forms from Emar: lu-ú ta-as-bu-ta . or i-ik-mi. The variation of vowel pattern Ha in Hebrew prefixes is inde­ pendent from the thematic vowel.32-34). for the stem with prefix n-. 40. However. Barth's law seems to be generally operative in Ugaritic. "He comforted. but spellings like i-ig-mu-ur. and from personal names in syllabic script (e. except in the first person singular where the laryngal occasions .2). "He kept peace. The vowel of the prefix is generally i (yi-/ti-). but the dual is normally replaced by the plural.. A l l the other stems take actor-affixes of Set I I .. Ig-ma-raiM /Yigmar-Haddu/. as it appears from the sequence 'i . The distinction of the third person singular masculine i.g.g. obviously under North Semitic influence. The personals of Set I are used for the basic stem (§41. suggest that the prefix may have been yi..is never indicated as such. d 5 40.C.. " I shall dispatch".20. especially the one from Byblos.and -tan.21).19. "he captured". 'iš'al... and lu-ú ta-ad-di-na. "may they both give". as shown e. "Haddu has completed". The distribution of the two sets of Assyro-Babylonian dual personals is the same as in Old Akka­ dian.372 MORPHOLOGY 40. and in texts from Kāmid el-Loz.g. The Old Akkadian prefix yi. Iš-la-ma-na /Yišlamānu/. "may the gods give you influence" (Kāmid el-Loz 6.18-19)."). "they will take ..g. e. 'il'ak. by Ia-an-ha-mu /Yanhamu/.'a of prefixed and thematic vowels in forms of the first person singular (e. The third masculine plural form is either yqtl(n) or tqtl(n).

There are third per­ son forms with preformatives /. and only «-prefixes are found in the canonical readings of the Qur'ān. among the non-canonical or sāúfd-readings some /-prefixes occur in verbal forms with the thematic vowel a. 40. In modern Arabic colloquials. "he will get up").g. However.is still preserved in early Syriac inscriptions (e. yāqūm. while the Middle Aramaic of Hatra and Ashur prefixes /. ta-. the latter follow a contrasting vocalization. The a-prefixes are used in Classical Arabic for the active con­ jugation of the basic stem (I) and of Stems V-XV. According to Sibawayh and other early philologists. and the u. as well as masculine and feminine plural.25. e. 40. yahhpūn. but it can change into a before a guttural (e.or n. e. "he will drink". 'a-. However. In Syriac. "he will .g.13).is an allophone of 'a-. "he will get up").g.which originated from the preca­ tive or optative (§39. ne. "they will pass over") and be reduced to a short d in open unstressed syllables before a mono­ syllabic radical CvC (e. yif'alu (yiqtal) in conformity with Barth's law.became the standard prefix of the third person masculine singular.g. as shown by the Jewish Babylonian vocalization and by the Jewish Yemenite traditional pronunciation. The vowel of the prefix is likewise a or its allophone e before a guttural (§27. The vowel of the prefixes is generally i in Aramaic. the /-pre­ fix is used when the thematic vowel is / or a. "do not go near this tree" (Qur'ān 2. and it is generally a > ā before monosyllabic verbal roots of the type CvC (e. e.10).prefixes ya-.g. "she will write") goes very likely back to an original a and implies a variant set of a. yhz'. na-.g. the vowel e attested in Galilean Aramaic and in Syriac (e.ACTOR AFFIXES 373 the change i > a. Hedjaz was the only region where the prefixes of the ^-imperfect had not the vowel i. the ^-prefixes are restricted to the Dosiri dialect as spoken in Kuwait and to other colloquials of the Persian Gulf region. ydqūm. lā tiqrabā hādī š-šiġra. but the imperfect prefix y. i. Insb.to the third person in the imperfect (e.g.e.33/35).g.or /-prefix appears when the thematic vowel is u. However.23. yuktub or yiktub. yaktib. tektob. the usual 'e. "he will write". In general. like in Classical Arabic. "he sees"). "he takes away").g.24. There is reason to believe therefore that the /-prefixes were old-inherited in Arabic and that the choice of the <at-prefix for the a-imperfects in Classical Arabic results from a systematization of the language. regardless of the the­ matic vowel. 40. but yišrab.

is probably due to the Aramaic substratum. niktib.26. except in Maghrebine Arabic where this prefix is ni. A characteristic feature of Mesopotamian vernaculars. nikitbu. "we shall write"). "let him break". " I shall write"). The prefix vowel can also be reduced to d. shared by Bedouin dialects in North and Central Arabia.g. However. is the ending -in.(e. nidmek.g. as suggested by the following examples taken from Syriac. since the vowel d in Ge'ez originates either from / or from u. from the Mardin dialect in Anatolia.g. 'd. This m-prefix of the singu­ lar is already attested in early Andalusian Arabic as transmitted by Pedro de Alcalá. fern. td-. However.374 MORPHOLOGY write". 40. The vowel a is char­ acteristic of the prefix of the first person singular (e. plur.pre­ fixes. but many verbs display the internal vowel change and the -i suffix. while the Qatabanic masculine plural is yf'lwn. aktib.28. yâskár. but its persistence in Mesopotamia. In Minaic there seems not to be any graphic differentiation between masculine singular and plural. The distinction between the second person singular masculine and feminine has disappeared in several Arabic colloquials. . tektdbln tektdbūn Mardin taktabln taktabūn Baghdad tkitbln tkitbūn Persian Gulf taktdbln taktdbūn 40. " I had slept". It is attested also in Western Neo-Aramaic with new for­ mations based on old participles. §40. 40. Depending on the type of verb. nsofar. ndktdb. the second person singular feminine is characterized in Śheri and in Soqotri by an internal vowel change which appears instead of the -i suffix.prefixes are found in the affirmative jussive in Gurage. The paradigm of the Sabaic simple imperfect (without the -n or -nn ending) is incomplete and the feminine plural is only dubiously attested. " I have cried".g. e. m. aktub. e. Thus in Gurage: yásbàr. and by dialects spoken along the Persian Gulf and in Dofar. " I travel". -ūn of the second person feminine singular and of the second and third persons plural. " I shall write"). The Ge'ez yd-. sing. " I shall cry".prefixes go most likely back to i. a.g.17). This form is common in Ugaritic and in West Semitic (cf. nkdtbu. 2 pers. The affixes of the prefix-conjugation are the same in all the Modern South Arabian languages. next to gayaht [sayaht]. from the Moslem dialect of Baghdad. like for the first person plural which ends in -u (e. and from the dialects of the Persian Gulf (ktb.27. despite the contrary use of Classical Arabic. and Gafat. Harari. "to write"): Syriac 2 pers. this morphological feature may occur also in Mehri. nigéh [nisêh].

fern.prefix (e. Besides.g. an East Gurage dialect.in Amharic (e. nàsbdr. "while I find").29. In Modern South Arabian.prefix appears in the first person when the verb is introduced by a conjunction (e. 40. fern. Chaha àràkdb. e. the prefix of the first person singu­ lar of the imperfect is also à in various Gurage dialects (e.g. "may they bury". The vowel d instead of á after the prefixes t.in Harari (e. and in East Gurage.in the jussive (e. tdnrak dm. Mehri terkēz. hsbàr.13) and their use was extended to the imperfect.g.g. nàsbàr. but also in Argobba. "let him arrive".g. and in Harari. "let me arrive"). especially in North Ethiopic Tigre (e. 40. and masc. after elision of the glottal stop or of the initial y-. "you find". The ^-prefixes were thus used at least in some of the South Ethiopian languages and their preservation marks an archaic state which parallels the North Ethiopic i. "let me break"). in East Gurage (e. "they go out") and perhaps in some Gurage dialects. "they find". hfagra. "let me break").g. the dis­ tinction is lost not only in Amharic. tdfagro. like Ge'ez.ACTOR AFFIXES 375 "let him be drunk". ydràkbàma. "you go out".30. while it is n. "they go out". "let me make"). "let him break". "let me break") and in other Gurage dialects (e. Selti làsbár. These prefixes are precative or optative preformatives (§39. "let her break") of the languages using the yâprefix is therefore to be explained by analogy with the vowel d of the imperfect.23). as indicated e. in the second and third persons plural.g. Idfagro.and n. Soqotri liqbdr. Idfagro. as shown e. in Gafat: yàltàm. where the n.31. The jussive prefix of the first person singular is /. "let me break"). in Harari: yàsbār. "may I stand up".g. the vowel of the prefix is a after any consonant. Gafat. Harari.g. by Tigre imperfect plural forms masc. like Chaha. nasâkkdt. Modern North Ethiopic makes a distinction in gender. tdsbár.g.16). /. ydrākbo. in Gafat (e. tdràkbo.is prefixed to subjunctive forms of the verb which begin with a vowel. by Chaha imperfect plural forms masc. and it is also ā in Soddo (North Gurage) in the first person singular after the n.g. like in Aramaic (§40. " I find"). fern. tdrākbāma. Idltam.prefixes underlying the Ge'ez set of dprefixes. The expected vowels of the prefixed Proto-Semitic personals in Set I are a in the singular and i in the plural (§40. but the assimila- .g. and masc. fern. The distinction is kept instead in West and North Gurage. y 40. In Selti. tdfagra.g. In South Ethiopic.

In particular. Aramaic.forms in North Semitic and in Old Canaanite (§40. Set I I 40. and Amharic. has 'a. while the plural ti. despite their use of the suffix yi-. are to be considered as results of later developments. as a rule. exactly as in modern Arabic collo­ quials.376 MORPHOLOGY tory effect of y. in Aramaic.21) confirm the antiquity of the /-vowel in the plural prefixe. the y/-prefix occurs in Palaeosyrian. the first person singular. Instead. South Arabian.instead of the expected 'u-. The feminine ending -ā is identical with that of the dual and it is used in Assyro-Babylonian for the second person plural of both genders. and Ethiopic. . Addi­ tional comments will be found in §40. Mehri. Ge'ez. regardless of the thematic vowel of the verb. The second set of the prefix and suffix elements of the prefixconjugations is characterized by the vowel u in the prefix. and in North Ethiopic. in a large area of Arabic. these languages are omitted in the follow­ ing table. very rightly.in the singular is confirmed by the sec­ ond and first persons where a is employed with the prefixes ta. However. Since no distinctive pattern is recognizable for Set I I in Sabaic.32. endings which are broadly reflected in Old Babylonian and in Ge'ez. Palaeosyrian and Amorite are not included because only some forms can be established on the basis of the available evidence.33-36. which is the unique form in Ugaritic where the vowel u would be recognizable. while i is predominant in the Amarna correspondence and in modern Arabic colloquials.on the following vowel occasioned the change of a into the homorganic i in most languages (§22.18. the original use of ya. in Old Akkadian.and 'aeven in Palaeosyrian. on their most ancient attestations. distinctive suffixes have to be posited for the genders of the third and the second persons plural: -ū for the masculine and -ā for the feminine. The vowel i is. The complete harmonization of the prefix vowels or the alternative contrasting vocalization. as formulated in Barth's law. but this assumed u is reduced to 9 in Hebrew. while a harmonizing ten­ dency obliterated the difference between singular and plural. Besides. and in Assyro-Babylonian. the best attested for the prefix of the first person plural. Therefore.14). the reconstruction of Proto-Semitic affixes is not based. in Old Akkadian.

. 1.-ū td-. Aramaic..g.-Ū tu-.-ā O.. III.. .-l 'u- 00»'t(i)t(i)t(i)-.Bab. 1 yu-.. "she puts on".-īn tdtd-.. .. " I shall cause to flow". . Uš-taš-ni-Èl /Yuštatnī-'El/.. f... Sing3 ni.... .lyuqattarāl.. Despite the incomplete evidence.. the prefix of the first person singular in Set I I is vocalized with a. . .. 40..16).43-47).-ūn yd-...-ā nu- (y)i-.-ā *tu-..-ūn td-.. . and Colloquial Arabic against East Semitic and Classical Arabic.34.. -ā tu-. 'aqrb.... 3 m.-nā td-.. . which should appear in this position as '6. . .-ū *(y)u-....-en t(i)-.of the first person singular.-ā (y)u-.-u (y)i-. 40.. tu-. ... 'a- tdtdte-. 2 m.-ān td-.-ā tu-.-n tt-.ACTOR AFFIXES 377 Aramaic C I .. . .. Here Ugaritic agrees with Hebrew. .23). but the 'a of the first person singular does not reflect 'u. f.. .-u t(í)-.-ā nuyu-.-Sem.. In Colloquial Arabic...g. . the prefix of the third person is na.-ā *tu-.. u -qá-ta-ra.36.33..-Ū yu-. "we let her go". and IV.. .. The reconstructed Old Akkadian forms parallel those of Set I (§40.-ū ylt-. . Ugaritic Hebrew yutututu-. .-n n- yd-.Akk. 2 m.-na tu-. .-ā tu-...-nā nd- yd-. Arab *Pr. .-ū yu-. 1'. "they will burn incense") and in Amorite (e.35.. 'ašhlk. e. tu-a-baáš /tulabbaš/. and for the passive forms of all the stems (§41... " 2 1 Plur. .-ī u- yttt'a- ydÎ3- yutututu-..-ū td-.-en n(i)- 40... 1 Dual 3 m. the use of the characteristic vowel u is nevertheless well attested both in Palaeosyrian (e..like in Set I (§40.-ū *tu-. Set I I is used in Classical Arabic for the active forms of Stems I I . . Arabic Coll.. The vowel u is reduced in Hebrew and in Aramaic to d. In Syriac. "El acted for the second time").-Ī (')u- uututu-. 9 40. .-ā nu- U-... with n. A 'â- 'â- yu-... tu-.. O. " I shall bring near". In Ugaritic. . there is an overwhelming use of the vowel i with the exception of the prefix 'a.-ā u-...-ā tu-.-ā tu-.g..-ā (y)u-.-ū tu-..-na nu- u-.(-n) t-.. 'u- (y)ututuÎU-. 1'..-ān nd- yu-. nu-wa-sa-ra-si /nuwaššaraši/.

3 m. I. or ū. This additional vowel variation should be distinguished from the stem-vowel or thematic vowel of the verb. which belongs to the root. and inflects for Suffix-Conjugation Libyco-Berber ("qualitative") Egyptian (old perfective) Old Babylonian Neo-Assyrian (stative) (stative) Ugaritic (perfect) Mishnaic Hebrew (perfect) Sing. the Semitic verb has a set of stems or themes in which formal changes correspond to certain semantic variations. which w i l l be examined in a separate section (§44). f. f. 1 hnin hnindt hnimd hninad hnindġ sdm(w) sdm-t{l) sdm-t(i) sdm-kwi I ki I k sdm-wy sdm-ty lamad lamdat lamdāt{a) lamdāti lamdāku lamad lamdat lamdāt(i)/āk(a) lamdāt(i) lamdāk(afu) katab katbat katabta katabti katabtu katbā katabtā 1 kātab kātabā kātabtā kātabt kātabtī (lamdā) ktbny sdm(wly) sdm-ti hninit sdm-tiwny sdm-wyn lamdū lamdā lamdātunu lamdātina lamdānu lamdū lamdā lamdātun{u) lamdātin{a) lamdān(i) I ākuniu) katbū *katbā (?) katabtum katabtin *katbān (?) kātdbū k3tabtem kdtabten kātabnū . The biconsonantal verbs. 2 1 Plur. different vowel patterns can also determine an active and a passive voice of the stem or theme (§41. generally have or initially had either a long thematic vowel ā. Stems and Voices 41. 2 m. 1 Dual 3 m. In West Semitic and in Modern South Arabian.43-47). tenses.1. i. f. 3 m.2. f. a) Basic Stem 41. and aspects. or a short vowel and a geminated second radical consonant. 2 m. It shows the three consonants of the root with the thematic vowel. or u. Besides moods. The triconsonantal verbs are divided into three classes characterized by the vowels a. There are traces of some other patterns as well. The simple or basic stem is either called Stem I or it is desig­ nated by the symbols B(asic) or G(rundstamm. f.378 MORPHOLOGY E . in German).

while the change i > a and u > d led to a twofold scheme in Ethiopic (e. Argobba. ya-ntin. Sometimes the vocalic variation has semantic implications. and some Gurage dialects Suffix-Conjugation Old Aramaic (perfect) Syriac (perfect) Cl. "he goes away". Gafat.g. "he kills"). and in West Semitic (e. "he strikes". "he tans". (perfect) Mehri (perfect) Ge'ez (perfect) Tigre (perfect) Amharic (perfect) katab katbat katabta katabti katabtu/i ktab ketbat ktabt ktabt ketbet kataba katabat katabta katabtì katabtu katabà katabatá katabtumā katab katbet katabt katabti katabt ktab katbat ktabti ktabti ktabt katob katabot katabk katabš katabk katabo katabto katabki katdbki nagara nagarat nagarka nagarki nagarku nagra nagrat nagarka nagarki nagarko nàggàrà nāggārāàà nāggārhlk nāggārš nággárhu/ku katabū katabā katabtūn katabtīn katabnā ktabūn ktabēn ktabton ktabtēn ktabn(an) katabtū katabna katabtum katabtunna katabnā katabu katabu katabtu katabtu katabna katbu katbu ktabtlw ktabtlw ktabna katowb katob katabkam katabkan katoban nagaru nagarâ nagarkammu nagarkan nagarna nagraw nagraya nagarkum nagarkan nagarna nàggàru nāggāru nàggāraàiahu nàggàraiíahu nâggárn . Assyro-Babylonian i-lmad. (perfect) Maghrebine Coll. "he separates". ya-qtul-u.16). "may he dress"). "he approached". e. Classical Arabic ya-dhab-u. Amorite ya-bhar.g. ya-dbiġ-u. but South Semitic certainly had a threefold vocalic scheme at an earlier stage. "he learned".17. ya-dbaġ-u. 40. before it became the basic stem of the entire system. in AssyroBabylonian. The threefold vocalic scheme is attested in East Semitic (e. e.STEMS AND VOICES 379 tense. i-prus. i-pqid. "he separated"). yd-lbas. "he remembered"). for mood. ya-dkur-. and for actor. viz.g. Ar. "he considers". Dialectal differences may affect the stem-vowel. Ge'ez yd-ngdr.g. Amharic. i-qrab or i-qrib. As suggested by the vowel ali of the prefixed personals. "he moves away". in Arabic ya-fsil-u. A formal trace of this shift is preserved by the gemination of the second radical consonant in the positive suffix-conjugation of several South Ethiopian languages. both verbs in Arabic. in functional opposition to Stem II.g. "he chose". ya-drib-u. and even ya-dbuġu. "he gave". ya-hsab-u or ya-hsib-u.g. and ya-fsul-u. in North Semitic (e. Stem I probably represented the conjuga­ tion of intransitive verbs (§38. (perfect) Damascene Coll. "he delegated". "may he speak".

380 MORPHOLOGY (§41. "to achieve"). "he said"). yi-dbil ti-dbil ti-dbil-a ti-dbil-i 'a-dbil yī-dbil tī-dbil tī-dbil-a ti-dbil-i 'ī-dbil i-lkâm td-lkâm td-lkâm-dd td-lkâm-dd lkâm-dġ yd-lkdm ta-lkam td-lkam-ad td-lkâm-dd dlkdm-dġ ilmad ilmad talmad talmadī almad ilmad talrnad talmad talmidī almad yaktub iaktub taktub taktubin 'aktub yiktob tiktob tiktob tiktabī 'ektob ual. ilmadà tn. or in the lack of finite forms of Stem I in the conjugation of a number of transi­ tive verbs (e. m. m.g. Other signs of this functional shift can still be traced back in the passage of some particular verbs from Stem I I to Stem I (§63. This gemination. f. East Semitic qu"ū.g. f. quwwū.3) when Stem I was developing into a transitive conjugation form. ti-dbil-na ni-dbil tī-dbil-na nī-dbil .53). Hebrew dibber. f. East Semitic gamāru and gummuru. extended from the positive perfect to the per­ fect throughout. in the synonymy of Stems I and I I in numerous other cases (e. Prefix-Conjugation: //?r«s-Type Bedja (past) ("condi­ tional") Libyco-Berber (preterite / (jussive) perfective) Old Babylonian (preterite) NeoAssyrian (preterite) Mishnaic Hebrew (imperfect) Ugaritic (perfective) mg. was most likely produced by analogy with the perfect of Stem I I or Ethiopic I.2/B (§41.2). m. f. "to expect". yi-dbil-na yī-dbil-na lkâm-dn Ikâm-mt td-lkâm-am ta-lkâm-ami m-lkâm dlkdm-dn dlkam-ant îa-lkdm-am îa-lkzm-dmî m-lkam ilmadū ilmadā talmadā talmadâ nilmad ilmudū ilmadā talmadā talmadā nilmad yaktubūina) taktubna taktubū{na) *taktubā(l) naktub yìkidbū yiktabū tìktdbū tìktabú niktob m. f.

Prefix-Conjugation: Damascene Coll. of the Bedja conjugation (-dbil-. are added for comparison. the / is a spirantized final t. Only attested Semitic verbs are used in this paragraph. Arabic (jussive) Maghrebine Coll.STEMS AND VOICES 381 The suffix. "to hear"). "to learn". The Libyco-Berber emphatics -d and -ġ repre­ sent Afro-Asiatic pharyngalized stops followed by a vowel. (imperfect) iprus-Typc Old Aramaic (imperfect) Syriac (imperfect) CI. "to speak".) Tigre (jussive) Amharic (subj.) Ge'ez (subj. "to be gracious") and in Tuareg (-dlkem-. The vocalization of Ugaritic and of Old Aramaic is based on analogy with vocalized proper names. "to write".and prefix-conjugation of the basic stem is inflected in the prin­ cipal Semitic languages as shown in the following tables where paradigms of the Egyptian old perfective or "pseudo-participle" (sdm. "to collect"). nagara.) xiktub tiklub liktub íiktubīn 'aktub nektob tektob tektob tektdbīn 'ektob yaktub taktub taktub taktubī 'aktub bydktob btektob btaktob btsktbi bdktob iktdb tdktdb tdktdb tdktdb ndktdb ydktēb tdktēb tdktēb tdktēbi tektēb ysngdr tdngar tangar tangari 'angar langar tangar tangar tangari 'angar yangàr tangàr tangār tangàri langàr yaktubā taktubā taktubā yaktdbo taktdbo tdktdbo hktdbo bysktbu bydkibu btdktbu htdkthu mnaktob īkatbu īkdtbu tkdtbu tkatbu rikatbu yakìēbdm tdktdbzn tdktēbzm tdktdban naktēb yangaru yangarā tangaru tangarā nangar langaro langara tangaro tangara nangar yangāru yangàru tangàru tangàru annangár yiktubūn yiktubān tiktubūn /iktubān niktub nektdbūn nektdbān tektdbūn tektdbān nektob yaktubū yaktubna taktubū taktubna naktub . (imperfect) Mehri (subj. lamādu. and of the Libyco-Berber verb. ktb. as inflected in Kabyle (hnin. thus -ta / -ti and ku. viz. "to fol­ low").

382 MORPHOLOGY Prefix-Conjugation: iparras-Type Bedja (imperfective) Sing. 2 m. it is likely that this stem originally represented the conjugation of transitive verbs (§38. Considering the function of Stem I I and the vowel u of the prefixed personals.17. are of a nongeminating type. as Modern South Arabian. just as Cushitic Bedja. and most Gurage dialects. 2 m. which exhibits ver­ bal formations with vocalic modifications alone that nevertheless match the two main functions of Stem I I in Semitic.16).2). It is generally designated by the symbol D(oubled or "Doppelungsstamm" in German) which alludes to the "doubling" of the second radical. f. in functional opposition to Stem I (§41. 3 m. f. 3 m. however. some Semitic languages. this characteristic of Stem I I is well preserved in the historically attested languages (2°). As result of particular developments. 1 Libyco-Berber (imperfective) i-lākkdm û-lâkkdm ti-lākkdm-dd ti-lākkdtn-dd lākkem-dġ Old Babylonian (present-future) ilammad ilammad talammad talammadī alammad Neo-Assyrian (present-future) ilammad talammad talammad talammidī alammad (yi-)danbīl (rì-)danbīl (ti-)danbīl-a (ti-)danbīl-i 'a-danbīl Dual 3 m. Harari. 2 ilammadā 1 Plur. 1° Stem I I is called also "intensive" in consideration of its function in expressing repetition or spatial dispersion. for instance. f. Stem I I with geminated or lengthened second radical consonant is attested over the whole Semitic area. f.3. f. 1 lākkdm-dn lākkdm-dnt td-lākkdm-9m td-lākkdm-dmt 'ē-dbìl-na nd-lâkkdm ilammadū ilammadā talammadā talammadā nilammad ilammudū ilammadā talammadā talammadā nilammad nē-dbīl tē-dbīl-na b) Stem with Geminated Second Radical Consonant 41. 40. and in indicating plurality of the object in the transitive verbs and plurality of the subject in the intran- .

Old Babylonian butuqtam ibattaq. Arabic tabata./ y. "to massacre".7).in sense (§41. "he will open breaches". but qatta'a. ginif. "he made fast"). "to cause carnage". "to k i l l " . but tabbata.g. Thus. "he heard". šibib. sibab. e. The . but bitter. 2° When inflecting intransitive verbs of Stem I . rimad. "he cut into pieces") and par­ ticularly of quantity (e.g. rimid. Here too. "to choke". but buîuqātim ubattaq. dar. "to arrive".g.STEMS AND VOICES 383 Prefix-Conjugation: Ugaritic (imperfective) *yakattub *takattub *takattub *takattubin * 'akattub iparras-Type Mehri (imperfect) yakotab takotab takotab takētab dkotdb Ge'ez (imperfect) yanaggar tanaggar tanaggar tanagri 'anaggar Tigre (imperfect) lanaggar tanaggar tanaggar tanagri 'anaggar Amharic (imperfect) yanagr tanàgr tanágr tanagri anāgr yaktabo taktabo taktabo aktabo *yakattabū takattubna *takattabū(na) *takattaba(l) *nakattab yakatbam takatban takatbam takatban nakotab yanagru yanagrā tanagru tanagrā nanaggar lanagro lanagra tanagro tanagra 'annaggar yanagru yanagru tanagru tanagru annanāgr sitive ones. "to make kneel down". both of qualitative result (e. fal. Arabic qata'a. "to pour out". There is a parallel rule in Bedja that the verb must be in the intensive stem when the subject or the direct object are in plural. sikal. "he was firm". thus approximating the stem with preformatives š. "he will open a breach". transitive dir. "he cut off". "he cut" one thing.g. kātim. but šimma'. it denotes intensity. Stem I I gives them a causative and transitive sense (e. "to kneel down".g. "to be choking". Bedja offers parallels like ginaf. "to see one's self". "he cut" several things). sikil. Hebrew šāma'. ///./ '. "to arrive repeatedly"./ h. "to avenge one's self". "he gave to hear"). then it is used also as a factitive of transitive verbs (e. "to avenge". intransitive kitim. "to overflow". but this stem is formed in Bedja by modifications of the stem vowel. Hebrew bātar. "to see".

A particular feature is the vowel e in the Ge'ez imper­ fect ydqettdl of Stem I. "he accused of lying"). This is a replacive vowel aimed at distin­ guishing the imperfect from the jussive ydqattdl. Arabic kadaba. 4° The corresponding Stem I.g. Whether a verb is of Stem I . jussive.2/B is then a question of vocabulary and of usage in the language. In consequence.4). this much discussed e does not result from the monoph­ thongization of a diphthong ay which would have been morphologically and phonologically unexplainable. from qdtoret. ydqattil. Hebrew qittēr.384 MORPHOLOGY causative form of these intransitive verbs is characterized by the vowel i like the Semitic D-stem (uparris.2/B of Ethiopic is no longer a derived stem. "he was near". §41. Maghrebine Coll. The following table is limited to the forms of the third person masculine sin­ gular in the Semitic languages presented in §41. Ge'ez qarba. Only one Assyro-Babylonian paradigm is given below for East Semitic.g. "he lied". but owe their pre­ sent shape to a misinterpretation of South Arabian imperfective forms of Stem I metanalyzed as Stem I I verbs. Ge'ez Tigre Amharic purrus qtl qittēl qtl qattel fa' 'ala fa' 'al fd"dl qattala qattala qattala 2 uparras yqtl 3 uparris yqtl ydqattēl yqtl tidqattel yufa' 'Hu blfa' 'el yifd' 'dl ydqattdl ydqattdl ydqattdl ydqettdl ydqattdl ydqattdl . but kaddaba. but a basic stem. e. 3° In Semitic. The three forms shown below belong to the suffix-conjugation (1).g. and West Semitic imperfect (3): 1 Assyro-Babylonian Ugaritic Hebrew Old Aramaic Syriac Classical Arabic Damascene Coll. "he made sacrificial smoke". It had to be e < ē in order to avoid its reducing to d in a form which already contained two d. the D-stem can have two supplementary functions related to the causative: it can be declarative (e.2. and to the preterite.2/B. There are very few exceptions. and the usual paradigmatic verbs are being used for the sake of clarity. and qarraba. yufa"il). except Mehri which is a language of a non-geminating type (cf. "he brought near". and it is quite often denom­ inative (e. It should be noticed that some Stem II verbs in dictionaries of Classical Ara­ bic probably not hark back to historically "intensive" forms. l / A or Stem I. to the East and South Semitic imperfective (2). "smoke of sacrifice").

already used in some grammars. Ge'ez degana. kātaba. South Arabian. Mehri arokdb. However. i. This stem is attested in Arabic (fā'ala. Stem I I I of Arabic indicates an action directed towards an object. A variant of Stem I I I with a diphthong derived from the long vowel. "he put (a pot) on the fire" (with a vowel prefixed to a voiced or glottalized first radical). Mehri Geez Tigre Amharic fā 'ala 2 3 yufā'ilu hīfā 'el yifā 'al fā'al fā'al fo'dl qātala qātala qattala yafa'hn yafo'dl ydqattdl ydqātdl ydqattdl ydqātdl ydqattdl ydqatl 41. "he tormented". m . q or a rounded consonant b . "he set fire to"). kaf ala. qātala. i. Besides. "he cut". is attested in Arabic (e. but as a basic stem. either an attempt to accomplish something (conative. represents not only the so-called third stem of Arabic (fā'ala).5. It must be distinguished from the Ethiopic verbs with the vowel o after the first radical which was originally a labiovelar g . k .4. and in Syriac (e. as con­ firmed by its either conative (Stem III) or intensive (Stem II) meaning. Maghrebine Coll. the long vowel may also replace the gemination of the second radical. zāmala.3: 1 Classical Arabic Damascene Coll.g. The presentation is the same as in §41. in Ethiopic (e.g. Tigre gādala. lāyana. horab. e.e. "he put on socks". but Stem I I I of Ethiopic (I.STEMS AND VOICES 385 c) Stem with Lengthened First Vowel 41. "he rang". and Ethiopic. Stem I I I with lengthened first vowel may be designated by the symbol L(engthening). yufā'ilu) and in Ethiopic (e. cut in many pieces". Ge'ez šāqaya.3/C) is no longer felt by native speakers as a derived stem.g. Mehri (d)CdCdC. Chaha ydbandr.g. "he kept company"). "he dissected.g. "he treated with kindness". this derived stem.g.g. e. "he tried to k i l l " .g. e. in consequence. ġawraba. gawzel. "he fought") or a correlative motion towards someone (e. "he demolishes"). f. "he corresponded". "he pursued"). The w w w w w w . e. but also the second stem (fa"ala). except in Tigre.e. "he song war songs"). The following table is limited to Arabic. but often reduced to 6 or ē. a stem with lengthened first vowel appears in Modem South Arabian.g. given that internal gemination is not a feature of derived verbal stems in Modem South Arabian. Amharic q dttdrd > qottdrd.

this variant stem appears mainly in denom­ inative verbs. ur ilkim. "he caused to pile up". In fact. but kātim. Bedja kitim.of Stem I I I are rare in Ethiopic and in Syriac. Arabic battala. This interpretation is supported by the use of these verbs with the causative reflexive affix at. Tuareg ilkâm. in particular the Gafat. "he didn't follow". by the intensive meaning of Stem I I I of many Arabic and Ethiopic verbs. ì) (1) {7) (7 41. sawġar. This opinion is confirmed by the situation in Modern South Arabian. as ilākkam. e. "he behaved violently". "to arrive". Gafat kimmàrā < *ka